【布莱克威尔哲学导读系列】心灵哲学 429页

TheBlackwellGuidetoPhilosophyofMind\nBlackwellPhilosophyGuidesSeriesEditor:StevenM.Cahn,CityUniversityofNewYorkGraduateSchoolWrittenbyaninternationalassemblyofdistinguishedphilosophers,theBlackwellPhilosophyGuidescreateagroundbreakingstudentresource–acompletecriticalsurveyofthecentralthemesandissuesofphilosophytoday.Focusingandadvanc-ingkeyargumentsthroughout,eachessayincorporatesessentialbackgroundmaterialservingtoclarifythehistoryandlogicoftherelevanttopic.Accordingly,thesevolumeswillbeavaluableresourceforabroadrangeofstudentsandreaders,includingprofessionalphilosophers.1TheBlackwellGuidetoEpistemologyEditedbyJohnGrecoandErnestSosa2TheBlackwellGuidetoEthicalTheoryEditedbyHughLaFollette3TheBlackwellGuidetotheModernPhilosophersEditedbyStevenM.Emmanuel4TheBlackwellGuidetoPhilosophicalLogicEditedbyLouGoble5TheBlackwellGuidetoSocialandPoliticalPhilosophyEditedbyRobertL.Simon6TheBlackwellGuidetoBusinessEthicsEditedbyNormanE.Bowie7TheBlackwellGuidetothePhilosophyofScienceEditedbyPeterMachamerandMichaelSilberstein8TheBlackwellGuidetoMetaphysicsEditedbyRichardM.Gale9TheBlackwellGuidetothePhilosophyofEducationEditedbyNigelBlake,PaulSmeyers,RichardSmith,andPaulStandish10TheBlackwellGuidetoPhilosophyofMindEditedbyStephenP.StichandTedA.Warfield\nTheBlackwellGuidetoPhilosophyofMindEditedbyStephenP.StichandTedA.Warfield\n©2003byBlackwellPublishingLtd350MainStreet,Malden,MA02148–5018,USA108CowleyRoad,OxfordOX41JF,UK550SwanstonStreet,CarltonSouth,Melbourne,Victoria3053,AustraliaKurfürstendamm57,10707Berlin,GermanyTherightofStephenP.StichandTedA.WarfieldtobeidentifiedastheAuthorsoftheEditorialMaterialinthisWorkhasbeenassertedinaccordancewiththeUKCopyright,Designs,andPatentsAct1988.Allrightsreserved.Nopartofthispublicationmaybereproduced,storedinaretrievalsystem,ortransmitted,inanyformorbyanymeans,electronic,mechanical,photocopying,recordingorotherwise,exceptaspermittedbytheUKCopyright,Designs,andPatentsAct1988,withoutthepriorpermissionofthepublisher.Firstpublished2003byBlackwellPublishingLtdLibraryofCongressCataloging-in-PublicationDataTheBlackwellguidetophilosophyofmind/editedbyStephenP.StichandTedA.Warfield.p.cm.–(Blackwellphilosophyguides;9)Includesbibliographicalreferencesandindex.ISBN0-631-21774-6(alk.paper)–ISBN0-631-21775-4(pbk.:alk.paper)1.Philosophyofmind.I.Stich,StephenP.II.Warfield,TedA.,1969–III.Series.BD418.3.B572003128′2–dc212002071221AcataloguerecordforthistitleisavailablefromtheBritishLibrary.Setin10/13ptGalliardbyGraphicraftLimited,HongKongPrintedandboundintheUnitedKingdombyMPGBooksLtd,Bodmin,CornwallForfurtherinformationonBlackwellPublishing,visitourwebsite:http://www.blackwellpublishing.com\nContentsContributorsviiIntroductionix1TheMind–BodyProblem:AnOverview1KirkLudwig2TheMind–BodyProblem47WilliamG.Lycan3Physicalism65AndrewMelnyk4Dualism85HowardRobinson5ConsciousnessanditsPlaceinNature102DavidJ.Chalmers6ThoughtsandTheirContents:NaturalizedSemantics143FredAdams7CognitiveArchitecture:TheStructureofCognitiveRepresentations172KennethAizawa8Concepts190EricMargolisandStephenLaurence9MentalCausation214JohnHeilv\nContents10FolkPsychology235StephenP.StichandShaunNichols11Individualism256RobertA.Wilson12Emotions288PaulE.Griffiths13ArtificialIntelligenceandtheManyFacesofReason309AndyClark14PhilosophyofMindandtheNeurosciences322JohnBickle15PersonalIdentity352EricT.Olson16FreedomoftheWill369RandolphClarkeIndex405vi\nContributorsFredAdamsisProfessorofPhilosophyattheUniversityofDelaware.KennethAizawaisCharlesT.BeairdProfessorofPhilosophyatCentenaryCollegeofLouisiana.JohnBickleisProfessorofPhilosophyandProfessorintheGraduateNeuroscienceProgramattheUniversityofCincinnati.DavidJ.ChalmersisProfessorofPhilosophyattheUniversityofArizona.AndyClarkisProfessorofPhilosophyandDirectoroftheCognitiveSciencesProgramatIndianaUniversity.RandolphClarkeisAssociateProfessorofPhilosophyattheUniversityofGeorgia.PaulE.GriffithsisProfessorofHistoryandPhilosophyofScienceattheUniver-sityofPittsburgh.JohnHeilisPaulB.FreelandProfessorofPhilosophyatDavidsonCollege.StephenLaurenceisaSeniorLecturerinPhilosophyattheUniversityofSheffield.KirkLudwigisAssociateProfessorofPhilosophyattheUniversityofFlorida.WilliamG.LycanisWilliamRandKenanJr.ProfessorofPhilosophyattheUniversityofNorthCarolina.EricMargolisisAssociateProfessorofPhilosophyatRiceUniversity.vii\nContributorsAndrewMelnykisAssociateProfessorofPhilosophyattheUniversityofMissouri.ShaunNicholsisAssociateProfessorofPhilosophyattheCollegeofCharleston.EricT.OlsonisUniversityLecturerinPhilosophyandFellowofChurchillCollege,UniversityofCambridge.HowardRobinsonisProfessorofPhilosophyatCentralEuropeanUniversity.StephenP.StichisBoardofGovernorsProfessorofPhilosophyandCognitiveScienceatRutgersUniversity.TedA.WarfieldisAssociateProfessorofPhilosophyattheUniversityofNotreDame.RobertA.WilsonisProfessorofPhilosophyattheUniversityofAlberta.viii\nIntroduction1ThisvolumeisanotherintheseriesofBlackwellPhilosophyGuides.Itcontains16newessayscoveringawiderangeofissuesincontemporaryphilosophyofmind.Authorswereinvitedtoprovideopinionatedoverviewsoftheirtopicandtocoverthetopicinanywaytheysawfit.Thisallowedthemthefreedomtomakeindividualscholarlycontributionstotheissuesunderdiscussion,whilesimultaneouslyintroducingtheirassignedtopic.Ihopethatthefinishedproductprovessuitableforuseinphilosophyofmindcoursesatvariouslevels.Thevolumeshouldbeagoodresourceforspecialistsandnon-specialistsseekingoverviewsofcentralissuesincontemporaryphilosophyofmind.InthisbriefintroductionIwilltrytoexplainsomeofthereasonswhyphilosophyofmindseemstobesuchanimportantsub-fieldofphilosophy.Iwillalsoexplainmyviewofthesourceofthegreatdiversityonefindswithinphilosophyofmind.Thisdiscussionwillleadtosomecommentaryonmethodologicalissuesfacingphi-2losophersofmindandphilosophersgenerally.Fewphilosopherswoulddisagreewiththeclaimthatphilosophyofmindisoneofthemostactiveandimportantsub-fieldsincontemporaryphilosophy.Philoso-phyofmindseemstohaveheldthisstatussinceatleastthelate1970s.Manywouldmakeanddefendthestrongerclaimthatphilosophyofmindisunequivo-callythemostimportantsub-fieldincontemporaryphilosophy.Itsstatuscanbeattributedtoatleasttworelatedfactors:theimportanceofthesubjectmatterandthediversityofthefield.Mentalphenomenaarecertainlyofgreatimportanceinmost,ifnotall,humanactivities.Ourhopes,dreams,fears,thoughts,anddesires,togivejustsomeexamples,allfigureinthemostimportantpartsofourlives.Somemaintainthatmentalityisessentialtohumannature:thatatleastsomesortofmentallifeisnecessaryforbeinghumanorforbeingfullyhuman.Othersmaintainthatspecificfeaturesofhumanmentality(perhapshumanrationality)distinguishhumansfromothercreatureswithminds.Whetherornottheseambitiousclaimsarecorrect,thementalisatleastofgreatimportancetoourlives.Whowoulddenythatix\nIntroductionthoughts,emotions,andothermentalphenomenaarecentrallyinvolvedinal-mosteverythingimportantaboutus?Thisobvioustruthonlypartlyexplainstheimportanceofphilosophyofmind.Thesizeanddiversityofthefieldalsodeserve3somecreditforthisstanding.Aquickglanceatthisvolume’stableofcontentswillgivesomeindicationof4thebreadthofthefield.Inadditiontoessaysontopicscentraltocontemporaryphilosophyofmind,suchasmentalcontent,mentalcausation,andconsciousness,wefindessaysconnectingthephilosophyofmindwithbroadlyempiricalworkofvariouskinds.Thisempiricallyorientedworkcoversareasinwhichphilosophersmakecontactwithbroadempiricalpsychologicalworkon,forexample,theemo-tionsandconcepts.Theintersectionsofphilosophywithbothneuroscienceandartificialintelligencearealsotopicsofseriouscontemporaryinterest.Incontrasttothisempiricallyorientedwork,wealsoseeessaysontraditionalphilosophicaltopicssuchasthemind–bodyproblem,personalidentity,andfreedomofthewill.Thesetopics(especiallythelattertwo)areoftenclassifiedasapartofcontemporarymetaphysicsbuttheyare,traditionally,apartofphilosophyofmindandsotheyareincludedinthisvolume.Despitetheseinitialclassificationsofworkaseither“traditional”or“empiricallyoriented,”oneshouldnotassumethatthisdistinctionmarksasharpdivide.Itispossibletoworkontraditionaltopicswhilebeingsensitivetorelevantempiricalwork;andmakinguseoftraditionalphilosophicaltools,suchassomekindofconceptualanalysis,isprobablynecessarywhendoingempiricallyorientedphilo-sophyofmind.Whatonefindsinthefieldarenotperfectlyprecisemethodologicaldivisions.Rather,onefindsdifferencesinthedegreetowhichvariousphiloso-phersbelieveempiricalworkisrelevanttophilosophyofmindanddifferencesin5thedegreetowhichphilosopherstrytoavoidtraditionalphilosophicalanalysis.Thebreadthanddiversityofphilosophyofmindisnotfullycapturedinasurveyoftopicsarisinginthefieldandinhighlightingdifferentapproachesthataretakentothoseprojects.Inadditiontoawiderangeoftopicsanddifferentapproachestothesetopics,wealsofindasomewhatsurprisinglistofdifferentexplanatorytargetswithinthisfield.Aphilosopherdoingphilosophyofmindmightbeprimarilyinterestedinunderstandingorexplainingthehumanmindor,moremodestly,somefeaturesofthehumanmind.Alternatively,onemightbeinterestedinexaminingthebroaderabstractnatureof“mentality”or“mindedness”(humanorotherwise).Onemightalsofocusonourconceptofthehumanmind,orourconceptofmindsgenerally,withorwithoutanyparticularviewofhowour6conceptofthesethingsrelatestotherealityofthesubjectmatter.Thesediffer-entpossibletargetsofinquiryatleastappeartoleadtoverydifferentkindsofquestions.Despitetheapparentdifferences,however,thislargevarietyofprojectfallsquitecomfortablyundertheumbrellaheadingof“philosophyofmind.”Thediversityofphilosophyofmindbecomesevenclearerwhenonerealizesthatonecanmixandmatchthevarioustargetsofinquiryandthedifferentmethodologies.Onemightbeinterestedinalargelyempiricalinquiryintoourconceptofthehumanmind.Alternatively,onemightbeinterestedinabroadlyx\nIntroductionconceptualinquiryintotheexactsamesubjectmatter.Thedifferentmethodolo-gies(andagain,recallthatthesedifferencesarebestthoughtofasdifferencesofdegreenotkind)canalsobeappliedininvestigationsofthenatureofthehumanmindorthenatureofmentality.Wemightexpectmethodologicaldisputestobreakoutasphilosopherstakedifferentapproachestodifferenttopicswithinphilosophyofmind.Forexample,thosefavoringtraditionalapriorimethodologymightchallengeempiricallyori-entedphilosopherswhoclaimtoreachconclusionsaboutthenatureofthehumanmindprimarilythroughempiricalworktoexplainhowtheybridgetheapparentgapbetweenthewayhumanmindsareandthewaytheymustbe.Similarly,empiricallyorientedphilosophersofmindmightchallengethosefavoringapriorimethodstoexplainwhytheythinksuchmethodscanreachconclusionsaboutanythingotherthantheconceptsofthosedoingtheanalysis.Why,forexample,shouldwethinkthatananalysisofourconceptofthemindisgoingtorevealanythingaboutthemind?Perhaps,thecriticismmightcontinue,ourcon-ceptofminddoesnotaccuratelyreflectthenatureofthemind.Unfortunatelyandsurprisingly,however,discussionsofthesemethodologicalissuesarenot7common.Fortunatelytheseandrelatedmethodologicalissuesalsoariseinotherareasofphilosophy,andthereseemstobeagrowinginterestinunderstandingandcommentinguponvariousapproachestophilosophicalinquiryinsideand8outsideofphilosophyofmind.Contributorstothisvolumewerenotaskedtocommentonmethodologicalissuesinphilosophyofmind.Theyweresimplyinvitedtointroduceanddiscusstheirassignedtopicinwhateverwaytheysawfit,usingwhatevermethodologytheychosetobringtothetask.Inadditiontothinkingaboutthefirst-orderphilosophicalissuesunderdiscussionintheseoutstandingessays,readersareinvitedtoreflectonthemethodologicalandmetaphilosophicalissuesrelevanttothediscussions.Perhapssuchreflectionwillhelpusbetterunderstandsomeorallofthetopicsweencounterinthephilosophyofmind.TedA.WarfieldNotes1Avolumeofthissortdoesnotcometogethereasily.Ithankthecontributorsfortheirvaryingdegreesofpatienceandsupportasweconfronteddifficultiesatvariousstagesofthisproject.Iespeciallythankmyco-editorforhisunwaveringsupportandguid-ance.Forhelpfuldiscussionofsomeoftheissuesarisinginthisbriefintroduction,IthankmycolleaguesLeopoldStubenbergandWilliamRamsey.Idonotthankmyemployer,theUniversityofNotreDame,thoughitdidkindlyallowmetheuseofacomputerandprinterwhileatworkonthisproject.2Thevolumecontainstwodistinctopeningessaysonthemind–bodyproblem.Inintroducingthevolume,Iresistthetemptationtowriteathirdsuchessayandinsteadfocusonafeworganizationalandmethodologicalissues.xi\nIntroduction3Thesepartialexplanationstogetherstilldonotfullyexplainthestatusofphilosophyofmindwithincontemporaryphilosophy.Ethics,forexample,istremendouslyimportantandisalsoalargeanddiversefield.Iamunabletofullyexplainthestatusofphiloso-phyofmind.Thoughnowabitdated,TylerBurge’simportantessay“PhilosophyofLanguageandMind:1950–1990”(PhilosophicalReview,101(1992),pp.3–51)con-tainssomehelpfulideasaboutthismatter.4Butnoonevolumecouldreallycoverthisentirefield.Onehelpfuladditionalresource,agoodsupplementalresourcetothisvolume,isTheBlackwellCompaniontoPhilosophyofMind,editedbySamuelGuttenplan(Blackwell,1994).5Thesamephilosophermighteventakedifferentgeneralmethodologicalapproachestodifferentproblemsoreventothesameproblematdifferenttimes.6Onecaneasilyimaginehowonemightconclude,forexample,thatourconceptofmindisinsomesensea“dualistic”concept,butnotthinkitfollowsfromthisthatdualismisthecorrectpositiononthemind–bodyproblem.7Somerecentdebatesaboutconsciousnesshaveincluded,ataveryhighlevelofsophis-tication,somemethodologicaldiscussionalongtheselines(see,forexample,DavidJ.ChalmersandFrankJackson’s“ConceptualAnalysisandReductiveExplanation,”Philo-sophicalReview,110(2001),315–60.8AnyonewishingtoexploretheseissuescouldprofitablybeginwithMichaelR.DePaulandWilliamRamsey(eds.),RethinkingIntuition(RowmanandLittlefield,1998).xii\nTheBlackwellGuidetoPhilosophyofMindTheMind–BodyProblem:AnOverviewEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter1TheMind–BodyProblem:AnOverviewKirkLudwigIhavesaidthatthesoulisnotmorethanthebody,AndIhavesaidthatthebodyisnotmorethanthesoul,Andnothing,notGod,isgreatertoonethanone’sselfis.WaltWhitman1.1IntroductionUnderstandingtheplaceofthoughtandfeelinginthenaturalworldiscentraltothatgeneralcomprehensionofnature,aswellasthatspecialself-understanding,whicharetheprimarygoalsofscienceandphilosophy.Thegeneralformoftheproject,whichhasexercisedscientistsandphilosopherssincetheancientworld,isgivenbythequestion,‘Whatistherelation,ingeneral,betweenmentalandphysicalphenomena?’Thereisnosettledagreementonthecorrectanswer.Thisisthesinglemostimportantgapinourunderstandingofthenaturalworld.Thetroubleisthatthequestionpresentsuswithaproblem:eachpossibleanswertoithasconsequencesthatappearunacceptable.Thisproblemhastraditionallygone1undertheheading‘TheMind–BodyProblem.’Myprimaryaiminthischapteristoexplaininwhatthistraditionalmind–bodyproblemconsists,whatitspossiblesolutionsare,andwhatobstacleslieinthewayofaresolution.Thediscussionwilldevelopintwophases.Thefirstphase,sections1.2–1.4,willbeconcernedtogetclearerabouttheimportofourinitialquestionasapreconditionofdevelopinganaccountofpossibleresponsestoit.Thesecondphase,sections1.5–1.6,explainshowaproblemarisesinourattemptstoanswerthequestionwehavecharacterized,andsurveysthevarioussolutionsthatcanbeandhavebeenoffered.Morespecifically,sections1.2–1.4areconcernedwithhowtounderstandthebasicelementsofourinitialquestion–howweshouldidentifythemental,onthe1\nKirkLudwigonehand,andthephysical,ontheother–andwithwhatsortsofrelationsbetweenthemweareconcerned.Section1.2identifiesandexplainsthetwotraditionalmarksofthemental,consciousnessandintentionality,anddiscusseshowtheyarerelated.Section1.3givesanaccountofhowweshouldunderstand‘physical’inourinitialquestionsoasnottoforecloseanyofthetraditionalpositionsonthemind–bodyproblem.Section1.4thenaddressesthethirdelementinourinitialquestion,mappingoutthebasicsortsofrelationsthatmayholdbetweenmentalandphysicalphenomena,andidentifyingsomeforspecialattention.Sections1.5–1.6areconcernedwithexplainingthesourceofthedifficultyinansweringourinitialquestion,andthekindsofsolutionsthathavebeenofferedtoit.Section1.5explainswhyourinitialquestiongivesrisetoaproblem,andgivesapreciseformtothemind–bodyproblem,whichispresentedasasetoffourpropositions,eachofwhich,whenpresentedindependently,seemscompelling,butwhicharejointlyinconsistent.Section1.6classifiesresponsestothemind–bodyproblemonthebasisofwhichofthepropositionsinourinconsistentsettheyreject,andprovidesabriefoverviewofthemainvarietiesineachcategory,togetherwithsomeofthedifficultiesthatariseforeach.Section1.7isabriefconclusionaboutthesourceofourdifficultiesinunderstandingtheplaceofmind2inthenaturalworld.1.2MarksoftheMentalThesuggestionthatconsciousnessisamarkofthementaltracesbackatleastto3Descartes.Consciousnessisthemostsalientfeatureofourmentallives.AsWilliamJamesputit,“Thefirstandforemostconcretefactwhicheveryonewillaffirmtobelongtohisinnerexperienceisthefactthatconsciousnessofsomesort4goeson”(James1910:71).Astateorevent(achangeofstateofanobject)ismental,onthisview,ifitisconscious.States,inturn,areindividuatedbythepropertiesthehavingofwhichbyobjectsconstitutestheirbeinginthem.Identifyingconsciousnessasamarkofthementalonlypushesourquestiononestepback.Wemustnowsaywhatitisforsomethingtobeconscious.Thisisnoteasytodo.Therearetwoimmediatedifficulties.First,inG.E.Moore’swords,“themomentwetrytofixourattentionuponconsciousnessandtoseewhat,distinctly,itis,itseemstovanish:itseemsasifwehadbeforeusamereemptiness...asifitwerediaphanous”(1903:25).Second,itisnotclearthatconsciousness,evenifwegetafixonit,isunderstandableinotherterms.Tosaysomethingsubstantiveaboutitistosaysomethingcontentiousaswell.Forpresentpurposes,however,itwillbeenoughtoindicatewhatweareinterestedininawaythateveryonewillbeabletoagreeupon.WhatIsaynowthenisnotintendedtoprovideananalysisofconsciousness,butrathertodrawattentionto,andtodescribe,thephenomenon,inmuchthesamewayanaturalistwoulddrawattentiontoacertainspeciesofinsectorplantbypointingoneout,ordescribing2\nTheMind–BodyProblem:AnOverviewconditionsunderwhichitisobserved,anddescribingitsfeatures,featureswhichanyoneinanappropriatepositioncanhimselfconfirmtobefeaturesofit.First,then,weareconsciouswhenweareawakeratherthanindreamlesssleep,and,insleep,whenwedream.Whenweareconscious,wehaveconsciousstates,whichwecandiscriminate,andrememberaswellasforget.Eachconsciousmentalstateisamode,orway,ofbeingconscious.Knowledgeofourconsciousmentalstates,evenwhenconnectedinperceptualexperienceswithknowledgeoftheworld,isyetdistinctfromit,asisshownbythepossibilityofindistinguish-ableyetnon-veridicalperceptualexperiences.Consciousmentalstatesincludeparadigmaticallyperceptualexperiences,somaticsensations,proprioception,painsanditches,feelingsadorangry,orhungerorthirst,andoccurrentthoughtsanddesires.InThomasNagel’sevocativephrase,anorganismhasconsciousmentalstatesifandonlyif“thereissomethingitisliketobethatorganism”(1979b:166).Thereis,incontrast,nothingitislikeintherelevantsense,itisusuallythought,tobeatoenail,orachair,orabladeofgrass.Intryingtocapturethekindsofdiscriminationwemakebetweenmodesofconsciousness(orwaysofbeingconscious),itissaidthatconsciousstateshaveaphenomenalorqualitativecharacter;thephenomenalqualitiesofconsciousmen-talstatesareoftencalled‘qualia’.Sometimesqualiaarereifiedandtreatedasiftheywereobjectsofawarenessinthewaytablesandchairsareobjectsofpercep-tion.Butthisisamistake.Whenoneisawareofone’sownconsciousmentalstatesortheirphenomenalqualities,theonlyobjectinquestionisoneself:whatoneisawareofisaparticularmodificationofthatobject,awayitisconscious.Similarly,whenweseearedapple,weseejusttheapple,andnottherednessasanotherthingalongsideit:rather,werepresenttheappleweseeasred.Astrikingfeatureofourconsciousmentalstatesisthatwehavenon-inferentialknowledgeofthem.Whenweareconscious,weknowthatweare,andweknowhowweareconscious,thatis,ourmodesofconsciousness,butwedonotinfer,whenweareconscious,thatweare,orhowweare,fromanythingofwhichwe5aremoredirectlyaware,orknowindependently.Itisnotoriouslydifficulttosaywhatthiskindofnon-inferentialknowledgecomesto.Itisdifficulttoseehowtoseparateitfromwhatwethinkofasthequalitativecharacterofconsciousmental6states.Arguablythis“first-person”knowledgeissuigeneris.Thereisarelatedasymmetryinourrelationtoourownandothers’consciousmentalstates.Wedonothavetoinferthatweareconscious,butothersmustdoso,typicallyfromourbehavior,andcannotknownon-inferentially.Othershave,atbest,“third-person”knowledgeofourmentalstates.Thesespecialfeaturesofconsciousstatesareconnectedwithsomeofthepuzzlesthatarisefromtheattempttoanswerouropeningquestion.Consciousnesshasoftenbeenseenasthecentralmysteryinthemind–bodyproblem,andtheprimaryobstacletoanadequatephysicalist7understandingofthemental.Theothertraditionalmarkofthemental,firstarticulatedclearlybyFranz8Brentano(1955[1874],bk2,ch.1),iscalled‘intentionality’.Theadjectivalformis‘intentional’.Butthisisatechnicalterm,anddoesnotjustinvolvethose3\nKirkLudwigstatesthatinEnglisharecalled‘intentions’(suchasmyintentiontohaveanothercupofcoffee).Intentionality,rather,isthefeatureofastateoreventthatmakesitaboutordirectedatsomething.Thebestwaytomakethiscleareristogivesomeexamples.UnlikethechairthatIamsittinginasIwrite,Ihavevariousbeliefsaboutmyself,mysurroundings,andmypastandfuture.IbelievethatIwillhaveanothercupofcoffeebeforethedayisout.Mychairhasnocorrespond-ingbelief,noranyother.Beliefsareparadigmaticallyintentionalstates.Theyrepresenttheworldasbeingacertainway.Theycanbetrueorfalse.Thisistheirparticularformofsatisfactioncondition.InJohnSearle’saptphrase,theyhavemind-to-worlddirectionoffit(1983:ch.1).Theyaresupposedtofittheworld.Anystatewithmind-to-worlddirectionoffit,anyrepresentationalstate,oratti-tude,isanintentionalstate(inthetechnicalsense).Falsebeliefsarejustasmuchintentionalstatesastrueones,evenifthereisnothingintheworldforthemtobeaboutofthesorttheyrepresent.Icanthinkaboutunicorns,thoughtherearenone.Therepresentationcanexistwithoutwhatitrepresents.Itisthissenseof‘aboutness’or‘directedness’thatisatissueinthinkingaboutintentionality.Thereareintentionalstateswithmind-to-worlddirectionoffitinadditiontobeliefs,suchasexpectations,suppositions,convictions,opinions,doubts,andsoon.Notallintentionalstateshavemind-to-worlddirectionoffit,however.Anotherimportantclassisexemplifiedbydesiresorwants.IbelieveIwill,butalsowanttohaveanothercupofcoffeesoon.Thisdesireisalsodirectedatorabouttheworld,andevenmoreobviouslythaninthecaseofbelief,thereneednotbeanythingintheworldcorresponding.Butincontrasttobelief,itsaimisnottogetitscontent(thatIhaveanothercupofcoffeesoon)tomatchtheworld,buttogettheworldtomatchitscontent.Ithasworld-to-minddirectionoffit.Adesiremaybesatisfiedorfailtobesatisfied,justasabeliefcanbetrueorfalse.Thisisitsparticularformofsatisfactioncondition.Anystatewithworld-to-minddirectionoffitislikewiseanintentionalstate.Clearlytherecanbesomethingincommonbetweenbeliefsanddesires.IbelievethatIwillhaveanothercupofcoffeesoon,andIdesirethatIwillhaveanothercupofcoffeesoon.Thesehaveincommontheircontent,anditisinvirtueoftheircontentthateachisanintentionalstate.(Elementsincommonbetweencontents,whichwouldbeexpressedusingageneralterm,aretypicallycalled‘concepts’;thus,theconceptofcoffeeissaidtobeaconstituentofthecontentofthebeliefthatcoffeeisabeverageandofthebeliefthatcoffeecontainscaffeine.)Thecontentineachmatchesorfailstomatchtheworld.Thedifferencebetweenbeliefsanddesiresliesintheirroleinourmentaleconomy:whethertheirpurposeistochangesothattheircontentmatchestheworld(beliefs)ortogettheworldtochangetomatchtheircontent(desires).Stateslikethesewithcontentsthatwecanexpressusingsentencesarecalled‘propositionalattitudes’(atermintroducedbyBertrandRussell,afterthesupposedobjectsoftheattitudes,propositions,namedordenotedbyphrasesoftheform‘thatp’,where‘p’isreplacedbyasentence).Propositionalattitudesareindividuatedbytheirpsychologicalmode(belief,supposition,doubt,desire,aspiration,etc.)andcontent.Stateswithworld-to-minddirectionoffitare4\nTheMind–BodyProblem:AnOverviewproor,ifnegative,conattitudes.Therearemanyvarietiesbesidesdesiresandwants,suchashopes,fears,likes,dislikes,andsoon.Itisnotclearthatallrepresentationalcontentisfullypropositional.Ourper-ceptualexperiences,e.g.,ourvisual,auditory,andtactileexperiences,representourenvironmentsasbeingacertainway.Theycanbeveridical(correctlyrepre-sent)ornon-veridical(incorrectlyrepresent),asbeliefscanbetrueorfalse.Theyhavemind-to-worlddirectionoffit,hence,representationalcontents,andinten-tionality.Butitisnotclearthatallthattheyrepresentcouldbecapturedpropositionally.Attitudesandperceptualexperiencesmightbesaidtobedifferentcurrenciesforwhichthereisnoprecisestandardofexchange.Cantherebestatesdirectedatoraboutsomethingwhichdonothavefullcontents?Someonecouldhaveafearofspiderswithouthavinganydesiresdirectedatparticularspiders,thoughthefearisinasensedirectedatoraboutspiders.Yetafearofspidersdoesentailadesiretoavoidcontactwith,orproximityto,spiders:anditisthistogetherwithaparticularemotionalaurawhichthinkingoforperceivingspidersevokeswhichwethinkofasthefearofspiders.Inanycase,wewillcallthisclassofstatesintentionalstatesaswell,thoughtheirintentionalityseemstobegroundedintheintentionalityofrepresentational,orproorconattitudes,whichunderliethem,or,aswecansay,onwhichtheydepend.Wemay,then,saythatanintentionalstateisastatewithacontent(inthesensewe’vecharacterized)orwhichdepends(inthesensejustindicated)onsuch9astate.Astatethenisamentalstate(orevent)ifandonlyif(iff)itiseitheraconsciousoranintentionalstate(orevent).Anobjectisathinkingthingiffithasmentalstates.Whatistherelationbetweenconsciousstatesandintentionalstates?Ifthetwosortsareindependent,thenourinitialquestionbreaksdownintotwosubquestions,oneabouttherelationofconsciousness,andoneaboutthatofintentionality,tothephysical.Ifthetwosortsarenotindependentofoneanother,anyanswertothegeneralquestionmusttacklebothsubquestionsatonce.Someintentionalstatesareclearlynotconsciousstates.YourbeliefthatAus-tralialiesintheAntipodeswasnotaconsciousbelief(oranoccurrentbelief)justamomentago.Youwerenotthinkingthat,thoughyoubelievedit.Itwasadispositional,asopposedtoanoccurrent,belief.Thedistinctiongeneralizestoallattitudetypes.Adesirecanbeoccurrent,mypresentdesireforacupofcoffee,forexample,ordispositional,mydesiretobuyacertainbookwhenIamnot10thinkingaboutit.Thisdoesnot,however,settlethequestionwhetherinten-tionalandconsciousmentalstatesareindependent.Itmaybeanecessarycondi-tiononourconceivingofdispositionalmentalstatesasintentionalattitudesthatamongtheirmanifestationpropertiesareoccurrentattitudeswiththesamemodeandcontent.Inthiscase,thestrategyofdivideandconquerwillbeunavailable:wewillnotbeabletoseparatetheprojectsofunderstandingtheintentionaland11theconscious,andproceedtotackleeachindependently.Someconsciousmentalstatesseemtolackintentionality,forexample,certainepisodesofeuphoriaoranxiety.Thoughtypicallycausedbyourbeliefsand5\nKirkLudwigdesires,itisnotclearthattheyarethemselvesaboutanything.Likewise,somaticsensationssuchasitchesandpainsseemtohavenon-representationalelements.Typicallysomaticsensationsrepresentsomething’soccurringinone’sbody.Aheadacheisrepresentedasinthehead,atoeacheasinthetoe.Butthequalityofpainitself,thoughitbetakentobeabiologicalindicatorof,say,damagetothebody,inthewaythatsmokeindicatescombustion,seemsnottohaveanyassoci-atedrepresentationalcontent.Paindoesnotrepresent(asopposedtoindicate)damage.And,thoughweusuallywishpainweexperiencetocease,thedesirethatone’spaincease,whichhasrepresentationalcontent,isnotthepainitself,any12morethanadesireforalargerhouseisitselfahouse.1.3ThePhysicalCharacterizingphysicalphenomenainawaythatcapturestheintentionofourinitialquestionisnotaseasyasitmayappear.Wecannotsaythatphysicalphenomenaconsistinwhatourcurrentphysicstalksabout.Physicaltheorychangesconstantly;currentphysicaltheorymayundergoradicalrevision,aspastphysicaltheoryhas.Themind–bodyproblemdoesn’tchangewithpassingphysicaltheory.Thereareatleastthreeotheroptions.Thefirstistocharacterizephysicalphenomenaaswhattheultimatelycorrectphysicaltheorytalksabout,wherewethinkofphysicaltheoryasthetheorythattellsusaboutthebasicconstituentsofthingsandtheirproperties.Thesecondistotreatphysicalphenomenaasbydefinitionnon-mental.Therearereasonstothinkthatneitherofthesecapturesthesenseofourinitialquestion.Oneresponsetothemind–bodyproblemisthatthebasicconstituentsofthingshaveirreduciblementalproperties.Onthefirstinterpretation,suchapositionwouldbeclassifiedasaversionofphysicalism(wewillgiveaprecisecharacterizationofthisattheendofsection1.4),sinceitholdsthatmentalpropertiesare,intherelevantsense,physicalproperties.Butthisposition,thatthebasicconstituentsofthingshaveirreduciblementalproperties,isusuallythoughttobeincompatiblewithphysicalism.Thesecondinterpretationinitsturndoesnotleaveopentheoptionofseeingmentalphenomenaasconceptuallyreducibletophysicalphenomena.Ifthephysicalisnon-mentalperse,thenshowingthatmentalpropertiesarereallypropertiesthatfallincategoryFwouldjustshowthatasubcategoryofpropertiesincategoryFwerenotphysicalproperties.Butwewantthetermsinwhichourinitialquestionisstatedtoleaveitopenwhethermentalpropertiesareconceptuallyreducibletophysicalproperties.(Wewillreturntowhatthiscouldcometobelow.)Athirdoptionistotakephysicalphenomenatobeofageneraltypeexemplifiedbyourcurrentphysics.Herewewouldaimtocharacterizeaclassofpropertiesthatsubsumesthoseappealedtobypastandcurrentphysicaltheories,fromthescientificrevolutiontothepresent,butwhichisbroadenoughtocoverpropertiesappealed6\nTheMind–BodyProblem:AnOverviewtoinanyextensionofourcurrentapproachtoexplainingthedynamicsofmaterialobjects.Thisinterpretationleavesopentheoptionsforeclosedbyourfirsttwointerpretations,andcomportswellwiththedevelopmentofconcernsabouttherelationofmentaltophysicalphenomenafromtheearlymodernperiodtothepresent.Itisnoteasytosayhowtocharacterizetheintendedclassofproperties.Thecoreconceptionofthemisgivenbythosequalitiesclassedasprimaryqualitiesintheseventeenthandeighteenthcenturies:size,shape,motion,number,solidity,texture,logicalconstructionsofthese,andpropertiescharacterizedessentiallyintermsoftheireffectsonthese(massandcharge,e.g.,arguablyfallinthelast13category).Itisnotclearthatthisisadequatetocovereverythingwemightwishtoinclude.Butitisfairtosaythat,typically,philosophershaveinmindthisconceptionofthephysicalinposingthequestionwebeganwith,withouthaving14adetailedconceptionofhowtodelineatetherelevantclassofproperties.1.4Mind–BodyRelationsThequestionoftherelationbetweenthementalandthephysicalcanbeposedequivalentlyasaboutmentalandphysicalproperties,concepts,orpredicates.Apropertyisafeatureofanobject,suchasbeinground,orbeingthreefeetfromtheearth’ssurface.Aconcept,aswehavesaid,isacommonelementindifferentthoughtcontentsexpressedbyageneralterm.Wedeployconceptsinthinkingaboutathing’sproperties.So,correspondingtothepropertyofbeingroundistheconceptofbeinground,orofroundness.WhenIthinkthatthisballisround,andsothinkofitashavingthepropertyofbeinground,Ihaveathoughtthatinvolvestheconceptofbeinground.Iamsaidtobringtheballundertheconceptofroundness.Predicatesexpressconcepts,andareusedtoattribute15propertiestoobjects.Thus,‘isround’expresses(inEnglish)theconceptofroundness,andisusedtoattributethepropertyofbeinground.Wemaysayitpicksoutthatproperty.Foreverypropertythereisauniqueconceptthatisaboutit,andviceversa.Morethanonepredicatecanexpressthesameconcept,and16pickoutthesameproperty,butthentheymustbesynonymous.Correspondingtoeachpropertycategory(mentalorphysical,e.g.)isacategoryofconceptsandpredicates.Thus,anyquestionweaskabouttherelationofmentalandphysicalpropertiescanberecastasaboutconceptsorpredicates,andviceversa.Thebasicoptionsinthinkingabouttherelationofmentalandphysicalproper-tiescanbeexplainedintermsofthefollowingthreesentenceforms,where‘isM’representsamentalpredicate,and‘isP’representsaphysicalpredicate(thisisgeneralizablestraightforwardlytorelationalterms).[A]Forallx,ifxisP,thenxisM[B]Forallx,ifxisM,thenxisP[C]Forallx,xisMifandonlyif(iff)xisP7\nKirkLudwigThough[C]isequivalenttotheconjunctionof[A]and[B],itwillbeusefultostateitseparately.Therelationofthementaltothephysicalisdeterminedbywhichinstancesof[A]–[C]aretrueorfalse,andonwhatgrounds.Onecouldholdeachtobenecessarilytrueornecessarilyfalse,inoneofthreesensesof“necessity”:conceptual,metaphysical(so-called),andnomological.Twonotionsthatfigureprominentlyindiscussionsofthemind–bodyprob-lemcanbecharacterizedinthisframework.Thefirstisthatofreduction,andthesecondthatofsupervenience.Eachcanbeconceptual,metaphysical,ornomological.Ibeginwithconceptualreductionandsupervenience.Conceptualnecessitiesaretruthsgroundedintheconceptsusedtoexpressthem.Thisisthestrongestsortofnecessity.Whatisconceptuallynecessaryissoineverymetaphysicallyandnomologicallypossibleworld,thoughnotviceversa.Knowledgeofconceptualtruthscanbeobtainedfromreflectionontheconceptsinvolved,andneednotrestonexperience(traditionally,knowledgeofone’sownconsciousmentalstatesiscountedasexperientialknowledge).Theyarethussaidtobeknowableapriori.Knowledgeobtainedinthiswayisaprioriknowledge.Apropositionknownonthebasisofexperienceisknownaposteriori,orempir-ically.Knowledgesobasedisaposterioriorempiricalknowledge.Conceptualtruthsarenotrefutablebythecontentsofanyexperiences.Asentenceexpressing(inalanguageL)aconceptualtruthisanalyticallytrue(inL),or,equivalently,analytic(inL)(henceforthIomittherelativization).Asentenceisanalyticiff17itstruthisentailedbytruemeaning-statementsaboutitsconstituents.Forexample,‘NoneoftheinhabitantsofDublinresideselsewhere’,or‘Thereis18nogreatestprimenumber’wouldtypicallyberegardedasanalytic.Conceptualreductionofmentaltophysicalproperties,orviceversa,isthestrongestconnectionthatcanobtainbetweenthem.(Wesayequivalently,inthiscase,thatmentalconcepts/predicatescanbeanalyzedintermsofphysicalcon-cepts/predicates,orviceversa.)Ifamentalpropertyisconceptuallyreducibletoaphysicalproperty,thentwoconditionsaremet:(a)theinstanceof[C],inwhich‘isM’isreplacedbyapredicatethatpicksoutthementalproperty,and‘isP’bya(possiblycomplex)predicatethatpicksoutthephysicalproperty,isconceptu-allynecessary,and(b)theconceptsexpressedby‘isP’areconceptuallypriortothoseexpressedby‘isM’,whichistosaythatwehavetohavetheconceptsexpressedby‘isP’inordertounderstandthoseexpressedby‘isM’,butnotviceversa(thinkoftheorderinwhichweconstructgeometricalconceptsasanexample).Thesecondclausegivescontenttotheideathatwehaveeffectedareduction,foritrequiresthephysicalconceptstobemorebasicthanthementalconcepts.Aconceptualreductionofamentalpropertytoaphysicalpropertyshowsthementalpropertytobeaspeciesofphysicalproperty.Thisamountstotheidentificationofamentalpropertywithaphysicalproperty.Similarlyforthereductionofaphysicalpropertytoamentalproperty.Onecouldholdthatinstancesof[C]wereconceptuallynecessarywithoutholdingthateitherthementalorthephysicalwasconceptuallyreducibletotheother.Inthiscase,theirnecessarycorrelationwouldbeexplainedbyappealtoanotherset8\nTheMind–BodyProblem:AnOverviewofconceptsneitherphysicalnormental,intermsofwhicheachcouldbeunder-stood.Forexample,itisconceptuallynecessarythateverytriangleisatrilateral,butneitherofthesenotionsprovidesaconceptualreductionoftheother.‘Supervenience’isatermofartusedinmuchcurrentphilosophicalliteratureonthemind–bodyproblem.Itmaybedoubtedthatitisneededinordertodiscussthemind–bodyproblem,butgivenitscurrentwidespreaduse,nocon-temporarysurveyofthemind–bodyproblemshouldomititsmention.Avarietyofrelatednotionshasbeenexpressedusingit.Thoughvaryinginstrengthamongthemselves,theyaregenerallyintendedtoexpressthesesweakerthanreductionism,invokingonlysufficiencyconditions,ratherthanconditionsthatarebothneces-19saryandsufficient.Supervenienceclaimsarenotsupposedtoprovideexplana-tions,butrathertoplaceconstraintsontheformofanexplanationofonesortofpropertiesintermsofanother.Iintroducehereadefinitionofonefamilyofpropertiessuperveningonanother,whichwillbeusefulforformulatingapositionwewillcall‘physicalism’,andwhichwillbeusefullaterinourdiscussionofapositionontherelationofmentaltophysicalpropertiesknownas‘functionalism’.Ibeginwith‘conceptualsupervenience’.F-propertiesconceptuallysuperveneonG-propertiesiffforanyx,ifxhasapropertyffromF,thenthereisapropertygfromG,suchthatxhasganditisconceptually20necessarythatifxhasg,thenxhasf.Conceptualreductionofonefamilyofpropertiestoanotherimpliesmutualconceptualsupervenience.Butthesupervenienceofonefamilyofpropertiesonanotherdoesnotimplytheirreducibilitytothem.Iwillcharacterize‘physicalism’asthepositionaccordingtowhich,whatevermentalpropertiesobjectshave,theyconceptuallysuperveneonthephysicalproperties21objectshave,andwhateverpsychologicallawsthereare,thephysicallawsentailthem.Thisallowssomeonewhothinksthatnothinghasmentalproperties,andthattherearenomentallaws,tocountasaphysicalist,whateverhisviewaboutthe22conceptualrelationsbetweenmentalandphysicalproperties.Thedefinitionhereisstipulative,thoughitisintendedtotrackawidespread(thoughnot23universal)usageinthephilosophicalliteratureonthemind–bodyproblem.Thequestionwhetherphysicalismistrue,sounderstood,marksafundamentaldivideinpositionsonthemind–bodyproblem.Nomologicalnecessitywecanexplainintermsofconceptualnecessityandthenotionofanaturallaw.AstatementthatpisnomologicallynecessaryiffitisconceptuallynecessarythatifL,itisthecasethatp,where“L”standsinforasentenceexpressingallthelawsofnature,whetherphysicalornot(adding“bound-aryconditions”to“L”yieldsmorerestrictivenotions).Iofferonlyanegativecharacterizationofmetaphysicalnecessity,whichhasreceivedconsiderableattentionincontemporarydiscussionofthemind–bodyproblem.Iwillargueinsection1.6thatnoconceptcorrespondstotheexpression“metaphysicalnecessity”inthesecontexts,despiteitswidespreaduse.Fornow,wecansaythatmetaphysical9\nKirkLudwignecessityissupposedtobeofasortthatcannotbediscoveredapriori,butwhichisstrongerthannomologicalnecessity,andweakerthanconceptualnecessity.Toobtaincorrespondingnotionsofmetaphysicalandnomologicalsupervenience,wesubstitute‘metaphysically’or‘nomologically’for‘conceptually’inourcharac-terizationabove.Metaphysicalandnomologicalreductionrequirethatbiconditionalsoftheform[C]aremetaphysicallyornomologicallynecessary(butnothingstronger),respectively.Butreductionisasymmetric.Sowemustalsogiveasensetotheideathatonesideofthebiconditionalexpressespropertiesthataremorebasic.Inpractice,thequestionishowtomakesenseoftheasymmetryformetaphysicalornomologicalreductionofthementaltothephysical.Thereisnothinginthecaseofmetaphysicalornomologicalnecessitythatcorrespondstoconceptualpriority.Itlooksasifthebestwecandoistogroundthedesiredasymmetryinphysicalpropertiesbeingbasicinourgeneralexplanatoryscheme.Thisisusuallyunder-stoodtomeanthatthephysicalconstitutesanexplanatorilyclosedsystem,whilethementaldoesnot.Thismeansthateveryeventcanbeexplainedbyinvokingphysicalantecedents,butnotbyinvokingmentalantecedents.1.5TheMind–BodyProblemAphilosophicalproblemisaknotinourthinkingaboutsomefundamentalmatterthatwehavedifficultyunraveling.Usually,thisinvolvesconceptualissuesthatareparticularlydifficulttosortthrough.Becausephilosophicalproblemsinvolvefoundationalissues,howweresolvethemhassignificantimportforourunderstandingofanentirefieldofinquiry.Often,aphilosophicalproblemcanbepresentedasasetofpropositionsallofwhichseemtrueonaninitialsurvey,orforallofwhichtherearepowerfulreasons,butwhicharejointlyinconsistent.Thisistheforminwhichtheproblemoffreedomofthewillandskepticismabouttheexternalworldpresentthemselves.Itisasignificantadvanceifwecanputaprobleminthisway.Forthewaysinwhichconsistencycanberestoredtoourviewsdeterminesthelogicalspaceofsolutionstoit.Themind–bodyproblemcanbeposedinthisway.Historicalandcontemporarypositionsontherelationofthementaltothephysicalcanthenbeclassifiedintermsofwhichofthepropositionstheychoosetorejecttorestoreconsistency.Theproblemarisesfromtheappealofthefollowingfourtheses.1Realism.Somethingshavementalproperties.2Conceptualautonomy.Mentalpropertiesarenotconceptuallyreducibletonon-mentalproperties,and,consequently,nonon-mentalpropositionentails24anymentalproposition.3Constituentexplanatorysufficiency.Acompletedescriptionofathinginterms25ofitsbasicconstituents,theirnon-relationalproperties,andrelationsto10\nTheMind–BodyProblem:AnOverview26oneanotherandtootherbasicconstituentsofthings,similarlydescribed(theconstituentdescription)entailsacompletedescriptionofit,i.e.,anaccountofallofathing’spropertiesfollowsfromitsconstituentdescription.4Constituentnon-mentalism.Thebasicconstituentsofthingsdonothave27mentalpropertiesassuch.Thelogicaldifficultycannowbepreciselystated.Theses(2)–(4)entailthenega-tionof(1).Forifthecorrectfundamentalphysicsinvokesnomentalproperties,(4),andeverynaturalphenomenon(i.e.,everyphenomenon)isdeduciblefromadescriptionofathingintermsofitsbasicconstituentsandtheirarrangements,(3),thengiventhatnonon-mentalpropositionsentailanymentalpropositions,(2),wecandeducethattherearenothingswithmentalproperties,whichisthenegationof(1).Thelogicaldifficultywouldbeeasytoresolvewereitnotforthefactthateachof(1)–(4)hasapowerfulappealforus.Thesis(1)seemsobviouslytrue.Weseemtohavedirect,non-inferentialknow-ledgeofourownconsciousmentalstates.Weattributetooneanothermentalstatesinexplainingwhatwedo,andbaseourpredictionsonwhatotherswilldoinpartonourbeliefsaboutwhatattitudestheyhaveandwhattheirconsciousstatesare.Relinquishing(1)seemsunimaginable.Proposition(2)isstronglysupportedbytheprimafacieintelligibilityofabodywhosebehaviorislikethatofathinkingbeingbutwhichhasnomentallifeofthesortweareawareoffromourownpointofview.Weimaginethatourmentalstatescauseourbehavior.Itseemsconceivablethatsuchbehaviorresultsfromothercauses.Indeed,itseemsconceivablethatitbecausedfromexactlythephysicalstatesofourbodiesthatwehaveindependentreasonstothinkanimatethemwithouttheaccompanyingchoirofconsciousness.Itislikewisesupportedbytheprimafacieintelligibilityofnon-materialthinkingbeings(suchasGodandHisangels,whomevenatheistshavetypicallytakentobeconceivable).Thus,itseems,primafacie,thathavingamaterialbodyisneitherconceptuallynecessarynorsufficientforhavingthesortsofmentalliveswedo.Thoughtexperimentsaskustoimagineapossiblycontrarytofactsituationandaskourselveswhetheritappearsbarelytomakesense(notjustwhetheritiscompatiblewithnaturallaw)thatacertainstateofaffairscouldthenobtain.Wetypicallytestconceptualconnectionsinthisway.Forexample,wecanaskour-selveswhetherwecanconceiveofanobjectthatisredbutnotextended.Theansweris‘no’.Wecanlikewiseaskwhetherwecanconceiveofanobjectthatisredandshapedlikeapenguin.Theansweris‘yes’.Thisprovidesevidencethatthefirstisconceptuallyimpossible–ruledoutbytheconceptsinvolvedinitsdescription–andthatthesecondisconceptuallypossible–notruledoutbythe28conceptsinvolved.Nooneislikelytodisputetheresultshere.Butwecanbemisled.Forexample,itmayseemeasytoconceiveofasetthatcontainsallandonlysetswhichdonotcontainthemselves(theRussellset).Foritiseasytoconceiveasetwhichcontainsnosets,andasetwhichcontainssetsonly,andso11\nKirkLudwigitcanseemeasytoconceiveofaspecialsetofsetswhosemembersarejustthosesetsnotcontainingthemselves.Butitispossibletoshowthatthisleadstoacontradiction.Callthesetofallsetsthatdonotcontainthemselves‘R’.IfRisamemberofR,itfailstomeetthemembershipconditionforR,andsoisnotamemberofitself.Butifitisnotamemberofitself,thenitmeetsthemembershipconditionandsoisamemberofitself.So,itisamemberofitselfiffitisnot,29whichisacontradiction,andnecessarilyfalse.Therecannotbesuchaset.Thus,somethingcanseemconceivabletousevenwhenitisnot.Inlightofthis,itisopenforsomeonetoobjectthatdespitetheapparentintelligibilityofthethoughtexperimentsthatsupport(2),wehavemadesomemistakeinthinkingthem30through.Proposition(3)issupportedbythesuccessofscienceinexplainingthebehaviorofcomplexsystemsintermsoflawsgoverningtheirconstituents.Whiletherearestillmanythingswedonotunderstandabouttherelationofmicrotomacrophenomena,itlooksasifthetechniquessofarappliedwithsuccesscanbeextendedtothosefeaturesofcomplexsystemswedon’tyetunderstandfullyintermsoftheirconstituents’properties–withthepossibleexceptionofpsycho-logicalphenomena.Proposition(3)expressesathoughtthathashadapowerfulideologicalholdonourthescientificworldview,thatnatureisultimatelyintelligibleasakindofvastmachine,acomplexsystemacompleteunderstandingofwhichcanbeobtainedbyanalyzingitsstructureandthelawsgoverningthepropertiesofitsparts.“Ithasbeen,”inE.O.Wilson’swords,“testedinacidbathsofexperimentandlogicandenjoyedrepeatedvindication”(1998:5).Thisthoughtmotivatesmuchscientificresearch,andtogiveitupevenwithrespecttoapartofthenaturalworldwouldbetogiveupacentralmethodologicaltenetofourcurrentscientificworldview.Itwouldbetoadmitthatnaturecontainssomebasicelementofarbitrariness,inthesensethattherewouldbefeaturesofobjectsthatwerenotexplicableasarisingfromtheirmannerofconstruction.Finally,proposition(4)issupportedalsobythesuccessofphysics(sofar)inaccountingforthephenomenathatfallinitsdomainwithoutappealtoanymentalproperties.Inthecatalogofpropertiesofparticlephysics,wefindmass,charge,velocity,position,size,spin,andthelike,butnothingthatbearstheleasthintofthemental,andnothingofthatsortlookstoberequiredtoexplainthe31interactionanddynamicsofthesmallestbitsofmatter.Itcanseemdifficulteventounderstandwhatitwouldbetoattributementalpropertiestothesmall-estconstituentsofmatter,whichareincapableofanyoftheoutwardsignsofmentalactivity.Thisthenisthemind–bodyproblem.Propositions(1)–(4)allseemtobetrue.Buttheycannotallbe,fortheyarejointlyinconsistent.Thatiswhyourinitialquestion,“Whatistherelation,ingeneral,betweenmentalandphysicalphenom-ena?,”givesrisetoaphilosophicalproblem.Eachanswerwemightliketogivewillinvolverejectingoneofourpropositions(1)–(4);yet,consideredindependently,eachofthesepropositionsseemstobeonewehavegoodreasonstoaccept.12\nTheMind–BodyProblem:AnOverview1.6TheLogicalSpaceofSolutionsProposedsolutionstothemind–bodyproblemcanbeclassifiedaccordingtowhichof(1)–(4)theyrejecttorestoreconsistency.Thereareonlyfourbasicpositions,sinceweseekaminimalrevision.Toreject(1)istoadoptirrealismoreliminativismaboutthemental.Toreject(2)istoadoptconceptualreductionismforthemental.Thisincludesneutralmonism,psychophysicalidentitytheories,functionalism,andfunctionalism-cum-externalism.Toreject(3)istoadoptconceptualanti-reductionism,butnotontologicalanti-reductionism.Neutralemergentismandemergentmaterialismfallintothiscategory.Toreject(4)istoadoptontologicalanti-reductionisminadditiontoconceptualanti-reductionism.Thissubsumesvarietiesofwhatmightbecalled‘mentalparticletheories’,andincludessubstancedualism,idealism,panpsychism,double(ordual)aspectthe-ories(onacertainconception),andwhatIwillcall‘specialparticletheories’.Wetakeupeachinreverseorder,sincethisrepresentstheirhistoricaldevelop-ment.Iprimarilydiscussviewsonthemind–bodyproblemfromthebeginningofthemodernperiodtothepresent,thoughinfactallthebasicpositionsexcept32eliminativismwereanticipatedinantiquity.1.6.1Ontologicalanti-reductionismRejectingproposition(4),thenon-mentalcharacterofthebasicconstituentsofthings,hasbeenhistoricallythemostpopularposition.Thegenericview,accordingtowhichsomebasicconstituentsofthingsassuchhavementalproperties,maybecalled‘thementalparticletheory’.Thesemaybefurtherdividedintopureandmixedmentalparticletheories,accordingtowhetherthementalparticlesarethoughttohaveonlymental,ortohavementalandphysicalproperties,andthen,dividedagainaccordingtowhetheralloronlysomethingshavementalproperties(universalvs.restricted).Themostprominent,andhistoricallyimportant,viewofthissortissubstance33dualism,whichtracesbacktotheancientviewofthesoulasasimplesubstance.Substancedualismholdsthattherearebothmaterialsubstancesandmentalsubstances:theformerhaveonlyphysicalproperties,andnonemental,thelatteronlymentalproperties,andnonephysical.Thisisarestrictedpurementalparticletheory.Descartes(1985[1641])isthemostprominentoftheearlymoderndefendersofdualism.Theappealofdualismliesinpartinitsabilitytofindaplaceforirreduciblementalpropertiesinaworldthatseemslargelytobeexplainableasamechanicalsystemreducibletopartswhichthemselvesareexhaustivelycharac-terizedintermsoftheirprimaryqualities.Descarteswroteatthebeginningofthescientificrevolution,andwashimselfamajorproponentofthenew‘mechanicalphilosophy’,whosefundamentalassumptionsprovidethoseformodernphysics.13\nKirkLudwigDualismwasDescartes’sanswertotheproblemthemechanicalphilosophypresentsforfindingaplaceformindinthenaturalworld.Descarteshashadsuchanenormousinfluenceonthedevelopmentofthewesterntraditioninphilosophythatitwillbeusefultoreviewbrieflyhisofficialargumentsfordualism.Thissetsthestageforsubsequentdiscussionsofthemind–bodyproblem.ToexplainDescartes’sarguments,however,wemustfirstgetcleareraboutthenotionofasubstance.Thisnotion,centraltophilosophical34discussionintheseventeenthandeighteenthcenturies,tracesbacktoAristotle’scharacterizationofitas“thatwhichisneithersaidofasubjectnorinasubject”(Categories(Cat)1b2–5;in1984:4).Thisistheconceptionofasubstanceasapropertybearer,somethingthatundergoesandpersiststhroughchange:“Asubstance...numericallyoneandthesame,isabletoreceivecontraries...paleatonetimeanddarkatanother”(Cat4a19–21;in1984:7).Thisgaveriseinmedievalphilosophy(inscholasticism,thetraditiontowhichtherecoveryofAristotle’sworksgaverise)totheviewofsubstancesasindependentexistents,becauseofthecontrastwithproperties,whichwerethoughttoexistonlyinasubject,notindependently.Descartesgivestwocharacterizationsofsubstance.Oneisasthatwhichisabsolutelyindependentofeverythingelse.Thisgeneralizesthescholasticnotion.Descartesheldthat,onthisconception,Godistheonlysubstance,sinceeverythingdependsonGodforitsexistence.ButDescartesadmitssubstancesaspropertybearersinasubsidiarysense,andallowstwofunda-mentallydifferentkindsinadditiontoGod:thinkingandcorporealsubstances(Princ.1644,I.51–2;in1985,vol.I:210).HenceforthIrestrictattentiontothelattersort.AcentralfeatureofDescartes’stheoryofsubstancekindsisthateachdifferentsubstancekindhasaprincipalindividuatingattribute,ofwhicheveryotherpropertyofasubstanceofthekindisamodification:extension,forcorpor-ealsubstances,andthought,forthinkingsubstances(Princ.1644,I.53–4;in1985,vol.I:210–11).Thisfeatureofthetheory,oftenoverlookedinintroduct-orydiscussions,isessentialforacorrectunderstandingoftheforceofDescartes’sargumentsforsubstancedualism.Thedoctrinethateachsubstancehasaprincipalattributeforcestheindividuatingandessentialpropertyofasubstancekindtobeafundamentalwayofbeingsome-thing,oracategoricalproperty.Acategoricalpropertyisadeterminablebutnotadeterminate.Adeterminableisapropertyanobjectcanhaveindifferentways,andmusthaveinsomeparticularway,as,e.g.,beingcolored.Somethingcanbecoloredbybeingblue,orgreen,orred,andsoon,andifcoloredmustbecoloredinsomedeterminateway(hencetheterminology,‘determinable’,‘determinate’).ExtensionandthoughtDescartesconceivedasdeterminables,andtheyarenot35themselvesapparentlydeterminatesofanyotherdeterminableproperty.Withthistheoryinplace,thereisaneasyargumenttomind–bodydualism.Iftherearetwomostgeneralwaysofbeing,andthingsthathavethem,itfollowsimmediatelythattherearetwokindsofsubstance.Descartesarguedthathehadaclearanddistinctconceptionofhimselfasathinkingthing,athingthatatleastcanexistindependentlyofhisbody,andlikewiseaclearanddistinctconception14\nTheMind–BodyProblem:AnOverviewofacorporealobjectasasolelyextendedthing,athingthatcanatleastexistwithoutthinking,and,moreover,thattheseconceptionsarecompleteandnotin36needofappealtoanymoregeneralconceptionofakind.Fromthis,itfollowsthatthinkingandextensionarecategoricalproperties.Fromthetheoryofsub-stances,itfollowsthatthinkingandextendedsubstancesarenecessarilydistinct.Theargumentisunquestionablyvalid:necessarily,ifitspremisesaretrue,soisitsconclusion.Whetherweshouldacceptitspremises(andsowhetheritissound,i.e.,hastruepremisesinadditiontobeingvalid)islessclear.Itsweakestpremiseistheassumptionthatdistinctkindsofsubstancemusthaveonlyonecategoricalattribute.ItisunclearwhyDescartesheldthis.Thethoughtthatsubstancesarepropertybearersprovidesinsufficientsupport.EvenSpinoza,whowasheavilyinfluencedbyDescartes,objectedthatpreciselybecausementalandcorporealpropertiesareconceptuallyindependent,therecanbenobarriertoonesubstancepossessingbothattributes(EthicsIP10Scholium;inSpinoza1994:90).And,asP.F.Strawson(1958)hasobserved,weroutinelyattributetotheverysamething,persons,bothmaterialandmentalproperties:Iwalk,andsleep,aswellasthinkandfeel.Descartesendorsedcausalinteractionismbetweenmentalandmaterialsubstancetoexplainwhyourlimbsmoveinaccordancewithwhatwewanttodo,andhowweareabletocorrectlyperceivethingsinourbodies’physicalsurroundings.Somephilosophers,includingmanyofDescartes’scontemporaries,haveobjectedthatwecannotconceiveofcausalinteractionbetweensuchfundamentallydiffer-entkindsofsubstanceasmindandbody,thelatterinspace,theformernot.(Thoughitishardtoseethisasaconceptualdifficulty;seeBedau1986.)Thisgivesrisetoaversionofepiphenomenalism,accordingtowhichthementalisnotcausallyrelevanttothephysical.Therejectionofcausalinteractionismtogetherwiththeobviouscorrelationsbetweenmentalandphysicaleventsgaverisetoparallelism,accordingtowhichmentalandphysicaleventsevolveindependentlybutinawaythatgivesrisetonon-causalcorrelations,asthehandsoftwoclocks,setindependentlyaminuteapart,mayappeartobecausallyinteractingbecauseof37thecorrelationsintheirpositions,thoughtheyarenot.ParallelismisusuallyexplainedbyreferencetoGod’sarrangingthingsoriginallysothatthementalandthephysicaldevelopinparallel(pre-establishedharmony),orthroughHisconstantinterventioninbringingaboutwhatevents,bothphysicalandmental,giverisetotheappearanceofinteraction(occasionalism).Barringareasontothinkthatapropertybearercannotpossessbothirreduciblymentalandphysicalproperties,atmostDescartes’sargumentsestablishthattherecouldbethingswhichhaveonlymentalproperties,aswellasthingswhichhaveonlyphysicalproperties,notthatthereareormustbe.Ifwecanestablishaprioriatmostthatdualismcouldbetrue,whetheritistrueistobedetermined,insofarasitcanbe,byempiricalinvestigation.Sofar,thereseemstobenoverygood38empiricalreasontosupposedualismistrue.Idealismisthehistoricalsuccessortodualism.Itisdualismwithoutmaterialsubstance.Thus,itisauniversal,purementalparticletheory.Theclassicalposition15\nKirkLudwigislaidoutinGeorgeBerkeley’sATreatiseConcerningthePrinciplesofHumanKnowledge(1710).Moresophisticatedmodernversionsarecalled‘phenomenal-39ism’.Idealismisoftenmotivatedbyaconcerntounderstandthepossibilityofknowledgeofobjectsofordinaryperception:forestsandmeadows,mountainsandrain,starsandwindowpanes.TheCartesianviewoftherelationofmindtoworldleavesitmysterioushowwecanhaveknowledgeofit:ifweknowinthefirstinstanceonlyourconsciousmentalstates,andwhateverwecanknowbyreasonalone,yetthementalandmaterialareconceptuallyindependent,itlooksasifwehavenoreasontobelievethatthereisamaterialworldcausingourconsciousexperiences.Berkeleysolvedtheproblembydenyingthatobjectsofperceptionwerematerial,andidentifyingtheminsteadwithcollectionsofideas(henceideal-ism).Morerecenttreatmentsidentifyordinaryobjectsofcommon-senseknowledgewithlogicalconstructionsoutofphenomenalstates.Berkeleydeniedalsothatwecouldevenmakesenseofmaterialsubstance.Leibniz(1714)likewiseheldthatthebasicconstituentsofthings,monads(unit,fromtheGreekmonos),wereasortofmind–thoughhedidnotholdthatallwereconscious–andthattalkofordinarythingswastobeunderstoodintermsofmonadsandtheirstates(asDavidArmstronghasputit,onLeibniz’sview,“materialobjectsarecoloniesofrudimentarysouls”(1968,p.5)).Kant(1781)issometimesalsointerpretedasaphenomenalist.Thisviewisnotnowwidelyembraced.Itseemstobepartofourconceptionoftheworldofwhichwethinkwehaveknowledgethatitisinde-pendentoftheexistenceofthinkingbeings,whoarecontingentplayersontheworldstage.Panpsychismholdsthateverythingisaprimarybearerofmentalproperties(notsimplybybeingrelatedtoaprimarybearer–asmychairhasthepropertyofbeingoccupiedbysomeonethinkingaboutthemind–bodyproblem).Panpsychismcomesinreductiveandnon-reductivevarieties.Itsrootcanbetracedbacktoantiquity(Annas1992:43–7).PanpsychistsarerepresentedamongtheRenaissancephilosophers,andamongprominentnineteenth-centuryphilosophers,including40Schopenhauer,W.K.Clifford,WilliamJames(atonetime),andC.S.Peirce.Panpsychismisassociatedoftenwith(whatseemstobe)arevisionarymetaphysics,withspecialmotivations,asinthecaseofidealism,whichisareductiveversionofpanpsychism.However,non-reductivepanpsychism,whichacceptsabasicmateri-alistontology,ismotivatedbythethoughtthatotherwiseitwouldbeinexplica-ble(aspeciesofmagic)thatcomplexobjectshavementalproperties.WilliamJames,inhismonumentalPrinciplesofPsychology(1890),laysoutthisargumentexplicitlyinchapterVI,“EvolutionaryPsychologydemandsaMind-dust.”Thomas41Nagel(1979a)hasmorerecentlyrevivedtheargument(seealsoMenzies1988).Panpsychismisauniversalmentalparticletheory,andmaybepureormixed.Thedoubleaspecttheoryshouldbethoughtofasafamilyoftheories,ratherthanasingledoctrine.Whatunifiesthefamilyistheiraffinityforbeingexpressedwiththesloganthatthementalandthephysicalaredifferentaspectsbywhichwecomprehendoneandthesamething,thoughthesloganmaybeunderstooddiffer-entlyondifferent“versions”ofthetheory.Spinoza’sdoctrineoftheparallelism16\nTheMind–BodyProblem:AnOverviewofthoughtandextensionistheoriginalofthedoubleaspecttheory,thoughhe42didnothimselfsodescribehisposition.Spinozaheldthattherewasasingle,infinite,eternal,andnecessarysubstance,whichhadeverypossiblecategoricalattribute,andsobothextensionandthought.Ordinarythingsweretobe(re)conceivedasmodes(modifications)oftheworldsubstance.Thinkingandextensionwererelatedinaccordancewiththeparallelismthesis:“Theorderandconnectionofideasisthesameastheorderandconnectionofthings”(Ethics,IIP7;in1994:119–20).AsSpinozafurtherexplainsitintheScholium:“thethinkingsubstanceandtheextendedsubstanceareoneandthesamesubstance,whichisnowcomprehendedunderthisattribute,nowunderthat.Soalsoamodeofextensionandtheideaofthatmodeareoneandthesamething,butexpressedintwoways”(ibid:119).Thisisnotanentirelypelluciddoctrine.WeunderstanditonlytotheextentthatweunderstandSpinoza’smetaphysics,itselfamatterofinterpretivedifficulty.Theideathatthementalandthephysicalaretwowaysofcomprehendingonething,however,cansurvivetherejectionofSpinoza’smeta-physics,andhasinspiredanumberofviewswhichappealtosimilarlanguage.Ifweallowamultitudeofsubstances,thedoubleaspecttheoryholdsthateveryobject,orsome,canbeviewedasmentalorphysical,dependingonhowwetakeit.InG.H.Lewes’simage(1877;repr.inVesey1964:155),tocomprehendathingasmentalorphysicalislikeseeingalineasconcaveorconvex:“Thecurvehasateverypointthiscontrastofconvexandconcave,andyetistheidenticallinethroughout.”Thedoubleaspecttheoryisnotcurrentlypopular.Partlythisisduetoitsunclarity.Itisintendedtobemorethantheclaimthatthereareobjectsthathavementalandphysicalproperties,neitherbeingconceptuallyreducibletotheother(thoughsometimesithasbeenusedinthisbroadersense),oreventhattherearesystematiccorrelationsbetweeneverythingphysicalandsomething43mental.Butthereseemstobenothingmoreingeneraltosayaboutwhatitcomesto,andwemustratherlooktoparticulartheoriestogiveitcontent.Itslackofpopularityispartlyduetofactorsindependentofthedetails,and,inparticular,tothedominanceofourcurrentscientificworldview,accordingtowhichtheworldoncecontainednothinkingthings,andhasevolvedtoitspresentstatebynaturallaw.Doubleaspecttheoriesmaybeeitheruniversalorrestricted,mixedmentalparticletheories.Somedoubleaspecttheoriesareversionsofpanpsychism,then,asinthecaseofSpinoza,sincehedoesmaintainthateverythinghasmentalproperties.Compatiblywiththeguidingidea,however,onemightalsomaintainthatsomeobjectshavetwoaspects,twowaysofcomprehendingthem,mental44andphysical,thoughnotalldo.Finally,thereiswhatIcallthespecialparticletheory,whichholdsthatsomebasicconstituentsofthings,whichareatleastspatiallylocated,havementalproperties,butnotall.Thiscountsasarestricted,mixedmentalparticletheory,countingspatiallocationasabroadlyphysicalproperty.SofarasIknow,thisisnotaviewthathasbeenrepresentedamongtraditionalresponsestothemind–45bodyproblem.17\nKirkLudwig1.6.2Conceptualanti-reductionismRejectingproposition(3)leadstoemergentism.Thereareinprincipletwovarieties,neutralemergentismandemergentmaterialism,accordingtowhetherbasiccon-stituentsareconceivedasphysicalorneitherphysicalnormental.Mostemergentistsarematerialists,andIconcentratethereforeonemergentmaterialism.Emergentmaterialistsholdthatthereareonlymaterialthings,butthatsomecomplexmaterialthings,thoughnosimpleonesconsideredindependentlyofcomplexesinwhichtheyparticipate,havementalproperties,andthatthosementalpropertiesarenotconceptuallyreducibletoanyofthephysicalpropertiesofthecomplexesthathavethem.Emergentismhistoricallywasaresponsetotherejectionofformsofdualismandidealisminfavorofamaterialistontology.Itisassociatedwiththeriseofsciencegenerallyinthenineteenthcentury,andthedevelopmentofthetheoryofevolutioninparticular.Itdispenseswiththeontological,butretainstheconceptualanti-reductionismofCartesiandualism.Latenineteenth-andearlytwentieth-centuryemergentistsincludedT.H.Huxley(“Darwin’sbulldog”;1901),SamuelAlexander(1920),C.LloydMorgan(1923),andC.D.Broad(1925).Theterm“emergent”waspressedintoservicebecausetheuniversewasthoughttohaveoncenotcontainedanyobjectsthathadanymentalproperties.Sinceallitsobjectsarematerialobjects,oncetheyhadnomentalproperties,butnowsomedo,andthosepropertiesarenotconceptuallyreducibletophysicalproper-ties,mentalpropertiesmustemergefrom,insomeway,certainorganizationsofmatter,thoughthiscannotbededucedfromacompletedescriptionoftheobjects46thathavementalpropertiesintermsoftheirphysicalproperties.Emergentiststakeseriouslytheevidencethatatleastsomeaspectsofthementalarenotinanysensephysicalphenomena.Thiswasthetraditionalview,andisundeniablyaninitiallyattractiveposition.Oncewehaveextricatedourselvesfromtheconfusionsthatleadtotheviewthattheremustbementalsubstancesdistinctfrommaterialsubstancestobearirreduciblementalproperties,theviewthatwearelatecomerstothephysicalworld–naturalobjectsthatarosebynaturalprocessesfrommaterialsthemselvesfallingwhollywithintherealmofmechanics–leadsnaturallytoemergentmaterialism.Varietiesofemergentismarisefromdifferentviewsabouttherelationbetweenfundamentalpropertiesandmentalproperties.Traditionalemergentmaterialistsheldthatthereweretype-typenomiccorrelationsbetweenphysicalandmentalstates.Thisistoholdthatforeverymentalpropertysomesentenceoftheform[C]obtainswiththeforceofnomologicalnecessity.Onemayholdthatmentalpropertiesmerelynomicallysuperveneonphysicalproperties,andthatthereare47notype-typecorrelations.Finally,onemightholdaversionofwhatiscalled‘anomalousmonism’.Anomalousmonismwasoriginallyproposedasathesisabouttherelationofmentalandphysicalevents(Davidson1980).Itholdsthat48everymentaleventistokenidenticalwithaphysicalevent,butthereareno18\nTheMind–BodyProblem:AnOverview49strictpsychophysicallaws,andsonostrictbridgelaws.Thisstillallowsloose,non-strict,nomicsupervenienceornomictypecorrelation.Astrongerversiondenieseventhatthereareloosenomicrelationsbetweenmentalandphysicaleventtypes.Theideacanbeadaptedtoobjectsastheviewthatthoughsomecomplexobjectshavementalproperties,therearenostrictnomiccorrelationsorsuperveniencerelationsbetweenphysicalandmentalproperties,or,inthestrongerversion,noneatall.50Emergentismisoften(nowadaysespecially)associatedwithepiphenomenalism.Epiphenomenalismholdsthatmentalpropertiesarenotcausallyrelevanttoany-thing(or,atleast,toanythingphysical).Amonglatenineteenth-andearlytwentieth-centuryemergentiststherewasdisagreementaboutthecausalefficacyofthemental.Some(e.g.MorganandBroad)heldthattherewerenotonlyemergentproperties,butalsoemergentlawsgoverningsystemsattheleveloftheemergentpropertieswhichcouldthenaffectthecourseofeventsatlowerlevels51(downwardcausation).Thisstreamintheemergentisttraditionhasnownearly52rundry(thoughseeSperry1986).Otherprominentemergentistssawthementalaswhollydependentonthephysical,andcausallyinert.Inafamousdis-cussion,T.H.Huxleyheldthatconsciousnesswas“thedirectfunctionofmater-ialchanges”(1874:141),butalsothatconsciousnesswasascompletelywithoutpowertoaffectthemovementsofourbodies“asthesteam-whistlewhichaccom-paniestheworkingofalocomotiveengineiswithoutinfluenceuponitsmachin-ery”(p.140).(SeealsoHodgson1870;G.J.Romanes1895.)Onthisview,mentalactivityisashadowcastbyneuralactivity,determinedbyit,butdetermin-ingnothinginturn:consciousmentalstatesare“nomologicaldanglers,”inFeigl’saptphrase(1958).Untilthesecondhalfofthetwentiethcentury,emergentistsbelievedthatthereweretype-typecorrelationsbetweenthestatesofourcentralnervoussystemsandmentalstatesthatheldasamatterofnaturallaw.Theselawswerenotpurelyphysical,butbridgelaws,sincetheirstatementinvolvedirreduciblybothmentalandphysicalpredicates.Epiphenomenalismismotivatedbythethoughtthattheuniversewouldproceedjustasithasphysicallyifweweresimplytosubtractfromitthebridgelaws:wedonotneedinprincipletorefertoanynon-physicaleventsorlawstoexplainanyphysicalevent.Justasthelocomotivewouldcontinueinitspathifweweretoremoveitswhistle,soourbodieswould53continueintheirtrajectoriesifweweretoremovetheirsouls.Theconjunctionoftheviewthattherearesuchtype-typenomiccorrelations,andtheviewthatthephysicalisaclosedsystem,isnomologicalreductionism.Obviously,thefurtherwemovefromnomictype-typecorrelations,thelessplausibleitbecomesthatwecanfindaplaceforthecausalefficacyofmentalproperties.Theperceivedthreatofepiphenomenalismhasbeenoneofthemotivationsforphysicalism.Itisanironythatsomepopularwaysoftryingtogroundphysicalismalsoraisedifficultiesforseeinghowmentalpropertiescouldbecausallyrelevanttowhattheyare54supposedtobe.19\nKirkLudwig1.6.3ConceptualreductionTorejectproposition(2)istoadoptconceptualreductionismformentalproperties.Weconsiderfirst,briefly,non-physicalistwaysofrejecting(2).Therearetwopossibilities:thatthementalisconceptuallyreducibleto,orsupervenesonsome-thingnon-physical.Whilethelatterpositionisanoption,ithasnotbeenoccu-pied.However,neutralmonism,theviewthatthementalandthephysicalmightbothbeunderstoodintermsofsomethingmorebasic,enjoyedabriefrunatthe55endofthenineteenthandinthefirsthalfofthetwentiethcentury.TheviewisassociatedwithWilliamJames(1904),whoarguedthat“pureexperience”istheprimalstuffoftheworldandmindsandobjectsweretobeconceivedofasdifferentsetsofexperiences,sothatthesameexperiencecouldbetakenwithonesetasathought,andwithanotherasacomponentofanobjectthoughtabout.Neutralmonism,asadvocatedbyJames,rejectstheviewthatthereisasubjectofexperience,andretainsonlywhatwastraditionallythoughtofasitsobject.AsJamesputit,“thosewhoclingtoitareclingingtoamereecho,thefaintrumorleftbehindbythedisappearing‘soul’upontheairofphilo-sophy”(pp.3–4).ErnestMach(1886)heldasimilarview,andBertrandRusselldevelopedaversionofneutralmonism,inspiredbyJames,inwhichsensibilia(or“sensations”asRussellputitinTheAnalysisofMind(1921)),introducedoriginallyasmind-independentobjectsofdirectawareness(1917),playedtheroleoftheneutralstuffoutofwhichmindsandphysicalobjectsweretobelogicallyconstructed(1921).Itmayseemasifthisviewshouldmoreproperlybedescribedasaversionofidealism,becausethetermsthatJames,Mach,andRussellusedtodescribetheneutralstuffareusuallyassociatedwithmentalphenomena.Buttheyheldthattheneutralstuffwasnotproperlythoughtofasmentalincharacter,butonlywhenitwasconsideredinacertainarrangement.Itmightthenseemreasonabletodescribeneutralmonismasadoubleaspecttheory,atleastinthesensethatittreatseachofthefundamentalthingsasathingthatcouldparticipateinaseriesofthingswhichconstitutedsomethingmental,aswellasinaseriesofthingswhichconstitutedsomethingphysical;thus,eachcouldbesaidtobeviewedunderaphysicaloramentalaspect.However,sincetalkofthoughtsandmaterialthingsisconceivedofastranslatableintotalkneithermentalnorphysical,neitherthementalnorthephysicalhasafundamentalstatusintheontologyofneutral56monism.Rather,bothbeartherelationtotheneutralstuffthatordinaryobjectsdotophenomenalexperienceaccordingtoidealisttheories.Justasidealistthe-oriesdonotcountenancegenuinematerialsubstance,neutralmonismdoesnotcountenancegenuinementalorphysicalsubstancesinitsfundamentalontology,thoughitgivesanaccountoftalkofeachsort.Neutralmonismhassometheoreticalvirtues.Itavoidsthedifficultiesassoci-atedwithtryingtoreduceeitherthementaltothephysicalorviceversa,and,ifsuccessful,providesafundamental,unifiedaccountofthingsofallkindsinterms20\nTheMind–BodyProblem:AnOverviewofafundamentalkind,thedreamofidealistsandphysicalistsalike.Despitethis,itisnotapopularview.Itattractsneitherthosewhothinkthementalisabasicfeatureofreality,northosewhodreamofthedesertlandscapeofphysics.More-over,itisdifficulttodeveloptheaccountindetail,anddifficulttounderstandthenatureoftheneutralstuffwhichitreliesupon.Weturnnowtophysicalistrejectionsofproposition(2).Thefirsttwentieth-centuryphysicalistpositiontogainpopularitywaslogicalbehaviorism,whichwasspurredoninpartbytheverificationismofthelogicalpositivistsbeforetheSecondWorldWar,theviewthatthemeaningofasentencewastobesoughtintheempiricalconditionsforconfirmingordisconfirmingit(a57viewwithrootsinclassicalBritishempiricism).Logicalbehaviorismhasastrongerandaweakerform.ThestrongformIwillcall‘translationalbehaviorism’,andtheweakerform‘criterialbehaviorism’.Translationalbehaviorismholdsthateverypsychologicalstatementcanbetranslatedintoastatementaboutactualandpotentialbehaviorofbodies.Criterialbehaviorismholds,incontrast,merelythattherearebehavioralanalyticallysufficientconditionsfortheapplicationofmentalpredicates.Logicalbehaviorismhaslongfallenoutoffashion.Thisisexplainedinpartbythefallfromfavorofverificationism,whichprovidedittheoreticalsupport,butalsobythefactthatnotonlywerenosatisfactorytranslationschemesadvanced,buttherearereasonstothinknonecouldbeforthcominginprinciple.Aparticu-larlytroublingproblemwasthatwhatbehavioralmanifestationswemayexpectfromsomeonewithacertainmentalstatedependsonwhatothermentalstateshehas.Consequently,therecanbenopiecemealtranslationofpsychologicalclaimsintobehavioralterms.Inaddition,behaviorismseemsincompatiblewithourconceptionofmentalstatesas(possible)causesofbehavior.Fortoreducetalkofmentalstatestotalkofbehavioristotreatitasmerelyamorecompendiousway58ofdescribingbehavior.Behavior,though,cannotcauseitself.Thetwoprincipalphysicalistresponsestothedefectsofbehaviorismwereanalyticfunctionalismandthepsychophysicalidentitytheory.Thoughthepsychophysicalidentitytheorycametoprominencebeforeanalyticfunctionalism,itwillbeusefultodiscussfunctionalismfirst,sinceitisthenaturalsuccessortologicalbehaviorism,andthiswillputusinapositiontousefullyclarifythepsychophysicalidentitytheory,whichinsomeearlyversionssufferedfromanumberofconfusionsandconflictingtendencies.Analyticfunctionalismholdsthatmentalstatesareconceptuallyreducibletofunctionalstates.Functionalstatesareheldtoconceptuallysupervene,inthe59sensedefinedinsection1.4,onphysicalstates.Theidentificationofmentalwithfunctionalstatesthenleadstophysicalismwithoutconceptualreductionofthementaltothephysicalperse.Afunctionalstate,intherelevantsense,isastateofanobjectdefinedintermsofitsrelationstoinputtoasystem,otherfunctionalstatesofthesystem,andoutputfromthesystem.Someofthelogicalbehaviorists,e.g.,GilbertRyleinTheConceptofMind(1949),canbeseentohavebeenmovingtowardsomethinglikethis(functionalismmaythereforebesaidtobethe21\nKirkLudwigeclosionofbehaviorism).Functionalismwasinspired,atleastinpart,bytherise60ofcomputertechnologyaftertheSecondWorldWar.Itsearliestforminthetwentiethcentury,machinetablefunctionalism,introducedbyHilaryPutnam(1967),wasdirectlyinspiredbytheoreticalworkonfinitestatemachines,which61iswhata(finitestate)computeris.Amachinetabledescribesasystemintermsofalistofexhaustiveandmutuallyexclusiveinputs,alistofpossiblestates,alistofoutputs,and,foreachpossiblestate,whatstateitmovestoandwhatoutputisproducedgiventhatitreceivesagiveninput.Theoperationofanycomputerrunningaprogramcanbedescribedexhaustivelyintermsofamachinetable.Forprogrammablecomputers,theprogramdetermineswhatmachinetableitinstan-tiates(relativetoadivisionofasystemintostatesofparticularinteresttous).Putnamgeneralizedthenotionofafinitestateautomaton(asystemdescribableusingafinitestatemachinetablewithdeterministicstatetransitions)toaprobabilisticfinitestateautomaton,inwhichtransitionsareprobabilistic.Thegeneralformoftheproposalisthatasystemisinacertainmentalstateiffithasanappropriatemachinetabledescriptionandappropriateinputsorappropriatestates.Putnamtreatedhisproposalasanempiricalhypothesis.Thisistypically62called‘psychofunctionalism’,followingBlock(1978).Itisnonethelessoneoftheprincipalinspirationsforanalyticfunctionalism,andiseasilyreconstruedasathesisaboutourconceptsofmentalstates.Theoreticalor,sometimes,causalrolefunctionalismisavariantonthetheme.Onthisview,westartwithatheorythatembedspsychologicalterms.Theconceptsexpressedbythesetermsaretakentobeconceptsofstatesthatarecharacterizedexhaustivelybytheirrelationstoother63statesandinputsandoutputsasspecifiedabstractlyinthetheory.Functionalismisattractive.Itaccommodatesathoughtthatmotivatedbehaviorism,namely,thatourmentalstatesareintimatelytiedupwithunder-standingofbehavior,butitdoessoinawaythatdistinguishesthemfrom,andtreatsthemascausesof,behavior.Moreover,functionalismallowsforthepos-sibilityofimmaterialthinkingbeings,sinceasystem’shavingacertainfunctionalorganizationdoesnotdependonwhatitismadeof,butratheronitscausalpowerswithrespecttoinputsandoutputs.Ithasmerelytosustaintherightorganizationmediatinginputsandoutputs.Functionalstatesaremultiplyrealiz-able.Thisaccommodatesoneofthethoughtexperimentsthatmotivatestheassumptionoftheconceptualindependenceofthementalandthephysical.Itfindsaplaceforthementalinthenaturalworldthatexhibitsitasgroundedinthephysical,inthesensethatitexhibitsthementalasconceptuallysuperveningonthephysical,withoutinsistingonaconceptualreductiontophysicalproperties.Ittherebyallowsthatthelanguageofpsychologyisdistinctfromthatofphysics,whileallowingthattherealizationofpsychologicalstatesrequiresnothingmorethanobjectshavingphysicalpropertiesgovernedbyphysicallaws.Themultiplerealizabilityoffunctionalstatesalso(primafacie)protectsfunctionalismfromachargeleveledagainstthepsychophysicalidentitytheory,namely,thatitwouldbeimplausible,andchauvinistic,toinsistthatonlythosephysicallylikeuscan64havementalstates.22\nTheMind–BodyProblem:AnOverviewAnalyticfunctionalismhascomeinforconsiderablecriticism,butremainspopu-lar,especiallyoutsidephilosophyinfieldscontributingtothenewdisciplineofcognitivescience.Afirstobjectiontofunctionalismisthatnoonehascomeupwithasuccessfulconceptualreductionofmentalconceptstofunctionalconcepts.Itmightbesaidthatthiscouldequallywellbeasignofthecomplexityofthesefunctionalconcepts.Asecondobjectiontofunctionalismisbasedontheprimafacieintelligibilityofsystemswhicharefunctionallyidenticaltousbutwhichhavenomentalstates.AnexampleisprovidedbyathoughtexperimentofNedBlock’s65(1978).Imaginearobotbodyactuatedbyaprograminstantiatingamachinetableforsomeperson.ImaginefurtherthatweinstantiatetheprogrambyprovidingeachmemberofthepopulationofChinawithatwo-wayradiowithadisplaythatshowsthecurrentinputtotheroboticsystemandanindicatorofwhetherthesystemisinhisstate.Eachpersonpressesabuttonontheradioappropriatefortheinputwhenhisstateisactive.Signalsarerelayedtothebodyforappropriateaction.SupposethattheChinesegetsogoodatthisthatourrobotandaccesso-riesconstituteasystemfunctionallyidenticaltoouroriginal.Doesthissystemnowconstituteanintelligent,consciousbeing?Mostpeople,firstconfrontedwiththethoughtexperiment,denythatwehavecreatedanewperson(whowilldie66whentheexerciseisterminated).AnotherimportantobjectionisalsoduetoNedBlock(1978).Functionalistsmustdecidehowtospecifyinputsandoutputstothesystem.Thispresentsthemwithadilemma.Ifwespecifytheinputsandoutputsphysicallyusingourselvesasmodels,itisnotdifficulttodescribesomesystemthatcouldhaveamindthatisincapableofcausingthoseoutputs,butcausesothersinstead(e.g.,wedonotwanttoruleout,apriori,intelligentjellyfish,orbeingswhoseinputsandoutputsarevariousportionsoftheelectromagneticspectrum,andsoon).Further,itisdifficulttoseehowwecouldputapriorilimitsonthephysicalcharacterofinputsandoutputs.However,iftheinputsandoutputsarespecifiedbarelyasdistinct,thenitisnotunlikelythatwecanfindmindsjustabouteverywhere,foritisplausiblethatmostcomplexsystemswilladmitofsomedivisionintostatesandinputsandoutputsthatwillinstantiatesomemachinetablesaidtobesufficientforhavingamind(e.g.,theworldeconomy).Italsohasbeenobjectedthatitiseasytoimaginefunctionalduplicateswhodifferinthequalitiesoftheirexperiences.Awell-knownthoughtexperimentdesignedtoshowthisisthatoftheinvertedspectrum.Weimaginetwoindi-vidualsfunctionallyindistinguishable,andthereforebehaviorallyindistinguishable,butimaginethattheirexperiencesofthecolorsofobjectsintheirenvironmentsareinvertedwithrespecttooneanother.Whereoneexperiencesaredobject,e.g.,theotherexperiencesagreenobject.Theybothutterthesamesentenceindescribingit,buteachseesitdifferently.Ifthisisconceivable,thentheircolorexperiencesarenotconceptuallyreducibletotheirfunctionalorganization,and,67hence,functionalismisfalsewithrespecttothesephenomenalqualities.Anotherdifficultyisthatitisunclearthatfunctionalstatescanbecausallyrelevanttotherightsortsofbehavior.Functionalismaccommodatesmentalstates23\nKirkLudwig68ascausesofbehaviorbydefinition.Butthismaysecurethecausalconnectioninthewrongway.Forastatedefinedintermsofitseffectsinvariouscircumstancescannotbethetypeinvirtueofwhichthoseeffectscomeabout.Causalrelationsbetweeneventsorstatesareunderlainbycontingentcausallawsconnectingtypes69underwhichtheyfall.Onetypeiscausallyrelevanttoanothertype(incertaincircumstances)ifftheyareconnectedbyacausallaw(inthecircumstances).However,therelationbetweenafunctionalstateandtheoutput(type)intermsofwhichitispartiallydefinedisnotcontingent.Thus,thestatetypeandoutputtypecannotfeatureappropriatelyinacontingentcausallaw.Therefore,functionalstatetypesarenotcausallyrelevanttooutputintermsofwhichtheyarede-70fined.Ifthisreasoningiscorrect,analyticfunctionalismentailsepiphenomenalismwithrespecttotheseoutputs.Anadvantageoffunctionalismoverbehaviorismwassupposedtobethatitmakesmentalstatescausesofbehavior.Thetroubleisthatitdoessoinawaythatundercutsthepossibilityofthosestatesbeingcausallyrelevanttowhatweexpectthemtobe.Worse,itseemsquiteplausiblethatwedoconceiveofourmentalstatesascausallyrelevanttothebehaviorthatwewouldusetodefinementalstatesonafunctionalanalysis.Ourbeliefsaboutthecausalrelevanceofmentalstatestobehaviormaybefalse.Itiscontingentonwhatcausallawshold.Butiftheyarenotnecessarilyfalse,thenfunctionalismcannotbetrue,sinceitprecludesthe71possibilityofourmentalstatesbeingcausallyrelevanttoourbehavior.Letusnowturntothepsychophysicalidentitytheory.Thisistheviewthatmentalpropertiesarephysicalproperties.IstartwithwhatIbelieveisthemostplausibleformofthepsychophysicalidentitytheory,whichisbasedonanapproachadvocatedbyDavidLewis(1966,1972).Theapproachmakesuseoffunctionalistdescriptionsofstatesextractedfroma“folktheory”ofpsychologytoidentifymentalstateswithphysicalstates.Analyticfunctionalismholdsthatpsychologicalconceptsandpropertiesarefunctionalconceptsandproperties.Thisshouldbedistinguishedfromtheviewthatpsychologicalpropertiesarepickedoutbyfunctionaldescriptions.Thisviewdoesnotreducementalpropertiestofunctionalproperties.Rather,ittreatsmentaltermsastheoreticalterms.Theoreticaltermsaretreatedaspickingoutpropertiesintheworld(andsoasexpressingwhateverconceptsareofthoseproperties)thatactuallyplaytherolethetheoryaccordstheminthesystemstowhichitisapplied.Werepresentourpsychologicaltheoryasasinglesentence,‘T(M1,M2,...,Mn)’,where‘M1’andsoonrepresentpsychologicaltermsreferringtoproperties.Thenwereplaceeachsuchtermwithacorrespondingvariable,‘x1’,‘x2’,andsoon,andprefacethewholewithaquantifierforeach,‘thereisauniquex1suchthat’(symbolizedas‘(∃!x1)’),etc.,toyield,‘(∃!x1)(∃!x2)...(∃!xn)T(x1,x2,...,xn)’.Theproperty“M1”picksoutcanbecharacterizedasfollows,whereweleaveoutthequantifierinfrontof‘T(...)’associatedwith‘x1’:M1istheuniquepropertyx1suchthat(∃!x2)...(∃!xn)T(x1,x2,...,xn)24\nTheMind–BodyProblem:AnOverviewInapplicationtohumanbeings,ontheassumptionthatthetheoreticaldescrip-tionofthispropertyissatisfiedbyaphysicalpropertyofourbodiesorcentralnervoussystems,itfollowsthatM1isthatphysicalproperty.Thus,wearriveatapsychophysicalidentitytheory.Givenhowwehavecharacterizedtherelationbetweenconcepts,predicates,states,andproperties,ifweidentifyamentalstateorpropertywithaphysicalstateorproperty,itfollowsthatthecorrespondingmentalconceptisaphysicalconcept.Therefore,theviewthatmentalpropertiesarepickedoutbyfunctionaldescriptionswillleadtotheconclusionthatmentalconceptsareconceptuallyreducibletophysicalconcepts,ifthosedescriptionspickoutphysicalstatesor72properties.Thisisnot,however,somethingwecouldknowapriori.Itcouldonlyemergeafterempiricalinvestigation.Foronthisview,theconceptsexpressedbyourtheoreticaltermsarehostagetothenatureofthephenomenatowhichweapplythem.Westartonlywithdescriptionsoftheproperties,andso,ineffect,onlywithdescriptionsoftheconceptsofthem.Wecanreasonaprioriusingtheconceptsonlyafterwehavediscoveredthemaposteriori.Thepsychophysicalidentitytheoryhastheadvantageoverfunctionalismandemergentisminsecuringthecausalrelevanceofmentalproperties.Noonedoubtsthatourphysicalstatesarecausallyrelevanttoourmovements.Identifyingmen-talstateswithphysicalstates,thepsychophysicalidentitytheorymakestheircausalrelevanceunproblematic.Somephilosophershavearguedthatsinceonlyidentifyingmentalwithphysicalstateswillsecuretheircausalefficacy,andmentalstatesarecausallyefficacious,wearejustifiedinidentifyingthem(Papineau1998).Thiscomesatacost,though.Onthisview,priortoempiricalinvestigationitisopenthattherearenomentalpropertiesatall,nopropertiesthatanswertothetheoreticaldescriptionswehaveofthem.Thisshowsthatthisviewhasincom-monwitheliminativismtheassumptionthatwedonotknowdirectlythatany-thinghasthepropertieswesupposetobepickedoutbyourpsychologicalterms.Aviewlikethisentailseliminativismwhencombinedwiththeclaimthatnophysical(oranyother)statesplaytherequiredroles.Totheextenttowhichwefinditimplausible,perhapsevenunintelligible,thatwecoulddiscoverwedon’thaveanymentalstates,weshouldfindequallyimplausibleorunintelligiblethe73argumentforthepsychophysicalidentitytheoryjustreviewed.Thepsychophysicalidentitytheory(alsocalled“centralstatematerialism”),likefunctionalism,hasantecedentsthatstretchbacktotheancientworld.Inthetwentiethcentury,itwasinfluentiallyadvocatedaftertheSecondWorldWarby74UllinPlace(1956),HerbertFeigl(1958),andJ.J.C.Smart(1959).PlaceandSmartheldthatsensationsweretobetheoreticallyidentifiedwithbrainprocesses,inthesamewaythatlightningwasidentifiedwithacertainsortofelectricaldischarge75(thiscanbegeneralizedstraightforwardlytostates;seeArmstrong1968).Theythoughtofthisasacontingentidentity,becauseitwasempiricallydiscovered.Thepositionisalsosometimescalled‘thetopicneutralapproach’,becauseSmartinparticulararguedthatinorderthatwenothaveirreduciblementalproperties,andyetmakesenseofthepossibilityofcontingentidentity,thedescriptionsbywhich25\nKirkLudwigwepickoutmentalprocesses(moregenerallymentalstates),whicharetobeempiricallyidentifiedwithphysicalones,mustleaveitopenwhethertheyarephysicalornot.Thispositioncameintoconsiderablecriticismfortheclaimthatidentitiescouldbecontingent(seeKripke1980:98–100,144–55).Ifwearespeakingaboutstrictidentityofthings–inthepresentcase,properties–thereisnoroomforcontingency,sinceidentityholdsofnecessitybetweeneverythinganditself,andbetweennodistinctthings.TheviewIhavepresentedbasedonLewis’sapproachisadescendantoftheseearlypsychophysicalidentitytheories.Itretainstheviewthatmentalpropertiesarephysicalproperties(ontheassumptionthatuniquephysicalpropertiesplaytherightroles).Butitrejectstheviewthatthisiscontingent(giventhatinfacttherearephysicalpropertiesplayingtherightroles).Seeingtheoreticaltermsasintroducedtotrackpropertiesthataretoplaycertainroleshelpsustoseehowthediscoveryofidentitiescanbeempiricalalthoughtheidentitiesarenecessary.Italsogivesprecisecontenttotheideathatthedescriptionsthatpickoutmentalstatesaretopicneutral,sincetheyaretobegivenbythestructureinducedbyourfolktheoryofpsychology.Atthispoint,anoteonmetaphysicalnecessityisinorder.Thismodalityisofteninvokedincontemporarydiscussionsofthemind–bodyproblem.Itissaidtobedistinctbothfromnomologicalandconceptualnecessity,strongerthantheformer,andweakerthanthelatter.Howdiditcometobeintroduced?Apara-digmofmetaphysicalnecessityissupposedtobethesortthatresultsfromthe-oreticalidentificationsinvolvingnaturalkinds,liketheidentificationofgoldwiththatelementwithatomicnumber79.Itisnotcontingentorjustamatterofnaturallaw,butnecessarythatgoldistheelementwithatomicnumber79,sincenothingthatdidnothaveatomicnumber79wouldcountasgoldeveninaworldwithdifferentnaturallaws.Still,itwasanempiricaldiscovery,andnotsomethingwecouldhaveknownpurelyapriori.Butsinceconceptualtruthsareknowableapriori,itmustbethatmetaphysicalnecessityisdistinctfromconcep-tualnecessity–orsotheargumentgoes.Theperceivedutilityofmetaphysicalnecessityisthatitprovidesawaytoargueforconnectionsbetweenthementalandthephysicalstrongerthannomologicalconnections,indeed,identities,whichatthesametimeisimmunetorefutationbythoughtexperimentsthatseemtoshowmentalandphysicalphenomenaareindependent.Sincemetaphysicalnecessityissupposednottobegovernedbywhatisconceptuallypossible,andsuchthoughtexperimentsare,theyfailtobear76ontheclaim.AsIsaidearlier,inmyviewnophilosopherhassucceededinexpressingacon-ceptby‘metaphysicalnecessity’thatanswerstothisargument.Thefirstthingthatshouldmakeussuspiciousabout“metaphysicalnecessity”isthatwedonothaveanyaccountofwhatgroundsclaimssupposedlyaboutit.Barringthis,itisdubiousthatwehaveanypreciseideaofwhatissupposedtobeexpressedherebytheterm‘metaphysical’.Thesecondthingthatshouldmakeussuspiciousisthatthereisavailableastraightforwardexplanationofthefactswhichmotivateintroducingmetaphysicalnecessitythatrequiresnomysteriousnewsortofnecessity.26\nTheMind–BodyProblem:AnOverviewOurreadingofLewis’saccountoftheoreticalidentificationsprovidesthekey.Onthataccount,weassociatewitheachtheoreticaltermadescriptionofthepropertythatitpicksout(thepropertyPwhichplayssuchandsucharoleinsuchandsuchsystems).Itisamatterforempiricalinvestigationwhatpropertyactuallysatisfiesthedescription(asitisindeterminingwhichindividualisthemayorofNewYork).However,theconceptatermexpressesis,aswehaveseen,whatdeterminesthepropertyitpicksout:theyareamatchedset.Thus,todiscoverwhatpropertyatheoreticaltermpicksoutbydiscoveringempiricallywhatsatis-fiestheassociateddescriptionislikewisetodiscoverempiricallywhatconceptthetermexpresses.Priortothat,wehadadescriptionofaconcept,butitwasnotgiventousdirectly.Thus,whenwediscoverthat‘isgold’picksouttheelementwithatomicnumber79,wediscoverwhatconceptitexpresses.Priortothis,wedidnotknowwhatconceptitexpressed.Onceweknow,weareinapositiontoseethat‘Goldisthatelementwithatomicnumber79’expressesaconceptualtruth,whichisknowableapriori.Whatwasnotknowableaprioriwasnotthatgoldisthatelementwithatomicnumber79,butthat‘gold’expressedtheconceptoftheelementwithatomicnumber79.Wecompetentlyusesuchnaturalkindtermspriortodiscoveringwhatconceptstheyexpress.Thisisexplainedbythefactthatwetreatsuchtermsastrackingpropertiesthatexplaineasilyiden-tifiablefeaturesofthingsweinpracticeapplythemto.Weapplythetermsinaccordancewiththosefeatures.Themistakeintheoriginalargumentwastoconfusecompetenceinapplyingnaturalkindtermswithgraspoftheconceptexpressed:giventhatwedonotknowwhatpropertyispickedout,welikewisedonotknowwhatconceptisexpressed.Whatweknowisjustwhatworkthepropertyissupposedtodo,whichenablesustodevelopanapplicationpracticewiththetermthatistopickitout.Thus,theintroductionof‘metaphysicalnecessity’isgratuitous.Wehavenoreasontosupposeanythingcorrespondstoit,andnoideaofwhatitwouldbeifitdid.Consequently,wecannotlooktometaphysicalnecessityfornewavenues77forthesolutionofthemind–bodyproblem.Beforeweleavethetopicofreductionism,itisimportanttoconsiderahybridviewthatcombinesfunctionalismandexternalismaboutthoughtcontent.Externalistaccountsofmentalstatesemphasizetheimportanceofourrelationstothingsinourenvironmentsinconceptuallyindividuatingthem.Atthesametimethatdifficultiesweremountingforfunctionalism,independentlysomeinfluentialargumentswereadvancedwhichsuggestedthatcontentpropertieswererelational78properties.Accordingtotheseaccounts,whatthoughtswehavedependsonwhatactualandpotentialcausalrelationswebeartothingsinourenvironments.(Relationallyindividuatedstatesareoftencalled‘widestates’intheliterature,andnon-relationallyindividuatedstates‘narrowstates’.)Themostimportantdivisionamongexternalistviewsisthatbetweenphysicalandsocialexternalism.Physicalexternalismholdsthatthoughtcontentsareindividuated(inpart)byrelationstoourphysicalenvironments.Socialexternalismholdsthatthoughtcontentsareindividuated(inpart)byhowothersinourlinguisticcommunities27\nKirkLudwig79usethewordsweintendtouseastheydo.Areductionistexternalistaccountofthoughtcontentwilltypicallyholdthatourconceptsatleastofcontentfulmentalstatescanbereducedtofunctionalandcausalconcepts,whereweincludesystem-aticcausalrelationstoexternalthingsinfixingthecontentsofthoughts.Externalisttheoriestoohavecomeinforconsiderablecriticism.Twoareworth80mentioningbecausetheyareconnectedwiththemesalreadytouchedon.Thefirstistheobjectionthatifexternalismweretrue,wewouldnotbeabletoknowthecontentsofourownthoughtswithoutempiricalinvestigation,butsincewemustinordertoundertakeempiricalinvestigationsinthefirstplace,externalism81entailsunacceptablythatwecanneverknowthecontentsofourownthoughts.Thesecondisconnectedwithadifficultyalreadynotedforfunctionalism.Itisthattreatingcontentpropertiesasindividuatedinpartintermsofrelationalpropertiesthreatenstomakethemunsuitableforexplainingourbehavior(describedphysic-ally).Theproblemisnotthatrelationalpropertiescannotbecausallyrelevanttoanything.Thereareprimafaciecounterexamplestothis.Thatsomethingisaplanet,forexample,maybecitedinexplainingwhyIcometobelievethatitis.Butthedifficultyforexternalismonlyrequiresthatthekindofrelationalproper-tiesthatcontentpropertieswouldturnouttobecouldnotbecausallyrelevanttoourbehavior.Forexternalisttheoriesexploitthepossibilityofbehavior(describedphysically)remainingthesamebecauseone’snon-relationalphysicalstatesremainthesamewhileone’sthoughtcontentsvary.Thisappearstoshowthattherela-tionalstatesare“screenedoff”fromtherelevanteffecttypesbythenon-relationalphysicalstates,whicharesufficienttoaccountforthebehaviorandareindepend-82entlynecessary.Theconceptionofour(atleastconscious)mentalstatesasofasortwhichare(a)non-inferentiallyknowablebytheirpossessor(ourconceptsofwhicharethereforenottheoreticalconcepts),thoughbynooneelse,and(b)as(possibly)causallyrelevanttoothersortsofthings(othermentaleventsandstatesaswellasnon-mentaleventsandstates)maybecalledthecoreoftheCartesianconceptionofthemind.Thedifficultieswehavebeenreviewingforreductionistproposalsaboutthementalareconnectedwiththesefeatures.Nophysicalstatesseemcapableofpossessingboth.Thefirstfeaturestandsinthewayoftheplausibilityofthepsychophysicalidentitytheory,and,arguably,ofexternalismaboutthoughtcontent.Thesecondseemstoprecludeconceptualreductiontostatescharacter-izedintermsoftheircausalrelationstootherthings,or,again,intermsoftheirrelationstothingsintheenvironment.1.6.4IrrealismFinally,weturntoeliminativism.Eliminativistsseekabsolutionthroughdenial.Accordingtoeliminativism,nothinghasmentalproperties.Prominentpropon-entsofthispositionarePaulChurchland(1981)andStephenStich(1983),83whoarguethatourmentalconceptsareempty.Theyareconceptsdeployedin28\nTheMind–BodyProblem:AnOverviewapre-scientificor“folk”theoryofbehavior,whichareripeforreplacementbyamoresophisticatedtheorydeployingdifferentcategories,whichanswerbettertoourexplanatoryinterests.Folkpsychologygoesthewayoftheoriesofdiseasethatappealtodemonicspirits.Thepsychologicalentitiesofourcommon-sensecon-ceptualschemetooarecreaturesofdarkness.Wemustnowmarchforwardintoabrighterfuture,outfromundertheshadowcastbysuperstitionsinculcatedinthechildhoodofcivilization,shrivenofthesinofbeliefinthemind.Eliminativismremains,notsurprisingly,aminorityposition.Ithassomeadvant-ages–asKarlPopperhassaid,“thedifficultbody–mindproblemsimplydisap-pears,whichisnodoubtveryconvenient:itsavesusthetroubleofsolvingit”(1994:8).Butitishardtocredit.Itmustrejecttheviewthatknowledgeofourownconsciousmentalstatesisepistemicallypriortoknowledgeofotherthings,whichseemstobeinconflictwithaverynaturalaccountofhowwecometoknowthingsabouttheworldaroundusthroughperceptualexperience.Therearealsocertaindifficultiesinvolvedinthinkingaboutourpositioninputtingforwardthetheory,andinaccountingforhowwecouldjustifyit.Forsurelyifsomeonemaintainsthatthetheoryistrue,thereisatleastonepersonwhobelievessome-thing,namely,thateliminativismistrue,inwhichcase,eliminativismisfalse.Thedifficultyisthatwehavenovocabularyfordescribingtheacceptance,rejection,andsupportoftheoriesthatdoesnotpresupposethattheoreticianshavementalstates.Eliminativistsmaintainthisismerelyapragmaticdifficulty,butitisnotonethattheyhaveovercome.1.7ConclusionThisconcludesoursurveyofthemind–bodyproblemandtheprincipalresponsestoit.Asummaryofthepositionswehaveconsideredisgiveninfigure1.1.Twobasicpositionsmarkthecontinentaldivideofthemind–bodyproblem.Allthepositionswehaveexaminedareexpressionsofoneortheotherofthem.Oneacceptsthementalasabasicfeatureofreality,notexplicableintermsofotherfeatures.Itsbasiccharacteristicisthatitacceptspropositions(1)and(2),realismandconceptualautonomy.Theotherinsiststhattheappearancethatthementalisabasicfeatureofrealitymustbeanillusion,andthatweandallourpropertiescanbeunderstoodexhaustivelyultimatelyintermsthatmakeintelli-gibletousatthesametimetheclearlynon-mentalphenomenaoftheworld.Itsbasiccharacteristicisthatitacceptspropositions(3)and(4),constituentexplanat-orysufficiencyandconstituentnon-mentalism.Thesecondview,constrainedbytheassumptionthatthebasicconstituentsofthingsarephysical(constituentphysicalism),isequivalenttophysicalism,witheliminativismasadegeneratecase.Thereasonthemind–bodyproblemdoesnotgoaway,despiteourbeingclearabouttheoptionsinrespondingtoit,isbecauseoftheconstantbattlebetweencommonsense,whichfavorstheviewthatthementalisabasicfeatureofreality,29\nKirkLudwigcriterialbehaviorismconceptual+externalismsupervenienceonthephysicalanalyticfunctionalismphysicalismreductiontothephysicaltranslationalbehaviorismReject(1):irrealismeliminativematerialismpsychologicalreductionism–bodyproblemReject(2):conceptualreductionismmaterialism)psychophysicalidentitytheories(akacentralstateneutralmonismAccept(3)and(4):constituentexplanatorysufficiencyandconstituentnon-mentalism:genericrestrictedtheoriesspecialparticleaspecttheory:restricteddoubleThelogicalspaceofsolutionstothemindmixedgenericemergentmaterialismuniversalnon-reductivepanpsychismFigure1.1type-type:theoryemergentismnoconnectionsdoubleaspectmerelysupervenientmentalparticletheoriesrestrictedparallelismpureinteractionismneutralReject(4):ontologicalanti-reductionismsubstancedualismemergentismReject(3):conceptualanti-reductionism:1.Realism2.Conceptualautonomy3.Constituentexplanatorysufficiency4.Constituentnon-mentalismuniversalreductiveidealismAccept(1)and(2):realismandconceptualautonomypanpsychismimmaterialism)(akaphenomenalism,Differentia30\nTheMind–BodyProblem:AnOverviewandthepulltoseeitasanauthoritativedeliveranceofsciencethatthisisnotso.Wefindourselvesconstantlypulledbetweenthesetwopoles,unabletoseeourmindsasnothingoverandabovethephysical,unwillingtoseetheuniverseascontaininganythingnotexplicableultimatelyintermsofitsbasic,apparentlynon-mental,constituents.Notes1Theterm‘themind–bodyproblem’isnotusedunivocally.Whatguidesmyusageisaninterestingettingatthepuzzlethathasgeneratedthegreatvarietyofpositionsthatwefindinphilosophicalandscientificdiscussionsoftherelationofmentalphenomenatophysicalphenomena.IfIamright,thereisapuzzlewecanarticulateclearlytowhichallthepositionsontherelationofthementaltothephysicalcanbeseenasresponses.Ifanyoneproblemdeservesthelabel‘themind–bodyproblem’itisthis.2Inthecourseofdiscussion,aconsiderableamountofterminologywillbeintroduced.Thisispartlytoenableustostateourproblemanditspossiblesolutionswithprecision.Moreterminologyisintroducedthanisstrictlynecessaryforthis.Theexcessisintendedtoprovideafoundationforfurtherreadingintherelevantliteratureonthetopic.Iwilloftenprovidereferencesrepresentativeofparticularviewsorarguments.Ilistheresomecollectionsofpaperswhichtogethergiveafairlycomprehensivepictureofthehistoricalandcontemporarydevelopmentofviewsonthemind–bodyproblem:Vesey(1964),Anderson(1964),O’Connor(1969),Borst(1970),Rosenthal(1971),Block(1980),Eccles(1985),Lycan(1990),Rosenthal(1991),BeakleyandLudlow(1992),WarnerandSzubka(1994),Blocketal.(1997),Cooney(2000).Rosenthal(1991)isparticularlycomprehensive.Vesey(1964)containshistoricalsourcesnotfoundintheothers.Anderson(1964)containsearlypapersonthecomputermodelofthemind.Eccles(1985)containscontributionsmostlybyscientists,bothphilosophicalandscientificincharacter.Blocketal.(1997)isdevotedspecificallytorecentworkonconsciousness.3“Bytheterm‘thought’,”Descartessays,“Iunderstandeverythingwhichweareawareofashappeningwithinus,insofaraswehaveawarenessofit”(1984,vol.I:195[1644:I.9]).ThiscorrespondstothefeatureofconsciousnessIdescribebelowasnon-inferentialknowledgeofourmodesofconsciousness.Descartesheldalsothatastateisamentalstateonlyifitisconscious,butthisiswidelyregardedastoostringentarequirement,forreasonsconsideredbelow.4Onthiscommon-senseconceptionofeventsaschanges,theyaredatableparticulars.Theymaybecomplexaswellassimple.Mysnappingmyfingersisanevent.SowastheSecondWorldWar.IfanobjectchangesfrombeingFtobeingnon-F,theeventisthechangingfrombeingFtobeingnon-F.Ifweindividuateeventsintermsofwhichobjects,times,andpropertiestheyarechangeswithrespectto,thequestionwhethermentaleventsarephysicaleventsisreducedtothequestionwhethermentalpropertiesarephysicalproperties.5Itissometimesthoughtthatthisistoostrong.Foronemightmistakenlythink,e.g.,thatoneisinpainbecauseoneexpectstobe,giventheoccurrenceofsomeeventonehadanticipatedandexpectedtocausepain.Forexample,someonemightthinkhewasinpainwhensomeoneputsanicecubeonthebackofhisneck,ifhehadbeen31\nKirkLudwigtoldthatapieceofmetalheatedredhotwasabouttobepressedagainstthebackofhisneck.Thepossibilityofhishavingafalsebeliefinthesecircumstancesdoesnotshow,however,thathedoesnotknowwhatheexperienced.Forhewillcorrecthismistake.Hewillrealizequicklythatheisnot,andwasnot,inpain.Forhecanrecallwhattheexperiencewaslike.Thatrequiresknowingwhatcharacterithadatthetime,sinceonecannotremembersomethingonedidnotoriginallyknow.Memorypreservesbutdoesnotcreateknowledge.6Fordiscussionofthisissue,seeessays20–24inBlocketal.(1997).7SeeNagel(1979b,1994,1998),andMcGinn(1989,1991,1999).McGinnandNagelthinktheremustbeawayofunderstandinghowtheoperationsofourbrainsgiverisetoconsciousness,butthatwecurrentlyhavenoconceptionofhowthatcouldbe.McGinnisthemorepessimistic,sincehethinkswhateverthecorrectexplanation,itisonethatwecannotinprincipleunderstand,givenourcognitivemake-up,whileNagelthinkswemayonedaydevelopappropriateconcepts.Theviewthatconscious-nessisthecentraldifficultyisasoldasdiscussionofthemind–bodyproblem.8Thisterminologytracesbacktomedievalphilosophy;itisderivedfromtheLatinverbintendere,for‘pointat’or‘aimat’;itwasusedtocharacterizetheobjectofathoughtwhenitdidnotexistinreality,buthadintentionalinexistence,orexistedonlyinten-tionallyinthethinkingsubject.9Somethingsbesidesattitudesofthesortswehavebeendiscussingcanbesaidtorepresentthings,andsotohaveintentionality;e.g.,asentence,oraportrait.How-ever,thesehaverepresentationalcontentonlybecauseagentstreatthemasrepresen-tationsinaccordancewithvariousrules.Thisisderived,asopposedtooriginal,intentionality(Searle1983,1984).Mentalstateshaveoriginalintentionality.Iuse‘intentionality’,withoutqualification,tomeanoriginalintentionality.10Adispositionisastateofanobjectthatconsistsinitssettledtendencytoundergosomechangeincertainconditions.Watersolubilityisasimpledispositionalstatepossessedbysaltandsugar:whenplacedinunsaturatedwaterinacertainrangeoftemperaturesandpressures,theydissolve.Thechangeundergonethatcharacterizesadispositionisitsmanifestationproperty,thepropertythatismanifested.Themanifes-tationconditionisthatunderwhichthemanifestationpropertyismanifested.Oftenbothoftheseareencodedinthenameofthedisposition,asin“watersolubility.”Dispositionalattitudesarenotsimpledispositions,butwhatGilbertRylecalled“multi-trackdispositions”(1949:43–4).Thismeansthattheymanifestthemselvesinvariousconditionsinvariousways.Moreover,theyareinterlockingdispositions:amongthemanifestationconditionsforanygivenattitudewillbeconditionsinvolvingwhatotherattitudesanagenthas.AdesiretobuyacertainbookwillnotbemanifestedunlessIbelieveIhavetheopportunitytopurchaseit,andhavenootherdesireswhosesatisfactionIrankabovethatforthepurchaseofthebook,andwhichIthinkIcansatisfyonlytoitsexclusion.11Manyrecenttheoriesofcognitiveactivityhaveappealedtoinprincipleunconsciousinferencesintheirexplanations,therebypresupposingthetwocanbeconceivedindependently.SeeLudwig(1996c)forcriticismoftheseviews.12Somephilosophershaverecentlyarguedthatconsciousstatesmaybeexhaustivelycharacterizedintermsoftheirrepresentationalcontent.ExamplesareLycan(1996),Dretske(1997),andTye(1997).Forcontraryviews,seeSearle(1993),Chalmers(1996),andSiewert(1998).Representationalaccountsofconsciousnesshaveoften32\nTheMind–BodyProblem:AnOverviewbeenmotivatedbythethoughtthatitiseasiertoseehowintentionalstatescouldbereducedtophysicalstatesthanhowconsciousnesscouldbe.Inmyview,whichIdonotargueforhere,intentionalityisultimatelytobeunderstoodasaformofcon-sciousness,ratherthantheotherwayaround,dispositionalintentionalstatesderivingtheircontentfromtheirmanifestationinconsciousness.Ifso,thequestionoftherelationofconsciousnesstothephysicalisbasic.13Importantly,Idonotcharacterizetheclassofphysicalpropertieshereaspersenon-mental,thoughgiventhelistofbasicproperties,theyareclearlynotmentalperse.Thisleavesitopenthatmentalpropertiescouldbeanalyzedaslogicalconstructionsofprimaryqualities,or,asconceptuallysuperveningonthem(seesection1.4).14SeePoland(1994:esp.pp.109–47)andPapineau(1993:29–32).15Moreproperly,afullymeaningfulpredicateinalanguageLexpressesaconceptandpicksoutaproperty.Indifferentlanguagesthesamewordmayexpressdifferentconcepts,ornone.Iomitthisrelativizationforbrevity,butitshouldbeunderstoodasimplicitwhereverweareconcernedwiththerelationoflinguisticitemstotruth,concepts,andproperties.Ialsoignore,forthemostpart,complicationsintroducedbytenseandothercontext-sensitiveelementsinnaturallanguages.16Thereareotherconceptsofpropertythatmightbe,andsometimesare,employedonwhichthiswouldnotbetrue.Forexample,onemightindividuatepropertiesintermsofthesetsofpossibleindividualswhopossessedthem.Thentwopredicateswouldpickoutthesamepropertyifftheywerenecessarilycoextensive,whichdoesnotrequiresynonymity(e.g.,‘istrilateral’and‘istriangular’).Butthethesesaboutpropertyidentitythatcouldbeexpressedinthiswaycanbeexpressedwithoutthedubiousontologyandunhelpfulinnovationinterminology,whichshouldnotbeencouraged.17Moregenerally,wewouldspeakofsentencesasanalyticrelativetooccasionsofutterance,sincewhatmanysentencesexpressinnaturallanguagesisrelativetocon-textofutterance.18Thereiscontroversyaboutwhetherthereareanalyticstatements,conceptualtruths,andtruthsknowableapriori,butinstatingthemind–bodyproblemitisnotneces-sarytotakeastandonthis.W.V.Quine’s“TwoDogmasofEmpiricism”(1953)isthelocusclassicusofthecaseagainstanalyticity.GriceandStrawson(1956)isanimportantearlyreply.19‘Supervenience’initscurrentuseisusuallysaidtohavebeenintroducedinthecontextofethicaltheorybyR.M.Hareintheearly1950stodescribetherelationofethicalpropertiestonaturalproperties,andthenimportedintodiscussionsinthephilosophyofmindbyDavidson(1980).Itwasinuseearlierintheemergentisttradition,thoughperhapsnotwithquiteasspecificameaning;seeKim(1993b:essay8).20Therearemanychangesonecanringonthisformulation.Forexample,ifweputin‘itisconceptuallynecessarythat’beforethewholeright-handsideofthebiconditional,wegetaversionofwhathasbeencalledstrongsupervenience(Kim1993b:essay4).Thereareweakervarietiesaswell.IusethisformulationbecauseIwishtoallowconceptualsupervenienceofthementalonthephysicaleventhoughtherecouldbeaworldofnon-materialobjectsthathadmentalproperties.Thisisapossibilitywhichfunctionalism,forexample,leavesopen.Thisgivescontenttotheideathatsuperveni-enceisstrictlyweakerthanreduction.Sometimessupervenienceclaimsareformulatedintermsofindiscernibilityclaims:F-propertiessuperveneonG-propertiesiffnecessarily33\nKirkLudwigthingswhicharealikewithrespecttotheirG-propertiesarealikewithrespecttotheirF-properties.SeetheessaysinKim(1993b)andSavellosandYalçin(1995)forfurtherdiscussionofthevariantsandtheirrelationstooneanother.21Therequirementthatpsychologicallaws(includinganypsychophysicallaws)beentailedbyphysicallawsisneededtoavoidtheproblemofluckymaterialism(Witmer2000).22Thispositionmayappearstrongerthanitis.Iputnoconstraintsonphysicalpro-pertiesotherthanthattheybephysical.Complexrelationalpropertiesmayfigureinthesuperveniencebase.Thus,itisequivalenttotheviewthatacompletephysicaldescriptionoftheworldentailsacompletepsychologicalaccountofit.23Ithasbeenusedinaweakersensetodenoteamaterialistontology,andinastrictersense,e.g.,bytheLogicalPositivists,tomeanthatallstatementsaretranslatableintothevocabularyofphysics.24By‘non-mentalproperties’hereImeanpropertiesthatareclassifiedintermsthatarenotmentalassuch,sothatsomemembersoftheclass,andcertainlyallbasic(i.e.,non-complex)members,arenotmental.Thisallowsthatmentalpropertiesmaybeasubclassofthepropertiesinquestion.Thatistosay,(2)assertsthattherearenoclassesofpropertiesthatarenotmentalpersetowhichmentalpropertiesareconcep-tuallyreducible.25Inthepresentcontext,byanon-relationalpropertywemeanapropertythatanindividualhaswhichdoesnotrequiretheexistenceofsomecontingentlyexistingindividualnotidenticalwiththeindividualpossessingthepropertyoranypartofit,anddoesnotrequirethenon-existenceofanythingorkindofthing.Forexample,beingmarriedandbeingaplanetarerelationalproperties,beingroundandbeingredarenot.26Thisrulesoutappealtopropertiesthatconstituentshavebecauseofemergentpro-pertiesofthewholestheycompose.27Thisleavesopenthattheymayhavementalpropertiesinthesensethattheyhaverelationalpropertieswhichentailthatsomethingpossessesmentalproperties,e.g.,becausetheycoexistwithorarepartofathingthathasirreduciblementalpropertiesbutwhichisnotitselfabasicconstituentofthings.Alsothisleavesopenthatthebasicconstituentsofthingshavepropertieswhichwemightnotrecognizeasbroadlyphysical,butitdoesnotallowthattheybemental.Thus,constituentnon-mentalismisamoreliberalthesisthanconstituentphysicalism.28SeeBealer(1992)forageneraldefenseofthesemethodsfordiscoveringwhatisnecessaryandpossible;amorerecentbook-lengthdefenseofconceptualanalysisisJackson(1998).29ThediscoveryofthisparadoxbyBertrandRussell,inMay1901,playedanimportantroleinfoundationalstudiesinsettheoryandmathematicsearlyinthetwentiethcentury.30Thequestionoftherelationofconsciousnessandintentionalitybecomesimportanthere,forthethoughtexperimentsmentionedseemtodependonourthinkingthataconsciouspointofviewcouldbemissinginabeingphysicallyandbehaviorallylikeus,orbepresentinabeingwithnoassociatedbodyatall.Ifintentionalstatesandconsciousstatesareindependent,thesupportofthesethoughtexperimentsfortheirreducibilityofthementaltoutcourtisreduced.31Withtheexception,however,oftheroleofthenotionofanobservationinquantummechanics:howseriouslythisistobetakenisamatterofcontroversy.34\nTheMind–BodyProblem:AnOverview32SeeAnnas(1992)forsurveyofancientphilosophyofmindconcentratingontheHellenisticperiod.33Thiswasaminoritypositioninantiquity.IntroducedbyPlato,itassumedtheimport-anceithasinthelaterwesterntraditionthroughtheinfluenceofPlato’sphilosophyonCatholictheology,throughwhichithaspermeatedideasaboutmindandbodyinwesternculture.34SeeWoolhouse(1993)fordiscussionofthenotionofsubstanceinearlymodernphilosophyrationalists.35Wemustexcludeheresuch“formal”propertiesashavingaproperty.36TheinitialmovesintheargumentaremadeinthesecondmeditationofDescartes’smasterpieceMeditationsonFirstPhilosophy(1985[1641])andconcludedinthesixth;seealsoPrinciplesofPhilosophy(1985[1644]:§63).37ThisanalogywasconceivedbyLeibniz,thoughhisbasicmetaphysicsrejectssub-stancedualism.38Thoughdualismisnotcurrentlyapopularviewamongphilosophersorscientists,itisstillnodoubtoneofthemostcommonly,ifunreflectively,heldviewsabouttherelationofmentaltophysicalphenomena,asitisthebackgroundmetaphysicsofanumberoftheworld’smajorreligions;anditisnotwithoutcontemporarypro-ponentsamongphilosophersandscientists,see,e.g.,Foster(1996),Eccles(1953:ch.8),PopperandEccles(1977:ch.E7).39ThreelandmarksofthetwentiethcenturyareCarnap(1928),Lewis(1929),andGoodman(1951).AmorerecentproponentisGrayling(1985).40AdetailedbibliographyofsourcesisavailableattheendofthearticleonpanpsychisminEdwards(1967).41Panpsychism,andothermentalparticletheories,asforreductivematerialism,isanexpressionoftheidea,asPopperandEccleshaveputit,thatthereisnothingnewunderthesun,whichisanexpressionofaformoftheprincipleofsufficientreason:nothingcancomefromnothing(PopperandEccles1977:14).42Nineteenth-centurydoubleaspecttheoristsincludeShadworthHodgson(1870:esp.ch.3)andG.H.Lewes(1877).43See,e.g.,thediscussionofMortonPrince(1885;repr.inVesey1964:187).44PerhapsStrawson’sviewthattheconceptofapersonismorebasicthanthatofaperson’smindorbodymaybeconstruedasofthissort(1958).45HowshouldweclassifyaviewsuchasHume’s“bundle”theoryoftheself?Onthisview,thereisnothingthatistheself,butrathereachselfistobeconstruedasconstitutedoutofasetofperceptionswhichbearappropriaterelationstooneanother.Theperceptionsareintrinsicallymentalincharacter,likementalatoms.Theyarenot,though,apparentlythoughtofasinspace.So,whileamentalparticleviewofasort,itismorelikesubstancedualismwithoutthebasicmentalsubstancesbeingthinkingbeings,butratherthoughtsconstitutiveofthinkingbeings.Ifwetake“perceptions”tobenon-mentalthemselves,andtakeboththeselfandordinaryobjectstobelogicalconstructionsoutofthem,wearriveataversionoftheneutralmonismadvocatedbyJames,Mach,andRussell(seebelow).46Itisimportanttodistinguishemergenceinthissensefromemergenceofhigherlevelsoforganizationofcomplexsystemsgovernedbysimplerulesthatisoftendiscussedinthecontextof“chaos”theory.Thepropertiesofthelattersortconceptuallysuper-veneontherulesgoverningtheconstituentsofthesystem,theirproperties,andtheir35\nKirkLudwigarrangement.Theyareemergentonlyinthesenseofbeingsurprisingtous,andsotheirstatusasemergent,inthissense,isafunctionofourinabilitytoeasilypredictthem.47Thisrequiresustodisallowindefinitelylongdisjunctionsfromexpressingrelevanttypes;otherwise,bydisjoiningallthenomicallysufficientconditionsstatedinphysicaltermsforagivenmentaltype,wecouldalwaysarriveatanomicallynecessarycondition.48Atokenisaninstanceofatype.Forexample,intheprevioussentence(inscription),therearefourtokensoftheletter“a.”Tokensarealwaysparticulars.Everytokenisidenticalwithitself.Wegetinformativestatementsabouttokenidentitywhenweusedifferentwaysofpickingoutthesamething.Itcanbeinformative,e.g.,tobetoldthatPlutoisthesmallestplanetintheSolarSystem.Type-typeidentity,strictlyspeaking,isaboutproperties.Again,everypropertyisidenticaltoitselfandtonodistinctthing.Informativetype-typeidentitystatementspickoutthepropertiesindifferentways.Wewillseeanexamplebelowofatype-typeidentitytheoryofthementalandthephysicalthatmakesthisaninterestingempiricaldiscovery.49Theconceptionofeventsarticulatedinnote4isincompatiblewithanomalousmonism,foritindividuateseventsintermsoftheobjectsandpropertiesthattheyarechangeswithrespectto.Thus,unlessmentalpropertiesarephysicalproperties,whichonthisviewtheyarenot,nomentaleventistokenidenticalwithanyphysicalevent.Therearevariousweakerrelationsthatcouldbearticulated.Forexample,itmightbesaidthateverymentaleventoccursatthesametimeasandinthesameobjectasaphysicalevent.Inanycase,itisnotclearthatmuchhingesonthis.Themorefundamentalquestionisaboutobjectsandpropertiesratherthanevents.50Inorigin,amedicaltermmeaning“symptomofanunderlyingcause”or“secondarysymptom.”51SeeMcLaughlin(1992)foradiscussionofthisparticularschoolinthebroaderemergentisttradition.BeawarethatMcLaughlinuses‘emergentism’inanarrowersensethanitisusedhere,namely,tocoverwhatIwouldcallemergentmaterialismwithdownwardcausation.‘Emergentism’istherighttermfortherejectionof(3);wecandistinguishepiphenomenalandnon-epiphenomenalversions,thelatterofwhichwillatleastincludeemergentismwithdownwardcausation.Alas,terminolo-gicalvariationinphilosophyisendemic.Broadhimself,whointroducedtheterm‘emergentmaterialism’,didnottakeittoimplydownwardcausation,whichheacceptedtentativelyasanempiricalhypothesisonthebasisofwhathetooktobetheevidenceofpsychicalresearch.52Thiscontrastanddebatebetweenepiphenomenalemergentistsanddownwardcausa-tionemergentistsreprisesasimilardebateinantiquitybetweenfollowersofAristotle(Caston1997).53Formorerecentdiscussions,seeArmstrong(1968:47)andKim(1993a,b).54Anoteisinorderontheterm‘propertydualism’,whichhasfiguredprominentlyinrecentliteratureonthemind–bodyproblem.Thislabelisoftenusedinapplicationtoemergentism,butappliestoanypositionthatholdsthatthereareobjectsthathavementalproperties,andthereareobjectsthathavephysicalpropertiesandthatbothsortsarebasicproperties,notconceptuallyreducibletoeachotheroranythingmorebasic.(Propertydualismisnotcoextensivewithanypositionthatholds(1)and(2)andeitherof(3)or(4),since,e.g.,idealismembraces(1)–(3),butreduceswhatareordinarilythoughtofasphysicalpropertiestomentalproperties.)Propertydualism36\nTheMind–BodyProblem:AnOverviewisaweakerviewthansubstancedualism,butisentailedbyit.“Propertydualism”isoftenusedasatermofabusebyphilosophersattractedbyreductionism,withtheideaofassociatingitsproponentswiththediscreditedviewofsubstancedualismbytheoverlapinthespellingoftheirlabels.Theintroductionof“propertydualism”intothephilosophicalvocabularyisnotanentirelyhappyterminologicalinnovation,andthatisonereasonitdoesnotfigureprominentlyinmydiscussion.Quiteapartfromitsassociationwithdemagoguery,thelabelfalselysuggeststhatthereareatmosttwofamiliesofpropertiesirreducibletoeachother:butevensettingasidethecurrentissue,therearemanymutuallyirreduciblefamiliesofproperties(colorandshapeproperties,forexample).55Therearesomepossiblethoughunoccupiedpositionsherethatwewillnotsurvey,suchastheviewthatthementalsupervenesonorisconceptuallyreducibletosomethingnon-mental,andthephysicalinturnsupervenesonorisconceptuallyreducibletothemental.56Inthiswayitislikethereductionofmathematicstologicandsettheory.Wecanretainouroldformsofspeech,butourontologyincludesonlysets,notnumbersinaddition.Therelationofthementalandphysicaltounderlyingrealityonneutralmonismisliketherelationofoddandevennumberstotheunderlyingrealityontheset-theoreticreductionofmathematics.Eachisdistinctfromtheother,andhasanessentialpropertytheothercannothave,buteachisexplainedasalogicalconstruc-tionoutofsomethingmorebasic.57Carnap(1931)andHempel(1935)provideearlyexamplesoflogicalbehaviorists;bothlaterretreatedfromtheearlyposition.Ryle’sTheConceptofMind(1949)wasanimportantandinfluentialbehavioristmanifesto(thoughRyledeniedthetermappliedtohisview).Wittgenstein’sPhilosophicalInvestigations(1950)wasanimport-antinspirationforcriterialbehaviorism.See,forexample,Malcolm(1958).Import-antpsychologicalbehavioristswereWatson(1925)andSkinner(1974),thoughtheirbehaviorismwasmethodologicalratherthanlogical.58SeePutnam(1968).Logicalbehaviorismseemstohavesuccumbedtoadangerthateveryreductiveprojectfaces.AsC.I.Lewisputit:“Confrontedwithproblemsofanalysiswhichthereistroubletoresolve,onemaysometimescircumventthembychangingthesubject”(1941:225).59Thisisheldtobetrueasamatteroffact.Ofcourse,iftherewerenon-physicalobjectsthathadinternalstructure,theywouldhavefunctionalstatesaswell.60Everyerahasitsfavoredmetaphorprovidedbyitsprestigetechnology.Intheseven-teenthandeighteenthcenturies,itwastheclockorthemill.Inthenineteenth,itwasthesteamengine.Inthelatterhalfofthetwentieth,itbecamethecomputer.61ThePythagoreansadvocatedthegeneralideaoffunctionalism,thathavingaminddependsonacertainorganizationofthebody,inantiquity.ItisoneofthepositionsthatSocratesrespondstointhePhaedoinSimmias’ssuggestionthatthesoulistothebodyastheattunementofitistoastringinstrument(Plato1989:69).62Therearetwowaysofunderstandingpsychofunctionalism’sempiricalcharacter.First,itcanbeunderstoodasaversionofemergentismwithbridgelawsconnectingfunc-tionalwithmentalstates(see,e.g.,Chalmers1996:ch.6).Second,itmightbemaintainedthattheidentificationofmentalwithfunctionalstatesisatheoreticalidentification,liketheidentificationoflightningasanelectricaldischarge(thisviewhasbeenadvocatedforintentionalstatesinRey1997).IfwhatIsaybelowaboutthis37\nKirkLudwigiscorrect,thisintroducesanempiricalelementintothediscovery,butnotinawaythatpreventsthisview,ifcorrect,fromcollapsingintoanalyticfunctionalism.Seethediscussionoftheidentitytheorybelow.63Amorerecentvariantonthegeneralthemeisconnectionism.Aconnectionistsystemconsistsofasetofinterconnectedunitsthatcantakeonactivationvalues:theinter-connectionsdeterminetheinfluenceoftheactivationvalueofagivennodeonthoseconnectedtoit.Throughtheirconnections,unitsmayinhibitorexciteotherunitstovariousdegreesdependingontheirownactivationstates.Certainunitsmaybedesign-atedinputunitsandothersoutputunits.Theactivationvaluescanbecontinuous,soaconnectionistsystemisnotafinitestatemachine.Butitfitsourinitialverygeneralcharacterizationofafunctionalsystem,sincedifferentconnectionistsystemsarewhollycharacterizedintermsoftheirstates’relationstoinputandoutputandotherstates.Thedifferencebetweenclassicalfunctionalismandconnectionismwillnotberelevantatthelevelofourdiscussionhere.64Theforceofthisobjectionisunclear.Eithermentalpropertiesareanalyzableasfunctionalpropertiesornot.Ifnot,thenthereisthequestionwhethertheyareanalyzableasphysicalproperties.Ifso,thatisanendtothematter,andthechargeofchauvinismisbootless.Ifnot,itisanempiricalmatterwhatphysicalstatetypes,ifany,mentalstatetypesarecorrelatedwith,andourhunchesorprejudicesaboutitareirrelevant.Thoughinthelattercase,clearlydifficultieswillarisewhenwetrytoconfirmordisconfirmclaimsaboutphysicalsystemsthatareverydifferentfromourselves.65SeealsoSearle’sChineseRoomthoughtexperiment(1980),andChalmers(1996:ch.3)forarecentdeploymentofso-calledzombiethoughtexperimentstoestablishtheirreducibilityofconsciousmentalstates.66Putnamwascarefultoexcludesystemsthatcontainpartsthathaveorganizationslikethewholetheyconstitute.ThiswouldruleoutthesysteminBlock’sthoughtexperi-mentasconstitutingaperson.However,itisdifficulttoseewhatjustifiestheexclu-sion.Forifourmentalconceptsarefunctionalconcepts,itshouldnotmatterhowthesystemthathastheappropriatefunctionalorganizationisconstituted.67SeeChalmers(1996:91–101)forasomewhatfullerdiscussionandsomeresponsestoobjectionsthathaveappearedintheliterature.68Afunctionalistneednotrequirethis.Afunctionalsystemcouldbecharacterizedintermsofnon-causaltransitionsbetweenstatesgiveninputandoutput.Butthisopensthedoortoagreatmanymoremachinetabledescriptionsofobjectsthatmayhavemindsthanafunctionalistwilltypicallywanttocountenance.69Theeventreportedintheheadlinesofthismorning’spapercausedextensivefloodingincoastalareasofFlorida,butitwasnotbyvirtueofbeingofthattypethatitdidso,butinvirtueofitsbeingthepassingofacategory3hurricaneoffthecoast.Causalrelationsholdbetweenparticulars,datableevents,orstates.Buttoexplainwhytheyholdbetweenthoseparticularswemustappealtotheirtypes.70SeeJacksonandPettit(1988),Block(1990),Fodor(1991),Dardis(1993),andLudwig(1994a,1998)fordiscussion.71Atthispointtoothequestionwhetherintentionalstatesareconceptuallyindepend-entofconsciousstatesisimportant,forourconvictionthatmentalstatesarecausallyrelevanttobehaviorseemstoattachinthefirstinstancetoconsciousmentalstates,andtodispositionalstatesonlythroughtheirmanifestationconditionsinconsciousness.38\nTheMind–BodyProblem:AnOverviewFordispositionalstatestooaredefinedintermsofmanifestationconditions,andsoassucharenotcausallyrelevanttothoseconditions.72TheaccountgivenheredepartsfromLewis’sown.Perhapsthedepartureislargelyinterminology,butitisstillimportant.Lewishasarguedthatdespitethetheoreticalidentificationofpainwithaphysicalpropertyinhumanbeings,itstillmakessensetosaythatsomebeing(aMartian,e.g.)couldbeinpainthoughhedoesnothavethatpropertywhichinhumanbeingsispain(Lewis1980).Howisthispossible?Itisnot,ifweunderstandtherelationbetweenpredicates,concepts,andpropertiesasIhaveintroducedthem.OntheaccountIhavegiven,thepredicate‘isinpain’expressestheconceptofpainandisusedtoattributethepropertyofbeinginpain,anditdoeseachoftheseinvirtueofitsmeaninginEnglish.Thepropertyis,sotospeak,theshadowofthemeaningofthepredicatecastontheworld,andtheconceptistheshadowitcastsinourthoughts.Ifthepropertyofbeinginpainisaphysicalproperty,so,onthisview,istheconceptofpainaphysicalconcept.Lewis,however,identifiessomethingelseastheconceptofpain.Toputitbriefly,Lewisuses‘conceptofpain’todenotetheconceptexpressedbythepredicate‘isathingthathasthepropertyPsuchthat,forthemostpart,T(P)forbeingsofkindK’where‘T(P)’isreplacedbytheappropriatepsychologicaltheorywith‘P’intheplaceofthevariablerepresentingthepropertyofpain.Thatconceptappliestoathinginvirtueofthatthing’shavingsomepropertythatplaysacertainrolemediatinginputandoutput.Itmighthavebeenthatadifferentpropertyplayedthatrole.Andindifferentkindsofbeings,perhaps,forthemostpart,differentpropertiesplaythatrole.However,Lewisdoesnotsaythatthepropertyofbeinginpainistheoneattributedusingthisformofpredicate.Rather,Lewiscallsthepropertythatactuallyplaystherolethepropertyofbeinginpain.Thisallowsthenthatindifferentkindsofbeingsadifferentpropertycanbe(called)thepropertyofbeinginpain.Italsoapparentlyallowsthatifthingshadbeendifferent,adifferentpropertyinuswouldhavebeen(calledbyus)thepropertyofbeinginpain.Apparently,however,Lewisdoeswanttotreatthepredic-ate‘isinpain’asifitattributedthepropertythatplaystherightrole.Thus,hesays“isinpain”isambiguouswhenweapplyittodifferentkindsofbeings,andwhenweconsideritindifferentpossibleworlds.Foradifferenceinpropertyattributedentailsadifferenceinthemeaningofthepredicate.Itisasifwehaddecidedtosaythatthepropertybeingrichisattributedusing‘hasalotmoremoneythanmostpeople’buttheconceptofrichnessisexpressedbythepredicate‘isLudwig’sfavoriteproperty’.Ikeeptheconceptofpainattachedtothepredicate‘isinpain’,andsomatchedwiththepropertyattributedusingit.Thisfollowsthetraditionalalignment,andprovidesusaclearerviewoftheissues.73Therearemanyargumentsagainstthepsychophysicalidentitytheoryandphysicalismmoregenerallythatrestonthoughtexperimentsdesignedtoshowthatnothingfollowsaboutwhatmentalpropertiesanobjecthasfromanexhaustivedescriptionofitsphysicalconstruction.Onestyleofargumentmuchdiscussedrecentlyhasbeendubbed‘theknowledgeargument’.SomedeploymentsoftheargumentinthelatterhalfofthetwentiethcenturyareMeehl(1966),Nagel(1979b),Jackson(1982,1986).LeibnizalreadygivesaversionofsuchanargumentinTheMonadology(1714:sec.16):“Ifweimaginethatthereisamachinewhosestructuremakesitthink,sense,andhaveperceptions,wecouldconceiveitenlarged,keepingthesameproportions,sothatwecouldenterintoit,asoneentersintoamill.Assumingthat,wheninspecting39\nKirkLudwigitsinterior,wewillonlyfindpartsthatpushoneanother,andwewillneverfindanythingtoexplainaperception.”Theseargumentsarecertainlydecisiveagainstanyversionofthepsychophysicalidentitythatsuggeststhatwecanperformanarmchairanalysisofourmentalconceptstodeterminethattheyinfactpickoutneuro-physiologicalproperties.Theydonotaddressversionsofthetheorythattreatourordinarytermsashavingtheirconceptsfixedbydescription:theburdensuchanapproachtakesup,though,isscarcelylessheavy,foritmustallow,aswehaveseen,thatourtermsmayfailtoexpressanyconceptsatall.74Theviewitselfwascertainlynotundiscussedpreviouslyinthetwentiethcentury.Broaddiscussesanddismissesit(1925:622–3).C.I.Lewisdiscussedandcriticizedaformoftheidentitytheory,whichhepresentsasproposingdescriptivedefinitionsofmentalterms,inmuchthesamespiritasthetheoryIhavepresented(1941:230–1).SomeofSmart’srepliestoobjectionsareclearlydirectedatBroad’sandLewis’searlierdiscussions.75Theyregardedpropositionalattitudesasunderstandablebehavioristically,orfunction-ally.However,thepositioncaneasilybegeneralizedtopropositionalattitudes.76SeeBealer(1987,1994)forargumentsagainstthisappealtowhatissometimescalledscientificessentialism.77Inanycase,itshouldbenotedthatthesameunclaritywouldattachtowhatevernotionofpropertyidentitywouldbehereinvokedasattachestometaphysicalneces-sity:ifwetrytoexplainitinaccordancewiththetradition,wemustadmitthatwhatwediscoveristhat,e.g.,“water”and“H2O”expressthesameconcept,contrarytothesupposition.78ThesebeganwithworkbyKripke(1980)onpropernamesandnaturalkindtermsandHilaryPutnam(1975)onnaturalkindtermsintheearly1970s.Initially,theseargumentsweredirectedtowardshowingthatthemeaningsofvariousnaturallan-guagetermsweredeterminedbytheircausalrelationswiththingsandkindsinourenvironments.Sinceweusethesesametermstocharacterizeourattitudes,however,itwassoonapparentthattheseargumentsmightbeusedtourgealsothatourthoughtcontentswereindividuatedrelativetowhatthingsandkindswereactuallyinourenvironments.79SeePutnam(1975)andBurge(1979,1982,1986).Widespreaduncriticalacceptanceofexternalismisasalientfeatureofdiscussionincontemporaryphilosophyofmind.80DifficultiesarediscussedinLudwig(1992a,1992b,1993a,1993b,1993c,1994a,1994b,1996a,1996b).81Theliteratureonthissubjectislarge.AnearlierpaperthatadvancedthisthesisparticularlyinresponsetoPutnam(1981)isBrueckner(1986).SeealsoBoghossian(1989,1993).82SeeJackson(1996)forafairlycomprehensivereviewofdiscussionofmentalcausation.83EarlyproponentswereFeyerabend(1963)andRorty(1965,1979).PerhapsWittgensteinendorsedeliminativismintheTractatusLogico-Philosophicus(1921),butifsoongroundsmoreabstractthanmorerecenteliminativists.Eliminativismmaybetheonemodernviewthatisnotrepresentedinancientphilosophy.Perhapstheatomists,LeucippusandDemocritus,mightbethoughttoendorseeliminativism,sincetheyheldthatrealityconsistedsolelyofatomsandthevoid.Buttheyshowednoinclinationtodenythattherewerepeoplewhothoughtandreasoned,and40\nTheMind–BodyProblem:AnOverviewDemocritusseemstohaveintendedtoexplainpsychologicalphenomenaintermsofhisatomisticmetaphysics(seeTaylor1999).Onecanbeapartial,aswellasawholesaleeliminativist.GeorgesReyarguesforfunctionalismforintentionalstates,buteliminativismforqualitativestates(1997).ReferencesAlexander,S.(1920).Space,TimeandDeity:TheGiffordLecturesatGlasgow1916–1918.London:Macmillan.Anderson,A.R.(ed.)(1964).MindsandMachines.EnglewoodCliffs:PrenticeHall.Annas,J.(1992).HellenisticPhilosophyofMind.Berkeley:UniversityofCalifornia.Aristotle.(1984).TheCompleteWorksofAristotle.Princeton:PrincetonUniversityPress.Armstrong,D.M.(1968).AMaterialistTheoryoftheMind.London:RoutledgeandKeganPaul,HumanitiesPress.Ayer,A.J.(ed.)(1959).LogicalPositivism.NewYork:FreePress.Beakley,B.andLudlow,P.(eds.)(1992).ThePhilosophyofMind:ClassicalProblems/ContemporaryIssues.Cambridge,MA:MITPress.Bealer,G.(1987).“ThePhilosophicalLimitsofScientificEssentialism.”PhilosophicalPerspectives,1:289–365.——(1992).“TheIncoherenceofEmpiricism.”ProceedingsoftheAristotelianSociety,66:99–138.——(1994).“MentalProperties.”TheJournalofPhilosophy,91:185–208.Beckermann,A.,Flohr,H.etal.(1992).EmergenceorReduction?EssaysontheProspectsofNonreductivePhysicalism.NewYork:W.deGruyter.Bedau,M.(1986).“CartesianInteraction.”MidwestStudiesinPhilosophy,10:483–502.Berkeley,G.(1710).ATreatiseConcerningthePrinciplesofHumanKnowledge.InMichaelR.Ayers(ed.),PhilosophicalWorks.London:TheGuernseyPress,1975:61–128.Block,N.(1978).“TroubleswithFunctionalism.”InC.W.Savage(ed.),PerceptionandCognition:IssuesintheFoundationsofPsychology.Minneapolis:UniversityofMinnesota:261–325.ReprintedinBlock(1980).——(ed.)(1980).ReadingsinthePhilosophyofPsychology.Cambridge,MA:HarvardUniversityPress.——(1990).“CantheMindChangetheWorld?”InG.Boolos(ed.),MeaningandMethod:EssaysinHonorofHilaryPutnam.NewYork:CambridgeUniversityPress:29–59.Block,N.,Flanagan,O.etal.(eds.)(1997).TheNatureofConsciousness:PhilosophicalDebates.Cambridge,MA:MITPress.Boghossian,P.(1989).“ContentandSelf-Knowledge.”PhilosophicalTopics,17:5–26.——(1993).“TheTransparencyofContent.”PhilosophicalPerspectives,8:33–50.Borst,C.V.(ed.)(1970).TheMind–BrainIdentityTheory.London:Macmillan.Brentano,F.(1955[1874]).PsychologievomEmpirischenStandpunkt(PsychologyfromanEmpiricalStandpoint).Hamburg:FelixMeiner.Broad,C.D.(1925).TheMindandItsPlaceinNature.NewYork:Harcourt,Braceandcompany.Brueckner,A.(1986).“BrainsinaVat.”TheJournalofPhilosophy,83:148–67.Burge,T.(1979).“IndividualismandtheMental.”MidwestStudiesinPhilosophy,4:73–121.41\nKirkLudwig——(1982).“OtherBodies.”InA.Woodfield(ed.),ThoughtandObject:EssaysonInten-tionality.Oxford:ClarendonPress:97–120.——(1986).“CartesianErrorandtheObjectivityofPerception.”InP.Pettit(ed.),Subject,Thought,andContext.NewYork:ClarendonPress:117–36.Carnap,R.(1928).TheLogicalStructureoftheWorld,trans.RolfA.George.Berkeley:UniversityofCaliforniaPress,1967.FirstpublishedinGermanin1928underthetitleDerLogischeAufbauderWelt.——(1931).“DiePhysikalischeSprachealsUniversalsprachederWissenschaft.”Erkenntnis,2:432–65.ReprintedinAyer(1959)as“PsychologyinPhysicalLanguage.”Caston,V.(1997).“Epiphenomenalisms,AncientandModern.”ThePhilosophicalReview,106:309–64.Chalmers,D.(1996).TheConsciousMind:InSearchofaFundamentalTheory.NewYork:OxfordUniversityPress.Churchland,P.(1981).“EliminativeMaterialismandthePropositionalAttitudes.”TheJournalofPhilosophy,78:67–90.Cooney,B.(ed.)(2000).ThePlaceofMind.Stamford:Wadsworth.Dardis,A.(1993).“Sunburn:IndependenceConditionsonCausalRelevance.”PhilosophyandPhenomenologicalResearch,53:577–98.Davidson,D.(1980).‘MentalEvents’.InEssaysonActionsandEvents.NewYork:ClarendonPress:207–24.Descartes,R.(1985[1641]).MeditationsonFirstPhilosophy.InThePhilosophicalWritingsofDescartes,vol.II,ed.andtrans.J.Cottingham,R.Stoothoff,andD.Murdoch.Cambridge:CambridgeUniversityPress:1–62.——(1984[1644]).ThePrinciplesofPhilosophy.InThePhilosophicalWritingsofDescartes,vol.I,ed.andtrans.J.Cottingham,R.Stoothoff,andD.Murdoch.Cambridge:Cam-bridgeUniversityPress:177–292.Dretske,F.(1997).NaturalizingtheMind.Cambridge,MA:MITPress.Eccles,J.C.(1953).TheNeurophysiologicalBasisofMind;ThePrinciplesofNeurophysiology.Oxford:ClarendonPress.——(ed.)(1985).MindandBrain:TheMany-FacetedProblems.NewYork:ParagonHouse.Edwards,P.(1967).TheEncyclopediaofPhilosophy.NewYork:Macmillan.Feigl,H.(1958).“TheMentalandthePhysical.”InH.Feigletal.,Concepts,TheoriesandtheMind–BodyProblem.Minneapolis:UniversityofMinnesotaPress:370–497.Feyerabend,P.(1963).“MentalEventsandtheBrain.”TheJournalofPhilosophy,60:295–96.Fodor,J.A.(1991).“AModalArgumentforNarrowContent.”TheJournalofPhilosophy,88(1):5–26.Foster,J.(1996).TheImmaterialSelf:ADefenceoftheCartesianDualistConceptionoftheMind.London:Routledge.Grayling,A.C.(1985).TheRefutationofScepticism.LaSalle,IL:OpenCourt.Grice,H.P.andStrawson,P.F.(1956).“InDefenseofaDogma.”PhilosophicalReview,65:141–58.Goodman,N.(1951).TheStructureofAppearance.Cambridge,MA:HarvardUniversityPress.Hempel,C.(1935).“TheLogicalAnalysisofBehavior.”InBlock(1980:14–23).Origin-allypublishedinRevuedeSynthese,1935.42\nTheMind–BodyProblem:AnOverviewHodgson,S.(1870).TheTheoryofPractice.London:Longmans,Green,ReaderandDyer.Huxley,T.H.(1874).“OntheHypothesisthatAnimalsareAutomata,anditsHistory.”InVesey(1964:134–43).Originallydeliveredin1874totheBritishAssociationfortheAdvancementofScience,Belfast.——(1901).MethodsandResults:Essays.NewYork:D.Appleton.Jackson,F.(1982).“EpiphenomenalQualia.”ThePhilosophicalQuarterly,32:127–36.——(1986).“WhatMaryDidn’tKnow.”TheJournalofPhilosophy,83:291–5.——(1996).“MentalCausation.”Mind,105(419):377–413.——(1998).FromMetaphysicstoEthics:ADefenseofConceptualAnalysis.NewYork:ClarendonPress.Jackson,F.andPettit,P.(1988).“FunctionalismandBroadContent.”Mind,97:381–400.James,W.(1950[1890]).ThePrinciplesofPsychology.NewYork:Dover.——(1910).Psychology.NewYork:HenryHoltandCo.PagereferencesinthetextaregiventoBlocketal.(1997),inwhichaportionofchapter11isreprinted.——(1904).“DoesConsciousnessExist?”TheJournalofPhilosophy,PsychologyandScien-tificMethods,1:477–91.Repr.inJames(1976[1912]).——(1976[1912]).EssaysinRadicalEmpiricism.Cambridge,MA:HarvardUniversityPress.Kant,I.(1997[1781]).TheCritiqueofPureReason.Cambridge:CambridgeUniversityPress.Originallypublishedin1781,amuchrevisedversionwaspublishedin1787.Kim,J.(1993a).“Mechanism,Purpose,andExplanatoryExclusion.”InKim(1993b:237–64).——(1993b).SupervenienceandMind.NewYork:CambridgeUniversityPress.Kripke,S.(1980).NamingandNecessity.Cambridge,MA:HarvardUniversityPress.Leibniz,G.W.(1714).PrinciplesofPhilosophy,or,TheMonadology.InPhilosophicalEssays.Indianapolis:Hackett,1989:213–24.Originallycomposedin1714forprivatecorre-spondenceasasummaryofhisphilosophicalviews.Lewes,G.H.(1877).ThePhysicalBasisofMind.London:TrubnerandCo.Lewis,C.I.(1929).MindandtheWorld-order:OutlineofaTheoryofKnowledge.NewYork:Scribner.——(1941).“SomeLogicalConsiderationsConcerningtheMental.”TheJournalofPhilosophy,38:225–33.Lewis,D.(1966).“AnArgumentfortheIdentityTheory.”TheJournalofPhilosophy,63:17–25.——(1972).“PsychophysicalandTheoreticalIdentifications.”TheAustralasianJournalofPhilosophy,50:249–58.Repr.inLewis(1999).——(1980).“MadPainandMartianPain.”InBlock(1980:216–22).——(1999).PapersinMetaphysicsandEpistemology.Cambridge:CambridgeUniversityPress.Ludwig,K.(1992a).“SkepticismandInterpretation.”PhilosophyandPhenomenologicalResearch,52(2):317–39.——(1992b).“BrainsinaVat,Subjectivity,andtheCausalTheoryofReference.”JournalofPhilosophicalResearch,17:313–45.——(1993a).“DirectReferenceinThoughtandSpeech.”CommunicationandCognition,26(1),49–76.——(1993b).“Externalism,Naturalism,andMethod.”InE.Villanueva(ed.),NaturalismandNormativity.Atascadero,Ridgeview:250–64.43\nKirkLudwig——(1993c).“DretskeonExplainingBehavior.”ActaAnalytica,811:111–24.——(1994a).“CausalRelevanceandThoughtContent.”PhilosophicalQuarterly,44:334–53.——(1994b).“FirstPersonKnowledgeandAuthority.”InG.Preyer,F.Siebelt,andA.Ulfig(eds.),Language,Mind,andEpistemology:OnDonaldDavidson’sPhilosophy.Dordrecht:Kluwer:367–98.——(1996a).“SingularThoughtandtheCartesianTheoryofMind.”Nous,30(4):434–60.——(1996b).“DuplicatingThoughts.”MindandLanguage,11(1):92–102.——(1996c).“ExplainingWhyThingsLooktheWayTheyDo.”InK.Akins(ed.),Perception.Oxford:OxfordUniversityPress:18–60.——(1998).“FunctionalismandCausalRelevance.”Psyche4(3)(http://psyche.cs.monash.edu.an/v4/psyche-4-03-ludwig.html).Lycan,W.G.(1996).ConsciousnessandExperience.Cambridge,MA:MITPress.——(ed.)(1990).MindandCognition:AReader.Oxford:Blackwell.Mach,E.(1897[1886]).ErnstMach:ContributionstotheAnalysisofSensations.Chicago:OpenCourt.Malcolm,N.(1958).“KnowledgeofOtherMinds.”TheJournalofPhilosophy,55:969–78.ReprintedinRosenthal(1991).McGinn,C.(1989).“CanWeSolvetheMind–BodyProblem?”Mind,98:349–66.——(1991).TheProblemofConsciousness.Oxford:Blackwell.——(1999).TheMysteriousFlame:ConsciousMindsinaMaterialWorld.NewYork:BasicBooks.McLaughlin,B.(1992).“TheRiseandFallofBritishEmergentism.”InBeckermann(1992).Meehl,P.E.(1966).“TheCompleatAutocerebroscopist:AThought-ExperimentonProfessorFeigl’sMind–BodyIdentityThesis.”InP.K.FeyerabendandG.Maxwell(eds.),Mind,Matter,andMethod:EssaysinPhilosophyandScienceinHonorofHerbertFeigl.Minneapolis:UniversityofMinnesotaPress:103–80.Menzies,P.(1988).“AgainstCausalReductionism.”Mind,97:551–74.Moore,G.E.(1959[1903]).“TheRefutationofIdealism.”PhilosophicalStudies.Paterson,N.J.:Littlefield,AdamsandCo.:1–30.OriginallypublishedinMind,13:433–53.Morgan,C.L.(1923).EmergentEvolution:TheGiffordLectures.London:WilliamsandNorgate.Nagel,T.(1979a).“Panpsychism.”InMortalQuestions.Cambridge:CambridgeUniver-sityPress:181–95.——(1979b).“WhatIsItLiketoBeaBat?”InMortalQuestions.Cambridge:CambridgeUniversityPress:165–80.——(1994).“ConsciousnessandObjectiveReality.”InRichardWarner(ed.),TheMind–BodyProblem:AGuidetotheCurrentDebate.Oxford:Blackwell:63–8.——(1998).“ConceivingtheImpossibleandtheMind–BodyProblem.”Philosophy,73(285):337–52.O’Connor,J.(ed.)(1969).ModernMaterialism:ReadingsonMind–BodyIdentity.NewYork:Harcourt,BraceandWorld.Papineau,D.(1998).“MindtheGap.”PhilosophicalPerspectives,12:373–88.——(1993).PhilosophicalNaturalism.Oxford:Blackwell.Place,U.T.(1956).“IsConsciousnessaBrainProcess?”BritishJournalofPsychology,45:243–55.44\nTheMind–BodyProblem:AnOverviewPlato(1989).TheCollectedDialoguesofPlato.Princeton:PrincetonUniversityPress.Poland,J.S.(1994).Physicalism,thePhilosophicalFoundations.Oxford:ClarendonPress.Popper,K.(1994).KnowledgeandtheBody–MindProblem:InDefenceofInteraction.London:Routledge.Popper,K.,andEccles,J.(1977).TheSelfanditsBrain.NewYork:Springer-Verlag.Prince,M.(1885).TheNatureofMindandHumanAutomatism.Philadelphia:J.B.LippincottCompany.Putnam,H.(1967).“PsychologicalPredicates.”InW.H.CapitanandD.D.Merrill(eds.),Art,MindandReligion.Pittsburgh:PittsburghUniversityPress:37–48.Reprintedunderthetitle“TheNatureofMentalStates”inRosenthal(1991).——(1968).“BrainsandBehavior.”InJ.R.Butter(ed.),AnalyticPhilosophy.Oxford:BasilBlackwell:1–19.ReprintedinRosenthal(1991).——(1975).“TheMeaningof‘Meaning’.”InMind,LanguageandReality:PhilosophicalPapers.Cambridge:CambridgeUniversityPress:215–71.——(1981).Reason,TruthandHistory.Cambridge:CambridgeUniversityPress.Quine,W.V.O.(1953).“TwoDogmasofEmpiricism.”InFromaLogicalPointofView.Cambridge,MA:HarvardUniversityPress:20–46.Rey,G.(1997).ContemporaryPhilosophyofMind:AContentiouslyClassicalApproach.NewYork:Blackwell.Romanes,G.J.(1895).MindandMotionandMonism.NewYork:LongmansGreen.Rorty,R.(1965).“Mind–BodyIdentity,Privacy,andCategories.”TheReviewofMetaphysics,19:24–54.——(1979).PhilosophyandtheMirrorofNature.Princeton:PrincetonUniversityPress.Rosenthal,D.(ed.)(1971).MaterialismandtheMind–BodyProblem.EnglewoodCliffs:Prentice-Hall.——(1991).TheNatureofMentalStates.NewYork:OxfordUniversityPress.Ryle,G.(1949).TheConceptofMind.London:HutchinsonandCompany,Ltd.Russell,B.(1917).“OntheRelationofSense-datatoPhysics.”InMysticismandLogic,andOtherEssays.London:AllenandUnwin:108–31.——(1921).TheAnalysisofMind.London:G.AllenandUnwin.Savellos,E.andYalçin,U.(eds.)(1995).Supervenience:NewEssays.NewYork:CambridgeUniversityPress.Searle,J.R.(1980).“Minds,Brains,andPrograms.”TheBehavioralandBrainSciences,3:417–24.ReprintedinRosenthal(1991).——(1983).Intentionality:AnEssayinthePhilosophyofMind.Cambridge:CambridgeUniversityPress.——(1984).“IntentionalityandItsPlaceinNature.”Dialectica,38:87–100.——(1993).TheRediscoveryoftheMind.Cambridge,MA:MITPress.Siewert,C.(1998).TheSignificanceofConsciousness.Princeton:PrincetonUniversityPress.Skinner,B.F.(1974).AboutBehaviorism.NewYork:Knopf.Smart,J.J.C.(1959).“SensationsandBrainProcesses.”ThePhilosophicalReview,68:141–56.Sperry,R.W.(1986).“Discussion:Macro-VersusMicro-Determination.”PhilosophyofScience,53:265–70.Spinoza,B.(1994).ASpinozaReader:TheEthicsandOtherWorks.Princeton:PrincetonUniversityPress.45\nKirkLudwigStich,S.(1983).FromFolkPsychologytoCognitiveScience:TheCaseAgainstBelief.Cam-bridge,MA:MITPress.Strawson,P.F.(1958).“Persons.”InHerbertFeigl,MichaelScriven,andGroverMaxwell(eds.),MinnesotaStudiesinthePhilosophyofScience2.Minneapolis:UniversityofMinnesotaPress:330–53.Taylor,C.C.W.(1999).“TheAtomists.”A.A.Long(ed.),TheCambridgeCompaniontoEarlyGreekPhilosophy.Cambridge:CambridgeUniversityPress:181–204.Tye,M.(1997).TenProblemsofConsciousness:ARepresentationalTheoryofthePhenomenalMind.Cambridge,MA:MITPress.Vesey,G.N.A.(ed.)(1964).BodyandMind:ReadingsinPhilosophy.London:GeorgeAllenandUnwin.Warner,R.andSzubka,T.(eds.)(1994).TheMind–BodyProblem:AGuidetotheCurrentDebate.Oxford:BasilBlackwell.Watson,J.B.(1925).Behaviorism.NewYork:W.W.Norton.Wilson,E.O.(1998).Consilience:TheUnityofKnowledge(1stedn).NewYork:Knopf.Witmer,D.G.(2000).“SufficiencyClaimsandPhysicalism:AFormulation.”InC.GillettandB.Loewer(eds.),PhysicalismandItsDiscontents.NewYork:CambridgeUniversityPress:57–73.Wittgenstein,L.(1961[1921]).TractatusLogico-Philosophicus.London:RoutledgeandKeganPaulLtd.Firstpublishedin1921inAnnalenderNaturphilosophie.——(1950).PhilosophicalInvestigations.London:Macmillan.Woolhouse,R.S.(1993).Descartes,Spinoza,Leibniz:TheConceptofSubstanceinSeven-teenthCenturyMetaphysics.London:Routledge.46\nTheBlackwellGuidetoPhilosophyofMindTheMind–BodyProblemEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter2TheMind–BodyProblemWilliamG.LycanHumanbeings,andperhapsothercreatures,havemindsaswellasbodies.Butwhatisamind,andwhatisitsrelationtobody,ortothephysicalingeneral?2.1Mind–BodyDualismThefirstanswertothemind–bodyquestionproposedsincemedievaltimeswasthatofDescartes,whoheldthatmindsarewhollydistinctfrombodiesandfromphysicalobjectsofanysort.AccordingtoCartesiandualism,mindsarepurelyspiritualandradicallynon-spatial,havingneithersizenorlocation.Onthisview,anormallivinghumanbeingorpersonisaduality,amindandabodypaired(thoughtherecanbebodieswithoutminds,andmindscansurvivethedestruc-tionoftheircorrespondingbodies).Mysteriously,despitetheradicaldistinctnessofmindsfrombodies,theyinteractcausally:bodilyhappeningscausesensationsandexperiencesandthoughtsinone’smind;conversely,mentalactivityleadstoactionandspeech,causingthephysicalmotionoflimbsorlips.Cartesiandualismhasstrongintuitiveappeal,sincefromtheinsideourmindsdonotfeelphysicalatall;andwecaneasilyimaginetheirexistingdisembodiedor,indeed,theirexistingintheabsenceofanyphysicalworldwhatever.Anduntilthe1950s,infact,thephilosophyofmindwasdominatedbyDescartes’s“first-person”perspective,ourviewofourselvesfromtheinside.Withfewexceptions,philosophershadacceptedthefollowingclaims:(1)thatone’sownmindisbetterknownthanone’sbody,(2)thatthemindismetaphysicallyinthebody’sdriver’sseat,and(3)thatthereisatleastatheoreticalproblemofhowwehumanintelligencescanknowthat“external,”everydayphysicalobjectsexistatall,eveniftherearetenablesolutionstothatproblem.Wehumansubjectsareimmuredwithinamovietheatreofthemind,thoughwemayhavesomedefensiblewaysofinferringwhatgoesonoutsidethetheatre.47\nWilliamG.LycanMidwaythroughthepast(twentieth)century,allthissuddenlychanged,fortworeasons.Thefirstreasonwastheaccumulatedimpactoflogicalpositivismandtheverificationtheoryofmeaning.Intersubjectiveverifiabilityortestabilitybecamethecriterionbothofscientificprobityandoflinguisticmeaningitself.Ifthemind,inparticular,wastoberespectedeitherscientificallyorevenasmean-ingfullydescribableinthefirstplace,mentalascriptionswouldhavetobepeggedtopublicly,physicallytestableverificationconditions.Sciencetakesaninter-subjective,third-personperspectiveoneverything;thetraditionalfirst-personpers-pectivehadtobeabandonedforscientificpurposesand,itwasfelt,forseriousmetaphysicalpurposesalso.ThesecondreasonwastheemergenceofanumberofpressingphilosophicalobjectionstoCartesiandualism,suchasthefollowing:1ImmaterialCartesianmindsandghostlynon-physicaleventswereincreasinglyseentofitillwithourotherwisephysicalandscientificpictureoftheworld,uncomfortablylikespooksorectoplasmthemselves.Theyarenotneededfortheexplanationofanypubliclyobservablefact,forneurophysiologypromisestoexplainthemotionsofourbodiesinparticularandtoexplainthemcom-pletely.Indeed,ghost-mindscouldnotverywellhelpinsuchanexplanation,sincenothingisknownofanypropertiesofspookstuffthatwouldbearonpublicphysicaloccurrences.2Sincehumanbeingsevolvedoveraeons,bypurelyphysicalprocessesofmuta-tionandnaturalselection,fromprimitivecreaturessuchasone-celledorgan-ismswhichdidnothaveminds,itisanomaloustosupposethatatsomepointMotherNature(intheformofpopulationgenetics)somehowcreatedimma-terialCartesianmindsinadditiontocellsandphysicalorgans.Thesamepointcanbeputintermsofthedevelopmentofasinglehumanzygoteintoanembryo,thenafetus,ababy,andfinallyachild.3Ifmindsreallyareimmaterialandutterlynon-spatial,howcantheypossiblyinteractcausallywithphysicalobjectsinspace?(Descarteshimselfwasveryuncomfortableaboutthis.Atonepointhesuggestedgravityasamodelfortheactionofsomethingimmaterialonaphysicalbody;butgravityisspatialinnatureeventhoughitisnottangibleinthewaythatbodiesare.)4Inanycaseitdoesnotseemthatimmaterialentitiescouldcausephysicalmotionconsistentlywiththeconservationlawsofphysics,suchasthoseregardingmotionandmatter-energy;physicalenergywouldhavetovanishandreappearinsidehumanbrains.2.2BehaviorismWhatalternativesaretheretodualism?First,Carnap(1932–3)andRyle(1949)notedthattheobviousverificationconditionsortestsformentalascriptionsare48\nTheMind–BodyProblembehavioral.Howcantherestofustellthatyouareinpain,savebyyourwincingandgroaningbehaviorincircumstancesofpresumabledamageordisorder,orthatyoubelievethatparsnipsaredangerous,savebyyourverbalavowalsandyouravoidanceofparsnips?Ifthetestsarebehavioral,then(itwasargued)theverymeaningsoftheascriptions,oratleasttheonlyfactsgenuinelydescribed,arenotghostlyorineffablebutbehavioral.Thusbehaviorismasatheoryofmindandaparadigmforpsychology.Inacademicpsychology,behaviorismtookprimarilyamethodologicalform,andthepsychologistsofficiallymadenometaphysicalclaims.Butinphilosophy,behaviorismdid(naturally)takeametaphysicalform:chieflythatofanalyticalbehaviorism,theclaimthatmentalascriptionssimplymeanthingsaboutbehavioralresponsestoenvironmentalimpingements.Thus,“Leoisinpain”means,notanythingaboutLeo’sputativeghostlyego,orevenaboutanyepisodetakingplacewithinLeo,butthateitherLeoisactuallybehavinginawincingandgroaningwayorheisdisposedsotobehave(inthathewouldsobehaveweresomethingnotkeepinghimfromdoingso).“Leobelievesthatparsnipsaredangerous”meansjustthat,ifasked,Leowouldassenttothatproposition,and,ifconfrontedbyaparsnip,Leowouldshunit,andsoforth.AnybehavioristwillsubscribetowhathascometobecalledtheTuringTest.Inresponsetotheperenniallypopularquestion“Canmachinesthink?”,AlanTuring(1964)repliedthatabetterquestionisthatofwhetherasophisticatedcomputercouldeverpassabatteryofverbaltests,totheextentoffoolingalimitedobserver(say,ahumanbeingcorrespondingwithitbymail)intothinkingitishumanandsentient.Ifamachinedidpasssuchtests,thentheputativelyfurtherquestionofwhetherthemachinereallythoughtwouldbeidleatbest,whatevermetaphysicalanalysisonemightattachtoit.BarringTuring’stenden-tiouslimitationofthemachine’sbehaviortoverbalasopposedtonon-verbalresponses,anybehaviorist,psychologicalorphilosophical,wouldagreethatpsy-chologicaldifferencescannotoutrunbehavioraltests;organisms(includingmachines)whoseactualandhypotheticalbehaviorisjustthesamearepsychologic-allyjustalike.Besidessolvingthemethodologicalproblemofintersubjectiveverification,philo-sophicalbehaviorismalsoadroitlyavoidedanumberoftheobjectionstoCartesiandualism,includingallof(1)–(4)listedabove.ItdispensedwithimmaterialCartesianegosandghostlynon-physicalevents,writingthemoffasmetaphysicalexcres-cences.ItdisposedofDescartes’sadmittedproblemofmind–bodyinteraction,sinceitpositednoimmaterial,non-spatialcausesofbehavior.Itraisednoscien-tificmysteriesconcerningtheinterventionofCartesiansubstancesinphysicsorbiology,sinceitcountenancednosuchintervention.Thusitisamaterialistview,asagainstDescartes’simmaterialism.Yetsometheoristswereuneasy;theyfeltthatinitstotalrepudiationoftheinner,theprivate,andthesubjective,behaviorismwasleavingoutsomethingrealandimportant.Whenthisworrywasvoiced,thebehavioristsoftenrepliedwithmockery,assimilatingthedoubterstoold-fashioneddualistswhobelievedin49\nWilliamG.Lycanghosts,ectoplasm,ortheEasterbunny;behaviorismwastheonly(evenhalfwaysensible)gameintown.Nonetheless,thedoubtersmadeseverallastingpointsagainstit.First,peoplewhoarehonestandnotanesthetizedknowperfectlywellthattheyexperience,andcanintrospect,actualinnermentalepisodesoroccur-rences,thatareneitheractuallyaccompaniedbycharacteristicbehaviornormerelystatichypotheticalfactsofhowtheywouldbehaveifsubjectedtosuch-and-suchastimulation.Place(1956)spokeofan“intractableresidue”ofconsciousmentalstatesthatbearnoclearrelationstobehaviorofanyparticularsort;seealsoArmstrong(1968:ch.5)andCampbell(1984).Secondly,contrarytotheTuringTest,itseemsperfectlypossiblefortwopeopletodifferpsychologicallydespitetotalsimilarityoftheiractualandhypotheticalbehavior,asinacaseof“invertedspectrum”ashypothesizedbyJohnLocke:itmightbethatwhenyouseearedobject,youhavethesortofcolorexperiencethatIhavewhenIseeagreenobject,andviceversa.Forthatmatter,acreaturemightexhibitalltheappropriatestimulus-responserelationsandlackamentallifeentirely;wecanimaginebuild-inga“zombie”orstupidrobotthatbehavesintherightwaysbutdoesnotreallyfeelorthinkanythingatall(BlockandFodor1972;Kirk1974;Block1981;Campbell1984).Thirdly,theanalyticalbehaviorist’sbehavioralanalysesofmen-talascriptionsseemadequateonlysolongasonemakessubstantiveassumptionsabouttherestofthesubject’smentality(Chisholm1957:ch.11;Geach1957:8;Block1981);forexample,ifLeobelievesthatparsnipsaredangerousandheisofferedparsnips,hewouldshunthemonlyifhedoesnotwanttodie.Therefore,thebehavioristanalysesareeithercircularorradicallyincomplete,sofarastheyaresupposedtoexhaustthementalgenerally.Somattersstoodinstalematebetweendualists,behaviorists,anddoubters,untilthelate1950s,whenU.T.Place(1956)andJ.J.C.Smart(1959)pro-posedamiddleway,aconciliatorycompromisesolution.2.3TheIdentityTheoryAccordingtoPlaceandSmart,contrarytothebehaviorists,atleastsomementalstatesandeventsaregenuinelyinnerandgenuinelyepisodicafterall.Theyarenottobeidentifiedwithoutwardbehaviororevenwithhypotheticaldispositionstobehave.But,contrarytothedualists,theepisodicmentalitemsareneitherghostlynornon-physical.Rather,theyareneurophysiological.Theyareidenticalwithstatesandeventsoccurringintheirowners’centralnervoussystems;moreprecisely,everymentalstateoreventisnumericallyidenticalwithsomesuchneurophysiologicalstateorevent.Tobeinpainis,forexample,tohaveone’sc-fibers,ormorelikelya-fibers,firinginthecentralnervoussystem;tobelievethatbroccoliwillkillyouistohaveone’sBbk-fibersfiring,andsoon.Bymakingthementalentirelyphysical,thisidentitytheoryofthemindsharedthebehavioristadvantageofavoidingtheobjectionstodualism.Butitalso50\nTheMind–BodyProblembrilliantlyaccommodatedtheinnerandtheepisodicasbehaviorismdidnot.For,accordingtotheidentitytheory,mentalstatesandeventsactuallyoccurintheirowners’centralnervoussystems.(HencetheyareinnerinanevenmoreliteralsensethancouldbegrantedbyDescartes.)Theidentitytheoryalsothoroughlyvin-dicatedtheideathatorganismscandiffermentallydespitetotaloutwardbehavioralsimilarity,sinceclearlyorganismscandifferneurophysiologicallyinmediatingtheiroutwardstimulus-responseregularities;thatwouldaffordthepossibilityofinvertedspectrum.Andofcoursetheconnectionbetweenabelieforadesireandtheusuallyaccompanyingbehaviorisdefeasiblebyothercurrentmentalstates,sincetheconnectionbetweenaB-orD-neuralstateanditsnormalbehavioraleffectisdefeasiblebyotherpsychologicallycharacterizableinteractingneuralstates.Theidentitytheorywastheidealresolutionofthedualist–behavioristimpasse.Moreover,therewasadirectdeductiveargumentfortheidentitytheory,hituponindependentlybyDavidLewis(1966,1972)andD.M.Armstrong(1968).LewisandArmstrongmaintainedthatmentaltermsweredefinedcausally,intermsofmentalitems’typicalcausesandeffects.Forinstance,theword“pain”meansastatethatistypicallybroughtaboutbyphysicaldamageandthattypicallycauseswithdrawal,favoring,complaint,desireforcessation,andsoon.(Armstrongclaimedtoestablishthisbystraightforward“conceptualanalysis.”Moreelabor-ately,Lewisheldthatmentaltermsarethetheoreticaltermsofacommon-sensical“folktheory,”andwiththepositiviststhatalltheoreticaltermsareimplicitlydefinedbythetheoriesinwhichtheyoccur.Thatcommon-sensetheoryhassincecometobecalled“folkpsychology.”)Nowif,bydefinition,painiswhateverstateoccupiesacertaincausalniche,andif,asisoverwhelminglylikely,scientificresearchwillrevealthatthatparticularnicheisinfactoccupiedbysuch-and-suchaneurophysiologicalstate,itfollowsstraightawaythatpainisthatneurophysiologicalstate;QED.Painretainsitsconceptualconnectiontobehavior,butalsounder-goesanempiricalidentificationwithaninnerstateofitsowner.(AnadvancedifconvolutedelaborationofthisalreadyhybridviewisdevelopedbyLewis1980;formeticulousdiscussion,seeBlock1978;Shoemaker1981;Tye1983;Owens1986.)NoticethatalthoughArmstrongandLewisbegantheirargumentswithaclaimaboutthemeaningsofmentalterms,their“common-sensecausal”versionoftheidentitytheorywasitselfnosuchclaim,anymorethanwastheoriginalidentitytheoryofPlaceandSmart.Rather,allfourphilosophersreliedontheideathatthingsorpropertiescansometimesbeidentifiedwith“other”thingsorpropertiesevenwhenthereisnosynonymyofterms;thereissuchathingassyntheticandaposterioriidentitythatisnonethelessgenuineidentity.Whiletheidentityoftriangleswithtrilateralsholdssimplyinvirtueofthemeaningsofthetwotermsandcanbeestablishedbyreasonalone,withoutempiricalinvestigation,thefollowingidentitiesarestandardexamplesofthesyntheticaposteriori,andwerediscoveredempirically:cloudswithmassesofwaterdroplets;waterwithH2O;lightningwithelectricaldischarge;theMorningStarwithVenus;MendeliangeneswithsegmentsofDNAmolecules;andtemperaturewithmeanmolecular51\nWilliamG.Lycankineticenergy.Theidentitytheorywasofferedsimilarly,inaspiritofscientificspeculation;onecouldnotproperlyobjectthatmentalexpressionsdonotmeananythingaboutbrainsorneuralfirings.Sothedualistswerewronginthinkingthatmentalitemsarenon-physicalbutrightinthinkingtheminnerandepisodic;thebehavioristswererightintheirmaterialismbutwrongtorepudiateinnermentalepisodes.Adelightfulsynthesis.Butalas,itwastoogoodtobetrue.2.4MachineFunctionalismQuitesoon,HilaryPutnam(1960,1967a,1967b)andJerryFodor(1968b)pointedoutapresumptuousimplicationoftheidentitytheoryunderstoodasatheoryof“types”orkindsofmentalitem:thatamentalstatesuchaspainhasalwaysandeverywheretheneurophysiologicalcharacterizationinitiallyassignedtoit.Forexample,iftheidentitytheoristidentifiedpainitselfwiththefiringsofc-fibers,itfollowedthatacreatureofanyspecies(earthlyorscience-fiction)couldbeinpainonlyifthatcreaturehadc-fibersandtheywerefiring.Butsuchaconstraintonthebiologyofanybeingcapableoffeelingpainisbothgratuitousandindefensible;whyshouldwesupposethatanyorganismmustbemadeofthesamechemicalmaterialsasweareinordertohavewhatcanbeaccuratelyrecog-nizedaspain?Theidentitytheoristhadoverreactedtothebehaviourists’difficul-tiesandfocusedtoonarrowlyonthespecificsofbiologicalhumans’actualinnerstates,andinsodoingtheyhadfallenintospecieschauvinism.PutnamandFodoradvocatedtheobviouscorrection:whatwasimportantwasnotitsbeingc-fibers(perse)thatwerefiring,butwhatthec-fiberfiringsweredoing,whattheycontributedtotheoperationoftheorganismasawhole.Theroleofthec-fiberscouldhavebeenperformedbyanymechanicallysuitablecomponent;solongasthatrolewasperformed,thepsychologyofthecontainingorganismwouldhavebeenunaffected.Thus,tobeinpainisnotpersetohavec-fibersthatarefiring,butmerelytobeinsomestateorother,ofwhateverbiochemicaldescription,thatplaysthesamecausalroleasdidthefiringsofc-fibersinthehumanbeingswehaveinvestigated.Wemaycontinuetomaintainthatpain“tokens”(individualinstancesofpainoccurringinparticularsubjectsatparticulartimes)arestrictlyidenticalwithparticularneurophysiologicalstatesofthosesubjectsatthosetimes–inotherwords,withthestatesthathappentobeplayingtheappropriateroles;thisisthethesisoftokenidentityor“token”mater-ialismorphysicalism.Butpainitself,thekind,universal,or“type,”canbeidentifiedonlywithsomethingmoreabstract:thecausalorfunctionalrolethatc-fiberfiringssharewiththeirpotentialreplacementsorsurrogates.Mentalstate-typesareidentifiednotwithneurophysiologicaltypesbutwithmoreabstractfunctionalroles,asspecifiedbystate-tokens’causalrelationstotheorganism’ssensoryinputs,behavioralresponses,andotherinterveningpsychologicalstates.52\nTheMind–BodyProblemFunctionalism,then,isthedoctrinethatwhatmakesamentalstatethetypeofstateitis–apain,asmellofviolets,abeliefthatkoalasarevenomous–isitsdistinctivesetoffunctionalrelations,itsroleinitssubject’sbehavioraleconomy.Putnamcomparedmentalstatestothefunctionalor“logical”statesofacom-puter:justasacomputerprogramcanberealizedorinstantiatedbyanyofanumberofphysicallydifferenthardwareconfigurations,socanapsychological“program”berealizedbydifferentorganismsofvariousphysiochemicalcomposi-tion,andthatiswhydifferentphysiologicalstatesoforganismsofdifferentspe-ciescanrealizeoneandthesamementalstate-type.Whereanidentitytheorist’stype-identificationwouldtaketheform,“TobeinmentalstateoftypeMistobeintheneurophysiologicalstateoftypeN,”Putnam’smachinefunctionalism,asIshallcallit,assertsthattobeinMistobemerelyinsomephysiologicalstateorotherthatplaysroleRintherelevantcomputerprogram(thatis,theprogramthatatasuitablelevelofabstractionmediatesthecreature’stotaloutputsgiventotalinputsandsoservesasthecreature’sglobalpsychology).Thephysiologicalstate“playsroleR”inthatitstandsinasetofrelationstophysicalinputs,outputs,andotherinnerstatesthatmatchesone-to-onetheabstractinput–output–logical-staterelationscodifiedinthecomputerprogram.Thefunctionalist,then,mobilizesthreedistinctlevelsofdescriptionbutappliesthemalltothesamefundamentalreality.Aphysicalstate-tokeninsomeone’sbrainataparticulartimehasaneurophysiologicaldescription,butitmayalsohaveafunctionaldescriptionrelativetoamachineprogramthatthebrainhappenstoberealizing,anditmayfurtherhaveamentaldescriptionifsomementalstateiscorrectlytype-identifiedwiththefunctionalcategoryitexemplifies.Andsothereisafterallasenseinwhich“themental”isdistinctfrom“thephysical.”Though,presumably,therearenonon-physicalsubstancesorstuffs,andeverymentaltokenisitselfentirelyphysical,mentalcharacterizationisnotphysicalcharacterization,andthepropertyofbeingapainisnotsimplythepropertyofbeingsuch-and-suchaneuralfiring.Moreover,unlikebehaviorismandtheidentitytheory,func-tionalismdoesnotstrictlyentailthatmindsarephysical;itmightbetrueofnon-physicalminds,solongasthosemindsrealizedtherelevantprograms.2.5HomuncularFunctionalismandOtherTeleologicalTheoriesMachinefunctionalismhasbeenchallengedonanumberofpoints,whichtogethermotivateaspecificallyteleologicalnotionof“function”:wearetothinkofathing’sfunctionaswhatthethingisfor,whatitsjobis,whatitissupposedtodo.Herearethreereasonsforthus“puttingthefunctionbackintofunctionalism”(Sober1985).First,themachinefunctionaliststillconceivedpsychologicalexplanationinthelogicalpositivists’termsofsubsumingobserveddataunderwiderandwider53\nWilliamG.Lycanuniversallaws.ButFodor(1968a),Dennett(1978),andCummins(1983)havedefendedacompetingpictureofpsychologicalexplanation,accordingtowhichbehavioraldataaretobeseenasmanifestationsofsubjects’psychologicalcapacit-ies,andthosecapacitiesaretobeexplainedbyunderstandingthesubjectsassystemsofinterconnectedcomponents.Eachcomponentisa“homunculus,”inthatitisthoughtofasalittleagentorbureaucratoperatingwithinitscontainingsubject;itisidentifiedbyreferencetothefunctionitperforms.Andthevarioushomuncularcomponentscooperatewitheachotherinsuchawayastoproduceoverallbehavioralresponsestostimuli.The“homunculi”arethemselvesbrokendownintosubcomponentswhosefunctionsandinteractionsaresimilarlyusedtoexplainthecapacitiesofthesubsystemstheycompose,andsoagainandagainuntilthesub-sub-...componentsareseentobeneurophysiologicalstructures.Thusbiologicalandmechanicalsystemsalikearehierarchicallyorganized.(Anautomobileworks–locomotes–byhavingafuelreservoir,afuelline,acarburetor,acombustionchamber,anignitionsystem,atransmission,andwheelsthatturn.Ifonewantstoknowhowthecarburetorworks,onewillbetoldwhatitspartsareandhowtheyworktogethertoinfuseoxygenintofuel;andsoon.)Butnothinginthispatternofexplanationcorrespondstothesubsumptionofdataunderwiderandwideruniversalgeneralizations.Thesecondreasonisthatthemachinefunctionalisttreatedfunctional“realiza-tion,”therelationbetweenanindividualphysicalorganismandtheabstractprogramitwassaidtoinstantiate,asasimplematterofone-to-onecorrespond-encebetweentheorganism’srepertoireofphysicalstimuli,structuralstates,andbehavior,ontheonehand,andtheprogram’sdefininginput–state–outputfunc-tionontheother.Butthiscriterionofrealizationwasseentobetooliberal;sincevirtuallyanythingbearsaone–onecorrelationofsomesorttovirtuallyanythingelse,“realization”inthesenseofmereone–onecorrespondenceisfartooeasilycomeby(Block1978;Lycan1987:ch.3);anymiddle-sizedphysicalobjecthassomesetofcomponentmolecularmotionsthathappentocorrespondone–onetoagivenmachineprogram.Sometheoristshaveproposedtoremedythisdefectbyimposingateleologicalrequirementonrealization:aphysicalstateofanorganismwillcountasrealizingsuch-and-suchafunctionaldescriptiononlyiftheorganismhasgenuineorganicintegrityandthestateplaysitsfunctionalroleproperlyfortheorganism,intheteleologicalsenseof“for”andintheteleolo-gicalsenseof“function.”Thestatemustdowhatitdoesasamatterof,sotospeak,itsbiologicalpurpose.(Machinefunctionalismtook“function”initssparemathematicalsenseratherthaninagenuinelyfunctionalsense.Oneshouldnotethat,asusedhere,theterm“machinefunctionalism”istiedtotheoriginalliberalconceptionof“realizing;”sotoimposeateleologicalrestrictionistoabandonmachinefunctionalism.)Thirdly,VanGulick(1980),Millikan(1984),Dretske(1988),Fodor(1990a),andothershavearguedpowerfullythatteleologymustenterintoanyadequateanalysisoftheintentionalityoraboutnessorreferentialcharacterofmentalstatessuchasbeliefsanddesires,byreferencetothestates’psychobiologicalfunctions.54\nTheMind–BodyProblemBeliefs,desires,andotherpropositionalattitudessuchassuspecting,intending,andwishingaredirecteduponstatesofaffairswhichmayormaynotactuallyobtain(forinstance,thattheRepublicancandidatewillwin),andareaboutindividualswhomayormaynotexist(suchasKingArthurorSherlockHolmes).FranzBrentano(1973[1874])drewadistinctionbetweenpsychologicalphe-nomena,whicharedirecteduponobjectsandstatesofaffairs,evennon-existingones,andphysicalobjects,whicharenotsodirected.Ifmentalitemsarephysical,however,thequestionariseshowanypurelyphysicalentityorstatecouldhavethepropertyofbeing“directedupon”oraboutanon-existentstateofaffairsorobject;thatisnotthesortoffeaturethatordinary,purelyphysicalobjects(suchasbricks)canhave.Accordingtotheteleologicaltheorists,aneurophysiologicalstateshouldcountasabeliefthatbroccoliwillkillyou,andinparticularasaboutbroccoli,onlyifthatstatehastherepresentingofbroccoliasinsomesenseoneofitspsychobiologicalfunctions.Ifteleologyisneededtoexplicateintentionality,andmachinefunctionalismaffordsnoteleology,thenmachinefunctionalismisnotadequatetoexplicateintentionality.Allthistalkofteleologyandbiologicalfunctionseemstopresupposethatbiologicalandother“structural”statesofphysicalsystemsreallydohavefunc-tionsintheteleologicalsense.Thelatterclaimis,tosaytheleast,controversial.But,fortunatelyfortheteleologicalfunctionalist,thereisavigorousindustrywhosepurposeistoexplicatebiologicalteleologyinnaturalisticterms,typicallyintermsofetiology.Forexample,atraitmaybesaidtohavethefunctionofdoingFinvirtueofitshavingbeenselectedbecauseitdidF;aheart’sfunctionistopumpbloodbecausehearts’pumpingbloodinthepasthasgiventhemaselec-tionadvantageandsoledtothesurvivalofmoreanimalswithhearts(Wright1973;Millikan1984).Functionalisminheritssomeofthesamedifficultiesthatearlierbesetbehaviorismandtheidentitytheory.Theseremainingobstaclesfallintotwomaincategories:qualiaproblemsandintentionalityproblems.2.6ProblemsoverQualiaandConsciousnessThequaleofamentalstateorevent(particularlyasensation)isthatstateorevent’sfeel,itsintrospectible“phenomenalcharacter,”itsnatureasitpresentsitselftoconsciousness.Manyphilosophershaveobjectedthatneitherfunctionalistmeta-physicsnoranyoftheallieddoctrinesaforementionedcan“explainconsciousness,”orilluminateoreventoleratethenotionofwhatitfeelsliketobeinamentalstateofsuch-and-suchasort.Yet,saythesephilosophers,thefeelsarequintessentiallymental–itisthefeelsthatmakethementalstatesthementalstatestheyare.Something,therefore,mustbedrasticallywrongwithfunctionalism.“The”problemofconsciousnessorqualiaisfamiliar.Indeed,itissofamiliarthatwetendtooverlookthemostimportantthingaboutit:thatitsnameis55\nWilliamG.Lycanlegion,foritismany.Thereisnosingleproblemofqualia;thereareatleastelevenquitedistinctobjectionsthathavebeenbroughtagainstfunctionalism(someofthemapplytomaterialistviewsgenerally).Tomentionafew:1Block(1978)andothershaveurgedvarious“zombie”-stylecounterexamplecasesagainstfunctionalism–examplesinwhichsomeentityseemstorealizetherightprogrambutwhichlacksoneofmentality’scrucialqualitativeaspects.(Typicallythe“entity”isagroupofhumanbeings,suchastheentirepopula-tionofChinaactingaccordingtoanelaboratesetofinstructions.Itdoesnotseemthatsuchagroupofindividualswouldcollectivelybefeelinganything.)Predictably,functionalistshaverejoinedbyarguing,foreachexample,eitherthattheproposedgroupentitydoesnotinfactsucceedinrealizingtherightprogram(forexample,becausetherequisiteteleologyislacking)orthatthereisnogoodreasonfordenyingthattheentitydoeshavetherelevantqualita-tivestates.2Nagel(1974)andJackson(1982)haveappealedtoadisparityinknowledge,asageneralanti-materialistargument:Icanknowwhatitisliketohavesuch-and-suchasensationonlyifIhavehadthatsensationmyself;noamountofobjective,third-personscientificinformationwouldsuffice.Inreply,function-alistshaveofferedanalysesof“perspectivalness,”completewithaccountsof“whatitislike”tohaveasensation,thatmakethosethingscompatiblewithfunctionalism.NagelandJacksonhaveargued,further,fortheexistenceofaspecial,intrinsicallyperspectivalkindoffact,thefactof“whatitislike”,whichintractablyandinprinciplecannotbecapturedorexplainedbyphysicalscience.Functionalistshaverespondedthattheargumentscommitalogicalfallacy(specifically,thatofapplyingLeibniz’sLawinanintensionalcontext);somehaveaddedthatinanycase,to“knowwhatitislike”ismerelytohaveanability,andinvolvesnofactofanysort,while,contrariwise,someothertheoristshavegrantedthattherearefactsof“whatitislike”butinsistedthatsuchfactscanafterallbeexplainedandpredictedbynaturalscience.3SaulKripke(1972)madeingenioususeofmodaldistinctionsagainsttypeoreventokenidentity,arguingthatunlessmentalitemsarenecessarilyidenticalwithneurophysiologicalones,whichtheyarenot,theycannotbeidenticalwiththematall.Kripke’sclosereasoninghasattractedconsiderablecriticalattention.Andevenmoresophisticatedvariantshavebeenoffered,e.g.,byJackson(1993)andChalmers(1996).4Jackson(1977)andothershavedefendedtheclaimthatinconsciousnesswearepresentedwithmentalindividualsthatthemselvesbearphenomenal,quali-tativeproperties.Forinstance,whenaredflashbulbgoesoffinyourface,yourvisualfieldexhibitsagreenblotch,an“after-image,”athingthatisreallygreenandhasafairlydefiniteshapeandexistsforafewsecondsbeforedisappearing.Iftherearesuchthings,theyareentirelydifferentfromanythingphysicaltobefoundinthebrainofa(healthy)humansubject.Beliefinsuch“phenomenalindividuals”asgenuinelygreenafter-imageshasbeenunpopular56\nTheMind–BodyProblemamongphilosophersforsomeyears,butitcanbepowerfullymotivated(seeLycan1987:83–93).Thisisaformidablequartetofobjections,and,onthefaceofit,eachisplausible.Materialistsandparticularlyfunctionalistsmustrespondindetail.Need-lesstosay,materialistshaverespondedatlength;someofthemostpowerfulrejoindersareformulatedinLycan(1987,1996).Yetrecentyearshaveseensomereactionagainsttheprevailingmaterialism,includingare-emergenceofsomeneo-dualistviews,asinRobinson(1988),Hart(1988),Strawson(1994),andChalmers(1996).2.7ProblemsoverIntentionalityTheproblemarisingfromourmentionofBrentanowastoexplainhowanypurelyphysicalentityorstatecouldhavethepropertyofbeingaboutor“directedupon”anon-existentstateofaffairs.ThestandardfunctionalistreplyisthatpropositionalattitudeshaveBrentano’sfeaturebecausetheinternalphysicalstatesandeventsthatrealizethemrepresentactualorpossiblestatesofaffairs.Whattheyrepresent(theircontent)isdeterminedatleastinpartbytheirfunctionalroles.Therearetwomaindifficulties.Oneisthatofsayingexactlyhowaphysicalitem’ssupposedrepresentationalcontentisdetermined;invirtueofwhatdoesaneurophysiologicalstaterepresentpreciselythattheRepublicancandidatewillwin?AnanswertothatgeneralquestioniswhatFodorhascalledapsychosemantics.Severalattemptshavebeenmade(Dretske1981;Millikan1984;Fodor1987,1990a,1990b,1994),butnoneisveryplausible.Inparticular,noneappliestoanycontentbutthatwhichinvolvesactualandpresentlyexistingphysicalobjects.Abstractentitiessuchasnumbers,futureentitiessuchasachildIhopeonedaytohave,andBrentano’snon-existentitems,arejustleftout.Theseconddifficultyisthatordinarypropositionalattitudecontentsdonotsuperveneonthestatesoftheirsubjects’nervoussystems,butareunderdeterminedbyeventhetotalstateofthatsubject’shead.Putnam’s(1975)TwinEarthandindexicalexamplesshowthat,surprisingasitmayseem,twohumanbeingscouldbemolecule-for-moleculealikeandstilldifferintheirbeliefsanddesires,depend-ingonvariousfactorsintheirspatialandhistoricalenvironments.Thuswecandistinguishbetween“narrow”properties,thosethataredeterminedbyasubject’sintrinsicphysicalcomposition,and“wide”properties,thosethatarenotsodeter-mined.Representationalcontentsarewide,yetfunctionalrolesare,ostensibly,narrow.How,then,canpropositionalattitudesbetype-identifiedwithfunctionalroles,orforthatmatterwithstatesofthebrainunderanynarrowdescription?Functionalistshaverespondedineitheroftwowaystotheseconddifficulty.Thefirstistounderstand“function”widelyaswell,specifyingfunctionalroleshistoricallyand/orbyreferencetofeaturesofthesubject’sactualenvironment.57\nWilliamG.LycanThesecondissimplytoabandonfunctionalismasanaccountofcontentinparticular,givingsomealternativepsychosemanticsforpropositionalattitudes,butpreservingfunctionalisminregardtoattitudetypes.(ThuswhatmakesastateadesirethatPisitsfunctionalrole,evenifsomethingelsemakesthestateadesirethatP).2.8TheEmotionsInalludingtosensorystatesandtomentalstateswithintentionalcontent,wehavesaidnothingspecificallyabouttheemotions.Sincetherejectionofbehavi-orism,theoriesofmindhavetendednottobeapplieddirectlytotheemotions;rather,theemotionshavebeengenerallythoughttobeconceptuallyanalyzableascomplexesofmorecentralor“core”mentalstates,typicallypropositionalattitudessuchasbeliefanddesire(andtheintentionalityofemotionshasaccordinglybeentracedbacktothatofattitudes).Armstrong(1968:ch.8,secnIII)essentiallytookthisline,asdoSolomon(1977)andGordon(1987).However,thereisaliteratureonfunctionalismandtheemotions;seeRey(1980)andsomeoftheotherpaperscollectedinRorty(1980).Griffiths(1997)takesagenerallyfunctionalistview,butarguesthat“theemotions”donotconstituteasinglekind.2.9InstrumentalismTheidentitytheoristsandthefunctionalists,machineorteleological,joinedcom-monsense(andcurrentcognitivepsychology)inunderstandingmentalstatesandeventsbothasinternaltohumansubjectsandascauses.Beliefsanddesiresinparticulararethoughttobecausedbyperceptualorothercognitiveeventsandasinturnconspiringfromwithintocausebehavior.IfArmstrong’sorLewis’stheoryofmindiscorrect,thisideaisnotonlycommon-sensicalbutaconceptualtruth;iffunctionalismiscorrect,itisatleastametaphysicalfact.Inrallyingtotheinner-causalstory,aswesawinsection2.3,theidentitytheoristsandfunctionalistsbrokewiththebehaviorists,forbehavioristsdidnotthinkofmentalitemsasentities,asinner,orascausesinanystrongersensethanthebarehypothetical.Behavioristseitherdispensedwiththementalisticidiomaltogether,orparaphrasedmentalascriptionsintermsofputativeresponsestohypotheticalstimuli.Morerecently,otherphilosophershavefollowedtheminrejectingtheideaofbeliefsanddesiresasinnercausesandinconstruingtheminamorepurelyoperationalorinstrumentalfashion.D.C.Dennett(1978,1987)hasbeenparticularlyconcernedtodenythatbeliefsanddesiresarecausallyactiveinnerstatesofpeople,andmaintainsinsteadthatbelief-ascriptionsanddesire-ascriptionsaremerelycalculationaldevices,whichhappentohavepredictive58\nTheMind–BodyProblemusefulnessforareasonthathegoesontoexplain.Suchascriptionsareoftenobjectivelytrue,hegrants,butnotinvirtueofdescribinginnermechanisms.ThusDennettisaninstrumentalistaboutpropositionalattitudessuchasbeliefanddesire.(Accordingtoacontemporaryinterpretation,an“instrumentalist”aboutXsisatheoristwhoclaimsthatalthoughsentencesabout“Xs”areoftentrue,theydonotreallydescribeentitiesofaspecialkind,butonlyservetosystematizemorefamiliarphenomena.Forinstance,weareallinstrumentalistsabout“theaverageAmericanhomeowner,”whoiswhite,male,andthefatherofexactly2.2children.)Toascribea“belief”ora“desire”isnottodescribesomesegmentofphysicalreality,Dennettsays,butismorelikemovingagroupofbeadsinanabacus.(ItshouldbenotedthatDennetthasmorerecentlymoder-atedhisline:see1991.)Dennettoffersbasicallyfourgroundsforhisrejectionofthecommon-sensicalinner-causethesis:1Hethinksitquiteunlikelythatanysciencewilleverturnupanydistinctiveinner-causalmechanismthatwouldbesharedbyallthepossiblesubjectsthathadaparticularbelief.2Hecomparesthebelief-desireinterpretationofhumanbeingstothatofloweranimals,chess-playingcomputers,andevenlightning-rods,arguingthat(a)intheircasewehavenoreasontothinkofbelief-ascriptionsanddesire-ascriptionsasotherthanmerecalculational-predictivedevicesand(b)wehavenomorereasonforthecaseofhumanstothinkofbelief-ascriptionsanddesire-ascriptionsasotherthanthat.3Dennettarguesfromtheverificationconditionsofbelief-ascriptionsanddesire-ascriptions–basicallyamatterofextrapolatingrationallyfromwhatasubjectoughttobelieveandwantinhisorhercircumstances–andthenheboldlyjustidentifiesthetruth-makersofthoseascriptionswiththeirverifica-tionconditions,challenginginner-causetheoriststoshowwhyinstrumentalismdoesnotaccommodatealltheactualevidence.4Hearguesthatinanycase,ifapurelynormativeassumption(the“rationalityassumption,”whichisthatpeoplewillgenerallybelievewhattheyoughttobelieveanddesirewhattheyshoulddesire)isrequiredforthelicensingofanascription,thentheascriptioncannotitselfbeapurelyfactualdescriptionofaplainstateofaffairs.Stich(1981)exploresandcriticizesDennett’sinstrumentalismatlength(per-hapsoddly,Stich(1983)goesontodefendaviewnearlyasdeprecatingasDennett’s,thoughclearlydistinctfromit).Dennett(1981)respondstoStich,bringingoutmoreclearlytheforceofthe“rationalityassumption”assumption.(OthercriticismsarelevelledagainstDennettbycommentatorsintheBehavioralandBrainSciencessymposiumthatisheadedbyDennett1988.)AclosecousinofDennett’sview,inthatitfocusesontherationalityassumption,isDonaldDavidson’s(1970)anomalousmonism.UnlikeDennett’sinstrumentalism,59\nWilliamG.Lycanitendorsestokenphysicalismandinsiststhatindividualmentaltokensarecauses,butitrejectsonsimilarlyepistemologicalgroundsthepossibilityofanyinterest-ingmaterialistictype-reductionofthepropositionalattitudes.2.10EliminativismandNeurophilosophyDennett’sinstrumentalismbreakswithcommonsenseandwithphilosophicaltraditionindenyingthatpropositionalattitudessuchasbeliefanddesirearerealinner-causalstatesofpeople.ButDennettconcedes–indeed,heurgentlyinsists–thatbelief-ascriptionsanddesire-ascriptionsaretrue,andobjectivelytrue,nonetheless.Otherphilosophershavetakenalessconciliatory,moreradicallyuncommon-sensicalview:thatmentalascriptionsarenottrueafterall,butaresimplyfalse.Commonsenseisjustmistakeninsupposingthatpeoplebelieveanddesirethings,andperhapsinsupposingthatpeoplehavesensationsandfeelings,disconcertingasthatnihilisticclaimmayseem.Followingstandardusage,letuscallthenihilisticclaim“eliminativematerial-ism,”or“eliminativism”forshort.Itisimportanttonoteacustomaryifunex-pectedalliancebetweentheeliminativistandthetokenphysicalist:theeliminativist,theidentitytheorist,andthefunctionalistallagreethatmentalitemsare,ifanything,realinner-causalstatesofpeople.Theydisagreeonlyontheempiricalquestionofwhetheranyrealneurophysiologicalstatesofpeopledoinfactanswertothecommon-sensicalmentalcategoriesof“folkpsychology.”Eliminativistspraiseidentitytheoristsandfunctionalistsfortheirforthrightwillingnesstostepupandtaketheirempiricalshot.Botheliminativistsandtokenphysicalistsscorntheinstrumentalist’ssleazyevasion.(Buteliminativistsagreewithinstrumentaliststhatfunctionalismisapipe-dream,andfunctionalistsagreewithinstrumentaliststhatmentalascriptionsareoftentrueandobviouslyso.Thethreeviewsformaneternaltriangleofanotuncommonsort.)PaulFeyerabend(1963a,1963b)wasthefirsttoargueopenlythatthementalcategoriesoffolkpsychologysimplyfailtocaptureanythinginphysicalrealityandthateverydaymentalascriptionswerethereforefalse.(Rorty(1965)tookanotoriouslyeliminativistlinealso,but,followingSellars(1963),triedtosoftenitsnihilism;LycanandPappas(1972)arguedthatthesofteningservedonlytocollapseRorty’spositionintoincoherence.)Feyerabendattractednogreatfollow-ing,presumablybecauseofhisview’soutrageousfloutingofcommonsense.ButeliminativismwasresurrectedbyPaulChurchland(1981)andothers,anddefendedinmoredetail.Churchlandarguesmainlyfromthepovertyof“folkpsychology;”heclaimsthathistorically,whenotherprimitivetheoriessuchasalchemyhavedoneasbadlyonscientificgroundsasfolkpsychologyhas,theyhavebeenabandoned,andrightlyso.P.S.Churchland(1986)andChurchlandandSejnowski(1990)emphasizethecomparativescientificrealityandcausalefficacyofneurobiological60\nTheMind–BodyProblemmechanisms:giventhescientificexcellenceofneurophysiologicalexplanationandthecontrastingdiffusenessandtype-irreducibilityoffolkpsychology,whyshouldwesuppose–evenforaminute,muchlessautomatically–thattheplatitudesoffolkpsychologyexpresstruths?Reasonsforrejectingeliminativismareobvious.First,wethinkweknowtherearepropositionalattitudesbecauseweintrospecttheminourselves.Secondly,theattitudesareindispensabletoprediction,reasoning,deliberation,andunderstand-ing,andtothecapturingofimportantmacroscopicgeneralizations.Wecouldnotoftenconversecoherentlywithoutmentionofthem.ButwhatofP.M.Churchland’sandP.S.ChurchlandandSejnowski’sarguments?Onemaydisputetheclaimthatfolkpsychologyisafailedorbadtheory;Kitcher(1984)andHorganandWoodward(1985)takethisline.Oronemaydisputethemorebasicclaimthatfolkpsychologyisatheoryatall.Ryle(1949)andWittgenstein(1953)staunchlyopposedthatclaimbeforeithadexplicitlybeenformulated.MorerecentcriticsincludeMorton(1980),Malcolm(1984),Baker(1988),McDonough(1991),andWilkes(1993).ReferencesArmstrong,D.M.(1968).AMaterialistTheoryoftheMind.London:RoutledgeandKeganPaul.Baker,L.R.(1988).SavingBelief.Princeton,NJ:PrincetonUniversityPress.Block,N.J.(1978).“TroubleswithFunctionalism.”InW.Savage(ed.),MinnesotaStud-iesinthePhilosophyofScience,Vol.IX:PerceptionandCognition.Minneapolis:UniversityofMinnesotaPress:261–325.ExcerptsreprintedinLycan(1990,1999).——(ed.)(1980).ReadingsinPhilosophyofPsychology,2vols.Cambridge,MA:HarvardUniversityPress.——(1981).“PsychologismandBehaviorism.”PhilosophicalReview,90:5–43.Block,N.J.andFodor,J.A.(1972).“WhatPsychologicalStatesAreNot.”PhilosophicalReview,81:159–81.ReprintedinBlock(1980).Brentano,F.(1973[1874]).PhilosophyfromanEmpiricalStandpoint.London:RoutledgeandKeganPaul.Campbell,K.(1984).BodyandMind(2ndedn).NotreDame,IN:UniversityofNotreDamePress.Carnap,R.(1932–3).“PsychologyinPhysicalLanguage.”Erkenntnis,3:107–42.ExcerptreprintedinLycan(1990).Chalmers,D.(1996).TheConsciousMind.Oxford:OxfordUniversityPress.Chisholm,R.M.(1957).Perceiving.Ithaca,NY:CornellUniversityPress.Churchland,P.M.(1981).“EliminativeMaterialismandthePropositionalAttitudes.”JournalofPhilosophy,78:67–90.ReprintedinLycan(1990,1999).Churchland,P.S.(1986).Neurophilosophy.Cambridge,MA:BradfordBooks/MITPress.Churchland,P.S.andSejnowski,T.(1990).“NeuralRepresentationandNeuralCom-putation.”InLycan(1990):224–52.ReprintedinLycan(1999).Cummins,R.(1983).TheNatureofPsychologicalExplanation.Cambridge,MA:MITPress/BradfordBooks.61\nWilliamG.LycanDavidson,D.(1970).“MentalEvents.”InL.FosterandJ.W.Swanson(eds.),ExperienceandTheory.Amherst,MA:UniversityofMassachusettsPress:79–101.ReprintedinBlock(1980)andinLycan(1999).Dennett,D.C.(1978).Brainstorms.Montgomery,VT:BradfordBooks.——(1981).“MakingSenseofOurselves.”PhilosophicalTopics,12:63–81.ReprintedinLycan(1990).——(1987).TheIntentionalStance.Cambridge,MA:BradfordBooks/MITPress.——(1988).“PrécisofTheIntentionalStance.”BehavioralandBrainSciences,11:495–505.——(1991).“RealPatterns.”JournalofPhilosophy,88:27–51.Dretske,F.(1981).KnowledgeandtheFlowofInformation.Cambridge,MA:BradfordBooks/MITPress.——(1988).ExplainingBehavior.Cambridge,MA:BradfordBooks/MITPress.Feyerabend,P.(1963a).“MaterialismandtheMind–BodyProblem.”ReviewofMetaphys-ics,17:49–66.——(1963b).“MentalEventsandtheBrain.”JournalofPhilosophy,60:295–6.Fodor,J.A.(1968a).“TheAppealtoTacitKnowledgeinPsychologicalExplanation.”JournalofPhilosophy,65:627–40.——(1968b).PsychologicalExplanation.NewYork,NY:RandomHouse.——(1987).Psychosemantics.Cambridge,MA:MITPress.——(1990a).“Psychosemantics.”InLycan(1990):312–37.——(1990b).ATheoryofContent.Cambridge,MA:BradfordBooks/MITPress.——(1994).TheElmandtheExpert.Cambridge,MA:BradfordBooks/MITPress.Geach,P.(1957).MentalActs.London:RoutledgeandKeganPaul.Gordon,R.M.(1987).TheStructureofEmotions.Cambridge:CambridgeUniversityPress.Griffiths,P.(1997).WhatEmotionsReallyAre.Chicago:UniversityofChicagoPress.Hart,W.D.(1988).EnginesoftheSoul.Cambridge:CambridgeUniversityPress.Horgan,T.andWoodward,J.(1985).“FolkPsychologyisHeretoStay.”PhilosophicalReview,94:197–226.ReprintedinLycan(1990,1999).Jackson,F.(1977).Perception.Cambridge:CambridgeUniversityPress.——(1982).“EpiphenomenalQualia.”PhilosophicalQuarterly,32:127–36.ReprintedinLycan(1990,1999).——(1993).“ArmchairMetaphysics.”InJ.O’Leary-HawthorneandM.Michael(eds.),PhilosophyinMind.Dordrecht:KluwerAcademicPublishing.Kirk,R.(1974).“Zombiesvs.Materialists.”AristotelianSocietySupplementaryVolume,48:135–52.Kitcher,P.(1984).“InDefenseofIntentionalPsychology.”JournalofPhilosophy,81:89–106.Kripke,S.(1972).NamingandNecessity.Cambridge,MA:HarvardUniversityPress.Lewis,D.(1966).“AnArgumentfortheIdentityTheory.”JournalofPhilosophy,63:17–25.——(1972).“PsychophysicalandTheoreticalIdentifications.”AustralasianJournalofPhilosophy,50:249–58.ReprintedinBlock(1980).——(1980).“MadPainandMartianPain.”InBlock(1980).Lycan,W.(1987).Consciousness.Cambridge,MA:MITPress/BradfordBooks.——(ed.)(1990).MindandCognition:AReader.Oxford:Blackwell.62\nTheMind–BodyProblem——(1996).ConsciousnessandExperience.Cambridge,MA:MITPress/BradfordBooks.——(ed.)(1999).MindandCognition:AnAnthology.Oxford:Blackwell.Lycan,W.andPappas,G.(1972).“WhatisEliminativeMaterialism?”AustralasianJour-nalofPhilosophy,50:149–59.Malcolm,N.(1984).“ConsciousnessandCausality.”InD.ArmstrongandN.Malcolm,ConsciousnessandCausality:ADebateontheNatureofMind.Oxford:Blackwell.McDonough,R.(1991).“ACulturalistAccountofFolkPsychology.”InJ.Greenwood(ed.),TheFutureofFolkPsychology.Cambridge:CambridgeUniversityPress:263–88.Millikan,R.G.(1984).Language,Thought,andOtherBiologicalCategories.Cambridge,MA:BradfordBooks/MITPress.Morton,A.(1980).FramesofMind.Oxford:OxfordUniversityPress.Nagel,T.(1974).“WhatIsItLiketobeaBat?”PhilosophicalReview,83:435–50.ReprintedinBlock(1980).Owens,J.(1986).“TheFailureofLewis’Functionalism.”PhilosophicalQuarterly,36:159–73.Place,U.T.(1956).“IsConsciousnessaBrainProcess?”BritishJournalofPsychology,47:44–50.ReprintedinLycan(1990,1999).Putnam,H.(1960).“MindsandMachines.”InS.Hook(ed.),DimensionsofMind.NewYork:CollierBooks:136–64.——(1967a).“TheMentalLifeofSomeMachines.”InH.-N.Castañeda(ed.),Intention-ality,Minds,andPerception.Detroit,MI:WayneStateUniversityPress:177–200.——(1967b).“PsychologicalPredicates.”InW.H.CapitanandD.Merrill(eds.),Art,Mind,andReligion,Pittsburgh,PA:UniversityofPittsburghPress:37–48.ReprintedinBlock(1980)underthetitle“TheNatureofMentalStates.”——(1975).“TheMeaningof‘Meaning’.”InPhilosophicalPapers.Cambridge:Cam-bridgeUniversityPress.Rey,G.(1980).“FunctionalismandtheEmotions.”InRorty(1980):163–95.Robinson,W.S.(1988).BrainsandPeople.Philadelphia,PA:TempleUniversityPress.Rorty,A.O.(ed.)(1980).ExplainingEmotions.BerkeleyandLosAngeles,CA:UniversityofCaliforniaPress.Rorty,R.(1965).“Mind–BodyIdentity,Privacy,andCategories.”ReviewofMetaphysics,19:24–54.Ryle,G.(1949).TheConceptofMind.NewYork,NY:BarnesandNoble.Sellars,W.(1963).Science,PerceptionandReality.London:RoutledgeandKeganPaul.Shoemaker,S.(1981).“SomeVarietiesofFunctionalism.”PhilosophicalTopics,12:93–119.Smart,J.J.C.(1959).“SensationsandBrainProcesses.”PhilosophicalReview,68:141–56.Sober,E.(1985).“PanglossianFunctionalismandthePhilosophyofMind.”Synthese,64:165–93.RevisedexcerptreprintedinLycan(1990,1999)underthetitle“PuttingtheFunctionBackIntoFunctionalism.”Solomon,R.(1977).ThePassions.NewYork,NY:Doubleday.Stich,S.(1981).“DennettonIntentionalSystems.”PhilosophicalTopics,12:39–62.ReprintedinLycan(1990,1999).——(1983).FromFolkPsychologytoCognitiveScience.Cambridge,MA:BradfordBooks/MITPress.Strawson,G.(1994).MentalReality.Cambridge,MA:BradfordBooks/MITPress.63\nWilliamG.LycanTuring,A.(1964).“ComputingMachineryandIntelligence.”InA.R.Anderson(ed.),MindsandMachines.EnglewoodCliffs,NJ:Prentice-Hall:4–30.Tye,M.(1983).“FunctionalismandTypePhysicalism.”PhilosophicalStudies,44:161–74.VanGulick,R.(1980).“Functionalism,Information,andContent.”NatureandSystem,2:139–62.Wilkes,K.(1993).“TheRelationshipBetweenScientificandCommonSensePsycho-logy.”InS.ChristensenandD.Turner(eds.),FolkPsychologyandthePhilosophyofMind.Hillsdale,NJ:LawrenceErlbaumAssociates:144–87.Wittgenstein,L.(1953).PhilosophicalInvestigations,trans.G.E.M.Anscombe.NewYork,NY:Macmillan.Wright,L.(1973).“Functions.”PhilosophicalReview,82:139–68.64\nTheBlackwellGuidetoPhilosophyofMindPhysicalismEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter3PhysicalismAndrewMelnykMostphilosophersofmindnowadaysprofesstobephysicalists(ormaterialists)ofonestripeoranother.Generally,however,ifnotinvariably,theyregardtheirphysicalismaboutthemindasaparticularapplicationtomentalphenomenaofaquitegeneralthesisofphysicalismtotheeffectthat,inwhateversenseof“phys-ical”itistruetosaythatthemindisphysical,everythingisphysical.Itiswiththisquitegeneralthesisofphysicalism(henceforth,physicalism)thatthepresentchapterwillbeconcerned.Onemightbetemptedtothinkthattheonlyseriousphilosophicalperplexitieswhichphysicalismprovokesariseinthephilosophyofmind;butitturnsout,asweshallsee,thatphysicalismprovidesmuchtothinkaboutevenifoneleavesasidetheproblemofgivingphysicalisticallyacceptableaccountsofsuchtraditionallyrecalcitrantmentalphenomenaasconsciousness,intentionality,andrationality.Inwhatfollows,Ishallfirstsurveyissuesthatariseinattemptingeventoformulatephysicalismadequately;thenconsiderattemptstojustifyphysicalism;andfinallydiscussthecharacterofobjectionstophysicalism.ThoughIaspiretofairtreatmentofviewsopposedtomyown,thereaderiswarnedthatmydiscussionwillinevitablyreflectsubstantivephilosophicalcom-1mitmentsthatotherwritersinthisareadonotshare.Bywayofbackground,however,letmepresentthephilosophicalproblemtowhichphysicalismcanbeplausiblyviewedasonepossiblesolution(thoughthereareothers).Evenacasualperusalofauniversitycoursedirectorywillrevealthattherearemanysciencesinadditiontophysics:meteorology,geology,zoology,biochemistry,neurophysiology,psychology,sociology,ecology,molecularbio-logy,andsoon,nottomentionhonorarysciencessuchasfolkpsychologyandfolkphysics.Eachofthesemanyscienceshasitsowncharacteristictheoreticalvocabularywithwhich,totheextentthatitgetsthingsright,itdescribesacharacteristicdomainofobjects,events,andproperties.Buttheexistenceofthemanysciencespromptsvariousquestions:howarethemanysciencesrelatedtooneanother?Andhowisthedomainofobjects,events,andpropertiespropri-etarytoeachsciencerelatedtotheproprietarydomainsoftheothers?Dothe65\nAndrewMelnykmanysciencessomehowspeakofdifferentaspectsofthesamethings?Ordotheyaddressthemselvestodistinctsegmentsofreality?Ifso,dothesedistinctseg-mentsofrealityexistquiteindependentlyofoneanother,saveperhapsforrela-tionsofspatio-temporalcontiguity,ordosomesegmentsdependininterestingwaysuponothers?Thisproblemofthemanysciences,aswemaycallit,isevidentlyageneralizationofthemind–bodyproblem,atleastifthatisunder-stoodastheproblemofaccountingfortherelationsbetweenfolkpsychology,on2theonehand,andscientificpsychology,ontheother.Physicalismprovidesaresponsetothisproblem.Anyresponsetoit,whetherphysicalistornot,mustofferasystematicaccountoftherelationsamongthemanysciences,andamongtheirmanydomains;itmustthereforeundertaketheambitiousprojectofsketchingapictureofthetotalityofrealityasrevealedtousbothbyscienceandbycommonsense.Aphysicalistresponsetotheproblem,however,isdistinguishedfromotherresponsesbythefactthatitsaccountoftherelationsamongthemanysciencesandtheirdomainshastheeffectofprivilegingphysicsanditsdomain,ofassigningtophysicsandthephysicalsomesortofdescriptiveandmetaphysicalprimacy;weshallsoonseesomeofthedifferentwaysinwhichthiscanbedone.Butnon-physicalistresponsesarepossibletoo.Inthecurrentclimateofopinion,twoareespeciallynoteworthy.Thefirstcorre-spondsmostcloselytotheintentionsoftraditionalmind–bodydualists,andclaims,ineffect,thatphysicalismisnearlytrue:whatphysicalismsaysabouttherelationbetweenthenon-physicalsciencesandphysicsistrueofeverynon-physicalscienceexceptfolkpsychology,whichmustinsteadbetreatedasdescrib-ingrealphenomenathatareeverybitasbasic,andthatwarrantjustasmuchprivilege,asthosedescribedbyfundamentalphysics.Thesecondnon-physicalistresponseclaimsthatphysicalismisentirelyfalse,alleginginsteadthatakindofpluralisticegalitarianismprevailsamongthevarioussciencesandhonorarysci-ences,sothateveryscienceisonanontologicalparwitheveryother,andtheworldturnsoutnottobestratifiedatall.Anadvocateofthissecondsortofresponsewilljoinwiththetraditionalmind–bodydualistindenyingthatthementalisphysical,butwilladdthatneitheristhegeologicalorthemeteorolo-gicalorthemicrobiological.Today’smostinfluentialanti-physicalistsseemto3favorthissecondresponse.3.1FormulatingPhysicalismAgoodplacetobeginiswiththephysicalistslogan,“Everythingisphysical.”Whatsortofthingsshouldfallwithinthescopeof“everything”?Oneimportant(thoughneglected)questionhereiswhetherthephysicalistmeanstomakeaclaimonlyaboutconcreteentities(e.g.,thephenomenadescribedbythespecialsciences),oralsoaboutabstractentities(e.g.,numbersorpropositionsasunder-stoodbyaPlatonist).Intheabsenceofanyliteraturetoreporton,letmemake66\nPhysicalismtwocomments.First,totheextentthatthereisaprincipleddistinctionbetweenconcreteandabstractentities,physicalistscanperfectlywellstipulatethatthescopeoftheirthesisberestrictedtoconcreteentities;thethesisthatresults,thoughlesscontroversialthananunrestrictedthesis,willstillbeamplycontrover-sial.Secondly,thecrucialquestiontoconsiderindecidingwhethersucharestric-tiononthescopeofphysicalismwouldbeobjectionablyarbitraryisthis:doestherationale,whateveritmightbe,forholdingthatallconcreteentitiesarephysicalcarryoverwithequalforcetothecaseofabstractentities?Onemight,forinstance,considertheproposedjustificationsforphysicalismsketchedbelowinsection3.2andask,ineachcase,whetheritcanbemodifiedtoyieldtheconclu-sionthatabstractentitiesarephysical.Asecondneglectedquestionaboutthescopeofphysicalismconcernsthecategoriesintowhichtheentitiesassertedtobephysicalfall.Letusassume,hereandhenceforth,thattheseentitiesareconcrete;thenthequestioniswhethertheseconcreteentitiesareobjects,events,properties,facts,orwhat.Intuitively,itwouldnotexpressthefullcontentofphysicalismtoclaimmerelythatallconcreteobjectsarephysical;forsurelytheirpropertiesmustbephysicaltoo.Butsincethesepropertiesthemselvesneednotbephysical,solongasalltheirinstancesare,physicalistsshouldperhapsclaimthatallconcreteobjectsandproperty-instances4arephysical.Forthesakeofclarityandsimplicity,Ishallassumethatthisisright.If,inadditiontoobjectsandproperty-instances,one’sontologyalsoincludeseventsorstatesorprocessesorconditionsorwhatever,andiftheseareirreducibletoobjectsandproperty-instances,then(token)events,states,processes,andsooncanandshouldbeincludedwithinthescopeofone’sphysicalism.Letusreturntotheslogan,“Everythingisphysical.”Howshouldweunder-stand“...isphysical”?Anattractiveideaistocountanentityasphysicalif,andonlyif,itisofakindexpressedbysomepredicateintheconsensustheoriesofcurrentphysics,wherethenatureofconsensustheoriesofcurrentphysicscanreadilybediscoveredbyconsultingsomeup-to-datephysicstextbooks;examplesofphysicalobjectsarethereforesuchthingsaselectronsandquarks,andexamplesofphysicalpropertiessuchpropertiesaschargeandmass.Givenourearlierassumptions,theresultingdoctrineofphysicalismwillthenclaimthateveryconcreteobjectandproperty-instanceisofakindexpressedbysomepredicateintheconsensustheoriesofcurrentphysics.Onitsface,physicalismofthissortseemswildlycounterintuitive,sinceitapparentlyentailsthenon-existenceofprettymucheverykindofthingdescribedbythespecialsciencesandbycommonsense.Forcabbagesandkings,embassiesandelephants,percolatorsandprices–noneofthesethingsisexpressedbyapredicateintheconsensustheoriesofcurrentphysics.Defendersofphysicalismofthissort,whenitisinterpretedinthisradicallyeliminativistway,neednotdenythattheworldcertainlyappearstocontainmorethantheirphysicalismcounten-ances,butwillaspiretoexplainthatfalseappearancebyappealonlytophysicalentities:theymaydenythatthereareanyelephants,anddenyinparticularthatanyelephantisthere,buttheywillinsistthatsomethingoccupiesthespacewhere67\nAndrewMelnykcommonsenselocatesanelephant–presumablysomeorotherspatio-temporalarrangementofmicrophysicalparticles(see,e.g.Maxwell1968).Infact,however,physicalism,whenunderstoodasclaimingthateveryconcreteobjectandproperty-instanceisofakindexpressedbysomepredicateintheconsensustheoriesofcurrentphysics,doesnotbyitselfentailanyeliminativistconsequencewhatsoever;itdoessoonlywhencombinedwiththeadditionalpremisethatthekindsofthingdescribedbythespecialsciencesandbycommonsensearenot(numerically)identicalwiththekindsofthingexpressedbypredic-atesintheconsensustheoriesofcurrentphysics.Nowthisadditionalpremiseiswidelyaccepted,mainlyonthegroundsthatthekindsofthingdescribedbythespecialsciencesandbycommonsensearemultiplyrealizable,inthesensethatphysicallyverydifferentassemblagesofmicrophysicalparticlesdo(ormerelycan)neverthelessconstituteindividualsofthesamespecial-scientificorcommon-sensekind.ForsupposethatanobjectofphysicalkindKandanobjectofincompatiblephysicalkindJbothconstitutepercolators;thenifbeingapercolatorweretheverysamepropertyasbeingofphysicalkindK,itwouldfollowthateverypercol-atorwasanobjectofphysicalkindK;butsinceatleastonepercolator(theoneofphysicalkindJ)isnotofphysicalkindK,itfollowsthatbeingapercolatorcannotbetheverysamepropertyasbeingofphysicalkindK.However,thisargumentagainstidentifyingspecial-scientificandcommon-sensekindswithphysicalkinds5maybechallenged.Foronething,theextentofactualmultiplerealizationofcertainphenomena(e.g.,ofphenomenalconsciousness)isnotclear,whilethesignificanceofthemereconceivabilityofmultiplerealizationcanbedoubtedbyanyphilosopherwhodoubtsthatconceivabilityisareliableguide,orevenanyguideatall,togenuinemetaphysicalpossibility.Foranotherthing,ifarbitrarydisjunctionsofphysicalpredicatesexpressauthentic,andauthenticallyphysical,kinds,thenitlooksasiftheargumentfrommultiplerealizabilitycanbeevadedbysimplyidentifyingapparentlynon-physicalkindswiththephysicalkindsexpressedbysuitabledisjunctionsof(conjunctionsof)physicalpredicates;forexample,perhapsbeingapercolatoristhephysicalpropertyofbeingofphysicalkindKorofphysicalkindJ.Multiplerealizabilityarguments,however,haveinfactpersuadednearlyallphilosophersthatspecial-scientificandhonorary-scientifickindscannotbeidenti-fiedwithphysicalkinds.Since,ifthisisright,theformulationofphysicalismwehavejustbeenconsideringentailsthestronglyeliminativistconclusionsmootedabove,philosophersinterestedinformulatingacomprehensivedoctrineofphys-icalismhaveinrecentdecadestakenadifferentapproach.Accordingtoit,phys-icalismclaimsthateveryconcreteobjectandproperty-instanceiseitherphysicalinsomenarrowsenseof“physical”orelseisphysicalinthebroadsenseofstandinginacertainrelationtothingsthatarephysicalinthenarrowsense.Thusformulated,physicalismneednotdenytheexistenceofthingsofthekindsdescribedbythespecialsciencesandbycommonsense;forthosethings,eveniftheyarenotphysicalinthenarrowsense,mayyetstandintherightrelation(whateveritturnsouttobe)tothingsthatare.Nor,onthisnewapproach,68\nPhysicalismwouldphysicalismhavetoclaimtheidentityofthekindsdescribedbythespecialsciencesandbycommonsensewith(narrowly)physicalkinds,for(presumably)therightrelationneedneitherbe,norentail,identity.How,onthisapproach,shouldweunderstand“physical”inthenarrowsense?Anobvioussuggestionistorecycletheaccountof“physical”simplicitergivenaboveandthereforesaythatsomethingisphysicalinthenarrowsense(hence-forth,“physicaln”)if,andonlyif,itisofakindexpressedbysomepredicateintheconsensustheoriesofcurrentphysics.Butif“physicaln”isunderstoodinthisway,thentheresultingformulationofphysicalismcanbetrueonlyifcurrentphysicsitselfisbothtrueandcomplete,somethingthatseemsmostunlikelygivenphys-ics’historicaltrack-recordoferrorandomission;sotheresultingformulationofphysicalismmustitselfbemostunlikelytobetrue,whichcertainlysoundsbadforphysicalism.Andthisdifficulty,ifgenuine,afflictsalltheformulationsofphysicalismdiscussedsofar.Soeitherphysicalismcannotbeformulatedatall(assomeanti-physicalistsallege)orweshouldunderstand“physicaln”insomeotherway.Buthow?Itwillnotdotosuggestthatsomethingisphysicalnif,andonlyif,itisofakindexpressedbysomepredicateofcompletedphysics.Forifnoconstraintisplaceduponwhatcompletedphysicsmightbelike,then,forallwenowknow,itmightpostulatetheexistenceofCartesianminds,contrarytotheintentionsofaspiringphysicalists.Moreover,itishardtoseehowanyscientificfindingscur-rentlyavailabletouscouldpossiblyconstituteevidenceforaphysicalismformu-latedbyappealtoacompletedphysicswhosecontentisatpresententirelyunknowntous.Somethirdwayofunderstanding“physicaln”thereforeseemsrequired.Perhapsitshouldappealtotheideaofamodestextensionofcurrentphysics,somethingsimilarenoughtocurrentphysicsforitscontentnottobeentirelyobscuretous,butflexibleenoughtowithstandthediscoverythatcurrentphysicsisbothincompleteand(insomerespects)false.Butwhetherathirdwaycanbefoundtoavoidtheproblemsbesettingthetwowaysjustconsideredisa6questionthatremainsunresolvedand,tillrecently,largelyunexplored.Somuchfor“physicaln.”How,ontheapproachwearepresentlyconsidering,shouldweunderstand“physical”inthebroadsense(henceforth,“physicalb”)?Overwhelmingly,themostpopularanswergivenoverthepastdecadeortwohasbeenthatsomethingisphysicalbif,andonlyif,itsupervenesuponthingsthatarephysicaln;andtheformulationofphysicalismityieldsclaimsthateveryconcreteobjectandproperty-instanceiseitheritselfphysicalnorelsesupervenesuponthingsthatarephysicaln.Theconceptofsupervenienceinphilosophy(layusageof“supervene”haslittletodowithitsphilosophicalusage)canbeexplainedintuitivelylikethis:themental(forexample)supervenesuponthephysicalif,andonlyif,oncethephysicalfactshavebeenfixed,thementalfactsaretherebyfixedalso;thewaythingsarementallycannotvary(andnotmerelydoesnotvary)withoutvariationalsointhewaythingsarephysically.Nowtheextensiveliteratureontheconceptofsupervenienceisfullofpro-posalsforhowtounderstandclaimsofsuperveniencemoreprecisely,andoneissuewhichsuperveniencephysicalists(aswemaycallphilosopherswhowishsome69\nAndrewMelnykthesisofsuperveniencetoplayanimportantroleintheformulationofphysicalism)mustresolveconcernstheexactkindofsupervenienceclaim,preciselyunder-stood,thataformulationofphysicalismshoulduse.Onemuch-discussednotion7ofsupervenienceisJaegwonKim’sstrongsupervenience.Accordingtothis,theclaimthatnon-physicalnpropertiessuperveneuponphysicalnpropertiesshouldbeunderstoodasfollows:non-physicalnpropertiesstronglysuperveneuponphysicalnpropertiesif,andonlyif,necessarily,ifathinghasanon-physicalnproperty,thenthereissomephysicalnpropertythatthethinghassuchthat,necessarily,anythingwiththatphysicalnpropertyalsohastheoriginalnon-physicalnproperty.Accord-ingtothissupervenienceclaim,however,allthenon-physicalnpropertiesofagivenobjectarefixedbyphysicalnpropertiesofthatveryobject(perhapsevenbyphysicalnpropertiesofthatveryobjectcontemporaneouswithitsnon-physicalnproperties).Andthisimplicationseemsinconsistentwithanumberofplausiblesug-gestionsastotheconstitutionofvariousnon-physicalnproperties;forexample,thesuggestionthatthegenuinenessofadollarbillispartiallyconstitutedbyits(historical)relationtoappropriateauthorities,thesuggestionthattheposses-sionofafunctionbyabiologicalentity(e.g.,aheart)ispartiallyconstitutedbytheselectionalhistoryofitsancestors,andtheexternalistsuggestionwidelyen-dorsedbyphilosophersofmindthattherepresentationalcontentofpropositionalattitudesispartiallyconstitutedbytheirrelationstostatesofaffairsexternaltotheirowners’heads.Forthisreason,superveniencephysicalistshavegenerallypreferredtoemployKim’snotionofglobalsupervenience.Averycrudefirststabatexpressingthedesiredsupervenienceclaimmightbethis:anytwopossibleworldsexactlyalikeinrespectofthephysicalnentitiesandproperty-instancestheycontainandthephysicallawsthatholdthereareexactlyalikeinrespectofallthe(concrete)entitiesandproperty-instancestheycontain.Theliteraturecontains8sophisticateddiscussionofhowsuchaclaimshouldberefined.Asecondissuewhichaspiringsuperveniencephysicalistsmustresolveconcernsthecharacterandappropriatestrengthofthemodalitytheirclaimsinvoke.Formodal-operatorformulations(e.g.,theclaimofstrongsupervenienceabove),thequestioniswhatkindofnecessitythenecessityoperatorsshouldexpress.Concep-tualnecessity?Metaphysicalnecessity?Nomologicalnecessity?Somethingelse?Forpossible-worldformulations(e.g.,theclaimofglobalsupervenienceabove),thequestioniswhichpossibleworldstheclaimshouldquantifyover.Literallyallpossibleworlds?Merelythosewhoselawsofnaturearethesameastheactualworld’slawsofnature?Someotherset?Anyadequateanswertosuchquestionsmustapparentlysteeramiddlecoursebetweentwoextremes.Letmeillustratewiththecaseofclaimsofglobalsupervenience.Ontheonehand,ifsuchclaimsquantifyoverliterallyallpossibleworlds,thentheyseemtobefalse.Fortheyentailthateveryworldphysicallynindistinguishablefrom(say)ourworldisalsoindistinguishablementallyfromourworld.Butsurelythereisaworldwhichisindistinguishablefromourworldphysicallyn,butdistinguishablementally,sinceinadditiontoitsphysicalncontentsitcontainsectoplasmicspiritswhichplaytherightfunctionalrolestocountasminds.Thisso-called“problemofextras”has70\nPhysicalismnogenerallyagreedsolution(forcriticaldiscussionofproposedsolutions,seeWitmer1999).Ontheotherhand,claimsofglobalsuperveniencecannotquan-tifymerelyovernomologicallypossibleworlds(i.e.,thoseinwhichthelawsofnatureareexactlythesameasthoseintheactualworld),forinthatcasethesupervenienceclaimscouldbetruewhilephysicalismwerefalse.Suppose,forinstance,thatphysicalismturnsouttobefalse,butonlybecausephenomenalpropertiesturnouttobenon-physicalbpropertiesthatareneverthelesslinkedbyfundamentalpsychophysicallaws,biconditionalinform,tocertainphysicalprop-erties;inthatcaseclaimsofglobalsuperveniencequantifyingovernomologicallypossibleworldswouldstillbetrue,sinceanynomologicallypossibleworldphysicallynindistinguishablefromtheactualworldwouldbeoneinwhichthosefundamentalpsychophysicallawsheldandthereforeoperatedtoproduceexactlythesamedistributionofphenomenalpropertiesasobtainsintheactualworld.Soclaimsofglobalsuperveniencemustquantifyoversomesetofpossibleworldsdistinctfromboththesetofallpossibleworldsandthesetofnomologicallypossibleworlds.Thereiscurrentlynoconsensusonwhatthatsetis.Athirdissuewhichaspiringsuperveniencephysicalistsmustresolveconcernstheexactrolethatasupervenienceclaim,howeverexpressed,isintendedtoplayintheoverallformulationofphysicalism.Theintendedrolevariesfromauthortoauthorandissometimesleftratherobscure.Someauthorsseemtotakeanappropriateglobalsupervenienceclaimtoconstitutethewholeofphysicalism.Othersholdthatatleastsomeadditionalclaimisrequired(totheeffectthateveryobjectiseitheraphysicalnobjectorelseaspatio-temporalsumofphysicalnobjects);butitisunclearwhatexactlytheyregardassufficient;perhapstheyregardasuperveniencethesisasexpressingphysicalismaboutproperties,whereasotherclaimsarerequiredtoexpressphysicalismaboutparticulars.Attheveryleastthereisalooseendtobetieduphere.Evenwhenalltheissuesjustmentioned,aswellasothers,havebeenresolved,however,theadequacyofasupervenienceformulationofphysicalismremainsopentodoubt;andenthusiasmforsuchformulationshasdeclinedsteadilyoverthelastdecade.Thecontentofthedoubtisthatalthoughanappropriateclaimofsuperveniencemaybealogicallynecessaryconditionforphysicalism,itfailsasalogicallysufficientcondition(evenforphysicalismaboutproperties).Thegroundofthedoubt,putbriefly,isthis:anysupervenienceclaimthathasbeenpressedintoserviceasaformulationofphysicalismismerelyavariationonthethemethatthephysicalnwaythingsarenecessitatesthenon-physicalnwaythingsare.Butthereisnoexplanation,entailedbythesupervenienceclaimitself,forhowandwhythisnecessitationoccurs;so,forallthatthesupervenienceclaimitselfsays,thenecessitationofthenon-physicalnbythephysicalnmightconstituteabrutemodalfact;butif,forallthatthesupervenienceclaimitselfsays,thenecessitationofthenon-physicalnbythephysicalnmightsimplybeabrutemodalfact,thenthesupervenienceclaimitselfyieldsnointuitivelysatisfactorysenseinwhichthementalisphysicaln.Nosupervenienceclaim,therefore,sufficesforphysicalism9aboutanything.71\nAndrewMelnykNoalternativetosuperveniencephysicalism’swayofunderstanding“physicalb”hasasyetachievedthepopularityonceenjoyedbythesupervenienceproposal,butanalternativetoit,thoughneglecteduntilrecently,hasindeedexistedforacoupleofdecades(Boyd1980;Lycan1981,1987:ch.4).Itsleadingideaisthatsomethingisphysicalbif,andonlyif,itisafunctionalkindofthingthatisrealizedbythephysicaln;andtheformulationofphysicalismityieldsclaimsthateveryconcreteobjectandproperty-instanceiseitheritselfphysicalnorelseisfunctionalandrealizedbysomethingthatisphysicaln.Realizationphysicalists(aswemaycallphilosopherswhoendorsesuchaformulationofphysicalism)arethereforecommittedtoholdingthatall(actual)non-physicalnkindsare,infact,functionalkinds;buttheyarenotcommittedtoanyconceptualorlinguisticthesiswhatever–nothesis,forexample,allegingthefunctionaldefinabilityofnon-physicalnconceptsorterms.(Realizationphysicalismcanthereforebeviewedasasortofgeneralizationofpsychofunctionalisminthephilosophyofmind.)Also,realizationphysicalistsdonotdenyappropriatelyexpressedclaimsofsuper-venience;indeed,theymayregardsuchclaimsaslogicallynecessaryconditionsofthetruthofphysicalism.Buttheyinsistthatwhatexplainsthesupervenienceofthenon-physicalnonthephysicaln(ifitdoessosupervene)isthefactthatthe10non-physicalnisfunctionalandrealizedbythephysicaln.Thecompanionnotionsofafunctionalkindandofrealizationthatrealizationphysicalistsexploitarefamiliar,ofcourse,fromthephilosophyofmind.Buttheveryheavyloadwhichrealizationphysicalismrequiresthemtobearhasrevealedthattheyareemployedeveninthephilosophyofmindinsensesthatareneitheruniformnorclear.Sorealizationphysicalistsneedtospellouthowtheyareunderstandingthem.Oneattractiveapproachistotreatfunctionalkindsashigher-orderkinds:afunctionalproperty,P,willthenbethepropertyofhavingsomeorotherpropertythatplaysroleso-and-so;afunctionalobject-kind,O,willbethekindofobjectthatexistsif,andonlyif,thereexistsanobjectofsomeorotherkindthatplayssuch-and-sucharole;andsoon.Therolesherereferredtomaybecausalornomicorcomputational–orofanyothersort,sinceplayingaroleisreallynomorethanmeetingacertainspecifiablecondition,andinprincipletheconditioncouldbeofanysort.Realizationcannowbeunderstoodasrole-playing.IffunctionalpropertyPisthepropertyofhavingsomeorotherpropertythatplaysroleso-and-so,thenanypropertyQthatplaysroleso-and-socanbesaidtorealizeP.Thisapproachtounderstandingrealization,however,needsmuchrefinementbeforerealizationphysicalistshaveinhandanotionadequateforformulatingphysicalism.Oneissueinparticularthatneedsattentioniswhetherrealizationphysicalistscangiveasatisfactoryaccountoftherealizationofindi-viduals(tokens),asopposedtokinds(types),andwhether,insodoing,theymustorshouldassertclaimsofidentitybetweennon-physicalnindividuals(objects,property-instances,andsoon)andthephysicalnindividualsthatrealize11them.Nowthatwehavesurveyedsomeattemptstoformulatephysicalismmoreprecisely,weareinapositionbrieflytoconsiderthreequestionsaboutthethesis72\nPhysicalismofphysicalism:(1)Isitcontingentornecessary?(2)Isitapriorioraposteriori?(3)Isitreductionist?Onitsface,physicalismseemscontingent.Since(whenexpressedinsloganform)itistruejustincaseeveryconcretethingisphysical,itisfalseifevenoneconcretethingexiststhatisnon-physical.Butitissurelycontingentwhetherornotthereexistanyconcretenon-physicalthings(e.g.,ectoplasmicghosts).Sophysicalismiscontingent.Moreover,thisconclusionstillholdswhenphysicalismismorepreciselyformulatedastheclaimthateveryconcreteobjectandproperty-instanceisofakindexpressedbysomepredicateintheconsensustheoriesofcurrentphysics;foritissurelycontingentwhetherornotanyconcreteobjectorproperty-instanceexiststhatisnotofakindexpressedbysomepredicateintheconsensustheoriesofcurrentphysics.Butdoesphysicalismremainacontingentthesiswhenitisformulatedintermsofsupervenienceorintermsofrealization?Itdoes.Admittedly,anysupervenienceformulationofphysicalismclaimsthatthephysicalnwaythingsarenecessitatesthenon-physicalnwaythingsare,whichcer-tainlysoundslikeanon-contingentclaim.Ontheotherhand,asupervenienceformulationofphysicalismmustapplytotheactualworld,implyingataminimumthattheactualworldissuchthatthephysicalnwaythingsareinitnecessitatesthenon-physicalnwaythingsareinit,i.e.,thatanyworldphysicallynjustliketheactualworldisalsonon-physicallynjustliketheactualworld.Nowonewaytoensurethatasupervenienceformulationofphysicalismsucceedsindoingthisistospellitoutasaglobalsupervenienceclaimthatquantifiesoverallpossibleworldswithoutexception;foriftheclaimquantifiesoverallpossibleworlds,assertingthatanytwoworldsexactlyalikephysicallynareexactlyalikeineveryway,thenitobviouslyentailsthatanyworldexactlyliketheactualworldphysicallynisexactlyliketheactualworldineveryway.Spelledoutasaquantificationoverliterallyallpossibleworlds,then,asupervenienceformulationofphysicalismdoesexpressanon-contingentclaim,notdependentforitstruthonwhattheactualworldturnsouttobelike.However,inordertoavoidthe“problemofextras”discussedabove,asuperveniencefor-mulationofphysicalismshouldquantifyoverfewerthanallthepossibleworlds;itshouldquantifyonlyoverallpossibleworldsthatmeetsomecontingentconditionX(whateverthatmightbe),thusclaimingmerelythatanytwoX-worldsexactlyalikephysicallynareexactlyalikeineveryway.Butinthatcasetheclaimappliestotheactualworld(i.e.,impliesthatanyworldphysicallynjustliketheactualworldisalsonon-physicallynjustliketheactualworld)onlyiftheactualworldmeetsconditionX,whichisacontingentmatter.So,strictlyspeaking,asupervenienceformulationofphysicalismmustincludenotonlyasupervenienceclaimwhichquantifiesoverasuitablyrestrictedsetofpossibleworlds,butalsothecontingentclaimthattheactualworldinfactbelongstothatrestrictedset.Aplausiblesupervenienceformulationofphysicalism,therefore,isacontingentthesis,dependentforitstruthonwhatsortofworldwehappentoinhabit.Somesuperveniencephysicalists,however,ensurethattheirformulationsofphysicalismapplytotheactualworldbymakingthemexplicitlyrefertotheactualworld;accordingtosuchformulations,anyworldexactlyliketheactualworld73\nAndrewMelnyk12physicallynisexactlyliketheactualworldnon-physicallyn.Nowlet“P”beacompletephysicalndescriptionoftheactualworld,and“Q”beacompletenon-physicalndescriptionoftheactualworld.Then,accordingtotheseformulations,ifphysicalismistrue,theconditional“IfPthenQ”expressesanecessarytruth.Soisphysicalismanon-contingentthesis,accordingtotheseformulations?No.Fortheconditional“IfPthenQ”isnotlogicallysufficientforphysicalism;itislogicallysufficientonlyifitisconjoinedwiththecontingentclaimthat“P”isacompletephysicalndescriptionoftheactualworld,andthat“Q”isacompletenon-physicalndescriptionoftheactualworld.Theconditional“IfPthenQ”simplydoesnotentailthatanyworldexactlyliketheactualworldphysicallynisexactlyliketheactualworldnon-physicallynunlessitis(contingently)truethat“P”expressesthephysicalnwaytheactualworldisandthat“Q”expressesthenon-physicalnwaytheactualworldis.Thinkofitthisway:youaregivenanextensivephysicalnworld-description“S,”andanextensivenon-physicalnworlddescription“T,”andyoufigureout(usingapriorimethods,letussuppose)that“IfSthenT”expressesanecessarytruth;haveyoutherebyfiguredoutthatphysicalismistrue?Obviouslynot,becauseyoudonotyetknowwhether“S”and“T”accuratelydescribetheactualworld;andwhethertheydoisamatterofcontingentfact.Strictlyspeaking,then,superveniencephysicalismformulatedsoastoreferexplicitlytotheactualworldisthethesisthat(1)“P”isacompletephysicalndescriptionoftheactualworld,(2)“Q”isacompletenon-physicalndescriptionoftheactualworld,and(3)“IfPthenQ”expressesanecessarytruth.Physicalismalsoremainscontingentwhenformulatedwithhelpfromthenotionofrealization.Since,whensoformulated,itclaimsthateveryconcreteobjectandproperty-instanceiseitheritselfphysicalnorelseisfunctionalandrealizedbysome-thingthatisphysicaln,itisfalseifconcretethingsexistthatareneitherphysicalnnorrealizedbythephysicaln.Butitissurelycontingentwhatconcretethingsexist;sophysicalism,thusformulated,iscontingent.Nor,forreasonsalreadyrehearsed,shouldthisconclusionberejectedonthegroundsthatarealizationformulationofphysicalismentailsasuitablyformulatedclaimofsupervenience.Letusturnnowtotheepistemicstatusofphysicalism.Since,onanyplausibleformulation,thethesisofphysicalismiscontingent,wecansafelypresumethatitisaposteriori.Surelythethesisthattheactualworld(thoughmaybenotothers)issuchthatthenon-physicalnphenomenaitcontainsareidenticalwith,orsup-erveneupon,orarerealizedby,thephysicalnphenomenaitcontainsisathesiswhosetruthorfalsitycouldonlybeestablishedbyexaminingtheactualworld.Still,whatabouttheepistemicstatusofthespecificmodalclaimstowhichphysicalistsarecommitted,oncethewaythingsactuallyarehasbeendiscoveredandspecified?Thatis,whatabouttheclaimthat(takethedefinitedescriptionsrigidlynow)thewaythingsactuallyarephysicallynnecessitatesthewaythingsactuallyarenon-physicallyn?Here,opinionsdiffer.Themajorityviewsupposesthatthisnecessitationholdsinvirtueofaposteriorinecessaryidentitiesbetweenthenon-physicaln,ontheonehand,andeitherthephysicalnorthefunctionalandphysicallynrealized,ontheother;onthisview,then,evenwhenwehavelearned74\nPhysicalism(aposteriori)whattheactualworldislikephysicallynandnon-physicallyn,itisstillaposterioriwhetherphysicalismistrue.Thereisaminorityview,however,accordingtowhich,ifphysicalismistrue,thensomeonewhohadacompletephysicalndescriptionoftheactualworldandwhopossessedalltheconceptsusedtoformulatenon-physicalnclaimscouldinprinciplededucehiswaytoacompletenon-physicalndescriptionoftheworld(Chalmers1996;Jackson1998).Ifthisviewiscorrect,then,bytakingacompletephysicalndescriptionoftheactualworldandbyusingone’sgraspofnon-physicalnconcepts,onecoulddeduceacompletenon-physicalndescriptionallegedlytrueoftheactualworldandthentestphysicalismbycomparingthisnon-physicalndescriptionwithacompletenon-physicalndescriptiondiscoveredempirically.Finally,letusaskwhetherphysicalismisreductionist,i.e.,whetheritentailsthatthenon-physicalnisreducibletothephysicaln.Accordingtoawidecon-sensus,ofthethreeformulationsofphysicalismwehaveconsidered,thefirstformulation(construedasnon-eliminativist)isreductionist,whilethesecond(supervenience)andthird(realization)arenot;indeed,theoriginatorsoftheselatterformulationsexplicitlyaimedtoformulateversionsofnon-reductionist13physicalism.Butthisconsensusmustrestontheassumptionofsomeorotheraccountofwhatreducibilitytothephysicalnis;andaccordingtotheaccountofreducibility(derivedfromErnestNagel)thatseemsinfacttobeassumed,thenon-physicalnisreducibletothephysicalnif,andonlyif,allnon-physicalnlawscanbededucedfromphysicalnlawsbymeansofadditionalpremises(i.e.,“bridgeprinciples”)assertingtheidentityofnon-physicalnkindswith(tractabledisjunctionsof)physicalnkinds.Supervenienceandrealizationformulationsofphysicalismdonotentailthereducibilityinthissenseofthenon-physicalntothephysicaln,sincebothareconsistentwiththemultiplerealizabilityofnon-physicalnkindsbyin-tractablymanydistinctphysicalnkinds.(Andinfactthefirstformulationdoesnotentailiteither,solongasitispermissibletoavoidtheproblemofmultiplerealizabilitybyidentifyingnon-physicalnkindswithintractable,perhapsinfinite,disjunctionsofphysicalnkinds.)Buttheneo-Nagelianaccountofreducibilityisasubstantivephilosophicalclaim.Whatifitisincorrect?Orwhatifitisnotuniquelycorrect(sothatthereisnosinglecorrectaccountofreducibility)?Thecoreideaofreducibilityseemstobethis:thenon-physicalnisreducibletothephysicalnjustincasethenon-physicalnissomehowexplainableintermsofthephysicaln.Theneo-Nagelianaccountiscertainlyonespecificationofthiscoreidea(construingexplanationasaspeciesofderivation),butitseemslikelythatotherspecificationsshouldbepossibletoo,andplausiblethatnooneofthemshouldbeuniquelycorrect.Willphysicalisminthatcasestillemergeasnon-reductionist,orasnon-reductionistinimportantways?Someexplorationofalternativeaccountsofreducibilitycanbefoundintheliterature(see,e.g.,Waters1990;Smith1992;Brooks1994;Melnyk1995;Chalmers1996;Bickle1998).Alsowelcomewouldbefurtherexplorationofthedifferentkindsofautonomythataspecialsciencelikepsychologycan–andcannot–enjoy,consistentlywiththetruthofphysicalism;itwouldbeniceto75\nAndrewMelnykknow,forexample,howfarthemethodologicalautonomyofpsychologyrequiresitsmetaphysicalautonomy.3.2JustifyingPhysicalismOntheassumptionthatathesisofphysicalismcanbesatisfactorilyformulated,andthat,soformulated,itisaposteriori,thequestionnaturallyariseswhetherthereisinfactanyempiricalevidencethatthethesisistrue.Whatsortofnon-deductivereasoningstrategiescouldinprincipleprovidesuchevidence?Anddoanyofthosestrategiesyieldevidenceforphysicalismwhenputintopractice?Foralltheenthusiasmforphysicalism,inphilosophyofmindandelsewhere,itissurprisinghowlittleattentiontheseissueshavereceived.Ontheotherhand,theyhavereceivedmoreattentionthansomeanti-physicalistrhetoricmightsuggest.Letusbrieflyreviewsomesuggestionsastohowphysicalismmightbeevidenced,andtheissuesthosesuggestionsraise.Oneproposal(modeledonDavidLewis’sargumentforthepsychophysicalidentitytheory)isthatatwo-premiseargumentcanbeusedtosupporttheconclusionthatsomenon-physicalnkindNisidenticalwithaphysicalnkindP(Lewis1966;thegeneralizationissuggestedinJackson1998).ThefirstpremisestatesthatNisthekindthatplaysso-and-sorole;thesecondpremisestatesthatPisthekindthatplaysso-and-sorole;andtheconclusionthatN=Pfollowsbythetransitivityofidentity.Thefirstpremiseisinferablefromtheallegedlyaprioriconceptualorlinguisticclaimthat“N”issemanticallyequivalentto“thekindthatplaysso-and-sorole”;thesecondpremiseisdiscoverableempiricallybycheckingoutwhatrolesphysicalnkindsinfactplay.Butmostphysicalistswoulddoubttheapplicabilityofthisargumentativestrategy,onthegroundsthatinpointoffacttheproprietaryconceptsortermsofthespecialsciencesarenotingeneral,perhapsnotever,semanticallyequivalenttodefinitedescriptions,asthefirstpremiseseemstorequire;suchadoubtwouldformpartofageneraldoubtaboutdescriptivisttheoriesofthemeaningsofconceptsorterms.AcurrentlyopenquestioniswhetherthegeneralizedLewisianargumentativestrategycanberepairedbysupposingthatspecialscientificconceptsorterms,thoughnotsemant-icallyequivalenttodefinitedescriptions,stillhavetheirreferencesfixedbymeansofaprioriknowabledefinitedescriptionsoftherigidifiedform“thekindthatactuallyplaysso-and-sorole.”Asecondproposalisthatconclusionsassertingtheidentityofnon-physicalnkindswithphysicalnkindscanbesupportedbyaninferencetothebestexplanationwhichtakesasitsdatatheobservedfactthatindividualsofthenon-physicalnkindoccurwhenandonlywhen,andwhereandonlywhere,individualsofthephys-icalnkindinquestionoccur.Suppose,then,thatweobservetheco-instantiation,inthissense,ofnon-physicalnkindNandphysicalnkindP.Surelyonehypo-thesiswhichcouldexplainthisobservedco-instantiationisthatN=P;certainly76\nPhysicalismifN=P,andindividualsofkindsNandPareobservedatall,theycannotfailtobeobservedtogether.AndthisidentityhypothesisisplausiblyregardedasabettercandidateexplanationthantherivalwhichassertsthedistinctnessofNandP,andwhichaccountsfortheirobservedco-instantiationbysupposingthemtobeconnectedbyafundamentallawofnature.Theidentityhypothesislooksbetterthanthisrivalbecause,intwoseparateways,itismoreeconomicalthantherival:itpostulatesjustonekind(=N,=P),whereastherivalpostulatestwo(NandP);andthenumberoflawsofnaturewhichitmusttreatasbruteandfundamentalisfewer(byone)thanthenumbertherivalmusttreatasbruteandfundamental.Accordingly,theobservedco-instantiationofNandPprovidesevidencethatN=P(forelaboration,seeHill1991:ch.2).Suchapatternofreasoningmightappeartobeoflimitedusefulness,sinceveryfewcontemporaryphysicalistswishtoendorsethesortofkind-to-kindidentityclaimswhichitsupports.Butthereasoningsuggestedcanbeextendedsothatitshowshowtosupportphysicalistconclusionsotherthanthosewhichasserttheidentityofnon-physicalnkindswithphysicalnkinds.Oneextensionisobvious:iftheempiricalevidencewithwhichthereasoningbeginsistheobservedco-instantiationofsomenon-physicalnkindNwithsomefunctional(ratherthanphysicaln)kindF,thenthereasoningcanpresumablybeusedtosupportthehypo-thesisthatN=F,ahypothesiswhicharealizationphysicalistwouldobviouslyfindcongenial.Ofcourse,todiscoverthatanon-physicalnkindNisafunctionalkindisnotyettodiscoverthatNisphysicallyrealized;so,sincethislatterconclusioniswhatrealizationphysicalismneeds,afurtherextensionoftheoriginallineofreasoningwouldbedesirable.Supposethattheobservationswhichserveasdataarethat,wheneverandwhereverthereisanindividualofphysicalnkindP,thereisalsoanindividualofnon-physicalnkindN;becauseofmultiplerealization,how-ever,theconverseisnotobserved.Theseobservationsarepotentiallyexplainablebyadoptingthehypothesisthat(roughly)NisidenticalwithsomefunctionalkindF,andisinfactrealizedbyphysicalnkindP;forifN(=F)isrealizedbyP,thenPissufficientforN,andsonaturallywheneverandwhereverthereisaP,therewillbeanN.Asbefore,ifthishypothesisissuperiortoitsrivalsinrespectofeconomy,thentheoriginalobservationsprovideevidencethatNisfunctionalandphysicallynrealized.Theseproposalsraiseatleasttwoimportantquestions.Oneiswhetheritislegitimatetoappealtoeconomy(orsimplicity)inthewayinwhichthesuggestedpatternsofreasoningdo;obviouslythisquestionturnsontheresolutionoflargeissuesinepistemology.Theotherquestioniswhetherwidelyacceptedscientificfindingscanbeusedtoconstructactualinstancesofthesepatternsofreasoningthathavetruepremises;thisquestionhashithertobeenprettymuchignored.Athirdlineofempiricalreasoninginsupportofphysicalism,andtheonethathasreceivedthemostattention,runsasfollows(see,e.g.,Peacocke1979:134–43;Melnyk1994).Thefirstpremiseistheso-calledcausalclosure(orcompleteness)ofthephysicaln.Itassertsthatthephysicalnisclosedinthesensethatonedoesnotneedtogooutsidetherealmofthephysicalninordertofindasufficientcauseof77\nAndrewMelnykphysicalnphenomena:everyphysicalneventhasasufficientphysicalncause(tothe14extentthatithasacauseatall).Thispremiseissupportedbycurrentphysics,whichhasinvestigatedeversomanyphysicalneventsbutwhichknowsofnonefortheexplanationofwhichitisnecessarytoinvokenon-physicalncauses.Thesecondpremiseisthatnon-physicalneventshavephysicalneffects.Certainlynon-physicalneventshavenon-physicalneffects(e.g.,hurricanesblowdowntrees).Butcurrentphysicsassuresus–andnoscientificrealistseriouslyconteststhistoday–thatnon-physicalneffectsatleasthavephysicalnparts.Sinceitishardtoseehowanon-physicalneventcouldhaveanon-physicalneffectwithoutalsohavingsomeeffectofsomekindonsomephysicalnpartofthatnon-physicalneffect,non-physicalneventsmusthavesomephysicalneffects.Fromthesetwopremises,togetherwiththeassumptionthatthenon-physicalneventswhichhavephysicalneffectsarenotphysicaleveninsomebroadsenseof“physical,”itfollowsthatsomephysicalneventsarecausallyoverdetermined;foreveryphysicalneventwhichhasanon-physicaln(andhence,bytheassumption,non-physicalb)causealsohasanentirelydistinctphysicalncause.Buttotheextentthat–andthisisthethirdpremise–itishighlyimplausiblethatphysicalneventsarecausallyoverdetermined,itisreasonabletorejecttheassumptionthatthenon-physicalneventswhichhavephysicalneffectsarenotphysicaleveninsomebroadsenseof“physical,”andhencetoacceptthat,insomebroadsenseof“physical,”theyarephysical.Afurther,enumerative-inductivestepleadstotheuniversalconclusionthatallnon-physicalnevents,andnotmerelythoseknowntohavephysicalneffects,arephysicalinsomebroadsense(i.e.,identicalwith,supervenienton,orrealizedby,thephysicaln).Thislineofreasoningpromptsmanyquestions(fordiscussion,seeMills1996;Sturgeon1998;Witmer2000;Papineau2001).Isthecausaloverdeterminationtowhichtherejectionofphysicalismallegedlyleadsreallysuchabadthing?Andifso,why?Isthecausalclosureofthephysicalnsomethingforwhichthereisevidencethatwouldbeacceptabletosomeonewhowasnotalreadyconvincedofphysicalism?Andaretherecounterexamplestoit?Isitreallytrue,inanynormalsenseof“cause,”thatnon-physicalneventscausephysicalneffects?Finally,cantheargumentbemodifiedtoaccommodatetheapparentlyindeterministiccharacterofthephysicalnrealm?Theliteraturecontainsothersuggestionsastohowphysicalismmightbeevi-denced(Loewer1995;Papineau1995;forcriticism,seeWitmer1998).Butinviewofthesurprisinglylittleattentionthatphilosophershavepaidtotheques-tionofjustifyingphysicalism,itstrikesmeasunlikelythatallthepossiblesugges-tionshaveyetbeenthoughtup.3.3ObjectingtoPhysicalismSincethethesisofphysicalism,aswehavebeenunderstandingit,hasthelogicalformofauniversalgeneralization,itisinprincipleopentocounterexamples:78\nPhysicalismconcreteobjectsorproperty-instancesthatareneitherphysicalnnorphysicalb.Onesortofobjectiontophysicalism,therefore,consistsingroundsforthinkingthatsuchentitiesreallydoexist.Suchentitiesmightbeones(e.g.,God,vitalforces,astralbodies)whosesheerexistenceisdeniedbyphysicalists,inwhichcasetheobjectormustprovideaposterioriorapriorigroundsforthinkingthattheydoexist.Alternatively,andmoreplausibly,suchentitiesmightbeones(e.g.,rationaldecisions,episodesinembryonicdevelopment)whosesheerexistence(atleastonaneutralconstrualofwhattheirexistenceentails)isundisputedbyphysicalists,butwhosecharacterizationasneitherphysicalnnorphysicalbisdis-puted,inwhichcasetheobjectormustprovideaposterioriorapriorigroundsforthinkingthattheentitiesinquestionareindeedneitherphysicalnnorphysicalb.Toillustrate:nophysicalistsdenythathumanbeingsexist(well,hardlyany);buthumanbeingsareacounterexampletophysicalismifitcanbeshown,asparapsy-chologicalresearchershavetriedsystematicallytoshow,thathumanbeingshavepowers(e.g.,psychokineticpowers)thathumanbeingssimplycouldnothaveiftheywerephysicalb.Mostoftheobjectionstophysicalismfamiliarfromthephilosophyofmindliteratureareobjectionsofthisfirstsort,i.e.,putativecounterexamples(see,e.g.,Robinson1993).Inrespondingtothem,physicalistswillnaturallywanttoexamineeachcaseonitsmerits,andweobviouslycannotenterintoanydetailshere.Butweshouldpausetonoticeacertainphilosophicaloutlookwhichislikelytounderliephysicalists’particulararguments,andwhichmaynotbesharedbytheiroppon-ents.Thisoutlookamountstoadeepsuspicionofanyallegedlyapriorigroundforholdingeitherthatsomeconcreteentityexistsorthat(itsmereexistencegranted)itisneitherphysicalnnorphysicalb.Accordingtothisoutlook,groundsforholdingthataconcreteentityofanykindexistshavetobeaposteriori.Likewise,anygroundsforholdingthataconcreteentityisneitherphysicalnnorphysicalbmustalsobeaposteriori,sincetheymustruleoutthepossibilitythattheentityisidenticalaposteriorieitherwithaphysicalnentityorwithanentitythatisfunctionalbutphysicallynrealized.Accordingly,physicalistsareunlikelytobeimpressedby,orperhapseventotakeseriously,objectionstophysicalismthat15startwithapremiseaboutwhatisconceivablebyhumans.Thesortofobjection,bycontrast,whichwouldreallyimpressaphysicalistwiththisoutlookwouldbetheidentificationofsomenon-tendentiousempiricalphenomenonforthebestexplanationofwhichitwasrequiredeithertopostulateanewsomethingneitherphysicalnnorphysicalbortoconstrueasneitherphysicalnnorphysicalbsomethingalreadyacknowledgedbyeveryonetoexist.Anexampleofsuchaphenomenonwouldbesometypeofhumanbehaviorwhichdemonstrablycouldnotbetheproductoftheoperationsofamerelyphysicalbsystem.Thephysicalist,however,deniesthatanysuchphenomenaactuallyexist,andtheplausibilityofthisdenial,inthelightofthepastcenturyofscience,goesalongwaytoexplainsuch16popularityasphysicalismenjoys.Nowanti-physicalistswhoadvanceobjectionsofthisfirstsortarelikelytoadoptapositiveresponsetotheproblemofthemanysciencesaccordingtowhich79\nAndrewMelnykphysicalismisrightaboutthephysicalbcharacterofmuchofwhatisnon-physicaln,thoughnotright,ofcourse,aboutitall.However,anti-physicalistswhoadvanceobjectionsofthethreeremainingsortsthatIshallconsiderarelikelytoadoptthepluralistic,egalitarianresponsetotheproblemofthemanysciencesaccordingtowhicheveryscience(orhonoraryscience)isonanontologicalparwitheveryother,sothatprettymuchnothingthatisnon-physicalnisphysicalb.Thefirstsuchobjectionisthatphysicalismcannotbetrue,becauseitcannotevenbeadequatelyformulated;anditcannotbeadequatelyformulatedbecause,forreasonsrehearsedinsection3.1,thereisnosatisfactorywaytodefine“physicaln.”Ifcorrect,thisobjectionisobviouslydevastatingtophysicalism.Butwhetheritiscorrectremainsundecidedandformsthetopicofongoingresearch.Thesecondobjectionclaimsthat,evenifphysicalismcanbesatisfactorilyformu-17lated,thereissimplynoreasonwhatevertothinkthatitistrue.(Somephilo-sophers,itseems,evenwishtoexplainthepopularityofphysicalismbyappealtosomesortofwhollyirrationalphysics-worship.)Ifthisiscorrect,then,sincesciencecertainlypresentsanappearanceofplurality,thereisnoreasonnottotakethisappearanceatfacevalueandthereforetotreatallthebranchesofscienceasmetaphysicallyequal.Butitisatbestprematuretoclaimthatthereisnoevidenceforphysicalism,since,aswehaveseen,theredoexistpromisinglinesofargumentforphysicalismandinanycasethematterthusfarhasonlybeenrathercursorilyinvestigatedbyphilosophers.Moreover,ifphysicalismisamereprejudice,thenitisanoticeablymoreprevalentoneamongthose(Imainlymeannon-philosophers)whohavesomeideaofwhatcondensedmatterphysicshastosayaboutfamiliarmacrophysicalphenomena,whatquantummechanicshastosayaboutchemistry,whatbiochemistryhastosayaboutcellbiology,andsoforth.Itispossiblethatthe(admittedlyimperfect)correlationbetweenbeingaphysicalistandbeingscientificallywellinformedcanbeexplainedsociologicallyorbyappealtosomesortofsystematicerrorinreasoning(thoughonewoulddearlyliketoseethehypothesisspelledout),butonthewholeitseemslikelierthatthepeopleinquestionseedimlythatwhattheyknowaboutsciencedoesconstituteevidencefor18physicalism,eveniftheycannotsayexactlyhowitdoes.Ifthisconjectureiscorrect,thenevidentlythereisworkforphilosopherstodoinhelpingthemout.ThethirdandfinalobjectionthatIshallconsideristhatphysicalismisimplaus-iblebecauseitimpliesthatnoeventsotherthanphysicalneventsareevercauses,andthatnopropertiesofeventsotherthanthephysicalnpropertiesofthoseeventsareevercausallyrelevantinthesenseofmakingadifferencetowhateffectstheeventshave(see,e.g.,Lowe1993;Moser1996).Ifphysicalismdoesimplythesethings,thenthatisbad;forsurelynon-physicalnevents(e.g.,decisions,earthquakes,chemicalreactions)aresometimescauses,andsurelyanevent(e.g.,acollisionwithasharpknife)cansometimeshavetheeffectithasbecauseitwasacollisionwithasharpknife,eventhoughtheeventkind,collisionwithasharpknife,isnotaphysicalneventkind.Andphysicalismdoesseemtoimplythesethings.Forif,foreverynon-physicalneffect,thereisanunderlyingphysicalnphenomenonsufficientforit(asphysicalismrequires),andifallsuchunderlying80\nPhysicalismphysicalnphenomenaarecompletelycausedbyearlierphysicalnphenomenainstrictaccordancewithphysicalnlaws,thenphysicalnphenomenaseemtobedoingalltherealcausalwork,andtheappearanceofnon-physicalncausationisjustanillusion.Butphysicalistswillhardlyallowthislineofreasoning,onwhichthethirdobjectionclearlyturns,togounchallenged;andtheymayinadditiontrytofashionanindependentlyplausibleaccountofcausationandcausalrelevancewhichdoesnotentailthat,ifphysicalismistrue,onlyphysicalneventsarecauses19andonlyphysicalnpropertiesarecausallyrelevant.Notes1TheinterestedreaderwillfindafullaccountofmyviewsinMelnyk(2003).2Themind–bodyproblemisfamouslycastinsuchtermsinChurchland(1981).3Anti-physicalistsofthissortappeartoincludeGoodman(1978),Putnam(1987),CraneandMellor(1990),andDupré(1993).4Aconcreteproperty-instanceisaninstanceofaconcreteproperty(e.g.,thepropertyofhavingmass);anabstractproperty-instanceisaninstanceofanabstractproperty(e.g.,thepropertyofbeingdivisiblebyfive).5OnechallengerisJaegwonKim;seeKim(1998).6Forthedifficultyhereandonepossiblesolution,seePoland(1994:ch.3).InMelnyk(1997),Idubbedthisdifficulty“Hempel’sdilemma,”inhonorofHempel(1980),andarguedthat,notwithstandingthereasoninginthetext,thereisnogoodobjectiontodefining“physical”intermsofcontemporaryphysics.However,seeDaly(1998),Montero(1999),CrookandGillett(2001).Foranotherapproach,seePapineau(2001);foracritique,seeWitmerandGillett(2001).7SeeKim(1984),thepaperwhich,intheUnitedStatesatleast,hassetthetermsofthedebateaboutsupervenience.Forfurtherdiscussion,seeKim(1993)andthepapersbyMcLaughlinandPostinSavellosandYalçin(1995).AnexcellentsurveyisHorgan(1993).8See,forexample,HellmanandThompson(1975),Haugeland(1982),Horgan(1982,1987),Lewis(1983),Post(1987),Jackson(1998).Foracloselyrelated(sincealsomodal)approach,seeKirk(1996).9Thisismywayofputtingthematter;seeMelnyk(1998,1999).Forsimilarconcerns,seeHorgan(1993).10ThefullestaccountofrealizationphysicalismisMelnyk(2003).SeealsoPoland(1994),whichadvocatesahybridformofphysicalismincorporatingbothsuperveni-enceandrealizationelements.11ThisistheonlyoccasiononwhichIshallmentionthethesisthateverytoken(e.g.,individualevent)isidenticalwithsomeorotherphysicalntoken.Thoughfamouslypropounded(see,e.g.,Fodor1974;Davidson1980),ithasplayedasurprisinglysmallroleinrecentdiscussionofphysicalism,perhapsbecausenooneseemstoregarditassufficientforphysicalism(unlesseventsaretreatedasKim-events,inwhichcaseitbecomesequivalenttotheunpopulartype-identityphysicalismdiscussedfirstinthetext).Davidsonhimself,ofcourse,advancedthethesisalongsideasuperveniencethesis.81\nAndrewMelnyk12SeeChalmers(1996)andJackson(1998).Inthetext,Imisrepresenttheseauthors,thoughharmlesslyasfarasthecurrentissuegoes:inordertohandlethe“problemofextras,”theywouldnotsay“exactlyliketheactualworldnon-physicallyn,”butrather“exactlyliketheactualworldwithregardtopositivenon-physicalnfacts.”13ThesourceofthisconsensusmaywellbeFodor(1974).SeealsoFodor(1997).14Confusingly,anotherclaimisalsosometimesreferredtoasthe“causalclosureofthephysical,”theclaimthatphysicalncausesaretheonlycausesofphysicalneffects.Thislatterclaimleadsswiftlytophysicalism,giventhefurtherpremisethatnon-physicalneventsarecausesofphysicalneffects;butforthatveryreasonitwillberegardedasquestion-beggingbyanti-physicalists.Itisnotentailedbytheclosureclaiminthetext.15DefenseofthisoutlookagainstthechallengetoitpresentedbyChalmers(1996)maybefoundinmy(2001).16Thedenialthatanysuchphenomenaexistis,Ibelieve,oneofthelinesofpro-physicalistthoughttobefoundinSmart’sclassic(1959).17Distastefulthoughitistomention,Ifearthatitmustbeasked,ofphilosopherswhoclaimtofindnoargumentsforphysicalismintheliterature,howhardtheyhavelooked.18AhypothesisofthisformisdefendedinPapineau(2001).19Nearlyallofthephilosophyofmindliteratureabouttheproblemsofmentalcausa-tionis,ofcourse,relevanthere.ReferencesBickle,John(1998).PsychoneuralReduction:TheNewWave.Cambridge,MA:TheMITPress.Boyd,Richard(1980).“MaterialismWithoutReductionism:WhatPhysicalismDoesNotEntail.”InNedBlock(ed.),ReadingsinthePhilosophyofPsychology,Vol.1.London:Methuen:268–305.Brooks,D.H.M.(1994).“HowToPerformAReduction.”PhilosophyandPhenomeno-logicalResearch,54:803–14.Chalmers,David(1996).TheConsciousMind:InSearchofaFundamentalTheory.NewYork:OxfordUniversityPress.Charles,David,andLennon,Kathleen(1992).Reduction,Explanation,andRealism.NewYork:OxfordUniversityPress.Churchland,Paul(1981).“EliminativeMaterialismandthePropositionalAttitudes.”TheJournalofPhilosophy,78:67–90.Crane,Tim,andMellor,D.H.(1990).“ThereIsNoQuestionOfPhysicalism.”Mind,90:185–206.Crook,SethandGillett,Carl(2001).“WhyPhysicsAloneCannotDefinethe‘Physical’:Materialism,Metaphysics,andtheFormulationofPhysicalism.”CanadianJournalofPhilosophy,31:333–60.Daly,Chris(1998).“WhatArePhysicalProperties?”PacificPhilosophicalQuarterly,79:196–217.Davidson,Donald(1980).EssaysonActionsandEvents.NewYork:OxfordUniversityPress.Dupré,John(1993).TheDisorderofThings:MetaphysicalFoundationsoftheDisunityofScience.Cambridge,MA:HarvardUniversityPress.82\nPhysicalismFodor,JerryA.(1974).“SpecialSciences,orTheDisunityofScienceAsAWorkingHypothesis.”Synthèse,28:97–115.——(1997).“SpecialSciences:StillAutonomousAfterAllTheseYears.”InJamesE.Tomberlin(ed.),PhilosophicalPerspectives,11,Mind,Causation,andWorld.Cambridge,MA:Blackwell:149–63.Gillett,CarlandLoewer,Barry(2001).PhysicalismandItsDiscontents.NewYork:Cam-bridgeUniversityPress.Goodman,Nelson(1978).WaysofWorldmaking.Sussex:HarvesterPress.Haugeland,John(1982).“WeakSupervenience.”AmericanPhilosophicalQuarterly,19:93–103.Hellman,Geoffrey,andThompson,Frank(1975).“Physicalism:Ontology,Determina-tion,andReduction.”TheJournalofPhilosophy,72:551–64.Hempel,CarlG.(1980).“CommentsonGoodman’sWaysofWorldmaking.”Synthèse,45:193–9.Hill,ChristopherS.(1991).Sensations:ADefenseofTypeMaterialism.NewYork:Cam-bridgeUniversityPress.Horgan,Terry(1982).“SupervenienceandMicrophysics.”PacificPhilosophicalQuarterly,63:29–43.——(1987).“SupervenientQualia.”PhilosophicalReview,96:491–520.——(1993).“FromSupervenienceToSuperdupervenience:MeetingtheDemandsofaMaterialWorld.”Mind,102:555–86.Jackson,Frank(1998).FromMetaphysicsToEthics:ADefenceofConceptualAnalysis.NewYork:OxfordUniversityPress.Kim,Jaegwon(1984).“ConceptsofSupervenience.”PhilosophyandPhenomenologicalResearch,45:153–76.ReprintedinKim(1993).——(1993).SupervenienceandMind:SelectedPhilosophicalEssays.NewYork:CambridgeUniversityPress.——(1998).MindinaPhysicalWorld:AnEssayontheMind–BodyProblemandMentalCausation.Cambridge,MA:TheMITPress.Kirk,Robert(1996).“StrictImplication,Supervenience,andPhysicalism.”AustralasianJournalofPhilosophy,74:244–57.Lewis,David(1966).“AnArgumentfortheIdentityTheory.”JournalofPhilosophy,63:17–25.——(1983).“NewWorkForATheoryOfUniversals.”AustralasianJournalofPhilo-sophy,61:343–77.Loewer,Barry(1995).“AnArgumentforStrongSupervenience.”InSavellosandYalçin(1995):218–25.Lowe,E.J.(1993).“TheCausalAutonomyoftheMental.”Mind,102:629–44.Lycan,WilliamG.(1981).“Form,Function,andFeel.”JournalofPhilosophy,78:24–50.——(1987).Consciousness.Cambridge,MA:TheMITPress.Maxwell,Grover(1968).“ScientificMethodologyandtheCausalTheoryofPerception.”InI.LakatosandA.Musgrave(eds.),ProblemsinthePhilosophyofScience.Holland:NorthHollandPublishingCompany.McLaughlin,BrianP.(1995).“VarietiesofSupervenience.”InSavellosandYalçin(1995):16–59.Melnyk,Andrew(1994).“BeingAPhysicalist:HowAnd(MoreImportantly)Why.”PhilosophicalStudies,74:221–41.83\nAndrewMelnyk——(1995).“TwoCheersForReductionism:Or,TheDimProspectsForNon-ReductiveMaterialism.”PhilosophyofScience,62:370–88.——(1997).“HowToKeepThe‘Physical’InPhysicalism.”TheJournalofPhilosophy,94:622–37.——(1998).“TheProspectsforKirk’sNon-ReductivePhysicalism.”AustralasianJournalofPhilosophy,76:323–32.——(1999).“Supercalifragilisticexpialidocious:ACriticalStudyofSavellosandYalçin’sSupervenience:NewEssays.”Nous,33:144–54.——(2001).“PhysicalismUnfalsified:Chalmers’InconclusiveConceivabilityArgument.”InGillettandLoewer(2001):329–47.——(2003).APhysicalistManifesto:ThoroughlyModernMaterialism.Cambridge:Cam-bridgeUniversityPress.Mills,Eugene(1996).“InteractionismandOverdetermination.”AmericanPhilosophicalQuarterly,33:105–17.Montero,Barbara(1999).“TheBodyProblem.”Nous,33:183–200.Moser,PaulK.(1996).“PhysicalismandMentalCauses:ContraPapineau.”Analysis,56:263–7.Papineau,David(1995).“ArgumentsforSupervenienceandPhysicalRealization.”InSavellosandYalçin(1995):226–43.——(2001).“TheRiseofPhysicalism.”InGillettandLoewer(2001):3–36.Peacocke,Christopher(1979).HolisticExplanation:Action,Space,Interpretation.NewYork:OxfordUniversityPress.Poland,Jeffrey(1994).Physicalism:ThePhilosophicalFoundations.NewYork:OxfordUniversityPress.Post,JohnF.(1987).TheFacesofExistence:AnEssayinNonreductiveMetaphysics.Ithaca,NY:CornellUniversityPress.——(1995).“‘Global’SupervenientDetermination:TooPermissive?”InSavellosandYalçin(1995):73–100.Putnam,Hilary(1987).TheManyFacesofRealism.Illinois:OpenCourt.Robinson,Howard(1993).ObjectionstoPhysicalism.NewYork:OxfordUniversityPress.Savellos,EliasE.andYalçin,ÜmitD.(eds.)(1995).Supervenience:NewEssays.NewYork:CambridgeUniversityPress.Smart,J.J.C.(1959).“SensationsandBrainProcesses.”PhilosophicalReview,68:141–56.Smith,Peter(1992).“ModestReductionsandtheUnityofScience.”InCharlesandLennon(1992):19–43.Sturgeon,Scott(1998).“PhysicalismandOverdetermination.”Mind,107:411–32.Waters,C.Kenneth(1990).“WhyTheAnti-ReductionistConsensusWon’tSurvive:TheCaseofClassicalMendelianGenetics.”InA.Fine,M.Forbes,andL.Wessels(eds.),PSA1990.EastLansing,MI:PhilosophyofScienceAssociation.Witmer,D.Gene(1998).“WhatisWrongwiththeManifestabilityArgumentforSuper-venience.”AustralasianJournalofPhilosophy,76:84–9.——(1999).“SuperveniencePhysicalismandtheProblemofExtras.”TheSouthernJour-nalofPhilosophy,37:315–31.——(2000).“LocatingtheOverdeterminationProblem.”TheBritishJournalforthePhilo-sophyofScience,51:273–86.Witmer,D.GeneandGillett,Carl(2001).“A‘Physical’Need:PhysicalismandtheViaNegativa.”Analysis,61:302–9.84\nTheBlackwellGuidetoPhilosophyofMindDualismEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter4DualismHowardRobinson4.1IntroductionDualisminthephilosophyofmindisthedoctrinethatmindandbody(ormentalstatesandphysicalstates)areofradicallydifferentnatures.Howexactlytoexpressthisdifferenceisamatterofcontroversy,butitisgenerallytakentocenterontwopropertiespossessedbythementalthatarealientothephysical.Oneoftheseistheprivacyorsubjectivityofstatesofconsciousness,ascontrastedtothepublicavailabilityofphysicalstates.Theotheristhepossessionofintentionalityor“aboutness”bymentalstates:physicalstatesstandinspatio-temporalandcausalrelationstoeachother,butarenotintrinsicallyaboutanything.Theprincipletaskforthephysicalististogiveanaccountofthesepropertiesinphysicalorphysical-1compatibleterms.Adualistissomeonewhothinksthatthiscannotbedone.Therearenormallythoughttobetwoformsofdualism,namelysubstancedualismandbundledualism.TheformerisprimarilyassociatedwithDescartes2andthelatterwithHume.Animportantdistinctionmustbemadeamongstbundledualists,however.Some,likeHume,donotbelieveineithermentalorphysicalsubstance,treatingbothasjustcollectionsofstates,properties,orevents(dependingonhowthetheoryisstated).Forothers,itisonlythemindthatisgiventhistreatment:bodiesaresubstantialentities,butmindsonlycollectionsofstates,properties,orevents.Thisconstitutesarelativedowngradingofthemindandamovetowardtheattributetheory.Accordingtothistheory,mentalstatesarenon-physicalattributesofaphysicalsubstance–thehumanbodyorbrain.Thistheorycanberegardedasthesoftestorleastreductiveformofmaterialism.Itismaterialisticbecauseitsaysthattheonlysubstancesarematerialsubstances.Itisalsoaformofdualism,becauseitallowstheirreducibilityofmentalstatesandproperties.Bothsubstanceandbundledualismsfacethesamethreeproblems.Thefirstproblemistoshowwhyweneedtobedualistsatall–whyamaterialistaccount85\nHowardRobinsonofthemindwillnotwork.Thesecondistoexplainthenatureoftheunityoftheimmaterialmind.FortheCartesian,thatmeansexplaininghowheunderstandsthenotionofimmaterialsubstance.FortheHumean,theissueistoexplainthenatureoftherelationshipbetweenthedifferentelementsinthebundlethatbinds3themintoonething.Neithertraditionhasbeennotablysuccessfulinthislattertask:indeed,Humedeclaredhimselfwhollymystifiedbytheproblem,rejecting4hisowninitialsolution(thoughquitewhyisnotclearfromthetext).Thethirdproblemistogiveasatisfactoryaccountoftherelationshipbetweentheimma-terialmindandthematerialbody.Whichmeans,forpreference,toexplainhowtheycaninteract,and,failingthis,torenderplausibleeitherepiphenomenalism(theviewthatthementalisproducedbythephysical,buthasnoinfluencebackonthephysical)orparallelism(theviewthatmentalandphysicalrealms“marchinstep,”butwithouteithercausallyinteractingwiththeother).Ishallusetheexcuseoflimitedspacefornotdealingwithalltheseissues.Rather,Ishallattempt,inCartesianspirit,toshow,first,thatthethinkingsubjecthastotranscendthephysicalworld;and,secondly,thatsuchsubjectsmustbeessentiallysimple.They(thatis,we)aremoreliketheimmaterialsubstanceinwhichDescartesbelieved,thanlikeaHumeanbundleofmentaleventsorstates.SoIshallbeconcernedwithwhyweshouldbedualists,andwhydualistsofaCartesianstripe.Howtoexplaintheunityofthemind–exceptbyshowingittobeessentiallysimple–andhowtoexplainourrelationstoourbodies,arenot5issuesIcandiscusshere.InordertoaccomplishthefirstofthetasksIhavesetmyself(thatis,toshowthatthethinkingsubjectmusttranscendthephysicalworld),Ishallintroduceaformofdualismnotsofarmentioned,andwhichisgenerallyneglectedindiscus-sionsofdualism,namelypredicatedualism.Thatisthetheorythatpsychologicalormentalisticpredicatesarenotreducibletophysicalisticpredicates.(WhatthismeansIshalldiscussinthenextsection.)Fewphilosophersnowadayseitherbelieveinsuchreductionorthinkthatitisnecessaryforphysicalism.Predicatedualismisonlydualismatthelevelofmeaning,andthisisgenerallythoughttohavenoontologicalconsequences.Ishallbearguingthatthisisamistake,andthatpredicatedualism–thefailureofreduction–isathreattophysicalismbecausetheirreducibilityofthespecialsciencesingeneralimpliesthatthemindisnotanintegralpartofthephysicalrealmwithwhichthosesciencesdeal.Thisconclusiondoesnotaloneforceustoadoptanyparticularformofdual-ism.Perhapsthemind,thoughittranscendsthephysicalworldaboutwhichitconstructsthesciences,isnomorethanabundleofmentalstatesorproperties,asHumethought.Perhaps,thatis,predicatedualismforcesustonothingmorethanpropertydualism,whichmaynotdriveonefurtherawayfromphysicalismthantheattributetheory.Ishallthenattempttoshowthatthisisnotso,forpropertydualismisnotadequatetocopewithcertainrespectsinwhichpersonalidentityisdemonstrablydifferentfromtheidentityconditionsforphysicalbodiesandothercomplexentities:theseconstraintsonpersonalidentitycanbemetonlybysubstancedualismofaroughlyCartesiankind.86\nDualism4.2TheArgumentforPredicateDualismIfphysicalismistrue,thenitshouldbepossible,inprinciple,togivewhatis,insomesense,atotaldescriptionoftheworldinthevocabularyofacompletedphysics.Toputitinthematerial,nottheformal,mode,allthepropertiesthatthereultimatelyareshouldbethoseofthebasicphysicalentities.Buttherearemanywaysoftalkingtrulyabouttheworldotherthanthatcouchedinthevocabularyofphysics;andthereare,insomeobvioussense,manypropertiesthattheworldpossessesthatarenotcontainedinthatphysics.Thesehigher-orderpredicatesandpropertiesareexpressedintheother–orspecial–sciences,suchaschemistry,biology,citology,epidemiology,geology,metereology,psychology,andthesupposedsocialsciences;nottomentionourordinarydiscourse,whichoftenexpressestruthsthatfindnoplaceinanythingwewouldnaturallycallascience.Howdoesthefundamentallevelofontology–whichwearepresuppos-ingtobecapturedideallyinphysics–sustainalltheseotherontologiesandmaketruetheseotherlevelsofdiscourse?Thelogicalpositivistshadasimpleanswertothisquestion.Anyrespectablelevelofdiscoursewasreducibletosomelevelbelowitandultimatelytophysicsitself.Thekindofreductionofwhichwearetalkinghasastrongformandaverystrongform.Accordingtotheverystrongform,allrespectablestatementsinthespecialsciencesandinordinarydiscoursecould,inprinciple,betranslatableintostatementsinthelanguageofphysics.Intheend,therefore,alltruthscouldbe6expressedusingthelanguageofphysics.Accordingtothemerelystrongform–whichwastheforminwhichreductionismwasgenerallydiscussed–therehadonlytobescientificlaws(called“bridginglaws”)connectingtheconceptsandlawsinahigher-ordersciencewiththoseinthenextlower,andultimatelyto7physics.Sotheconceptsandlawsofpsychologywouldbenomicallyconnectedtothoseofsomebiologicalscience,andthese,inturn,withchemistry,andchemistrywouldbenomicallyreducibletophysics.So“reducibleto,”inthissense,meantthattheentitiesandpropertiesinvokedinthenon-basicdiscourseweretypeidenticalwithcertainbasicstructures.Forexample,ourordinarycon-ceptwaterisreducibletothechemicaltypeH2O,andthischemicalmoleculealwaysconsistsofthesameatomicarrangements.Thispatternmakesiteasytounderstandintuitivelyhowtheexistenceofwaterandthetruthsofsentencesreferringtowaterneedinvolvenothingmorethantheexistenceofthingsintheontologyofphysics.Butnotallconceptsinthespecialsciences,letaloneordinarydiscourseandthesocialsciences,canbefittedintothispattern.Noteveryhurricanethatmightbeinvokedinmetereology,oreverytectonicshiftthatmightbementionedingeology,willhavethesamechemicalorphysicalconstitution.Indeed,itisbarelyconceivablethatanytwowouldbesimilarinthisway.Norwilleveryinfectiousdisease,oreverycancerousgrowth,nottomentioneverydevaluationofthecurrencyoreverycoupd’étatsharesimilarstructuresindepth.JerryFodor,inhis87\nHowardRobinsonimportantarticle“SpecialSciences”(1974),correctlyclaimsthatthedoctrineofreductionismrequiresthatallourscientificallylegitimateconceptsbenaturalkindconceptsand–likewater–carrytheirsimilaritiesdowntothefoundations,andthatthisisnotplausibleformostofourusefulexplanatoryconcepts.Itisparticularlynotplausiblefortheconceptsofpsychologicalscience,understoodinfunctionalistterms,norfortheconceptsinourlaymentalisticvocabulary.Alltheseconceptsaremultiplyrealizable,whichmeansthatdifferentinstancesofthesamekindofthingcanbequitedifferentatlowerlevels–intheir“hardwear”–andthatitisonlybyapplyingtheconceptsfromthespecialsciencethatthedifferentcasescanbeseenassalientlysimilaratall.Whereasyoucouldeliminatetheword“water”andspeakalwaysof“H2O”withnolossofcommunicativepower,youcouldnotdothisfor“livinganimal,”“thoughtoftheEiffelTower,”“continentaldrift,”etc.Fodor(1974)thinksthatthisisnothreattophysicalism,becauseeachinstanceofahigher-orderconceptwillbeidenticalwithsomestructuredescribableintermsofbasicphysics,andnothingmore.Tokenreductionismisallthatphysicalismandtheunityofthesciencesrequire:typereductionisunnecessary.Ishallnowtrytoexplainwhy,contrarytoappearances,thisiswrong.4.3WhyPredicateDualismleadstoDualismProperFodorisquiterighttothinkthattheverysamesubjectmattercanbedescribedinirreduciblydifferentwaysandstillbejustthatsubjectmatter.What,inmyview,hefailstonoticeisthatsuchdifferentexplanatoryframeworkspresupposeaperspectiveonthatsubjectmatterwhichis,primafacie,fromoutsideofit.Theoutlineofmypositionisasfollows.Onarealistconstrual,thecompletedphysicscutsphysicalrealityupatitsultimatejoints:anyspecialsciencewhichisnomicallystrictlyreducibletophysicsalso,invirtueofthisreduction,itcouldbeargued,cutsrealityatitsjoints,butnotatitsminutestones.Bycontrast,asciencewhichisnotnomicallyreducibletophysicsdoesnottakeitslegitimationfromtheunderlyingrealityinthisdirectway;rather,itisformedfromthecollaborationbetween,ontheonehand,objectivesimilaritiesintheworldand,ontheother,perspectivesandinterestsofthosethatdevisethescience.Ifscientificrealismistrue,acompletedphysicswilltellonehowtheworldis,independentlyofanyspecialinterestorconcern:itisjusthowtheworldis.Platetectonics,however,tellyouhowitisfromtheperspectiveofaninterestinthedevelopmentofcontinents,andtalkabouthurricanesandcoldfrontsfromtheperspectiveofaninterestintheweather.Aselectionofphenomenawithacertainteleologyinmindisrequiredbeforethesestructuresorpatternsarereified.Thepointisthatthesesciencesandtheentitiesthattheypostulateexistfromcertainintellectualperspectives,andaperspective,whetherperceptualorintellectual,isexternaltothatonwhichitisa8perspective.Theproblemforthephysicalististosaywhatitisforaperspective88\nDualismonthephysicalworldtobesomethingwithinit.Aunifiednaturalisticviewoftheworldwouldrequirethattheobserver’sperspectiverequiredbythesesciencesbeintegratedintotherealityheobserves.Theintegrationofperspectivesandinterestsintotheoneworldrequirestheintegrationofpsychologicalstatesofbothperceptualandintentionalkindsintothephysicalworld.These,however,areparadigmsofthekindsofstatethatseemtoresistnomic–type–reductiontophysics.Thereare,ofcourse,famousargumentsthatappealtothephenomenologyofconsciousnessforthinkingthattokenreductionsfail:butnoappealtotheseis9involvedinthecurrentargument.Eveniftokenreductionismofthemindcouldmeetthephenomenologicalproblems,thefactthatitistoken,nottype,meansthatitpresupposestheexistenceofaperspectivefromwhichthephysicalworldisseeninordertobringoutthesefacts.Theperspectivethatmakespossiblethenomicallyirreduciblesciences,beingitselfirreducible,coulditselfexist(ifitwerephysical)onlyfromaperspectiveonphysicalreality.Asthissecondperspectiveisessentiallyofthesamekindastheonewearetryingtoexplain,namelyapsycho-logicalorintellectualperspective,thereisnoprospectofanon-viciousregresshere.Wecannowunderstandthemotivationforfull-blownreduction.Atruebasicphysicsrepresentstheworldasitisinitself,andifthespecialscienceswerereducible,thentheexistenceoftheirontologieswouldmakesenseasexpressionsofthephysical,notjustaswaysofseeingorinterpretingit–theycouldbeunderstood“fromthebottomup,”notfromabovedown.Theirreducibilityofthespecialsciencescreatesnoproblemforthedualist,whoseestheexplanatoryendeavorofthephysicalsciencesassomethingcarriedonfromaperspectiveconceptuallyoutsideofthephysicalworld.Norneeditworryaphysicalist,ifhecanreducepsychology,forthenhecouldunderstand“fromthebottomup”theacts(withtheirinternal,intentionalcontents)whichcreatedtheirreducibleontologiesoftheothersciences.Butpsychologyisoneoftheleastlikelyofsciencestobereduced.4.4IstheTalkof“Perspectives”Legitimate?Someonewhowishedtoresistthislineofargumentmightdenytheclaimthatthenomicallyirreduciblesciencescannotbegivenafullyrealistinterpretation,butareaperspectiveonthereality.HemightarguethatthefoundationsofthespecialsciencesarewhatDennett(1991)calls“realpatterns”inreality,andthattheseareasobjectiveasthestructuresoftheultimateandreduciblesciences.Thismissesthepoint.Mypositionisnottodenythatthe“realpatterns”onwhichthespecialsciencesarebasedareobjectiveandgenuine,butthat,aswellasthisfundamentuminre,thosesciencesrequireaninterpretativecomponentwhichtakesthesesimilaritiesandpicksthemoutasinterestingforcertainpurposes.TherelationbetweenanidealphysicsandthenomicallyirreduciblespecialsciencesislikethatbetweenstraightforwardphenomenaandGestaltphenomena.89\nHowardRobinsonEntitiesinphysicsareanalogoustoaperfectlycircularobject,whichneedsnointerpretationtobetakenasacircle:thoseinirreduciblespecialsciencesarelikeaseriesofdiscontinuousdotsormarksarrangedroughlyinacirclewhichoneseesascircular.Twohurricanes,forexample,arenotperfectlysimilarandwouldpresentthemselvesasakindonlytosomeonewithaninterestinweather:platetectonicsexistonlygivenaninterestinthehabitabilityoftheearth.Fromawhollydetachedviewpoint,boththesephenomenacould,perfectlycorrectly,beregardedsimplyasby-productsofmorefundamentalprocesses,andnotasconsti-tutingnaturalkindsatall.Theworldinitselfisacontinuousflowofevents–whichisnottosaythatitstextureiseverywherethesame.Takingsomepointasthestartorendofsomeprocessisonlynon-arbitrarywhenseeninthelightofsomeinterestorconcern.4.5ASurprisingAllySupportformytreatmentof(mostof)thespecialsciencescanbedrawnfromArmstrong’saccountofuniversals(1980:vol.2;1989).Armstrongisarealist,butnotforallproperties,onlyforthoserequiredbybasicscience.Nowitmightbethoughtthatthisincludesthoseinthespecialsciences,butIthinkthatitdoesnot.Arealuniversalisonethatmakesadistinctivecausalcontribution,butnon-microentities,casebycase,addnothingtothecausalcontributionofthemicrobase.WhateverreservationsImayhaveaboutArmstrong’sclosetyingoftheidentityofuniversalsandpropertiestotheircausalpowers,Ithinkitisnotunreasonable,inthiscontext,totakethematterofwhetherauniversal“doeswork”initsparticularinstancesascriterialofwhetherarealuniversalisthereneeded.ThiscanperhapsbereinforcedbyappealtoArmstrong’sclaimthatthere10arenodisjunctiveuniversals(1980:19–23;1989:82–4).Thepropertiesofanyspecialsciencenotrelatedbysimplebridginglawstophysicswillbedisjuncts–perhapsopen-endeddisjuncts–ofmoreatomicuniversals.Thisreinforcesthesenseinwhichirreducibleuniversalsarenotstrictlynecessary:thecorrespondingpredicatesarenecessaryfortheschemesofexplanationthatconstitutethespecialsciences,butpredicates,asopposedtouniversals,arecreaturesofhumanthought11andtalk,andsopresupposethementalperspectiveonthesubjectmatter.4.6TheOptionalityofNon-basicLevelsandtheUnavoidabilityofPsychologyIwanttotakethematterfurtherbydiscussingthesuggestionthat,ifabeingcouldunderstandtheworldinallitsphysical(meaning,onthelevelofphysics)detail,butignoredthegrosserlevels,itwouldbemissingoutonnothing.The90\nDualismpurposeofthediscussionistoshowthatamongstthespecialsciencesonlypsychologycouldnotbeomittedwithoutloss,andthatthisshowstheessentialdifferenceofthementalfromthephysical.Imagineasemi-divinebeingwhofollowseverythingatthelevelofphysics,buttakesnonoticeofanyofthemoremacroscopicpatternsofevents.Becauseofhisintelligence,hecanpredictthepositionofeverythingwithasmuchaccuracyasisinprinciplepossible.Arewetosaythathisfailuretoconcernhimselfwithgrosserpatternsisaformofsubstantiveignorance,orthathemerelyignorescertainmacropatternsthatareessentialtousforunderstandingbecausewecannotgraspthedetail:theyare,forus,anecessaryshorthandandforhim,notnecessaryatall?Someonewhothoughtsuchabeingwassubstantiallyignorantmightstartbyclaimingthatfailuretonoticepatternsandoperativelawsconstitutesignorance.Butsupposethatoursemi-divinitywerecapableofnoticingthesethings,butfoundthemofnointerest,givenhisabilitytodoeverythingintermsofphysics.Itwouldbenecessarytoarguethatthenon-basiclevelswere,insomeway,significantintheirownright,endsinthemselves.Theissueiscloselyparalleltothatoftheirreducibilityofteleologicalexplanation.Supposingthetruthofmech-anism,doteleologicalexplanationsdoextra,non-heuristicwork?Thesituationisatitsmostcrucialforpsychology,asisbroughtoutbyDennett’sdiscussionin“TrueBelievers”(in1987:13–42).Dennettarguesthatevenanomnicompetentobserverwhowasabletopredictthebehaviorofhumansbypredictingthebehavioroftheindividualatomsthatmakethemupwouldneedfolkpsychology.Hewouldneeditifhewishedtounderstandtheutterancesofhumanswhenhetalkedtothem,and,morefundamentally,hewouldneedittounderstandwhathehimselfwasdoing.Sothefolkpsychologicallevelofdescrip-tionisineliminable,thoughitcarriesnofundamentalontologicalclout.TheproblemwithDennett’spositionisthattherecanbenoexplanationofwhywemustadoptthefolkpsychologicalperspective.Ifwearealljustcloudsofatoms,whyareweobligedtoseeourselvesinthisparticularontologicallynon-basicway?Itistruethatwecannotseeourselvesaspeopleorunderstandouractionsunlessweadoptthisperspective,butwhyseeourselvesintheseways?Aneliminativistwouldarguethatitisjustconceptualconservatism.Butifonerejectstheideathatwejusthappentobehookedonthiswayofseeingourselvesandagreethattheapplicabilityofthesecategoriesistrulyfundamental,thenthereistheprob-lemofexplainingwhythisshouldbeso.Areductionistbelievesthatstatementsonthislevelcanbetrue,becausetheyarereducible.Butthisfactdoesnotexplainwhy,amongstallpossiblenon-basiclevelsofdiscourse,thisoneshouldbeunavoidable,ratherthanmerelyavailableifrequired.Itispossibletoarguethatthequestion“whyshouldweseeourselvesaspersons?”answersitself,becausetheuseof“we”alreadypresupposesthepersonalperspective.Butthismissesthepoint.Thebehaviorofthephysicalstructuresthatwecall“people”cannotbeunderstoodinawaythatseemscompleteorremotelyadequatewithoutthepersonalperspective.Physicalisticallyspeaking,thereshouldbeno“we”thatexistsatsomeparticularlevel.But,evenifonetriestothinkinaphysicalistic91\nHowardRobinsonmanner,onecannotavoidthinkingthat,atacertainlevelofcomplexity,thereemergessomethingwhichisneitheramatterofseeingorinterpretingtheorgan-isminacertainwayfromoutside(onpainofregress)norisitjustoneofthoselevelsofcomplexitywhichonemightnoticeorignore.Thereispresentthere,inamannerwhollydifferentfromotherformsofemergentcomplexity(becauseothersareeitherorbothinterpretativeandignorable)somethingofwhichitmakesnosensetosayonemightignoreit.ThisisatleasttheseedofwhatDescartesexpressesinthecogito.Thetruthisthatevenifwewereabletodoallthepredictingthatphysicalomnisciencewouldmakepossible,itwouldbeimpossibletorestrictone’sunder-standingofoneselftothephysicalterms.TheCartesiancertaintythatIthinkisabsolute,notrelativetoadoptingonepossiblebut,likealltherest,optionallevelofdiscourse.Ourexistenceonthepersonallevelisafundamental,notaprag-matic,fact.Thereisnowayitcanbethoughtofasafunctionofacertainwayofthinkingorconceptualizing:itisabasicfact,inthesenseofbeingunavoidableinamorethanpragmaticway,anditcouldnotbethusbasicifthephysicalistontologywerecorrect.4.7WhyBundleDualismWillNotDoIfwhatissaidaboveiscorrect,themindtranscendsthephysicalworldandis,ipsofacto,non-physical.Butthisdoesnotindicatewhetheritisasubstanceoronlyacollectionofstates.Ishallarguethatbundledualismwillnotsufficebecausethiswouldmakeitintoacomplexentityandonlybysupposingittobesimplecanweaccommodatecertainirresistibleintuitionsconcerningpersonalidentity.Thereisalongtradition,datingatleastfromReid,forarguingthattheidentityofpersonsovertimeisnotamatterofconventionordegreeinthewaythattheidentityofother(complex)substancesis.Criticismoftheseargumentsandoftheintuitionsonwhichtheyrest,runningfromHumetoParfit,hasleftuswithaninconclusiveclashofintuitions.Myargumentdoesnotconcernidentitythroughtime,buttheconsequencesforidentityofcertaincounterfactualsconcerningorigin.Itcan,Ihope,therefore,breakthestalematewhichfacesthedebateoverdiachronicidentity.Myclaimwillbethatthebroadlyconventionalistways,whichareusedtodealwithproblemcasesthroughtimeforbothpersonsandmaterialobjects,andwhichcanalsobeemployedincasesofcounterfactualsconcerningoriginforbodies,cannotbeusedforsimilarcounterfactualsconcerningpersonsorminds.Itisnowadaysrespectabletomaintainthatindividualshaveessentialproperties,thoughitissomewhatlessgenerallyagreedthattheyhaveessences.Kripke’sclaimthataparticularwoodentablecouldnothavebeenmadeoficeseemstobewidelyaccepted,sothereisatleastonenecessaryconditionfortheexistenceofthatindividualtable:butwhethertherearenecessaryandsufficientconditions–92\nDualismi.e.anessence–aswellasmerelynecessaryconditionsforitbeingtheobjectitis,ismorecontroversial(Kripke1980:39–53).Evengrantedthatthetablehassomeessentialproperties,itisdoubtfulwhetherithasanessence.Wecanscalesentencesasfollows:1Thistablemighthavebeenmadeofice.2Thistablemighthavebeenmadeofadifferentsortofwood.3Thistablemighthavebeenmadeof95percentofthewooditwasmadeofand5percentofsomeotherwood.Therewillcomeapointalongthespectrumillustratedby(1)and(2)andtowards(3)wherethequestionofwhetherthehypothesizedtablewouldbethesameastheonethatactuallyexistshasnoobviousanswer.Itseemsthatthequestionofwhetherit“really”isthesameonehasnoclearmeaning:itisof,say,75percentthesamematterandof25percentdifferentmatter.Thesearetheonlygenuinefactsinthecase;thequestionofnumericalidentitycanbedecidedinanyconvenientfashion,orleftunresolved.Therewillthusbeapenumbraofcounterfactualcaseswherethequestionofwhethertwothingswouldbethesameisnotamatteroffact.Supposethatagivenhumanindividualhadhadoriginsdifferentfromthosewhichheinfacthad,suchthatwhetherthatdifferenceaffectedwhohewaswasnotobvioustointuition.Whatwouldcountassuchacasemightbeamatterofcontroversy,buttheremustbeone.Perhapsitisunclearwhether,iftherehadbeenacounterparttoJones’sbodyfromthesameeggbutadifferentthoughgenetic-allyidenticalspermfromthesamefather,thepersonthereembodiedwouldhavebeenJones.Somephilosophersmightregarditasobviousthatsamenessofspermisessentialtotheidentityofahumanbodyandtopersonalidentity.Inthatcaseimagineacounterpartsperminwhichsomeofthemoleculesinthespermaredifferent;wouldthatbethesamesperm?Ifonepursuesthematterfarenoughtherewillbeindeterminacywhichwillinfectthatoftheresultingbody.Theremustthereforebesomedifferencesuchthatneithernaturallanguagenorintui-tiontellsuswhetherthedifferencealterstheidentityofthehumanbody;apoint,thatis,wherethequestionofwhetherwehavethesamebodyisnotamatteroffact.Howoneistodescribethesecasesis,insomerespects,amatterofcontroversy.Somephilosophersthinkonecantalkofvagueidentityorpartialidentity,othersthinkthatsuchexpressionsarenonsensical.Idonothavethespacetodiscussthisissue.Iamassuming,however,thatquestionsofhowoneisallowedtousetheconceptofidentityaffectonlythecarewithwhichoneshouldcharacterizethesecases,notanysubstantivematteroffact.Therearecasesofsubstantialoverlapofconstitutioninwhichthatfactistheonlybedrockfactinthecase:thereisnofurtherfactaboutwhethertheyare“really”thesameobject.Iftherewerethentherewouldhavetobeahaecceitasorthisnessbelongingtoandindividuatingeachcomplexphysicalobject,andthisIamassumingtobeimplausibleifnot93\nHowardRobinsonunintelligible.(Moreabouttheconditionsunderwhichhaecceitascanmakesensewillbefoundbelow.)Myclaimisthatnosimilaroverlapofconstitutioncanbeappliedtothecounterfactualidentityofminds.InGeoffreyMadell’swords:“Butwhilemypresentbodycanthushaveitspartialcounterpartinsomepossibleworld,mypresentconsciousnesscannot.AnypresentstateofconsciousnessthatIcanimagineeitherisorisnotmine.Thereisnoquestionofdegreehere”12(1981:xx).Whyisthisso?ImaginethecasewherewearenotsurewhetheritwouldhavebeenJones’sbody–and,hence,Jones–thatwouldhavebeencreatedbytheslightlymodifiedspermandthesameegg.Canwesay,aswewouldforanobjectwithnoconsciousness,thatthestorysomethingthesame,somethingdifferentisthewholestory,thatoverlapofconstitutionisallthereistoit?FortheJonesbodyassuch,thisapproachwoulddoaswellasforanyotherphysicalobject.ButsupposeJones,inreflectivemood,askshimself“ifthathadhappened,wouldIhaveexisted?”Thereareatleastthreeanswershemightgivetohimself:(1)“Ieitherwouldorwouldnot,butIcannottell;”(2)“ThereisnofactofthematterwhetherIwouldorwouldnothaveexisted:itisjustamis-posedquestion;”(3)“Insomeways,ortosomedegree,Iwouldhave,andinsomeways,ortosomedegree,Iwouldnot.Thecreaturewhowouldhaveexistedwouldhavehadakindofoverlapofpsychicconstitutionwithme.”Thethirdanswerparallelstheresponsewewouldgiveinthecaseofbodies.Butasanaccountofthesubjectivesituation,itmakesnosense.Callthecreaturethatwouldhaveemergedfromtheslightlymodifiedsperm“Jones2.”Istheoverlapsuggestionthat,justas,say,85percentofJones2’soriginalbodywouldhavebeenidenticalwithJones’s,about85percentofhispsychiclifewouldhavebeenJones’s?ThatitwouldhavebeenlikeJones’s–indeed,thatJones2mighthavehadapsychiclife100percentlikeJones’s–makesperfectsense,butthathemighthavebeentothatdegree,thesamepsyche–thatJones“85percentexisted”–makesnosense.TakethecaseinwhichJonesandJones2haveexactlysimilarlivesthroughout:which85percentofthe100percentsimilarmentaleventsdotheyshare?NordoesitmakesensetosuggestthatJonesmighthaveparticipatedinthewholeofJones2’spsychiclife,butinaratherghostlyonly85percenttheremanner.Clearly,thenotionofoverlapofnumericallyidenticalpsychicpartscannotbeappliedinthewaythatoverlapofactualbodilypartconstitutionquiteunproblematicallycan.Thismightmakeonetrythesecondanswer.Wecanapplythe“overlap”answertotheJonesbody,butthequestionofwhetherthemindsorsubjectswouldhavebeenthesamehasnoclearsense.Itisdifficulttoseewhyitdoesnot.SupposeJonesfoundoutthathehadoriginallybeenoneoftwins,inthesensethatthezygotefromwhichhedevelopedhaddivided,butthattheotherhalfhaddiedsoonafterwards.Hecanentertainthethoughtthatifithadbeenhishalfthathaddied,hewouldneverhaveexistedasaconsciousbeing,thoughsomeonewould,whoselife,bothinnerandouter,mighthavebeenverysimilartohis.Hemightfeelratherguiltilygratefulthatitwastheotherhalfthatdied.Itwouldbe94\nDualismstrangetothinkthatJonesiswrongtothinkthatthereisamatteroffactaboutthis.Andhowisoneto“manage”thetransitionfromthecasewherethereisamatteroffacttothecasewherethereisnot?Thisonlyleavesuswiththefirstoption.Therehastobeanabsolutematteroffactfromthesubjectivepointofview.Butthephysicalexampleswehaveconsid-eredshowthatwhensomethingisessentiallycomplex,thiscannotbethecase.Whenthereisconstitution,degreeandoverlapofconstitutionareinevitablypossible.Sothemindmustbesimple,andthisispossibleonlyifitissomethinglikeaCartesiansubstance.4.8TwoReflectionsonthisConclusionThefirstreflectionconcernsthedifferencebetweenJones’sfailuretoimaginehisrelationtotheexistenceofJones2,andothermoretraditionalproblemsinper-sonalidentity.Unliketheothercases,Jones’sisnotamatterofwhatonemightcallempatheticdistance.SupposethatmyparentshademigratedtoChinawhilstmymotherwaspreg-nantwithme,andthat,shortlyaftermybirth,bothmyparentshaddied.IwasthentakeninbyChinesefosterparents,livedthroughtherevolutionandendedupbeingbroughtupinwhateverwayanalienwouldhavebeenbroughtupinMao’sChina.Noneofthisperson’spost-uterineexperienceswouldhavebeenlikemine.Itseems,ontheonehand,thatthispersonwouldobviouslyhavebeenme,and,ontheother,thatitisutterlyunclearwhatkindofempatheticconnec-tionIcanfeeltothisother“me.”IfIask,likeJones,“wouldthishavebeenme?,”Iamdividedbetweentheconvictionthat,asthestoryistold,itobviouslywould,andacompleteinabilitytofeelmyselfintothepositionIwouldthenhaveoccupied.Thiskindoffailureofempathyplaysanimportantroleinmanystoriesthataremeanttothrowdoubtontheabsolutenessofpersonalidentity.ItisimportanttotheattempttothrowdoubtonwhetherIamthesamepersonasIwouldbecomeinfiftyyearstime,orwhetherbraindamagewouldrenderme“adifferentperson”inmorethanametaphoricalsense.Itisalsoobviouslysome-thingthatcanbeamatterofdegree:somedifferencesaremoreempatheticallyimaginablethanothers.Inallthesecasesourintuitionsareindecisiveabouttheeffectonidentity.Itisanimportantfactthatproblemsofempathyplaynoroleinmyargument.Thetwinwhomighthavesurvivedinmystead,orthepersonwhowouldhaveexistedifthespermhadbeenslightlydifferent,couldhavehadasexactlysimilarapsychiclifetomineasyoucaretoimagine.ThisshowsthedifferencebetweenthecasesIhavediscussedandtheproblematiccasesthatinvolveidentitythroughtime.Inthosecasestheideaof“similarbutnotquitethesame”getsempiricalpurchase.Myfutureselffeels,inhismemory,much,butnotall,ofwhatInowfeel.Inthesecases,overlapofconsciousconstitutionisclearlyintelligible.Butinthecounterfactualcases,imaginativeorempathetic95\nHowardRobinsondistanceplaysnoessentialrole,andtheaccompanyingrelativityofidentificationgetsnogrip.Secondly,Ithinkthattheargumentisreinforcedbythelightitthrowsontheconceptofhaecceitas.Inthecaseofcomplexphysicalbodiesitisimpossibletoimaginewhatahaecceitaswouldconsistinorhowitrelatestotheotherfeaturesoftheobject,andsothesuggestionthatthereissuchathingseemstobepuremystery-mongering.Bycontrast,inthecaseofmindswedohaveaformofhaecceitaswhich,inasense,weallunderstand,namelysubjectivity.Itisbecauseweintuitivelyunderstandthisthatwefeelwecangiveaclearsensetothesuggestionthatitwould,orwouldnot,havebeenourselvestowhichsome-thinghadhappened,ifithadhappened:andthatwefeelwecanunderstandveryradicalcounterfactuals–e.g.thatImighthavebeenanancientGreekorevenanon-human–whereassuchradicalcounterfactualswhenappliedtomerebodies–e.g.thatthiswoodentablemighthavebeentheothertableinthecornerorevenapyramid–makesnointuitivesense.Itispossibletoarguethatthesuggestionthatmymindmighthavebeeninanotherbodyultimatelymakesnosense,butitmakesaprimafaciesense–itseemstohavecontent–inawaythatasimilarsuggestionformerebodiesdoesnot.Theveryfactthatthecounter-factualsforsubjectsseemtomakesenseexhibitssomethingnotpresentintheothercases,whichisavailabletofunctionintheroleofhaecceitas.OnlywithconsciousnessunderstoodinaCartesianfashioncanhaecceitasbegivenanempiricalinterpretation.4.9AnObjectionOneresponsesometimesmadetothisargumentisthatitiscorrectasanaccount13ofourconceptofthemind,butnotcorrectabouttheactualnatureofthemind.Realityis,sotospeak,deconstructiveoftheconceptthatwehave.SoourconceptualschemedoescommitustosomethingliketheCartesianconceptionofthemind,butwehaveothergroundsforthinkingthatthisisamistake.Asitstands,thisismoreanexpressionofuneasethanaworkedoutobjection.Ishallconsidertwowaysoffillingitout.First,onemightargueasfollows.Ifwesupposethemindtobeonlyacollec-tionofmentalstatesrelatedbyaco-consciousnessrelation,thephenomenologywouldstillseemtobetousasitisinfact.Theargumentdoesnot,therefore,showthatthebundletheoryisfalse,forevenifthebundletheoryweretrue,itwouldseemtousasifweweresimplesubstances.Itcouldbecomparedtowhata“harddeterminist”mightsayaboutfreewill,namelywecannothelpbutfeelwe14haveit,butthefeelingismistaken.Therearetwoproblemswiththisargument.First,itdoesnothelpJonestoanswerhisquestion.Inordertoavoidanswer(1)–thatheeitherwouldorwouldnotbeidenticalwithJones2–hewouldhavetomakesenseofoneoftheotheralternatives,andthisobjectiongiveshimno96\nDualismhelpwiththat.IsthesuggestionthatwhenJonestriestoimagineoverlapofpsychicconstitution,ourconceptspreventhimfromdoingso,but,inreality,suchathingwouldbepossible?Ifso,Idonotthinkthisveryplausible.Itseemstometobearealfactthatthismakesnosense.Myobjectionsabovetotheotheroption–thatthereisnofactofthematter–seemalsotobeuntouched.Secondly,theargumentisquestion-begging.Itisamootpointbetweenthebundletheoristandthesubstantivalistwhethertherecouldbeaco-consciousnessrelationthatwouldproduceanexperientiallyunitedmind.Myargumentsup-portstheviewthatexperientialunityinvolvesasimplesubstanceandsosupportstheviewthatthereisnosuchthingasaself-standingco-consciousnessrelation.Soitisnotpropersimplytoclaimthatitcouldbethesameforusifthebundletheoryweretrue,ifthatconditionisinfactanimpossibleone.Theanalogywithfreewill,thoughillustrativeofwhattheobjectorisdrivingat,doeslittletoshowthatheiscorrect.First,thecoherenceoftheharddeterminist’spositioniscon-troversial.Secondly,thedeterministcangivearationaleforwhywemustfeelfreeintermsoftheconceptualimpossibilityofreplacingone’sownpracticeofdecid-ingbyoneofmerelypredictingone’sownbehavior.Thereseemstobenoparallelexplanationofwhyitseemsallornothingforcounterfactualidentity.Thisisespeciallymysteriousgiventhatitcanseembeamatterofdegreeincases15thatturnonempatheticdistance.Thereisacompletelydifferentwayoffillingtheobjectionout.Itconcernsmyuseofcounterfactuals.CounterfactualsareacontroversialmatterandImakenoattempttodiscussthem.IblatantlyassumethefalsehoodofLewis’scounterpartanalysis,forifJones’squestionwhetherhewouldexistonlyenquiredwhethertherewouldbeacounterpartwhichpossessedstatesverylikehisown,thentherewouldbenophenomenologicalproblem.Allcounterpartsarestrictlydifferentobjects.However,Iamquitehappy,alongwithalmostallotherphilosophers,simplytodenyLewis’stheory.Butitisnotfromthissourcethatthechallengecomes,butfromsomeonewhotakesanon-realistattitudetocounterfactuals.Thereisanempiricisttraditionwhichdeniestruthvaluestocounterfactualsandsaysthattheyexpresspoliciesorattitudes.Therewillbenotruthaboutwhatwouldhavehappenediftherelevantspermhadbeenslightlydifferent.Itisnotpossibletogetdeeplyengagedinadiscussionofcounterfactualshere.Iwouldmaketwopoints.First,mostphilosophersdoacceptarealistaccountofcounterfactuals–theanti-realistviewisnotveryplausible–andtheargumentwouldgothroughforthem.Secondly,theanti-realistapproachhasaweakerandastrongerform.Theweakerversionsimplydeniestruthvaluetocounterfactuals:thereisnofactofthematteraboutwhetheritisaorbthatwouldhavehappenedifChadobtained.Ccouldhaveobtainedand,ifithad,eitheraorb(orsomethingelse)wouldhaveoccurred:thereisjustnotruthfromtheperspectiveoftheactualworldaboutwhichitwouldhavebeen.Thisdoesnotaffectmyargumentatall,whichonlyrequiresthattheonlyoptionsaboutwhatmighthavehappenedareallornothing,notthatthereisafactaboutwhich.Thestrongerversionsaysthatthewholenotionofmighthavebeenotherwiseisaprojectionof97\nHowardRobinsonourmodeofthought–ofourabilitytoimaginethings–notsomethingthatobtainsinreality.Thisisnottosay–asitmightseem–thattheactualworldisnecessary(becausethereisnothingelsethatmighthavebeen)butonlythatallthesemodalcategoriesaremereprojections.Evenifweacceptedthis–whichIdonotrecommend–itwouldnotentirelydeflatetheargument.Itwouldstillshowsomethinginterestingaboutthenatureofmind,namelythatitmadenosensetotreatitinthesamewayasbodieswithinthelogicalspaceofpossibilitythatwecreatebyprojection.Thefactthatwecreatethatspacedoesnotimplythatwhatweexpresswithinitdoesnotreflectrealdifferencesbetweentheobjectsaboutwhichwearetalking.4.10ConclusionMyargumentsinthischapterhavebeeninaCartesianspirit.First,insec-tions4.2–4.6Iarguedthatthethinkingsubjecthastotranscendthephysicalworldaboutwhich(amongotherthings)itthinks.Onlyifastrongreductionismweretruecoulditsthinkingbepartofthatphysicalworld.Then,insections4.7–4.9,Iarguedthatthethinkingsubjecthastobeasimplesubstance,onpainofentertainingincoherentcounterfactuals.Theseargumentscomplementeachother,buttheyarelogicallyindependentandthesecondcanestablishitsconclusiononitsown.Notes1Idonotenterfurtherintoafullerdiscussionofthesepropertieshere,forthatbelongsprincipallytoanexaminationoftheproblemsformaterialism.Forafullerdescriptionofthesepropertiesandabriefoutlineofthestrategiesthatmodernmaterialistshaveemployedtocopewiththem,seeRobinson(1999).2Descartes’SixthMeditationisthelocusclassicusforsubstancedualism.ModerndefensesofthetheorycanbefoundinPopperandEccles(1977),Swinburne(1986),andFoster(1991).HumedevelopshistheoryintheTreatise(BkI,Partiv,Section6)andexpresseshisdissatisfactionwithwhathehassaidintheAppendixtotheTreatise.Thereareseveralmodernphilosopherswhoaccountfortheunityofthemindintermsoftherelationsbetweenmentalevents,andsocouldbesaidtohaveabundletheory,buttheydonottendtobedualists.Parfit(1971;1984)isamaterialistandDainton(2000)isneutralonontologicalquestions.3Itmightbethoughtthattheattributetheoryalreadyhasanaccountoftheunityofthemind,intermsofthedependenceofalltheelementsinagivenbundleonthesamebrain.But,thoughthismaybeacausalexplanation,itisnotananalysis,ofunity.Meredependenceonthesamebraindoesnotconceptuallyguaranteeunityofconsciousness.SeeFoster(1968)inreplytoAyer(1963).4Forthedoubts,seetheAppendixtotheTreatise.98\nDualism5Idiscussembodiment–thoughnotspecificallytheproblemofinteraction–inRobinson(1989).6ExamplesoftranslationreductionismareHempel(1980)andCarnap(1934).7TheclassicsourceforthisisErnestNagel(1961).8Thewithdrawalfromgenuinereductionisminpsychology,then,beganwhenSkinneracceptedthatastimulus-responsemodelwasinadequate,anddevelopedthenotionofoperantconditioning.Whereastheformerrequiredonlymechanisticcausalconcepts,thelatterisirreduciblyteleological.Thebehavioroftheratwhichislearninghowtogetthefoodpelletmayhaveamechanicaldescriptiononalowerlevel,buttheunderstandingofitasoperantconditioninghastobeteleological,foritconcernswhattheratistryingtoachieve,orthepointofitsbehavior.Furthermore,thebehavioristisprevented,byhisownprincipleddisinterestinwhathappensinside,fromhavingviewsaboutthenatureoftheprocessinwhichthelearningisrealized.Thisbringsouttheambiguityoftheconceptofreductionwhenappliedtothephilosophyofmind.Itscentralconcernistoeliminate“theghostinthemachine”–thatis,anythingirreduciblyprivateorsubjective.Thisformofreductionisentirelyirrelevanttoanyofthephysicalsciences.Thesecondelementistheeliminationoranalysisawayofconceptsofakindthathavenoplaceinapurelyphysicalscience.Operantconditioningmeetsthefirstobjectivebutnotthesecond.Itisthebruntoftheargumentofthispartofmychapterthat,contraFodor,thesecondobjectiveisasessentialtothephysicalistasthefirst.9Thesearethemuchdiscussedqualiaobjectionstophysicalism.See,forexample,Jackson(1982),Robinson(1993).10Armstrong’sacceptanceofconjunctiveuniversalsalsoreinforcestheintuitionthatstrongreductionpreservesfullrealismforthespecialsciences.Waterisaconjunctionofinstancesoftheuniversalsofhydrogen-nessandoxygen-nessinacertainspatialarrangement.These,intheirturn,areconjunctionsofmoreatomicuniversals.11Itfollowsfromthis,ofcourse,thatifpsychology(whichincludesnotonlythescience,butourordinarymentalisticconcepts)isnotreducibleinastrongsense,its“properties”areonlypredicatesanditssubjectmatterisinpartcreatedbyanactofthemind–themindnotbeingpresentuntilthatacthasbeenperformed.Armstrong’stheorybecomeslessdifferentfromDennett’sinterpretativetheory,withtheattendantthreatofregress,thanwastheintention.12Madell’sbookisanexcellenttreatmentofthetopicsIdiscussinthissection.13Thisobjectionhasbeenmadetome,ondifferentoccasions,bySimonBlackburn,DerekParfit,andKatalinFarkas.Itisworthnotingthatthisobjectioninvolvesamajorconcession.IftheargumentIhavepresentedshowsthatwearecommittedbythewaywethinkofourselvestoaCartesianconceptoftheself,thiswasnotinvirtueofsomeeasilyrevisabledefinition.Theargumentwasnotaderivationoflogicalconsequencesfromsomenecessaryandsufficientconditionsforbeingasubject,leavingtheoptionofalteringthoseconditions.Itproceededonthebasisofwhatwasconceivableforaconscioussubject.Theassociatedconceptoftheselfmustbeunavoidableina“Kantian”manner.Thesuggestionthatitismistakenis,therefore,aformofskepticalnihilism,whichwecanonlylivethroughbypretendingtoignore.14IowethecomparisonwithfreewilltoKatalinFarkas.15Thereisamorecomplicatedversionoftheargumentpresentedinsections4.6–4.9,whichwouldresisttheobjection.Ibelievethatitcanbearguedthatvaguepredicates99\nHowardRobinsonareneverontologicallybasicandcan,inprinciple,beeliminated.Amongstthesewillbethenotionofidentityundercounterfactualcircumstancesforphysicalbodiesofallkinds.Thereisnorealfactualdifferencebetweenanassertionthatsomephysicalbodywouldhaveexistedifsuchandsuchhadbeendifferent,andanassertionthattherewouldhavebeena“counterpart”bodyofasimilarkindunderthosecircumstances.Thisappliesevenifthecounterfactualchangedoesnotdirectlyinvolvetheobjectinquestion.Butthistreatmentiswhollyunacceptableforsubjects.Supposethat,con-trarytofact,someonehadcoughedontheothersideoftheworldjustbeforeyouwereconceived.Ontheprinciplethatappliestobodies,thereisnofactualdifferencebetweenthepropositionthatyouwouldstillhavecomeintoexistence,andthepropositionthatsomeonewiththesamequalitiesasyouwouldhave.Asthetwinexampleshows,thisdifferenceisrealandnoteliminable.Thefullversionofthisargumentisnotinprint,butfordiscussionofsomeoftherelevantissuesconcerningvagueness,seeRobinson(2001).ReferencesArmstrong,D.M.(1980).UniversalsandScientificRealism(2vols).Cambridge:Cam-bridgeUniversityPress.——(1989).Universals:AnOpinionatedIntroduction.Boulder,CO:WestviewPress.Ayer,A.J.(1963).“TheConceptofaPerson.”InTheConceptofaPersonandotherEssays.London:Macmillan:82–128.Carnap,R.(1934).TheUnityofScience.London:KeganPaul.Dainton,B.(2000).StreamofConsciousness.London:Routledge.Dennett,D.(1987).TheIntentionalStance.Cambridge,MA:MITPress.——(1991).“RealPatterns.”JournalofPhilosophy,89:27–51.Descartes,R.(1984–5).ThePhilosophicalWritingsofDescartes,trans.J.Cottingham,R.Stoothof,andD.Murdoch(2vols).Cambridge:CambridgeUniversityPress.Fodor,J.(1974).“SpecialSciencesortheDisunityofScienceasaWorkingHypothesis.”Synthese,28:77–115.Foster,J.(1968).“PsychophysicalCausalRelations.”AmericanPhilosophicalQuarterly,5.——(1991).TheImmaterialSelf.London:Routledge.Hempel,C.G.(1980).“TheLogicalAnalysisofPsychology.”InN.Block(ed.),ReadingsinPhilosophyofPsychology,vol.1.London:Methuen:14–23.(OriginallypublishedinFrenchin1935.)Hume,D.(1978).ATreatiseonHumanNature,ed.P.H.Nidditch.Oxford:ClarendonPress.Jackson,F.(1982).“EpiphenomenalQualia.”PhilosophicalQuarterly,32:127–36.Kripke,S.(1980).NamingandNecessity.Oxford:Blackwell.Madell,G.(1981).TheIdentityoftheSelf.Edinburgh:EdinburghUniversityPress.Nagel,E.(1961).TheStructureofScience.London:RoutledgeandKeganPaul.Parfit,D.(1971).“PersonalIdentity.”PhilosophicalReview,80:3–27.——(1984).ReasonsandPersons.Oxford:ClarendonPress.Popper,K.R.andEccles,J.C.(1977).TheSelfanditsBrain.Berlin:SpringerInternational.Robinson,H.(1989).“ADualistAccountofEmbodiment.”InSmythiesandBeloff(1989):43–57.100\nDualism——(1993).“TheAnti-materialistStrategyandthe‘KnowledgeArgument’.”InH.Robinson(ed.),ObjectionstoPhysicalism.Oxford:ClarendonPress:159–83.——(1999).“MaterialismandtheMind–BodyProblem.”InE.Graig(ed.),TheRoutledgeEncyclopaediaofPhilosophy.London:Routledge.——(2001).“Vagueness,Realism,LanguageandThought.”InT.HorganandM.Potric(eds.),EssaysonVagueness.Oxford:OxfordUniversityPress.Smythies,J.R.andBeloff,J.(eds.)(1989).TheCaseforDualism.Charlottesville:Univer-sityofVirginiaPress.Swinburne,R.(1986).TheEvolutionoftheSoul.Oxford:ClarendonPress.101\nTheBlackwellGuidetoPhilosophyofMindDavidJ.ChalmersEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter5ConsciousnessanditsPlaceinNatureDavidJ.Chalmers15.1IntroductionConsciousnessfitsuneasilyintoourconceptionofthenaturalworld.Onthemostcommonconceptionofnature,thenaturalworldisthephysicalworld.Butonthemostcommonconceptionofconsciousness,itisnoteasytoseehowitcouldbepartofthephysicalworld.Soitseemsthattofindaplaceforconsciousnesswithinthenaturalorder,wemusteitherreviseourconceptionofconsciousness,orreviseourconceptionofnature.Intwentieth-centuryphilosophy,thisdilemmaisposedmostacutelyinC.D.Broad’sTheMindanditsPlaceinNature(1925).Thephenomenaofmind,forBroad,arethephenomenaofconsciousness.Thecentralproblemisthatoflocatingmindwithrespecttothephysicalworld.Broad’sexhaustivediscussionoftheproblemculminatesinataxonomyofseventeendifferentviewsofthemental–2physicalrelation.OnBroad’staxonomy,aviewmightseethementalasnon-existent(“delusive”),asreducible,asemergent,orasabasicpropertyofasubstance(a“differentiating”attribute).Thephysicalmightbeseeninoneofthesamefourways.Soafour-by-fourmatrixofviewsresults.(TheseventeenthentryarisesfromBroad’sdivisionofthesubstance/substanceviewaccordingtowhetheronesubstanceortwoisinvolved.)Attheend,threeviewsareleftstanding:thoseonwhichmentalityisanemergentcharacteristicofeitheraphysicalsubstanceoraneutralsubstance,whereinthelattercase,thephysicalmightbeeitheremergentordelusive.InthischapterItakemycuefromBroad,approachingtheproblemofcon-sciousnessbyastrategyofdivide-and-conquer.IwillnotadoptBroad’scategor-ies:ourunderstandingofthemind–bodyproblemhasadvancedsincethe1920s,anditwouldbenicetothinkthatwehaveabetterunderstandingofthecrucialissues.Onmyview,themostimportantviewsonthemetaphysicsofconscious-nesscanbedividedalmostexhaustivelyintosixclasses,whichIwilllabel“type102\nConsciousnessanditsPlaceinNatureA”through“typeF.”Threeofthese(AthroughC)involvebroadlyreductiveviews,seeingconsciousnessasaphysicalprocessthatinvolvesnoexpansionofaphysicalontology.Theotherthree(DthroughF)involvebroadlynon-reductiveviews,onwhichconsciousnessinvolvessomethingirreducibleinnature,andrequiresexpansionorreconceptionofaphysicalontology.Thediscussionwillbecastatanabstractlevel,givinganoverviewofthemetaphysicallandscape.Ratherthanengagingtheempiricalscienceofconscious-ness,ordetailedphilosophicaltheoriesofconsciousness,Iwillbeexaminingsomegeneralclassesintowhichtheoriesofconsciousnessmightfall.Iwillnotpretendtobeneutralinthisdiscussion.Ithinkthateachofthereductiveviewsisincorrect,whileeachofthenon-reductiveviewsholdssomepromise.Sothefirstpartofthischaptercanbeseenasanextendedargumentagainstreductiveviewsofconsciousness,whilethesecondpartcanbeseenasaninvestigationofwherewegofromthere.5.2TheProblemTheword“consciousness”isusedinmanydifferentways.Itissometimesusedfortheabilitytodiscriminatestimuli,ortoreportinformation,ortomonitorinternalstates,ortocontrolbehavior.Wecanthinkofthesephenomenaasposingthe“easyproblems”ofconsciousness.Theseareimportantphenomena,andthereismuchthatisnotunderstoodaboutthem,buttheproblemsofexplainingthemhavethecharacterofpuzzlesratherthanmysteries.Thereseemstobenodeepprobleminprinciplewiththeideathataphysicalsystemcouldbe“conscious”inthesesenses,andthereisnoobviousobstacletoaneventualexplanationofthesephenomenainneurobiologicalorcomputationalterms.Thehardproblemofconsciousnessistheproblemofexperience.Humanbeingshavesubjectiveexperience:thereissomethingitisliketobethem.Wecansaythatabeingisconsciousinthissense–orisphenomenallyconscious,asitissometimesput–whenthereissomethingitisliketobethatbeing.Amentalstateisconsciouswhenthereissomethingitisliketobeinthatstate.Consciousstatesincludestatesofperceptualexperience,bodilysensation,mentalimagery,emotionalexperience,occurrentthought,andmore.Thereissomethingitisliketoseeavividgreen,tofeelasharppain,tovisualizetheEiffelTower,tofeeladeepregret,andtothinkthatoneislate.Eachofthesestateshasaphenomenalcharacter,withphenomenalproperties(orqualia)characterizingwhatitisliketo3beinthestate.Thereisnoquestionthatexperienceiscloselyassociatedwithphysicalpro-cessesinsystemssuchasbrains.Itseemsthatphysicalprocessesgiverisetoexperience,atleastinthesensethatproducingaphysicalsystem(suchasabrain)withtherightphysicalpropertiesinevitablyyieldscorrespondingstatesofexperi-ence.Buthowandwhydophysicalprocessesgiverisetoexperience?Whydonot103\nDavidJ.Chalmerstheseprocessestakeplace“inthedark,”withoutanyaccompanyingstatesofexperience?Thisisthecentralmysteryofconsciousness.Whatmakestheeasyproblemseasy?Fortheseproblems,thetaskistoexplaincertainbehavioralorcognitivefunctions:thatis,toexplainhowsomecausalroleisplayedinthecognitivesystem,ultimatelyintheproductionofbehavior.Toexplaintheperformanceofsuchafunction,oneneedonlyspecifyamechanismthatplaystherelevantrole.Andthereisgoodreasontobelievethatneuralorcomputationalmechanismscanplaythoseroles.Whatmakesthehardproblemhard?Here,thetaskisnottoexplainbehavioralandcognitivefunctions:evenonceonehasanexplanationofalltherelevantfunctionsinthevicinityofconsciousness–discrimination,integration,access,report,control–theremaystillremainafurtherquestion:whyistheperformanceofthesefunctionsaccompaniedbyexperience?Becauseofthis,thehardproblemseemstobeadifferentsortofproblem,requiringadifferentsortofsolution.Asolutiontothehardproblemwouldinvolveanaccountoftherelationbetweenphysicalprocessesandconsciousness,explainingonthebasisofnaturalprincipleshowandwhyitisthatphysicalprocessesareassociatedwithstatesofexperience.Areductiveexplanationofconsciousnesswillexplainthiswhollyonthebasisofphysicalprinciplesthatdonotthemselvesmakeanyappealtocon-4sciousness.Amaterialist(orphysicalist)solutionwillbeasolutiononwhichconsciousnessisitselfseenasaphysicalprocess.Anon-materialist(ornon-physicalist)solutionwillbeasolutiononwhichconsciousnessisseenasnon-physical(evenifcloselyassociatedwithphysicalprocesses).Anon-reductivesolutionwillbeoneonwhichconsciousness(orprinciplesinvolvingconsciousness)isadmittedasabasicpartoftheexplanation.Itisnaturaltohopethattherewillbeamaterialistsolutiontothehardproblemandareductiveexplanationofconsciousness,justastherehavebeenreductiveexplanationsofmanyotherphenomenainmanyotherdomains.Butconsciousnessseemstoresistmaterialistexplanationinawaythatotherphenom-enadonot.Thisresistancecanbeencapsulatedinthreerelatedargumentsagainstmaterialism,summarizedinwhatfollows.5.3ArgumentsAgainstMaterialism55.3.1TheexplanatoryargumentThefirstargumentisgroundedinthedifferencebetweentheeasyproblemsandthehardproblem,ascharacterizedabove:theeasyproblemsconcerntheexplana-tionofbehavioralandcognitivefunctions,butthehardproblemdoesnot.Onecanarguethatbythecharacterofphysicalexplanation,physicalaccountsexplainonlystructureandfunction,wheretherelevantstructuresarespatio-temporalstructures,andtherelevantfunctionsarecausalrolesintheproductionofa104\nConsciousnessanditsPlaceinNaturesystem’sbehavior.Andonecanargueasabovethatexplainingstructuresandfunctionsdoesnotsufficetoexplainconsciousness.Ifso,nophysicalaccountcanexplainconsciousness.Wecancallthistheexplanatoryargument:(1)Physicalaccountsexplainatmoststructureandfunction.(2)Explainingstructureandfunctiondoesnotsufficetoexplainconsciousness.——(3)Nophysicalaccountcanexplainconsciousness.Ifthisisright,thenwhilephysicalaccountscansolvetheeasyproblems(whichinvolveonlyexplainingfunctions),somethingmoreisneededtosolvethehardproblem.Itwouldseemthatnoreductiveexplanationofconsciousnesscouldsucceed.Andifweaddthepremisethatwhatcannotbephysicallyexplainedisnotitselfphysical(thiscanbeconsideredanadditionalfinalstepoftheexplanat-oryargument),thenmaterialismaboutconsciousnessisfalse,andthenaturalworldcontainsmorethanthephysicalworld.Ofcourse,thissortofargumentiscontroversial.Butbeforeexaminingvariouswaysofresponding,itisusefultoexaminetwocloselyrelatedargumentsthatalsoaimtoestablishthatmaterialismaboutconsciousnessisfalse.65.3.2TheconceivabilityargumentAccordingtothisargument,itisconceivablethattherebeasystemthatisphysicallyidenticaltoaconsciousbeing,butthatlacksatleastsomeofthatbeing’sconsciousstates.Suchasystemmightbeazombie:asystemthatisphysicallyidenticaltoaconsciousbeingbutthatlacksconsciousnessentirely.Itmightalsobeaninvert,withsomeoftheoriginalbeing’sexperiencesreplacedbydifferentexperiences,orapartialzombie,withsomeexperiencesabsent,oracombinationthereof.Thesesystemswilllookidenticaltoanormalconsciousbeingfromthethird-personperspective:inparticular,theirbrainprocesseswillbemolecule-for-moleculeidenticalwiththeoriginal,andtheirbehaviorwillbeindistinguishable.Butthingswillbedifferentfromthefirst-personpointofview.Whatitisliketobeaninvertorapartialzombiewilldifferfromwhatitisliketobetheoriginalbeing.Andthereisnothingitisliketobeazombie.Thereislittlereasontobelievethatzombiesexistintheactualworld.Butmanyholdthattheyareatleastconceivable:wecancoherentlyimaginezombies,andthereisnocontradictionintheideathatrevealsitselfevenonreflection.Asanextensionoftheidea,manyholdthatthesamegoesforazombieworld:auniversephysicallyidenticaltoours,butinwhichthereisnoconsciousness.Somethingsimilarappliestoinvertsandotherduplicates.Fromtheconceivabilityofzombies,proponentsoftheargumentinfertheirmetaphysicalpossibility.Zombiesareprobablynotnaturallypossible:theyprobably105\nDavidJ.Chalmerscannotexistinourworld,withitslawsofnature.Buttheargumentholdsthatzombiescouldhaveexisted,perhapsinaverydifferentsortofuniverse.Forexample,itissometimessuggestedthatGodcouldhavecreatedazombieworld,ifhehadsochosen.Fromhere,itisinferredthatconsciousnessmustbenon-physical.Ifthereisametaphysicallypossibleuniversethatisphysicallyidenticaltooursbutthatlacksconsciousness,thenconsciousnessmustbeafurther,non-physicalcomponentofouruniverse.IfGodcouldhavecreatedazombieworld,then(asKripkeputsit)aftercreatingthephysicalprocessesinourworld,hehadtodomoreworktoensurethatitcontainedconsciousness.Wecanputtheargument,initssimplestform,asfollows:(1)Itisconceivablethattherebezombies.(2)Ifitisconceivablethattherebezombies,itismetaphysicallypossiblethattherebezombies.(3)Ifitismetaphysicallypossiblethattherebezombies,thenconsciousnessisnon-physical.——(4)Consciousnessisnon-physical.AsomewhatmoregeneralandpreciseversionoftheargumentappealstoP,theconjunctionofallmicrophysicaltruthsabouttheuniverse,andQ,anarbitraryphenomenaltruthabouttheuniverse.(Here“∧”represents“and”and“¬”represents“not”.)(1)ItisconceivablethatP∧¬Q.(2)IfitisconceivablethatP∧¬Q,itismetaphysicallypossiblethatP∧¬Q.(3)IfitismetaphysicallypossiblethatP∧¬Q,thenmaterialismisfalse.——(4)Materialismisfalse.75.3.3TheknowledgeargumentAccordingtotheknowledgeargument,therearefactsaboutconsciousnessthatarenotdeduciblefromphysicalfacts.Someonecouldknowallthephysicalfacts,beaperfectreasoner,andstillbeunabletoknowallthefactsaboutconsciousnessonthatbasis.FrankJackson’scanonicalversionoftheargumentprovidesavividillustration.Onthisversion,Maryisaneuroscientistwhoknowseverythingthereistoknowaboutthephysicalprocessesrelevanttocolorvision.ButMaryhasbeenbrought8upinablack-and-whiteroom(onanalternativeversion,sheiscolorblind)andhasneverexperiencedred.Despiteallherknowledge,itseemsthatthereissomethingveryimportantaboutcolorvisionthatMarydoesnotknow:shedoesnotknowwhatitisliketoseered.Evencompletephysicalknowledgeand106\nConsciousnessanditsPlaceinNatureunrestrictedpowersofdeductiondonotenablehertoknowthis.Later,ifshecomestoexperienceredforthefirsttime,shewilllearnanewfactofwhichshewaspreviouslyignorant:shewilllearnwhatitisliketoseered.Jackson’sversionoftheargumentcanbeputasfollows(herethepremisesconcernMary’sknowledgewhenshehasnotyetexperiencedred):(1)Maryknowsallthephysicalfacts.(2)Marydoesnotknowallthefacts.——(3)Thephysicalfactsdonotexhaustallthefacts.Onecanputtheknowledgeargumentmoregenerally:(1)Therearetruthsaboutconsciousnessthatarenotdeduciblefromphysicaltruths.(2)Iftherearetruthsaboutconsciousnessthatarenotdeduciblefromphys-icaltruths,thenmaterialismisfalse.——(3)Materialismisfalse.5.3.4TheshapeoftheargumentsThesethreesortsofargumentarecloselyrelated.Theyallstartbyestablishinganepistemicgapbetweenthephysicalandphenomenaldomains.Eachdeniesacertainsortofcloseepistemicrelationbetweenthedomains:arelationinvolvingwhatwecanknow,orconceive,orexplain.Inparticular,eachofthemdeniesacertainsortofepistemicentailmentfromphysicaltruthsPtothephenomenaltruthsQ:deducibilityofQfromP,orexplainabilityofQintermsofP,orconceivingofQuponreflectiveconceivingofP.Perhapsthemostbasicsortofepistemicentailmentisapriorientailment,orimplication.Onthisnotion,PimpliesQwhenthematerialconditionalP⊃Qisapriori;thatis,whenasubjectcanknowthatifPisthecasethenQisthecase,withjustificationindependentofexperience.AllofthethreeargumentsabovecanbeseenasmakingacaseagainstanapriorientailmentofQbyP.IfasubjectwhoknowsonlyPcannotdeducethatQ(astheknowledgeargumentsuggests),orifonecanrationallyconceiveofPwithoutQ(astheconceivabilityargumentsug-gests),thenitseemsthatPdoesnotimplyQ.TheexplanatoryargumentcanbeseenasturningontheclaimthatanimplicationfromPtoQwouldrequireafunctionalanalysisofconsciousness,andthattheconceptofconsciousnessisnotafunctionalconcept.Afterestablishinganepistemicgap,theseargumentsproceedbyinferringanontologicalgap,whereontologyconcernsthenatureofthingsintheworld.Theconceivabilityargumentinfersfromconceivabilitytometaphysicalpossibility;the107\nDavidJ.Chalmersknowledgeargumentinfersfromfailureofdeducibilitytodifferenceinfacts;andtheexplanatoryargumentinfersfromfailureofphysicalexplanationtonon-physicality.Onemightsaythattheseargumentsinferfromafailureofepistemicentailmenttoafailureofontologicalentailment.Theparadigmaticsortofonto-logicalentailmentisnecessitation:PnecessitatesQwhenthematerialconditionalP⊃Qismetaphysicallynecessary,orwhenitismetaphysicallyimpossibleforPtoholdwithoutQholding.ItiswidelyagreedthatmaterialismrequiresthatPnecessitatesalltruths(perhapswithminorqualifications).Soiftherearephenom-enaltruthsQthatPdoesnotnecessitate,thenmaterialismisfalse.Wemightcalltheseargumentsepistemicargumentsagainstmaterialism.EpistemicargumentsarguablydescendfromDescartes’sargumentsagainstmaterialism(althoughthesehaveaslightlydifferentform),andaregiventheirfirstthorough9airinginBroad’sbook,whichcontainselementsofallthreeargumentsabove.Thegeneralformofanepistemicargumentagainstmaterialismisasfollows:(1)Thereisanepistemicgapbetweenphysicalandphenomenaltruths.(2)Ifthereisanepistemicgapbetweenphysicalandphenomenaltruths,thenthereisanontologicalgap,andmaterialismisfalse.——(3)Materialismisfalse.Ofcourse,thiswayoflookingatthingsoversimplifiesmatters,andabstractsaway10fromthedifferencesbetweenthearguments.Thesamegoesforthepreciseanalysisintermsofimplicationandnecessitation.Nevertheless,thisanalysispro-videsausefullensthroughwhichtoseewhattheargumentshaveincommon,andthroughwhichtoanalyzevariousresponsestothearguments.Thereareroughlythreewaysthatamaterialistmightresisttheepistemicarguments.Atype-Amaterialistdeniesthatthereistherelevantsortofepistemicgap.Atype-Bmaterialistacceptsthatthereisanunclosableepistemicgap,butdeniesthatthereisanontologicalgap.Andatype-Cmaterialistacceptsthatthereisadeepepistemicgap,butholdsthatitwilleventuallybeclosed.Inwhatfollows,Idiscussallthreeofthesestrategies.5.4Type-AMaterialismAccordingtotype-Amaterialism,thereisnoepistemicgapbetweenphysicalandphenomenaltruths;oratleast,anyapparentepistemicgapiseasilyclosed.Ac-cordingtothisview,itisnotconceivable(atleastonreflection)thattherebeduplicatesofconsciousbeingsthathaveabsentorinvertedconsciousstates.Onthisview,therearenophenomenaltruthsofwhichMaryisignorantinprinciplefrominsideherblack-and-whiteroom(whensheleavestheroom,shegainsatmostanability).Andonthisview,onreflectionthereisno“hardproblem”of108\nConsciousnessanditsPlaceinNatureexplainingconsciousnessthatremainsonceonehassolvedtheeasyproblemsof11explainingthevariouscognitive,behavioral,andenvironmentalfunctions.Type-Amaterialismsometimestakestheformofeliminativism,holdingthatconsciousnessdoesnotexist,andthattherearenophenomenaltruths.Itsome-timestakestheformofanalyticfunctionalismorlogicalbehaviorism,holdingthatconsciousnessexists,wheretheconceptof“consciousness”isdefinedinwhollyfunctionalorbehavioralterms(e.g.,wheretobeconsciousmightbetohavecertainsortsofaccesstoinformation,and/orcertainsortsofdispositionstomakeverbalreports).Forourpurposes,thedifferencebetweenthesetwoviewscanbeseenasterminological.Bothagreethatweareconsciousinthesenseofhavingthefunc-tionalcapacitiesofaccess,report,control,andthelike;andtheyagreethatwearenotconsciousinanyfurther(non-functionallydefined)sense.Theanalyticfunc-tionalistthinksthatordinarytermssuchas“conscious”shouldbeusedinthefirstsortofsense(expressingafunctionalconcept),whiletheeliminativistthinksthattheyshouldbeusedinthesecond.Beyondthisterminologicaldisagreementabouttheuseofexistingtermsandconcepts,thesubstanceoftheviewsisthesame.Somephilosophersandscientistswhodonotexplicitlyembraceeliminativism,analyticfunctionalism,andthelikeareneverthelessrecognizablytype-Amaterial-ists.Thecharacteristicfeatureofthetype-Amaterialististheviewthatonreflec-tionthereisnothinginthevicinityofconsciousnessthatneedsexplainingoverandaboveexplainingthevariousfunctions:toexplainthesethingsistoexplaineverythinginthevicinitythatneedstobeexplained.Therelevantfunctionsmaybequitesubtleandcomplex,involvingfine-grainedcapacitiesforaccess,self-monitoring,report,control,andtheirinteraction,forexample.Theymayalsobetakentoincludeallsortsofenvironmentalrelations.Andtheexplanationofthesefunctionswillprobablyinvolvemuchneurobiologicaldetail.Soviewsthatareputforwardasrejectingfunctionalismonthegroundsthatitneglectsbiologyorneglectstheroleoftheenvironmentmaystillbetype-Aviews.Onemightthinkthatthereisroominlogicalspaceforaviewthatdenieseventhissortofbroadlyfunctionalistviewofconsciousness,butstillholdsthatthereisnoepistemicgapbetweenphysicalandphenomenaltruths.Inpractice,thereappearstobelittleroomforsuchaview,forreasonsthatIwilldiscussundertype12C,andtherearefewexamplesofsuchviewsinpractice.SoIwilltakeitforgrantedthatatype-Aviewisonethatholdsthatexplainingthefunctionsexplainseverything,andwillclassotherviewsthatholdthatthereisnounclosableepistemicgapundertypeC.Theobviousproblemwithtype-Amaterialismisthatitappearstodenythemanifest.Itisanuncontestedtruththatwehavethevariousfunctionalcapacitiesofaccess,control,report,andthelike,andthesephenomenaposeuncontestedexplananda(phenomenainneedofexplanation)forascienceofconsciousness.Butinaddition,itseemstobeafurthertruththatweareconscious,andthisphenomenonseemstoposeafurtherexplanandum.Itisthisexplanandumthatraisestheinterestingproblemsofconsciousness.Toflatlydenythefurthertruth,ortodenywithoutargumentthatthereisahardproblemofconsciousnessover109\nDavidJ.Chalmersandabovetheeasyproblems,wouldbetomakeahighlycounterintuitiveclaimthatbegstheimportantquestions.Thisisnottosaythathighlycounterintuitiveclaimsarealwaysfalse,buttheyneedtobesupportedbyextremelystrongargu-ments.Sothecrucialquestionis:arethereanycompellingargumentsfortheclaimthat,onreflection,explainingthefunctionsexplainseverything?Type-Amaterialistsoftenarguebyanalogy.Theypointoutthatinotherareasofscience,weacceptthatexplainingthevariousfunctionsexplainsthephenom-ena,soweshouldacceptthesamehere.Inresponse,anopponentmaywellacceptthatinotherdomainsthefunctionsareallweneedtoexplain.Inexplain-inglife,forexample,theonlyphenomenathatpresentthemselvesasneedingexplanationarephenomenaofadaptation,growth,metabolism,reproduction,andsoon,andthereisnothingelsethatevencallsoutforexplanation.Buttheopponentholdsthatthecaseofconsciousnessisdifferentandpossiblyunique,preciselybecausethereissomethingelse,phenomenalexperience,thatcallsoutforexplanation.Thetype-Amaterialistmusteitherdenyeventheappearanceofafurtherexplanandum,whichseemstodenytheobvious,oraccepttheapparentdisanalogyandgivefurthersubstantialargumentsforwhy,contrarytoappear-ances,onlythefunctionsneedtobeexplained.Atthispoint,type-Amaterialistsoftenpressadifferentsortofanalogy,holdingthatatvariouspointsinthepast,thinkersheldthattherewasananalogousepistemicgapforotherphenomena,butthattheseturnedouttobephysicallyexplained.Forexample,Dennett(1996)suggeststhatavitalistmighthaveheldthattherewasafurther“hardproblem”oflifeoverandaboveexplainingthebiologicalfunction,butthatthiswouldhavebeenmisguided.Onexaminingthecases,however,theanalogiesdonotsupportthetype-Amaterialist.Vitaliststypicallyaccepted,implicitlyorexplicitly,thatthebiologicalfunctionsinquestionwerewhatneededexplaining.Theirvitalismarosebecausetheythoughtthatthefunctions(adaptation,growth,reproduction,andsoon)wouldnotbephysicallyexplained.Sothisisquitedifferentfromthecaseofconsciousness.ThedisanalogyisveryclearinthecaseofBroad.Broadwasavitalistaboutlife,holdingthatthefunctionswouldrequireanon-mechanicalexplanation.Butatthesametime,heheldthatinthecaseoflife,unlikethecaseofconsciousness,theonlyevidencewehaveforthephenomenonisbehavioral,andthat“beingalive”meansexhibitingcertainsortsofbehavior.Othervitalistswerelessexplicit,butveryfewofthemheldthatsomethingmorethanthefunctionsneededexplaining(exceptconsciousnessitself,insomecases).Ifavitalisthadheldthis,theobviousreplywouldhavebeenthatthereisnoreasonto13believeinsuchanexplanandum.Sothereisnoanalogyhere.Sotheseargumentsbyanalogyhavenoforceforthetype-Amaterialist.Inothercases,itwasalwaysclearthatstructureandfunctionexhaustedtheapparentexplananda,apartfromthosetieddirectlytoconsciousnessitself.Sothetype-Amaterialistneedstoaddresstheapparentfurtherexplananduminthecaseofconsciousnessheadon:eitherflatlydenyingit,orgivingsubstantialargumentstodissolveit.110\nConsciousnessanditsPlaceinNatureSomeargumentsfortype-Amaterialistsproceedindirectly,bypointingouttheunsavorymetaphysicalorepistemologicalconsequencesofrejectingtheview:e.g.,thattherejectionleadstodualism,ortoproblemsinvolvingknow-14ledgeofconsciousness.Anopponentwilleitherembracetheconsequencesordenythattheyareconsequences.Aslongastheconsequencesarenotcompletelyuntenable,thenforthetype-Amaterialisttomakeprogress,thissortofargumentneedstobesupplementedbyasubstantialdirectargumentagainstthefurtherexplanandum.Suchdirectargumentsaresurprisinglyhardtofind.Manyargumentsfortype-Amaterialismenduppresupposingtheconclusionatcrucialpoints.Forexample,itissometimesargued(e.g.,Rey1995)thatthereisnoreasontopostulatequalia,sincetheyarenotneededtoexplainbehavior;butthisargumentpresupposesthatonlybehaviorneedsexplaining.Theopponentwillholdthatqualiaareanexplanandumintheirownright.Similarly,Dennett’s(1991)useof“heterophenomenology”(verbalreports)astheprimarydatatogroundhistheoryofconsciousnessappearstorestontheassumptionthatthesereportsarewhatneedexplaining,orthattheonly“seemings”thatneedexplainingaredispositionstoreactandreport.Onewaytoarguefortype-Amaterialismistoarguethatthereissomeinterme-diateXsuchthat(i)explainingfunctionssufficestoexplainX,and(ii)explainingXsufficestoexplainconsciousness.OnepossibleXhereisrepresentation:itisoftenheldboththatconsciousstatesarerepresentationalstates,representingthingsintheworld,andthatwecanexplainrepresentationinfunctionalterms.Ifso,itmayseemtofollowthatwecanexplainconsciousnessinfunctionalterms.Onexamination,though,thisargumentappealstoanambiguityinthenotionofrepresentation.Thereisanotionoffunctionalrepresentation,onwhichPisrepresentedroughlywhenasystemrespondstoPand/orproducesbehaviorappropriateforP.Inthissense,explainingfunctioningmayexplainrepresenta-tion,butexplainingrepresentationdoesnotexplainconsciousness.Thereisalsoanotionofphenomenalrepresentation,onwhichPisrepresentedroughlywhenasystemhasaconsciousexperienceasifP.Inthissense,explainingrepresentationmayexplainconsciousness,butexplainingfunctioningdoesnotexplainrepresen-tation.Eitherway,theepistemicgapbetweenthefunctionalandthephenomenalremainsaswideasever.SimilarsortsofequivocationcanbefoundwithotherX’sthatmightbeappealedtohere,suchas“perception”or“information.”Perhapsthemostinterestingargumentsfortype-Amaterialismarethosethatarguethatwecangiveaphysicalexplanationofourbeliefsaboutconsciousness,suchasthebeliefthatweareconscious,thebeliefthatconsciousnessisafurtherexplanandum,andthebeliefthatconsciousnessisnon-physical.Fromhereitisarguedthatoncewehaveexplainedthebelief,wehavedoneenoughtoexplain,ortoexplainaway,thephenomenon(e.g.,Clark2000,Dennettforthcoming).Hereitisworthnotingthatthisonlyworksifthebeliefsthemselvesarefunctionallyanalyzable;Chalmers(2002a)givesreasontodenythis.Butevenifoneacceptsthatbeliefsareultimatelyfunctional,thisclaimthenreducestotheclaimthatexplaining111\nDavidJ.Chalmersourdispositionstotalkaboutconsciousness(andthelike)explainseverything.Anopponentwilldenythisclaim:explainingthedispositionstoreportmayremovethethird-personwarrant(basedonobservationofothers)foracceptingafurtherexplanandum,butitdoesnotremovethecrucialfirst-personwarrant(fromone’sowncase).Still,thisisastrategythatdeservesextendeddiscussion.Atacertainpoint,thedebatebetweentype-Amaterialistsandtheiropponentsusuallycomesdowntointuition:mostcentrally,theintuitionthatconsciousness(inanon-functionallydefinedsense)exists,orthatthereissomethingthatneedstobeexplained(overandaboveexplainingthefunctions).Thisclaimdoesnotgainitssupportfromargument,butfromasortofobservation,alongwithrebuttalofcounterarguments.Theintuitionappearstobesharedbythelargemajorityofphilosophers,scientists,andothers;anditissostrongthattodenyit,atype-Amaterialistneedsexceptionallypowerfularguments.Theresultisthatevenamongmaterialists,type-Amaterialistsareadistinctminority.155.5Type-BMaterialismAccordingtotype-Bmaterialism,thereisanepistemicgapbetweenthephysicalandphenomenaldomains,butthereisnoontologicalgap.Accordingtothisview,zombiesandthelikeareconceivable,buttheyarenotmetaphysicallypossible.Onthisview,Maryisignorantofsomephenomenaltruthsfrominsideherroom,butneverthelessthesetruthsconcernanunderlyingphysicalreality(whensheleavestheroom,shelearnsoldfactsinanewway).Andonthisview,whilethereisahardproblemdistinctfromtheeasyproblems,itdoesnotcorre-spondtoadistinctontologicaldomain.Themostcommonformoftype-Bmaterialismholdsthatphenomenalstatescanbeidentifiedwithcertainphysicalorfunctionalstates.Thisidentityisheldtobeanalogousincertainrespects(althoughperhapsnotinallrespects)withthe16identitybetweenwaterandH2O,orbetweengenesandDNA.Theseidentitiesarenotderivedthroughconceptualanalysis,butarediscoveredempirically:theconceptwaterisdifferentfromtheconceptH2O,buttheyarefoundtorefertothesamethinginnature.Onthetype-Bview,somethingsimilarappliestoconsciousness:theconceptofconsciousnessisdistinctfromanyphysicalorfunc-tionalconcepts,butwemaydiscoverempiricallythattheserefertothesamethinginnature.Inthisway,wecanexplainwhythereisanepistemicgapbetweenthephysicalandphenomenaldomains,whiledenyinganyontologicalgap.Thisyieldstheattractivepossibilitythatwecanacknowledgethedeepepistemicproblemsofconsciousnesswhileretainingamaterialistworldview.Althoughsuchaviewisattractive,itfacesimmediatedifficulties.Thesedifficult-iesstemfromthefactthatthecharacteroftheepistemicgapwithconsciousnessseemstodifferfromthatofepistemicgapsinotherdomains.Forastart,theredonotseemtobeanalogsoftheepistemicargumentsaboveinthecasesofwater,112\nConsciousnessanditsPlaceinNaturegenes,andsoon.Toexplaingenes,wemerelyhavetoexplainwhysystemsfunctionacertainwayintransmittinghereditarycharacteristics;toexplainwater,wehavetoexplainwhyasubstancehasacertainobjectivestructureandbehavior.Givenacompletephysicaldescriptionoftheworld,Marywouldbeabletodeducealltherelevanttruthsaboutwaterandaboutgenes,bydeducingwhichsystemshavetheappropriatestructureandfunction.Finally,itseemsthatwecannotcoherentlyconceiveofaworldphysicallyidenticaltoourown,inwhichthereisnowater,orinwhichtherearenogenes.Sothereisnoepistemicgapbetweenthecompletephysicaltruthabouttheworldandthetruthaboutwaterandgenesthatisanalogoustotheepistemicgapwithconsciousness.(Except,perhaps,forepistemicgapsthatderivefromtheepistemicgapforconsciousness.Forexample,perhapsMarycouldnotdeduceorexplaintheper-ceptualappearanceofwaterfromthephysicaltruthabouttheworld.Butthiswouldjustbeanotherinstanceoftheproblemweareconcernedwith,andsocannothelpthetype-Bmaterialist.)Soitseemsthatthereissomethinguniqueaboutthecaseofconsciousness.WecanputthisbysayingthatwhiletheidentitybetweengenesandDNAisempir-ical,itisnotepistemicallyprimitive:theidentityisitselfdeduciblefromthecompletephysicaltruthabouttheworld.Bycontrast,thetype-Bmaterialistmustholdthattheidentificationbetweenconsciousnessandphysicalorfunctionalstatesisepistemicallyprimitive:theidentityisnotdeduciblefromthecompletephysicaltruth.(Ifitwerededucible,type-Amaterialismwouldbetrueinstead.)Sotheidentitybetweenconsciousnessandaphysicalstatewillbeasortofprimitiveprincipleinone’stheoryoftheworld.Here,onemightsuggestthatsomethinghasgonewrong.Elsewhere,theonlysortofplacethatonefindsthissortofprimitiveprincipleisinthefundamentallawsofphysics.Indeed,itisoftenheldthatthissortofprimitiveness–theinabilitytobededucedfrommorebasicprinciples–isthemarkofafundamentallawofnature.Ineffect,thetype-Bmaterialistrecognizesaprinciplethathastheepistemicstatusofafundamentallaw,butgivesittheontologicalstatusofanidentity.Anopponentwillholdthatthismoveismoreakintotheftthantohonesttoil:elsewhere,identificationsaregroundedinexplanations,andprimitiveprinciplesareacknowledgedasfundamentallaws.Itisnaturaltosuggestthatthesameshouldapplyhere.Ifoneacknowledgestheepistemicallyprimitiveconnectionbetweenphysicalstatesandconsciousnessasafundamentallaw,itwillfollowthatconsciousnessisdistinctfromanyphysicalproperty,sincefundamentallawsalwaysconnectdistinctproperties.Sotheusualstandardwillleadtooneofthenon-reductiveviewsdiscussedinthesecondhalfofthischapter.Bycontrast,thetype-Bmaterialisttakesanobservedconnectionbetweenphysicalandphenomenalstates,unexplainableinmorebasicterms,andsuggeststhatitisanidentity.Thissuggestionismadelargelyinordertopreserveapriorcommitmenttomaterialism.Unlessthereisanindependentcaseforprimitiveidentities,thesuggestionwillseematbestadhocandmysterious,andatworstincoherent.113\nDavidJ.ChalmersAtype-Bmaterialistmightrespondinvariousways.First,some(e.g.,Papineau1993)suggestthatidentitiesdonotneedtobeexplained,soarealwaysprimitive.Butwehaveseenthatidentitiesinotherdomainscanatleastbededucedfrommorebasictruths,andsoarenotprimitiveintherelevantsense.Secondly,some(e.g.,BlockandStalnaker1999)suggestthateventruthsinvolvingwaterandgenescannotbededucedfromunderlyingphysicaltruths.Thismatteristoo17complextogointohere(seeChalmersandJackson2001foraresponse),butonecannotethattheepistemicargumentsoutlinedatthebeginningsuggestaverystrongdisanalogybetweenconsciousnessandothercases.Thirdly,some(e.g.,Loar1990/1997)acknowledgethatidentitiesinvolvingconsciousnessareunlikeotheridentitiesbybeingepistemicallyprimitive,butseektoexplainthisuniquenessbyappealingtouniquefeaturesoftheconceptofconsciousness.Thisresponseisperhapsthemostinteresting,andIwillreturntoit.Thereisanotherlinethatatype-Bmaterialistcantake.Onecanfirstnotethatanidentitybetweenconsciousnessandphysicalstatesisnotstrictlyrequiredforamaterialistposition.Rather,onecanplausiblyholdthatmaterialismaboutcon-sciousnesssimplyrequiresthatphysicalstatesnecessitatephenomenalstates,inthatitismetaphysicallyimpossibleforthephysicalstatestobepresentwhilethephenomenalstatesareabsentordifferent.Thatis,materialismrequiresthatentailmentsP⊃Qbenecessary,wherePisthecompletephysicaltruthabouttheworldandQisanarbitraryphenomenaltruth.Atthispoint,atype-BmaterialistcannaturallyappealtotheworkofKripke(1980),whichsuggeststhatsometruthsarenecessarilytruewithoutbeingapriori.Forexample,Kripkesuggeststhat“waterisH2O”isnecessary–trueinallpossibleworlds–butnotknowableapriori.Here,atype-BmaterialistcansuggestthatP⊃QmaybeaKripkeanaposteriorinecessity,like“waterisH2O”(thoughitshouldbenotedthatKripkehimselfdeniesthisclaim).Ifso,thenwewouldexpecttheretobeanepistemicgap,sincethereisnoapriorientailmentfromPtoQ,butatthesametimetherewillbenoontologicalgap.Inthisway,Kripke’sworkcanseemtobejustwhatthetype-Bmaterialistneeds.Here,someoftheissuesthatarosepreviouslyariseagain.Onecanarguethatinotherdomains,necessitiesarenotepistemicallyprimitive.Thenecessarycon-nectionbetweenwaterandH2Omaybeaposteriori,butitcanitselfbededucedfromacompletephysicaldescriptionoftheworld(onecandeducethatwaterisidenticaltoH2O,fromwhichitfollowsthatwaterisnecessarilyH2O).ThesameappliestotheothernecessitiesthatKripkediscusses.Bycontrast,thetype-Bmaterialistmustholdthattheconnectionbetweenphysicalstatesandconscious-nessisepistemicallyprimitive,inthatitcannotbededucedfromthecompletephysicaltruthabouttheworld.Again,onecansuggestthatthissortofprimitivenecessaryconnectionismysteriousandadhoc,andthattheconnectionshouldinsteadbeviewedasafundamentallawofnature.Iwilldiscussfurtherproblemswiththesenecessitiesinthenextsection.Buthere,itisworthnotingthatthereisasenseinwhichanytype-Bmaterialistpositiongivesuponreductiveexplanation.Eveniftype-Bmaterialismistrue,we114\nConsciousnessanditsPlaceinNaturecannotgiveconsciousnessthesamesortofexplanationthatwegivegenesandthelike,inpurelyphysicalterms.Rather,ourexplanationwillalwaysrequireexplanatorilyprimitiveprinciplestobridgethegapfromthephysicaltothephenomenal.Theexplanatorystructureofatheoryofconsciousness,onsuchaview,willbeverymuchunlikethatofamaterialisttheoryinotherdomains,andverymuchliketheexplanatorystructureofthenon-reductivetheoriesdescribedbelow.Bylabelingtheseprinciplesidentitiesornecessitiesratherthanlaws,theviewmaypreservetheletterofmaterialism;butbyrequiringprimitivebridgingprinciples,itsacrificesmuchofmaterialism’sspirit.5.6TheTwo-DimensionalArgumentAgainstType-BMaterialismAsdiscussedabove,thetype-Bmaterialistholdsthatzombieworldsandthelikeareconceivable(thereisnocontradictioninP¬Q)butarenotmetaphysicallypossible.Thatis,P⊃Qisheldtobeanaposteriorinecessity,akintosuchaposteriorinecessitiesas“waterisH2O.”WecananalyzethispositioninmoredepthbytakingacloserlookattheKripkeancasesofaposteriorinecessity.Thismaterialissomewhattechnical(hencetheseparatesection)andcanbeskippedifnecessaryonafirstreading.ItisoftensaidthatinKripkeancases,conceivabilitydoesnotentailpossibility:itisconceivablethatwaterisnotH2O(inthatitiscoherenttosupposethatwaterisnotH2O),butitisnotpossiblethatwaterisnotH2O.Butatthesametime,itseemsthatthereissomepossibilityinthevicinityofwhatoneconceives.WhenoneconceivesthatwaterisnotH2O,oneconceivesofaworldW(theXYZ-world)inwhichthewateryliquidintheoceansisnotH2O,butXYZ,say.ThereisnoreasontodoubtthattheXYZ-worldismetaphysicallypossible.IfKripkeiscorrect,theXYZ-worldisnotcorrectlydescribedasoneinwhichwaterisXYZ.Nevertheless,thisworldisrelevanttothetruthof“waterisXYZ”inaslightlydifferentway,whichcanbebroughtoutasfollows.OnecansaythattheXYZ-worldcouldturnouttobeactual,inthatforallweknowapriori,theactualworldisjustliketheXYZ-world.AndonecansaythatiftheXYZ-worldturnsouttobeactual,itwillturnoutthatwaterisXYZ.Similarly:ifwehypothesizethattheXYZ-worldisactual,weshouldrationallyconcludeonthatbasisthatwaterisnotH2O.Thatis,thereisadeepepistemicconnectionbetweentheXYZ-worldand“waterisnotH2O.”EvenKripkeallowsthatitisepistemicallypossiblethatwaterisnotH2O(inthebroadsensethatthisisnotruledoutapriori).ItseemsthattheepistemicpossibilitythattheXYZ-worldisactualisaspecificinstanceoftheepistemicpossibilitythatwaterisnotH2O.Here,weadoptaspecialattitudetoaworldW.WethinkofWasanepistemicpossibility:asawaytheworldmightactuallybe.Whenwedothis,weconsiderWasactual.WhenwethinkofWasactual,itmaymakeagivensentenceStrueor115\nDavidJ.Chalmersfalse.Forexample,whenthinkingoftheXYZ-worldasactual,itmakes“waterisnotH2O”true.ThisisbroughtoutintheintuitivejudgmentthatifWturnsouttobeactual,itwillturnoutthatwaterisnotH2O,andthattheepistemicpossibilitythatWisactualisaninstanceoftheepistemicpossibilitythatwaterisH2O.Bycontrast,onecanalsoconsideraworldWascounterfactual.Whenwedothis,weacknowledgethatthecharacteroftheactualworldisalreadyfixed,andwethinkofWasacounterfactualwaythingsmighthavebeenbutarenot.IfKripkeisright,thenifthewaterystuffhadbeenXYZ,XYZwouldneverthelessnothavebeenwater.SowhenweconsidertheXYZ-worldascounterfactual,itdoesnotmake“waterisnotH2O”true.Consideredascounterfactual,wedescribetheXYZ-worldinlightoftheactual-worldfactthatwaterisH2O,andweconcludethatXYZisnotwaterbutmerelywaterystuff.Theseresultsdonotconflict:theysimplyinvolvetwodifferentwaysofconsideringanddescribingpossibleworlds.Kripke’sclaimsconsidercounterfactualevaluationofworlds,whereastheclaimsinthepreviousparagraphconcerntheepistemicevaluationofworlds.18Onecanformalizethisusingtwo-dimensionalsemantics.WecansaythatifWconsideredasactualmakesStrue,thenWverifiesS,andthatifWconsideredascounterfactualmakesStrue,thenWsatisfiesS.Verificationinvolvestheepistemicevaluationofworlds,whereassatisfactioninvolvesthecounterfactualevaluationofworlds.Correspondingly,wecanassociateSwithdifferentintensions,orfunc-tionsfromworldstotruthvalues.Theprimary(orepistemic)intensionofSisafunctionthatistrueataworldWiffWverifiesS,andthesecondary(orsubjunct-ive)intensionisafunctionthatistrueataworldWifWsatisfiesS.Forexample,whereSis“waterisnotH2O,”andWistheXYZ-world,wecansaythatWverifiesSbutWdoesnotsatisfyS;andwecansaythattheprimaryintensionofSistrueatW,butthesecondaryintensionofSisfalseatW.Withthisinmind,onecansuggestthatwhenastatementSisconceivable–thatis,whenitstruthcannotberuledoutapriori–thenthereissomeworldthatverifiesS,orequivalently,thereissomeworldatwhichS’sprimaryintensionistrue.Thismakesintuitivesense:whenSisconceivable,Srepresentsanepistemicpossibility.Itisnaturaltosuggestthatcorrespondingtotheseepistemicpossibil-itiesarespecificworldsW,suchthatwhentheseareconsideredasepistemicpossibilities,theyverifyS.Thatis,Wissuchthatintuitively,ifWturnsouttobeactual,itwouldturnoutthatS.ThismodelseemstofitallofKripke’scases.Forexample,Kripkeholdsthatitisanaposteriorinecessitythatheatisthemotionofmolecules.Soitisconceiv-ableintherelevantsensethatheatisnotthemotionofmolecules.Correspond-ingtothisconceivablescenarioisaworldWinwhichheatsensationsarecausedbysomethingotherthanthemotionofmolecules.Wrepresentsanepistemicpossibility:andwecansaythatifWturnsouttobeactual,itwillturnoutthatheatisnotthemotionofmolecules.Thesamegoesinmanyothercases.ThemoralisthattheseKripkeanphenomenainvolvetwodifferentwaysofthinkingofpossibleworlds,withjustoneunderlyingspaceofworlds.116\nConsciousnessanditsPlaceinNatureIfthisprincipleisappliedtothecaseoftype-Bmaterialism,troubleimmedi-atelyarises.Asbefore,letPbethecompletephysicaltruthabouttheworld,andletQbeaphenomenaltruth.LetussaythatSisconceivablewhenthetruthof19Sisnotruledoutapriori.Thenonecanmountanargumentasfollows:(1)P∧¬Qisconceivable.(2)IfP∧¬Qisconceivable,thenaworldverifiesP∧¬Q.(3)IfaworldverifiesP∧¬Q,thenaworldsatisfiesP∧¬Qortype-Fmonismistrue.(4)IfaworldsatisfiesP∧¬Q,materialismisfalse.——(5)Materialismisfalseortype-Fmonismistrue.Thetype-Bmaterialistgrantspremise(1):todenythiswouldbetoaccepttype-Amaterialism.Premise(2)isaninstanceofthegeneralprinciplediscussedabove.Premise(4)canbetakenasdefinitiveofmaterialism.Asforpremise(3):ingeneralonecannotimmediatelymovefromaworldverifyingStoaworldsatisfyingS,asthecaseof“waterisH2O”(andtheXYZ-world)suggests.ButinthecaseofP∧¬Q,alittlereflectiononthenatureofPandQtakesusinthatdirection,asfollows.First,Q.Here,itisplausiblethatifWverifies“thereisconsciousness,”thenWsatisfies“thereisconsciousness,”andviceversa.ThiscorrespondstotheKripkeanpointthatinthecaseofconsciousness,thereisnodistinctionanalogoustothatbetweenwateritselfandmerewaterystuff.Toputitintuitively,ifWverifies“thereisconsciousness,”itcontainssomethingthatatleastfeelsconscious,andifsomethingfeelsconscious,itisconscious.Onecanholdmoregenerallythattheprimaryandsecondaryintensionsofourcorephenomenalconceptsarethesame(seeChalmers2002a).ItfollowsthatifworldWverifies¬Q,Wsatisfies¬Q.(Thisclaimisnotrequiredfortheargumenttogothrough,butitisplausibleandmakesthingsmorestraightforward.)Second,P.Atype-BmaterialistmightseektoevadetheargumentbyarguingthatwhileWverifiesP,itdoesnotsatisfyP.Onreflection,theonlywaythismightworkisasfollows.IfaworldverifiesP,itmusthaveatleastthestructureoftheactualphysicalworld.TheonlyreasonwhyWmightnotsatisfyPisthatitlackstheintrinsicpropertiesunderlyingthisstructureintheactualworld.(Onthisview,theprimaryintensionofaphysicalconceptpicksoutwhateverpropertyplaysacertainroleinagivenworld,andthesecondaryintensionpicksouttheactualintrinsicpropertyacrossallworlds.)IfthisdifferenceinWisresponsiblefortheabsenceofconsciousnessinW,itfollowsthatconsciousnessintheactualworldisnotnecessitatedbythestructuralaspectsofphysics,butbyitsunderlyingintrinsicnature.ThisispreciselythepositionIcalltype-Fmonism,or“panprotopsychism.”Type-Fmonismisaninterestingandimportantposition,butitismuchmoreradicalthantype-Bmaterialismasusuallyconceived,andIcountitasadifferentposition.Iwilldeferdiscussionofthereasoningandoftheresultingpositionuntillater.117\nDavidJ.ChalmersItfollowsthatpremise(4)iscorrect.IfaworldverifiesP∧¬Q,theneitheraworldsatisfiesP∧¬Q,ortype-Fmonismistrue.Settingasidetype-Fmonismfornow,itfollowsthatthephysicaltruthaboutourworlddoesnotnecessitatethephenomenaltruth,andmaterialismisfalse.Thisconclusionisineffectaconsequenceof(i)theclaimthatP∧¬Qiscon-ceivable(intherelevantsense),(ii)theclaimthatwhenSisconceivable,thereisaworldthatverifiesS,and(iii)somestraightforwardreasoning.Amaterialistmightrespondbydenying(i),butthatissimplytodenytherelevantepistemicgapbetweenthephysicalandthephenomenal,andsotodenytype-Bmaterialism.Ithinkthereislittlepromiseforthetype-Bmaterialistindenyingthereasoninginvolvedin(iii).Sotheonlyhopeforthetype-Bmaterialististodenythecentral20thesis(ii).Todothis,atype-Bmaterialistcoulddenythecoherenceofthedistinctionbetweenverificationandsatisfaction,oracceptthatthedistinctioniscoherentbutdenythatthesis(ii)holdseveninthestandardKripkeancases,oracceptthatthesis(ii)holdsinthestandardKripkeancasesbutdenythatitholdsinthespecialcaseofconsciousness.Thefirsttwooptionsdeserveexploration,butIthinktheyareultimatelyunpromising,asthedistinctionandthethesisappeartofittheKripkeanphenomenaverywell.Ultimately,Ithinkatype-Bmaterialistmustholdthatthecaseofconsciousnessisspecial,andthatthethesisthatholdselsewherefailshere.Onthisview,theaposteriorinecessitiesconnectingthephysicalandphenom-enaldomainsaremuchstrongerthanthoseinotherdomainsinthattheyareverifiedbyallworlds.Elsewhere,Ihavecalledtheseunusualaposteriorinecessit-iesstrongnecessities,andhavearguedthatthereisnogoodreasontobelievetheyexist.Aswithexplanatorilyprimitiveidentities,theyappeartobeprimitivefactspostulatedinanadhocway,largelyinordertosaveatheory,withnosupportfromcaseselsewhere.Further,onecanarguethatthisviewleadstoanunderlyingmodaldualism,withindependentprimitivedomainsoflogicalandmetaphysicalpossibility;andonecanarguethatthisisunacceptable.Perhapsthemostinterestingresponsefromatype-Bmaterialististoacknow-ledgethatstrongnecessitiesareuniquetothecaseofconsciousness,andtotrytoexplainthisuniquenessintermsofspecialfeaturesofourconceptualsystem.Forexample,ChristopherHill(1997)hasarguedthatonecanpredicttheepistemicgapinthecaseofconsciousnessfromthefactthatphysicalconceptsandphe-nomenalconceptshavedifferentconceptualroles.BrianLoar(1990/1997)hasappealedtotheclaimthatphenomenalconceptsarerecognitionalconceptsthatlackcontingentmodesofpresentation.JosephLevine(2000)hasarguedthatphenomenalconceptshavenon-ascriptivemodesofpresentation.Inresponse,Ihaveargued(Chalmers1999)thattheseresponsesdonotwork,andthatthere21aresystematicreasonswhytheycannotwork.Butitislikelythatfurtherattemptsinthisdirectionwillbeforthcoming.Thisremainsoneofthekeyareasofdebateonthemetaphysicsofconsciousness.Overall,myownviewisthatthereislittlereasontothinkthatexplanatorilyprimitiveidentitiesorstrongnecessitiesexist.Thereisnogoodindependent118\nConsciousnessanditsPlaceinNaturereasontobelieveinthem:thebestreasontopostulatethemistosavematerial-ism,butinthecontextofadebateoverwhethermaterialismistruethisreasoningisuncompelling,especiallyifthereareviablealternatives.Nevertheless,furtherinvestigationintothekeyissuesunderlyingthisdebateislikelytobephilosophic-allyfruitful.5.7Type-CMaterialismAccordingtotype-Cmaterialism,thereisadeepepistemicgapbetweenthephysicalandphenomenaldomains,butitisclosableinprinciple.Onthisview,zombiesandthelikeareconceivableforusnow,buttheywillnotbeconceivableinthelimit.Onthisview,itcurrentlyseemsthatMarylacksinformationaboutthephenomenal,butinthelimittherewouldbenoinformationthatshelacks.Andonthisview,whilewecannotseenowhowtosolvethehardprobleminphysicalterms,theproblemissolvableinprinciple.Thisviewisinitiallyveryattractive.Itappearstoacknowledgethedeepex-planatorygapwithwhichweseemtobefaced,whileatthesametimeallowingthattheapparentgapmaybeduetoourownlimitations.Therearedifferentversionsoftheview.Nagel(1974)hassuggestedthatjustasthePresocraticscouldnothaveunderstoodhowmattercouldbeenergy,wecannotunderstandhowconsciousnesscouldbephysical,butaconceptualrevolutionmightallowtherelevantunderstanding.Churchland(1997)suggeststhatevenifwecannotnowimaginehowconsciousnesscouldbeaphysicalprocess,thatissimplyapsycho-logicallimitationonourpartthatfurtherprogressinsciencewillovercome.VanGulick(1993)suggeststhatconceivabilityargumentsarequestion-begging,sinceoncewehaveagoodexplanationofconsciousness,zombiesandthelikewillnolongerbeconceivable.McGinn(1989)hassuggestedthattheproblemmaybeunsolvablebyhumansbecauseofdeeplimitationsinourcognitiveabilities,butthatitneverthelesshasasolutioninprinciple.Onewaytoputtheviewisasfollows.Zombiesandthelikeareprimafacieconceivable(forusnow,withourcurrentcognitiveprocesses),buttheyarenotideallyconceivable(underidealizedrationalreflection).Orwecouldsay:phe-nomenaltruthsarededucibleinprinciplefromphysicaltruths,butthededucibilityisakintothatofacomplextruthofmathematics:itisaccessibleinprinciple(perhapsaccessibleapriori),butisnotaccessibletousnow,perhapsbecausethereasoningrequirediscurrentlybeyondus,orperhapsbecausewedonotcur-rentlygraspalltherequiredphysicaltruths.Ifthisisso,thenitwillappeartousthatthereisagapbetweenphysicalprocessesandconsciousness,buttherewillbenogapinnature.Despiteitsappeal,Ithinkthatthetype-Cviewisinherentlyunstable.Uponexamination,itturnsouteithertobeuntenable,ortocollapseintooneoftheotherviewsonthetable.Inparticular,itseemsthattheviewmustcollapseinto119\nDavidJ.Chalmersaversionoftype-Amaterialism,type-Bmaterialism,type-Ddualism,ortype-Fmonism,andsoisnotultimatelyadistinctoption.Onewaytoholdthattheepistemicgapmightbeclosedinthelimitistoholdthatinthelimit,wewillseethatexplainingthefunctionsexplainseverything,andthatthereisnofurtherexplanandum.Itisatleastcoherenttoholdthatwecurrentlysufferfromsomesortofconceptualconfusionorunclaritythatleadsustobelievethatthereisafurtherexplanandum,andthatthissituationcouldbeclearedupbybetterreasoning.Iwillcountthispositionasaversionoftype-Amaterialism,nottype-Cmaterialism:itisobviouslycloselyrelatedtostandardtype-Amaterialism(themaindifferenceiswhetherwehaveyethadtherelevantinsight),andthesameissuesarise.Likestandardtype-Amaterialism,thisviewultimatelystandsorfallswiththestrengthof(actualandpotential)first-orderargumentsthatdissolveanyapparentfurtherexplanandum.Oncetype-Amaterialismissetaside,thepotentialoptionsforclosingtheepistemicgaparehighlyconstrained.Theseconstraintsaregroundedinthenatureofphysicalconcepts,andinthenatureoftheconceptofconsciousness.Thebasicproblemhasalreadybeenmentioned.First:physicaldescriptionsoftheworldcharacterizetheworldintermsofstructureanddynamics.Secondly:fromtruthsaboutstructureanddynamics,onecandeduceonlyfurthertruthsaboutstructureanddynamics.Andthirdly:truthsaboutconsciousnessarenottruthsaboutstructureanddynamics.Butwecantakethesestepsoneatatime.First,amicrophysicaldescriptionoftheworldspecifiesadistributionofpar-ticles,fields,andwavesinspaceandtime.Thesebasicsystemsarecharacterizedbytheirspatio-temporalproperties,andpropertiessuchasmass,charge,andquan-tumwavefunctionstate.Theselatterpropertiesareultimatelydefinedintermsofspacesofstatesthathaveacertainabstractstructure(e.g.,thespaceofcontinu-ouslyvaryingrealquantities,orofHilbertspacestates),suchthatthestatesplayacertaincausalrolewithrespecttootherstates.Wecansubsumespatio-temporaldescriptionsanddescriptionsintermsofpropertiesintheseformalspacesundertherubricofstructuraldescriptions.Thestateofthesesystemscanchangeovertimeinaccordwithdynamicprinciplesdefinedovertherelevantproperties.Theresultisadescriptionoftheworldintermsofitsunderlyingspatio-temporalandformalstructure,anddynamicevolutionoverthisstructure.Sometype-Cmaterialistsholdwedonotyethaveacompletephysics,sowecannotknowwhatsuchaphysicsmightexplain.Butherewedonotneedtohaveacompletephysics:wesimplyneedtheclaimthatphysicaldescriptionsareintermsofstructureanddynamics.Thispointisgeneralacrossphysicaltheories.Suchnoveltheoriesasrelativity,quantummechanics,andthelikemayintroducenewstructures,andnewdynamicsoverthosestructures,butthegeneralpoint(andthegapwithconsciousness)remains.Atype-Cmaterialistmightholdthattherecouldbenewphysicaltheoriesthatgobeyondstructureanddynamics.Butgiventhecharacterofphysicalexplanation,itisunclearwhatsortoftheorythiscouldbe.Novelphysicalpropertiesarepostu-latedfortheirpotentialinexplainingexistingphysicalphenomena,themselves120\nConsciousnessanditsPlaceinNaturecharacterizedintermsofstructureanddynamics,anditseemsthatstructureanddynamicsalwayssufficehere.Onepossibilityisthatinsteadofpostulatingnovelproperties,physicsmightendupappealingtoconsciousnessitself,inthewaythatsometheoristsholdthatquantummechanicsdoes.Thispossibilitycannotbeexcluded,butitleadstoaviewonwhichconsciousnessisitselfirreducible,andisthereforetobeclassedinanon-reductivecategory(typeDortypeF).Thereisoneappealtoa“completephysics”thatshouldbetakenseriously.Thisistheideathatcurrentphysicscharacterizesitsunderlyingproperties(suchasmassandcharge)intermsofabstractstructuresandrelations,butitleavesopentheirintrinsicnatures.Onthisview,acompletephysicaldescriptionoftheworldmustalsocharacterizetheintrinsicpropertiesthatgroundthesestructuresandrelations;andoncesuchintrinsicpropertiesareinvoked,physicswillgobeyondstructureanddynamics,insuchawaythattruthsaboutconsciousnessmaybeentailed.Therelevantintrinsicpropertiesareunknowntous,buttheyareknowableinprinciple.Thisisanimportantposition,butitispreciselytheposi-tiondiscussedundertypeF,soIdeferdiscussionofituntilthen.Secondly,whatcanbeinferredfromthissortofdescriptionintermsofstruc-tureanddynamics?Alow-levelmicrophysicaldescriptioncanentailallsortsofsurprisingandinterestingmacroscopicproperties,aswiththeemergenceofchem-istryfromphysics,ofbiologyfromchemistry,ormoregenerallyofcomplexemergentbehaviorsincomplexsystemstheory.Butinallthesecases,thecomplexpropertiesthatareentailedareneverthelessstructuralanddynamic:theydescribecomplexspatio-temporalstructuresandcomplexdynamicpatternsofbehavioroverthosestructures.Sothesecasessupportthegeneralprinciplethat,fromstructureanddynamics,onecaninferonlystructureanddynamics.Atype-Cmaterialistmightsuggesttherearesometruthsthatarenotthem-selvesstructural-dynamicalthatareneverthelessimpliedbyastructural-dynamicaldescription.Itmightbeargued,perhaps,thattruthsaboutrepresentationorbeliefhavethischaracter.Butaswesawearlier,itseemsclearthatanysenseinwhichthesetruthsareimpliedbyastructural-dynamicdescriptioninvolvesatacitlyfunctionalsenseofrepresentationorofbelief.Thisiswhatwewouldexpect:ifclaimsinvolvingthesecanbeseen(onconceptualgrounds)tobetrueinvirtueofastructural-dynamicdescriptionsholding,thenotionsinvolvedmustthemselvesbestructural-dynamic,atsomelevel.OnemightholdthatthereissomeintermediatenotionX,suchthattruthsaboutXholdinvirtueofstructural-dynamicdescriptions,andtruthsaboutcon-sciousnessholdinvirtueofX.Butasinthecaseoftype-Amaterialism,eitherXisfunctionallyanalyzable(inthebroadsense),inwhichcasethesecondstepfails,orXisnotfunctionallyanalyzable,inwhichcasethefirststepfails.Thisisbroughtoutclearlyinthecaseofrepresentation:forthenotionoffunctionalrepresentation,thefirststepfails,andforthenotionofphenomenalrepresenta-tion,thesecondstepfails.Sothissortofstrategycanonlyworkbyequivocation.Thirdly,doesexplainingordeducingcomplexstructureanddynamicssufficetoexplainordeduceconsciousness?Itseemsclearlynot,fortheusualreasons.Mary121\nDavidJ.Chalmerscouldknowfromherblack-and-whiteroomallaboutthespatio-temporalstruc-tureanddynamicsoftheworldatalllevels,butthiswillnottellherwhatitisliketoseered.Foranycomplexmacroscopicstructuralordynamicdescriptionofasystem,onecanconceiveofthatdescriptionbeinginstantiatedwithoutcon-sciousness.Andexplainingstructureanddynamicsofahumansystemisonlytosolvetheeasyproblems,whileleavingthehardproblemsuntouched.Toresistthislaststep,anopponentwouldhavetoholdthatexplainingstructureanddynamicstherebysufficestoexplainconsciousness.Theonlyremotelytenablewaytodothiswouldbetoembracetype-Amaterialism,whichwehavesetaside.Atype-Cmaterialistmightsuggestthatinsteadofleaningondynamics(asatype-Amaterialistdoes),onecouldleanonstructure.Here,spatio-temporalstruc-tureseemsveryunpromising:toexplainasystem’ssize,shape,position,motion,andsoonisclearlynottoexplainconsciousness.Afinalpossibilityisleaningonthestructurepresentinconsciousstatesthemselves.Consciousstateshavestruc-ture:thereisbothinternalstructurewithinasinglecomplexconsciousstate,andtherearepatternsofsimilaritiesanddifferencesbetweenconsciousstates.Butthisstructureisadistinctivelyphenomenalstructure,quitedifferentinkindfromthespatio-temporalandformalstructurepresentinphysics.Thestructureofacom-plexphenomenalstateisnotspatio-temporalstructure(althoughitmayinvolvetherepresentationofspatio-temporalstructure),andthesimilaritiesanddiffer-encesbetweenphenomenalstatesarenotformalsimilaritiesanddifferences,butdifferencesbetweenspecificphenomenalcharacters.Thisisreflectedinthefactthatonecanconceiveofanyspatio-temporalstructureandformalstructurewith-outanyassociatedphenomenalstructure;onecanknowaboutthefirstwithoutknowingaboutthesecond;andsoon.Sotheepistemicgapisaswideasever.Thebasicproblemwithanytype-Cmaterialiststrategyisthatepistemicimpli-cationfromAtoBrequiressomesortofconceptualhookbyvirtueofwhichtheconditiondescribedinAcansatisfytheconceptualrequirementsforthetruthofB.Whenaphysicalaccountimpliestruthsaboutlife,forexample,itdoessoinvirtueofimplyinginformationaboutthemacroscopicfunctioningofphysicalsystems,ofthesortrequiredforlife:here,broadlyfunctionalnotionsprovidetheconceptualhook.Butinthecaseofconsciousness,nosuchconceptualhookisavailable,giventhestructural-dynamiccharacterofphysicalconcepts,andthequitedifferentcharacteroftheconceptofconsciousness.Ultimately,itseemsthatanytype-Cstrategyisdoomedforfamiliarreasons.Onceweacceptthattheconceptofconsciousnessisnotitselfafunctionalcon-cept,andthatphysicaldescriptionsoftheworldarestructural-dynamicdescrip-tions,thereissimplynoconceptualroomforittobeimpliedbyaphysicaldescription.Sotheonlyroomleftistoholdthatconsciousnessisabroadlyfunctionalconceptafterall(acceptingtype-Amaterialism),toholdthatthereismoreinphysicsthanstructureanddynamics(acceptingtype-Ddualismortype-Fmonism),ortoholdthatthetruthofmaterialismdoesnotrequireanimplica-22tionfromphysicstoconsciousness(acceptingtype-Bmaterialism).Sointheend,thereisnoseparatespaceforthetype-Cmaterialist.122\nConsciousnessanditsPlaceinNature5.8InterludeArethereanyotheroptionsforthematerialist?Onefurtheroptionistorejectthedistinctionsonwhichthistaxonomyrests.Forexample,somephilosophers,espe-ciallyfollowersofQuine(1951),rejectanydistinctionbetweenconceptualtruthandempiricaltruth,orbetweentheaprioriandtheaposteriori,orbetweenthecontingentandthenecessary.OnewhoissufficientlyQuineanmightthereforerejectthedistinctionbetweentype-Aandtype-Bmaterialism,holdingthattalkofepistemicimplicationand/ormodalentailmentisungrounded,butthatmaterial-ismistruenevertheless.Wemightcallsuchaviewtype-Qmaterialism.Still,evenonthisview,similarissuesarise.SomeQuineansholdthatexplainingthefunc-tionsexplainseverything(Dennettmaybeanexample);ifso,alltheproblemsoftype-Amaterialismarise.Othersholdthatwecanpostulateidentitiesbetweenphysicalstatesandconsciousstatesinvirtueofthestrongisomorphicconnectionsbetweentheminnature(PaulChurchlandmaybeanexample);ifso,theprob-lemsoftype-Bmaterialismarise.Othersmayappealtonovelfuturesortsofexplanation;ifso,theproblemsoftype-Cmaterialismarise.SotheQuineanapproachcannotavoidtherelevantproblems.Leavingthissortofviewaside,itlooksliketheonlyremotelyviableoptionsforthematerialistaretype-Amaterialismandtype-Bmaterialism.Ithinkthatotherviewsareeitherultimatelyunstable,orcollapseintooneofthese(orthethree23remainingoptions).Itseemstomethatthecostsoftheseviews–denyingthemanifestexplananduminthefirstcase,andembracingprimitiveidentitiesorstrongnecessitiesinthesecondcase–suggestverystronglythattheyaretobeavoidedunlesstherearenoviablealternatives.Sotheresidualquestioniswhetherthereareviablealternatives.Ifconscious-nessisnotnecessitatedbyphysicaltruths,thenitmustinvolvesomethingontologicallynovelintheworld:touseKripke’smetaphor,afterfixingallthephysicaltruths,Godhadtodomoreworktofixallthetruthsaboutconscious-ness.Thatis,theremustbeontologicallyfundamentalfeaturesoftheworldoverandabovethefeaturescharacterizedbyphysicaltheory.Weareusedtotheideathatsomefeaturesoftheworldarefundamental:inphysics,featuressuchasspacetime,mass,andchargearetakenasfundamentalandnotfurtherexplained.Iftheargumentsagainstmaterialismarecorrect,thesefeaturesfromphysicsdonotexhaustthefundamentalfeaturesoftheworld:weneedtoexpandourcatalogoftheworld’sbasicfeatures.Therearetwopossibilitieshere.First,itcouldbethatconsciousnessisitselfafundamentalfeatureoftheworld,likespacetimeandmass.Inthiscase,wecansaythatphenomenalpropertiesarefundamental.Secondly,itcouldbethatcon-sciousnessisnotitselffundamental,butisnecessitatedbysomemoreprimitivefundamentalfeatureXthatisnotitselfnecessitatedbyphysics.Inthiscase,wemightcallXaprotophenomenalproperty,andwecansaythatprotophenomenalpropertiesarefundamental.Iwilltypicallyputthingsintermsofthefirstpossibility123\nDavidJ.Chalmersforeaseofdiscussion,butthediscussionthatfollowsappliesequallytothesecond.Eitherway,consciousnessinvolvessomethingnovelandfundamentalintheworld.Thequestionthenarises:howdothesenovelfundamentalpropertiesrelatetothealreadyacknowledgedfundamentalpropertiesoftheworld,namelythoseinvokedinmicrophysics?Ingeneral,wheretherearefundamentalproperties,therearefundamentallaws.Sowecanexpectthattherewillbesomesortoffundamentalprinciples–psychophysicallaws–connectingphysicalandphenom-enalproperties.Likethefundamentallawsofrelativityorquantummechanics,thesepsychophysicallawswillnotbededuciblefrommorebasicprinciples,butinsteadwillbetakenasprimitive.Butwhatisthecharacteroftheselaws?Animmediateworryisthatthemicrophysicalaspectsoftheworldareoftenheldtobecausallyclosed,inthateverymicrophysicalstatehasamicrophysicalsufficientcause.Howarefundamen-talphenomenalpropertiestobeintegratedwiththiscausallyclosednetwork?Thereseemtobethreemainoptionsforthenon-reductionisthere.First,onecoulddenythecausalclosureofthemicrophysical,holdingthattherearecausalgapsinmicrophysicaldynamicsthatarefilledbyacausalrolefordistinctphe-nomenalproperties:thisistype-Ddualism.Secondly,onecouldacceptthecausalclosureofthemicrophysicalandholdthatphenomenalpropertiesplaynocausalrolewithrespecttothephysicalnetwork:thisistype-Edualism.Thirdly,onecouldacceptthatthemicrophysicalnetworkiscausallyclosed,butholdthatphenomenalpropertiesareneverthelessintegratedwithitandplayacausalrole,byvirtueofconstitutingtheintrinsicnatureofthephysical:thisistype-Fmonism.Inwhatfollows,Iwilldiscusseachoftheseviews.Thediscussionisnecessarilyspeculativeincertainrespects,andIdonotclaimtoestablishthatanyoneoftheviewsistrueorcompletelyunproblematic.ButIdoaimtosuggestthatnoneofthemhasobviousfatalflaws,andthateachdeservesfurtherinvestigation.5.9Type-DDualismType-Ddualismholdsthatmicrophysicsisnotcausallyclosed,andthatphenom-24enalpropertiesplayacausalroleinaffectingthephysicalworld.Onthisview,usuallyknownasinteractionism,physicalstateswillcausephenomenalstates,andphenomenalstatescausephysicalstates.Thecorrespondingpsychophysicallawswillruninbothdirections.Onthisview,theevolutionofmicrophysicalstateswillnotbedeterminedbyphysicalprinciplesalone.Psychophysicalprinciplesspecifyingtheeffectofphenomenalstatesonphysicalstateswillalsoplayanirreduciblerole.ThemostfamiliarversionofthissortofviewisDescartes’ssubstancedualism(henceDforDescartes),onwhichthereareseparateinteractingmentalandphysicalsubstancesorentities.Butthissortofviewisalsocompatiblewitha124\nConsciousnessanditsPlaceinNaturepropertydualism,onwhichthereisjustonesortofsubstanceorentitywithbothphysicalandphenomenalfundamentalproperties,suchthatthephenomenalprop-ertiesplayanirreducibleroleinaffectingthephysicalproperties.Inparticular,theviewiscompatiblewithan“emergentist”viewsuchasBroad’s,onwhichphe-nomenalpropertiesareontologicallynovelpropertiesofphysicalsystems(notdeduciblefrommicrophysicalpropertiesalone),andhavenoveleffectsonmicrophysicalproperties(notdeduciblefrommicrophysicalprinciplesalone).Suchaviewwouldinvolvebasicprinciplesof“downward”causationofthementalonthemicrophysical(hencealsoDfordownwardcausation).Itissometimesobjectedthatdistinctphysicalandmentalstatescouldnotinteract,sincethereisnocausalnexusbetweenthem.ButonelessonfromHumeandfrommodernscienceisthatthesamegoesforanyfundamentalcausalinteractions,includingthosefoundinphysics.Newtoniansciencerevealsnocausalnexusbywhichgravitationworks,forexample;rather,therelevantlawsaresimplyfundamental.Thesamegoesforbasiclawsinotherphysicaltheories.Andthesame,presumably,appliestofundamentalpsychophysicallaws:thereisnoneedforacausalnexusdistinctfromthephysicalandmentalpropertiesthemselves.Byfarthemostinfluentialobjectiontointeractionismisthatitisincompatiblewithphysics.Itiswidelyheldthatsciencetellsusthatthemicrophysicalrealmiscausallyclosed,sothatthereisnoroomformentalstatestohaveanyeffects.Aninteractionistmightrespondinvariousways.Forexample,itcouldbesuggestedthatalthoughnoexperimentalstudieshaverevealedtheseeffects,nonehasruledthemout.Itmightfurtherbesuggestedthatphysicaltheoryallowsanynumberofbasicforces(fourasthingsstand,butthereisalwaysroomformore),andthatanextraforceassociatedwithamentalfieldwouldbeareasonableextensionofexistingphysicaltheory.Thesesuggestionswouldinvokesignificantrevisionstophysicaltheory,soarenottobemadelightly;butonecouldarguethatnothingrulesthemout.Byfarthestrongestresponsetothisobjection,however,istosuggestthatfarfromrulingoutinteractionism,contemporaryphysicsispositivelyencouragingtothepossibility.Onthestandardformulationofquantummechanics,thestateoftheworldisdescribedbyawavefunction,accordingtowhichphysicalentitiesareofteninasuperposedstate(e.g.,inasuperpositionoftwodifferentpositions),eventhoughsuperpositionsareneverdirectlyobserved.Onthestandarddynamics,thewavefunctioncanevolveintwoways:linearevolutionbytheSchrödingerequation(whichtendstoproducesuperposedstates),andnon-linearcollapsesfromsuperposedstatesintonon-superposedstates.Schrödingerevolutionisdeterministic,butcollapseisnon-deterministic.Schrödingerevolutioniscon-stantlyongoing,butonthestandardformulation,collapsesoccuronlyoccasion-ally,onmeasurement.Thecollapsedynamicsleavesadoorwideopenforaninteractionistinterpretation.Anyphysicalnon-determinismmightbeheldtoleaveroomfornon-physicaleffects,buttheprinciplesofcollapsedomuchmorethanthat.Collapseissupposedto125\nDavidJ.Chalmersoccuronmeasurement.Thereisnowidelyagreeddefinitionofwhatameasure-mentis,butthereisonesortofeventthateveryoneagreesisameasurement:observationbyaconsciousobserver.Further,itseemsthatnopurelyphysicalcriterionforameasurementcanwork,sincepurelyphysicalsystemsaregovernedbythelinearSchrödingerdynamics.Assuch,itisnaturaltosuggestthatameasurementispreciselyaconsciousobservation,andthatthisconsciousobser-vationcausesacollapse.Theclaimshouldnotbetoostrong:quantummechanicsdoesnotforcethisinterpretationofthesituationontous,andtherearealternativeinterpretationsofquantummechanicsonwhichtherearenocollapses,oronwhichmeasurement25hasnospecialroleincollapse.Nevertheless,quantummechanicsappearstobeperfectlycompatiblewithsuchaninterpretation.Infact,onemightarguethatifoneweretodesignelegantlawsofphysicsthatallowarolefortheconsciousmind,onecouldnotdomuchbetterthanthebipartitedynamicsofstandardquantummechanics:oneprinciplegoverningdeterministicevolutioninnormalcases,andoneprinciplegoverningnon-deterministicevolutioninspecialsitu-ationsthathaveaprimafacielinktothemental.Ofcoursesuchaninterpretationofquantummechanicsiscontroversial.Manyphysicistsrejectitpreciselybecauseitisdualistic,givingafundamentalroletoconsciousness.Thisrejectionisnotsurprising,butitcarriesnoforcewhenwehaveindependentreasontoholdthatconsciousnessmaybefundamental.Thereissomeironyinthefactthatphilosophersrejectinteractionismonlargelyphysical26grounds(itisincompatiblewithphysicaltheory),whilephysicistsrejectaninteractionistinterpretationofquantummechanicsonlargelyphilosophicalgrounds(itisdualistic).Takenconjointly,thesereasonscarrylittleforce,especiallyinlightoftheargumentsagainstmaterialismelsewhereinthischapter.27Thissortofinterpretationneedstobeformulatedindetailtobeassessed.Ithinkthemostpromisingversionofsuchaninterpretationallowsconsciousstatestobecorrelatedwiththetotalquantumstateofasystem,withtheextracon-straintthatconsciousstates(unlikephysicalstates)canneverbesuperposed.Inaconsciousphysicalsystemsuchasabrain,thephysicalandphenomenalstatesofthesystemwillbecorrelatedina(non-superposed)quantumstate.Uponobser-vationofasuperposedsystem,thenSchrödingerevolutionatthemomentofobservationwouldcausetheobservedsystemtobecomecorrelatedwiththebrain,yieldingaresultingsuperpositionofbrainstatesandso(bypsychophysicalcorrelation)asuperpositionofconsciousstates.Butsuchasuperpositioncannotoccur,sooneofthepotentialresultingconsciousstatesissomehowselected(presumablybyanon-deterministicdynamicprincipleatthephenomenallevel).Theresultisthat(bypsychophysicalcorrelation)adefinitebrainstateandadefinitestateoftheobservedobjectarealsoselected.Thesamemightapplytotheconnectionbetweenconsciousnessandnon-consciousprocessesinthebrain:whensuperposednon-consciousprocessesthreatentoaffectconsciousness,therewillbesomesortofselection.Inthisway,thereisacausalroleforconsciousnessinthephysicalworld.126\nConsciousnessanditsPlaceinNature(Interestingly,suchatheorymaybeempiricallytestable.Inquantummechanics,collapsetheoriesyieldpredictionsslightlydifferentfromno-collapsetheories,anddifferenthypothesesaboutthelocationofcollapseyieldpredictionsthatdifferfromeachother,althoughthedifferencesareextremelysubtleandarecurrentlyimpos-sibletomeasure.Iftherelevantexperimentscanonedaybeperformed,someoutcomeswouldgiveusstrongreasontoacceptacollapsetheory,andmightinturngiveusgroundstoacceptaroleforconsciousness.Asabonus,thiscouldevenyieldanempiricalcriterionforthepresenceofconsciousness.)Thereareanynumberoffurtherquestionsconcerningthepreciseformulationofsuchaview,itscompatibilitywithphysicaltheorymoregenerally(e.g.,relativityandquantumfieldtheory),anditsphilosophicaltenability(e.g.,doesthisviewyieldthesortofcausalrolethatweareinclinedtothinkconsciousnessmusthave).Butattheveryleast,itcannotbesaidthatphysicaltheoryimmediatelyrulesoutthepossibilityofaninteractionisttheory.Thosewhomakethisclaimoftenraisetheireyebrowswhenaspecifictheorysuchasquantummechanicsismentioned;butthisisquiteclearlyaninconsistentsetofattitudes.Ifphysicsissupposedtoruleoutinteractionism,thencarefulattentiontothedetailofphysicaltheoryisrequired.Allthissuggeststhatthereisatleastroomforaviableinteractionismtobeexplored,andthatthemostcommonobjectiontointeractionismhaslittleforce.Ofcourseitdoesnotentailthatinteractionismistrue.Thereismuchthatisattractiveabouttheviewofthephysicalworldascausallyclosed,andthereislittledirectevidencefromcognitivescienceofthehypothesisthatbehaviorcannotbewhollyexplainedintermsofphysicalcauses.Still,ifwehaveindependentreasontothinkthatconsciousnessisirreducible,andifwewishtoretaintheintuitiveviewthatconsciousnessplaysacausalrole,thenthisisaviewtobetakenveryseriously.5.10Type-EDualismType-Edualismholdsthatphenomenalpropertiesareontologicallydistinctfrom28physicalproperties,andthatthephenomenalhasnoeffectonthephysical.Thisistheviewusuallyknownasepiphenomenalism(hencetype-E):physicalstatescausephenomenalstates,butnotviceversa.Onthisview,psychophysicallawsruninonedirectiononly,fromphysicaltophenomenal.Theviewisnaturallycombinedwiththeviewthatthephysicalrealmiscausallyclosed:thisfurtherclaimisnotessentialtotype-Edualism,butitprovidesmuchofthemotivationfortheview.Aswithtype-Ddualism,type-Edualismiscompatiblewithasubstancedualismwithdistinctphysicalandmentalsubstancesorentities,andisalsocompatiblewithapropertydualismwithonesortofsubstanceorentityandtwosortsofproperty.Again,itiscompatiblewithanemergentismsuchasBroad’s,onwhichmentalpropertiesareontologicallynovelemergentpropertiesofanunderlyingentity,butinthiscasealthoughthereareemergentqualities,thereisnoemergentdownwardcausation.127\nDavidJ.ChalmersType-Edualismisusuallyputforwardasrespectingbothconsciousnessandscience:itsimultaneouslyaccommodatestheanti-materialistargumentsaboutconsciousnessandthecausalclosureofthephysical.Atthesametime,type-Edualismisfrequentlyrejectedasdeeplycounterintuitive.Iftype-Edualismiscorrect,thenphenomenalstateshavenoeffectonouractions,physicallycon-strued.Forexample,asensationofpainwillplaynocausalroleinmyhand’smovingawayfromaflame;myexperienceofdecisionwillplaynocausalroleinmymovingtoanewcountry;andasensationofredwillplaynocausalroleinmyproducingtheutterance“Iamexperiencingrednow.”Theseconsequencesareoftenheldtobeobviouslyfalse,oratleastunacceptable.Still,thetype-Edualistcanreplythatthereisnodirectevidencethatcontradictstheirview.Ourevidencerevealsonlyregularconnectionsbetweenphenomenalstatesandactions,sothatcertainsortsofexperiencearetypicallyfollowedbycertainsortsofaction.Beingexposedtothissortofconstantconjunctionproducesastrongbeliefinacausalconnection(asHumepointedoutinanothercontext);butitisneverthelesscompatiblewiththeabsenceofacausalconnection.Indeed,itseemsthatifepiphenomenalismweretrue,wewouldhaveexactlythesameevidence,andbeledtobelievethatconsciousnesshasacausalroleformuchthesamereasons.Soifepiphenomenalismisotherwisecoherentandacceptable,it29seemsthattheseconsiderationsdonotprovidestrongreasonstorejectit.Anotherobjectionholdsthatifconsciousnessisepiphenomenal,itcouldnothaveevolvedbynaturalselection.Thetype-Edualisthasastraightforwardreply,however.Onthetype-Eview,therearefundamentalpsychophysicallawsassoci-atingphysicalandphenomenalproperties.Ifevolutionselectsappropriatephys-icalproperties(perhapsinvolvingphysicalorinformationalconfigurationsinthebrain),thenthepsychophysicallawswillensurethatphenomenalpropertiesareinstantiated,too.Ifthelawshavetherightform,onecanevenexpectthat,asmorecomplexphysicalsystemsareselected,morecomplexstatesofconsciousnesswillevolve.Inthisway,physicalevolutionwillcarrytheevolutionofconscious-nessalongwithitasasortofby-product.Perhapsthemostinterestingobjectionstoepiphenomenalismfocusontherelationbetweenconsciousnessandrepresentationsofconsciousness.Itiscer-tainlyatleaststrangetosuggestthatconsciousnessplaysnocausalroleinmyutterancesof“Iamconscious.”Somehavesuggestedmorestronglythatthisrulesoutanyknowledgeofconsciousness.ItisoftenheldthatifabeliefaboutXistoqualifyasknowledge,thebeliefmustbecausedinsomefashionbyX.Butifconsciousnessdoesnotaffectphysicalstates,andifbeliefsarephysicallyconsti-tuted,thenconsciousnesscannotcausebeliefs.Andevenifbeliefsarenotphys-icallyconstituted,itisnotclearhowepiphenomenalismcanaccommodateacausalconnectionbetweenconsciousnessandbelief.Inresponse,anepiphenomenalistcandenythatknowledgealwaysrequiresacausalconnection.Onecanargueonindependentgroundsthatthereisastrongerconnectionbetweenconsciousnessandbeliefsaboutconsciousness:consciousnessplaysaroleinconstitutingphenomenalconceptsandphenomenalbeliefs.Ared128\nConsciousnessanditsPlaceinNatureexperienceplaysaroleinconstitutingabeliefthatoneishavingaredexperience,forexample.Ifso,thereisnocausaldistancebetweentheexperienceandthebelief.Andonecanarguethatthisimmediateconnectiontoexperienceandbeliefallowsforthebelieftobejustified.Ifthisisright,thenepiphenomenalismposesnoobstacletoknowledgeofconsciousness.Arelatedobjectionholdsthatmyzombietwinwouldproducethesamereports(e.g.,“Iamconscious”),causedbythesamemechanisms,andthathisreportsareunjustified;ifso,myownreportsareunjustified.Inresponse,onecanholdthatthetruebearersofjustificationarebeliefs,andthatmyzombietwinandIhavedifferentbeliefs,involvingdifferentconcepts,becauseoftherolethatconsciousnessplaysinconstitutingmyconceptsbutnotthezombie’s.Further,thefactthatweproduceisomorphicreportsimpliesthatathird-personobservermightnotbeanymorejustifiedinbelievingthatIamconsciousthanthatthezombieisconscious,butitdoesnotimplyadifferenceinfirst-personjustification.Thefirst-personjustificationformybeliefthatIamconsciousisnotgroundedinanywayinmyreportsbutratherinmyexperiencesthemselves,experiencesthatthezombielacks.Ithinkthatthereisnoknock-downobjectiontoepiphenomenalismhere.Still,itmustbeacknowledgedthatthesituationisatleastoddandcounterintuitive.Theoddnessofepiphenomenalismisexacerbatedbythefactthattherelationshipbetweenconsciousnessandreportsaboutconsciousnessseemstobesomethingofaluckycoincidence,ontheepiphenomenalistview.Afterall,ifpsychophysicallawsareindependentofphysicalevolution,thentherewillbepossibleworldswherephysicalevolutionisthesameasoursbutthepsychophysicallawsareverydifferent,sothatthereisaradicalmismatchbetweenreportsandexperiences.Itseemsluckythatweareinaworldwhosepsychophysicallawsmatchthemupsowell.Inresponse,anepiphenomenalistmighttrytomakethecasethattheselawsaresomehowthemost“natural”andaretobeexpected;butthereisatleastasignificantburdenofproofhere.Overall,Ithinkthatepiphenomenalismisacoherentviewwithoutfatalprob-lems.Atthesametime,itisaninelegantview,producingafragmentedpictureofnature,onwhichphysicalandphenomenalpropertiesareonlyveryweaklyinte-gratedinthenaturalworld.Andofcourseitisacounterintuitiveviewthatmanypeoplefinddifficulttoaccept.Ineleganceandcounterintuitivenessarebetterthanincoherence;soifgoodargumentsforceustoepiphenomenalismasthemostcoherentview,thenweshouldtakeitseriously.Butatthesametime,wehavegoodreasontoexamineotherviewsverycarefully.5.11Type-FMonismType-Fmonismistheviewthatconsciousnessisconstitutedbytheintrinsicpropertiesoffundamentalphysicalentities:thatis,bythecategoricalbasesof30fundamentalphysicaldispositions.Onthisview,phenomenalorprotophenomenal129\nDavidJ.Chalmerspropertiesarelocatedatthefundamentallevelofphysicalreality,and,inacertainsense,underliephysicalrealityitself.ThisviewtakesitscuefromBertrandRussell’sdiscussionofphysicsinTheAnalysisofMatter(1927).Russellpointedoutthatphysicscharacterizesphysicalentitiesandpropertiesbytheirrelationstooneanotherandtous.Forexample,aquarkischaracterizedbyitsrelationstootherphysicalentities,andapropertysuchasmassischaracterizedbyanassociateddispositionalrole,suchasthetendencytoresistacceleration.Atthesametime,physicssaysnothingabouttheintrinsicnatureoftheseentitiesandproperties.Wherewehaverelationsanddispositions,weexpectsomeunderlyingintrinsicpropertiesthatgroundthe31dispositions,characterizingtheentitiesthatstandintheserelations.Butphysicsissilentabouttheintrinsicnatureofaquark,orabouttheintrinsicpropertiesthatplaytheroleassociatedwithmass.Sothisisonemetaphysicalproblem:whataretheintrinsicpropertiesoffundamentalphysicalsystems?Atthesametime,thereisanothermetaphysicalproblem:howcanphenomenalpropertiesbeintegratedwiththephysicalworld?Phenomenalpropertiesseemtobeintrinsicpropertiesthatarehardtofitinwiththestructural/dynamiccharacterofphysicaltheory;andarguably,theyaretheonlyintrinsicpropertiesofwhichwehavedirectknowledge.Russell’sinsightwasthatwemightsolveboththeseproblemsatonce.Perhapstheintrinsicpropertiesofthephysicalworldarethem-selvesphenomenalproperties.Orperhapstheintrinsicpropertiesofthephysicalworldarenotphenomenalproperties,butneverthelessconstitutephenomenalproperties:thatis,perhapstheyareprotophenomenalproperties.Ifso,thenconsciousnessandphysicalrealityaredeeplyintertwined.Thisviewholdsthepromiseofintegratingphenomenalandphysicalpropertiesverytightlyinthenaturalworld.Here,natureconsistsofentitieswithintrinsic(proto)phenomenalqualitiesstandingincausalrelationswithinaspacetimemani-fold.Physicsasweknowitemergesfromtherelationsbetweentheseentities,whereasconsciousnessasweknowitemergesfromtheirintrinsicnature.Asabonus,thisviewisperfectlycompatiblewiththecausalclosureofthemicrophysical,andindeedwithexistingphysicallaws.Theviewcanretainthestructureofphysicaltheoryasitalreadyexists;itsimplysupplementsthisstructurewithanintrinsicnature.Andtheviewacknowledgesaclearcausalroleforconsciousnessinthephysicalworld:(proto)phenomenalpropertiesserveastheultimatecat-egoricalbasisofallphysicalcausation.Thisviewhaselementsincommonwithbothmaterialismanddualism.Fromoneperspective,itcanbeseenasasortofmaterialism.Ifoneholdsthatphysicaltermsrefernottodispositionalpropertiesbuttheunderlyingintrinsicproperties,thentheprotophenomenalpropertiescanbeseenasphysicalproperties,thuspreservingasortofmaterialism.Fromanotherperspective,itcanbeseenasasortofdualism.Theviewacknowledgesphenomenalorprotophenomenalpropert-iesasontologicallyfundamental,anditretainsanunderlyingdualitybetweenstructural-dispositionalproperties(thosedirectlycharacterizedinphysicaltheory)andintrinsicprotophenomenalproperties(thoseresponsibleforconsciousness).130\nConsciousnessanditsPlaceinNatureOnemightsuggestthatwhiletheviewarguablyfitstheletterofmaterialism,itsharesthespiritofanti-materialism.Initsprotophenomenalform,theviewcanbeseenasasortofneutralmonism:thereareunderlyingneutralpropertiesX(theprotophenomenalproperties),suchthattheXpropertiesaresimultaneouslyresponsibleforconstitutingthephysicaldomain(bytheirrelations)andthephenomenaldomain(bytheircollectiveintrinsicnature).Initsphenomenalform,itcanbeseenasasortofidealism,suchthatmentalpropertiesconstitutephysicalproperties,althoughtheseneednotbementalpropertiesinthemindofanobserver,andtheymayneedtobesupple-mentedbycausalandspatio-temporalpropertiesinaddition.Onecouldalsocharacterizethisformoftheviewasasortofpanpsychism,withphenomenalpropertiesubiquitousatthefundamentallevel.Onecouldgivetheviewinitsmostgeneralformthenamepanprotopsychism,witheitherprotophenomenalorphenomenalpropertiesunderlyingallofphysicalreality.Atype-Fmonistmayhaveoneofanumberofattitudestothezombieargu-mentagainstmaterialism.Sometype-Fmonistsmayholdthatacompletephysicaldescriptionmustbeexpandedtoincludeanintrinsicdescription,andmaycon-sequentlydenythatzombiesareconceivable.(Weonlythinkweareconceivingofaphysicallyidenticalsystembecauseweoverlookintrinsicproperties.)Otherscouldmaintainthatexistingphysicalconceptsreferviadispositionstothoseintrinsicpropertiesthatgroundthedispositions.Ifso,theseconceptshavediffer-entprimaryandsecondaryintensions,andatype-Fmonistcouldcorrespondinglyacceptconceivabilitybutdenypossibility:wemisdescribetheconceivedworldas32physicallyidenticaltoours,wheninfactitisjuststructurallyidentical.Finally,atype-Fmonistmightholdthatphysicalconceptsrefertodispositionalproperties,sothatzombiesarebothconceivableandpossible,andtheintrinsicpropertiesarenotphysicalproperties.Thedifferencesbetweenthesethreeattitudesseemtobeultimatelyterminologicalratherthansubstantive.Asfortheknowledgeargument,atype-FmonistmightinsistthatforMarytohavecompletephysicalknowledge,shewouldhavetohaveadescriptionoftheworldinvolvingconceptsthatdirectlycharacterizetheintrinsicproperties;ifshehadthis(asopposedtoherimpoverisheddescriptioninvolvingdispositionalcon-cepts),shemighttherebybeinapositiontoknowwhatitisliketoseered.Regardingtheexplanatoryargument,atype-Fmonistmightholdthatphysicalaccountsinvolvingintrinsicpropertiescanexplainmorethanstructureandfunc-tion.Alternatively,atype-Fmonistwhostickstodispositionalphysicalconceptswillmakeresponsesanalogoustooneoftheothertworesponsesabove.Thetype-Fviewisadmittedlyspeculative,anditcansoundstrangeatfirsthearing.Manyfinditextremelycounterintuitivetosupposethatfundamentalphysicalsystemshavephenomenalproperties:e.g.,thatthereissomethingitisliketobeanelectron.Theprotophenomenalversionoftheviewrejectsthisclaim,butretainssomethingofitsstrangeness:itseemsthatanypropertiesre-sponsibleforconstitutingconsciousnessmustbestrangeandunusualproperties,ofasortthatwemightnotexpecttofindinmicrophysicalreality.Still,itisnot131\nDavidJ.Chalmersclearthatthisstrangenessyieldsanystrongobjections.Likeepiphenomenalism,theviewappearstobecompatiblewithallourevidence,andthereisnodirectevidenceagainstit.Onecanarguethatiftheviewweretrue,thingswouldappeartousjustastheyinfactappear.Andwehavelearnedfrommodernphysicsthattheworldisastrangeplace:wecannotexpectittoobeyallthedictatesofcommonsense.Onemightalsoobjectthatwedonothaveanyconceptionofwhatproto-phenomenalpropertiesmightbelike,orofhowtheycouldconstitutephenomenalproperties.Thisistrue,butonecouldsuggestthatthisismerelyaproductofourignorance.Inthecaseoffamiliarphysicalproperties,therewereprincipledreasons(basedonthecharacterofphysicalconcepts)fordenyingaconstitutiveconnec-tiontophenomenalproperties.Here,therearenosuchprincipledreasons.Atmost,thereisignoranceandabsenceofaconnection.Ofcourseitwouldbeverydesirabletoformapositiveconceptionofprotophenomenalproperties.Perhapswecandothisindirectly,bysomesortoftheoreticalinferencefromthecharacterofphenomenalpropertiestotheirunderlyingconstituents;orperhapsknowledgeofthenatureofprotophenomenalpropertieswillremainbeyondus.Eitherway,33thisisnoreasontorejectthetruthoftheview.Thereisonesortofprincipledprobleminthevicinity,pointedoutbyWilliamJames(1890:ch.6).Ourphenomenologyhasarichandspecificstructure:itisunified,bounded,differentiatedintomanydifferentaspects,butwithanunder-lyinghomogeneitytomanyoftheaspects,andappearstohaveasinglesubjectofexperience.Itisnoteasytoseehowadistributionofalargenumberofindividualmicrophysicalsystems,eachwiththeirownprotophenomenalpro-perties,couldsomehowadduptothisrichandspecificstructure.Shouldonenotexpectsomethingmorelikeadisunified,jaggedcollectionofphenomenalspikes?Thisisaversionofthecombinationproblemforpanpsychism(Seagar1995),orwhatStoljar(2001)callsthestructuralmismatchproblemfortheRussellianview(seealsoFoster1991:119–30).Toanswerit,itseemsthatweneedamuchbetterunderstandingofthecompositionalprinciplesofphenomenology:thatis,theprinciplesbywhichphenomenalpropertiescanbecomposedorconstitutedfromunderlyingphenomenalproperties,orprotophenomenalproperties.Wehaveagoodunderstandingoftheprinciplesofphysicalcomposition,butnorealunderstandingoftheprinciplesofphenomenalcomposition.Thisisanareathatdeservesmuchcloseattention:Ithinkitiseasilythemostseriousproblemforthetype-Fmonistview.Atthispoint,itisanopenquestionwhetherornottheproblemcanbesolved.Sometype-Fmonistsappeartoholdthattheycanavoidthecombinationproblembyholdingthatphenomenalpropertiesaretheintrinsicpropertiesofhigh-levelphysicaldispositions(e.g.,thoseinvolvedinneuralstates),andneednotbeconstitutedbytheintrinsicpropertiesofmicrophysicalstates(hencetheymayalsodenypanprotopsychism).Butthisseemstobeuntenable:ifthelow-levelnetworkiscausallyclosedandthehigh-levelintrinsicpropertiesarenot132\nConsciousnessanditsPlaceinNatureconstitutedbylow-levelintrinsicproperties,thehigh-levelintrinsicpropertieswillbeepiphenomenalalloveragain,forfamiliarreasons.Theonlywaytoembracethispositionwouldseemtobeincombinationwithadenialofmicrophysicalcausalclosure,holdingthattherearefundamentaldispositionsabovethemicro-physicallevel,whichhavephenomenalpropertiesastheirgrounds.Butsucha34viewwouldbeindistinguishablefromtype-Ddualism.Soadistinctivetype-Fmonismwillhavetofacethecombinationproblemdirectly.Overall,type-Fmonismpromisesadeeplyintegratedandelegantviewofna-ture.Noonehasyetdevelopedanysortofdetailedtheoryinthisclass,anditisnotyetclearwhethersuchatheorycanbedeveloped.Butatthesametime,thereappeartobenostrongreasonstorejecttheview.Assuch,type-Fmonismislikelytoprovidefertilegroundsforfurtherinvestigation,anditmayultimatelyprovidethebestintegrationofthephysicalandthephenomenalwithinthenaturalworld.5.12ConclusionsArethereanyotheroptionsforthenon-reductionist?Therearetwoviewsthatmaynotfitstraightforwardlyintothecategoriesabove.First,somenon-materialistsholdthatphenomenalpropertiesareontologicallywhollydistinctfromphysicalproperties,thatmicrophysicsiscausallyclosed,butthatphenomenalpropertiesplayacausalrolewithrespecttothephysicalnever-theless.Onewaythismighthappenisbyasortofcausaloverdetermination:physicalstatescausallydeterminebehavior,butphenomenalstatescausebehavioratthesametime.Anotherisbycausalmediation:itmightbethatinatleastsomeinstancesofmicrophysicalcausationfromAtoB,thereisactuallyacausalcon-nectionfromAtothemindtoB,sothatthemindentersthecausalnexuswithoutalteringthestructureofthenetwork.Andtheremaybefurtherstrategieshere.Wemightcallthisclasstype-Odualism(takingoverdeterminationasaparadigmcase).Theseviewssharemuchofthestructureofthetype-Eview(causallyclosedphysicalworld,distinctphenomenalproperties),butescapesthechargeofepiphenomenalism.Thespecialcausalsetupsoftheseviewsmaybehardtoswallow,andtheysharesomeofthesameproblemsasthetype-Eview(e.g.,thefragmentedviewofnature,andthe“lucky”psychophysicallaws),but35thisclassshouldneverthelessbeputonthetableasanoption.Second,somenon-materialistsareidealists(inaBerkeleyansense),holdingthatthephysicalworldisitselfconstitutedbytheconsciousstatesofanobservingagent.Wemightcallthisviewtype-Imonism.Itshareswithtype-Fmonismthepropertythatphenomenalstatesplayaroleinconstitutingphysicalreality,butonthetype-Iviewthishappensinaverydifferentway:notbyhavingseparate“microscopic”phenomenalstatesunderlyingeachphysicalstate,butratherbyhavingphysicalstatesconstitutedholisticallybya“macroscopic”phenomenal133\nDavidJ.Chalmersmind.Thisviewseemstobenon-naturalisticinamuchdeepersensethananyoftheviewsabove,andinparticularseemstosufferfromanabsenceofcausalorexplanatoryclosureinnature:oncethenaturalexplanationintermsoftheexter-nalworldisremoved,highlycomplexregularitiesamongphenomenalstateshavetobetakenasunexplainedintermsofsimplerprinciples.Butagain,thissortofviewshouldatleastbeacknowledged.AsIseethings,thebestoptionsforanon-reductionistaretype-Ddualism,type-Edualism,ortype-Fmonism:thatis,interactionism,epiphenomenalism,orpanprotopsychism.Ifweacknowledgetheepistemicgapbetweenthephysicalandthephenomenal,andweruleoutprimitiveidentitiesandstrongnecessities,thenweareledtoadisjunctionofthesethreeviews.Eachoftheviewshasatleastsomepromise,andnonehasclearfatalflaws.Formypart,Igivesomecredencetoeachofthem.Ithinkthatinsomewaysthetype-Fviewisthemostappealing,butthissenseislargelygroundedinaestheticconsiderationswhoseforceisunclear.Thechoicebetweenthesethreeviewsmaydependinlargepartonthedevel-opmentofspecifictheorieswithintheseframeworks.Especiallyforthetype-Dviewandtype-Fview,furthertheoreticalworkiscrucialinassessingthetheories(e.g.,inexplicatingquantuminteractionism,orinunderstandingphenomenalcomposition).Itmayalsobethattheempiricalscienceofconsciousnesswillgivesomeguidance.Asthescienceprogresses,wewillbeledtoinfersimpleprinciplesthatunderliecorrelationsbetweenphysicalandphenomenalstates.Itmaybethattheseprinciplesturnouttopointstronglytowardoneortheotheroftheseviews:e.g.,ifsimpleprinciplesconnectingmicrophysicalstatestophenomenalorprotophenomenalstatescandotheexplanatorywork,thenwemayhavereasontofavoratype-Fview,whileiftheprincipleslatchontothephysicalworldatahigherlevel,thenwemayhavereasontofavoratype-Dortype-Eview.Andifconsciousnesshasaspecificpatternofeffectsonthephysicalworld,asthetype-Dviewsuggests,thenempiricalstudiesoughtinprincipletobeabletofindtheseeffects,althoughperhapsonlywithgreatdifficulty.Noteveryonewillagreethateachoftheseviewsisviable.Itmaybethatfurtherexaminationwillrevealdeepproblemswithsomeoftheseviews.Butthisfurtherexaminationneedstobeperformed.Therehasbeenlittlecriticalexaminationoftype-Fviewstodate,forexample;wehaveseenthatthestandardargumentsagainsttype-Dviewscarryverylittleweight;andwhileargumentsagainsttype-Eviewscarrysomeintuitiveforce,theyarefarfrommakingaknock-downcaseagainsttheviews.Isuspectthateveniffurtherexaminationrevealsdeepproblemsforsomeviewsinthisvicinity,itisveryunlikelythatallsuchviewswillbeeliminated.Inanycase,thisgivesussomeperspectiveonthemind–bodyproblem.Itisoftenheldthateventhoughitishardtoseehowmaterialismcouldbetrue,materialismmustbetrue,sincethealternativesareunacceptable.AsIseeit,thereareatleastthreeprimafacieacceptablealternativestomaterialismonthetable,eachofwhichiscompatiblewithabroadlynaturalistic(evenifnotmaterialistic)134\nConsciousnessanditsPlaceinNatureworldview,andnoneofwhichhasfatalproblems.Sogiventheclearargumentsagainstmaterialism,itseemstomethatweshouldatleasttentativelyembracetheconclusionthatoneoftheseviewsiscorrect.Ofcoursealloftheviewsdiscussedinthischapterneedtobedevelopedinmuchmoredetail,andexaminedinlightofallrelevantscientificandphilosophicaldevelopments,inordertobecompre-hensivelyassessed.Butasthingsstand,Ithinkthatwehavegoodreasontosupposethatconsciousnesshasafundamentalplaceinnature.Notes1Thischapterisanoverviewofissuesconcerningthemetaphysicsofconsciousness.Muchofthediscussioninthischapter(especiallythefirstpart)recapitulatesdiscus-sioninChalmers(1995;1996;1997),althoughitoftentakesadifferentform,andsometimesgoesbeyondthediscussionthere.Igiveamoredetailedtreatmentofmanyoftheissuesdiscussedhereintheworkscitedinthebibliography.2Thetaxonomyisinthefinalchapter,chapter14,ofBroad’sbook(setoutonpp.607–11,anddiscusseduntilp.650).ThedramatizationofBroad’staxonomyasa4×4matrixisillustratedonAndrewChrucky’swebsitedevotedtoBroad,athttp://www.ditext.com/broad/mpn14.html#t.3Onmyusage,qualiaaresimplythosepropertiesthatcharacterizeconsciousstatesaccordingtowhatitisliketohavethem.Thedefinitiondoesnotbuildinanyfurthersubstantiverequirements,suchastherequirementthatqualiaareintrinsicornon-intentional.Ifqualiaareintrinsicornon-intentional,thiswillbeasubstantiveratherthanadefinitionalpoint(sotheclaimthatthepropertiesofconsciousnessarenon-intrinsicorthattheyarewhollyintentionalshouldnotbetakentoentailthattherearenoqualia).Phenomenalpropertiescanalsobetakentobepropertiesofindividuals(e.g.,people)ratherthanofmentalstates,characterizingaspectsofwhatitisliketobethematagiventime;thedifferencewillnotmattermuchforpresentpurposes.4NotethatIuse“reductive”inabroadersensethanitissometimesused.Reductiveexplanationrequiresonlythathigh-levelphenomenacanbeexplainedwhollyintermsoflow-levelphenomena.Thisiscompatiblewiththe“multiplerealizability”ofhigh-levelphenomenainlow-levelphenomena.Forexample,theremaybemanydifferentwaysinwhichdigestioncouldberealizedinaphysiologicalsystem,butonecannever-thelessreductivelyexplainasystem’sdigestionintermsofunderlyingphysiology.Anothersubtletyconcernsthepossibilityofaviewonwhichconsciousnesscanbeexplainedintermsofprincipleswhichdonotmakeappealtoconsciousnessbutcannotthemselvesbephysicallyexplained.Thedefinitionsabovecountsuchaviewasneitherreductivenornon-reductive.Itcouldreasonablybeclassifiedeitherway,butIwillgenerallyassimilateitwiththenon-reductiveclass.5AversionoftheexplanatoryargumentasformulatedhereisgiveninChalmers(1995).Forrelatedconsiderationsaboutexplanation,seeLevine(1983)onthe“explanatorygap”andNagel(1974).SeealsothepapersinShear(1997).6VersionsoftheconceivabilityargumentareputforwardbyCampbell(1970),Kirk(1974),Kripke(1980),Bealer(1994),andChalmers(1996),amongothers.Import-antpredecessorsincludeDescartes’sconceivabilityargumentaboutdisembodiment,andLeibniz’s“mill”argument.135\nDavidJ.Chalmers7SourcesfortheknowledgeargumentincludeNagel(1974),Maxwell(1968),Jackson(1982),andothers.PredecessorsoftheargumentarepresentinBroad’sdiscussionofa“mathematicalarchangel”whocannotdeducethesmellofammoniafromphysicalfacts(1925:70–1),andFeigl’sdiscussionofa“Martiansuperscientist”whocannotknowwhatcolorslooklikeandwhatmusicaltonessoundlike(1967[1958]:64,68,140).8ThisversionofthethoughtexperimenthasareallifeexemplarinKnutNordby,aNorwegiansensorybiologistwhoisarodmonochromat(lackingconesinhisretinaforcolorvision),andwhoworksonthephysiologyofcolorvision.SeeNordby(1990).9Forlimitedversionsoftheconceivabilityargumentandtheexplanatoryargument,seeBroad(1925:614–15).Fortheknowledgeargument,seepp.70–2,whereBroadarguesthatevena“mathematicalarchangel”couldnotdeducethesmellofammoniafrommicroscopicknowledgeofatoms.Broadisarguingagainst“mechanism,”whichisroughlyequivalenttocontemporarymaterialism.PerhapsthebiggestlacunainBroad’sargument,tocontemporaryeyes,isanyconsiderationofthepossibilitythatthereisanepistemicbutnotanontologicalgap.10Foradiscussionoftherelationshipbetweentheconceivabilityargumentandtheknowledgeargument,seeChalmers(1996and2002b).11Type-AmaterialistsincludeRyle(1949),Lewis(1988),Dennett(1991),Dretske(1995),Rey(1995),andHarman(1990).12Twospecificviewsmaybeworthmentioning:(1)Someviews(e.g.,Dretske1995)denyanepistemicgapwhileatthesametimedenyingfunctionalism,byholdingthatconsciousnessinvolvesnotjustfunctionalrolebutalsocausalandhistoricalrelationstoobjectsintheenvironment.Icounttheseastype-Aviews:wecanviewtherelevantrelationsaspartoffunctionalrole,broadlyconstrued,andexactlythesameconsidera-tionsarise.(2)Someviews(e.g.,Strawson2000andStoljar2001)denyanepistemicgapnotbyfunctionallyanalyzingconsciousnessbutbyexpandingourviewofthephysicalbasetoincludeunderlyingintrinsicproperties.Theseviewsarediscussedundertype-F(sectn5.11).13Inanotheranalogy,Churchland(1996)suggeststhatsomeoneinGoethe’stimemighthavemountedanalogousepistemicargumentsagainstthereductiveexplana-tionof“luminescence.”Butonacloselook,itisnothardtoseethattheonlyfurtherexplanandumthatcouldhavecauseddoubtshereistheexperienceofseeinglight(seeChalmers1997).Thispointisnohelptothetype-Amaterialist,sincethisexplanandumremainsunexplained.14Foranargumentfromunsavorymetaphysicalconsequences,seeWhite(1986).Foranargumentfromunsavoryepistemologicalconsequences,seeShoemaker(1975).Themetaphysicalconsequencesareaddressedinthesecondhalfofthischapter.TheepistemologicalconsequencesareaddressedinChalmers2002a.15Type-BmaterialistsincludeLevine(1983),Loar(1990/1997),Papineau(1993),Tye(1995),Lycan(1996),Hill(1997),BlockandStalnaker(1999),andPerry(2001).16Incertainrespects,wheretype-AmaterialismcanbeseenasderivingfromthelogicalbehaviorismofRyleandCarnap,type-BmaterialismcanbeseenasderivingfromtheidentitytheoryofPlaceandSmart.Thematteriscomplicated,however,bythefactthattheearlyidentitytheoristsadvocated“topic-neutral”(functional)analysesofphenomenalproperties,suggestinganunderlyingtype-Amaterialism.136\nConsciousnessanditsPlaceinNature17BlockandStalnaker(1999)argueagainstdeducibilityinpartbyarguingthatthereisusuallynoexplicitconceptualanalysisofhigh-leveltermssuchas“water”inmicrophysicalterms,orinanyothertermsthatcouldgroundanapriorientailmentfrommicrophysicaltruthstotruthsaboutwater.Inresponse,ChalmersandJackson(2001)arguethatexplicitconceptualanalysesarenotrequiredforapriorientailments,andthatthereisgoodreasontobelievethatsuchentailmentsexistinthesecases.18Two-dimensionalsemanticframeworksoriginateintheworkofStalnaker(1978),Evans(1979),andKaplan(1989).Theversionusedintheseargumentsissomewhatdifferent:fordiscussionofthedifferences,seeChalmers(forthcoming).19ThisisaslightlymoreformalversionofanargumentinChalmers(1996:131–6).ItisquitecloselyrelatedtoKripke’smodalargumentagainsttheidentitytheory,thoughdifferentinsomeimportantrespects.Thecentralpremise2canbeseenasawayofformalizingKripke’sclaimthatwherethereis“apparentcontingency,”thereissomemisdescribedpossibilityinthebackground.Theargumentcanalsobeseenasawayofformalizingaversionofthe“dualproperty”objectionattributedtoMaxBlackbySmart(1959),anddevelopedbyJackson(1979)andWhite(1986).Relatedapplica-tionsofthetwo-dimensionalframeworktoquestionsaboutmaterialismaregivenbyJackson(1994)andLewis(1994).20Ihavepassedoverafewsubtletieshere.Oneconcernstheroleofindexicals:tohandleclaimssuchas“Iamhere,”primaryintensionsaredefinedovercenteredworlds:worldswithamarkedindividualandtime,correspondingtoindexical“locatinginfor-mation”aboutone’spositionintheworld.Thischangedoesnothelpthetype-Bmaterialist,however.EvenifwesupplementPwithindexicallocatinginformationI(e.g.,tellingMaryaboutherlocationintheworld),thereisasmuchofanepistemicgapwithQasever;soP∧I∧¬Qisconceivable.AndgiventhatthereisacenteredworldthatverifiesP∧I∧¬Q,onecanseeasabovethateitherthereisaworldsatisfyingP∧¬Q,ortype-Fmonismistrue.21Hill(1997)triestoexplainawayourmodalintuitionsaboutconsciousnessincogn-itiveterms.Chalmers(1999)respondsthatanymodalintuitionmightbeexplainedincognitiveterms(asimilarargumentcould“explainaway”ourintuitionthattheremightberedsquares),butthatthishasnotendencytosuggestthattheintuitionisincorrect.Ifsuchanaccounttellsusthatmodalintuitionsaboutconsciousnessareunreliable,thesamegoesforallmodalintuitions.Whatisreallyneededisnotanexplanationofourmodalintuitionsaboutconsciousness,butanexplanationofwhytheseintuitionsinparticularshouldbeunreliable.Loar(1990/1997)attemptstoprovidesuchanexplanationintermsoftheuniquefeaturesofphenomenalconcepts.Hesuggeststhat(1)phenomenalconceptsarerecognitionalconcepts(“thatsortofthing”);that(2)likeotherrecognitionalcon-cepts,theycanco-referwithphysicalconceptsthatarecognitivelydistinct;andthat(3)unlikeotherrecognitionalconcepts,theylackcontingentmodesofpresentation(i.e.,theirprimaryandsecondaryintensionscoincide).If(2)and(3)bothhold(andifweassumethatphysicalconceptsalsolackcontingentmodesofpresentation),thenaphenomenal-physicalidentitywillbeastrongnecessityinthesenseabove.Inresponse,Chalmers(1999)arguesthat(2)and(3)cannotbothhold.Theco-referenceofotherrecognitionalconceptswiththeoreticalconceptsisgroundedintheircontingentmodesofpresentation;intheabsenceofsuchmodesofpresentation,thereisnoreasontothinkthattheseconceptscanco-refer.Soaccepting(3)undercutsanysupportfor(2).137\nDavidJ.ChalmersChalmers(1999)alsoarguesthatbyassumingthatphysicalpropertiescanhavephenomenalmodesofpresentationnon-contingently,Loar’saccountisineffectpre-supposingratherthanexplainingtherelevantstrongnecessities.22Ofthosementionedaboveasapparentlysympatheticwithtype-Cmaterialism,IthinkMcGinnisultimatelyatype-Fmonist,Nageliseitheratype-Bmaterialistoratype-Fmonist,andChurchlandiseitheratype-Bmaterialistoratype-Qmaterialist(below).23Onemightaskaboutspecificreductiveviews,suchasrepresentationalism(whichidentifiesconsciousnesswithcertainrepresentationalstates),andhigher-orderthoughttheory(whichidentifiesconsciousnesswiththeobjectsofhigher-orderthoughts).Howtheseviewsareclassifieddependsonhowagiventheoristregardstherepresen-tationalorhigher-orderstates(e.g.,functionallydefinableornot)andtheirconnec-tiontoconsciousness(e.g.,conceptualorempirical).Amongrepresentationalists,IthinkthatHarman(1990)andDretske(1995)aretype-Amaterialists,whileTye(1995)andLycan(1996)aretype-Bmaterialists.Amonghigher-orderthoughtthe-orists,Carruthers(2000)isclearlyatype-Bmaterialist,whileRosenthal(1997)iseithertype-Aortype-B.Onecouldalsoinprincipleholdnon-materialistversionsofeachoftheseviews.24Type-DdualistsincludePopperandEccles(1977),Sellars(1981),Swinburne(1986),Foster(1991),Hodgson(1991),andStapp(1993).25No-collapseinterpretationsincludeBohm’s“hidden-variable”interpretations,andEverett’s“many-worlds”(or“many-minds”)interpretation.AcollapseinterpretationthatdoesnotinvokemeasurementistheGhirardi-Rimini-Weberinterpretation(withrandomoccasionalcollapses).Eachoftheseinterpretationsrequiresasignificantrevi-siontothestandarddynamicsofquantummechanics,andeachiscontroversial,althougheachhasitsbenefits(seeAlbert1993fordiscussionoftheseandotherinterpretations).Itisnotablethatthereseemstobenoremotelytenableinterpreta-tionthatpreservesthestandardclaimthatcollapsesoccuruponmeasurement,exceptfortheinterpretationinvolvingconsciousness.26Ihavebeenasguiltyofthisasanyone,settingasideinteractionisminChalmers(1996)partlyforreasonsofcompatibilitywithphysics.Iamstillnotespeciallyin-clinedtoendorseinteractionism,butInowthinkthattheargumentfromphysicsismuchtooglib.ThreefurtherreasonsforrejectingtheviewarementionedinChalmers(1996).First,ifconsciousnessistomakeaninterestingqualitativedifferencetobehavior,thisrequiresthatitactnon-randomly,inviolationoftheprobabilisticrequirementsofquantummechanics.Ithinkthereissomethingtothis,butonecouldbitethebulletonnon-randomnessinresponse,oronecouldholdthatevenarandomcausalroleforconsciousnessisgoodenough.Secondly,Iarguedthatdeny-ingcausalclosureyieldsnospecialadvantage,asaviewwithcausalclosurecanachievemuchthesameeffectviatype-Fmonism.Againthereissomethingtothis,butthetype-Dviewdoeshavethesignificantadvantageofavoidingthetype-Fview’s“combinationproblem.”Thirdly,itisnotclearthatthecollapseinterpretationyieldsthesortofcausalroleforconsciousnessthatweexpectittohave.Ithinkthatthisisanimportantopenquestionthatrequiresdetailedinvestigation.27Consciousness-collapseinterpretationsofquantummechanicshavebeenputforwardbyWigner(1961),Hodgson(1991),andStapp(1993).OnlyStappgoesintomuchdetail,withaninterestingbutsomewhatidiosyncraticaccountthatgoesinadirectiondifferentfromthatsuggestedabove.138\nConsciousnessanditsPlaceinNature28Type-EdualistsincludeHuxley(1874),Campbell(1970),Jackson(1982),andRobinson(1988).29Someaccusetheepiphenomenalistofadoublestandard:relyingonintuitioninmakingthecaseagainstmaterialism,butgoingcountertointuitionindenyingacausalroleforconsciousness.Butintuitionsmustbeassessedagainstthebackgroundofreasonsandevidence.Todenytherelevantintuitionsintheanti-materialistargu-ment(inparticular,theintuitionofafurtherexplanandum)appearstocontradicttheavailablefirst-personevidence;butdenyingacausalroleforconsciousnessappearstobecompatibleonreflectionwithallourevidence,includingfirst-personevidence.30Versionsoftype-FmonismhavebeenputforwardbyRussell(1927),Feigl(1967[1958]),Maxwell(1979),Lockwood(1989),Chalmers(1996),Griffin(1998),Strawson(2000),andStoljar(2001).31Thereisphilosophicaldebateoverthethesisthatalldispositionshaveacategoricalbasis.Ifthethesisisaccepted,thecasefortype-Fmonismisparticularlystrong,sincemicrophysicaldispositionalmusthaveacategoricalbasis,andwehavenoindependentcharacterizationofthatbasis.Butevenifthethesisisrejected,type-Fmonismisstillviable.Weneedonlythethesisthatmicrophysicaldispositionsmayhaveacategoricalbasistoopenroomforintrinsicpropertieshere.32Hencetype-Fmonismisthesortof“physicalism”thatemergesfromtheloopholementionedinthetwo-dimensionalargumentagainsttype-Bmaterialism.Theonlywaya“zombieworld”WcouldsatisfytheprimaryintensionbutnotthesecondaryintensionofPisforittosharethedispositionalstructureofourworldbutnottheunderlyingintrinsicmicrophysicalproperties.IfthisdifferenceisresponsibleforthelackofconsciousnessinW,thentheintrinsicmicrophysicalpropertiesinourworldareresponsibleforconstitutingconsciousness.Maxwell(1979)exploitsthissortofloopholeinreplyingtoKripke’sargument.NotethatsuchaWmustinvolveeitheradifferentcorpusofintrinsicpropertiesfromthoseinourworld,ornointrinsicpropertiesatall.Atype-Fmonistwhoholdsthattheonlycoherentintrinsicpropertiesareprotophenomenalpropertiesmightendupdenyingtheconceivabilityofzombies,evenunderastructural-functionaldescrip-tionoftheirphysicalstate–forreasonsverydifferentfromthoseofthetype-Amaterialist.33McGinn(1989)canbereadasadvocatingatype-Fview,whiledenyingthatwecanknowthenatureoftheprotophenomenalproperties.Hisargumentsrestsontheclaimthatthesepropertiescannotbeknowneitherthroughperceptionorthroughintrospection.Butthisdoesnotruleoutthepossibilitythattheymightbeknownthroughsomesortofinferencetothebestexplanationof(introspected)phenom-enology,subjecttotheadditionalconstraintsof(perceived)physicalstructure.34Inthisway,wecanseethattype-Dviewsandtype-Fviewsarequitecloselyrelated.Wecanimaginethatifatype-Dviewistrueandtherearemicrophysicalcausalgaps,wecouldbeledthroughphysicalobservationalonetopostulatehigher-levelentitiestofillthesegaps–“psychons,”say–wherethesearecharacterizedinwhollystruc-tural/dispositionalterms.Thetype-Dviewaddstothisthesuggestionthatpsychonshaveanintrinsicphenomenalnature.Themaindifferencebetweenthetype-Dviewandthetype-Fviewisthatthetype-Dviewinvolvesfundamentalcausationabovethemicrophysicallevel.Thiswillinvolveamoreradicalviewofphysics,butitmighthavetheadvantageofavoidingthecombinationproblem.139\nDavidJ.Chalmers35Type-OpositionsareadvocatedbyLowe(1996),Mills(1996),andBealer(forthcoming).ReferencesAlbert,D.Z.(1993).QuantumMechanicsandExperience.Cambridge,MA:HarvardUniversityPress.Bealer,G.(1994).“MentalProperties.”JournalofPhilosophy,91:185–208.Bealer,G.(forthcoming).“MentalCausation.”Block,N.andStalnaker,R.(1999).“ConceptualAnalysis,Dualism,andtheExplanatoryGap.”PhilosophicalReview,108:1–46.Broad,C.D.(1925).TheMindanditsPlaceinNature.London:RoutledgeandKeganPaul.Campbell,K.K.(1970).BodyandMind.London:Doubleday.Carruthers,P.(2000).PhenomenalConsciousness:ANaturalisticTheory.Cambridge:Cam-bridgeUniversityPress.Chalmers,D.J.(1995).“FacinguptotheProblemofConsciousness.”JournalofCon-sciousnessStudies,2:200–19.ReprintedinShear(1997).http://consc.net/papers/facing.html.——(1996).TheConsciousMind:InSearchofaFundamentalTheory.Oxford:OxfordUniversityPress.——(1997).“MovingForwardontheProblemofConsciousness.”JournalofConscious-nessStudies,4:3–46.ReprintedinShear(1997).http://consc.net/papers/moving.html.——(1999).“MaterialismandtheMetaphysicsofModality.”PhilosophyandPhenom-enologicalResearch,59:473–93.http://consc.net/papers/modality.html.——(2002a).“TheContentandEpistemologyofPhenomenalBelief.”InQ.SmithandA.Jokic(eds.),Consciousness:NewPhilosophicalEssays.Oxford:OxfordUniversityPress.http://consc.net/papers/belief.html.——(2002b).“DoesConceivabilityEntailPossibility?”InT.GendlerandJ.Hawthorne(eds.),ConceivabilityandPossibility.Oxford:OxfordUniversityPress.http://consc.net/papers/conceivability.html.——(forthcoming).“TheFoundationsofTwo-DimensionalSemantics.”http://consc.net/papers/foundations.html.Chalmers,D.J.andJackson,F.(2001).“ConceptualAnalysisandReductiveExplana-tion.”PhilosophicalReview,110:315–61.http://consc.net/papers/analysis.html.Churchland,P.M.(1996).“TheRediscoveryofLight.”JournalofPhilosophy,93:211–28.Churchland,P.S.(1997).“TheHornswoggleProblem.”InShear(1997).Clark,A.(2000).“ACaseWhereAccessImpliesQualia?”Analysis,60:30–8.Dennett,D.C.(1991).ConsciousnessExplained.Boston,MA:Little,Brown.——(1996).“FacingBackwardontheProblemofConsciousness.”JournalofConscious-nessStudies,3:4–6.——(forthcoming).“TheFantasyofFirst-PersonScience.”http://ase.tufts.edu/cogstud/papers/chalmersdeb3dft.htm.Dretske,F.(1995).NaturalizingtheMind.Cambridge,MA:MITPress.Evans,G.(1979).“ReferenceandContingency.”TheMonist,62:161–89.140\nConsciousnessanditsPlaceinNatureFeigl,H.(1967[1958]).“The‘Mental’andthe‘Physical’.”MinnesotaStudiesinthePhilosophyofScience,2:370–497.Reprinted(withapostscript)asThe“Mental”andthe“Physical”.UniversityofMinnesotaPress.Foster,J.(1991).TheImmaterialSelf:ADefenceoftheCartesianDualistConceptionoftheMind.Oxford:OxfordUniversityPress.Griffin,D.R.(1998).UnsnarlingtheWorld-Knot:Consciousness,Freedom,andtheMind-BodyProblem.Berkeley:UniversityofCaliforniaPress.Harman,G.(1990).“TheIntrinsicQualityofExperience.”PhilosophicalPerspectives,4:31–52.Hill,C.S.(1997).“Imaginability,Conceivability,Possibility,andtheMind–BodyProb-lem.”PhilosophicalStudies,87:61–85.Hodgson,D.(1991).TheMindMatters:ConsciousnessandChoiceinaQuantumWorld.Oxford:OxfordUniversityPress.Huxley,T.(1874).“OntheHypothesisthatAnimalsareAutomata,anditsHistory.”FortnightlyReview,95:555–80.ReprintedinCollectedEssays.London,1893.Jackson,F.(1979).“ANoteonPhysicalismandHeat.”AustralasianJournalofPhilosophy,58:26–34.——(1982).“EpiphenomenalQualia.”PhilosophicalQuarterly,32:127–36.——(1994).“FindingtheMindintheNaturalWorld.”InR.Casati,B.Smith,andG.White(eds.),PhilosophyandtheCognitiveSciences.Vienna:Holder-Pichler-Tempsky.James,W.(1890).ThePrinciplesofPsychology.HenryHoltandCo.Kaplan,D.(1989).“Demonstratives.”InJ.Almog,J.Perry,andH.Wettstein(eds.),ThemesfromKaplan.NewYork:OxfordUniversityPress.Kirk,R.(1974).“ZombiesvsMaterialists.”ProceedingsoftheAristotelianSociety(Supple-mentaryVolume),48:135–52.Kripke,S.A.(1980).NamingandNecessity.Cambridge,MA:HarvardUniversityPress.Levine,J.(1983).“MaterialismandQualia:TheExplanatoryGap.”PacificPhilosophicalQuarterly,64:354–61.——(2000).PurpleHaze:ThePuzzleofConsciousExperience.Cambridge,MA:MITPress.Lewis,D.(1988).“WhatExperienceTeaches.”ProceedingsoftheRussellianSociety(Uni-versityofSydney).——(1994).“ReductionofMind.”InS.Guttenplan(ed.),CompaniontothePhilosophyofMind.Oxford:Blackwell.Loar,B.(1990/1997).“PhenomenalStates.”PhilosophicalPerspectives,4:81–108.RevisededitioninN.Block,O.Flanagan,andG.Güzeldere(eds.),TheNatureofConsciousness.Cambridge,MA:MITPress.Lockwood,M.(1989).Mind,Brain,andtheQuantum.Oxford:OxfordUniversityPress.Lowe,E.J.(1996).SubjectsofExperience.Cambridge:CambridgeUniversityPress.Lycan,W.G.(1996).ConsciousnessandExperience.Harvard,MA:MITPress.Maxwell,G.(1979).“RigidDesignatorsandMind–BrainIdentity.”MinnesotaStudiesinthePhilosophyofScience,9:365–403.Maxwell,N.(1968).“UnderstandingSensations.”AustralasianJournalofPhilosophy,46:127–45.McGinn,C.(1989).“CanWeSolvetheMind–BodyProblem?”Mind,98:349–66.Mills,E.(1996).“InteractionismandOverdetermination.”AmericanPhilosophicalQuar-terly,33:105–15.141\nDavidJ.ChalmersNagel,T.(1974).“WhatIsItLikeToBeaBat?”PhilosophicalReview,83:435–50.Nordby,K.(1990).“VisioninaCompleteAchromat:APersonalAccount.”InR.Hess,L.Sharpe,andK.Nordby(eds.),NightVision:Basic,Clinical,andAppliedAspects.Cambridge:CambridgeUniversityPress.Papineau,D.(1993).“Physicalism,Consciousness,andtheAntipatheticFallacy.”Aus-tralasianJournalofPhilosophy,71:169–83.Perry,J.(2001).Knowledge,Possibility,andConsciousness.Cambridge,MA:MITPress.Popper,K.andEccles,J.(1977).TheSelfandItsBrain:AnArgumentforInteractionism.NewYork:Springer.Quine,W.V.(1951).“TwoDogmasofEmpiricism.”PhilosophicalReview,60:20–43.Rey,G.(1995).“TowardaProjectivistAccountofConsciousExperience.”InT.Metzinger(ed.),ConsciousExperience.Paderborn:FerdinandSchöningh.Robinson,W.S.(1988).BrainsandPeople:AnEssayonMentalityanditsCausalCondi-tions.Philadelphia:TempleUniversityPress.Rosenthal,D.M.(1997).“ATheoryofConsciousness.”InN.Block,O.Flanagan,andG.Güzeldere(eds.),TheNatureofConsciousness.Cambridge,MA:MITPress.Russell,B.(1927).TheAnalysisofMatter.London:KeganPaul.Ryle,G.(1949).TheConceptofMind.London:HutchinsonandCo.Seagar,W.(1995).“Consciousness,InformationandPanpsychism.”JournalofConsciousnessStudies,2.Sellars,W.(1981).“IsConsciousnessPhysical?”TheMonist,64:66–90.Shear,J.(ed.)(1997).ExplainingConsciousness:TheHardProblem.Cambridge,MA:MITPress.Shoemaker,S.(1975).“FunctionalismandQualia.”PhilosophicalStudies,27:291–315.Smart,J.J.C.(1959).“SensationsandBrainProcesses.”PhilosophicalReview,68:141–56.Stalnaker,R.(1978).“Assertion.”InP.Cole(ed.),SyntaxandSemantics:Pragmatics,Vol.9.NewYork:AcademicPress.Stapp,H.(1993).Mind,Matter,andQuantumMechanics.Berlin:Springer-Verlag.Stoljar,D.(2001).“TwoConceptionsofthePhysical.”PhilosophyandPhenomenologicalResearch,62:253–81.Strawson,G.(2000).“RealisticMaterialistMonism.”InS.Hameroff,A.Kaszniak,andD.Chalmers(eds.),TowardaScienceofConsciousnessIII.Cambridge,MA:MITPress.Swinburne,R.(1986).TheEvolutionoftheSoul.Oxford:OxfordUniversityPress.Tye,M.(1995).TenProblemsofConsciousness:ARepresentationalTheoryofthePhenom-enalMind.Cambridge,MA:MITPress.VanGulick,R.(1993).“UnderstandingthePhenomenalMind:AreWeAllJustArmadil-los?”InM.DaviesandG.Humphreys(eds.),Consciousness:PhilosophicalandPsycholo-gicalAspects.Oxford:Blackwell.White,S.(1986).“CurseoftheQualia.”Synthese,68:333–68.Wigner,E.P.(1961).“RemarksontheMind–BodyQuestion.”InI.J.Good(ed.),TheScientistSpeculates.London:BasicBooks.142\nTheBlackwellGuidetoPhilosophyofMindThoughtsandTheirContentsEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter6ThoughtsandTheirContents:NaturalizedSemanticsFredAdams6.1OverviewFamously,Wittgensteinaskedthequestion“Whatmakesmythoughtaboutyouathoughtaboutyou?”IfIdohaveathoughtaboutyou,let’ssaythatyouareapartofthecontentofmythought.Youareapartofwhatmythoughtisabout.Wecanthinkaboutallsortsofthings:objects(theEiffelTower),properties(beingafamouslandmark),relations(beingEastofLondon),events(thetower’sconstruction),andthoughtsthemselves(thethoughtthattheEiffelTowerisoneofParis’smostfamouslandmarks).Thisisnotintendedtobeexhaustive,buttohelpbroadenthequestionto“whatmakesone’sthoughtaboutxathought12aboutx?”Weknowwehavethoughtsaboutthings.Whatwewillbeinterestedinhereareaccountsofhowthishappens.Wewillfocusonthoughtsandtheircontents,butbeliefs,desires,hopes,wishes,intentions,andsoonareoftenlooselyconsideredthoughts.AndDescartes,amongothers,wouldhaveincludedsensationsaskindsofthoughts,butitiscustomarytoconsiderthemdifferently,sincetheyarenotpropositionalattitudesanddonothavetruth-values(thoughtheymaybeveridicalornon-veridical).Sensationsclearlyhavecontentsandonsomeaccounts(Dretske1995)thereisaremarkablesimilaritytohowtheyandthoughtsacquiretheircontents.Sincethelate1970sandearly1980stherehavebeenseveralattemptstonaturalizesemantics.Whiletherearesubtledifferencesbetweenthevariousattempts,theysharetheviewthatmindsarenaturalphysicalobjects,andthatthewaytheyacquirecontentisalsoanatural(orphysical)affair.Atleastsincethemid1970s,externalistictheoriesofcontenthaveurgedthatthoughtcontentsdependcruciallyuponone’senvironment,anddonotdependsolelyuponwhatisinsidethehead(formostthoughts).Theverysamesortofphysicalstatethatisathoughtofwater(H2O)143\nFredAdamsinAl’sheadonEarthmaybeathoughtoftwin-water(XYZ)inTwin-Al’sheadonTwin-Earth.ThedifferenceofthoughtcontentisnotduetoanythinginternaltoAlorTwin-Al(themselvesphysicalduplicates),butduetodifferencesinthewaterysubstancesintheirrespectiveenvironments.Whatthenaturalizersofmean-ingaddtothepictureofmeaningexternalismisamechanism.Weneedanaccountofthemechanismthatexplainshowexternalphysicalobjectsbecomecorrelatedwiththeinternalphysicalstatesofone’shead(mind)suchthattheinternalphysicalstatescometomeanorbeabouttheexternalphysicalobjects.Naturalistictheoriesofcontentoffernaturalisticmechanisms.Meaningmechanismscannotrelyuponmeaningorcontent.Thegoalistonaturalizemeaningandexplainhowmeaningfulbitsofnatureariseoutofnon-meaningfulbits.Sowecannotrelyonthemeaningsofwordsorintentionsofagentstoexplainhowthoughtsacquirecontents.Ofcourse,oncecontentfulthoughtsexistandmeaningfullanguageexists,thesemayexplainhowfurthermean-ingorcontentarises.Butweneedsomeunmeantmeanerstogetthingsrolling.Naturalisticaccountsofthoughtcontentmustappealtomechanismsthatgeneratethoughtsandcontentwithoutusingthoughtsorcontentintheexplanation–atleastinitially.Perhapsawaytothinkaboutnaturalismistoaskhowthefirstmind3couldthinkitsfirstthought(s).Whatconditionswouldmakethispossible?Inthischapter,wewilllookattwoofthemoreprominenttheoriesthatattempttonaturalizesemantics.Wewillconsidermechanismsthatgeneratethoughtcontentonthesetheories,andthenconsiderimportantobjections.Therearefartoomanytheoriesandissuestocoveralloftheimportantones,butwhatweloseinbreadthwewillgainindepth.Manyoftheissuesarisefortheothertheoriesaswell.6.2AMediumforThoughtInorderforthoughtstoacquirecontenttheyneednotonlyamechanismbutalsoamedium.WhenAlthinksthattheEiffelTowerisinParis,histhoughtisinpartabouttheToweritself,inpartaboutParis,andinpartaboutthegeographicalrelationoftheonetotheother.Howthoughtsareabletodothis,tobesensitivetoobjects,properties,andrelations,isindispute.Otherchaptersinthisvolumewillemphasizetheoptionsforthecognitivearchitectureofamind:classicism,connectionism,andmore.Thecorrectviewmustshowhowdifferentpartsofthoughtarededicatedtodifferentpartsorfeaturesoftheenvironment.Thiswillinvolvedifferentiationofphysicalstatesofthemind(brain)toserveasdifferentially4representativevehiclesforthought.SomethingthatwascompletelyuniformwouldnotbeabletorepresentorthinkthattheEiffelTowerisinParis.Onewaythismightgoisifthereisalanguageofthought(LOT),asymbolsys-temthatmirrorsapublic,naturallanguageinstructure.AverygoodreasontothinkthatLOTisnotapublic,naturallanguageisthatweneedtheresourcesofalan-guageinordertolearnafirstnaturallanguage,viz.hypothesisformationabout144\nThoughtsandTheirContentswhatwordsandphrasesmeanandconfirmationprocedurestotestthosehypo-theses(Fodor1975).Thus,wehavetobeabletothinkinordertolearnourfirst5naturallanguage.Itisevenarguedthatjustaboutallofouradultthoughtsareinanaturallanguage–atleastforconsciousthoughts(Carruthers1996).Whether6ornotsomethoughtsareactuallythoughtinapublic,naturallanguage,theyseemexpressibleinanaturallanguage.WhenAlthinksthattheEiffelTowerisinParis,itiswidelyagreedthatAlcanexpresshisthoughtinEnglish(asanEnglishspeaker).Onewaythismightbetrueisif,correspondingtoeachelementintheexpressionofthethought,thereisanelementinthethoughtitself.Therewouldbeanelementforthedefinitearticle“the,”anelementtostandfortheEiffelTower,anelementtostandforParis,andanelementtostandfortherelationofbeingin.Essentially,ifthelanguageofthoughtisasymbolsystemwithacompositionalsyntaxandsemanticsthatareisomorphictothecompositionallogicalsyntaxandsemanticsofnaturallanguage(Harman1973),plusorminusabit(Fodor1981),thatwouldexplainhowthoughtscanbeexpressedinnaturallanguage.Ofcourse,mattersarenevereasy,andmanyissuesaboutsuchanisomorphismremainunresolved(Fodor1975).Butsurelytherearedependencyrelationsandfunctionsfromtheonetotheotherthatpreservecontent.Thatmuchseemsclear.Anditseemssafetosaythatthoughtsarepartofasymbolicsystembecausetheyhaverepresenta-tionalcharacteristicsthatdependontheirstructure(Harman1973:59).Thought’smediummakesitproductiveinjustthewaythatnaturallanguagesareproductive.Wecanthinkthat1isthepositiveintegerthatislessthan2,whichistheintegerthatislessthan3,whichistheintegerthatislessthan4,...Yougetthepicture.Wecandothistypeofiterationandcompositionforthoughtsofunboundedcomplexity.Themediumofthoughtalsomakesitsystematic–ifonecanthinkthatobjectastandsinrelationRtoobjectb,thenonecanthinkthatobjectbstandsinrelationRtoobjecta.Whatexplainsthesefeaturesisindispute,7butatleastonethingthatseemsjustrightforexplainingitisalanguageofthought.Afurtherreasontothinkthereisasymbolicmediumofthoughtalsotiesintothenaturalisticprogram.Intelligentbehaviorseemstodependuponthecontentsofourthoughts(beliefs,desires,intentions,sensations,andsoon).Jerry’sgoingtothefridgetogetabeer(Dretske1988)seemstorequirethatJerry’sreasonsforgoingtothefridgearerepresentedinhismind.Somethinggetshimtothefridge.Some-thingelsegetshimreachingforthebeer.Folkpsychology(andcognitivescience,too)seemstocryoutforvehiclesofthoughtthatplayanexplanatoryroleinguid-ingJerrytothefridgeandtheninguidinghisreachingforthebeer.Theseseemtorequiredifferentcausalelementsguidingdifferentportionsofhistotaltrajectory.Furthermore,considerlinguisticbehavior.IfIsaythatKenistallerthanGarybecauseIbelievethisanddesiretocommunicateit,therewouldseemtobedistinctelementsproducingmysaying“Ken”and“Gary,”etc.TheintentionalrealistandsemanticnaturalistwhoalsoembracesLOTtriestoexplainpurposivebehaviorbyappealtothecontentsofone’spropositionalattitudesandotherthoughts.Onetriestoaccountforthecontentsofone’sthoughtsascomputationaloperations(takenquiteliterally)overinternalformulae(orsequencesofformulae)(Fodor145\nFredAdams1975)inLOT.Onthisview,thinkingthataisFisstandinginthecomputationalrelationtoasymbolinthelanguageofthoughtthatmeansthataisF.Therefore,thethoughtsymbolsfor“a”andfor“F”havetobeabletocauseorexplainthebehaviorthatoneintelligentlyproduces,withrespecttoaandtoF(leta=theparticularbottleofbeerandF=beingopenedbyJerry).IfwearetoexplainJerry’sopeningthebeerbyappealtohisdesiretoopenitandhisbeliefthathecandosothusly(themannerofopening),hisinternalthoughtsymbolsorvehiclesmust8beabletocausebehavior(ormovements)invirtueoftheircontents.Alreadywecanseehowinterestingthingscanget.Howdoesthisworkforvacuousthoughtswheretheterm“a”or“F”isvacuous(planetVulcanissmall,phlogistonhasnegativeweight)?Howcanbehaviorbeexplainedinvirtueofthecontentsofone’sthoughtsinthosecases(Adamsetal.1993)?Indeed,whatisthecontentofone’sthoughtinthosecases(AdamsandStecker1994;EverettandHofweber2000)?Howcouldoneeverthinktrulyathoughtoftheform“adoesnotexist?”AndifaandFareexternaltotheheadandarethecontentsofthethoughtvehicles“a”and“F,”howcantheexternalcontent(whatisknownas“wide”or“broad”content)becausallyrelevanttowhat“a”and“F”cancause?Howcanbroadcontentberelevanttotheexplanationofintelligentbehavior(Adamsetal.1990;Adams1991;Adamsetal.1993)?StandardFregepuzzlescanbeseenfromthiscontext,aswell.IfIthink“aisF”anda=b,haveItherebythoughtthatbisF,ornot?Indeed,standardFregecasesprovideanotherexcellentreasonwhytherejustaboutmustbethoughtvehicles.WhenIthinkthataisF,ImightbehavedifferentlythanwhenIthinkthatbisF,evenwhena=b.Howisthatpossible?LOTprovidestheanswerthatthethoughtvehicle“a”isnotidenticalto“b.”Mymindmayconcatenate“a”with“F,”butnot“b”with“F,”eventhougha=b(butIdon’tknowthat).SowhileIamblindfolded,BernieSchwartzmayentertheroom.Imaybelievethathedid,butnotaskhimaboutJamieLeeoraskforhisautograph.AlthoughBernieSchwartz=TonyCurtis(someonewhomIwouldaskaboutJamieLeeCurtisorfromwhomIwouldrequestanautograph),myminddoesnotconcatenatemymentalvehicleforTonyCurtis“b”withmymentalvehicleforbeinginthisroom“F.”Iliterallyhavean“Fa”(“BernieSchwartzishere”)inmymindbutno“Fb”(“TonyCurtisishere”)(AdamsandFuller1992).Letusnowsupposethatthoughtshavevehiclesandtheytakeexternalobjects,properties,andrelationsascontents(atleastveryoften,ifnotalways),andthatweareworkingwithnaturalcauses.Nowletusconsidersomemeaningmechanisms.6.3Naturalization9Thenaturalizationofsemanticsisreallyaboutthemechanismsthatconnectthoughtvehicles(symbols)withtheircontents.ThelineofinfluencegoesbackatleasttoGrice(1957),andrunsthroughStampe(1977),Dretske(1981,1988),146\nThoughtsandTheirContentsandFodor(1987,1990a),tonameonlysomeofthekeyplayers.Naturalizationisanattempttocapturethemechanismsofcontentandexplainhowobjectsofthoughtbecomepairedwiththoughtvehicles.ThestorybeginswithGrice’snotionof“naturalmeaning.”Thisnotioniscloselylinkedwiththenotionsof“information”and“indication.”Allthreeareaboutpropertycorrelations(anddependencies).If,underlocallystableenviron-mentalconditions,thingswithpropertyGarecorrelatedwiththingswithprop-ertyF,inarelationofnomicdependency,thentheoccurrenceofsomething’sbeingGcanbeanaturalsignorindicatorofsomething’sbeingF.Smoke(G)naturallymeansfire(F).Footprintsinthesnowindicatethatsomeonewalkedthroughthesnow.Ringsinthetreecarryinformationabouttheageofthetree.Thethermometer’srisingindicatesrisingtemperature.Fornaturalmeaning(indicationorinformation)toexist,thesepropertyde-pendenciesmustbelocallystable.Theremustnotbecausaloverdetermination(artificialsmoke,artificialfootprints,ortree-boars),andtheremustbenootherfactorsthatwoulddisruptsuchdependencies(seasonsofnon-tree-growth,imper-ceptiblecracksinthethermometer).Theneedtospecifythesedependenciesled10Dretskeawayfromanearlyformulation(“therewouldn’tbesmokeunlesstherewerefire”)toaninformationtheoreticone(“theprobabilityoffire,givensmoke,mustbe1(unity)”).Subtledifferencesaside,naturalmeaning(orindicationorinformation)hasbeentherefromthestartofthenaturalizationproject–withgoodreason.IfsomethinginAl’sheadisgoingtomeanorbeaboutfire,thenAlneedsathoughtvehiclethatcannaturallymeanfireassurelyassmokenaturallymeansfire.Perhapsthethoughtvehicleitselfiscausedbyperceptualmechanismsthataretriggeredbysensorydetectionoffire(ortherearesymbolsinthepercep-11tualsystemthatnaturallymeanfireandinturncausesymbolsinthecentralsystemthatcometomeanfire).Thisrequiresanidentitybetweentheenvironmental(orecological)conditionsnecessaryforknowledgeandthosenecessaryforunivocalcontent(Dretske1989).Supposethatinone’senvironmentitisnotpossibletoknowthatsomethingisFbyevidencethatsomethingisG.Supposethisisbecause,inthisenvironment,thingsthatareGarealsonomicallycorrelatedwith(anddependentupon)thingsthatareH–supposeGsarealternatelycausedbyFsorHs.How,insuchanenvironment,couldonebuildadetectormechanismforFs,outofadetectorofGs?Onecouldnot.SinceGsaredependentonFsorHs,suchadetectorwouldbeofFsorHs,notofFsalone.InanenvironmentwhereGsarereliablydependentupon(andcorrelatedwith)FsorHs,something’sbeingGdetectsthatsome-thingisForH.Callthisthedisjunctionproblem.Withrespecttoknowledge,themostsuchadetectorcouldtellusisthatsomethingisForH.ThisisbecausethemostitcouldindicateornaturallymeanisthatsomethingisForH.Withrespecttothoughts,ifthoughtcontentderivesfromnaturalmeaning,fromdisjunctivenaturalmeaning,disjunctivethoughtcontentderives.Toavoidthis,thenaturaliza-tionprojecthastosolvethedisjunctionproblemandexplainhowathoughtsymbolmayhaveunivocalmeaning.Inthecaseathand,if“G”wereathought147\nFredAdamssymbolthathadonlydisjunctivenaturalmeaning,atbestitwouldallowonetothinkadisjunctivethoughtaboutFsorHs(notaboutFsalone).Further,tobeathoughtsymbolatall,“G”wouldhavetorisetoalevelabovenaturalmeaning–asweshallsoonsee.Fornaturalmeaningtobepartofamechanismthatgeneratesunivocalthoughtcontent,itmustspringfromnon-disjunctivenaturalmeaning.Wecanalsoseethatsomething’shavingunivocalnaturalmeaningisjustthesortofthingthatisrequiredtoknowthatsomethingisFbyitsbeingG.Thisiswhythereisaconnectionbetweenknowledgeandthoughtcontent.Ifone’senvironmentisnotlocallystableenoughtoknowthatsomethingisF(univocallybysomeG),thenitisnotstableenoughtoacquireanon-disjunctivethoughtsymbol“G”inthedeploymentofthethoughtthatsomethingisF(alone).Putnam(1975)givestheexampleofjadeite(F)andnephrite(H).SupposethatIdon’tknowthatjadecomesintwovarieties.I’veheardtheterm“jade”butdon’tknowwhatitmeans.Youshowmesomejadeite,butbothjadeiteandnephritelookexactlyaliketome(G).ThenIcannotbytheirlook(G)knowifI’mseeingjadeite(F)ornephrite(H).NorcouldIformaunivocalthought(“thisisjade”)ofjadeitealone.Mythoughtwouldbeasmuchaboutjadeiteasnephrite–thoughIhaveonlythethoughtsymbol“jade”(whateverthatwouldbeinLOT).Naturalmeaning(indicationorinformation),therefore,isanimportantingre-dientofthemechanismthatunderwritesthoughtcontent.Still,thoughtcontentcannotbemerelyamatterofnaturalmeaning,forindicationandthoughthavedivergentproperties.WhenAlseesaparticularbeer,hisperceptualsymbolsofthebeermaynaturallyindicatepresenceofbeer.AndthismaycauseAltothinkthat12thereisabeerpresent.ButAlmaythinkofbeerwhenthereisnonepresent,whenhewishessomewere.Al’sperceptualmechanismsdon’tworkthisway(barringdreaming,hallucination,orsomethingoutoftheordinary).Perceptualmechanismsaretunedintowhatishappeningnow.Thoughtsareabletofocusonthehereandnow(via“here”and“now”),butarenotboundbythepresent.Thisgivesthoughtanelementoffreedomthatperceptionandsensation(whenveridical)donothave.Unfortunately,thesamecognitiveabilitythatfreesustothinkfreesustothinkfalsely.WhenAl’sperceptualmechanismsareworkingproperly,hewillnotsee13beerthatisn’tthere.Butwhenhisthinkingmechanismsareworkingproperly,hemaywellthinkthereisabeer(inthefridge)thatisn’tthere(hemaylosecount,someonehetrustsmaytellhimthereisoneleft).Thatthiscanhappenisnotacognitivedeficit.Indeed,itisabenefitthatthemindcanfreeitselffromitsimmediateenvironmentalcontingencies.ButthefactthatAlcanfalselythinksomethingoftheform“Fa”tellsusthatthoughtcontentisnotnaturalmeaning.Gricecalledit“non-naturalmeaning.”Iprefertocallitsemanticcontent–contentthatcanbefalselytokened.If“Fa”hasnaturalmeaning,amustbeF,butif“Fa”hassemanticcontent,aneednotbeF.Thequestionbecomeshowsomethinggoesfromnaturalmeaningtosemanticthoughtcontent?Forinstance,148\nThoughtsandTheirContentssmokenaturallymeansfire.Al,thinkinghesmellssmoke,maymentallytoken“smoke.”But“smoke”semanticallymeanssmoke,notfire.Howisthispossible?Howdoessomething“fire”gofromindicatingfire(therebyrequiringfire’spresence)tosemanticallymeaningfire(notrequiringfire’spresence)?6.4MechanismsofMeaningWhatisrequiredtomakethejumpfromnaturalmeaningtosemanticthought14contentisthatasymbolbecomesdedicatedtoitscontent.For“F”tohaveFsasitssemanticcontent,thesymbolmustbecomededicatedtothepropertyofbeinganF.ItmustmeanFswhetheroneiscurrentlythinkingofsomethingthatisF,thatitisForofsomethingthatisnotF,thatitisF.Indeed,itmusthavethecontentthatsomethingisFevenifitistokenedbythoughtsunrelatedtowhethersomethingisanF.Thiswouldsecurethepossibilitybothofrobustandfalsetokening–twoofthepropertiesthatdistinguishthoughtsfromperceptsandotheritemswithonlynaturalmeaning.Sotheproblemistoarticulateamechanismofdedication.Dretske(1981)oncesuggestedthepossibilityofalearningperiod–atimeperiodduringwhichaconceptformedandacquireditsmeaning.Letusthinkofaconcept,forourpurposeshere,asathoughtsymbolorvehicle.Dretske’ssuggestionwasthatsomeonemightacquiretheconcept(an“F”)ofanFbybeingshownFsandnon-FsunderconditionsappropriatefordetectingFs.IfthepropertyofbeinganFisthemostspecificpieceofinformationthesubjectbecomesselectivelysensit-iveto(indigitalformofrepresentation),thenthesubject’s“F”-tokens(or“F”s)becomededicatedorlockedtoFs,aswemightputit.“F”sbecomeactivatedbyFsandFsonlyasthesubjectlearnstodiscriminateFsandnon-Fsshownduringthelearningperiod.Theideaofalearningperiodmakesperfectsense,ifonethinksofathoughtsymbollockingtoitscontentalongthelinesofababyduck’simprintingonits“mother.”Awindowofopportunityforcontentacquisitionopens,thesymbolisreceptiveofamostspecificpieceofinformation,lockstoit,andthewindowofopportunityforcontentacquisition(thelearningperiod)closes.Onsuchaview,alearningperiodmightjustwork.Itseemstoworkjustfineforimprintinginbabyducks.Theproblems(Fodor1990a)withalearningperiodarethatthereisnogoodreasontothinkthatconceptacquisitionisanythinglikeimprinting–withawindowclosingafteracertaintimeperiod.Andevenifthereweresuchawindowofopportunityforconceptstoformonspecificinstancesofobjectspresentedtoalearner,thereisnoguaranteethattheinformationdeliveredtothelearnerisexhaustedbythepropertiesofitemspresented.Consider“jade”again.Since,aswesupposed,Icannotdiscriminatejadeitefromnephrite,ifmythoughtsymbol“jade”istokenedexclusivelybyshowingmejadeiteduringthelearningperiod,theinformationdeliveredmaystillbethatsomethingisjadeiteornephrite.Somy149\nFredAdamssymbol“jade”islockingtojadeiteornephrite,eventhoughIamexclusivelyshownjadeite.Afterthelearningperiod,ifIamshownnephriteandittokensmysymbol“jade,”Iamnotfalselybelievingthatthisisjadeite,buttrulybelievingthatitisjadeiteornephrite.ThoughIwouldnotputitthatway,thisisthecontentofmythought.So,ineffect,thisexampleillustratestheproblemsforthelearningperiodapproach.Itneithersolvesthedisjunctionproblem,norexplainsthepossibilityoffalsehood.Dretskerevisedhisaccountofhowmisrepresentationwaspossible(1986),andfinallysettledonanevendifferentaccount(1988)whichnotonlyattemptstoexplainhowsymbolslocktotheircontents,butalsohowtheirhavingcontentisexplanatorilyrelevantforbehavior.However,beforelookingatthisaccount,letusconsiderFodor’sownapproachtomeaningmechanisms.6.5Fodor’sMeaningMechanisms15Fodor(1987,1990a,1994)offersconditionssufficientforasymbol“X”tomeansomethingX.Sinceheofferssufficientconditionsonly,hisviewinspiresconcernsthathisconditionsdon’tapplytous(ortoanyone).AndFodorisperfectlyhappyifthereareothersufficientconditionsformeaning(sincehisaren’tintendedtobenecessary).Asmuchaspossible,IhopetominimizetheseissuesbecauseitisprettyclearthatFodorwouldnotbeofferingtheseconditionsifhethoughttheydidn’tapplytous.Sowewillproceedasthoughhisconditionsaresupposedtoexplainthemechanismsbywhichourthoughtshavethecontentsthattheydo.Let’salsobeclearthatFodorisofferingconditionsforthemeaningsofprim-itive,non-logicalthoughtsymbols.Thismaywellbepartoftheexplanationofwhyheseeshisconditionsasonlysufficientformeaning.Thelogicalsymbolsandsomeotherthoughtsymbolsmaycomebytheirmeaningsdifferently.Symbolswithnon-primitive(molecular)contentmayderivefromprimitiveoratomicsymbolsbydecomposingintoatomicclusters.Itisanempiricalquestionwhensomethingisaprimitiveterm,andFodoristhefirsttorecognizethis.Stillhetriestoseehowfarhisaccountcanextendbytryingtodeterminewhetheritwouldapplytomanytermsnotnormallytakentobeprimitive(“unicorn,”“doorknob”).Fodor’sconditionshavechangedovertimeandarenotlistedbyhimanywhereintheexactformbelow,butIbelievethistobethebestrepresentationofhis16currentconsideredtheory.(ThisversionisculledfromFodor1987,1990a,and1994.)Thetheorysaysthat“X”meansXif:(1)“Xscause‘X’s”isalaw,(2)forallYsnot=Xs,ifYsquaYsactuallycause“X”s,thenY’scausing“X”sisasymmetricallydependentonXscausing“X”s,150\nThoughtsandTheirContents(3)therearesomenon-X-caused“X”s,(4)thedependencein(2)issynchronic(notdiachronic).Condition(1)representsFodor’sversionofnaturalmeaning(information,indi-cation).IfitisalawthatXscause“X”s,thenatokened“X”mayindicateanX.Whetheritdoeswilldependonone’senvironmentanditslaws,butthiscondi-17tionaffordsnaturalmeaningaroletoplayinthismeaningmechanism.Itisclearthatthisconditionisnotsufficienttomakethejumpfromnaturalmeaningtosemanticcontent.For“X”tobecomeasymbolforXsrequiresmorethanbeingtokenedbyXs.“X”smustbededicatedto,faithfulto,lockedtoXsfortheircontent.Condition(2)isdesignedtocapturethejumpfromnaturalmeaningtosemanticcontentandsolvethedisjunctionproblem,atthesametime.ItdoestheworkofDretske’slearningperiod,givingusanewmechanismforlocking“X”stoXs.Ratherthanawindowopeningandclosingwhere“X”sbecomededicatedtoXs,Fodor’sfixistomakeallnon-X-tokeningsof“X”snomicallydependentuponX-tokeningsof“X”sfromtheverystart.Thereisthennoneedfora18learningperiod.Theconditionsaysthatnotonlywilltherebealawconnectingasymbol“X”withitscontentX,butforanyotheritemsthatarelawfullycon-nectedwiththesymbol“X”,thereisanasymmetricaldependencyoflawsorconnections.Theasymmetryissuchthat,whileotherthings(Ys)arecapableofcausingthesymboltobetokened,theY→“X”lawdependsupontheX→“X”law,butnotviceversa.Butforthelatter,theformerwouldnothold.Hence,theasymmetricaldependenceoflawslocksthesymboltoitscontent.Condition(3)establishes“robust”tokening.Itacknowledgesthattherearenon-X-caused“X”s.Someoftheseareduetofalsethoughtcontent,aswhenImistakeahorseonadarknightforacow,andfalselytoken“cow”(believingthatthereisacowpresent).Othersareduetomereassociations,aswhenoneassoci-atesthingsfoundonafarmwithcowsandtokens“cow”(butnotacaseoffalsebelief).Thesetokeningsdonotcorruptthemeaningof“cow”because“cow”isdedicatedtocowsinvirtueofcondition(2).Condition(4)isdesignedtocircumventpotentialproblemsduetokindsofasymmetricaldependencethatarenotmeaningconferring(Fodor1987:109).ConsiderPavlovianconditioning.Foodcausessalivationinthedog.Thenabellcausessalivationinthedog.Itislikelythatthebellcausessalivationonlybecausethefoodcausesit.Yet,salivationhardlymeansfood.Itmaywellnaturallymeanthatfoodispresent,butitisnotathoughtorthoughtcontentanditisnotripeforfalsesemantictokening.Condition(4)allowsFodortoblocksayingthat19salivationitselfhasthesemanticcontentthatfoodispresent,foritsbell-causeddependencyuponitsfood-causeddependencyisdiachronic,notsynchronic.Firstthereistheunconditionedresponsetotheunconditionedstimulus,then,overtime,therecomestobetheconditionedresponsetotheconditioningstimulus.Fodor’sstipulationthatthedependenciesbesynchronicnotdiachronicscreensoffPavlovianconditioningandmanyothertypesofdiachronicdependencies,aswell.151\nFredAdamsThatwouldbetheendofthediscussionofFodor’smeaningmechanisms,ifitwerenotforafurtherhistoricalinstantiationcondition(HIC)thathasshownup(Fodor1990a),andsubsequentlydisappeared(Fodor1994),inFodor’swritings.ItisunquestionablethatthemiddleversionsofhistheorystateHIC.HIC:Some“X”sareactuallycausedbyXs.Thefactthatthiswasonceastatedconditioncomplicatesmatters.WithHICincludedasafifthcondition,(2)–(4)seemtobeconditionsonactualinstancesofcausation,notjustcounterfactuals(Warfield1994).Thismakesaratherimport-antdifference.ItmakesFodor’stheoryhistoricalinvirtueofrequiringactualempiricalencounterswithobjectsandtheirproperties.Thisisaproblem,giventhatFodorwantstosayasymbolsuchas“unicorn”maylocktouninstantiatedproperties,suchasthepropertyofbeingaunicorn(Fodor1990a,and1991).Further,condition(4)seemsonlytomakesenseifweincludesomethinglikeHIC.Withoutit,whatsensewoulditmaketosaythatadependencybetweenlawsisdiachronic(AdamsandAizawa1993)?Lawsaretimeless.SowithoutHIC,conditions(1)–(2),atleast,seemonlytobeaboutcounterfactuals,notinstancesoflaws(Fodor1994).HICmakesperfectlygoodsenseifoneisworriedaboutexcludingthoughtsforDavidson’s(1987)Swampmanoraccountingforthedifferencesofcontentof“water”hereoronTwin-Earth.Letmeexplain.Firstconsiderthecontentof“water.”InJerry,thethoughtsymbol“water”meanswater(ourwater,H2O).InTwin-Jerry,thethoughtsymbol“water”meanstwin-water(XYZ).Howisthatpossibleonconditions(1)–(4)alone?ThereisanH2O→“water”law.ButthereisalsoanXYZ→“water”law.SinceJerryandTwin-Jerryarephysicallytype-identical,thesamelawsholdofeach.Thereexistsnoasymmetricaldependencyoflawstofixunivocalcontent.ItmighthelptoinvoketheHIC.ForJerrydoesnotinstantiatetheXYZ→“water”lawandTwin-JerrydoesnotinstantiatetheH2O→“water”law.Thus,itwouldbepossibleforJerry’s“water”symboltolocktoonething,duetoactualcausalcontactwiththatkindofsubstance,whileTwin-Jerry’s“water”symbollockstoanotherkindofsubstanceviaactualcausalcontactwithit.Byincluding(HIC),atleastprimafaciethetheorywouldbeable20toexplainthesedifferencesofbroadcontent.Forthenthedependenciesof(2)wouldholdonlyfortheinstantiatedlaws.Inthesameway,thetheorywouldbeabletoexplainwhyDavidson’sSwampmanlacksthoughts.Hisvehicleslackcontent.AlthoughthesamecounterfactualsmaybetrueofJerryandofSwampJerry,sinceSwampJerryhasnocausaltruckwiththesameobjectsandpropertiesasJerry,SwampJerryfailstosatisfyhistoricalconditionHIC.Usefulthoughthisconditionmaybe,Fodorjettisonsitbecausehenow(1994)deniesthatTwin-Earthexamplesareproblemsthatneedtobeaddressed.HealsonowacceptsthatSwampJerryhasthesamethoughtsasJerry.Therefore,Fodor’sconsideredtheorydropsthiscondition.Laterwewillconsiderwhetherthisis152\nThoughtsandTheirContentswise.NextwewilllookatDretske’sconsideredviewandthenwewillexamineproblemsforbothnaturalizedtheories.6.6Dretske’sMeaningMechanismsDretske’srecipeforcontentinvolvesthreeinterlockingpieces.(i)Thecontentofasymbol“C”mustbetiedtoitsnaturalmeaningF(Fs–objectsthatareF).(ii)Naturalmeaning(indication,information)mustbetransformedtosemanticcontent.Theremustbeatransformationofperceptuallyacquiredinformationcontentintocognitive(semantic)content–encodedinaformcapableofbeingharnessedtobeliefsanddesiresinserviceoftheproductionofbehaviorM.(iii)ThecausalexplanationoftheresultantbehaviorMmustbeinvirtueofthecontentsofthecognitivestates(viatheirpossessionofcontent).Thus,ifasymbol“C”causesbodilymovementsMbecausetokeningsof“C”indicate(naturallymean)Fs,then“C”iselevatedfrommerelynaturallymeaningFstohavingthesemanticcontentthatsomethingisF.F←indicates“C”andcauses→M(becauseitindicatesF)WhileFodorflirtedwithanhistoricalaccountofcontent(viaHIC),Dretske’saccountiswaybeyondflirtation.Hisaccountisessentiallyhistorical.Indifferentenvironments,thesamephysicalnaturalsignsmaysignifydifferentthings,andhavedifferentnaturalmeaning.OnEarth,Al’sfingerprintsarenaturalsignsorindicatorsofAl’spresence.OnTwin-Earth,thesamephysicaltypesofprintsindicateTwin-Al’spresence,notAl’s.Forthistobetrue,theremustbesome-21thinglikeanecologicalboundarythatscreensoffwhatispossibleinoneenviron-mentfromwhatispossibleinanother.OnEarth,forAl’sprintstoindicateAl’spresence,theremustbeazeroprobabilityofthesetypesofprintsbeingleftbyTwin-Al(whocan’tgetherefromTwin-Earth,orwouldnotcomehere,letussuppose).Indeed,theremustbeazeroprobabilitythat,giventheoccurrenceoftheseprints,anythingbutAlmadethem.Ifthemoblearnshowtofakeprints,noprintsmayhaveunivocalnaturalmeaning.Sowhetheranaturalsignhasonenaturalmeaningoranotherwilldependupontheecologicalconditionsinwhichthesignoccurs.ThismakesDretske’stheoryhistoricaltothemax.Alllawsexisteverywhere,butnotalllawsareinstantiatedeverywhere.Sowhichlawsarerel-evantdependsuponwhereyouare,andyourhistoryofinteractionwithyourenvironment.Physicallyidenticalthoughtsymbols“S”indifferent,butqualitat-ivelysimilarorganisms,indifferentenvironments,mayacquiredifferentthought22contents.Whatcontentsthesymbolsacquirewilldependonwhatnaturalmean-ingstheycouldacquire,intheirrespectiveecologicalniches.Dretske’ssolutiontothedisjunctionproblemhasatleasttwocomponents.Thefirstcomponenthasalreadybeenaddressed.Thesymbol“C”muststartout153\nFredAdamswiththeabilitytonaturallymeanFs(andonlyFs).IfitindicatesFsorGs,thenadisjunctivecontentistheonlysemanticcontentitcouldacquire.Thesecondcomponentisthejumptosemanticcontent.Evenif“C”sindicateFsonly,toacquiresemanticcontent,asymbolmustloseitsguaranteeofpossessingnaturalmeaning.ItneedstobecomelockedtoFsandpermitrobust,andevenfalse,tokening,withoutinfectingitssemanticcontent.We’veseenwhyanappealtoa“learningperiod”doesn’tquitework,unembellished.Andwe’veseenthatFodortriestoturnthistrickwithasymmetricalcausaldependenciesoflaws.WhereFodorusesasymmetricaldependenciesoftokeningsof“C,”Dretskeappealstotheexplanatoryrelevanceofthenaturalmeaning.ForDretske,itisnotjustwhatcauses“C”s,butwhat“C”sinturncause,andwhytheycausethisthatisimportantinlocking“C”stotheircontent(F).Let’ssupposethatagroundsquirrelneedstodetectFs(predators)tostayalive.IfFscause“C”sinthegroundsquirrel,thenthetokeningof“C”sindicateFs.Dretskeclaimsthat“C”scometohavethecontentthatsomethingisanF,when“C”scometohavethefunctionofindicatingthepresenceofFs.Whenwillthatbe?Foreverypredatorisnotjustapredator,itisananimal(G),aphysicalobject(H),alivingbeing(I),andsoonformanyproperties.Hence,tokensof“C”willindicateallofthese,notjustFs.Dretske’sansweristhatwhen“C”sindicationofFs(alone)explainstheanimal’sbehavior,then“C”sacquirethesemanticcontentthatsomethingisapredator(F).Hence,itistheintensionalityofexplanatory23rolethatlocks“C”stoF,nottoGorHorI.ForDretske,behaviorisacomplexofamentalstate’scausingabodilymove-ment.Sowhen“C”causessomebodilymovementM(say,theanimal’smove-mentintoitshole),theanimal’smovementconsistsofitstrajectoryintoitshole.Theanimal’sbehaviorisitscausingthattrajectory.Theanimal’sbehavior–runningintoitshole–consistsof“C”scausingM(“C”→M).Thereisnospecificbehaviorthatisrequiredtoacquireanindicatorfunction.Sometimestheanimalslipsintoitshole(M1).Sometimesitfreezes(M2).Sometimesitscurriesaway(M3).Thisaccountsaysthat“C”sbecomerecruitedtocausesuchmovementsbecauseofwhat“C”sindicate(naturallymean).TheanimalneedstokeeptrackofFsanditneedstobehaveappropriatelyinthepresenceofFs(toavoidpreda-tion).Hence,theanimalthinksthereisapredatorwhenitstoken“C”causessomeappropriatemovementM(andhencetheanimalbehaves)becauseof“C”’sindication(naturalmeaning).Notuntil“C”’snaturalmeaninghasanexplanatoryroledoes“C”locktoitssemanticcontentF.So“C”’sacquiredfunctiontoindicateordetectpredatorselevatesitscontenttothenext,semanticlevel.Now“C”canbefalselyorotherwiserobustlytokened.Theanimalmayrunintoitsholebecauseitthinksthereisapredator,evenwhenspookedonlybyasoundorashadow,aslongasthepresenceofsoundsorshadowsdoesn’texplain24whythe“C”scauserelevantMs(don’texplaintheanimal’sbehavior).Soevenwhenfalselyorrobustlytokened,thesemanticcontentofthe“C”sisnotinfectedwithdisjunctivecontent.154\nThoughtsandTheirContentsNoticethatthisdoesn’texactlyrequirealearningperiodorwindowtoopenorclose,butitdoesrequireadeterminatefunction.ForDretskeanindicatorfunctionbecomesfixedwhenanexplanatoryroleforitsindicationisfixed(ordeterminate).25Couldthischangeovertime?Yes,itcould.Functionsofanykindcanchangewhentheconditionsoftheirsustainedcausingchange.Cognitivesystemsmayadapttochangesintheexternalenvironmentortheinternaleconomyofthecognitiveagent.Onthisview,indicatorfunctionsarelikeothernaturalfunctions,suchasthefunctionoftheheartorkidneysorperceptualmechanisms.TheaccountofnaturalfunctionsfavoredbyDretskeisoneonwhichtheXacquiresafunctiontodoYwhendoingYcontributessomepositiveeffectorbenefittoanorganismandsodoinghelpsexplainwhytheorganismsurvives.ThenthereisatypeofselectionfororganismswithXsthatdoY.Consequently,partofthereasonXsarestillpresent,stilldoingY,isthatatypeofselectionforsuchorganismshastakenplace.Ofcourse,thisdoesn’texplainhowXgotthereorbegandoingYinthebeginning.Naturally,theselectionforindicatorfunctionshastobewithinanorganism’slifetime,notacrossgenerations.Dretskethinksofthiskindofselectionasatypeofbiologicalprocessof“recruitment”or“learning”thatconformswithstandard,etiologicalmodelsofnaturalfunctions(Adams1979;AdamsandEnc1988;EncandAdams1998).Nowthethirdpieceofthepuzzleistoshowthatthecontentof“C”atsomelevelisrelevanttotheexplanationoftheorganism’sbehavior.“C”maycauseM,butnotbecauseofitsnaturalmeaning.“C”’smeaningmaybeidle.Forthispurpose,Dretskedistinguishestriggeringandstructuringcauses.Atriggeringcausemaybethethingthatcauses“C”tocauseMrightnow.Whereas,astructuringcauseiswhatexplainswhy“C”causesM,ratherthansomeothermovementN.Or,alternatively,structuringcausesmayexplainwhyitis“C”ratherthansomeotherstateofthebrainDthatcausesM.Sostructuringcauseshighlightcontrastives:(a)why“C”scauseM,oralternatively,(b)why“C”scauseM.Ineithercase,ifitisbecauseof“C”’snaturalmeaning,thenwehaveacaseofstructuringcausa-tion,andcontentplaysaroleonthisaccountofmeaningmechanisms.Let’sillustratethiswithacomparisonofanon-intelligentrobotcatandanintel-26ligentcat.Bothmayproducetheidenticalmovements,buttheirbehaviorsmaynotbethesame.Supposethatbothappeartobestalkingamouse.Thisdoesnotmeanthatbotharestalking,noteveniftherobotcat’s“brain”hasstructuresthatresemblethebrainofthecat.Fortheremaybenothingintherobotcatthatisinanywayaboutamouse.Therewillbesomethinginthecatthatisaboutthis,givenitslearninghistory.Itwillalsohaveadesiretocatchthemouseandbeliefsabouthowtodoso.Thoughtheirbodilymovementsmaybephysicallysimilar,itwouldbestretchingthingstosaytherobotcatwas“stalking”themouse,andteleologicalnonsensetotrytoexplainwhy.Sinceunintelligent,thebestwecoulddoisexplainhow.Itwouldbequitesensibletosaythatthecatwasstalkingthemouseanditwouldmakeperfectsenseteleologicallytoexplainwhy–itisstalkinginordertocatchthemouse.Insidetheworkingsoftherobotcat’s“brain”therearetriggering155\nFredAdamscausesofitsmovements.Butsincetherobotcat’sinternalstateshavenosemantic27content,therearenostructuringcausesthatproducemovementsbecauseofwhattheyindicateornaturallymean.Thus,thereisnointelligentbehaviorhere,unlikeinthecaseofthecat.Hence,semanticcontentmakesanimportantdifferenceintheoriginandexplanationofintelligentbehavior,onthisview.ThiscompletesthebasicsketchofthemeaningmechanismsofFodorandDretske.Let’sconsidersomeobjectionsthathavebeenraisedtobothaccounts.Thiswillhelpusseemoredeeplyintothenatureofthesetheoriesanddetecttheirstrengthsandweaknesses.6.7ObjectionsNamesNeitherFodor’snorDretske’stheoriesaredesignedtohandlenamesandtheircontents.Forboththeoriesaredesignedtoexplainhowthoughtsymbolsbecomelockedtoproperties.Sinceobjectshaveawealthofproperties,unlesstheyhaveindividualessences,thesetheoriesdonoteasilyaccountforthecontentsofnames.Namesanddemonstrativesarewidelythoughttohavetheirreferentsdeterminedbycausalchainsthatconnecttheirintroductionintoalanguage(orthoughtsystem).Aristotle’sfamilynamedhimandused“Aristotle”torefertohim.Thementalsymbolthatcorrespondstotheterminnaturallanguagealsogetsitsreferencedeterminedviathiscausalchain.Andthischaincanbepassedonfrompersontoperson,generationtogeneration.PerhapstheproblemiseasiesttoseeonFodor’stheory,sincehestateshistheoryintermsoflaws.For“Aristotle”tomeanAristotle,whenwelookatcondition(1)weseethatitmustbealawthat“Aristotlecauses‘Aristotle’s.”Thedifficultyisimmediatelyapparent.ThetheoryrequirestheindividualAristotletofeatureinalaw.Butlawsfeaturekindsofproperties,notindividuals.Sothetheoryisnotdesignedtohandlecontentsofnames(AdamsandAizawa1994a).Fodornoticedandtriedtofixthisproblem(1994:118)bysuggestingthattherelevantlawin(1)wouldbethis:“PropertyofbeingAristotle→‘Aristotle’s.”WhilehegetsanAforeffort,thisstillseemstomake“Aristotle”meanaproperty,notAristotle(theman)(AdamsandAizawa,1997a).Fodormaywanttoinsistthatforeveryindividual,thereisapropertyofbeingthatindividual.Butifitwerethiseasyfortheretobeproperties,whywouldanyoneeverhavethoughtthatindividualsdonotfeatureinlaws?Therecouldbeasmanysuchlawsasyouplease.Itseemsmuchmorelikelythatthereisadifferencebetweenpropertiesandindividualsandthatnameslike“Aristotle”nametheindividualandthatphraseslike“propertyofbeingAristotle”nameaproperty.SinceFodorisgivingonlysufficientconditionsformeaning,itwouldnotbetheendoftheworldifhistheorydidn’tapplytonames.Hehimselfsuggeststhat156\nThoughtsandTheirContentsitdoesn’tapplytodemonstrativesorlogicalterms.Perhapsacausaltheoryofreference,suchasthedirectreferencetheory,isadequatefornamesanddemonstratives(AdamsandStecker1994;AdamsandFuller1992).Namesinthoughtmayconnectonedirectlytoanindividual,supplyingthatindividualforthepropositionalcontentofathought(consistingofthatindividualandaprop-erty,relation,orsequence).Dretske’stheorytoohastobeableexplainhow“C”canmeanAristotle.Some-28thingmustindicateAristotle(forexample,fingerprintswould,DNAwould).Pre-sumably,Aristotlehadfeaturesviawhichhisfamilyrecognizedhim.“Aristotle”doesnotmeanthesefeaturesorproperties.“Aristotle”meansAristotle,butaconstella-tionoffeaturesinthatspaceandtimeuniquetoAristotlewouldpermitastructure“C”tobeselectivelysensitivetoAristotle’spresenceinvirtueofthem,andthereby29tonaturallymeanthatheispresent.Ofcourse,theremustbeacausalchainlinkingAristotletoperceptsofAristotleandperceptsto“C”inthosewhonamedhim(Dretske1981:66–7,andch.6).Dretskecantelltherestofhisstoryabouthow“C”causessomerelevantMinvirtueofnaturallymeaningAristotle.ArelevantMmayhavebeenhismother’scallinghim“Aristotle,”forexample.ThiswouldmakeAristotle(theindividual)thecontentofthethoughtsymbol“Aristotle.”UninstantiatedpropertiesPeoplecanthinkaboutunicornsandfountainsofyouthandsoon,butnoneofthesethingsexists.Soitisanimportantquestionhowuninstantiatedproperties30mightbethesemanticcontentsofsuchthoughts.Onewayisifsuchcontentsofthoughtsymbolsarecomplexanddecomposeintomeaningfulprimitiveconstitu-ents.So,forexample,thecontentofthoughtsaboutunicornsmaydecomposeintocontentofhorseswithhorns.“Horse,”“horn,”and“possession”maybeprimitivesymbolswithprimitivecontents(andifnot,theymayfurtherdecom-pose).Theseprimitivesymbolsmayhaveinstantiatedpropertiesastheircontents.Thisisastandardstrategyofempiricists,andisfollowedbyDretske(1981).Itisclearthat“unicorn”scannotnaturallymeanorindicateunicorns,ifunicornsdon’texist.Thus,meaningfulsymbolshavingcomplexuninstantiatedpropertiesastheircontentswoulddecomposeintotheirmeaningfulparts(withsimpler,instantiatedpropertiesascontents).Noticethatsuchaviewmustmaintainthattherearenomeaningfulprimitivetermsthathaveuninstantiatedpropertiesastheircontents.Fodor,beingarationalist,hasahardertimewithuninstantiatedpropertiesascontentsofthoughts.Hehasopentohimthestrategyofdecomposition,buthebelievesthatitisatleastpossiblethat“unicorn”isaprimitivethoughtsymbol.Sosupposethat“unicorn”isaprimitive.Onewaytogetanorganismtolocktoapropertyistorubitsnoseininstantiations.Thisisabithardwhentherearenoinstantiationsoftheunicorns→“unicorns”law.OnesuspectsthatitisforreasonslikethisthatFodordroppedHIC.157\nFredAdamsAnotherwayistosuggestthatnon-unicorn-caused“unicorn”sinthisworldasymmetricallydependonunicorn-caused“unicorn”sinclosepossibleworlds(Fodor1991).Ofcoursethisdoesn’ttelluswhatmetrictouseforclosenessofworlds(Cummins1989;Sterelny1990;Loar1991).Worseyet,itdoesn’ttellushowunicornscause“unicorns”intheclosepossibleworlds.Presumablyitisbecausethepropertyisinstantiatedinthosecloseworlds.Ifso,thentheHICconditionseemstobeemployedinthoseworldsandneedstobeputbackintoFodor’stheoryinsomefashion.Whenpressed,Fodor(1991)notesthathecanalwaysretreatandsaythat31“unicorn”isacomplexterm,notprimitive,afterall.Butheisreluctanttodoso.Whatismore,hisreluctancebaitsothers(Wallis1995)intoattemptstoinventprimitivetermsfornomicallyuninstantiableproperties.Supposeagiantant(gant)isanomologicalimpossibilityforbiologicalreasons–itslegswouldcrushunderitsownweightanditscirculationwouldnotallowsufficientheattransfer.ThenWalliswouldcontendthattherearenocloseworldswherethegant→“gant”lawisinstantiated.WereFodorstubbornlytosticktohisstory,hewouldsaythat“gant”lockstothepropertyofbeingagant,becauseintheclosestworldswherethelawsofnaturearedifferentfromoursgantscause“gants”s.Whatevercauses“gant”sinusheredoessoonlybecausegantscause“gant”sthere(andnotviceversa).Howplausiblethisisbecomesthequestion.Ofcourse,Fodorhimselfnotesthathemustusethedecompositionalstrategyforlogicallyimpossiblepropertiessuchasbeingaroundsquare(Fodor1998a).(Let“roundsquare”suggestaprimitiveterm.)Therearenoworldswherearoundsquares→“roundsquare”slawholds.Ofcourse,ifFodorreallyrejectsHICevenwhenappealingtocloseworldsthatgroundasymmetricaldependenciesoflaws,hecouldmaintainthatitisnotthatroundsquaresorgantsorunicornsdocause“roundsquares”or“gants”or“unicorns,”buttheywouldiftheyweretobeinstantiated.However,itishighlydoubtfulthatFodorwouldsaysuchathing.Forthenthemechanismsofmeaningevaporate.Thiswouldbetoresortnotonlytotheuninstantiated,buttotheuninstantiableandthereisnoreasontobelieveinsuchametaphysicsofsemanticmechanisms.Thedisjunctionproblem–againCriticsarguethatsemanticnaturalistsstillhavenotsolvedthedisjunctionprob-lem.Fodor(1990a)alertedustoitoriginallyinresponsetoDretske’sappealtoalearningperiod.Dretskemodifiedhisaccountsothatitwasnotdependentuponalearningperiod,temporallyconstrued.However,thereremainsaresiduallearn-ingelementinDretske’snewaccount(1988,1995)ofindicatorfunctions.ItremainstruethatduringaprocessofwhatDretskecalls“recruitment”someinternalstructureacquiresitsindicatorfunction,andtherebyacquiresitsrep-resentationalcontent.Thisisnottemporallydeterminedanditisnotarbitrary.However,itdoesrequireastructure“C”havingitsindicatorfunctionbecome158\nThoughtsandTheirContentsfixedorset.Aswenotedabove,Dretskethinksthatindicatorfunctionsbecomefixedinwayssimilartothewaysanynaturalfunctionbecomesfixedorset.Themostskepticalcriticsworrythatallfunctionattributionsareindeterminate(Enc,manuscript).Othersworrythatfunctionsarefarlessdeterminatethanisrequiredfordeterminatesemanticcontent.TheaboveisarepresentativesampleoftheobjectionstoDretske’ssolutiontothedisjunctionproblem.ThereisasimilarrangeofattacksuponFodor’ssolu-tion.TheobjectionstoFodor’suseofasymmetricaldependenciesbeganearly(Dennett1987a,1987b;AdamsandAizawa1992).AizawaandIpointedoutthatTwin-EarthexamplesshouldbeaproblemforFodor.SinceAlandhisTwinarephysicallysimilarineveryrelevantway,ifalawwouldapplytoAl,itwouldapplytoTwin-Al.Hence,ifthereisanH2O→“water”lawandanXYZ→“water”32lawtherecanhardlybeanasymmetricaldependencyoflaws.Breakingeitherlawshouldbreaktheother,since,byhypothesisofTwin-Earthcases,AlandTwin-Alcannotdiscriminatewaterfromtwin-water.Asnotedabove,FodormighttrytouseHICtoexplainthatAlinstantiatesthefirstlaw(aboutwater)andTwin-Alinstantiatesthesecondlaw(aboutXYZ)andneitherinstantiatesboth.So,thatiswhyAl’s“waters”meanwaterandTwin-Al’smeantwin-water(Warfield1994).Ultimately,Idon’tthinkthishelps(AdamsandAizawa1994a,1994b)andFodordropsHIC,anyway.HistheorynolongerblockssayingthatAl’s“water”tokenssymmetricallydependonboththewaterlawandthetwin-waterlaw,therebyhavingdisjunctivemeaning.Fodor(1994)seemsnolongerworriedaboutTwin-Earthcases–metaphysicalpossibilitiesaretooremotetobeworrisome.Hemaybecorrectthatmerepossibilitiesaresoremotethattheyare,asifbyanecologicalboundary,screenedoff.Twin-waterisscreenedofffromAl’senvironment(andviceversaforTwin-Al).Thesecasesarenot“relevantalternatives,”touseafamiliartermfromtheepistemologyliterature.Still,asDennett(1987b)andalonglineofothers(Baker1989;Cummins1989;Godfrey-Smith1989;Maloney1990;Sterelny1990;Boghossian1991;Jonesetal.1991;AdamsandAizawa1992,1994a;ManfrediandSummerfield1992;Wallis1994)havepointedout,Twin-Earthmaynotbearelevantalternative,butotherthingsare(ormightbe).Wecanassumethatthereisnothingmetaphysicallyoutreaboutlookalikes.Whatkeeps“X”frommeaningXorX-lookalike?Cummins(1989)picksmiceforhisXsandshrewsforhisX-lookalikes.Itwouldbeeasyforsomeonetoconfusethesetwoanimalsbytheirlooks.Therewillbeamouse→“mouse”law,satisfyingFodor’scondition(1),buttherewillalsobeashrew→“mouse”law.Thequestioniswhetherthesecondlawisasym-metricallydependentuponthefirstlaw.Cumminsconsidersthevariouswaysofexplainingwhythisasymmetryseemsunlikely.ItseemsclearthatforAl,histhoughtsymbol“mouse”mightsymmetricallydependuponmiceorshrews.His33thoughtsymbol“mouse”wouldlocktomouseorshrew.Ofcourse,thereareotherpropertiesthan“mouseylooks”thatmightbeinvolvedingettingAltolocktomice.Theremaybepropertiesthatmicehaveandthatshrewslack,such159\nFredAdamsthatifmicedidn’thavetheirproperties,shrewswouldn’tbeabletopoachuponthemouse→“mouse”law.ThisiswhatFodorneedstoexplainhow“mouse”lockstothepropertyofbeingamouse,forAl.ButitseemsatleastplausiblethatAl’s“mouse”symbolmighthavedisjunctivesemanticcontent,byFodor’sconditions.Cummins’sexamplemaynotbeaproblemforFodor.I’llexplainwhy.Fodorcansurelyacceptthatitispossiblethatone’sidiolect,oritsequivalentinthought,hasdisjunctivemeaning.IfAlreallymistakesmiceforshrews,thisistobeexpected.WhatFodordoesn’twantisthatnotokensof“mouse”meanmouse,byhisconditions.Toavoidthis,hemightappealtoadivisionoflinguisticlabor.Sincethereisadivisionoflaborintheintroductionoftermsintoournaturallanguage,aslongastheexpertscantellmicefromshrews,theEnglishword“mouse”maystillmeanmouse(alone).IfAlacquireshisthoughtsymbol“mouse”fromexpertsandEnglishspeakers,therecanbesemanticborrowing.SemanticborrowingoccurswhenpersonAacquiresatermfrompersonBandA’stermtherebymeanswhatA’stermmeans.IfAlhearsFranktalkabouttheAustralianechidna,butAlhasnotseentheseanimals,Alcanstillthinkaboutechidnas.Hecanwonderwhattheylooklike,whattheyeat,andsoon.Al’sthoughtsymbolforechidnasmayberatherimpoverished,butlocktoechidnasnonetheless.Sothoughtsymbolscanlocktotheirsemanticcontentviacausalchainsgoingthroughotherminds.WemusttakeCumminstobearguingthattherearenoexpertsinthemouse/shrewcase.ThenAl’sthoughtsymbol“mouse”willnotderiveunivocalcontentfromtheEnglishword.Still,Fodorcouldacceptthat“mouse”lockstomouseorshrewforAl.Itevencouldlocktosomethingdisjunctiveforeveryone,ifnoonecantellmicefromshrews.Butsurelythisisnottrue.Tobeaproblemonemustshowthat“mouse”isunivocal,butwouldbedisjunctiveonFodor’sconditions(andnotbecauseofsemanticborrowing).Baker(1989,1991)usescatsforXsandrobot-catsforX-lookalikestoarguethatFodor’stheorygivesthewrongcontentassignment.SheimaginesJerryfirstseeingrobot-cats,laterseeingcats,anddiscoveringstilllaterthathewaswrongaboutcats(thinkingthattheywerenotrobots).Therearebothofthefollowinglaws:robot-cats→“cat”sandcats→“cat”s.WhatisthecontentofJerry’sthoughtsymbol“cat”?Bakerarguesstrenuouslythat“cat”cannotmeancat,forJerry(andIthinksheisright,ifweexcludethepossibilityofsemanticborrowing).Bakeralsoarguesthat“cat”cannothavecatsasitssemanticcontent(heretoo,Iagree).TheasymmetricaldependencyclauseofFodor’sconditions(condition2)isnotsatisfiedforeitherofthesecontents.BakeralsoclaimsthatJerry’s“cat”scannothavethedisjunctivecontentcatorrobot-catbecauseifitdid,Jerrycouldnotlaterdiscoverthathewasmistakenaboutcats.Butitseemstome,andFodor(1991)agrees,thatthisisacaseofdisjunctivecontent.Thereisacatsorrobot-cats→“cat”slawuponwhichallothertokeningsof“cat”sasymmetricallydepend.TherestofFodor’sconditionsareeasilymet,consistentwiththisinterpretation,andBaker’sclaimaboutFodor’sdiscoveringhismistakeaboutcatsisconsistentwiththisinterpretation.Itbecomesasecond-ordermistake.Fodor’slaterdiscoveryis160\nThoughtsandTheirContentsthathisformerthoughtsaboutcatsweremistakenbecausehefindsoutthatthecontentofhisthoughtswasdisjunctive(wherehethoughttheywerenon-34disjunctivelyaboutrobot-cats).SoBaker’sexamplemaynotbeaproblemforFodor,afterall.ManfrediandSummerfield(1992)tryadifferenttack.Theysuggestthatathoughtsymbol“cow”mayremainlockedtocows,evenifthecow→“cow”slawisbroken.TheyaskustoimaginethatJerryhasseenlotsofcowsandacquiredathoughtsymbol“cow.”SupposethatallofFodor’sconditionsaremetandthencowschangetheirappearance(throughevolutionorradiation,say).Theyarguethatthechangemaybreakthecow→“cow”law,butnotchangethecontentofJerry’s“cow”s.Barringsemanticborrowing,aplausiblereplyisthatthecow→“cow”lawhasnotbeenbroken,justmasked.Aslongastheessenceofbeingacowhasnotchanged,thecow→“cow”lawmaymanifestitselfthroughdifferentappearancesovertime.Nodoubttheearliestcowsinhistorylookeddifferentfromwhattheydonow.Thefactthatoneofthoseearlycowsmightnotcausea“cow”inJerrydoesn’tshowthatthecow→“cow”lawisbroken.Whyshoulditiftherewereasuddenchangeintheappearance,ratherthanaslowgradualchange?ThatJerrywouldn’trecognizecowsbytheirappearancewouldnotbeaproblemforFodor’stheory(thoughitmightpresentpracticalproblemsforJerry).This,too,doesn’tseemtopresentaninsurmountableworry.Toomuchmeaning(semanticpromiscuity)AdamsandAizawa(1994a)havearguedthatFodor’stheoryattributesmean-ingtothingsthatitshouldn’t–attributestoomuchmeaning,ifyouwill.Dretske’stheorymayhavethisdifficultyaswell.AninterestingexamplebroughttomyattentionbyColinAllenseemstoapplytoboththeories.Ifsemanticcontentisaseasytocomebyasitappearsinthisexample,itmaybeubiquitousonnaturalizedtheories.Kuduantelopeeatthebarkoftheacaciatree.Consequently,thetreeemitstanninthatthekududon’tlike.Notonlythat,thewindcarriesthisdownwindtoothertreeswhichemittannintoo.Wereahumantodisturbthebarkoftheacaciatree,itwouldemittannintoo.Ifwelettanninmoleculescountassymbols,allofFodor’sconditionsaresatisfied.Kudubites→tannin(condition1).Humandisturbance→tannin(condition3).Law(3)isasymmetr-icallydependentuponlaw(1)(Condition2).Thedependenciesaresynchronic(condition4).ForDretske’stheorytoo,somestructure(C)intheacaciadetectedandnatur-allymeantkudu.Thatstructurealsohadtheabilitytoturnonthetanninproduction(M).Hence,Cbecamelockedtokudu,whenthefunctionofindicat-ingkuduexplainedthetanninproductionintheacacia.C’sindicatorfunctionbecamelockedtokudu(whowouldotherwisehavedecimatedtheacaciaforests).Theeasywayout,Ithink,istorestrictboththeoriestosymbolsinLOT(or35itsproducts).ThereisnothingintheacaciatreethatcomesclosetoLOT.161\nFredAdamsFodororiginally(1990c)hadsucharequirement,butdroppedit,andDretske’stheoryseemstobedesignedforcreaturesthathaveconsciousexperience,beliefs,36desires,andafullcognitiveeconomy.Stillexampleslikethis,ifsuccessful,yieldwhatmaybeasurprisingresulttosome:thatsemanticcontentcanexistoutside37ofminds.ProximalprojectionsBothFodorandDretskefacetheproblemofwhyathoughtsymbolmeansitsdistalcause(cow)andnotamoreproximalcause(retinalprojectionofacow)thatservestomediatebetweenthoughtandreality(Sterelny1990;AntonyandLevine1991).Dretske’s(1981)solutionwastosaythatconstancymechanismsmayoperateandresultin“cow”indicatingcowswithouttherebyindicatingproximalprojectionsofcows.ThisisbecauseDretskebelievestherearemultipleprojections(P1vP2...vPn)andthemostspecificpieceofinformationthat“cow”carriesindigitalizedformwillbethatacowispresent,notthatP1,notthatP2,...notthatPn.Notethat“cow”willstillcarrytheconjunctiveinformationthatF(acowispresent)andP(someorotherproximalprojections–P1vP2...vPn)areoccurring.IthinkDretskeshouldsaywhathenow(1988)saysinreplyto“C”’sindicatingthatthereisapredator(F)andananimal(G)andaphysicalobject(H)present.Namely,if“C”sindicationofFsexplainswhyitcausesrelevantMsthenitsemanticallymeansFs,eventhoughitindicatesFandGandH.Similarly,“cow”mayindicatecowsandP(wherePisthefinitedisjunctiveproperty).Still“cow”maycauserelevantMsbecauseitindicatescows,enablingustoperceivethecowandthinkaboutthecow,nottheproximalprojections.Fodor’ssolutiontotheproblemofproximalprojectionsreliesonhiscondition(3).Allofthecow-caused“cows”arealsoproximalprojectionsofcow-caused“cow”s.Sothereisnorobustcausationof“cow”sbyproximalprojectionsofcows,and“cow”cannotmeanprojectionofcow.ForFodor,itseemsfalsethatthereisnotrobustcausationof“cow”evenifallperceptionsofcowsasymmetricallydependonproximalprojectionsofcows.Forthoughtsofcowscause“cow”sandplausiblythisasymmetricallydependsuponproximalprojectionscausing“cow”s.Ifso,thecontentof“cow”shouldbeproximalprojectionofcows(AdamsandAizawa1997b).Thus,thisstillseemstobeaproblemforFodor’saccount.Swampman38HerewehaveasignificantdifferencebetweenDretskeandFodor(minusHIC).Clearly,onDretske’sview,Swampmanhasnothoughtcontentwhenheinstan-taneouslymaterializes.Noneofhissymbolshasfunctionstoindicate.Nosymbols162\nThoughtsandTheirContentshavesemanticcontents.Ofcourse,whetherSwampmancanacquiresemanticcontentsisupforgrabs,evenonDretske’saccount.ThatwoulddependonwhetherhisinternalneuralstatesarecapableofnaturalmeaningandsustainedcausingofrelevantmovementsMbecauseoftheirnaturalmeaning(i.e.,learn-ing).Ifpossible,thenintimetherewouldbenoreasoninprinciplewhySwampmancouldnotacquiresemanticthoughtcontent.However,Fodor(1994)claimsthatSwampmanhasthoughtsfromtheinstantofhismaterialization.HebelievesthatthesamemeaningmechanismsthatapplytoJerryapplytoSwampJerry.Fodor’sjustificationsformaintainingthisaresix,noneofwhichseemssufficient(AdamsandAizawa1997a).First,itissimpler(“moreaesthetic”)tohaveonemeaningmechanismforall.Iamforunifiedtheories,butFodor’stheoryhasitswarts.Hehastohandledemonstrativesandnamesandlogicaltermsdifferently.WhynotSwampman?Secondly,henotesthatonemaytoken“X”sintheabsenceofXs(implyingtherejectionofsomethinglikeHIC).Butthisistruewhetherthe“X”shavesemanticcontentornot.“Giz”canbetokenedinabsenceofgizs,but“giz”doesn’tmeananything.Thirdly,Fodor’sintuitionsarestrongthatSwampmanhasthoughts.Yes,andEuclid’sintuitionswerestrongthattheparallelpostulatewastrueofalllinesandpointsoffthem.Fourthly,FodorthinksthattheonlyexplanationofwhySwampmansays“Wednesday”whenaskedthecurrentdateisthathethinksitisWednesday.However,asyntacticbutnon-contentful“todayisWednesday”inwhatwouldbe3940hisbeliefboxwouldexplainitaswell.Acurrentthoughtlesscomputerpro-gramwithaspeechmoduledrivenbyitssyntaxprovesthis.Thesyntaxismean-ingfultous,butnottoexistingcomputers(compuJerry,ifyouwill).Fifthly,FodorclaimsthatthebestexplanationofwhyitismoreplausibletosaythatSwampJerry’s“water”tokensmeanH2OandTwin-SwampJerry’smeanXYZisthattheyhavetheserespectivesemanticcontents.IwouldmaintainthatitisasgoodtoexplainthatifSwampJerry’sthoughtshadcontent,they’dhavethecontentofthemostproximatepopulationofbelievers(viz.Earth),whileTwin-SwampJerrywouldhavethecontentofhismostproximatepopulationofbelievers(viz.Twin-Earth).IftheseSwampmenhadthoughts,thesearethethoughtstheywouldhave,buttheyhavenone.Thesecountersare,Ibelieve,justasstrongorstrongerthanFodor’s.MinddependenceShope(1999)objectsthatDretske’saccountemploystheconceptofexplanationandexplanationisamind-dependentactivity.SoDretske’saccountisnotreallyanaturalizedaccount.Thisistrueonlyiftheappealtoexplanationisineliminable.Anaturalizedaccountneedsintensionality(withan“s”),suchthat“C”scauseMsbecauseoftheirindicatingFs(notGs,thoughtheyindicateGs).Thisintensionalityisfullysuppliedbythatoflawsandteleologicalfunctions.Sothemind-dependentactivityofexplainingiseliminable.163\nFredAdamsFunctionsdon’tworklikethatManycriticscomplainthatDretskedoesn’thavetherighttheoryoffunctionsorthatthereisnoconsensusaboutteleologicalfunctions.Dretske(1990)wouldbehappytoabandontheterm“function”ifneedbe.Hisaccountwouldbethesameifanewtermfor“indicatorfunction”weresubstituted.AlsoShope(1999)andGodfrey-Smith(1992),amongothers,arguethatitisnotnecessaryforanorganismtohaveasymbolthatnaturallymeansFstoacquirethefunctionofindicatingFs(asemanticcontentF).Somethinglesswilldo.Perhapsapredatoryanimalwillrecruita“C”thatisbestcorrelatedwithprey.Thesuggestionisthat“C”stillwillmeanprey.IthinktheproperquestioniswhetherShopeorGodfrey-Smith(oranyone)hasawayaroundtheexampleabovewhere“jade”seemssemanticallytomeanjadeiteornephriteforpreciselythereasonsthatitnaturallyindicatesthisdisjunction.SofarasIcansee,ShopeandGodfrey-Smithandothersassertthatthiscanhappen(“F”meansFwithouteverhavingnaturallymeantFs),buttheydon’texplainhowitcanhappen.Untiltheydo,theyhavenotestablishedthatithappens.Theexamplestheygiveareconsistentwiththeanimals’havingdisjunctivecontents,despitetheirclaimstothecontrary.VacuityThefactthatFodorgivesonlysufficientconditionsforcontentinvitestheworrythat,whileingenious,histheoryisvacuous.Itmayapplytonoactualmeaningfulitems,or,ifitdoes,itmayyieldthewrongcontents(Baker1991;Seager1993;AdamsandAizawa1992,1994a).Watermaybecapableofcausing“water”sinJanet,butsomayhallucinogenicdrugs,braintumors,orhighfevers.SinceFodordropsHIC,theabilitiesofeachofthesetocause“water”sinJanetmustasym-metricallydependonthewater→“water”law.Butdothey?Whywouldthey?“Water”sarestructuresinthebrainthatareidentifiableindependentlyofcon-41tent(byFodor’sownconditions).Sowhywouldn’tsomethinginthebrainbecapableofcausingsuchastructure,independentlyofthestructure’scontent?Icantype“Giz”whether“Giz”hasameaningornot.Whycouldn’tmybraindosomethingsimilarwith“water”?Ontheassumptionthatitcan,Fodor’scondi-tionsalonedonotexplainhowJanet’s“water”slocktowater.Janet’sandourthoughtshavecontent,butnotbecauseoftheconditionsofFodor’stheory.Hence,hisconditionsarevacuous.AnaturalwayoutofthisworryistobringbackHIC.Indeed,thisiswhatDretskewoulddo.Hewouldsaythatitmaybepossibleinsomepeoplethatasymbol42“water”istriggeredbysomethingotherthanwater(priorto“water”sacquiringitssemanticcontent).Ifso,“water”doesnothavewaterasitsnaturalmeaninganditcouldnotacquirewaterasitssemanticcontent,forthoseindividualsin164\nThoughtsandTheirContentsthosecontexts.Buttheremayverywellbestableconditionsthatscreenoffthesecausesinpersonsfreeofdrugs,tumors,fevers,XYZ,andsoon.Fodorneedsto43explainwhytheseotherthingsarescreenedoffand,minusHIC,simplyhasnomechanismtodothis(Warfield1994;AdamsandAizawa1994b).Fodorwillhavetosaythatworldswherewatercauses“water”sinJanetarecloserthanworldswherepathologicalcausesdo.Butnoworldcouldbecloserthantheactualworld.AndpeoplejustlikeJanetinallotherrelevantphysicalrespectsseemperfectlycapableofhavingthesekindsofdeviantcausesofthingsinthebrain,inthisworld.SoFodor’stheorymaynotapplytothem(orJanet).Whichcamefirst:meaningorasymmetry?Manyauthorshavedoubtedwhetherasymmetricaldependenciesgeneratemean-ing(Seager1993;Gibson1996;AdamsandAizawa1994a,1994b;Wallis1995).Fodor’sasymmetriesaresupposedtobringmeaningintotheworld,notresultfromit.IfYscause“X”sonlybecauseXsdo,thismustnotbebecauseofanysemanticfactsabout“X”s.Whatsortofmechanismwouldbringaboutsuchsyntacticasymmetricdependencies?Infact,whywouldn’tlotsofthingsbeabletocause“X”sbesidesXs,quiteindependentlyofthefactthatXsdo?Theinstan-tiationof“X”sinthebrainissomesetofneurochemicalevents.Thereshouldbenaturalcausescapableofproducingsucheventsinone’sbrain(andunderavarietyofcircumstances).Whyonearthwouldsteaksbeabletocause“cow”sinusonlybecausecowscan(giventhat“cows”areuninterpretedneuralevents)?Isitbrute?Often,inexplainingtheexistenceofsuchasymmetries,Fodorreliesonthe“experts,”ontheirintentionstouseterms(1990c:115).But,ofcourse,thiswon’tdo.Onecannotappealtomeaningstoexplaintheexistenceofunderivedmeanings.Sowheredotheunderivedasymmetriescomefrom?Mybestguessisthatitgoeslikethis:“cow”meanscow,“steak”meanssteak,weassociatesteakswithcowsandthatiswhysteakscause“cow”sonlybecausecowscause“cow”s.Wewouldn’tassociatesteakswith“cow”sunlessweassociated“cow”swithcowsandsteakswithcows.Thisexplanationoftheasymmetricaldependencyexploitsmeanings–itdoesnotgeneratethem.Unlessthereisabetterexplanationofsuchasymmetricaldependencies,itmaywellbethatFodor’stheoryismisguidedtoattempttorestmeaninguponthem.6.8ConclusionWartsandall,theseareamongthebesttheoriesofthoughtcontentthatwehave.Theyarenottheonlytheories,buttheyexhibitthebasicprojecttonaturalizecontent.Thedifferencesbetweenthesetwotheoriesandothernaturalizedtheories165\nFredAdamsarerelativelyminor.Andthesetheoriesarenotreallytoobad,especiallywhenyou44considerthealternatives–butthatisaprojectforanothertime.Notes1IfDretske(manuscript)isright,itmaybehardertoexplainhowweknowthesethingsthanwepreviouslybelieved.2Inthischapter,Iwillbeadoptingtheviewofanintentionalrealist.Intentionalrealistsmaintainthatthoughtandothermentalstateshavecontentandexplainbehavior(andothermentalstates)invirtueofhavingcontent.Thereareviewsthatmaintainthatattributionsofcontentstothoughtsisamatterofinterpretation,butthathavingcontentisnotamatterthatcoulddoexplanatorywork.3Thereisadisputebetweencontentholistsandcontentatomists.Holistswouldsaythatthoughtscomeinclusters–nomindcouldhavejustone.Atomistsbelievethatthoughtsandmindscouldbepunctate–amindcouldhavejustonethought.Wemaynotbeabletogodeeplyintothisdisputehere,butseeFodorandLepore(1992).4Thisisawayinwhichmeaningisdifferentfrominformation.IsupposeauniformsignalwouldbeabletoindicateorinformthattheEiffelTowerwasinParis.Supposeweprearrangedthataspecificlight’sgoingonwillsignalthattheTowerisinParis.Thenalight’sgoingonwouldbeabletoinformonewhodidnotknowthattheTowerisinParis.Butalight’sgoingonwouldnotbeabletoconstitutethethoughtthattheTowerisinParis.5Anothergoodreason–towhichalmosteveryoneappeals–isthatnon-verbalinfantsandanimalsthink.Ofcourse,therearedissenters(Davidson1982;Carruthers1996).6Harman(1973)andCarruthers(1996)claimthatmostthoughtsareinnaturalpubliclanguages.DissentersincludeFodor(1998b).7FodorandPylyshyn(1988),Fodor,andFodorandMcLaughlininFodor(1998b)andAizawa(1997)fordissentontheefficacyofsomeLOTargumentsforsystematicity.8Atleast,theymustifintentionalrealismistrue.9IhopeitisclearthatwhenItalkaboutmechanisms,Iamabstractingfromthematerialbasisofthoughtinhumans(theparticularstructuresofneuronsorchemistryofneurotransmitters),andevenfromtheparticularpsychophysicalmechanismsofperception.I’mtalkingabouttheinformationalrequirements,notparticularphysicalorpsychologicalimplementationsthatmeetthoserequirements.10Dretskeoriginally(1971)cameupwiththenotionofa“conclusivereason”wherethethingthatwasthereasonR(whichcouldbeSmith’sfingerprintsonthegun)wouldn’tbethecaseunlessp(Smithtouchedthegun).R’sbeingthecasewouldallowonetoknowthatpwastrue.Dretskelater(1981)turnedtoinformationtheorytofindamoreexactspecificationoftherelationbetweenpropertiesnecessarytohaveknowledge(necessarytoknowSmithtouchedthegun).11SeeBarsalou(1999)fortheviewthatperceptualsymbolsintheperceptualsystemarethemselvesusedasthoughtsymbolsorvehicles.Fordissent,seeAdamsandCampbell(1999)andmanyoftheotherpeercommentaries.12Fodor(1990b)makesmuchofthisandeventually(1990a)dubsit“robustness.”13Descartesmaydissent(MeditationIV)aboutwhetheroneisusinghiscognitiveabilitiesproperlywhenthinkingfalsely.166\nThoughtsandTheirContents14Fodornow,aptly,callsthis“locking”toaproperty(orcontent).15Fodor’sconditionsformeaningareinfluxand(subtly)changeacrossthesethreeworks.16Belowwewillconsideranotherincarnationofthetheorythataddsaconditionanddiscusswhyhemayhaveaddedandthendroppedthatcondition.FormoreaboutthisseeAdamsandAizawa(1994b).17Fodorlikestorefertohisviewasan“informational”semantics(1994).18Noristhereaneedforlearning(period)–consistentwithFodor’spenchantfornativism.19Onemightthinkthatitdoesn’tneedblockingbecausesalivationisnotavehicleinthelanguageofthought.ButFodordoesnotrestricthistheorytoitemsinLOT.So,inprinciple,eventhingsoutsidetheheadcanhavemeaning.20Actually,therearestillproblemsaboutwhethertherearedisjunctivelawsoftheform“waterortwin-water”→“water”s(AdamsandAizawa1994a)orwhetherthereisasymmetricaldependenceoflawshere,butIshallignorethosefornow.21Let’sthinkofanecologicalboundaryasakintowhatDretske(1981)callsa“channelcondition.”22Perhapsitgoeswithoutsaying,but,becauseofthis,Dretskeisanempiricist.Thesamecognitivestructures“water”maybeinAlandTwin-Alinnately,butsincetheAlshavedifferenthistories,theirthoughts(via“water”)willacquiredifferentcontents.23WewillreturntothislaterwhenconsideringanobjectionbyShope.24Sticksandstonesmaybreakone’sbones,butshadowsandsoundscannotharmyou.Everygroundsquirrelknowsthis.Sono“C”isrecruitedtobeanindicatorofshadowsorsounds.Predators–thatisaltogetherdifferent.25SeeDretske(1988:150).26It’sokaywithmeiftheintelligentcatisarobottoo,butithasgottobeabletothink.Formypurposes,aDavidsonianswampcatwoulddoaswellforthenon-intelligentcat.27Theremaybethingsintherobot-catthatcausethingsbecausetheengineerswiredituptocausethosethings.Buttherewillnotbestructuresthatcausethingsbecausethestructuresindicatetothecatthatthereisamousepresent,therebycausingbodilymovementsinconjunctionwithbeliefsanddesires.Fortherobot-cathasnobeliefsanddesires,beingunintelligent.28HereiswhereDretske’stheorymayhaveaneasiertimeofit.ForafingerprinttocarryinformationaboutAristotle,onDretske’sinformation-theoreticaccount(1981),theprobabilitythatAristotletouchedanobject,giventhathisfingerprintisonit,mustbeone.Onthefaceofit,thisdoesn’tsaythatAristotleentersintoalaw.NooneknewaboutfingerprintsinAristotle’sday,butthatisn’tthepoint.Hisappear-ancemayhavebeenasindividuatingasafingerprint.Soidentifyingpropertiesofindividualsmayenableonetotrackinformationaboutthatindividual,withouttherebysayingtheindividualentersintolaws.29ThisiswhyperceptsofAristotlemaybequalitativelyidenticaltothoseofTwin-Aristotle,buttheynaturallymeanthatAristotleispresent(notTwin-Aristotle)becauseAristotlecausedthem(notTwin-Aristotle).30Arelatedproblemexists,ofcourse,forvacuousnames(AdamsandStecker1994;EverettandHofweber2000),buttherewon’tbetimetodiscussthesehere.31IthinkitisFodor’shatredofsemanticholismthataccountsforhisavoidanceofthisstrategy(FodorandLepore1992).32Clearlytheremustbeboth,sinceEarthwatercauses“water”sinAlandtwin-watercauses“water”sinTwin-Al.167\nFredAdams33TherearemovesonecouldmakebybringinginHIC,butIwillleavethosetothereader.34Onceagain,barringthepossibilityofsemanticborrowing,Dretske’stheorywouldconcludethesamethingasFodor’sonBaker’sexample–thecontentisdisjunctive.Itmustbe,ifJerryhasnowaytodistinguishcatsfromrobot-catsbytheirappearance.ThenaturalmeaningofmentalstatesfromwhichJerry’s“cat”symbolsderivetheirindicatorfunctionisitselfdisjunctive.35Theneedforthisisobvious.Symbolsinnaturallanguagesexistoutsidethemindandhavemeaning,buttheirmeaningisderivedfrommentalcontent.36Ofcourse,ifonehopedtoidentifyhavingamindwithhavingsemanticcontents,thiswouldbeadisappointingmove.37Idon’tknowwhatFodor’sreactionwouldbetothispossibility,butDretske(per-sonalcommunication)toldmethathefiguredallalongthatstatesoftheearlyvisualsystem(andpossiblyothers)wouldsatisfyhisconditionsforsemanticcontent.Dretskejokedabouttheacaciathat“itsoundslikeaprettyboringmentallife.”Butthisisonlya“mentallife”atall,boringornot,ifoneattemptstoidentifyhavingmindswithhavingsemanticcontent.38WithHICFodorandDretskewouldbothdenythatSwampmanhasthoughts.ForSwampman,byhypothesis,hasnohistoryofinstantiationofrelevantlawsbetweenpropertiesandsymbols.39BystipulationoftheSwampmanthoughtexperiment,JerryandSwampJerryhaveallthesamesyntacticobjectsintheirheads(wheresyntaxsupervenesonpurelyphysicalstates).Butthesyntacticobjectsmaynotbelockedtoproperties.40Notethattherearestillverygoodreasonswhycontentisstillrelevanttotheexplana-tionofbehavior,andwhyonemaynotretreattoapurelysyntactictheory,suchasStich’s(Adamsetal.1990).41Semanticcontentisaproductofasymmetricaldependency,notasourceofit,onthistheory.42Ifthishappensaftersemanticcontentislocked,itisafalse(orotherwiserobust)tokening.43TherearesimilarproblemsforFodor’stheoryevenwithHIC(AdamsandAizawa1994a,1994b).Forexample,with(HIC),ifweshowJanetonlyjadeite,sheinstanti-atesonlythe“jade”→jadeitelaw.Andanythingthatrobustlyorfalselytokens“jade”wouldthusasymmetricallydependuponjadeite’stokening“jade”.Sothetheorywouldsay“jade”lockstojadeitebecauseshedoesn’tinstantiatenephrite→“jade.”Butthisseemstobeaclassiccasewhere“jade”wouldstillhavethecontentjadeiteornephritebecausethereareplentyofbotharoundandJanetcannottellthemapart.SoFodor’stheorywith(HIC)wouldstillbeintrouble.44ThankstoKenAizawaandFredDretskeforconversationsandadvice.ReferencesAdams,F.(1979).“AGoal-StateTheoryofFunctionAttribution.”CanadianJournalofPhilosophy,9:493–518.——(1991).“CausalContents.”InB.McLaughlin(ed.),DretskeandHisCritics.Oxford:BasilBlackwell.168\nThoughtsandTheirContentsAdams,F.andAizawa,K.(1992).“‘X’MeansX:SemanticsFodor-Style.”MindsandMachines,2:175–83.——(1993).“FodorianSemantics,Pathologies,and‘Block’sProblem’.”MindsandMachines,3:97–104.——(1994a).“FodorianSemantics.”InS.StichandT.Warfield(eds.),MentalRepresen-tations.Oxford:BasilBlackwell.——(1994b).“‘X’MeansX:Fodor/WarfieldSemantics.”MindsandMachines,4:215–31.——(1997a).“RockBeatsScissors:HistoricalismFightsBack.”Analysis,57:273–81.——(1997b).“Fodor’sAsymmetricalCausalDependencyandProximalProjections.”TheSouthernJournalofPhilosophy,35:433–7.Adams,F.andCampbell,K.(1999).“ModalityandAbstractConcepts.”BehavioralandBrainSciences,22(4):610.Adams,F.andEnc,B.(1988).“NotQuiteByAccident.”Dialogue,27:287–97.Adams,F.andFuller,G.(1992).“Names,Contents,andCauses.”Mind&Language,7:205–21.Adams,F.andStecker,R.(1994).“VacuousSingularTerms.”Mind&Language,9:387–401.Adams,F.,Drebushenko,D.,Fuller,G.,andStecker,R.(1990).“NarrowContent:Fodor’sFolly.”Mind&Language,5:213–29.Adams,F.,Fuller,G.,andStecker,R.(1993).“ThoughtsWithoutObjects.”Mind&Language,8:90–104.Aizawa,K.(1997).“ExplainingSystematicity.”Mind&Language,12:115–36.Antony,L.andLevine,J.(1991).“TheNomicandtheRobust.”InB.LoewerandG.Rey(eds.),MeaninginMind:FodorandHisCrtics.Oxford:BasilBlackwell.Baker,L.(1989).“OnaCausalTheoryofContent.”PhilosophicalPerspectives,3:165–86.——(1991).“HasContentBeenNaturalized?”InB.LoewerandG.Rey(eds.),MeaninginMind:FodorandHisCritics.Oxford:BasilBlackwell.Barsalou,L.(1999).“PerceptualSymbolSystems.”BehavioralandBrainSciences,22:577–660.Boghossian,P.(1991).“NaturalizingContent.”InB.LoewerandG.Rey(eds.),MeaninginMind:FodorandHisCritics.Oxford:BasilBlackwell.Carruthers,P.(1996).LanguageThoughtandConsciousness:AnEssayinPhilosophicalPsychology.Cambridge:CambridgeUniversityPress.Cummins,R.(1989).MeaningandMentalRepresentation.Cambridge,MA:MIT/Bradford.Davidson,D.(1982).“RationalAnimals.”Dialectica,36:317–27.——(1987).“KnowingOne’sOwnMind.”ProceedingsandAddressesoftheAmericanPhilosophicalAssociation,60:441–58.Dennett,D.(1987a).TheIntentionalStance.Cambridge,MA:MITPress.——(1987b).“ReviewofJ.Fodor’sPsychosemantics.”JournalofPhilosophy,85:384–9.Dretske,F.(1971).“ConclusiveReasons.”AustralasianJournalofPhilosophy,49:1–22.——(1981).KnowledgeandtheFlowofInformation.Cambridge,MA:MIT/BradfordPress.——(1986).“Misrepresentation.”InR.Bogdan(ed.),Belief.Oxford:OxfordUniversityPress.——(1988).ExplainingBehavior:ReasoninaWorldofCauses.Cambridge,MA:MIT/BradfordPress.169\nFredAdams——(1989).“TheNeedtoKnow.”InM.ClayandK.Lehrer(eds.),KnowledgeandSkepticsm.Boulder:WestviewPress.——(1990).“RepliestoReviewers.”PhilosophyandPhenomenologicalResearch,50:819–39.——(1995).NaturalizingtheMind.Cambridge,MA:MIT/BradfordPress.——(manuscript)“HowDoYouKnowYouAreNotAZombie?”Enc,B.(manuscript).“IndeterminacyofFunctionAttributions.”Enc,B.andAdams,F.(1998).“FunctionsandGoal-Directedness.”InC.Allen,M.Bekoff,andG.Lauder(eds.),Nature’sPurposes.Cambridge,MA:MIT/Bradford.Everett,A.andHofweber,T.(2000).EmptyNames,FictionandthePuzzlesofNon-Existence.Stanford:CSLIPublications.Fodor,J.(1975).TheLanguageofThought.NewYork:ThomasCrowell.——(1981).Representations:PhilosophicalEssaysontheFoundationsofCognitiveScience.Cambridge,MA:MIT/Bradford.——(1987).Psychosemantics.Cambridge,MA:MIT/BradfordPress.——(1990a).ATheoryofContentandOtherEssays.Cambridge,MA:MIT/BradfordPress.——(1990b).“InformationandRepresentation.”InP.Hanson(ed.),Information,Lan-guage,andCognition.Vancouver:UniversityofBritishColumbiaPress.——(1990c).“Psychosemanticsor:WhereDoTruthConditionscomefrom?”InW.Lycan(ed.),MindandCognition.Oxford:BasilBlackwell.——(1991).“Replies.”InB.LoewerandG.Rey(eds.),MeaninginMind:FodorandHisCritics.Oxford:BasilBlackwell.——(1994).TheElmandtheExpert:MentaleseanditsSemantics.Cambridge,MA:MIT/BradfordPress.——(1998a).Concepts:WhereCognitiveScienceWentWrong.Oxford:OxfordUniversityPress.——(1998b).InCriticalCondition:PolemicalEssaysonCognitiveScienceandthePhilo-sophyofMind.Cambridge,MA:MIT/BradfordPress.Fodor,J.andLepore,E.(1992).Holism:AShopper’sGuide.Oxford:Blackwell.Fodor,J.andPylyshyn,Z.(1988).“ConnectionismandCognitiveArchitecture:ACriticalAnalysis.”Cognition,28:3–71.Gibson,M.(1996).“AsymmetricDependencies,IdealConditions,andMeaning.”Philo-sophicalPsychology,9:235–59.Godfrey-Smith,P.(1989).“Misinformation.”CanadianJournalofPhilosophy,19:533–50.——(1992).“IndicationandAdaptation.”Synthese,92:283–312.Grice,H.P.(1957).“Meaning.”PhilosophicalReview,66:377–88.Harman,G.(1973).Thought.Princeton:PrincetonUniversityPress.Jacob,Pierre(1997).WhatMindsCanDo.Cambridge:CambridgeUniversityPress.Jones,T.,Mulaire,E.,andStich,S.(1991).“StavingOffCatastrophe:ACriticalNoticeofJerryFodor’sPsychosemantics.”Mind&Language,6:58–82.Loar,B.(1991).“CanWeExplainIntentionality?”InB.LoewerandG.Rey(eds.),MeaninginMind:FodorandHisCritics.Oxford:BasilBlackwell.Maloney,C.(1990).“MentalRepresentation.”PhilosophyofScience,57:445–8.Manfredi,P.andSummerfield,D.(1992).“RobustnessWithoutAsymmetry:AFlawinFodor’sTheoryofContent.”PhilosophicalStudies,66:261–83.170\nThoughtsandTheirContentsPutnam,H.(1975).“TheMeaningof‘Meaning.’”InMind,LanguageandReality.Cambridge:CambridgeUniversityPress.Seager,W.(1993).“Fodor’sTheoryofContent:ProblemsandObjections.”PhilosophyofScience,60:262–77.Shope,R.(1999).TheNatureofMeaningfulness:Representing,PowersandMeaning.Boston:RowmanandLittlefield.Stampe,D.(1977).“TowardsaCausalTheoryofLinguisticRepresentation.”InMidwestStudiesinPhilosophy.Vol.2,ContemporaryPerspectivesinthePhilosophyofLanguage,ed.P.French,T.UehlingJr.,andH.Wettstein.Minneapolis:UniversityofMinnesotaPress.Sterelny,K.(1990).TheRepresentationalTheoryofMind.Oxford:Blackwell.Wallis,C.(1994).“RepresentationandtheImperfectIdeal.”PhilosophyofScience,61:407–28.–—(1995).“AsymmetricalDependence,Representation,andCognitiveScience.”TheSouth-ernJournalofPhilosophy,33:373–401.Warfield,T.(1994).“FodorianSemantics:AReplytoAdamsandAizawa.”MindsandMachines,4:205–14.171\nTheBlackwellGuidetoPhilosophyofMindKennethAizawaEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter7CognitiveArchitecture:TheStructureofCognitiveRepresentationsKennethAizawaAlthoughtheoriesofcognitivearchitectureareconcernedwiththenatureofthebasicstructuresandprocessesinvolvedincognition,philosophicalinterestinthis1areahaslargelyfocusedonthestructureofhypotheticalcognitiverepresentations.Theclassicaltheoryofcognitivearchitecture,forexample,maintainsthat1Thereexistsyntacticallyandsemanticallycombinatorialmentalrepresenta-tions,i.e.,thereisadistinctiontobemadebetweensyntacticallyandsemant-icallyatomicandsyntacticallyandsemanticallymolecularrepresentations.2Eachtokenofamolecularrepresentationliterallycontainsatokenofeachoftherepresentationsofwhichitisconstructed.3Themeaningofamolecularrepresentationisafunctionofthemeaningsofitspartsandthewayinwhichthosepartsareputtogether.4Eachofthesyntacticpartsofamolecularrepresentationhasthesamecontentinwhatevercontextitoccurs.5Thereexistcomputationalmechanismsthataresensitivetothestructureofthementalrepresentations.Onealternativetoclassicismisatomicrepresentationalism(AR).ARmaintainsthatcognitiverepresentationsareoneandallsyntacticallyandsemanticallyatomic,henceitrejects(1)–(5).Anotherrival,withaconsiderablefollowinginsome2quarters,isfunctionalcombinatorialism(FC).FCmaintainsthat,whiletherearecombinatorialrepresentations,theyarenotofthesortpostulatedbyclassicism.Somewhatmorespecifically,FCassertsthatmolecularrepresentationsaremerely(computable)functionsoftheiratoms,hencethattheatomsneednotbeliteralpartsofthemoleculesfromwhichtheyarederived.Ratherthanattempttosurveythewholeofthefieldofcognitivearchitecture,oreventhewholeofthedebatesovercognitiverepresentations,thispaperwillfocusonJerryFodorandZenonPylyshyn’ssystematicityargumentsforclassi-cism.Inroughoutline,theargumentsaresimple.Therearecertainfeaturesof172\nCognitiveArchitecturethought,namely,thesystematicityofinference,thesystematicityofthought,andthecompositionalityofrepresentations,whicharebestexplainedbyclassicism,hencewehavesomedefeasiblereasontobelievethatclassicismistrue.Thissurveyoftheargumentshasanumberofgoals.Inthefirstplace,itaimstoaddressarangeofmisunderstandingsaboutwhatistobeexplainedintheargu-ments.Secondly,itwilldrawattentiontoarelativelyunderappreciatedfeatureofthesystematicityarguments,namely,thatthereissomeprincipleofbetterexplanationatwork.Thirdly,itwillindicatehow,classicistcontentionsnotwith-standing,theusualformulationsofthesystematicityargumentsdonotinfactsupportclassicism.Finally,itwilldrawattentiontoanotherkindofsystematicityargumentsuggestedbyFodorandPylyshyn’scritique.ThisargumenthastheexplanatoryvirtueFodorandPylyshynhaveinmindandshowsastrengthofclassicismlackinginARandFC.Theplanofthischapterwillbetosurveythesystematicityofinference,thesystematicityofcognitiverepresentations,thecompositionalityofrepresentationsarguments,andanewtypeofsystematicityargument.Eachargumentwillbeintroducedviaanexplanandum,alongwithpossibleARandclassicalexplana-tions.Afterthefirstpassthroughthearguments,wewillreturntoseehowthesystematicityargumentsbearonaspecificversionofFC,namely,thehypothesisthatcognitiverepresentationshavethestructureofGödelnumerals.7.1TheSystematicityofInferenceFodorandPylyshynsuggestthatcognitionhasthefollowinggeneralfeature:“inferencesthatareofsimilarlogicaltypeought,prettygenerally,toelicitcorre-spondinglysimilarcognitivecapacities.Youshouldn’t,forexample,findakindofmentallifeinwhichyougetinferencesfromP&Q&RtoPbutdon’tgetinfer-encesfromP&QtoP”(1988:47).Further:Thehedge[“prettygenerally”]ismeanttoexcludecaseswhereinferencesofthesamelogicaltypeneverthelessdifferincomplexityinvirtueof,forexample,thelengthoftheirpremises.Theinferencefrom(AvBvCvDvE)and(¬B&¬C&¬D&¬E)toAisofthesamelogicaltypeastheinferencefromAvBand¬BtoA.Butitwouldn’tbeverysurprising,orveryinteresting,ifthereweremindsthatcouldhandlethesecondinferencebutnotthefirst.(Ibid.:fn.28)Thequestion,then,arises,“Whyisitthatinferentialcapacitiesaresystematic?”Anumberoffeaturesoftheexplanandumbearcomment.Inthefirstplace,the3explananduminvolvesinferencesofthesamelogicaltype.Anormalcognitiveagentthatcanperformoneinstanceof,say,conjunctioneliminationcan,ceterisparibus,performanotherinstance.Anormalcognitiveagentthatcanperformoneinstanceofmodusponenscan,ceterisparibus,performanotherinstance.For173\nKennethAizawasimplicity,inwhatfollowswewillconsideronlyalimitedrangeofsystematicityofconjunctionelimination.Inthesecondplace,theexplananduminvolvescognitivecapacities,orcognitivecompetences,inlogicalinference.TheexplanandumheredoesnotmaintainthatanynormalcognitiveagentthatinfersPfromP&QwillalsoinferPfromP&Q&R.Becausetheexplanandumisconcernedwithcapacitiesforinference,ratherthanactualperformanceininference,theexperimentalliter-atureonhumanperformanceinreasoning–theliteraturethathasdetectedvariouscontenteffects,frequencyeffects,andsoforth–doesnot,asitstands,directly4addressFodorandPylyshyn’sexplanandum.Thirdly,thesystematicityargu-mentsassumeonlyafinitehumancognitivecompetence.Theydonotrelyontheviewthathumancompetenceinvolvesanunboundedrepresentationalcapacity.FodorandPylyshynwrite:[W]eproposetoviewthestatusofproductivityargumentsforClassicalarchitecturesasmoot;we’reabouttopresentadifferentsortofargumentfortheclaimthatmentalrepresentationsneedanarticulatedinternalstructure.Itiscloselyrelatedtotheproductivityargument,butitdoesn’trequiretheidealizationtounboundedcompetence.Itsassumptionsshouldthusbeacceptableeventotheoristswho–likeConnectionists–holdthatthefinitisticcharacterofcognitivecapacitiesisintrinsic5totheirarchitecture.(1988:36–7)Fourthly,itiscrucialtoseethatinforegoingrecoursetotheideaofanun-boundedrepresentationalcapacity,classicistsdonottherebyforgorecoursetothecompetence/performancedistinction.Clearly,classicistsbelievethatactual6humanperformanceinreasoningisafunctionofmanycapacities.Oneoftheseisalogicalinferentialcapacity,butonemustalsoadmitrecognitional,attentional,andmemorycapacities.Indeed,anycompetentexperimentalistwillrecognizethattherearemanyfeaturesofanexperimentalsituation–suchasthoseaffectingmotivation,recognition,attention,andmemory–thatmustbecontrolledinordertodetectacapacityforlogicalinference.Recognizingthismultiplicityoffactorsistheessenceofrecognizingthecompetence/performancedistinction.So,eventhoughFodorandPylyshynproposetorunthesystematicityofinfer-enceargumentwithoutrelyingonthesuppositionthatthereisanunboundedcapacityforinference,theydonottherebyproposetodowithoutthecompet-7ence/performancedistinctionintoto.Somuchfortheexplanandum.Whatabouttheexplanans?Somecriticsofthesystematicityargumentshaveobservedthatitispossibletodevelopsystemsthat8displayvariousformsofsystematicrelations.Fromthis,theyconcludethattherealissueinthesystematicitydebateisoverexactlywhatsortsofsystematicrelationsexistinhumanthoughtandtheextenttowhichagiventheoryofcognitioncangeneratethosesystematicrelations.Whiledata-fitisanimportantfactorinrationalscientifictheorychoice,itisnottheonlyone.Moreimport-antly,itisnottheoneFodorandPylyshyninvokeinthesystematicityarguments.Theissueintheseargumentsisnotoneofmerelyaccommodatingtheavailable174\nCognitiveArchitecturedata,butoneofaccountingforitinacertainimportantsortofway.FodorandPylyshyn’scommentaryonthesystematicityofinferenceargumentbearsthispointoutnicely:AConnectionistcancertainlymodelamentallifeinwhich,ifyoucanreasonfromP&Q&RtoP,thenyoucanalsoreasonfromP&QtoP....ButnoticethataConnectionistcanequallymodelamentallifeinwhichyougetoneoftheseinferencesandnottheother.Inthepresentcase,sincethereisnostructuralrelationbetweentheP&Q&RnodeandtheP&Qnode...there’snoreasonwhyamindthatcontainsthefirstshouldalsocontainthesecond,orviceversa.Analogously,there’snoreasonwhyyoushouldn’tgetmindsthatsimplifythepremiseJohnlovesMaryandBillhatesMarybutnoothers;ormindsthatsimplifypremiseswith1,3,or5conjuncts,butdon’tsimplifypremiseswith2,4,or6conjuncts;or,forthatmatter,mindsthatsimplifypremisesthatwereacquiredonTuesdays...etc.Infact,theConnectionistarchitectureisutterlyindifferentasamongthesepossibilities.(1988:47–8)TheideathatthereismoreatstakeinthesystematicityargumentsthanmerelyfittingthedataisfurthersupportedbyalaterpassagebyFodorandMcLaughlin:Nodoubtitispossiblefor[aConnectionist]towireanetworksothatitsupportsavectorthatrepresentsaRbifandonlyifitsupportsavectorthatrepresentsbRa...Thetroubleisthat,althoughthearchitecturepermitsthis,itequallypermits[aConnectionist]towireanetworksothatitsupportsavectorthatrepresentsaRbifandonlyifitsupportsavectorthatrepresentszSq;or,forthatmatter,ifandonlyifitsupportsavectorthatrepresentsTheLastoftheMohicans.Thearchitecturewouldappeartobeabsolutelyindifferentasamongtheseoptions.(1990:202)Clearly,moreisatstakeinexplainingthesystematicrelationsinthoughtthansimplycoveringthedata.Manycriticshaverespondedtotheforegoingpassages,indicatingweaknesses9inthewayinwhichFodoretal.developthisidea.Whiletherearegenuineweaknessesintheformulation,intheend,Fodoretal.appeartobeontosomethingthatisofscientificimport,somethingthatphilosophersofsciencewoulddowelltoanalyze,andsomethingtowhichcognitivescientistsoughtto10paygreaterattention.Givenspacelimitations,thesecontentionscanbesup-portedonlywithanapparentlyanalogouscasefromthehistoryofscience.IntheOriginofSpecies,CharlesDarwinnotesregularitiesinmorphology,taxonomy,embryology,andbiogeographythathetakestobebetterexplainedbyevolution11thanbyatheoryofdivinecreation.Theideaisthat,althoughbothevolutionandcreationismhaveaccountsoftheseputativeregularities,theevolutionaryaccountdoesnotrelyonarbitraryhypothesesinthewaythecreationistaccountdoes.Oneinstanceinvolvesthebiogeographyofbatrachians:BorySt.VincentlongagoremarkedthatBatrachians(frogs,toads,newts)haveneverbeenfoundonanyofthemanyislandswithwhichthegreatoceansare175\nKennethAizawastudded.Ihavetakenpainstoverifythisassertion,andIhavefounditstrictlytrue.Ihave,however,beenassuredthatafrogexistsonthemountainsofthegreatislandofNewZealand;butIsuspectthatthisexception(iftheinformationbecorrect)maybeexplainedthroughglacialagency.Thisgeneralabsenceoffrogs,toads,andnewtsonsomanyoceanicislandscannotbeaccountedforbytheirphysicalcon-ditions;indeeditseemsthatislandsarepeculiarlywellfittedfortheseanimals;forfrogshavebeenintroducedintoMadeira,theAzores,andMauritius,andhavemultipliedsoastobecomeanuisance.Butastheseanimalsandtheirspawnareknowntobeimmediatelykilledbysea-water,onmyviewwecanseethattherewouldbegreatdifficultyintheirtransportalacrossthesea,andthereforewhy,onthetheoryofcreation,theyshouldnothavebeencreatedthere,itwouldbeverydifficulttoexplain.(1859:393)Accordingtothetheoryofevolution,batrachianformsfirstappearedonthemainland,butbecauseseawaterkillsthem,therebyhinderingtheirmigrationacrossoceans,onefindsthat(almostwithoutexception)therearenobatrachiansonoceanicislands.Accordingtocreationism,Goddistributedlifeontheplanetaccordingtosomeplan.TheproblemisthatitappearsthatGod’splancouldaseasilyhaveplacedbatrachiansonoceanicislandsasnot.TheevidenceforthislatterclaimisthatnaturalistshadalreadyobservedthatitispossibleforhumanstotransportbatrachianstoMadeira,theAzores,andMauritiusandhavethemsurvivequitewell.Creationismandevolutionhavewhatmightbeidentifiedascentralhypothesesandauxiliaryhypotheses.Thecentralhypothesisincreationismis,ofcourse,thatspeciesaretheproductofdivinecreation,wherethecentralhypothesisinevolu-tionis,ofcourse,thatspeciesaretheproductofdescentwithmodification.Thedifferenceintheaccountsthetheoriesofferliesintheirappealstoauxiliaryhypotheses.Theevolutionaryaccountreliesonauxiliaryhypothesesthatarecon-firmedindependentlyoftheexplanatorytaskathand.Theevolutionaryaccountassumesthatthemainlandisolderthanoceanicislands,afactthatisconfirmedbygeologicalobservationsoferosion.Theevolutionaryaccountalsoassumesthatsaltwaterconstitutesamigrationbarriertobatrachians,afacteasilyconfirmedbysimpleexperiments.Bycontrast,creationisthypothesesconcerningGod’splanfordistributinglifeformsarenotindependentlyconfirmable;thenatureofGod’splanincreationwouldseemtobeinaccessibleunlessonehadantecedentlyveri-fiedthatGoddid,infact,separatelycreateorganismsaccordingtoaplan.Inthissense,thecreationistreliesonanarbitraryhypothesis.Withthisroughcharacterizationoftheexplanatorystandardatworkinthesystematicityarguments,wecanconsiderwhatARmighthavetosayaboutthesystematicityofconjunctionelimination.Theatomicrepresentationalistwillpos-tulateasetofsyntacticallyatomicrepresentations{α,β,γ},whereαmeansJohnlovesMaryandBilllovesMaryandAlicelovesMary,βmeansJohnlovesMaryandBilllovesMary,andγmeansJohnlovesMary.176\nCognitiveArchitectureAsystemforinferringthatJohnlovesMaryfromthepremisethatJohnlovesMaryandBilllovesMaryandAlicelovesMaryandfromthepremisethatJohnlovesMaryandBilllovesMarymighthavetheTuring-machine-likeprogram(P1)s0αγs1s0βγs1Thisprogramissuchthat,ifthesystemisinstates0scanninganαorβ,thenitwillprintaγ,andgointostates1.But,suchasystemmightjustaseasilyhavetheprogram(P2)s0αγs1(P1)allowsasystemtoinferPfromP&QandfromP&Q&R,where(P2)onlyallowsasystemtoinferPfromP&Q.So,recallingwhatFodorandPylyshynhadtosayaboutthispossibility,ARcancertainlymodelamentallifeinwhichanagentcanreasonfrombothP&QandP&Q&RtoP,butcanequallymodelamentallifeinwhichyougetoneoftheseinferencesandnottheother.Onecan,ofcourse,addtothecentralARhypothesisconcerningtheexistenceofatomicmentalrepresentationstheauxiliaryhypothesisthattheARsystemhasaprogramlike(P1),ratherthan(P2),butherewehaveanobjectionableauxiliary.ThisauxiliarycannotbeconfirmedindependentlyofthetruthofARjustasthecreationisthypothesisabouttheplanofGodincreationcouldnotbeconfirmedindependentlyofthetruthofcreationism.Aclassicalaccountofthesystematicityofconjunctioneliminationwillbeginwiththesyntacticallyatomicsymbols,{α,β,γ,&,b},whereαmeansJohnlovesMary,βmeansBilllovesMary,γmeansAlicelovesMary,&meansconjunction,12andbisameaninglessblanksymbol.OneTuring-machine-likeprogramthatenablesasystemtoinferitsfirstconjunctregardlessofwhethertherearetwoormoreconjunctionsinthewholeis(P3)s0αRs0s2αbs3s0βRs0s2βbs3s0γRs0s2γbs3s0&bs1s2&bs3s1bRs2s3bRs2Theinstructionsintheleftcolumndirectthesystemtoscanoverthefirstsymbolonthetapeanderasethefirst“&,”whilethoseinthesecondcolumndirectittoerasethenon-blanksymbolsfromtheremainderofthetape.Theevidentprob-lemwiththisapproachisthat,whilethereareclassicalprograms,suchas(P3),thatgiverisetosystematicityofconjunctionelimination,thereareotherpro-gramsmeetingclassicistspecificationsthatdonotgiverisetothesystematicityofconjunctionelimination.Inotherwords,givenclassicism,onecanaseasilyhave177\nKennethAizawaaprogramthatgivesrisetothesystematicityofconjunctioneliminationassomeotherprogramthatdoesnot.Ofcourse,onemightaddsomefurtherauxiliaryhypothesistoclassicism,sayingthattheprogramofthemindislike(P3),ratherthannot,butthisauxiliarywouldappeartobeinaccessibletoconfirmationindependentofthetruthofclassicism.Theclassicistcannobetterindependentlyconfirmthisauxiliarythancouldthecreationistindependentlyconfirmherauxili-aryaboutGod’splanofcreation.Further,theclassicistcannobetterindepend-entlyconfirmthisauxiliarythancouldtheatomicrepresentationalistconfirmherauxiliaryaboutthenatureofthecomputerprogramofthemind.Theupshot,therefore,isthatneitherclassicismnorARhasanadequateexplanationofthesystematicityofinference.7.2TheSystematicityofCognitiveRepresentationsForthisargument,FodorandPylyshynclaimthat,innormalcognitiveagents,theabilitytohavesomethoughtsisintrinsicallyconnectedtotheabilitytohavecertainotherthoughts.Thisintrinsicconnectionmightbespelledoutintermsoftwotypesofpsychological-leveldependenciesamongthoughtcapacities.Ontheonehand,wereanormalcognitiveagenttolackthecapacityforcertainthoughts,13thatagentwouldalsolackthecapacityforcertainotherthoughts.Ontheotherhand,wereanormalcognitiveagenttohavethecapacityforcertainthoughts,thatagentwouldtherebyhavethecapacityforcertainotherthoughts.Itshouldbenotedthatthisfeatureofnormalcognitiveagentsisnotlogicallyorconceptu-allynecessary.Mindscouldbeentirelypunctateinthesensethattheabilitytohavecertainthoughtsmighthavenoconsequencesatallforthepossessionofanyotherthoughts.So,whatneedstobeexplainediswhynormalcognitiveagentshavesystematic,ratherthanpunctate,minds.Asawayoffurtherclarifyingtheputativeexplanandum,considerwheretheclassicistexpectstofindthesedependencies.Foranormalcognitiveagent,oneexpectstofindanintrinsicconnectionbetweenthecapacityforthethoughtthatJohnlovesMaryandthecapacityforthethoughtthatMarylovesJohn.WereanormalcognitiveagenttolackthecapacitytohavethethoughtthatJohnlovesMary,thenthatagentwouldalsolackthecapacitytohavethethoughtthatMarylovesJohn.Thisisoneaspectoftheideaofintrinsicconnectionsamongthoughts;hereisanother.WereanormalcognitiveagenttobeabletothinkthatJohnlovesMary,thenthatcognitiveagentwouldalsobeabletothinkthatMarylovesJohn.Bearinmindthatthethingtobeexplainedinthesystematicityofcognitiverepresentationsargumentistheveryexistenceofsystematicrelationsamongthoughts,notwherethoserelationslie.Bringingoutwherethesedependencieslieismerelyanexpositorymovethatmaylendsomeintuitivecredibilitytotheclaimthatthoughtissystematic.Onceagain,wewanttoknowwhysomethoughtsareconnectedtoothers,ratherthantononeatall.178\nCognitiveArchitectureHow,then,mightARattempttoexplainthesystematicityofthought?ARmightconjecturethat,becauseofthestructureofthecomputerprogramofthemind,thelossofarepresentationcorrespondingtoonethoughtbringswithitalossoftherepresentationcorrespondingtooneormoreadditionalthoughts.Further,theprogramissuchthattheadditionofrepresentationscorrespondingtosomethoughtsbringswithittheadditionofrepresentationscorrespondingtoadditionalthoughts.Sincerepresentationsare(partof)theunderlyingbasisforthoughts,havingtheconnectionsbetweenthevariousrepresentationswouldconstitute(inpart)theconnectionsbetweenthecapacitiesforthecorrespondingthoughts.TheproblemwiththeARaccountofthesystematicityofthoughtisessentiallythesameasthatwiththeARaccountofthesystematicityofinference.Letussuppose,ifonlyforthesakeofargument,thatonecaninfactprogramacom-putertogiverisetodependenciesamongthecapacitiesfortokeningvariousrepresentations.SupposethereisaclassofARcomputerprogramsthatdisplaydependenciesamongrepresentations.Evenifthereexistsuchcomputerpro-grams,itisclearthattherealsoexistcomputerprogramsthatmeettheconditionsofAR,yetdonotgiverisetodependenciesamongrepresentations.OnemightsaythatanARcomputerprogramcanaseasilybesystematic,asnot.Onecan,ofcourse,addtoARsomeauxiliaryhypothesistotheeffectthatthecomputerprogramofthemindissuchastogiverisetothedependenciesamongthoughts,ratherthannot.But,onceagain,thishypothesiswouldappeartobeinaccessibletoindependentconfirmationshortofconfirmingAR.AlthoughweseewhyARdoesnotexplainthesystematicityofthought,wemustalsoconsiderwhetherclassicismcanpassmusterbythesameexplanatorystandard.Supposethatthereisasetofsyntacticallyatomicrepresentations,Γ={John,Jane,Mary,Lisa,loves,hates}.Thereare,ofcourse,computerprogramsthatcombinetheatomsinΓsoastoyieldtheformulasinthesetΓ*1={JohnlovesJohnJohnlovesJaneJohnlovesMaryJohnlovesLisaJanelovesJohnJanelovesJaneJanelovesMaryJanelovesLisaMarylovesJohnMarylovesJaneMarylovesMaryMarylovesLisaLisalovesJohnLisalovesJaneLisalovesMaryLisalovesLisaJohnhatesJohnJohnhatesJaneJohnhatesMaryJohnhatesLisaJanehatesJohnJanehatesJaneJanehatesMaryJanehatesLisaMaryhatesJohnMaryhatesJaneMaryhatesMaryMaryhatesLisaLisahatesJohnLisahatesJaneLisahatesMaryLisahatesLisa}buttherearealsocomputerprogramsthatcombinetheatomsofΓsoastoyieldtheformulasinthesetΓ*2={JohnlovesMary,JanehatesLisa}.Γ*1issystematic,whereΓ*2isnot;therearedependenciesamongtherepresentationsinΓ*1,butnotamongthoseinΓ*2.So,giventhatonehasaclassicalsystemofrepresentation,onecanaseasilyhaveasystematicsetofrepresentations,asnot.Theclassicistwill,thus,wishtoaddsomeauxiliaryhypothesistotheeffectthatΓformsaset179\nKennethAizawalikeΓ*1,ratherthanasetlikeΓ*2.Therefrain,however,isthatwelackindepend-entconfirmationofthisauxiliary.So,again,eventhoughARlacksasatisfactoryaccountofthesystematicityofthought,classicismisinnobettershapeinthisregard.7.3TheCompositionalityofRepresentationsThesystematicityofcognitiverepresentationsisamatterofsomethoughtsbeingdependentonotherthoughts.Thecompositionalityofrepresentationshastodowithanadditionalpropertyofthoughts:possibleoccurrentthoughtsaresemant-icallyrelated.Roughlyspeaking,thoughtspredicatethesamepropertiesandrelationsofthesameobjects.Thus,theprevioussectionindicatedwhereweshouldexpecttofindintrinsicconnectionsamongthoughts,namely,amongthosethataresemanticallyrelated.Now,thissemanticrelatednessisconvertedintoanexplanandum.FodorandPylyshyn(1988:41)suggestthatsystematicityiscloselyrelatedtocompositionalityandthattheymightbestbeviewedastwoaspectsofasinglephenomenon.Bethisasitmay,systematicityandcompositionalityarelogicallydistinctproperties.So,ontheonehand,itislogicallypossiblethatwereone,asamatterofpsychologicalfact,tolosethecapacitytohavethethoughtthatJohnlovesMary,onemighttherebylosethecapacitytohavethethoughtthatAris-totlewasashippingmagnate.Itisalsologicallypossiblethat,asamatterofpsychologicalfact,wereonetohavethecapacitytohavethethoughtsthatJohnlovesMaryandthatMarylovesherself,onewouldalsohavethecapacitytohavethethoughtthatAristotleisashippingmagnate.Thediscoveryofcognitiveagentsthatweresystematic,butnotcompositional,wouldbepuzzlingintheextreme,butsuchadiscoveryisnonethelessapossibility.Ontheotherhand,itisalsopossibletohavethoughtsthatarecontentfullyrelatedwithouttheirbeinginterdependent.OnecouldhavethecapacityforthethoughtsthatJohnlovesMary,thatMarylovesJohn,thatJohnhatesMary,andthatMaryhatesJohnwithoutthelossofoneofthesecapacitiesprecipitatingthelossofanyothers;further,onecouldhavethecapacityforthethoughtsthatJohnlovesMary,thatMarylovesJohn,andthatJohnhatesMary,withouthavingthecapacityforthethoughtthatMaryhatesJohn.Dependenceamongthoughtsdoesnotlogicallyimplycontentfulrelationsamongthethoughts,andcontentfulrelationsamongthoughtsdoesnotlogicallyimplydependenceamongthoughts.HowthenmightARexplaintheputativefactthatthesetofpossibleoccurrentthoughtsforanormalcognitiveagentarecontentfullyrelated?ARwillsaythatthethoughtthatJohnlovesMaryinvolvesasyntacticatomαthatmeansthatJohnlovesMary,andthatthethoughtthatMarylovesJohninvolvesasyntacticatomβthatmeansthatMarylovesJohn.Nowwhiletheremaywellbecomputerprogramsmeetingthisdescription,therearealsoclearlycomputerprogramsthat180\nCognitiveArchitecturedonotmeetthisdescription.Computerscanbeaseasilyprogrammedtobelikethisasnot,henceARalonedoesnotleadtothecompositionalityofthought.Moreover,shouldARaddanauxiliaryhypothesistotheeffectthatthecomputerprogramofthemindissuchthatweretheagentnottobeabletohandleαitwouldnotbeabletohandleβ,thesituationisnotimproved.ThisauxiliaryisofexactlythesortthatadmitsofnoconfirmationindependentofthetruthofAR.Theclassicalaccountofthecompositionalityofrepresentationsinvokesthehypothesisthatthoughtsinvolveasetofsyntacticatomsandsomewayofcom-posingthemintosyntacticmolecules.Yet,therearewaysofbuildingmoleculesandtherearewaysofbuildingmolecules.AsetofsyntacticatomsΣ={John,Jane,loves,hates}canbecombinedtoformthesetofstringsΣ*1={JohnlovesJohnJohnlovesJaneJanelovesJohnJanelovesJaneJohnhatesJohnJohnhatesJaneJanehatesJohnJanehatesJane}oritcanbecombinedtoformthesetofstringsΣ*2={JohnlovesJohn,JanehatesJane}.Thesetofclassicisthypotheseswehaveenumerateddoesnotleadtotherebeingcontentrelationsamongthoughts.Classicismmust,therefore,invokeanauxiliaryhypothesistotheeffectthatΣiscombinedtoformsyntacticitemsinasetlikeΣ*1,ratherthanthoseinasetlikeΣ*2.Butsuchanauxiliaryisnotindependentlyconfirmed,leavingclassicismwithoutabonafideexplanationofthecompositionalityofrepresentations.Again,wefindthatneitherARnorclas-sicismhasanexplanationofthecompositionalityofthought.7.4AnotherSystematicityArgumentThusfarwehavehadthreeillustrationsofthebasicweaknessincurrentattemptstousesystematicityargumentstojustifyhypothesizingaclassicalsystemofcognit-iverepresentations.Allcurrentattemptsrelyonauxiliaryhypothesesthatareinsomesensearbitrary.Thismightsuggestthattheexplanatorystandardbeing14invokedinthesystematicityargumentsisunrealisticallyhigh.Theargumentofthissectionwillshowthatthestandardisnottoohigh.ItwillmaintaintheexplanatorystandardimplicitinFodorandPylyshyn’swork,butinvokeanotherexplananduminanothersystematicityargument.Thisapproachwillnot,ofcourse,showushowtoexplainthesystematicrelationsthatFodorandPylyshynhaveintroduced,butitwillprovideuswithsomedefeasiblereasontobelievethatthereexistsacombinatoriallanguageofthought.Inthelastsection,wenotedthelogicalseparabilityofthesystematicityandcompositionalityofrepresentations.Wecanhaveoneasapsychologicalfactwithouttheother.Here,however,isanotherpsychologicalfact.Ifanormalcognitiveagenthasasystematicmind,thenitalsohasacompositionalmind.Whyisthis?Whyisitthattheinterdependentthoughtsare,inaddition,contentfully181\nKennethAizawarelated?WehavenotedthatneitherARnorclassicismhasanappropriateaccountoftheseindependentregularities,butwhataboutonefollowingupontheother?Classicismappearstohavetherightsortofaccountoftheco-occurrence.Giventheapparatusclassicismneedstoaccountforsystematicity,onehas,withoutfurtherassumption,theapparatusnecessarytoaccountforcompositionality.Thus,aclassicalaccountofsystematicityreliesonthehypothesisthatthereexistsyntactic-allyatomicrepresentationsthatcombinetoformsyntacticallymolecularrepresenta-tionsandthattheseatomicrepresentationssatisfytheprincipleofsemanticcompositionality.Thereasonthatsomethoughtsaredependentonothersisthattheyhavecommonatomicormolecularrepresentations.Thus,thereasonthethoughtthatJohnlovesMaryisdependentonthethoughtthatMarylovesJohnisthattheyshare(a)anatomicrepresentation“John”(whichmeansJohninboththecontextof“–lovesMary”and“Maryloves–,”(b)anatomicrepresentation“Mary”(whichmeansMaryinboththecontextof“Johnloves–”and“–lovesJohn,”(c)anatomicrepresentation“loves”(whichmeanslovinginboththecontextof“John–Mary”and“Mary–John,”and(d)acommongrammaticalstructure.GiventhissortofaccountoftheinterdependenceoftheJohnlovesMarythoughtandtheMarylovesJohnthought,thefactthatthoughtswillbecontent-relatedfollowswithoutadditionalassumption.Thesetofclassicalassumptionsthatareneededinordertoaccountforsystematicityentailcompositionality.Bycontrast,ARhasnosatisfactorymethodforconnectingthesystematicityofthoughtwiththecompositionalityofthought.Thisarisesbecausethecontentofonesyntacticallyatomicrepresentationiscompletelyindependentofthecontentofanyothersyntacticallyatomicrepresentation.GivenwhatARneedsinordertoaccountfortheinterdependenceofthoughts,thereisnoreasonwhythoseinter-dependentthoughtsshouldatthesametimebecontentfullyrelated.EveniftheARtheoristcanmakegoodonthehypothesisthattheprogramofthemindissuchthattwosyntacticitemsαandβ,withtheirrespectivecontents,aredepend-entoneachother,itwouldrequireanauxiliaryhypothesisregardingthespecificsemanticcontentofαandβtohaveitworkoutthatαandβarealsocontentfullyrelated.Suchanadditionalhypothesis,however,wouldbejustthesortofhypo-thesisthatcouldnotbeconfirmedindependentofthehypothesisofanARsystemofmentalrepresentation.Thestrengthofthissortofexplanatoryargumentisborneoutinanexample.Ancientastronomershadobservedthat,asaverygrossapproximation,thesuper-iorplanetsMars,Jupiter,andSaturnmovethroughthefixedstarsfromwesttoeast.Thisverygeneraltendency,however,isperiodicallyinterruptedbyaperiodofretrogrademotionwhichinvolvesthesuperiorplanetsslowingintheirnormaleastwardmotion,stopping,movingforatimeinawestwardretrogrademanner,beforeagainslowing,stopping,andfinallyresuminganormaleastwardmotion.Ptolemaicastronomerswereawareoftheseirregularitiesandwereabletoprovideaqualitativelycorrectmodelofthem.Thebasicideaistohaveasuperiorplanet,suchasMars,orbitingonanepicycle.Thisepicyclethenorbitsattheendofa182\nCognitiveArchitecturedeferent.Bycarefuladjustmentoftherelativesizesandrelativeratesofrotationoftheepicycleanddeferent,itispossibletogenerate,toafirstapproximation,theobservedmotionsofthesuperiorplanets.TheCopernicanaccountofretro-grademotionsisfundamentallydifferent.AccordingtoCopernicans,retrogrademotionsaremerelyapparentmotionsthatarisefromtheEarth’sovertakingasuperiorplanetasbothorbittheSun.WheretheCopernicanaccountprovestobefarsuperiortothePtolemaicaccountisinitsabilitytoaccountforaparticularfeatureofretrogrademotions:theyalwaysoccurwhenthesuperiorplanetstandsinoppositiontothesun.Wheneveraplanetisintheverymiddleofitswestwardretrogrademotion,itisfoundtobeseparatedfromtheSunby180°.Byclevermanipulationoffeaturesoftheepicycleonthedeferentsystem,Ptolemaicastro-nomycouldprovideanaccountofthisfeatureofretrogrademotion,buttheCopernicansystemgeneratedthefurtherfactwithoutanyadditionalhypothesis.SimplygiventheproposednatureofretrogrademotionsontheCopernicansystem,itfollowsofnecessitythatretrogrademotionswilloccuratopposition.ThenecessaryelementsoftheCopernicanaccountofretrogrademotionssufficetoaccountforretrogrademotionsoccurringatopposition.ThePtolemaicac-countdoesn’thavethisstrength.7.5CanFunctionalCombinatorialismExplaintheSystematicRelationsinThought?Tothispoint,wehaveconsideredhowarangeofsystematicityargumentsbearsonclassicismandAR.Oneresponsetothesearguments,however,hasbeentoclaimthatcognitioninvolvesathirdformofrepresentationalism,anon-classicalFC.AccordingtoFC,molecularrepresentationsaremerely(computable)func-tionsoftheiratoms;theatomsneednotbeliteralspatio-temporalpartsofthemoleculesfromwhichtheyarederived.OnewaythisideaisfleshedoutisthroughPaulSmolensky’s(1995)TensorProductTheory.AnotherwayisthroughGödelnumerals.Infact,Gödelnumeralsarefrequentlycitedtoshowtheimpec-15cablescientificstatureoffunctionallycombinatorialrepresentations.FodorandMcLaughlin(1990)andFodor(1996)haveraisedanumberoftechnicalandconceptualproblemswithSmolensky’stheory,ultimatelycarryingthediscussionindirectionswedonothavetimetoexplorehere.This,however,givesusanopeningtoexploreamoreconservativelineofcriticism.Wecanpresstheex-planatorystandardimplicitinFodorandPylyshyn(1988)toshowthatthekindoffunctionallycombinatorialrepresentationsembodiedinGödelnumeralscan-notexplainthesystematicrelationsinthought.SupposewetrytouseGödelnumeralstoexplainhowacognitiveagentcaninferJohnlovesMaryfromJohnlovesMaryandJohnlovesJaneandfromJohnlovesMaryandJohnlovesJaneandJohnlovesAlice.TheGödelnumeralsstorymightbeginwiththefollowingatomicrepresentations183\nKennethAizawa“1”meansJohnlovesMary“2”meansJohnlovesJane“3”meansJohnlovesAlice“4”meansand.Thenatomicrepresentationsthatwillconstituteamolecularrepresentationgiveusasequenceofnnumerals.Thus,torepresentthepropositionthatJohnlovesMaryandJohnlovesJanewewillusethesequence<1,4,2>,whiletorepresentthepropositionthatJohnlovesMaryandJohnlovesJaneandJohnlovesAlicewewillusethesequence<1,4,2,4,3>.Thissequenceofnnumeralsgivesusexponentsforthefirstnprimenumbers,whicharethenmultipliedinordertocompleteourGödelrepresentations.So,wehaveitthat“1”meansJohnlovesMary.42“144”(=1×2×3)meansJohnlovesMaryandJohnlovesJane.4243“30870000”(=1×2×3×5×7)meansJohnlovesMaryandJohnlovesJaneandJohnlovesAlice.Herethesystemofrepresentationisnon-classical,sinceatokenofagivensyn-tacticmolecule,suchas“30870000,”neednotliterallycontainatokenofeachofthesyntacticatomsofwhichitisconstructed,i.e.,tokensof“1,”“2,”“3,”16or“4.”Togetthesystematicityofconjunctioneliminationinthesecases,wesimplyhypothesizethatthereexistsacomputerprogramthatproducesa“1”inresponsetoboth“144”and“30870000.”Theproblemhereiswhatwehavecometoexpect.Speakingloosely,itisjustaseasytoproduceacomputerprogramthatwrites“1”inresponsetoboth“144”and“30870000”asitistoproduceacomputerprogramthatwrites“1”inresponseto“144”butnotto“30870000.”Addingsomehypothesistotheeffectthattheprogramdoesproduce“144”and“30870000”will,however,beunproductive,sincesuchanhypothesiscannotbeconfirmedshortofconfirmingthehypothesisthatthesystemusesGödelnumeralsascognitiverepresentations.Whataboutthesystematicityofthought?SupposewehavethesetofpropositionsJohnlovesJohnJohnlovesMaryJohnlovesJaneMarylovesJohnMarylovesMaryMarylovesJaneJanelovesJohnJanelovesMaryJanelovesJane.WebeginsettingupaGödelnumeralrepresentationofthesepropositionsusingnumeralsfromthefamiliarbasetenArabicsystemandgivingthemthefollowingsemanticinterpretations:“1”meansJohn,“2”meansMary,184\nCognitiveArchitecture“3”meansJane,and“4”meansloving.HerewehaveourGödelnumeralsystem’satomicrepresentations.Wenextasso-ciatewitheachpropositionasequenceofnumerals.Thus,thepropositionJohnlovesMaryisassociatedwiththesequence<1,4,2>andthepropositionthatJohnlovesJaneisassociatedwiththesequence<1,4,3>.Wetakethennumbersrepresentedbythennumeralsinthesequenceandusethemasthepowersofnprimenumbers.TheproductofthesenprimenumbersyieldsanothernumberwhoseArabicdecimalrepresentationwecanthentaketobetherepresentationofourproposition.Thus,wetakethethree-membersequence<1,4,2>(whichisassociatedwithJohnlovesMary)andapplyittothreeprimenumberstogiveus42thenumbertwothousandandtwentyfive(=2×3×5)whichiswritteninArabicnotationas“2025.”Similarly,wetakethethree-membersequence<1,4,3>(whichisassociatedwiththepropositionJohnlovesJane)andusethisinconjunctionwiththreeprimenumbers,sothatJohnlovesJaneisassociatedwith43thenumeral“20250”(=2×3×5).Followingthisarrangement,werepresentoursetofpropositionswiththefollowingnumerals:4810(=2×3×5)representsJohnlovesJohn422025(=2×3×5)representsJohnlovesMary4320250(=2×3×5)representsJohnlovesJane241620(=2×3×5)representsMarylovesJohn2428100(=2×3×5)representsMarylovesMary24340500(=2×3×5)representsMarylovesJane343240(=2×3×5)representsJanelovesJohn34216200(=2×3×5)representsJanelovesMary34381000(=2×3×5)representsJanelovesJane.Inspectingtherepresentationsgeneratedinthisway,weseethatthemutualdependenceofrepresentationsononeanothergivesrisetosystematicrelationsamongthoughts.Thatis,giventhat,say,therepresentationofJohnlovingJaneandtherepresentationofJanelovingJohnbothdependonthecapacityforhavinga“0”intheonesplace,wecanseethattherewillbeadependencybetweenthecapacityforthinkingthatJohnlovesJaneandthecapacityforJanelovesJohn.Fromthepreviousdiscussion,however,weshouldhavelearnedthatthefore-goingonlyshowsthatGödelnumeralscanexhibitadependenceamongthoughts.ItdoesnotshowthatGödelnumeralscanexplaintheinterdependenceofthoughts.Wehavetoconsiderwhetherornotthereissomearbitraryauxiliaryhypothesisintheaccount.Moreover,aswemayhavecometoexpect,thereis.OneassumptionunderlyingthesystemaboveisthattheGödelnumerals(i.e.,theproductsoftheexponeniatedprimes)areexpressedinthefamiliarbasetennotation.Invirtueofthisassumptionandthechoiceofnumeralsfortheatomicrepresentations,it185\nKennethAizawaturnsoutthatsomeoftheGödelnumeralsforthepropositionshavecommonelements,hencethattherearedependenciesamongthemolecularrepresenta-tions.Theassumptionthatthemindusesabasetenrepresentationalsystemfortheproductsoftheprimesis,however,arbitrary.ToputmattersasFodorandPylyshynwould,onecanaseasilyuseabasetenrepresentationalsystemasnotallthewhileremainingwithintheframeworkofaGödelnumeralsystem.Alternat-ively,wemaysaythatthehypothesisthataGödelnumeralsystemofrepresenta-tionisabasetensystemisnotconfirmedindependentlyofthepresentexplanatorychallenge.AnalternativeassumptionisthattheGödelnumeralsoccurin,say,abase100,000systeminwhichnoneofthe100,000atomicsymbolswillhaveanythingsyntacticincommon.So,thesetofatomicnumeralsinthesystemmightbesomethinglike{0,1,2,3,4,5,6,7,8,9,a,A,b,B,...,z,Z,...,♥,♦,♣,♠}.Insuchasystem,noneofthenumeralsforourpropositionswouldhavecommonelements,hencetherewouldbenointerdependenciesamonganyofthenumeralsrepresentingthepropositionsinourset,hencenointerdepend-enceamongthecorrespondingthoughts.So,Gödelnumeralscannotexplainthe17interdependenciesamongthoughts.Next,whatiswrongwiththeGödelnumeralaccountofthecontentrelationsamongpossiblethoughts?Essentially,thesamethingthatwaswrongwiththeclassicalaccount:a“Gödelnumeralsgrammar”canaseasilygenerateacontentfullyrelatedsetofmolecularrepresentationsasnot.Take,again,thesetofatomicrepresentations,{1,2,3,4},where“1”meansJohn,“2”meansMary,“3”meansJane,“4”meansloving,and“5”meanshatingThissetofatomicrepresentationscangeneratetherepresentationswiththecontentsJohnlovesJohnJohnlovesJaneJohnlovesMaryJohnlovesLisaJanelovesJohnJanelovesJaneJanelovesMaryJanelovesLisaMarylovesJohnMarylovesJaneMarylovesMaryMarylovesLisaLisalovesJohnLisalovesJaneLisalovesMaryLisalovesLisaJohnhatesJohnJohnhatesJaneJohnhatesMaryJohnhatesLisaJanehatesJohnJanehatesJaneJanehatesMaryJanehatesLisaMaryhatesJohnMaryhatesJaneMaryhatesMaryMaryhatesLisaLisahatesJohnLisahatesJaneLisahatesMaryLisahatesLisa,aseasilyasitcangeneraterepresentationswiththecontentsJohnlovesMaryJanehatesLisa.186\nCognitiveArchitectureBearinmind,Gödelnumeralsmustallowsomewayofgeneratingdifferentsetsofmolecularrepresentationsonagivensetofatomicrepresentations.Forpur-posesofcognitivetheory,theremustbesomeprincipleorhypothesisthatmakesitthecasethata“Gödelnumeralgrammar”doesnotgeneraterepresentationswithsuch“contents”asJohnJohnJohn,JohnJohnloves,andJohnlovesloves.WhateverhypothesisatheoryofGödelnumeralshastodo,thiswillbethepro-blematichypothesisthatistheundoingofitsaccountofthecontentrelationsinthought.Thishypothesiswillbeoneforwhichtherewillbenoindependentconfirmation.So,likeclassicism,Gödelnumeralscannotexplainthesystematicityofthought.Thisbringsustothelessfamiliarfeatureofsystematicityexaminedabove,theco-occurrenceofsystematicityandsemanticrelatedness.HowdoGödelnumeralsfarehere?Itshouldbeclearthat,whileitispossibletogeneratedependenciesbetweenrepresentationsthataresemanticallyrelated,itisalsopossibletogeneratedependenciesbetweenrepresentationsthatarenotsemanticallyrelated.Therep-resentation“20250”(whichinourexampleaboverepresentedJohnlovesJane)isintrinsicallyconnectedtotherepresentation“3240”(whichinourexampleaboverepresentedJanelovesJohn),butthisconnectionisindependentofthecontentsof“20250”and“3240.”WhatatheoryofGödelnumeralshypothesizesinordertoaccountfortheintrinsicconnectionamongthoughtsdoesnotimplythattheremustbesemanticrelationsamongthoughts,hencedoesnotexplainthecorrelation.7.6ConclusionAsidefromtherelativelyminortaskofclarifyingthearguments,thischapterhashadothermoreimportantobjectives.First,itdrawsgreaterattentiontothefactthatthesystematicityargumentsinvolvesomeprincipleconcerningchoiceamongcompetingexplanations.Thereismuchthatneedstobesaidabouttheprinciple,butatheartitappearstohavesomethingtodowithexplanationshavingtoavoidadhocauxiliaryhypotheses.Secondly,whiledefendingtheclassicistcontentionthattheoriessuchasatomicrepresentationalismandfunctionalcombinatorialismdonotexplainthesystematicityrelationsinthought,thischapterurgesthatevenclassicismfailstoexplainthesystematicrelationsinthought.Thirdly,thechapterpointsoutanotherkindofsystematicityargumentsuggestedbyFodorandPylyshyn’scritique.ThisargumenthastheexplanatoryvirtueFodorandPylyshynhaveinmindandshowsastrengthofclassicismlackinginARandFC.Notes1Alas,atraditionofsemanticeliminativismthatwouldwarmabehaviorist’sheartlivesonintherepresentationaleliminativismofBrooks(1997),andvanGelder(1997).187\nKennethAizawaAnothersubstantialareaofinvestigationthathasconcernedphilosophersisthe“modularityofmind”(cf.Fodor1983;Karmiloff-Smith1992).Anyadequatediscus-sionofthemodularityofmindwould,however,havetobethesubjectofanotherchapter.2Cf.,e.g.,Smolensky(1995),Cummins(1996),andHorganandTienson(1996).3SeeCummins(1996:612),wherethispointappearstobemissed.4BothvanGelderandNiklasson(1994)andCummins(1996)overestimatethesigni-ficanceofthehumanreasoningliteratureforthesystematicityofinferenceargument.5See,aswell,ibid.:37,38,40,wherethelimitationimposedonthesystematicityargumentistoanhypothesisofaboundedcognitivecapacity.6That,afterall,waspartofthepointofthenote5.7BothNiklassonandvanGelder(1994)andCummins(1996)seemtomissthispoint.8Cf.NiklassonandvanGelder(1994),Cummins(1996),HadleyandHayward(1996).9See,forexample,thediscussionsofnecessitatingtheexplanandumandprincipledexplanationsinSmolensky(1995),Cummins(1996),andHadley(1997).10Theseclaims,undefendedhere,aredefendedinAizawa(inpreparation).11Aizawa(1997a,1997b)examineadditionalillustrations.12Hereα,β,andγabbreviatesentences,ratherthanformulaeinfirst-orderlogic.Thisisstillaclassicalaccount,sincecombinatorialrepresentations,structuresensitivity,theprincipleofcompositionality,andsofortharestillinplay.Havingα,β,andγrepresentsentencesmerelysimplifiesthediscussion.13Theforceofthiscounterfactualisnotthat,wereonetoperformabrainlesionthatremovesonethought,atleastoneotherthoughtwouldtherebybelesioned.Suchanexplanandumwouldpresumablybeanimplementationalfact,hencenotthesortoffacttobeexplainedbyapurelypsychological-leveltheory.ThedependenceFodoretal.areaimingformustbeunderstoodasapurelypsychological-leveldependence.14Hadley(1997)offersthisresponsetothewayinwhichwehaveformulatedthesystematicityarguments.15Cf.,e.g.,vanGelder(1990).16Ofcourse,“30870000”doescontainatokenof“3,”whichisoneoftheatomsfromwhichitisderived,butthisisaccidental.17Apointofclarificationisinorderhere.Recallthat,forthesystematicityarguments,wesupposethatonlyafinitestockofthoughtsisinvolved.Notethat,foranyfinitestockofthoughts,therewillbesomebasefortheexpressingtheGödelnumeralssuchthatthebasewillnotleadtodependenciesamongthethoughts.Giventhis,theGödelnumeralsproposalcannotexplaintheinterdependenciesamongthoughts.ReferencesAizawa,K.(1997a).“ExhibitingversusExplainingSystematicity:AReplytoHadleyandHayward.”MindsandMachines,7:39–55.——(1997b).“ExplainingSystematicity.”MindandLanguage,12:115–36.——(inpreparation).TheSystematicityArguments.Brooks,R.(1997).“IntelligencewithoutRepresentation.”InJ.Haugeland(ed.),MindDesignII.Cambridge,MA:MITPress:395–420.Cummins,R.(1996).“Systematicity.”JournalofPhilosophy,93:591–614.188\nCognitiveArchitectureDarwin,C.(1859).TheOriginofSpecies.London:JohnMurray.Fodor,J.(1983).TheModularityofMind.Cambridge,MA:MITPress.——(1996).“ConnectionismandtheProblemofSystematicity(continued):WhySmolensky’sSolutionStillDoesn’tWork.”Cognition,62:109–19.Fodor,J.andMcLaughlin,B.(1990).“ConnectionismandtheProblemofSystematicity:WhySmolensky’sSolutionDoesn’tWork.”Cognition,35:183–204.Fodor,J.andPylyshyn,Z.(1988).“ConnectionismandCognitiveArchitecture:ACriticalAnalysis.”Cognition,28:3–71.Hadley,R.(1997).“ExplainingSystematicity:AReplytoKennethAizawa.”MindsandMachines,7:571–9.Hadley,R.andHayward,M.(1996).“StrongSemanticSystematicityfromHebbianConnectionistLearning.”MindsandMachines,7:1–37.Horgan,T.andTienson,J.(1996).ConnectionismandthePhilosophyofPsychology.Cam-bridge,MA:MITPress.Karmiloff-Smith,A.(1992).BeyondModularity:ADevelopmentalPerspectiveonCognitiveScience.Cambridge,MA:MITPress.Niklasson,L.andvanGelder,T.(1994).“OnBeingSystematicallyConnectionist.”MindandLanguage,9:288–302.Smolensky,P.(1995).“Reply:ConstituentStructureandExplanationinanIntegratedConnectionist/SymbolicCognitiveArchitecture.”InC.MacDonaldandG.MacDonald(eds.),Connectionism:DebatesonPsychologicalExplanation:223–90.vanGelder,T.(1990).“Compositionality:AConnectionistVariationonaClassicalTheme.”CognitiveScience,14:355–84.vanGelder,T.(1997).“DynamicsandCognition.”InJ.Haugeland(ed.),MindDesignII.Cambridge,MA:MITPress:421–50.vanGelder,T.andNiklasson,L.F.(1994).“ClassicismandCognitiveArchitecture.”ProceedingsoftheSixteenthAnnualConferenceoftheCognitiveScienceSociety.Atlanta,GA:905–9.189\nTheBlackwellGuidetoPhilosophyofMindEricMargolisandStephenLaurenceEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter8ConceptsEricMargolisandStephenLaurenceThehumanmindhasaprodigiouscapacityforrepresentation.Wearen’tlimitedtothinkingaboutthehereandnow,justaswearen’tlimitedtothinkingabouttheobjectsandpropertiesthatarerelevanttoourmostimmediateneeds.Instead,wecanthinkaboutthingsthatarefarawayinspaceortime(e.g.,AbrahamLincoln,AlphaCentauri)andthingsthatinvolveconsiderableabstractionfromimmediatesensoryexperience(e.g.,democracy,thenumberpi).Wecaneventhinkaboutthingsthatneverhaveorneverwillexistintheactualworld(e.g.,SantaClaus,unicorns,andphlogiston).Oneofthecentralquestionsinthehistoryofphilosophyhasbeenhowweareabletodothis.Howisitthatweareabletorepresenttheworldtoourselvesinthought?Inansweringthisquestion,philosophersandpsychologistsoftentakeourcapacityforthoughttobegroundedinourconceptualabilities.Thoughtsareseenashavingconstituentsorparts,1namely,concepts.Asaresult,allofscience,literature,andthearts–aswellaseverydaythought–canbeseentostemfromtheastoundingexpressivepowerofthehumanconceptualsystem.Giventhefoundationalrolethatconceptshaveforunderstandingthenatureofcognition,it’snotpossibletoprovideatheoryofconceptswithouttakingsidesonanumberoffundamentalquestionsaboutthemind.Infact,thetheoryofconceptshasbecomeafocalpointfordemarcatingvastlydifferentapproachestothemindandevendifferentworldviews.Forexample,itinteractswithsuchquestionsaswhethertherereallyarethoughtsatallandwhethersemanticprop-2ertiesarerelevanttothestudyofhumanaction.Similarly,itisattherootofthedisagreementaboutwhetherphilosophyisanapriorienterprise.Needlesstosay,wewillnotdiscussallofthesesortsofissueshere.Inordertokeepthediscussionfocusedandmanageableitwillbenecessarytomakecertainassumptionsaboutmattersthatremaincontroversialbothwithinthephilosophyofmindingeneral3andwithinthetheoryofconceptsinparticular.Thetheoryofconceptshasbeenoneofthemostactiveareasofresearchinbothphilosophyandpsychologyinthepast50years,withmanyimportantand190\nConceptslastingresults.Inwhatfollows,wewillsurveyanumberofthemostinfluentialtheorieswithaneyetowardthekeyissuethatdividesthem–theissueofconcep-4tualstructure.Wewillarguethatnoneofthevarioustypesofconceptualstruc-turecurrentlyonofferisentirelysatisfactory.Thishasledustorethinkthenatureofconceptualstructureitselfandtodistinguishseveralcategoricallydifferenttypesofstructure.8.1DefinitionalStructureTheorizingaboutthenatureofconceptshasbeendominatedsinceantiquitybyanaccountknownastheClassicalTheoryofconcepts.Sodominanthasthisaccountbeenthatitwasnotuntilthe1970sthatseriousalternativesfirstbegantobedeveloped.Moreover,thoughthesealternativetheoriesareinsomerespectsradicallydifferentfromtheClassicalaccount,theyarealldeeplyindebtedtoit.Infact,itwouldhardlybeanexaggerationtosaythatallexistingtheoriesofcon-ceptsare,ineffect,reactionstotheClassicalTheoryanditsfailings.Soappreciat-ingthemotivationsfortheClassicalTheoryanditspitfallsisessentialtounderstandingworkonthenatureofconcepts.AccordingtotheClassicalTheory,conceptsarecomplexmentalrepresentations5whosestructuregenerallyencodesaspecificationofnecessaryandsufficientcon-6ditionsfortheirownapplication.Consider,forexample,theconceptBACHELOR.TheideaisthatBACHELORisactuallyacomplexmentalrepresentationwhoseconstituentsareUNMARRIEDandMAN.Somethingfallsunder,orisintheextensionof,BACHELORjustincaseitsatisfieseachoftheseconstituentconcepts.Or,totakeanotherexample,theconceptKNOWLEDGEmightbeanalyzedasJUSTIFIEDTRUEBELIEF.Inthatcase,somethingfallsundertheconceptKNOWLEDGEjustincaseit7isaninstanceofatruebeliefthat’sjustified.Thissimpleandintuitivelyappealingtheoryhasmuchtorecommendit.Agooddealofthepowerandeleganceofthetheoryderivesfromthefactthatitisabletoprovideaccountsofavarietyofkeypsychologicalphenomena,accountsthatseamlesslymeshwiththetreatmentofreferencedeterminationjustsketched.Categorization,forexample,isoneofthemostfundamentalofallprocessesinvolvingconcepts.Mostofourhighercognitiveabilities–nottomentionourownsurvival–dependuponourabilitytoquicklyandreliablydeterminewhichcategoriesdifferentobjectsinourenvironmentbelongto.TheClassicalTheory’saccountofthiscapacityisnaturalandcompelling.Whathappensincategorizingsomethingasabird,forexample,isthatoneaccessesanddecomposesthecon-ceptBIRDandcheckswhetheritsconstituentsapplytotheobjectinquestion.Ifeachdoes,thentheobjectisdeemedabird;ifatleastonedoesn’t,thentheobjectisnot.TheClassicalTheoryoffersanequallypowerfulaccountofconceptlearning.Theprocessofconceptlearningworksinmuchthesamewayascategor-ization,buttheprocessrunsbackwards.Thatis,toacquireaconceptonestarts191\nEricMargolisandStephenLaurenceoutwithitsconstituentsandassemblestheminlightofone’sexperience.Learn-ing,onthisview,isaconstructiveoperation.Onehascertainconceptstobeginwithandbringsthesetogethertoformnovel,complexconcepts.Inshort,theClassicalTheoryoffersanelegantlyunifiedaccountofreferencedetermination,8categorization,andlearning.Asattractiveasitmaybe,theClassicalTheoryhasfewadherentstoday.Thisisbecauseitfacesanumberofextremelychallengingobjections.Intheremainderofthissectionwebrieflyreviewsomeoftheseobjectionstobringoutcertainmotivationsbehindcompetingtheoriesandtohighlightanumberofthemesthatwillberelevantlateron.PerhapsthemostpressingobjectiontotheClassicalTheoryisthesheerlackofuncontroversialexamplesofdefinitions.Thiswouldn’tbesuchaproblemiftheClassicalTheorywerepartofanewresearchprogram.Butthetruthisthatinspiteofmorethantwothousandyearsofintensivesustainedphilosophicalana-lysis,therearefew,ifany,viablecaseswhereaconceptcanbesaidtohavebeendefined.Infact,thefailuresofthisresearchprogramarenotorious.Totakeonewell-knownexample,considerthedefinitionthatwecitedamomentagofortheconceptKNOWLEDGE.TheproposalwasthatKNOWLEDGEcanbeanalyzedasJUSTIFIEDTRUEBELIEF.Asplausibleasthisdefinitionsoundsatfirst,itissubjecttoafamilyofpowerfulcounterexamples,firstnoticedbyEdmundGettier.Thefollow-ingexampleisadaptedfromDancy(1985).HenryisfollowingtheWimbledonmen’ssinglestournament.HeturnsonthetelevisiontowatchthefinalmatchandseesMcEnroetriumphoverConnors.Asaresult,HenrycomestobelievethatMcEnroewonthematchandhehaseveryreasontoinferthatMcEnroeisthisyear’schampion.ButwhatHenrydoesn’tknowisthat,duetoaproblemwiththenetwork’scameras,thegamecan’tbeshownasittakesplaceand,instead,arecord-ingoflastyear’sgameisbeingshown.Still,atthisyear’stournament,McEnroerepeatslastyear’sperformance,beatingConnorsinthefinalmatch.SoHenry’sbeliefthatMcEnroeisthisyear’schampionistrueandjustifiedaswell,butfewpeoplewouldwanttosaythatheknowsthatMcEnroeischampionthisyear.It’snotjustphilosophicallyinterestingconceptsthathaveproblemslikethis.AsWittgensteinfamouslyarguedinhisPhilosophicalInvestigations,ordinaryconceptsdon’tseemtobeanymoredefinablethanphilosophicalones.OneofWittgenstein’smainexamplesistheconceptGAME,forwhichheconsidersanumberofinitiallyplausibledefinitions,eachofwhichendsupbeingsubjecttoadevastatingcounterexample.Evenphilosophy’sstockexample,BACHELOR,isn’tunproblematic.IsthePopeabachelor?Howaboutaself-declaredgaymanwholiveswithhisloverinamonogomouslong-termrelationship?Botharecasesofunmarriedmen,yetneitherseemstobeabachelor.DefendersoftheClassicalTheorycouldrespondthatwhiledefinitionsareindeedhardtocomeby,thisdoesn’tnecessarilymeanthattherearen’tany.Perhapsdefinitionsaretacitandsonoteasilyaccessibletointrospection(see,e.g.,Rey1993;Peacocke1998).Thegeneralfeeling,however,isthatthemostlikelyreasonwhydefinitionsaresohardtofindissimplythattherearen’tany.192\nConceptsAnotherproblemfortheClassicalTheoryisthat,becauseofitscommitmenttodefinitions,itisalsocommittedtoaformoftheanalytic/syntheticdistinction–adistinctionwhich,inthewakeofQuine’sfamouscritique,isthoughtbymanyphilosopherstobedeeplyproblematic.OnestrandofQuine’scriticismcentersaroundhisviewthatconfirmationisholistic.Confirmationinvolvesglobalprop-ertiessuchassimplicity,conservatism,overallcoherence,andthelike.Moreover,sinceconfirmationreliesuponauxiliaryhypotheses,whenatheoreticalclaimisconfrontedbyrecalcitrantdata,onecan’tsayinadvancewhetherit’sthisclaimratherthansomeauxiliaryhypothesisthatneedstobeabandoned.Allofthisseemstoshowthatwedon’thaveaprioriaccesstotruthsthatarewithintherealmofscientificinvestigation.Moreover,wedon’tknowinadvancejusthowfarthereachofscienceis.Whatmaylooklikeaconceptualnecessity(andthere-forelookanalyticandimmunetorevision)mayturnouttobeacasewherepeoplearebeingmisledbytheirownlackoftheoreticalimagination.Notice,however,thatifaconcepthasadefinition,thisdefinitionwillstronglyconstraintheoreticaldevelopmentsinscienceandplaceapriorilimitsonwhatwearecapableofdiscoveringabouttheworld.Forexample,iftheproperanalysisofSTRAIGHTLINEwereSHORTESTDISTANCEBETWEENTWOPOINTS,then,itwouldseem,onecouldn’tdiscoverthatastraightlineisn’talwaystheshortestdistancebe-tweentwopoints.AndiftheproperanalysisofCATwere(SUCHANDSUCHTYPEOF)ANIMAL,thenonecouldn’tdiscoverthatcatsaren’tanimals.Thesesortsofdefini-tionswouldseemtobeaboutasplausibleandunassailableastheycome.Yet,asHilaryPutnam(1962)haspointedout,thesituationisn’tsosimple.Withthediscoverythatspaceisnon-Euclidian,wecannowseethatthefirstdefinitionisactuallywrong.Andwiththehelpofalittlesciencefiction,wecanseethatitatleastseemspossibletodiscoverthatthesecondiswrongtoo.(PerhapscatsareactuallyMartian-controlledrobots,andnotanimalsatall.)ButifSTRAIGHTLINEandCAThadthedefinitionsthattheClassicalTheorysuggests,thenthesediscov-erieswouldbeentirelyprohibited;theywouldn’tbepossibleatall.ExampleslikethesethreatentheveryfoundationsoftheClassicalTheory.Adefinitionmayappeartocapturethestructureofaconcept,buttheappearancemayonlybean9illusionwhichlaterdiscoverieshelpustoseebeyond.Relatedtocasessuchasthese,onefindsotherconsiderationsthatargueagainstdefinitions–inparticular,SaulKripke’sandHilaryPutnam’sinfluentialworkonthesemanticsofnamesandnaturalkindterms(seeesp.Kripke1972/1980;Putnam1970,1975).Kripke’sandPutnam’stargetwasthedescriptiontheoryofreference,accordingtowhichsomeoneisabletouseanameorkindtermbyvirtueofknowingadescriptionthatpicksoutitsreference.Notice,however,thattheClassicalTheoryjustisaformofthedescriptiontheory,onlyitholdsatthelevelofconceptnotwords.Forthisreason,allofKripke’sandPutnam’sargu-mentsarepertinenttoitsevaluation.OneoftheirargumentsisanelaborationoftheQuineanpointthatwecanmakediscoveriesaboutakindthatrevealthatwewerewrongaboutitsnature–theproblemoferror.Closelyrelatedistheprob-lemofignorance:ifpeoplearesometimeswrongaboutcertainpropertiesofa193\nEricMargolisandStephenLaurence10kind,theyarealsooftenignorantofthefeaturesthatreallyareessentialtoit.Whatturnsouttobecrucialtotheidentityofgoldisitsatomicnumber,andnot,forexample,itscolor,orweight.Similarly,thecrucialfeatureofthebubonicplagueisitsbacterialsource,andnotthechills,fever,ornauseathatitisassoci-atedwith,andcertainlynotaconnectionwithsinfuldeeds(inspiteofthewidespreadbeliefthattheplaguewasaformofdivineretribution).Whatbearsemphasizinghereisthatsuchignorancedoesn’tpreventpeoplefrompossessingtheconceptGOLDorPLAGUE.Ifitdid,peoplewouldn’tbeablegenuinelytodisagreewithoneanotheraboutthecauseoftheplague;they’dalwaysenduptalkingatcrosspurposes.ThephilosophicalconsiderationsweighingagainsttheClassicalTheoryareimpressive.Butitsworriesdon’tendthere.TheClassicalTheoryalsofacesanumberofdauntingproblemsbasedonpsychologicalconsiderations.Perhapsthemostglaringoftheseisthatdefinitionshavefailedtoshowupinexperimentalsituationsthatareexplicitlydesignedtotestforthepsychologicalcomplexityofconcepts(see,e.g.,Kintsch1974;J.D.Fodoretal.1975;J.A.Fodoretal.1980).If,forexample,CONVINCEisanalyzedasCAUSETOBELIEVE(followingstandardClassicaltreatments),onewouldexpectthatCONVINCEwouldimposeagreaterprocessingburdenthanBELIEVE;afterall,CONVINCEissupposedtohaveBELIEVEasaconstituent.Yetthissortofeffecthasneverbeendemon-stratedinthelaboratory.Notonlydodefinitionsfailtorevealthemselvesinprocessingstudies,thereisalsonoevidenceoftheminlexicalacquisitioneither(Carey1982).Ofcourseitisalwayspossiblethattheseexperimentsaren’tsubtleenoughorthatthereissomeotherexplanationofwhydefinitionsfailtohavedetectablepsychologicaleffects.Butitcertainlydoesn’thelptheClassicalThe-ory’scasethatdefinitionsrefusetorevealthemselvesexperimentally.ThemostpowerfulpsychologicalargumentsagainsttheClassicalTheory,how-ever,arebaseduponso-calledtypicalityeffects.Typicalityeffectsareavarietyofpsychologicalphenomenaconnectedtothefactthatpeoplewillinglyratesubcategoriesforhowtypicalorrepresentativetheyareforagivencategory.Forexample,subjectstendtosaythatrobinsarebetterexamplesofthecategorybirdthanchickensare;i.e.,theysayrobinsaremore“typical”ofbird.Inandofitself,thisresultmaynotbeterriblyinteresting.Whatmakestypicalityjudgmentsimportantisthefactthattheytrackavarietyofothersignificantpsychologicalvariables(forreviews,seeRosch1978;SmithandMedin1981;foramorecriticalreview,seeBarsalou1987).EleanorRoschandCarolynMervis(1975)foundthatwhensubjectsareaskedtolistpropertiesthatareassociatedwithagivencategoryanditssubordinates,thedistributionofpropertiesontheselistsispredictedbyindependenttypicalityrankings.Themoretypicalasubordinateisjudgedtobe,themorepropertiesitwillsharewithotherexemplarsofthesamecategory.Forinstance,robinsaretakentohavemanyofthesamepropertiesasotherbirds,and,correspondingly,robinsarejudgedtobehighlytypicalbirds;incontrast,chickensaretakentohavefewerpropertiesincommonwithotherbirds,andchickensarejudgedtobe194\nConceptslesstypicalbirds.Anotherfindingisthattypicalityhasadirectreflectionincategorization.IncaseswheresubjectsareaskedtojudgewhetheranXisaY,independentmeasuresoftypicalitypredictthespeedofcorrectaffirmatives.Sub-jectsarequickerintheircorrectresponseto“Isarobinabird?”thanto“Isachickenabird?”Errorrates,aswell,arepredictedbytypicality.Themoretypicaltheprobe(X)relativetothetargetcategory(Y),thefewertheerrors.Typicalityalsocorrelateswithlexicalacquisitionandavarietyofotherphenomena,suchastheorderinwhichsubjectswillprovideexemplarsforagivencategory–moretypicalitemsarecitedfirst.Insum,typicalityeffectsseemtopermeateeveryaspectofaconcept’slife,significantlydeterminingitsacquisition,use,andevenmisuse.It’snowonderthatpsychologistshaverequiredthatatheoryofconceptsdojusticetothesedata.It’sinthiscontextthatmostpsychologistshavegivenupontheClassicalTheory.TheproblemisthattheClassicalTheorysimplyhasnothingtosayaboutanyofthesephenomena.Theclassicalmodelsofcategorizationandconceptacquisitionthatwesketchedabovedon’tpredictanyoftheeffects,andclassicalattemptstoaccommodatethemappearadhocandquicklyrunintofurtherproblems.Moreover,aswe’llseeinthenextsection,therearealternativetheoriesofconceptsthatpro-videnaturalandhighlyexplanatoryaccountsofthefullrangeoftypicalityeffects.TheClassicalTheoryfacesabatteryofpowerfulphilosophicalandpsycholo-gicalobjections.Definitionsareveryhardtocomeby,theydon’thaveanypsycho-logicaleffects,theycan’texplainanyofthemostsignificantpsychologicalfactsthatareknownaboutconcepts,theyflyinthefaceofQuine’scritiqueoftheanalytic–syntheticdistinction,andtheyaren’tequippedtoexplainhowtherefer-enceofaconceptisdetermined.Asaresult,it’shardtoresistthethoughtthat,inspiteofitsconsiderableattractions,theClassicalTheoryisn’tworthsaving.8.2ProbabilisticStructureThe1970ssawthedevelopmentofanewtheoryofconcepts,onethatgainedconsiderablesupportasanalternativetotheClassicalTheory.Thisnewtheory–thePrototypeTheory–gaveupontheideathataconcept’sinternalstructure11providesadefinitionoftheconcept.Instead,thePrototypeTheoryadoptedaprobabilistictreatmentofconceptualstructure.AccordingtothePrototypeTheory,mostlexicalconceptsarecomplexmentalrepresentationswhosestructureen-codesnotdefiningnecessaryandsufficientconditions,but,rather,conditionsthatitemsintheirextensiontendtohave.SoincontrastwiththeClassicalTheory,foranobjecttobeintheextensionofaconcept,itneedn’tsatisfyeachandeverypropertyencodedintheconcept’sstructureaslongasitsatisfiesasufficientnumberofthem.Notice,rightoff,thatoneoftheadvantagesofthePrototypeTheoryisthatitdoesn’trequirethatconceptshavedefinitions.It’snoproblemforthePrototype195\nEricMargolisandStephenLaurenceTheorythatpeoplehavehadsomuchdifficultyformulatingthem.AccordingtothePrototypeTheory,concepts,byandlarge,lackdefinitionalstructure;theyhaveprototypestructureinstead.Forthisreason,italsoshouldn’tbeasurprisethatdefinitionsnevershowupinstudiesofpsychologicalprocessing.Infact,it’swhenweturntotheempiricalpsychologicaldatathatPrototypeTheorybecomesespeciallyappealing.Thewaythetheoryisgenerallyunderstood,ittakescat-egorizationtobeafeature-matchingprocesswhereanexemplarorindividualiscomparedtoatargetcategoryforhowsimilartheyare.Solongasenoughfeaturesmatch,theyaredeemedsufficientlysimilarandonecomestojudgethattheitemfallsunderthecategory.Thisrelianceonsimilarityprovidesthere-sourcesforanextremelynaturalexplanationofthetypicalityphenomena(see,e.g.,Smith1995).Oneneedonlyassumethattypicalityjudgmentsarealsoformedbytheverysameprocess.Inotherwords,thereasonwhyrobinsarejudgedtobemoretypicalbirdsthanchickensisbecauseROBINsharesmorefeatureswithBIRD;itrankshigherinthesimilarity-comparisonprocess.ConsideralsothefindingbyRoschandMervis,thattypicalityjudgmentstrackthenumberoffeaturesthataconceptshareswithotherexemplarsforasuperordinatecategory.Again,thePrototypeTheoryhasanaturalexplanationofwhythishappens.Thereasonisbecausethepropertiesthatsubjectslistthatarecommonamongthesubordinatecategoriescorrespondtothefeaturesofthesuperordinateconcept;thatis,theycharacterizethestructureofthesuperordinateconcept.Asaresult,conceptsthatsharemanyfeatureswiththeirfellowsubordin-ateswillautomaticallysharemanyfeatureswiththesuperordinate.StickingtotheexampleoftheconceptBIRD,theideaisthatthepropertiesthatarecom-monlycitedacrosscategoriessuchasrobin,sparrow,ostrich,hawk,andsoon,aretheverypropertiesthatareencodedbythestructureofBIRD.SinceROBINhasmanyofthesamestructuralelements,andCHICKENhasfew,robinswillbejudgedtobemoretypicalbirdsthanchickensare.Inshort,thePrototypeTheoryhastremendouspsychologicaladvantages.It’snowonderthatthepsychologicalcommunityembracedthetheoryasanalternat-ivetotheClassicalTheory.ButthePrototypeTheoryisn’twithoutitsdifficultieseither,andafullappreciationofsomeofthesedifficultiesisessentialtoarrivingatasatisfactorytheoryofconcepts.Tokeepthingsbrief,we’llmentiononlythree.ThefirstproblemisthatthePrototypeTheoryissubjecttotheproblemsofignoranceanderror,justliketheClassicalTheory.Onceagain,theproblemisthatpeoplecanpossessaconceptandyethaveerroneousinformationabouttheitemsinitsextensionorlackasufficientamountofcorrectinformationtopickthemoutuniquely.Moreover,prototypesarenotoriouslybadindealingwiththequestionofreferencedetermination.Take,forexample,theconceptGRANDMOTHER.Prototypicalgrandmothersarewomenwithgrayhair,theyhavewrinkledskin,theywearglasses,andsoon.Yetweallknowthattherearepeoplewhofailtoexhibitthesecharacteristicswhoaregrandmothers,andthattherearepeoplewhodoexhibitthesecharacteristicswhoarenot.Mrs.Doubtfire(theRobinWilliamscharacter)maylooklikeagrandmother,butTinaTurnerreallyisagrandmother.196\nConceptsThesecondproblemisthatmanyconceptssimplylackprototypes.Thisisespeciallyclearinthecaseofcertaincomplexconcepts.AsJerryFodorputsit:“[T]heremaybeprototypicalgrandmothers(MaryWorth)andtheremaybeprototypicalpropertiesofgrandmothers(good,oldMaryWorth).Buttherearesurelynoprototypicalpropertiesof,sayChaucer’sgrandmothers,andtherearenoprototypicalpropertiesofgrandmothersmostofwhosegrandchildrenaremarriedtodentists”(1981:297;seealsoFodor1998).Thethirdproblemisthatprototypesdon’tappeartocomposeinaccordancewiththeprinciplesofacompositionalsemantics(seeFodor1998;FodorandLepore1996).Thedifficultyisthat,onthestandardaccountofhowtheconceptualsystemisproductive(i.e.,ofhowwearecapableofentertaininganunboundednumberofconcepts),conceptsmusthaveacompositionalsemantics.Fodorillus-tratestheargumentwiththeconceptPETFISH.ThePETprototypeencodespropertiesthatareassociatedwithdogsandcats,andtheFISHprototypeencodespropertiesthatareassociatedwiththingsliketrout,yetthePETFISHprototypeencodespropertiesthatareassociatedwithgoldfishandothersmallcolorfulfish.Soit’shardtoseehowtheprototypeforPETFISHcouldbecomputedfromtheproto-typesforPETandFISH.Together,thesethreecriticismsposeaseriousthreattothePrototypeTheory.However,prototypetheoristsdostillhavesomeroomtomaneuver.Whatallthreeobjectionspresupposeisthatprototypetheoristsmustholdthataconcept’sstructureisexhaustedbyitsprototype.Butprototypetheoristscouldsimplyabandonthisconstraint.Theycouldmaintain,instead,thataconcept’sprototypeisacrucialpartofitsstructure,butthatthereismoretoaconceptthanitsprototype.Infact,anumberofprototypetheoristshavesuggestedtheoriesalongjusttheselinesinordertodealwiththefirstofourthreecriticisms,viz.,theproblemthatprototypesaren’tsuitedtodeterminingreference.AccordingtothisDualTheory,aconcepthastwotypesofstructure,onetypeconstitutestheconcept’s“core”andthesecondits“identificationprocedure”(OshersonandSmith1981;Smithetal.1984;Landau1982).Prototypesaresupposedtobeconfinedtoidentificationprocedures.Theyaccountforquickcategorizationprocessesaswellasallofthetypicalityeffects.Ontheotherhand,coresaresupposedtohavesomeothertypeofstructurethataccountsforreferencedeterminationandisrespons-ibleforourmostconsideredcategorizationjudgments–thedefaultviewbeing12thatcoresexhibitclassicalstructure.TheDualTheoryhandlesthefirstobjectionbyitscommitmenttoconceptualcores.Theideaisthatit’sperfectlyfineifprototypescan’tdeterminereference,sincebyhypothesiscoresfulfilthatrole.Ithandlesthesecondobjectionbyaddingthatsomeconceptslackprototypesbutthatthisdoesn’tprohibitanyonefrompossessingtheconcepts;theyneedonlygraspthecoresoftheseconcepts.Finally,ithandlesthethirdobjectionbymaintainingthattheproductivityoftheconceptualsystemisestablishedsolongasconceptualcorescombineinaccord-ancewithacompositionalsemantics,andthatexamplessuchasPETFISHdon’ttellagainstthispossibility.197\nEricMargolisandStephenLaurenceThoughnoneoftheseresponsesiswithoutmerit,noticethattheyworkbyinsulatingprototypestructurefrommanyofthetheoreticalrolesforwhichcon-ceptualstructureisintroducedinthefirstplace.Asaresult,theDualTheoryplacesagreatdealofweightontheconceptualstructureassociatedwithacon-cept’score.Totheextentthatthisotherstructureissupposedtobeclassicalstructure,theDualTheoryinheritsmostoftheproblemsthatwereassociatedwiththeClassicalTheory.Forexample,theDualTheoryfacestheproblemofignoranceanderror,ithastoovercomeQuineanobjectionstotheanalytic–syntheticdistinction,ithastoconfrontthedifficultythattherearefewexamplesoftruedefinitions,andsoon.Inshort,theDualTheorymayexpandthelogicalspacesomewhat,but,withoutanadequateaccountofconceptualcores,itisn’tmuchofanimprovementoneithertheClassicalTheoryorthePrototypeTheory.8.3TheoryStructureTheDualTheorycontinuestoenjoywidespreadsupportinspiteofthesedifficult-ies.WesuspectthatthisisbecauseofthefeelingthatpsychologyhasfoundawaytoabandonitsresidualtiestotheClassicalTheory.Theideaisthatconcep-tualcoresshouldbeunderstoodintermsoftheTheoryTheory(see,e.g.,Keil1994).Thisistheviewthatconceptsareembeddedinmentalstructuresthatareinimportantwayslikescientifictheoriesandthattheyapplytothethingsthatsatisfythedescriptivecontentgivenbytherolesthattheyhavewithintheirrespectivementaltheories(see,e.g.,Carey1985;MurphyandMedin1985;13GopnikandMeltzoff1997).Foramentalstructuretobetheory-like,itmustembodyanexplanatoryschema,thatis,asetofprinciplesorrulesthatathinkerusesintryingtomakesenseofaneventinthecourseofcategorizingit.Examplesofsuchtheoriesincludeso-calledcommon-sensepsychology,common-sensephysics,andcommon-sensebiology–thesetsofprinciplesthatordinarypeople14useinexplainingpsychological,physical,andbiologicalevents.OneofthemainadvantagesoftheTheoryTheoryisthemodelofcategoriza-tionthatitencourages.Manypsychologistshaveexpresseddissatisfactionwithearliertheoriesofconceptsonthegroundsthattheyfailtoincorporatepeople’stendencytowardessentialistthinking–aviewthatMedinandOrtony(1989)havedubbedpsychologicalessentialism.Accordingtopsychologicalessentialism,peopleareapttoviewcategorymembershipforsomekindsasbeinglessamatterofaninstance’sexhibitingcertainobservablepropertiesthantheitem’shavinganappropriateinternalstructureorsomeother“hidden”property(including,per-haps,relationalandhistoricalproperties).TheTheoryTheoryreadilyaccommod-atespsychologicalessentialismsincetheTheoryTheorytakespeopletoappealtoamentallyrepresentedtheoryinmakingcertaincategorydecisions.Ratherthanpassingquicklyoveracheck-listofproperties,peopleaskwhethertheitemhastherighthiddenproperty.Thisisn’ttosaythattheTheoryTheoryrequires198\nConceptsthatpeoplehaveadetailedunderstandingofgeneticsandchemistry.Theyneedn’tevenhaveclearlydevelopedviewsaboutthespecificnatureoftheproperty.AsMedinandOrtonyputit,peoplemayhavelittlemorethanan“essenceplaceholder”(1989:184).Thissuggeststhatdifferentpeoplerepresentdifferentsortsofinformationinthinkingofakindashavinganessence.Insomecasestheymayhavedetailedviewsabouttheessence.Inmost,theywillhaveaschematicview,forinstance,thebeliefthatgeneticmakeupiswhatmatters,eveniftheydon’trepresentparticulargeneticpropertiesandknowverylittleaboutgeneticsingeneral.TheTheoryTheoryisbestsuitedtoexplainingourconsideredactsofcategor-ization.Whatmattersinsuchcasesisnotsomuchanobject’sgrossperceptualproperties,but,rather,thepropertiesthataretakentobeessentialtoitsnature.Atthesametime,theTheoryTheoryisnotterriblywellsuitedtoexplainingourmorerapidcategorizationjudgmentswhereconceptsaredeployedunderpres-suresoftimeandresources.Andingeneral,theTheoryTheorymakeslittlecon-tactwithtypicalityeffects;liketheClassicalTheory,ithasnothingtosayaboutwhysomeexemplarsseemmoretypicalthanothersandwhytypicalitycorrelateswithsomanyothervariables.Ontheotherhand,iftheTheoryTheorywerecombinedwithPrototypeTheory,theresultingversionoftheDualTheorywouldseemtohaveconsiderablepromise.Coreswiththeorystructurewouldseemtobeavastimprovementoncoreswithclassicalstructure.Unfortunately,thisrevisedDualTheorystillfacesanumberofseriousdifficulties.WewillmentiontwothatarespecificallyassociatedwiththeTheoryTheoryasanaccountofconceptualcores.Thefirstproblemisonethathasalreadycroppedup,soitshouldn’tbemuchofasurprise(theproblemofreferencedetermination);theotherproblemisnew(theproblemofstability).TheproblemofreferencedeterminationaffectstheTheoryTheoryinseveralways.Foronething,we’veseenthattheorytheoriststypicallyallowthatpeoplecanhaverathersketchytheories,wheretheessenceplaceholderforaconceptincludesrelativelylittleinformation.Notice,however,thattotheextentthatthisistrue,conceptswillmostlikelyencodeinadequateinformationtopickoutacorrectanddeterminateextension.Ifpeopledon’trepresentanessenceforcatsordogsapartfromsomethinideasaboutgeneticendowment,thentheconceptsCATandDOGwillbeembeddedintheoriesthatlookaboutthesame.Dependingonhowanemicthetheoriesare,theremaythenbenothingtopullaparttheirconceptsCATandDOG.Ontheotherhand,peoplemayhavedetailedenoughtheoriestodifferentiateanynumberofconcepts,yetthiscomeswiththedangerthattheymayhaveincorporatedincorrectinformationintotheirtheories.Toreturntoourearlierexample,someonemightholdthattheplagueiscausedbydivineretribution,orthattheillnessitselfinvolvesthepossessionofevilspirits.But,again,someonewhobelievessuchthingsshouldstillbecapableofentertainingtheverysameconceptaswedo–thePLAGUE.Indeed,itisnecessaryforthemtohavetheverysameconceptinordertomakesenseoftheideathatwecandisagreewiththem199\nEricMargolisandStephenLaurenceaboutthenatureandcauseofthedisease.IgnoranceanderrorareasproblematicfortheTheoryTheoryastheywerefortheClassicalTheory.15Still,whethertwopeopleareemployingthesameconceptornotisadifficultquestion.Wesupposethatmanytheoristswouldclaimthatit’ssimplyinappropriatetoinsistthattheverysameconceptmayoccurdespiteadifferenceinsurroundingbeliefs.Thealternativesuggestionisthatpeopleneedonlyhavesimilarconcepts.Theideaisthatdifferencesinbeliefdoyielddistinctconcepts,butthisisnotproblematicbecausetwoconceptsmightstillbesimilarenoughincontentthattheywouldbesubsumedbythesamepsychologicalgeneralizations–andperhapsthat’sallthatreallymatters.Astemptingasthispositionmaybe,itisactuallyfraughtwithdifficulty.Theproblemisthatwhenthenotionofcontentsimilarityisunpackeditgenerallypresupposesapriornotionofcontentidentity(FodorandLepore1992).Forexample,acommonstrategyformeasuringcontentsimilarityisintermsofthenumberofconstituentsthattwoconceptsshare.Iftheyoverlapinmanyoftheirconstituents,thentheyaresaidtohavesimilarcontents(see,e.g.,Smithetal.1984).Butnoticethatthisproposalworksonlyontheassumptionthattheshared,overlappingconstituentsarethesame.Sothenotionofcontentsimilarityisillicitlybuildingontheverynotionitissupposedtoreplace.Sincethescopeofthisproblemhasn’tbeenabsorbedineitherphilosophicalorpsychologicalcircles,itpaystoexploresomeotherproposedsolutions.Consider,forexample,asuggestionbyEricLormand(1996).Lormandclaimsthatevenacompletelyholistictheoryofcontentneedn’thaveanydifficultieswithstability;inotherwords,stabilityisn’tsupposedtobeaproblemevenforatheorythatclaimsthatanychangeinthetotalbeliefsystemchangesthecontentofeverysinglebelief.Thetricktoestablishingstability,Lormandclaims,istheideathatagivensymbolhasmultiplemeanings.Eachofitsmeaningsisgivenintermsofasubsetofitscausal/inferentiallinks.Lormandcallsthesesubsetsunitsandasksustothinkofaunit“asaseparableroughtestfortheacceptableuseofthatrepresentation”(1996:57).Theproposal,then,isthataholisticsystemofrepre-sentationcanallowforstabilityofcontent,since,asthesystemexhibitschanges,someofaconcept’smeaningschange,butsomedon’t.Totheextentthatitkeepssomeofitsunitsintact,itpreservesthosemeanings.Unfortunately,thissuggestiondoesn’twork.SinceLormand’sunitsarethem-selvesrepresentations,theyarepartoftheholisticnetworkthatdeterminesthecontentofeveryconceptinthesystem.Asaresult,everyconceptembeddedinanyunitwillchangeitsmeaningastheothermeaningsintheinferentialnetworkchange.Andiftheychangetheirmeaning,theycan’tbethebasisofthestabilityforotherconcepts(MargolisandLaurence1998).PaulChurchland(1998)hasproposedadifferentsolution.Forsometime,Churchlandhasbeendevelopinganapproachtomentalcontentknownasstate-spacesemantics.State-spacesemanticsisatheoryofcontentforneuralnetworkswherecontentissupposedtobeholistic.Toafirstapproximation,thecontentofanactivationvector–i.e.,apatternofactivationacrossanassemblyofnodesin200\nConceptssuchanetwork–issupposedtobedeterminedbyitspositionwithinthelargerstructureofthenetwork.Sincethispositionwillberelativetothepositionsofmanyothernodesinthenetwork,state-spacesemanticsshouldhaveconsiderabledifficultiesinachievingcontentstability.Asaresult,Churchlandisquicktorejectcontentidentityinfavorofcontentsimilarity.Inearlierwork,ChurchlandadoptedamodelmuchliketheoneinSmithetal.(1984).Imagineaconnectionistnetworkwithaseriesofinputnodes,outputnodes,andanintermediarysetofso-calledhiddennodes.Takingthehiddennodesasspecifyingcontentfuldimensions,wecanconstructasemanticspaceofasmanydimensionsastherearehiddennodes,wherepointswithinthespacecorrespondtopatternsofactivationacrossthehiddennodes.Supposingforsimplicitythatthereareonlythreehiddennodes,theresultingsemanticspacewouldbeacube,eachofwhoseaxescorrespondstoaparticularhiddennodeanditslevelofactivation.OnChurchland’searlytreatments,contentsimilaritywasunderstoodasrelativeclosenessinaspaceofthissort.ButthisapproachrunsintomuchthesameproblemastheSmithetal.account.Itonlyexplainssimilarityofcontentbypresupposingapriornotionofidentityofcontent,onethatappliestotheconstitutingdimensionsofthespace.Inlightofthisdifficulty,Churchlandhasrecentlyputforwardanewaccountofsimilarityofcontent.Inthenewmodel,Churchlandsuggests:Apointinactivationspaceacquiresaspecificsemanticcontentnotasafunctionofitspositionrelativetotheconstitutingaxesofthatspace,butratherasafunctionof(1)itsspatialpositionrelativetoalloftheothercontentfulpointswithinthatspace;and(2)itscausalrelationstostableandobjectivemacrofeaturesoftheexternalenvironment.(1998:8)Thisnewposition,Churchlandtellsus,“constitute[s]adecisiveanswertoFodorandLepore’schallenge”(ibid:5)toprovideaworkableholisticaccountofcon-tentsimilarity.Yetfarfrombeingadecisiveanswertothechallenge,Churchland’snewac-countisreallynoimprovementatall.Hisfirstdeterminantofcontent–spatialpositionrelativetoothercontentfulpointsinthespace–immediatelyconfrontsaseriousdifficulty.Supposingthattwonetworksdohavenodeswiththesameoverallrelativepositions,thisalonedoesn’tsufficetofixtheircontents;onemightwellwonderwhyanygivennodeineithernetworkhastheparticularcontentithas(andnotsomeothercontent).Forexample,Churchlanddescribesonetypeofnetworkasrepresentingdistinctfamiliesasitextractsfourprototypicalfacesgivenphotographsasinput.Butwhatmakesitthecasethatthenetwork’snodesrepresentfamiliesandfacesasopposedtoanyofawidevarietyofpotentialobjects?Inresponsetothisproblem,Churchlandcanonlyappealtotheresourcesofhisseconddeterminantofcontent–causalrelationstofeaturesoftheenviron-ment.Theproblemwiththisanswer,however,isthatthisisn’taversionoftheTheoryTheoryatall.Rather,itreliesonanatomistictheoryofcontentofthe201\nEricMargolisandStephenLaurencesortwediscussinthenextsection.Therelationofthenodetoitssurroundingnodesturnsouttohavenothingtodowithitscontent;whatmattersforcontent16isjusttheexistenceofareliablecausallinktofeaturesoftheenvironment.Ofcourse,thesereliablelinksprovidestability,butthat’sbecausetheyunderwriteatheoryofcontentidentity:Twonodeshaveidenticalcontentsjustincasetheyarelinkedtothesameenvironmentalfeature.Soit’snosurprisethatChurchlandcanhaveanotionofsimilarcontent,sincehehelpshimselftoanindependent17accountofsamenessofcontent,despitehisrhetorictothecontrary.Stability,itturnsout,isarobustconstraintonatheoryofconcepts.WhatthismeansfortheTheoryTheoryisthatmentaltheoriesmakeforbadcores.TheyhaveasmuchtroubleasthePrototypeTheorywhenitcomestoreference,andtheyareespeciallybadinsecuringstability.IfaversionoftheDualTheoryofconceptsistosucceed,itlookslikeit’snotgoingtobeonewhosecoreshaveeitherclassicalstructureortheorystructure.8.4ConceptsWithoutStructureWe’veseenthatthemainviewsofconceptualstructureareallproblematic.Inlightofthesedifficulties,anumberoftheoristshaveproposedtoexplorethepossibilitythatlexicalconceptsdon’thaveanystructure–aviewknownasConceptualAtomism(see,e.g.,Fodor1998;Leslie2000;Millikan1998,2000).CentraltoConceptualAtomismisthethesisthataconcept’scontentisn’tdeter-minedbyitsrelationtoanyotherparticularconcepts.Instead,it’sdeterminedbyamind–worldrelation,thatis,acausalorhistoricalrelationbetweenthesymbolandwhatitrepresents.Notsurprisingly,AtomismfindsitsinspirationinKripke’sandPutnam’streatmentofnaturalkindterms,onlyit’sintendedtocoverabroaderrangeofsemanticitemsandisdirected,inthefirstinstance,tothenatureoftheconceptualsystem,nottolanguage.Themostdifficulttaskforanatomististoprovideasufficientlydetailedaccountofthemind–worldrelationthat’ssupposedtodetermineconceptualcontent.Onegeneralstrategyistoexplaincontentintermsofthenotionofco-variation(thesamenotionthatwesawwasillicitlyatplayinChurchland’streat-mentofstability).Theideaisthataconceptrepresentswhatitcausallyco-varieswith.Forexample,iftheconceptDweretokenedasareliablecausalconsequenceofthepresenceofdogs,then,onthepresentaccount,thesymbolwouldexpressthepropertydogandbetheconceptDOG.Notice,however,thatthissimpleaccountwon’tdo.ThereasonisbecauseallsortsofotherthingswillreliablycausetokeningsofthesymbolD.Thismighthappen,forexample,asaresultofperceptualerror.Onadarknightyoumightcatchafoxoutofthecornerofyoureyeandmistakeitforadogrunningpastyourcar.Atomistshaveanumberofresourcesforrulingoutthenon-dogs.Oneistoaddthefurtherconditionthataconceptrepresentswhatitwouldco-varywith202\nConceptsunderidealconditions(allowingforthepossibilitythatnon-dogscauseDOGswhentheconditionsaren’tideal;see,e.g.,Stampe1977;Fodor1981/90).Anotheroptionistosaythataconceptrepresentswhatithasthefunctionofco-varyingwith(allowingforthepossibilitythattheconcept,orthesystemthatproducesit,isn’tfunctioningproperlyinthenon-dogcases;see,e.g.,Dretske1995;Millikan1984,1993).Yetanotherpossibilityistosaythatthedog/DOGdependenceis,inasense,morebasicthanthenon-dog-yet-dog-like/DOGdependence.Forinstance,theformerdependencemayholdwhetherornotthelatterdoes,butnottheotherwayaround(Fodor1990).Thougheachofthesestrategieshasitsowndifficulties,wewanttofocusonmoregeneralproblemswithAtomism,onesthataren’ttiedtothedetailsofanyparticularatomistictheory.We’llmentionthree.Thefirstobjectionconcernstheexplanatoryroleofconcepts.Mosttheoriestieaconcept’sexplanatorypotentialtoitsstructure.Thisisevidentintheothertheorieswe’vereviewed.Forinstance,thePrototypeTheoryexplainsawidevarietyofpsychologicalphenomenabyreferencetoconceptualstructure–cat-egorization,typicalityjudgments,efficiencyofuse,andsoon.TheproblemwithConceptualAtomism,however,isthatitsaysthatconceptshavenostructure.Soitwouldseemthattheycan’treallyexplainanything.Thenwhatgoodarethey?ThesecondobjectionistheworrythatConceptualAtomismiscommittedtoanextremelyimplausibledegreeofinnateness.Infact,JerryFodor,themostvocaldefenderofAtomism,hasmadethisconnectionexplicitly,defendingtheclaimthatvirtuallyalllexicalconceptsareinnate,includingsuchunlikelycandi-datesasCARBURETORandQUARK.AsFodorseesit,theonlywaythataconceptcouldbelearnedisviaaprocessofconstruction,whereitisassembledfromitsconstituents.SinceAtomismmaintainsthatlexicalconceptshavenoconstituents,theymustallbeinnate(Fodor1981).ButifCARBURETORisinnate,somethinghasdefinitelygonewrong;maybethatsomethingisAtomismitself.Thethirdobjectionisthatatomistictheoriesindividuateconceptstoocoarsely.Sincetheyreducecontenttoacausalorhistoricalrelationbetweenarepresenta-tionandwhatitrepresents,conceptswouldseemtobenomorefinelyindividuatedthantheworldlyitemstheypickout.Yetsurelythatisn’tfineenough.TheconceptWATERisn’tthesamethingastheconceptH2O–someonecouldhavetheonewithouttheother–butpresumablytheypickouttheverysameproperty.Ortotakeamoreextremecase,theconceptUNICORNisn’tthesamethingastheconceptCENTAUR,yetbecausetheyareemptyconcepts,theywouldseemtopickouttheverysamething,viz.,nothing.Soit’shardtoseehowanatomistictheorycouldteasesuchconceptsapart.Let’staketheseobjectionsinreverseorder.Nodoubt,theproblemofachiev-ingafine-grainedindividuationisaseriousconcernforAtomism,butatomistsdohaveafewresourcestheycancallupon.Forinstance,inthecaseofemptyconcepts,theycanmaintainthatthecontentdeterminingco-variationrelationisanomicrelationbetweenproperties.Thishelpsbecauseit’splausibletherecanbenomicrelationsbetweenpropertieseveniftheyareuninstantiated(Fodor1990).203\nEricMargolisandStephenLaurenceWithotherexamples,atomistscandistinguishco-referentialconceptsbyinsistingthatoneoftheconceptsisreallycomplexandthatitscomplexityisn’tindispute.Presumably,thisishowtheywouldhandletheWATER/H2Ocase–bymaintainingthattheconceptH2Oincorporates,amongotherthings,theconceptHYDROGEN(Fodor1990).Ofcourse,thereareotherchallengingcasesforwhichneitherofthesestrategieswillwork.Herewehaveinmindpairsofprimitiveconceptsthatexpressnomologicallyco-extensiveproperties(e.g.,BUYING/SELLING,CHASING/FLEEING,EXTENDED/SHAPED).Theseprovetobethemostdifficultcases,sincethenaturalsolutionfordistinguishingthemistosaytheyareassociatedwithdifferentcontent-determininginferences.Whetheratomistshaveanalternativesolutionisveryhardtosay.Butlet’sturntotheotherobjectionstoAtomism,which,onthefaceofit,leavetheatomistwithevenlessroomtomaneuver.IfAtomismsaysthatlexicalconceptshavenostructure,musttheyallbeinnate?Andiflexicalconceptshavenostructure,whyaren’ttheyexplanatorilyinert?Fodor’sargumentforradicalconceptnativismhascausedquiteastirinphilo-sophyofmind,withtheoristsofdifferentsortsdroppinganydoctrinethoughtto18betiedupwiththethesis.Asaresult,theargumenthasnotreceivedthesortofcarefulcriticalscrutinythatitdeserves.WebelievethatAtomismhasbeenun-fairlyburdenedwithFodor’sstrongnativistthesis,andthatinfactitispossibletoprovideasatisfyingaccountofhownewprimitiveconceptscanbeacquiredinawaythatiscompatiblewithConceptualAtomism.Thekeyhereisthenotionofasustainingmechanism.Sustainingmechanismsaremechanismsthatunderwritethemind–worldrelationthatdeterminesaconcept’scontent.Thesewilltypicallybeinferentialmechanismsofonesortoranother,sincepeopleclearlylacktrans-ducersformostofthepropertiestheycanrepresent.Importantly,however,theseinferentialmechanismsneedn’tgiverisetoanyanalyticitiesortoaconcept’shavinganysemanticstructure,sincenoparticularinferenceisrequiredforcon-ceptpossession.Thus,suchinferentialmechanismsarefullycompatiblewithConceptualAtomism.WearenowinapositiontoseewhyAtomismisnotcommittedtoradicalconceptnativism.Whattheatomistoughttosayisthatthegeneralquestionofhowtoacquireaconceptshouldbeframedintermsofthemorerefinedquestionofhow,giventhecorrecttheoryofcontent,someonecomestobeinastateofmindthatsatisfiesthetheory(Margolis1998;LaurenceandMargolis2002).Onanatomistictreatmentofcontentthisistobeunderstoodintermsofthepossessionofasuitablesustainingmechanism.Sothequestionofacquisitionjustisthequestionofhowsustainingmechanismsareassembled.Andheretherearemanythingsthatanatomistcansay,allconsistentwiththeclaimthatconceptshavenostructure.Forexample,onetypeofsustainingmechanismthatwe’veexploredindetailsupportsthepossessionofnaturalkindconcepts(seeMargolis1998;LaurenceandMargolis,forthcoming).Themodelisbasedonwhatwecallasyndrome-basedsustainingmechanism,onethatincorporateshighlyindicativeperceptualinformationaboutakindtogetherwithadispositiontotreatsomethingasamemberofthe204\nConceptssamekindsolongasitsharesthesameconstitutivehiddenproperties(andnotnecessarilythesameperceptualproperties)asthecategory’sparadigmaticinstances.Thesuggestionisthatpeoplehaveageneraltendencytoassemblesyndrome-basedsustainingmechanismsinaccordancewiththeirexperience.Suchamechanismthenestablishesthemind–worldrelationthatatomistssayisconstitutiveofcon-tent,andtogetherwithenvironmentalinputiscapableofdeliveringawiderangeofunstructuredconcepts.Sincethemechanismrespectsthecharacterofone’sexperience–acquisitionproceedsbythecollection,storage,andmanipulationofinformationtoproducearepresentationthattracksthingsintheconcept’sexten-sion–wethinkitisfairtosaythatthisisalearningmodel.TurningfinallytothechargethatAtomismleavesconceptsexplanatorilyinert,thebeststrategyfortheatomististosaythattheexplanatoryrolesthatareoftenaccountedforbyaconcept’sstructureneedn’tactuallybeexplaineddirectlyintermsoftheconcept’snature.Theideaisthattheatomistcanappealtoinforma-tionthathappenstobeassociatedwiththeconcept;thatis,theatomistcanmakeuseoftherelationsthataconceptCbearstootherconcepts,eventhoughtheseothersaren’tconstitutiveofC.Thismayseemadrasticstep,butvirtuallyanytheoryofconceptswilldothesameinordertoexplainatleastsomeinferencesinwhichconceptsparticipate.Perhapsasachildyouwerefrightenedbyadogandasaresultyou’vecometobelievethatdogsaredangerous.Thisbeliefmaywellexplainquitealotofyourbehaviortowarddogs.Nonetheless,aclassicaltheoristwouldnotlikelysupposethatitwaspartofthedefinitionofDOGthatdogsaredangerous.Alltheoriesofconceptssaythatsomeofaconcept’srelationstootherconceptsareconstitutiveofitsidentityandsomearenot.Andhavingmadethatdistinction,it’ssometimesgoingtobethecasethathowaconceptisdeployedwillreflectitsnon-constitutiverelations.Theatomistsimplytakesthispositiontothelimitandsaysthatthisisalwaysthecase.Aconcept’sroleinthoughtcan’thelpbutreflectitsnon-constitutiverelations,sincewhat’sconstitutiveofacon-ceptisn’titsrelationtoanyotherparticularconceptsbutjusthowitiscausally(orhistorically)relatedtothingsintheworld.Onewonders,however,whethertheatomisthasgonetoofar.Coulditreallybethatnoneofthewaysinwhichaconceptisdeployedisexplainedbyitsnature?8.5RethinkingConceptualstructureThere’ssomethingunsettlingabouttheclaimthattheexplanatoryfunctionsofconceptsarehandledbytheirincidentalrelations.Consideronceagaintypicalityeffects.Typicalityeffectsaresopervasiveandsorichintheirpsychologicalimportthattheyconstituteoneofthecentralexplanandaofanytheoryofconcepts.Indeed,itislargelybecauseoftheClassicalTheory’sfailuretoaccountfortheseeffectsthatpsychologistsabandonedtheClassicalTheoryindroves.Notice,how-ever,thatConceptualAtomismisnodifferentthantheClassicalTheoryinits205\nEricMargolisandStephenLaurencecapacitytodealwithtypicalityeffects.Bymaintainingthatconceptshavenostructure,atomistsarecommittedtotheviewthataconcept’snaturehasnobearingwhatsoeveronitsroleintypicalityeffects.Ofcourse,thisdoesn’tmeanthatatomistshavetodenytheexistenceoftypicalityeffects.Yetitispuzzlingthatsomeofthemostimportantpsychologicaldatainvolvingconceptsenduphavingnothingatalltodowiththeirnature.Atthesametime,therearecompellingpressuresmitigatinginfavorofAtomism’scentralclaimthatconceptsdon’thaveanystructure.Inparticular,allattemptstoexplainreferencedeterminationintermsofaconcept’sstructurerunintoformid-abledifficulties.TheClassicalTheory,thePrototypeTheory,andtheTheoryTheoryallfallpreytotheproblemsofignoranceanderror,andeachtheoryhasitsownpeculiardifficultiesaswell.ThewayoutofthisimpasseliesintworelatedinsightsaboutconceptualstructurethatareimplicitintheDualTheory.Thefirstoftheseissimplythatconceptscanhavemultiplestructures.ThusintheoriginalDualTheoryconceptsweretakentohavecoresandidentificationprocedures.Thesecondinsightislessobviousbutit’sreallythecrucialone.Thisisthatconceptscanhavecategorically19differenttypesofstructureansweringtoverydifferentexplanatoryfunctions.TheDualTheoryimplicitlyrecognizesthispossibilityinthedistinctmotivationsthatitassociateswithcoresandidentificationprocedures.Butoncethepointismadeexplicit,andonceitismadeinperfectlygeneralterms,awholenewrangeoftheoreticalpossibilitiesemerges.ThemostimmediateeffectistheDualTheory’srecognitionthatthefunctionofexplainingreferencemayhavetobeteasedapartfromcertainotherfunctionsofconcepts.Thiswouldfreetheothertypesofstructurethataconcepthasfromaheavyburdenand,crucially,wouldimplythatnotallconceptualstructureisreference-determiningstructure.Havingtakenthisstep,onecantheninquireaboutwhatothertypesofconceptualstructurethereareandaboutthespecificfunctionstheyanswerto.Wesuggestthatthereareatleastfourcentraltypesofstructure:Compositionalreference-determiningstructureThisisstructurethatcontributestothecontentandreferenceofaconceptviaacompositionalsemantics.ThistypeofstructureisfamiliarfromtheClassicalTheory.Whetheranylexicalconceptshavethistypeofstructurewilldependonwhethertheproblemsofanalyticityandignoranceanderrorcanbemetandwhetherdefinitionscanactuallybefound.However,itismoreorlessuncontroversialthatphrasalconceptssuchasBROWNDOGhavethiskindofstructure.BROWNDOGiscomposedofBROWNandDOGanditsreferenceiscompositionallydeterminedbythereferentialpropertiesofitsconstituents:SomethingfallsunderBROWNDOGjustincaseit’sbrownandadog.Non-semanticstructureThisisstructurethatdoesn’tcontributetothecont-entofaconceptbutdoescontributesignificantlytosomeothertheoretically206\nConceptsimportantexplanatoryfunctionofconcepts.ThoughtheDualTheoryisnotexplicitaboutthis,itseemsplausibletothinkofDualTheory’scommitmenttoprototypesasacommitmenttonon-semanticstructure.Non-referentialsemanticstructureThisisstructurethatcontributestothecon-tentofaconceptbutisisolatedfromreferentialconsequences.Thoughourdiscussionofthemeaningorcontentofconceptshasfocusedontheirreferentialproperties,thesemaywellnotexhaustthesemanticpropertiesthatconceptspossess.Thistypeofstructurewouldapplyto,amongotherthings,so-called20narrowcontent.SustainingmechanismstructureThisisstructurethatcontributestothecontentofaconceptindirectlybyfiguringinatheoreticallysignificantsustainingmech-anism.Sustainingmechanismstructuredeterminesthereferentialpropertiesofaconcept,butnotviaacompositionalsemantics.Rather,thistypeofstructuresupportsthemind–worldrelationthat(directly)determinesaconcept’scontent.Thesefourdifferenttypesofstructurepointtoarangeofnewtheoreticaloptionsthatbearexploring.Bywayofillustration,wewillbrieflysketcharesolutiontotheimpassebetweenConceptualAtomismandthepressuretoappealtoacon-cept’sstructureinexplainingitsmostsalientbehavior.IfwelookbackattheDualTheory,themainproblemsitfacescenterarounditstreatmentofconceptualcores.We’veseenthatbothdefinitionalstructureandtheorystructureareequallyproblematicinthisregard.Neitherisespeciallysuitedtoreferencedetermination;and,inanycase,definitionshaveproventobequiteelusive,whiletheorystructurehasitsdifficultieswithstability.Notice,however,thatthereisnowanalternativeaccountofcoresavailable.Giventhedistinctionswehavejustdrawnamongthefourtypesofconceptualstructure,ConceptualAtomismisbestconstruednotintermsoftheglobalclaimthatlexicalconceptshavenostructureatall,butratherasclaimingthattheyhavenocompositionalreference-determiningstructure.Thisopensthepossibilitythatthecoresofcon-ceptsmightbeatomic.Indeed,atomsseemtobealmostperfectlysuitedtofilltheexplanatoryrolesassociatedwithconceptualcores.Ifcoresareatomic,thenonedoesn’thavetoworryaboutthefactthatconceptsaren’tdefinable.Atomismimpliesthattheyaren’t.Similarly,ifcoresareatomic,thenonedoesn’thavetoworryaboutstability.Atomismimpliesthataconcept’srelationstootherconceptscanchangeasmuchasyoulikesolongasthemind–worldrelationthatdeterminesreferenceremainsinplace.Atomiccoresalsoexplaintheproductivityofconcepts:complexconceptsaregeneratedthroughtheclassicalcompositionalityofatomiccores.Theonlyexplanatoryroleassociatedwithcoresthatatomsseemtohavetroublewithisaccountingforourmostconsideredjudgmentsaboutcategorymember-ship.However,it’shardlyclearthatthisisalegitimatedesideratumforatheoryofconceptualcoresinthefirstplace.IfQuine’sworkonanalyticityshows207\nEricMargolisandStephenLaurenceanything,it’sthatpeople’smostconsideredjudgmentsofthissortareholistic,soit’snotplausibletosupposethatallofthisinformationcouldbeisolatedforeachconcepttakenindividually.Droppingthislastdesideratum,then,thereisagoodcasetobemadeforthinkingthatcoresshouldbeatomic.Atthesametime,amodelofthissortavoidstheobjectionthatAtomismispsychologicallyunexplanatory.Wecanagreewithatomiststhatlexicalconceptsgenerallylackcompositionalreference-determiningstructure,butthisdoesn’tmeanwehavetosaythatconceptsareentirelyunstructured.Forexample,proto-typesandsustainingmechanismsmayverywellbepartofaconcept’sstructure.It’sjustthatthisstructuredoesn’tdirectlydetermineitsreference;referenceisfixedbythemind–worldrelationthatimplicatescores,leavingprototypes(andothertypesofstructure)toexplainotherthings.Andprototypes,forone,doexplainmanyotherthings.Giventheirtremendouspsychologicalsignificance,prototypesshouldbetakentobepartlyconstitutiveofconceptsifanythingis.Conceptsarepsychologicalkinds.Asweseeit,thebesttheoryofconceptsisonethattakestheirpsychologicalcharacterseriously.Thewaytodothisistoadoptatheorythatadmitsdifferenttypesofconceptualstructurewhiletyingthemtogetherbymaintainingthatconceptshaveatomiccores.Inanyevent,itpaystofocusonthenatureofconceptualstructureitself.Articulatingthediffer-entexplanatoryrolesforpostulatingconceptualstructureandteasingtheseapartopensuparangeofunexploredandpotentiallyverypromisingtheoreticaloptionsinthestudyofconcepts.NotesThispaperwasfullycollaborative;theorderoftheauthors’namesisarbitrary.1Thisviewofthenatureofthoughtisnotentirelyuncontroversial.Yetit’sdifficulttoseehowfinitecreatureswithoutaccesstoastructuredsystemofrepresentationcouldbecapableofentertainingthevastnumberofthoughtsthathumanshaveavailabletothem.Evenifwesticktorelativelysimplethoughts,thenumberoftheseistruly18astronomical.Forexample,thereare10simplestatementsofsumsinvolvingnum-berslessthanamillion.ThisismorethanthenumberofsecondssincethebeginningoftheUniverseandmorethanamilliontimesthenumberofneuronsinthehumanbrain.Howcouldatheoryofthoughtaccommodatethesefactswithoutpostulatingastructuredrepresentationalsysteminwhichthesameelements–concepts–canoccurindifferentpositionswithinastructuredassembly?Inanyevent,ifatheoryreallysaysthatthoughtsdon’thaveconstituents,perhapsthebestthingtosayisthat,accordingtothattheory,therearen’tanysuchthingsasconcepts.2Wewillassumethatthoughtsandconceptshavesemanticpropertiesandthatchiefamongthesearetheirtruth-theoreticproperties.Wetakeittobeanimportantconstraintonatheoryofconceptsthat,e.g.,theconceptDOGreferstodogs.3Still,itisworthnotingthatthetheorieswediscusscanbeadaptedwithslightmodifica-tiontoalternativeframeworksthattakedifferentstandsonthesefoundationalquestions.208\nConcepts4Formoredetailedsurveysanddevelopmentoftheviewshere,seeLaurenceandMargolis(1999;inprep.).SeealsoSmithandMedin(1981).5Themainreasonforthequalificationisthat,accordingtotheClassicalTheory,someconceptshavetohavenostructure;thesearetheprimitiveconceptsoutofwhichallothersarecomposed.Classicaltheoristshavehadlittletosayabouthowthereferenceofaprimitiveconceptisfixed.Butthemostvenerableaccount,owingtotheBritishempiricists,isthatprimitiveconceptsexpresssensorypropertiesandthattheyrefertothesesimplybecausetheyarecausallylinkedtosuchpropertiesviasensorytransducers.6Workonthetheoryofconceptshasbecomeincreasinglyinterdisciplinary,andmanyofthetheorieswewilldiscussbearthemarksofideasandmotivationswhichhavebeentransferredacrossdisciplinaryboundaries,particularlybetweenpsychologyandphilosophy.Inlinewithmuchofthisresearch,wetakeconceptstobementalrepres-entations(andthusmentalparticulars),sincethisperspectivemakesthemostsenseofthevariouspsychologicalexplanandathathaverightlyexertedconsiderablepres-sureontheorizingaboutconcepts–eveninphilosophicalcircles.Thereadershouldnotethatthisisnotauniversallysharedperspectiveandthatmanyphilosophersinsistonconstruingconceptsasabstractentitiesofonesortoranother.Nonetheless,theoristswhotakeconceptstobeabstractaalsotakeadeepinterestinquestionsaboutconceptualstructure.It’sjustthatthestructureinquestionissupposedtobethestructureofabstractentities.See,e.g.,Peacocke(1992)andBealer(1982).7Astheexampleshereindicate,theClassicalTheory(andindeedallthetheorieswewillbediscussing)is,inthefirstinstance,atheoryaboutthenatureofconceptsthatcorrespondtowordsinnaturallanguage–whatarecalledlexicalconcepts.Thisisbecausetheoristsinterestedinconceptsassumethattherepresentationscorrespond-ingtonaturallanguagephrasesorsentencesarestructured.8ThemotivationfortheClassicalTheoryisbynomeanslimitedtothesevirtues.Forexample,anotherinfluentialpointinfavorofthistheoryisitsabilitytoexplainourintuitionsthatcertainstatementsorargumentsarevalideventhough,onthefaceofit,theyfailtoexpresslogicaltruths,e.g.,“Johnisabachelor,soJohnisunmarried”(see,e.g.,Katz1972).9Classicaltheoristshavehadlittletosayindefenseofthenotionofanalyticity.E.g.,ChristopherPeacocke’sseminalbookonconcepts(1992)fallssquarelyintheclassicaltradition,especiallyinitscommitmenttodefinitions,yetPeacocketakeslittlenoticeoftheproblemsassociatedwithanalyticity,simplystatinginafootnotethatheiscommittedtosomeversionoftheanalytic/syntheticdistinction(seep.244,fn7).SeeKatz(1997),however,forarareclassicaldefenseofanalyticity,especiallyinthefaceofthepresentconsiderations.10Inthemostextremecases,peopleknowhardlyanyinformationatall.Forinstance,Putnamremarksthathecan’tdistinguishelmsfrombeeches,thatforhimtheyarebothjusttrees.Yetarguably,hestillhastwodistinctconceptsthatreferseparatelytoelmsandbeeches.Thatwouldn’tbepossibleifthemechanismofreferencehadtobeaninternalizeddefinition.11Whatwearecalling“thePrototypeTheory”isanidealizedversionofabroadclassoftheories,onethatabstractsfrommanydifferencesofdetail.Thisistrueofeachofthetheorieswepresent,thoughthediversityisperhapsmorepronouncedinthecaseofthePrototypeTheory.Fordiscussionofsomeofthedifferentvarieties,seeSmithandMedin(1981).209\nEricMargolisandStephenLaurence12TheDualTheoryshouldnotbeconfusedwithso-calledTwoFactortheoriesinphilosophy.Thoughtherearesimilarities,theDualTheoryandTwoFactortheoriesaddressdifferentissues.TwoFactortheoriesareprimarilyconcernedwithdistinguish-ingtwodifferenttypes,oraspects,ofcontent.Onefactoraccountsforallaspectsofcontentthatsuperveneonaperson’sbodyorthatwouldbesharedbymoleculeformoleculeduplicates(“narrowcontent”).Theotherfactoraccountsforaspectsofcontentthatgobeyondthis,involvingtheperson’srelationtoherenvironment(“widecontent”).Asaresult,thetwotypesofstructureintheDualTheorycross-classifythetwoaspectsofcontentinTwoFactortheories(seenote20below).13AccordingtotheTheoryTheory,thestructureofaconceptisconstitutedbyitsrelationstotheotherconceptsthatareimplicatedinanembeddingtheory.Noticethatonthisaccountthestructureofaconceptcan’tbeunderstoodintermsofpart/wholerelations.Forthisreason,wehavedistinguishedtwomodelsofconceptualstructure(seeLaurenceandMargolis1999).Thefirst,theContainmentModel,saysthatoneconcept,C1,isincludedinthestructureofanother,C2,justincaseC1isliterallycontainedin(i.e.,isaproperpartof)C2.Thesecond,theInferentialModel,saysthatC1isincludedinthestructureofC2justincaseC1standsinaprivilegedinferentialrelationtoC2.Asshouldbeevidentfromthischaracterization,theTheoryTheoryhastobeconstruedintermsoftheInferentialModel,buttheClassicalTheoryandtheProto-typeTheorycouldbeconstruedintermsofeithermodel,dependingontheexactmotivationsthatsupportthepostulationofclassicalandprototypestructure.14Theseparticulardomainshavebeenthesubjectofintenseinterdisciplinaryinvestiga-tioninrecentyears.Forcommon-sensepsychology,seeDaviesandStone(1995a,1995b),Carruthers(1996);forcommon-sensephysics,seeSpelke(1990),Baillargeon(1993),XuandCarey(1996);forcommon-sensebiology,seeMedinandAtran(1999).15Or,forthatmatter,whetherthesamepersonisemployingthesameconceptovertime.16Atbest,Churchland’smodelshowshowpsychologicalprocessescouldbeholistic.Theyareholisticbecausetheyinvolveactivationpatternsacrossmassivelyconnectednodesinanetwork.Butthisdoesn’tmeanthatthesemanticsofthenetworkareholistic.17ItshouldbenotedthatChurchlandissomethingofamovingtargetontheseissues,thoughheoftenneglectstoacknowledgechangesinhisview.Forinstance,inadditiontothepositionsmentionedinthetext,Churchlandalsotriesmaintainingthatcontentsimilarityisamatterofsimilarityof“downstreamprocessing”(seeesp.1996:276),Itisthisdownstreamaspectofthevector’scomputationalrolethatissovitallyimportantforreckoningsamenessofcognitivecontentacrossindividuals,oracrosscultures.Apersonorculturethatdiscriminatedkittensreliablyenoughfromtheenvironment,buttreatedtheminabsolutelyeveryrespectasavariantformofwharf-rat,mustbeascribedsomeconceptionof“kitten”importantlydifferentfromourown.Ontheotherhand,analienpersonorspecieswhoseexpectationsofandbehaviortowardskittenspreciselymirrorourownmustbeascribedthesameconcept“kitten,”eventhoughtheymightdiscriminatekittensprincipallybymeansofalienolfactionandhigh-frequencysonarsbeamedfromtheirforeheads.210\nConceptsApartfrommakinghis“statespacesemantics”havenothingwhatsoevertodowiththestatespace,thispositionfallspreytoexactlythesamesortsofproblemsasChurchland’sfirstposition,namely,itpresupposesanotionofcontentidentityforthe“downstream”statesthatfixthecontentofthekittenvector.18See,e.g.,Churchland(1986)andPutnam(1988).19Thesetwopointsgohandinhand,sinceit’stobeexpectedthatifaconcepthasmultiplestructuresthatthesewouldbeofcategoricallydifferenttypes.20Thenatureofnarrowcontentiscontroversialbutthemainideaisthatnarrowcontentissharedbymolecule-for-moleculeduplicateseveniftheyinhabitdifferentenvironments.OnsomeTwoFactortheories(seenote12),aconcept’snarrowcontentisdeterminedbyitsinferentialrole–aviewthatcloselyresemblestheTheoryTheory’saccountofconceptualstructure.Thedifferenceisthat,onaTwoFactortheory,theinferentialroleofaconceptisn’tsupposedtodetermineitsreference.ReferencesBaillargeon,R.(1993).“TheObjectConceptRevisited:NewDirectionsintheInvestiga-tionofInfants’PhysicalKnowledge.”InC.Granrund(ed.),VisualPerceptionandCogni-tioninInfancy.Hillsdale,NJ:LawrenceErlbaumAssociates.Bealer,G.(1982).QualityandConcept.Oxford:ClarendonPress.Barsalou,L.(1987).“TheInstabilityofGradedStructure:ImplicationsfortheNatureofConcepts.”InU.Neisser(ed.),ConceptsandConceptualDevelopment:EcologicalandIntellectualFactorsinCategorization.NewYork:CambridgeUniversityPress.Carey,S.(1982).“SemanticDevelopment:TheStateoftheArt.”InE.WannerandL.Gleitman(eds.),LanguageAcquisition:TheStateoftheArt.NewYork:CambridgeUniversityPress.——(1985).ConceptualChangeinChildhood.Cambridge,MA:MITPress.Carruthers,P.(ed.)(1996).TheoriesofTheoriesofMind.Cambridge:CambridgeUniver-sityPress.Churchland,P.M.(1996).“FodorandLepore:State-SpaceSemanticsandMeaningHolism.”InR.McCauley(ed.),TheChurchlandsandTheirCritics.Cambridge,MA:Blackwell.——(1998).“ConceptualSimilarityacrossSensoryandNeuralDiversity:TheFodor/LeporeChallengeAnswered.”JournalofPhilosophy,XCV(1):5–32.Churchland,P.S.(1986).Neurophilosophy:TowardaUnifiedScienceoftheMind/Brain.Cambridge,MA:TheMITPress.Dancy,J.(1985).IntroductiontoContemporaryEpistemology.Cambridge,MA:Blackwell.Davies,M.andStone,T.(eds.)(1995a).FolkPsychology.Oxford:Blackwell.——(eds.)(1995b).MentalSimulation.Oxford:Blackwell.Dretske,F.(1995).NaturalizingtheMind.Cambridge,MA:MITPress.Fodor,J.D.,Fodor,J.A.,andGarrett,M.(1975).“ThePsychologicalUnrealityofSemanticRepresentations.”LinguisticInquiry,6:515–32.Fodor,J.A.(1981).“ThePresentStatusoftheInnatenessControversy.”InRepresentations:PhilosophicalEssaysontheFoundationsofCognitiveScience.Cambridge,MA:MITPress.——(1981/90).“Psychosemantics;or,WhereDoTruthConditionsComeFrom?”InN.G.Lycan(ed.),MindandCognition.Oxford:Blackwell.211\nEricMargolisandStephenLaurence——(1990).“ATheoryofContent,II:TheTheory.”InATheoryofContentandOtherEssays.Cambridge,MA:MITPress.——(1998).Concepts:WhereCognitiveScienceWentWrong.NewYork:OxfordUniver-sityPress.Fodor,J.A.,Garrett,M.,Walker,E.,andParkes,C.(1980).“AgainstDefinitions.”Cognition,8:263–367.Fodor,J.A.andLepore,E.(1992).Holism:AShopper’sGuide.Cambridge,MA:BasilBlackwell.——(1996).“TheRedHerringandthePetFish:WhyConceptsStillCan’tBeProto-types.”Cognition,58:253–70.Gettier,E.(1963).“IsJustifiedTrueBeliefKnowledge?”Analysis,23:121–3.Gopnik,A.andMeltzoff,A.(1997).Words,Thoughts,andTheories.Cambridge,MA:MITPress.Katz,J.(1972).SemanticTheory.NewYork:HarperandRow.——(1997).“Analyticity,Necessity,andtheEpistemologyofSemantics.”PhilosophyandPhenomenologicalResearch,LVII:1–28.Keil,F.(1994).“Explanation,Association,andtheAcquisitionofWordMeaning.”InL.GleitmanandB.Landau(eds.),TheAcquisitionoftheLexicon,Cambridge,MA:MITPress.Kintsch,W.(1974).TheRepresentationofMeaninginMemory.Hillsdale,NJ:LawrenceErlbaumAssociates.Kripke,S.(1972/1980).NamingandNecessity.Cambridge,MA:HarvardUniversityPress.Landau,B.(1982).“WilltheRealGrandmotherPleaseStandUp?ThePsychologicalRealityofDualMeaningRepresentations.”JournalofPsycholinguisticResearch,11(1):47–62.Laurence,S.andMargolis,E.(1999).“ConceptsandCognitiveScience.”InE.MargolisandS.Laurence(eds.),Concepts:CoreReadings.Cambridge,MA:MITPress.——(2002).“RadicalConceptNativism.”Cognition,86(1):22–55.——(inpreparation).TheBuildingBlocksofThought.Leslie,A.(2002).“HowtoAcquirea‘RepresentationalTheoryofMind’.”InD.SperberandS.Davis(eds.),Metarepresentations.VancouverStudiesinCognitiveScience,vol.10.Oxford:OxfordUniversityPress.Lormand,E.(1996).“HowtoBeaMeaningHolist.”JournalofPhilosophy,XCIII:51–73.Margolis,E.(1998).“HowtoAcquireaConcept.”MindandLanguage,13(3):347–69.Margolis,E.,andLaurence,S.(1998).“MultipleMeaningsandtheStabilityofContent.”JournalofPhilosophy,XCV(5):255–63.Medin,D.andAtran,S.(1999).Folkbiology.Cambridge,Mass:MITPress.Medin,D.andOrtony,A.(1989).“PsychologicalEssentialism.”InS.VosniadouandA.Ortony(eds.),SimilarityandAnalogicalReasoning.NewYork:CambridgeUniversityPress.Millikan,R.(1984).Language,Thought,andOtherBiologicalCategories:NewFoundationsforRealism.Cambridge,MA:MITPress.——(1993).WhiteQueenPsychologyandOtherEssaysforAlice.Cambridge,MA:MITPress.——(1998).“ACommonStructureforConceptsofIndividuals,Stuffs,andRealKinds:MoreMama,MoreMilk,andMoreMouse.”BehavioralandBrainSciences,21:55–65.212\nConcepts——(2000).OnClearandConfusedIdeas:AnEssayaboutSubstanceConcepts.NewYork:CambridgeUniversityPress.Murphy,G.andMedin,D.(1985).“TheRoleofTheoriesinConceptualCoherence.”PsychologicalReview,92(3):289–316.Osherson,D.andSmith,E.(1981).“OntheAdequacyofPrototypeTheoryasaTheoryofConcepts.”Cognition,9:35–58.Peacocke,C.(1992).AStudyofConcepts.Cambridge,MA:MITPress.——(1998).“ImplicitConceptions,UnderstandingandRationality.”InE.Villanueva(ed.),PhilosophicalIssues,9:Concepts.Atascadero,CA:RidgeviewPublishingCompany.Putnam,H.(1962).“TheAnalyticandtheSynthetic.”InH.FeiglandG.Maxwell(eds.),MinnesotaStudiesinthePhilosophyofScience,VolumeIII.Minneapolis:UniversityofMinnesotaPress.——(1970).“IsSemanticsPossible?”InH.KieferandM.Munitz(eds.),Languages,BeliefandMetaphysics.NewYork:StateUniversityofNewYorkPress:50–63.——(1975).“TheMeaningof‘Meaning’.”InK.Gunderson(ed.),Language,MindandKnowledge.Minneapolis:UniversityofMinnesotaPress.——(1988).RepresentationandReality.Cambridge,Mass.:MITPress.Quine,W.(1951/1980).“TwoDogmasofEmpiricism.”InFromaLogicalPointofView:NineLogico-PhilosophicalEssays.Cambridge,MA:HarvardUniversityPress:20–46.Rey,G.(1993).“TheUnavailabilityofWhatWeMean:AReplytoQuine,FodorandLepore.”InJ.A.FodorandE.Lepore(eds.),Holism:AConsumerUpdate.Atlanta:RodopiBV:61–101.Rosch,E.(1978).“PrinciplesofCategorization.”InE.RoschandB.Lloyd(eds.),Cogni-tionandCategorization.Hillsdale,NJ:LawrenceErlbaumAssociates.Rosch,E.andMervis,C.(1975).“FamilyResemblances:StudiesintheInternalStructureofCategories.”CognitivePsychology,7:573–605.Smith,E.(1995).“ConceptsandCategorization.”InE.SmithandD.Osherson(eds.),Thinking:AnInvitationtoCognitiveScience,Vol.3.SecondEdition.Cambridge,MA:MITPress.Smith,E.andMedin,D.(1981).CategoriesandConcepts.Cambridge,MA:HarvardUniversityPress.Smith,E.,Medin,D.,andRips,L.(1984).“APsychologicalApproachtoConcepts:CommentsonRey’s‘ConceptsandStereotypes’.”Cognition,17:265–74.Spelke,E.(1990).“PrinciplesofObjectPerception.”CognitiveScience,14:29–56.Stampe,D.(1977).“TowardsaCausalTheoryofLinguisticRepresentation.”MidwestStudiesinPhilosophy.Minneapolis:UniversityofMinnesotaPress.Wittgenstein,L.(1953/1958).PhilosophicalInvestigations,trans.E.Anscombe.3rdedi-tion.Oxford:Blackwell.Xu,F.andCarey,S.(1996).“Infants’Metaphysics:TheCaseofNumericalIdentity.”CognitivePsychology,30:111–53.213\nTheBlackwellGuidetoPhilosophyofMindJohnHeilEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter9MentalCausationJohnHeil9.1TheCartesianBackgroundDescartessetthetoneforthemoderndiscussionoftherelationofmindstobodies.AccordingtoDescartes,mindsandbodiesaredistinctkindsofsubstance.(Inthiscontext,asubstanceshouldbethoughtof,notasakindofstuff,some-thingthatmightstainyourshirtorsticktothebottomofyourshoe,forinstance,butasaparticularobjectorentity:thetreeoutsideyourwindow,apebble,theMoon,yourrightear.)Bodies,Descartesthought,arespatiallyextendedsub-stances,incapableoffeelingorthought;minds,incontrast,areunextended,thinking,feelingsubstances.Youmightbeledtosuchaviewbyconsideringmentalandphysicalcharacter-istics.Theseseemvastlydifferentonthefaceofit.Statesofmindexhibitqualitiesthatappeartofalloutsidethephysicalrealm:afeelingyouhavewhenyoubumpyourelbow,thesmellofpeat,thesoundofamosquitocirclingyourheadseemtodifferqualitativelyfromanythingbelongingtothephysicalworld.Thecausesoftheseexperiencesareperfectlyunexceptionalphysicaloccurrences.Thementaleffectsofthesecauses,however–theirappearances–seemtoincludequalitiesnotlocatableinthephysicalworld.Fortheirpart,physicalbodiesexhibitcharac-teristicsthatappeardecisivelynon-mental.Astonehasaparticularsize,shape,mass,anddefinitespatiallocation.Sensationsandthoughts,incontrast,appar-entlylackthesecharacteristics.Apaincanbeintense,butnotthreeincheslong;yourthoughtsofanimpendingholidaylackmass.Tobesure,wesaythatthoughtsoccurintheheadandthatapaininthetoeisinthetoe.Thissuggeststhatstatesofmindareatleastspatiallylocatable.Thesenseinwhichapainorathoughthasaspatiallocationapparentlydiffersfromthesenseinwhichaphysicalobjecthasaspatiallocation,however.Descarteswaswellawareofthephenomenonofphantompain:theapparentoccurrenceofpainsinamputatedlimbs.Thissuggeststhat,indescribingapainasoccurringinyour214\nMentalCausationtoe,whatyouarereallydescribingisasensationofaparticularkind:asensationasofapaininyourtoe;apain-in-the-toekindofsensation.Suchasensationmightoccur–andindeedsuchsensationsdooccur–inagentswhosetoeshavebeenamputated.Mentalandphysicalitemsappeartodifferinanotherrespectaswell.Yourthoughtsandfeelingsareprivate.Otherscanguessorinferwhatyouarethinkingandfeeling,butonlyyouhave“direct”accesstoyourthoughtsandfeelings.YouandIstandingsidebysidecanobservethesametreeorthesameperson.Icanobserveyourhavingathoughtorexperiencingapain.ButIcannot,asyouevidentlycan,encounteryourthoughtorpain.Myexperienceisnotofyourpainbutofitseffectsonyouandyourbehavior.ConsiderationsliketheseencourageustofollowDescartesandplacesensationandthoughtoutsidethephysicalworld.ForDescartes,thismeantthatmentalqualitiesmustbequalitiesofmentalsubstances,entitiesdistinctfromphysicalsubstances,themselvesentitiespossessingdistinctivecharacteristics.Whatweshouldregardasmentalpropertiesare,Descartescontended,modesofthought:waysofbeingathinkingsubstance.Incontrast,physicalpropertiesaremodesofexten-sion:waysofbeingextendedinoroccupyingspace.Onceweembracethispicture,thequestionarises:howcouldmentalandphysicalsubstancesinteractcausally?InalettertoDescartes,PrincessElizabethofBohemiaobservesthat“itwouldbeeasierformetoattributematterandextensiontothesoul,thantoattributetoanimmaterialbodythecapacitytomoveandbemovedbyabody”(Kenny1970:140).OnewaytoseethedifficultyfacingDescartesistonotethatcausalinteractionofmentalandphysicalsubstancesapparentlyobligesustoabandontheideathatthephysicalworldiscausallyautonomous.Physicstreatsthephysicalworldasaclosedsystem.Occurrencesinthissystemreflectonlyoccurrenceselsewhereinthesystem(perhapstogetherwith“boundaryconditions”).Supposetheseoccur-rencesareultimatelymotionsofelementaryparticles.Thesemotionsareaffectedonlybythemotionsofotherparticles.Ifweimagineanon-physicalentityinter-actingcausallywithaphysicalsystem,weshouldhavetocountenancemotionsof1particlesnotproducedbymotionsofotherparticles.Thisappealtothecausalautonomyofthephysicalrealmisnotintendedasanaprioriargumentagainstthepossibilityofcausallinksbetweenthementalandthephysical.Rather,itisareminderthattheprospectofcausalinteractionbetweenanon-physicalmindandapurelyphysicalworldwouldobligeustorethinkthecharacterofthefundamentalnaturallawsandbroadenournotionofwhatconstitutestheworldasawhole.Weshouldexpecttodiscoverparticlesbehavinginwaysthatcouldnotbeaccountedforsolelybyreferencetolawsgoverninginter-particlerelations.Thisneednotimplythepossibilityofnon-materialcausesviolatingnaturallaw.Naturallaws,unlikelegalstatutes,areinviolable.Aleafflutteringslowlytothegrounddoesnotviolatelawsofgravity.Rather,anexplana-tionoftheleaf’sbehaviorrequiresappealtocomplexfeaturesofasystemthatincludesthefallingleaf,theEarth,andtheswirlinggaseousatmospherethrough215\nJohnHeilwhichtheleafisfalling.Theintroductionofnon-physicalcauseswouldcomplicatethecausalpicture,notbycountenancingviolationsofphysicallaw,butbyintro-ducingheretoforeunanticipatedcausalfactors.TheworryexpressedbyPrincessElizabethreflectsaworrycommonamongDescartes’scontemporaries.Ifmindsarespatiallyunextendedentities,howcouldtheyaffectaspatiallyextendedworld?Howcouldanon-spatialthing,asitwere,getagriponaspatialthing?Whatisthenatureofthecausalmechanism?Toseetheforceoftheseworries,thinkofasimplecaseofmechanicalcausation:arotatinggearengagesasecondgearcausingthesecondgeartorotate.Inthiscasewehaveadistinctivecausalmechanism:wecanseehowthesecondgear’sturningisbroughtaboutbytherotationofthefirstgear.Nowimaginethefirstgear’sbeingreplacedbyanon-spatialentity.Howcouldsuchanentityengagethesecondgear?Ofcourse,rotatinggearsaffordmerelyoneexampleofaneasilyvisualizablecausalnexus.ThinkoftheactionofamagnetonironfilingsortheeffectsoftheMoononthetides.Inneithercasecanweobserveanythinglikeamechanicalconnectionbetweencauseandeffect,yetwedonotregardcasesofthiskindasworrisome.Thisisdue,inpart,tothefactthatsuchphenomenaaresofamiliar,andinparttoourhavingacceptedtheideathatobjectscanaffectoneanotheratadistancewhentheobjectsarecontainedwithinafield.Amagnetcreatesamagneticfield.Ironfilingsareaffectedbycharacteristicsofthisfield.TheEarthandMoonalterthecontoursofagravitationalfieldthatincludesthemboth,andbywayofthisfieldaffectobjectsinit.Perhapsmindsactonbodies,notbypushingthosebodiesaround,butbycreatingoraffectingthecontoursoffieldswhichinturnaffectthebehaviorofbodiesinthem.Youmightstillworrythatafieldhasadefinitelocation,butaCartesianmindisutterlynon-spatial.Howcouldsomethingthatisnothere–indeednotanywhere–bearresponsibilityforthecharacterofafieldpresentinadefinitespatialregion?This,however,istomisunderstandtheideathatmindsarenon-extended.Apointisnon-extended,yetpossessesadefinitespatiallocation.Itwouldseempossible,then,foranon-materialsubstance,utterlylackinginextension,toexistataspatiallocationormovefromplacetoplaceandtoaffectcontiguousspatialregions.Ifallthiswereso,however,weshouldhavetoincludementalsubstancesamongthefundamentalentitiesmakingupourworld.Thiswouldrequire,attheveryleast,supplementinglawswenowtaketogoverntheelementaryconstitu-ents.Alinkofthiskindbetweenthementalandthephysicalmightsuggestthatwearelosingthedistinctionbetweenthementalandthephysicalandineffectsubsumingthementalunderthephysical.Atthispoint,weshoulddowelltoremindourselvesofthevastgulfbetweenmentalandphysicalqualities.Itishardtofindaplaceforthequalitiesofconsci-ousexperiencealongsidethequalitiesofordinaryobjects.Thehardness,sphericity,andmassofabilliardballseemtobenothingatalllikethequalityofyourexperienceofaheadacheorthetasteofamango.Itisnotjustthatmentalandphysicalphenomenadifferqualitatively:thereareendlessqualitativedifferences216\nMentalCausationamongphysicalphenomena.Rather,thequalitiesofstatesofmindseemnottooverlapinanywayatallwithphysicalqualities.Itmaybepossibletounderstandhowthequalitiesofabilliardball,forin-stance,couldbegroundedinfeaturesofthebilliardball’sconstituentparticles.Butitisanothermatteraltogethertounderstandhowmentalqualitiesmightbegroundedinfeaturesoftheparticlesthatmakeusup(orforthatmattermakeupourbrains).InthiscaseweseemtobefacedwithwhatJosephLevine(1983)hascalledan“explanatorygap.”Evenifweacceptthefamiliarideathatmindsaresomehowdependentonbrains,wehavenoclearideaofthenatureofthisdependence.Themental–physicalrelationappearsutterlymysterious.9.2IntentionalityDifficultiesconcerningthecausalroleofmentalqualitiesmakeuponecompon-entoftheproblemofmentalcausation.Aseconddifficultyishardertomotiv-ate,andisbesttackledinstages.Thedifficultyinquestionstemsfromthefactthatmanystatesofmindexhibitrepresentationalcontent.(Philosopherscallsuchstatesofmindintentionalstates.)Whenyoustubyourtoe,youexperienceaqualitativelydistinctivekindofexperience.YoumayalsocometoformathoughtyoumightexpressinEnglishbysaying“I’vestubbedmytoe!”Thisthought,2unlikeyourfeelingofpain,isrepresentational.Letusbracketforthemomentincipientworriesaboutmentalqualities,andconsideraninfluentialattemptbyDonaldDavidsontocometotermswithintentionalstatesofmind(seeDavidson1970,1974).Davidson’saccountoftherelationthatmentaleventsbeartophysicaleventsisstandardlycharacterizedasatokenidentitytheory.Davidsonarguesthatalthoughmentalpropertiesortypesarenotreducibleto(thatis,analyzableintermsoforidentifiablewith)physical3types,everymentaltokenisidenticalwithsomephysicaltoken.Yourbeinginpainatmidnightis(letusimagine)identicalwithsomephysical(presumablyneurological)eventoccurringinyourbodyatmidnight,althoughthereisnoprospectoftranslatingtalkofpainintoneurologicaltalk.Davidsondoesnotappealtofamiliarargumentsforthe“multiplerealizability”ofmentaltypes,althoughtheseargumentsmightbetakentosupporthisposition.(Ishalldiscussmultiplerealizabilitypresently.)Physicallyindiscernibleagentsmustbementallyindiscernible,accordingtoDavidson(themental“supervenes”onthephysical),butthisdoesnotimplythatagentsinthesamestateofmindmustbephysicallyindiscernible.Youandanoctopusmaybothbeinpain,butyourphysicalcondi-tionisverydifferentfromthatoftheoctopus.Davidsonhopedtosolvetheproblemofmentalcausationbyappealingtotokenidentity.Ifevery(particular,token)mentaleventisidenticalwithsome(particular,token)physicalevent,andifphysicaleventsareunproblematicallycausesandeffects,thenmentaleventscanbecausesandeffectsaswell.Howcan217\nJohnHeilmentaleventsbeidenticalwithphysicaleventsifmentalpropertiesortypesarenotreducibletophysicalpropertiesortypes?Davidson’sideaisthataneventcountsasamentaleventifitfallsunderamentaldescription.Similarly,aneventisaphysicaleventjustincaseitfallsunderaphysicaldescription.Oneandthesameevent(anoccurrenceinyourbrain,forinstance)couldfallunderamentaldescription(“beingapain”)andsatisfyaphysicaldescription(“beinganeuro-logicaloccurrenceofkindN”).Theprinciplesweusetoascribestatesofmindtoagentsdifferimportantlyfromthoseweusetoascribeneurologicalstates,how-ever.Thismeansthatalthoughevery(true)ascriptionofastateofmindtoanagentholdsinvirtueofthatagent’sbeinginaparticularphysicalstate,thereisnowaytoreconstitutetalkofstatesofmindinneurologicalterms.Indeed,inapplyingmentaltermstoagents,weneedhavenoideawhatcomplexphysicalfeaturesofthoseagentsanswertothoseterms.Thisissoquitegenerally.WhenIcorrectlyascribeaheadachetoyou,Idosoonthebasisofyourbehavior:whatyousayanddo.Butwhatmakesmyascriptioncorrectisnotyourbehavior,butsomecomplexstateofyourbrainaboutwhichImaybeutterlyignorant.Davidson’scontentionthatmentaltermscannotbereducedtophysicaltermscanbeillustratedbymeansofananalogy.Wheneverabatterhitsintoadoubleplay,thedoubleplayisconstitutedbyasequenceofphysicalevents.Itdoesnotfollow,however,thatwecouldredefine“doubleplay”intermsofprecisesequencesofphysicalmotions.Thisissodespitethefactthat,ifaparticularphysicalsequenceconstitutesadoubleplay,anyphysicallyindiscerniblesequencewouldconstituteadoubleplayaswell.(Sobeingadoubleplay“supervenes”onphysicalsequences.)Davidson’sproposedsolutiontotheproblemofmentalcausation,althoughinfluential,hasbeenwidelyattacked.Ingeneral,theattackshavehadthefollow-ingform.Supposeweconcedetokenidentity:everymentaleventisidenticalwithsomephysicaleventorother.Supposeweconcede,aswell,thateverysuchphysicaleventiscausallyunproblematic.SupposeyourhavingaheadachetonightatmidnightisidenticalwithyourthenbeinginneurologicalstateN,andsupposeyourbeinginneurologicalstateNcausesaparticularbodilymotion(youreachforabottleofaspirin).Wecan,itseems,stillask:didyoureachfortheaspirininvirtueofbeinginpainorinvirtueofbeinginstateN?(Thequestionissome-timesputlikethis:didtheeventthatcausedacertainbodilymotiondosoquabeingapainorquabeingneurologicalstateN?)Consideraparallelcase.TheballhitbyMarkMcGwireforhis65thhomerunoftheseasonstrikesGusinthehead,causingaconcussion.Theball’sstrikingthGusisGus’sbeingstruckbyMcGwire’s65homerunball,buttheball’sbeingthMcGwire’s65homerunballisirrelevanttoitshavingthisphysicaleffect.(Onewaytoseethisistonotethatanyobjectwiththeball’smassandvelocitywouldhavehadpreciselythesameeffect.)Theworryisthatmentalstatescouldbelikethis.Mentaleventsmightfigureincausaltransactionsbutnotinvirtueofbeingmental(notquamental),onlyinvirtueoftheirphysicalcharacteristics–charac-teristicspickedoutbypurelyphysicaldescriptions.218\nMentalCausation9.2.1“Broad”statesofmindYoumightbesuspiciousofthisexample.Afterall,abaseball’sbeingonehitbyMarkMcGwireisnotanintrinsic(“builtin”)propertyoftheball,butafeatureoftheballitpossessesonlybyvirtueofstandinginaparticularrelationtosomethingelse(forstarters,itwashitbyMarkMcGwire).Anditishardtoseehowanyrelationalfeatureofanobjectcouldaffectthatobject’scausalcapacities.TheaspirintabletyoutakeforaheadachecouldbethemillionthtabletproducedinMaybytheBayerCompany,butthisfeatureofthetabletplaysnoroleintheoperationoftheaspirininyourbloodstream.Ingeneral,itwouldseemthatonlyanobject’sintrinsic–built-in–featurescouldaffectitscausalcapacities.Youcouldconcedethatanobject’srelationalpropertiesarecausallyirrelevanttowhatitdoesorcoulddo,butwonderwhatthishastodowithmentalcausation.Theanswer,accordingtomanyphilosophers:everything!Atraditionintwentieth-centuryphilosophyofmindextendingfromWittgensteinthroughPutnamandBurgerejectstheCartesianpictureofthemindasaself-containedentitythatundergoessensations,entertainsthoughts,andmanipulatesthebody.Sensations,perhaps,canbeunderstoodasstatesandprocessesintrinsictoagents.Intentionalstatesofmind,however,beliefs,desires,intentions,andthelike,areheldtoincorporateanineliminablerelationalcomponent.Thethesismightbeillustratedbyimaginingtwointrinsicallyindiscernibleagentssituatedindistinctenvironments.Oneofthese,Wayne,livesonEarth.WhenWayneentertainsthoughtshewouldexpressbyutteringsentencessuchas“Theglassisfullofwater,”histhoughtsconcernwater.Wayne’stwin,Dwayne,exactlyresemblesWayneintrinsically(WayneandDwayneare“molecularduplic-ates”).DwayneinhabitsaplanetphysicallyresemblingEarthdowntothelastdetail,withoneimportantexception.OnDwayne’splanet(whichDwaynecalls“Earth,”butweshallcall“TwinEarth”),thestuffinrivers,bathtubs,andicetrays,althoughcalled“water”isnotwateratall,butXYZ,averydifferentchemicalsubstancethatsuperficiallyresembleswater:XYZlooks,feels,tastes,andbehavesasordinarywaterdoesonEarth.WhenDwayneentertainsathoughthewouldexpressbyuttering“Theglassisfullofwater,”histhoughtsdonotconcernwater(waterafterallisH2O)butXYZ,twin-water(aclearcolorlessliquidwithadistinctivechemicalmake-up).Theguidingideahereisthatthecontentsofthoughtsdependnotmerelyonagents’intrinsicfeatures,butalso,andcrucially,ontheircontext.Ifitisessentialtoabelief,desire,orintentionthatithaveaparticularcontent(ifbeliefB1andbeliefB2differincontent,thenB1≠B2),thenbeliefs,desires,intentions–inten-tionalstatesgenerally–dependonagents’contexts.Supposethisisso.ReturningtoDavidson,imagineacaseinwhichanagent,A,isinagivenneurologicalstate,N,andthatNisidenticalwithabelief,B(thatis,byvirtueofbeinginneurologicalstateN,AcanbesaidtohaveB).Imagine,aswell,thatNcausessomebodilymotion,M.A’sbelief,B,isN,but,giventhat219\nJohnHeilNisBpartlyinvirtueofA’scontext,itishardtoseehowN’sbeingabelief(asopposedtobeingacertainkindofneurologicalstate)hasanybearingontheoccurrenceofM.NisA’sbelief,butthisissoonlybecause,inthiscontext,AcanbesaidtohavebeliefB.WhyshouldthispurelyextrinsicfactaboutNhaveanybearingatallonwhatNcauses?Aparticularphysicalobjectisadollarbill.Itsbeingadollarbilldependsonahostofbroadlycontextualfactors:thebillhasacertainkindofcausalhistory:itwasprintedbytheUSTreasuryDepartment.Thesecontextualfactors,althoughessentialtothebill’sbeingadollarbill,playnorolewhateverintheoperationofavendingmachineintowhichthebillisinserted.Aneventinvolvingaparticularobjectcausesthemachinetodispenseacandybar,andtheobjectinquestionisadollarbill.Buttheobject’sbeingadollarbillisirrelevanttotheoperationofthemachine.Theconclusionappearsinescapable.EvenifDavidsonisright,andeverymen-taleventisidenticalwithsomephysical(causallyunproblematic)event,itseemsnottofollowthateventshavephysicaleffectsinvirtueofbeingmental(quamental).Atleast,thisseemssoforeventsinvolvingintentionalityifwegrantthatintentionalcharacteriscontextual.Wheredoesthisleaveus?Wehaveuncoveredtwokindsofworryconcerningmentalcausation.Oneworryconcernsmentalqualities.Suchqualitiesseemnottoengagewithphysicalgoingson.Asecond,lessobvious,worryfocusesonintentionalstatesofmind–beliefs,desires,intentions,andthelike.Suchstatesofmindappeartohaveirreduciblycontextualorrelationalcomponentsthatdisqualifythemascandidatecausesofphysicaleffects(bodilymotionsofagents,forinstance).Ineithercase,theactionofmindonthephysicalworldishardtounderstand.Perhapsthisistoohasty,however.Wehaveyettolookatthemostinfluentialaccountofthemindtoday:functionalism.Functionalismpurportstoofferawaythroughthethicketofproblemsassociatedwithmental–physicalcausalinteraction.9.3FunctionalismFunctionalismhasmanysources,butasanexplicitconceptionofminditcanbe4tracedtoHilaryPutnam’s1967paper,“PsychologicalPredicates.”Functionalistsholdthatstatesofmindarefunctionalstatesofcreaturestowhomtheyareascribed.Theideaofafunctionalstateismosteasilyunderstoodbyreferencetothenotionofafunctionalcharacterization.Whatisanegg-beater?Anegg-beaterisadevicethefunctionofwhichistobeateggs.Egg-beaterscantakemanyforms.Anegg-beatermightbeawirewhisk,ahand-crankeddevicemadeofmetalorplastic,oragleamingsolidstateCuisinart.Thinkofeachofthesedevicesasbeinganegg-beaterdifferentlyembodiedor“realized.”Eachcountsasanegg-beaterbecauseeachperformsaparticularfunction:eachtakesunbeateneggsasinputsandyieldsasoutputsbeateneggs.220\nMentalCausationHowcouldstatesofmindbefunctionalstates?Thinkofpain.Yourbeinginpainisamatterofyourbeinginastatewithaparticularkindofcausalrole.Painsarecausedbytissuedamageormalfunction,forinstance,painsgiverisetovariousbodilyresponses,andpainshaveassortedmentaleffectsaswell.Whenyoustubyourtoeyougointoastateproducedbyacollisionbetweenyourtoeandsomebulkyobject,youreactbyrubbingyourtoe,andyouformthebeliefthatyourtoehurtsandadesiretotakeappropriatemedicinalaction.Lookedatthisway,yourbeinginpainisamatterofyourbeinginastateofakindwithcharacteristickindsofcauseandeffect.Thisstateis“realized”inyoubyaparticularphysiologicalstate.Whatisimportanttothatstate’srealizingyourpainisnotitsintrinsicmake-up,butthefactthatthestateoccupiestherightsortofcausalrole.Otherkindsofcreature–octopodes,forinstance,orreptiles–mighthaveutterlydifferentkindsofphysicalconstitution,yetbecapableofgoingintostateswithsimilarcausalprofiles:theyarebroughtaboutbytissuedamage,andtheyresultinaversiveresponses.Thesestatesaresaidtorealizepainincreaturesofthosekinds.Supposeweencounteredabeingfromaremotegalaxywithasilicon-based“biology.”Couldsuchacreaturefeelpain?Weshouldbeinclinedtosayso,functionalistsargue,insofaraswehaveevidencethatthealiencreatureshaveacapacityforgoingintostatesthatresembleourpainstatesintheircharacteristiccausesandeffects.If,whenthealienssufferbodilyinjury,theycryout,withdraw,andseekrelief,functionalistssensiblycontend,itwouldbechurlishtodenythatthealienssufferpainsolelyonthegroundsthattheirbodilymake-updiffersfromthatofterrestrialspecies.Wearethusledtotheviewthatpains(andstatesofmindgenerally)arefunctionalstates,statescharacterizablenotbytheirintrinsicmake-up,butbytheiroccupyinganappropriatecausalrole.Aviewofthissortappearstosolvetheproblemofmentalcausationinastroke.Ifstatesofmindarefunctionalstates,statesthatowetheirnaturetopatternsofphysicalcausesandeffects,itwouldseemthattherecanbenomental–physical“gap.”Statesofmind,afterall,arestatesofmindbyvirtueofwhatcausesthemandwhattheycause.9.3.1MultiplerealizabilityMattersarenotsoclear,however.Functionalistsdonotidentifystatesofmindwithphysicalstatesoftheirpossessors.Onthecontrary,functionalistsregardsuchstatesastherealizersofmentalstates.Painisrealizedinyoubyonekindofphysicalstate;butitisrealizedinothercreatures(andotherpossiblecreatures)bystatesofverydifferentsorts.Statesofmindareinthiswaymultiplyrealizable.YouareinmentalstateSbyvirtueofbeinginphysicalstateP1;anoctopusisinstateSbyvirtueofbeinginaverydifferentkindofphysicalstate,P2;andanAlphaCentaurianisinstateSbyvirtueofbeinginstateP3.P1,P2,andP3areverydifferentkindsofphysicalrealizerofS.Paincannotbeidentifiedwithanyoneofthesekindsofstatewithouttherebyexcludingtherest.Whatmakesapainapain,221\nJohnHeilt1t2H1realizescausesP1P2Figure9.1functionalistshold,isnotthephysicalcharacterofthestatethatrealizesthepain,butthefactthatthestatehastherightkindofcausalprofile.Thisisnottheplacetodiscusstheprosandconsoffunctionalism(seechapter1).Wecan,however,seehowtheproblemofmentalcausationarisesforafunctionalist.Afunctionalstateisnotidentifiablewiththephysicalstatesthatrealizeit.Functionalistsputthisbydescribingfunctionalstatesas“higher-level”states:statescreaturesareinbyvirtueofbeinginparticularlower-levelrealizing5states.Now,however,itishardtoseehowahigher-levelstatecouldhavealower-leveleffect.Inthecaseofyourbeinginpain(which,wearesupposing,isrealizedinyoubyyourbeinginsomephysicalstateP1),itlooksasthoughthephysicalrealizerofyourpain–P1–isresponsibleforanyphysicalresponsesassociatedwiththepain.Thepainitselfappearsmerelytofloataboveitsphysicalrealizer,andsotodonothing.Thesituationisillustratedinfigure9.1.(H1isahigher-levelstate–yourbeinginpain,forinstance–P1isthatstate’slower-levelrealizer,P2issomephysicaleffect–yourtakingaspirin,forinstance–andt1andt2reflectthepassageoftime.)Thisdifficultyattachesnotmerelytofunctionalismbuttoanyaccountofmentalitythatregardsstatesofmindas“higher-level”states,statesrealizedby,butdistinctfrom,lower-levelphysicalstates.Davidsonisnotafunctionalist,butmanyhavefounditnaturaltoreadhimasendorsingtheideathatstatesofmindarehigher-levelstates“supervenient”on,butnotreducibleto,lower-levelphys-icalstates.Indeed,philosophersofmanydifferentpersuasionshavebeenattractedtotheideathatmentalproperties,althoughinsomewaydependentonphysicalproperties,arenottherebyreducibletophysicalproperties.Theargumentformultiplerealizabilityseemstoestablishthatwhilestatesofmindarerealizedbyphysicalstates,mentalstatesarenotreducibletophysicalstates.Ifthementalisnotreducibletothephysical,however,wemusteitherassumethatmentalstatesaresomething“overandabove”theirphysicalrealizersor–moreradically–tosupposethattherearenomentalstatesatall,onlyphysicalstatesandgoings-on.Thelatterposition,eliminativism(seechapter2),strikesmostpeopleasanon-starter.Surely,itmightbeargued,ourhavingmentalstatesisadatumtobeexplained,notacandidateforelimination.Eliminativistscitecasesinwhichentit-ies(caloric,phlogiston,theetherarethreecommonlycitedexamples)postulatedbyscientifictheoriesweresubsequentlyabandoned.Withtheabandonmentofthetheoriescameabandonmentoftheentities.Suchcasesdonotfitwellourowndirectexperienceofmentality.Statesofmindandtheirpropertiesarenot222\nMentalCausationtheoreticalpositslikecaloric,phlogiston,ortheether,butobjectsofdirectexperience(orbetter:experiences).Whatsensecouldtherebeinthethoughtthattherearenopains,nofeelingsofgrieforhappiness,nothoughts,onlyneuro-logicalgoings-on?Eliminativismastostatesofmindappearsunpromising,adesperatemovethatattemptstoanswerthequestionofhowstatesofmindarerelatedtothephysicalworldbysubtractingstatesofmindandleavingonlythephysicalworld.Thealternatives–theideathatmindsarehigher-levelentitiesandreductionism–appeartohaveseriousproblemsoftheirown,however.Wehavenotedalreadythatreductionismisatoddswiththeevidentpossibilityofmultiplerealizability.Thatworryaside,manytheoristsconsiderreductionismtobeacovertformofeliminativism,aconceptionofmindthatstemsfromthebenightedthoughtthatscienceisthemeasureofallthings.Weshouldadmit(suchtheoristsargue)thatmentalstatesareaspeciesofirreduciblehigher-levelphenomenathatdeservetreatmentontheirownterms.Ifwehavedifficultyreconcilingtheroleofsuchphenomenainthecausalstructureofthephysicalworld,weshouldnotdoubtthephenomena,butabandonthe“scientistic”ideathatallgenuinecausalrelationsarereducibletobasicphysicalprocesses(see,forinstance,Post1991,Dupré1993,andPoland1994).9.4LevelsofRealityThesensethatwearefacedwiththreeunpalatableoptionsisdueperhapslesstothenatureofthingsandmoretothenetworkofconceptsthatdictatestheseoptions.Weshall(asmanagementtextsadvise)needto“thinkoutsidethebox”ifwearetomakeprogressinourunderstandingofthemind’splaceinnature.Consider,first,whatitistobearealistaboutstatesofmind.Ingeneral,youarearealistaboutaphenomenonoradomainofphenomenatotheextentthatyoubelievethatphenomenaofthesortinquestionexistindependentlyofyourthoughtsaboutthem.Mostofusarerealistsabouttables,chairs,mountains,andgalaxies,butnotaboutghosts,witches,orphlogiston.Somephilosophersendorserealismaboutvalue,regardingobjectsasvaluableornotquiteindependentlyofourvaluingthem.Othersarevalueanti-realists,preferringtothinkofanobject’svalueasdependinginsomewayonattitudesvaluerstakeuptowardit.Whatisrequiredforrealismaboutstatesofmind?Aswillbecomeevident,thewaythissomewhatobscurequestionisansweredcandramaticallyaffecttheconceptualframeworkwithinwhichquestionsaboutmentalcausationareposedandanswered.Philosophersaretrainedtothinkaboutrealisminaparticularway.Youarearealistaboutghosts,orquarks,ortheetherifyouthinkthatghosts,orquarks,ortheetherexistindependentlyofyourthoughtsaboutsuchthings.Philosophers,seekingprecision,prefertocharacterizerealismintermsofattitudesweevincetowardtermsorpredicates:223\nJohnHeil(P)YouarearealistaboutF’sifyouthinktheterm,“F,”designatesa6propertysharedbyeveryobjecttowhichit(truly)applies.Youarearealistaboutcolors,forinstance,ifyouthinkthat“isred”designatesapropertypossessedbysomeobjectsandsharedbyeveryobjecttowhich“isred”correctlyapplies.Note,first,that(P)goesbeyondtheinnocuousclaimthatwhen“isred”appliestoanobject,itdoessoinvirtueofsomepropertypossessedbythatobject.Thisclaimisinnocuousbecauseitdoesnotimply,as(P)does,thateveryobjecttowhich“isred”appliespossessesoneandthesameproperty–presum-ablythepropertyofbeingred.Toseethedifference,thinkoftworedballs.Oneballiscrimson,oneisscarlet.Bothballs(invirtueoftheirrespectivecolors)answertothepredicate“isred.”Butitiseasytodoubtthatthereissomeoneproperty,thatofbeingred,possessedbyeachballinadditiontothatball’sbeingsomeparticularshadeofred.Isitfalse,then,thatbothballsarered?Onlya7philosopherwouldsaythis;onlyaphilosopherwhoaccepted(P).Itseemsmorenaturaltosaythat“isred”applies–trulyandliterally–toobjectsbyvirtueofthoseobjectspossessinganyofa(possiblyopen-ended)rangeofcolors,thoseweclassifyasshadesofred.Mysuggestionisthatasimilarpointholdsforthekindsofmentaltermthoughttorangeovermultiplyrealizablepropertiesandstates.“Isinpain,”forinstance,mightbetakentoholdofdiversecreaturesinvirtueofthosecreatures8beingindistinctkindsofstate.Althoughthestatesdiffer,theyarepertinentlysimilar.Ifthefunctionalistsareright,thentheyaresimilarwithrespecttothekindsofeventthatevokethemandthekindsofeventtheythemselvesevoke.Ifthefunctionalistsarewrong,ifthereismoretobeinginpainthanbeinginaparticularkindoffunctionalstate,thenacreatureanswersto“isinpain”invirtueofbeinginastatethatisrelevantlysimilar–perhapssimilarqualitatively–tostatesofothercreaturestowhom“isinpain”applies.Theoperativewordhereis“similar.”Whenyouandanoctopusareinpain,youareinsimilarbutnotperfectlysimilarstates.Thereisnoneedtopostulate,asthefunctionalistsdo,somefurtherhigher-levelstatethatbothyouandtheoctopusarein,astatewithrespecttowhichyouandtheoctopusareperfectlysimilar,astateansweringdirectlytothepainpredicatebutdifferentlyrealizedinyouandtheoctopus.Ifthisisright,wecanatleastseeourwayaroundoneprominentpuzzleaboutmentalcausation:howcouldhigher-levelstatesorpropertieshavelower-leveleffects?Mysuggestionisthatthehigher-levelstatesandpropertiesarephilo-sophicalartifacts,traceabletoacovertacceptanceofsomethinglikeprinciple(P).Wecanturnourbacksonhigher-levelstatesandpropertieswithoutgivinguprealismaboutputativelyhigher-levelitems.“Isinpain”mightapplytrulyandliterallytoyou,toanoctopus,andtoanAlphaCentaurian.Thepainpredicateappliestoyou,theoctopus,andtheAlphaCentaurian,notbecauseyoushareasinglehigher-levelproperty,butbecauseyou,theoctopus,andtheAlphaCentaurianpossesssimilar,althoughnotpreciselysimilar,properties.224\nMentalCausationTheviewIamadvancingisontologicallyreductivebutnotreductionistinthesensethatmanyanti-reductionistsfindobjectionable.Idonotimagineforamomentthatwecouldtranslatetalkofstatesofmindintotalkofbiologicalstates,muchlessintotalkofstatesinvolvingelementaryparticles.Noristhepositionadvancedhereaformofeliminativism.Particularmentaltermscanapplytrulyandliterallytocreaturesinvirtueofthosecreatures’beinginanyofaverylargenumberofrelevantlysimilarphysicalstates.Ineachindividualcase,adefinitephysicalstateanswerstotheterm.Onsuchaview,allthatiseliminatedisanallegedhigher-levelstateorproperty,apurelyphilosophicalposit.Theabandonmentofthelevelspictureresolvesonecomponentoftheproblemofmentalcausation.Itdoesnotprovideanexhaustivesolution,however.Weareleftwithatleasttworesidualissues:theproblemof“broad”contextuallydeter-minedstatesofmind,andthequaliaproblem.9.5CausationandBroadStatesofMindRecalltheideathatintentionalstates–beliefs,desires,intentions,andthelike–9varywithcontext.Oneresultofsuchaconceptionofintentionalityisthat,differentlysituated,intrinsicallyindiscernibleagents(“molecularduplicates”)mightbeentertainingutterlydifferentthoughts.This,coupledwiththeideathatanobject’scausalpowersarewhollyafunctionofitsintrinsicmake-up,leadstotheideathatdifferencesinthecontentsofstatesofmindmakenocausaldifference.Totheextentthatthecontentofyourmentalstatesdependsonfactorsexternaltoyou(broadlyspeaking,onyourcontext),thecontentsofthosestates–whattheyconcern–couldmakenodifferencetoyourbehavior.Butsurelywhatyoubelieve,ordesireorintend,doesmakeadifferencetowhatyoudo.Again,wearefacedwithvariousoptions.Oneoptionistobitethebullet.Yourintrinsicpropertiesdeterminewhatyoudo(howyourespondtoincomingstimuli,forinstance).Thefactthat,invirtueofyourintrinsicstatetogetherwithyourcontext,itistrueofyouthatyouhaveparticularbeliefsanddesiresisbesidethepointcausally.Aviewofthiskindissimplyanextensionoftheeliminativistimpulsetoanewdomain.Ineithercaseitishardtoseeeliminativismasmuchmorethananadmissionofdefeat.Asecondoptionparallelsfunctionalistappealstolevelsofreality.Supposewereplacetalkofcausationwithtalkofexplanation.Weroutinelydescribeandexplainoneanother’sbehavior(and,significantly,thebehaviorofnon-humancreatures)byappealingtointentionalstatesofmind.Whydidyouvisitthepantry?Becauseyouwantedsomecheeseandbelievedtherewascheeseinthepantry.Yousubsequentlyformedanintentiontovisitthepantryand,onthebasisofthisintention,visitedthepantry.Weseemwillingtoacceptsuchexplanationsasfundamental.Ourconceptionofcausation,itcouldbeargued,isfoundedonourgraspofthiskindofexplanation.Ifthatisso,itisnogoodtryingtoundercut225\nJohnHeilasuccessfulexplanatorypracticebyappealstometaphysicalqualmsaboutthenatureofcausation.Causesarewhatweappealtoinsuccessfulexplanationsofthebehaviorofobjects.Itisaconceptualmistaketoimaginethatwecould10discoverthat,ingeneral,successfulexplanationslackedacausalgrounding.Ifyouregardtheproblemofmentalcausationasasignificantmetaphysicalproblem,thenyouwillnotbeattractedtopurportedsolutionsthatseektoreplacetalkofcausationwithtalkofexplanation.Todosowouldbetoputthecartbeforethehorse.Thismightappeartobegthequestionagainsttheideathatcausalexplanationisconceptuallypriortocausation.Wearefacedwithastand-offbetweentwopositions,eachofwhichreliesonpremisesthatentailthedenialoftheother.Howmightwebreakthedeadlock?Wearrivedatworriesaboutmentalcausationviadeeplyheldmetaphysicalcon-victionsaboutcausation.Fewtheoristswoulddisputethepracticeofexplainingagents’intelligentbehaviorbyreferencetoagents’statesofmind.Onequestionishowwemightaccommodatesuchapracticetootherpracticesthatseemnolessfundamental–includingthepracticeofexplainingthebehaviorofphysicalbodiesexclusivelybyreferencetotheintrinsicphysicalpropertiesofthosebodies.Fewphilosopherswouldbetemptedtodismissfine-grainedphysicalaccountsofacake’sfallingonthegroundsthatwealreadyhaveaperfectlyacceptableeverydayexplana-tionofthisevent:thecakefellbecauseLilianslammedtheovendoor.Thesameholds,Ibelieve,formentalcausation.True,therewouldbesomethingfishyaboutanyviewthatentailedtheutterfalsehoodofeverydayexplanationsofintelligentbehavior.Thisdoesnotimplythatsuchexplanationsconstitutebedrock,however.Bypersevering,wecanhopetofindanaccountofmentalcausationthataccom-modatessuchexplanations.Insodoing,wemayfinditconvenienttomodifythewaywethinkofintelligentbehaviorandtheintentionalstatesthatseemtodriveit.Whatmightwesay,then,about“broad”(contextuallydetermined)statesofmindandtheircausalefficacy?First,weshouldnotbequicktoabandontheideathatthecausalpowersofanobject(hencewhatitdoesorcoulddo)depend11whollyonitsintrinsicmake-up.Second,weshouldlookmorecarefullyatthecontextualmodelofintentionality.Perhapstheprojectivecharacterofstatesofmindcan,afterall,beaccountedforbyreferencetoagents’intrinsicmake-up.Aviewofthissortcouldallowthatwhatourthoughtsconcern–inthesenseofwhattheyincludereferenceto–dependsonthewaytheworldisindependentofagents.Butthisrelationalmatter,althoughitcanenterintodescriptionsofstatesofmind,neednotobligeustoimaginethatthosestatesofmindarethemselves12constitutedbyrelationsbetweenagentsandexternalfactors.9.6QualiaManyreaderswillbynowhavegrownimpatient.Thedeepworryaboutmentalcausation–indeedthedeepproblemforaccountsofmindsgenerally(whatDavid226\nMentalCausationChalmers(1996)calls“thehardproblem”)–istheworryabouthowtofitthequalitiesofconsciousexperiences,theso-calledqualia,intoacausallyself-containedphysicalworld.Letmefocusonjusttwodifficultiesposedbyqualiaforanyaccountofmentalcausation.ImaginethatyouaregazingatcherrytreesinbloomaroundtheJeffersonMemorial.Youhaveavividvisualexperienceyouwouldfinddifficulttoputintowords.Imaginenowthatascientistwhobelievesthatexperiencesaregoings-oninthebraincarefullyinspectsyourbrainwhileyouareundergoingthisexperi-ence.Thescientistobservesadullgraymass.Oncloserinspection(andwiththeaidofexpensiveinstruments),thescientistobservesfine-grainedneuralactivity:cellsfiring,chemicalreactionsalongaxons,andthelike.Theseactivitiesmightcorrespondtoyourexperience.Yourexperiencehasadefinitequalitativechar-acter,butthescientist’sobservationrevealsnothingofthis,onlyboringneurolo-gicalqualities.Wherearethesequalitiesofyourexperience,ifnotinyourbrain?Perhapstheylieoutsidethephysicalworld.13Thislineofreasoningbetraysaconfusionoverqualitiesofexperience.Yourexperienceisofpinkblossoms,whitemarble,andshimmeringwater.Pinkness,whiteness,andtheshimmeringcharacteryouperceivearequalitiesoftheobjectsyouperceive,notqualitiesofyourexperience.Whenyouperceiveaball,theball,butnotyourexperience,isspherical.Ifyouweretothink,then,thatonlookingintoyourbrainandobservingyourexperiencesascientistoughttoobservepink,white,shimmering,orsphericalitems,youwouldbeinerror.Ifneurologicalgoings-oninyourbrainareyourexperiences,thosegoings-onneednothavequalitiesascrib-abletotheobjectsyouareexperiencing.Thescientistlookingintoyourbrainexperi-encesoccurrencesofyourexperiences,letussuppose.Butthescientist’sexperiencesneednotresembleyours;yourexperiencesareofcherrytrees,thescientist’sexperiencesareofsomethingquitedifferent:yourexperiencesofcherrytrees.Thisisnottosaythatexperienceslackqualities(doesanythinglackqualities?),onlythatwemusttakecaretodistinguishqualitiesofexperiencesfromqualitiesofobjectsexperienced.Whenwedothis,isitsoclearthatthequalitiesofexperiencesdifferradicallyfromthequalitiesofbrains?Inconsideringthisques-tion,wetendtoforgetthat,indescribingbrainsasgray,mushy,andthelike,wearedescribingthewaybrainslooktous:brainsasweexperiencethem.Thereisnoreasontothinkthatthequalitiesofourexperiencesofbrainsoughttoresemblequalitiesofexperiencesofobjectsotherthanbrains(cherryblossoms,forin-stance).Consideredinthislight,thereisnoobviousproblemwiththethoughtthatyourexperienceofcherryblossomsandtheJeffersonMemorialisanoccur-renceinyourbrainandthatitsqualitiesarequalitiesofthatneurologicaloccur-rence.Thismeansthat,ifthereisnospecialproblemwiththeideathatthequalitativechangesinyourbrainaffectyourbehavior,thenconsciousqualitiesposenospecialproblemofmentalcausation.Manyphilosopherswilldisagree.Qualities,theywillsuppose,arecausallyinert.Whenabaseballcausesaconcussion,itisthecausalpowers(dispositionalities)ofthebaseball,notitsqualities,thatmatter.Aviewofthiskindisfoundedonthe227\nJohnHeilpracticeofdistinguishingdispositionalproperties(propertiesregardedasbestow-ingcausalpowersontheirpossessors)andcategoricalproperties(purelyqualit-14ativenon-causalproperties).Oneseldomnoticedproblemwithsuchaviewisthatitapparentlyfliesinthefaceofordinaryexperience.Youenjoyviewingcherryblossomsbecauseoftheirvisuallyperceivedqualities;youtakepleasureineatingicecreambecauseofitsgustatoryqualities.Anartistselectsaparticularmediuminwhichtoworkinpartbecauseofthequalitiesofthatmedium.Ineachofthesecases,itlooksforalltheworldasthoughagentsarerespondingtoqualitiesoftheobjectsinquestion.Howmightwereconcilethesedeliverancesofcommonsensewithaviewofpropertiesaccordingtowhichpropertiesmustbeeithercausalorqualitative?Perhapsweshouldrejectthedivisionofpropertiesintoexhaustiveandmutuallyexclusiveclasses:dispositionalandcategorical.Philosopherswhohavedonesohavetypicallyattemptedtoreduceoneclasstotheother–arguing,forinstance,thatpropertiesareexclusivelydispositional(Shoemaker1980).Amoreattractivepossibilityisthateveryproperty(andhereIhaveinmindnaturalpropertiesofconcreteobjects,notabstracta)issimultaneouslydispositionalandqualitative.Thisissometimesputbysayingthatpropertieshavedispositionalandqualitativeaspects.Talkofaspects,however,bringstomindpropertiesandleadstothethoughtthateverypropertymightbe(ormightbemadeupof)twoproperties,onequalitativeonedispositional.IprefertofollowC.B.Martinandseequalitiesanddispositionsasstrictlyidentical(seeMartin1997;MartinandHeil1999;Heil1998:ch.6).Considerthesphericityofaparticularball.Theball’ssphericityisaparticularqualitypossessedbytheballanditisinvirtueofthisqualitythattheballisdisposedtoroll.Thequalityanddispositiondonotmerelyco-vary,theyareoneandthesamepropertydifferentlyconsideredanddescribed.Pretendthatsomethinglikethisisright.Itwouldfollowthatthereisnoparticularmysteryastohowthequalitiesofexperiencebearoncausaltransac-tionsinthephysicalrealm–providingwearewillingtocountenancethepossibil-itythatconsciousexperiencesare,atbottom,physicalevents.This,ofcourse,isaweightyproviso,onemanytheoristswouldnotconcedewithoutafight.Ratherthanattemptadefenseoftheseideashere,Iproposetoapplythemtoaparticularpuzzlecaseandnotehowtheystackup.9.7ZombiesBeforewarned:thephilosophicalnotionofazombiediffersdramaticallyfromthe15popularconception.Thephilosopher’szombieisnotamemberofthe“undead,”requiringhumanbloodinordertoremain,ifnotexactlyalive,atleastundead.Aphilosopher’szombieisabeingindistinguishablefromordinarypeopleineveryrespect,saveone:zombieslackconsciousexperiences.Youmighthaveazombiecounterpart.Thiscounterpartwouldbehaveexactlyasyoudo,wouldexhibitall228\nMentalCausationyourpreferencesandprejudices.Indeed,noobservercoulddistinguishyoufromyourzombietwin.Wheninterrogated,yourzombietwinwilldenybeingazom-bie(andindeed,ifyoubelieveyouareconscious,yourtwinwillbelievethesame).Fromtimetotimeyourtwincomplainsofheadachesandreportsafond-nessforchocolate.Differencesbetweenyouandzombie-youareliterallyinvisibletoanyobserver,evenaneuroscientistarmedwithbrain-scanninginstruments.Zombieshavefiguredinthoughtexperimentsattackingfunctionalism.Theideaisstraightforward.Youandyourzombietwinarefunctionallyindiscernible.You,however,enjoyconsciousexperiences,whileyourtwin,despiteprotestationstothecontrary,doesnot:“allisdarkinside.”Ifsuchcasesareconceivable(theyneednotbereallypossible,“possibleintheactualworld,”onlylogicallypossibleinthewayyourleapingtallbuildingswithasingleboundislogicallypossible),itwouldseemthatfunctionalismfallsshortofprovidingacompleteaccountofthenatureofmind.Functionalismleavesoutacentralfeatureofminds:consciousness.Somefunctionalistsdenythepossibilityofzombiesonthegroundsthatanybeingwiththerightsortoffunctionalarchitecturetherebyhasamind.Thisamountstotheclaimthatthereisnothingmoretohavingamindthanhavingtherightkindoffunctionalarchitecture,however;thisistheverypointatissue.Inarecent,muchdiscussedbook,DavidChalmers(1996)takesadifferenttack.Chalmersdefendsfunctionalisminthefaceofthepossibilityofzombies.Hearguesthat,whilezombiesarelogicallypossible,theyarenaturallyimpossible.Lawsofnature,hethinks,tieconsciousnesstoparticularsortsoffunctionalstate.Intheactualworld,acreaturewithafunctionalarchitectureidenticaltoyourswouldbeconscious.Consciousness“emerges”fromfunctionalarchitecturebyvirtueoftheholdingofcertainirreduciblelawsofnature.Zombiesarepossible,16butonlyinaworldlackingthesebasiclaws.Chalmers’sviewisintendedtoworkconsciousnessintothephysicalworldwhileatthesametimeshowingwhyconsciousnessdoesnotfindaplaceinordinaryaccountsofphysicalprocesses.Consciousexperiencesaresalienttocon-sciousagents,butbecausethequalitiesofsuchexperiencearemerelyemergentby-productsoffunctionalsystems,theyhavenodirecteffectsonphysicalpro-cesses.(Hencethepossibilityofzombies.)Apositionofthiskindmandatesfundamentallawsofnaturerelatingpropertiesofconsciousexperiences–consciousqualities,qualia–tofunctionalpropertiesofcreaturestowhomtheexperiencesbelong.Chalmersacceptsthefunctionalistcontentionthatfunctionalpropertiesarehigher-levelproperties,propertiespos-sessedbyagentsbyvirtueofthoseagent’spossessionofsomelower-levelprop-erty(seesection9.3above).You,forinstance,possessaparticularfunctionalproperty,F,byvirtueofpossessinganeurologicalproperty,P1.AnoctopuspossessesFbyvirtueofpossessingaverydifferentneurologicalproperty,P2;andanAlphaCentaurianpossessesFbyvirtueofpossessinganaltogetherdifferentphysicalproperty,P3.ItisimportantforfunctionalismthattheclassofphysicalrealizersofF(P1,P2,P3,...)isopen-ended.Functionalpropertiesarenotinanysensereducibletophysicalproperties.(Toimaginethatfunctionalpropertiesare229\nJohnHeil–possiblyinfinite–disjunctionsofphysicalpropertiesistomakehashofthenotionofaproperty.)Thisleadstoaremarkablepictureofthebasiclawsofnature.Theemergenceofconsciouspropertiesfromfunctionalpropertiesre-quiresbasiclawsofnaturethattiesimplepropertiesofconsciousnesstoopen-endeddisjunctionsofphysicalproperties(therealizersofthefunctionalpropertiesfromwhichthelatteremerge).Suchlawswouldbeveryoddindeed,unlikeanythingthusfarencounteredinbasicphysics.Worriesofthiskindaside,letuslookattheimplicationsoftheconceptionofmentalitysketchedearlierforthepossibilityofzombies.Ihavearguedagainstthefunctionalistthesisthatstatesofmindarehigher-levelstates,anditscorollarythethesisthatmentalpropertiesarehigher-levelproperties–propertiespossessedbyagentsbyvirtueoftheirpossessionoflower-levelrealizingproperties.Thefunc-tionalistideathatmentalpropertiesaremultiplyrealizedisbettercapturedbytheideathatmentalpredicates(“isinpain,”forinstance)holdofdiverseagents,notinvirtueofthoseagents’sharingasinglemultiplyrealizedhigher-levelproperty,butbyvirtueofthoseagents’possessinganypropertyfromamongasprawling,somewhatunrulyfamilyofsimilarproperties.Mentalpredicatesare“projectable”(theyfigureinexplanationsofagents’behavior,forinstance)becausetheyholdofagentsinvirtueofthoseagents’possessingcausallysimilarproperties.Inadditiontobestowing“causalpowers”ontheirpossessors,however,thesesamepropertiescontributetotheagent’squalitativenature.Therearenottwokindsofproperty,qualitativeanddispositional,onlypropertiesthemselvesdiffer-entlyconsidered.Ifthis“identitytheory”ofpropertiesisright,thenitisflatlyimpossibletovarydispositionsandqualitiesindependently.Ifazombieisapreciseduplicateofyoudispositionally(henceafunctionalreplica),thezombiemustbeaqualitativeduplicateaswell.Thisresult,coupledwiththeideathatyourstatesofmindarephysicalstates(statesofyournervoussystem,forin-stance),yieldsthefurtherresultthatzombiesareimpossible–notjustimpossiblegivenlawsofnatureinourworld,butflatlyimpossible.Theremightbecreaturesfunctionallysimilartousinsomerespectsthatdifferintheirconsciousexperi-ences(orevenlackthemaltogether).Buttherecouldnotbecreatureswithallourphysicalpropertieswhodifferedfromusinthisregard.Youmaybeunimpressedbythisresult.Itdepends,afterall,oncertainsub-stantive,hencecontroversial,philosophicaltheses.Butattheveryleastitshouldservetounderminetheairofinevitabilitythatoftenaccompaniesdiscussionofthe“hardproblem”ofconsciousness.Therearestillproblems,tobesure,puzzlesremainingtobeanswered.Butphilosophycanillaffordtomakehardproblemsharder.9.8ConclusionCartesianworriesaboutmentalcausationstemfromthethoughtthatmindsarenon-physicalsubstances.Theproblemthenarises:howcansomething230\nMentalCausationnon-physicalhavephysicaleffects(forthatmatter,howcouldsomethingphysicalhavenon-physicaleffects)?Thisisthevenerablemind–bodyproblem,nowadaysreferredtoastheproblemofmentalcausation.PhilosophershavelargelygivenupCartesiandualism.Propertydualism–thethesisthattherearetwodistinctivekindsofproperty,mentalandphysical–iswidelyaccepted,however.TheresultisacollectionofproblemseverybitasresistanttosolutionasthoseissuingfromCartesianism.Ifaccountsofsensationandthoughtrequirethepostulationofmentalproperties,whatisthebearingofthesepropertiesonphysicalprocesses,inparticularonthosephysicalprocessesthatunderlieintelligentaction?Ihavesuggestedthattheframeofmindrequiredtoregardmentalpropertiesasspecialkindsofnon-physicalpropertyisinducedbycertainphilosophicalthesesweneednotaccept.Thesethesesareripeforreplace-ment.Insodoing,wecanseeourwaypastmanyoftheworriesthatdogthemind–bodydebate.Difficultiesundoubtedlyremain.Inphilosophywemustrestcontentwiththekindofprogressthatresultswhenwecanseeourwayaroundself-imposedbarriers.Soitisinthecaseofmentalcausation.NotesIamindebtedtoDavidsonCollegeforfundingaresearchleaveduring2000–1andtotheDepartmentofPhilosophy,MonashUniversityforitshospitalityandforsupportinganinvigoratingphilosophicalenvironment.MygreatestdebtistoC.B.Martin,themostontologicallyseriousoftheontologicallyserious.1Descartes(1596–1650),whodiedbeforeNewton(1642–1727)producedhismonu-mentalworkonphysics,heldthatthemindmightaffecttheparticles,notbyimpart-ingmotiontothem,butbyalteringtheirdirection.Inthiswaymotioninthephysicalsystemwasconserved.Newton’slawsrequiredconservationofmomentum,however,andthisisviolatedifchangesinthedirectioninwhichparticlesmovehasanon-materialsource.2Somephilosophers,hopingtoassimilatequalitiesofexperiencestorepresentations,regardsensoryexperiencesaspurelyrepresentational(see,e.g.,Harman1990;Dretske1995;Tye1995;Lycan1996).AlthoughIhavedoubtsaboutanysuchview,nothingIsayhererequiresthatitbeacceptedorrejected.Ifyoudoacceptit,thenworriesabouttheplaceofmentalqualitiesinaphysicalworldarereplacedbyworriesaboutthecausalsignificanceofrepresentationalstates.3Considerthebox:SaginawSaginaw.Howmanywordsdoesthisboxcontain?Well,youmightsaytheboxcontainstwooccurrencesorinstancesofoneword.Philo-sopherssaythattheboxincludestwotokensofasingletype.4Aristotleembracedaspeciesoffunctionalism,andfunctionalexplanationhashadalonghistoryinthebiologicalandsocialsciences(seeWinch1958).Inlightofwhatfollows,itisperhapsworthnotingthatPutnamsubsequently(and,asIshallsuggest,inappropriately!)retitledthepaper“TheNatureofMentalStates.”5SeeBlock(1980)foradiscussionoftwospeciesoffunctionalism:thefunctionalidentitytheory(whatIhavebeencallingfunctionalism),andthefunctionalspecifier231\nJohnHeiltheory(aviewassociatedwithLewis1966;Armstrong1968;andSmart1959).Onthelatterview,yourbeinginpainisyourbeinginaparticularphysicalstate.Ontheformerview,yourphysicalstateisthoughtmerelytobetherealizerofpaininyou(forreasonsmentionedearlier).Thisissometimesputbysayingthatastateofmindisafunctionalrole,nottheoccupantoftherole.Today,mostfunctionalistsembracethefunctionalidentityconceptionoffunctionalism,soIshallomitdiscussionofthefunctionalspecifierversionhere.Functionalspecifiersavoidproblemsstemmingfromtheidentificationoffunctionalstateswithhigher-levelstates,butatthecostofsacrificingthecorefunctionalistthesisthatfunctionalstatesaremultiplyrealizable–orsofunctionalidentitytheoristsinsist.6See,forinstance,Boghossian(1990:161).Thisformulationholdsfor“characterizingpredicates”(roughly,terms,suchas“isred”or“iswise,”usedtoascribeproperties).Aslightlydifferentformulationisrequiredfor“substantialpredicates”(those,suchas“isahorse”or“isgold”usedtoclassifyobjectsaskinds).Whetherornotthedistinctionisadeeponeisamatterofcontroversy.Ishallignoreithereinordertokeepthediscussionsimple.7Somephilosophersareanti-realistsaboutcolorsquitegenerally.Iusecolorheremerelyasastalkinghorse,however.Ifyoudoubtthecolors,substitutesomeotherproperty–beingtriangular,forinstance,orhavingmass.8Thissuggestioncallstomindthefunctionalspecifierversionoffunctionalism(seenote5).9Ifsuchaconceptionofintentionalstatesofmindstillseemsodd,thinkofacom-ponentinapainting–asmilingface.Imaginethisfacetransferredfromonepictorialcontexttoanother.Inonepainting,thefaceappearsinthemidstofajoyousweddingscene.Inanotherpainting,thefacebelongstoasoldierinaconcentrationcamp.Contextaffectsthesignificanceoftheexpressionontheface,despitetherebeingnointrinsicdifferencesinthefacesthemselves.10ThisismyreadingofBaker(1993)and,perhaps,Burge(1993).11Somephilosophers(e.g.Teller1986)havearguedthatcertainkindsofquantumstateviolatethisprinciple.Evenifcorrect,itishardtoseehowthiscouldhelpresolvethepuzzleposedby“broad”statesofmind.12Thisisahugeissue,notonetobeaddressedinafewparagraphs.TheinterestedreaderisreferredtoMartinandHeil(1998)andHeil(1998:115–19,148–58)foramoredetaileddiscussion.Itisworthnotingherethatcausesareroutinelydescribedbyreferencetotheirextrinsicfeatures:ifmyflippingthelightswitchresultsintheroom’sbeingilluminated,myaction–flippingtheswitch–canbedescribedasmyilluminatingtheroom,andwecansaythatthisactionfrightenedaburglar.13TheconfusionisdiscussedbySmart(1959)andbyPlace(1956),whodubsit“thephenomenologicalfallacy.”(Forfurtherdiscussion,seeHeil1998:78–81,206–9.)14Sometheoristsarguethatdispositionalpropertiesaregroundedincategoricalpropert-ies,othersthatallpropertiesaredispositional.Foradiscussionofthepossibilities,seeMumford(1998);seealsoHeil(1998:ch.6).15ZombieswerefirstusedasaphilosophicalexamplebyRobertKirk(1974).Morerecently,theyhavebeendiscussedatlengthbyDavidChalmers(1996).16Oneconsequenceofthisviewisthatthelawsunderlyingconsciousnessmustbebasicinthestrongsensethattheyarenotderivablefromlawsgoverningthebasicparticlesandforces.232\nMentalCausationReferencesArmstrong,D.M.(1968).AMaterialistTheoryoftheMind.London:RoutledgeandKeganPaul.Baker,LynneRudder(1993).“MetaphysicsandMentalCausation.”InHeilandMele(1993):75–95.Block,Ned(1980).“WhatisFunctionalism?”InReadingsinPhilosophyofPsychology,vol.I.Cambridge,MA:HarvardUniversityPress:171–84.Boghossian,Paul(1990).“TheStatusofContent.”PhilosophicalReview,99:157–84.Burge,Tyler(1993).“Mind–BodyCausationandExplanatoryPractice.”InHeilandMele(1993):97–120.Capitan,W.H.andMerrill,D.D.(eds.)(1967).Art,Mind,andReligion.Pittsburgh:UniversityofPittsburghPress.Chalmers,David(1996).TheConsciousMind:InSearchofaFundamentalTheory.NewYork:OxfordUniversityPress.Davidson,D.(1970).“MentalEvents.”InL.FosterandJ.W.Swanson(eds.),ExperienceandTheory.Amherst,MA:UniversityofMassachusettsPress:79–101.ReprintedinDavidson(1980).——(1974).“PsychologyasPhilosophy.”InS.C.Brown(ed.),PhilosophyofPsy-chology.NewYork:BarnesandNobleBooks:41–52.ReprintedinDavidson(1980):231–9.——(1980).EssaysonActionsandEvents.Oxford:ClarendonPress.Dretske,Fred(1995).NaturalizingtheMind.Cambridge,MA:MITPress.Dupré,John(1993).TheDisorderofThings:MetaphysicalFoundationsoftheDisunityofScience.Cambridge,MA:HarvardUniversityPress.Harman,Gilbert(1990).“TheIntrinsicQualityofExperience.”PhilosophicalPerspectives,4:31–52.Heil,John(1998).PhilosophyofMind:AContemporaryIntroduction.London:Routledge.Heil,JohnandMele,Alfred(eds.)(1993).MentalCausation.Oxford:ClarendonPress.Kenny,A.,trans.anded.(1970).Descartes:PhilosophicalLetters.Oxford:ClarendonPress.Kirk,Robert(1974).“Zombiesvs.Materialists.”ProceedingsoftheAristotelianSociety,Supplementaryvol.48:135–52.Levine,Joseph(1983).“MaterialismandQualia:TheExplanatoryGap.”PacificPhilo-sophicalQuarterly,64:354–61.Lewis,David(1966).“AnArgumentfortheIdentityTheory.”JournalofPhilosophy,63:17–25.ReprintedinPhilosophicalPapers,vol.1.NewYork:OxfordUniversityPress(1983):99–107.Lycan,W.G.(1996).ConsciousnessandExperience.Cambridge,MA:MITPress.Martin,C.B.(1997).“OntheNeedforProperties:TheRoadtoPythagoreanismandBack.”Synthese,112:193–231.Martin,C.B.andHeil,John(1998).“RulesandPowers.”PhilosophicalPerspectives,12:283–312.——(1999).“TheOntologicalTurn.”MidwestStudiesinPhilosophy,23:34–60.Mumford,Stephen(1998).Dispositions.Oxford:ClarendonPress.Place,U.T.(1956).“IsConsciousnessABrainProcess?”TheBritishJournalofPsychology,47:44–50.233\nJohnHeilPoland,Jeffrey(1994).Physicalism:ThePhilosophicalFoundations.Oxford:ClarendonPress.Post,JohnF.(1991).Metaphysics:AContemporaryIntroduction.NewYork:ParagonHouse.Putnam,Hilary(1967).“PsychologicalPredicates.”InW.H.CapitanandD.D.Merrill(eds.),Art,Mind,andReligion.Pittsburgh:UniversityofPittsburghPress:37–48.Reprintedas“TheNatureofMentalStates,”inPutnam’sMind,Language,andReality:PhilosophicalPapers,vol.2.Cambridge:CambridgeUniversityPress(1975):429–40;andinNedBlock(ed.),ReadingsinPhilosophyofPsychology,vol.I.Cambridge,MA:HarvardUniversityPress(1980):223–31.Shoemaker,Sydney(1980).“CausalityandProperties.”InPetervanInwagen(ed.),TimeandCause.Dordrecht:ReidelPublishingCo.:109–35.ReprintedinIdentity,Cause,andMind:PhilosophicalEssays.Cambridge:CambridgeUniversityPress(1984):206–33.Smart,J.J.C.(1959).“SensationsandBrainProcesses.”PhilosophicalReview,68:141–56.Teller,Paul(1986).“RelationalHolismandQuantumMechanics.”BritishJournalforPhilosophyofScience,37:71–81.Tye,M.(1995).TenProblemsofConsciousness:ARepresentationalTheoryofthePhenom-enalMind.Cambridge:MITPress.Winch,Peter(1958).TheIdeaofaSocialScienceanditsRelationtoPhilosophy.London:RoutledgeandKeganPaul.234\nTheBlackwellGuidetoPhilosophyofMindFolkPsychologyEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter10FolkPsychologyStephenP.StichandShaunNicholsDiscussionsanddebatesaboutcommon-sensepsychology(or“folkpsychology,”asitisoftencalled)havebeencenterstageincontemporaryphilosophyofmind.Therehavebeenheateddisagreementsbothaboutwhatfolkpsychologyisandabouthowitisrelatedtothescientificunderstandingofthemind/brainthatisemerginginpsychologyandtheneurosciences.Inthischapterwewillbeginbyexplainingwhyfolkpsychologyplayssuchanimportantroleinthephilosophyofmind.Doingthatwillrequireaquicklookatabitofthehistoryofphilosophicaldiscussionsaboutthemind.Wewillthenturnourattentiontothelivelycontem-porarydiscussionsaimedatclarifyingthephilosophicalrolethatfolkpsychologyisexpectedtoplayandatusingfindingsinthecognitivesciencestogetaclearerunderstandingoftheexactnatureoffolkpsychology.10.1WhyDoesFolkPsychologyPlayanImportantRoleinthePhilosophyofMind?Toappreciatephilosophers’fascinationwithfolkpsychology,itwillbeusefultobeginwithabriefreminderaboutthetwomostimportantquestionsinthephilosophyofmind,andtheproblemsengenderedbywhatwasforcenturiesthemostinfluentialanswertooneofthosequestions.Thequestionsarethemind–bodyproblem,whichaskshowmentalphenomenaarerelatedtophysicalphe-nomena,andtheproblemofotherminds,whichaskshowwecanknowaboutthementalstatesofotherpeople.OnDescartes’sproposedsolutiontothemind–bodyproblem,therearetwoquitedifferentsortsofsubstanceintheuniverse:physicalsubstance,whichislocatedinspaceandtime,andmentalsubstance,whichislocatedintimebutnotinspace.Mentalphenomena,accordingtoDescartes,areeventsorstatesoccurringinamentalsubstance,whilephysicalphenomenaareeventsorstatesoccurringinaphysicalsubstance.Descartesinsistedthatthere235\nStephenP.StichandShaunNicholsistwo-waycausalinteractionbetweenthementalandthephysical,thoughmanyphilosophersfinditpuzzlinghowthetwocouldinteractifoneisinspaceandtheotherisn’t.AnotherproblemwiththeCartesianviewisthatitseemstomaketheothermindsproblemquiteintractable.If,asDescartesbelieved,Iamtheonlypersonwhocanexperiencemymentalstates,thenthereseemstobenowayforyoutoruleoutthehypothesisthatIamamindlesszombie–aphysicalbodythatmerelybehavesasthoughitwascausallylinkedtoamind.Inthemiddleofthetwentiethcenturytheverificationistaccountofmeaninghadamajorimpactonphilosophicalthought.Accordingtotheverificationists,themeaningofanempiricalclaimiscloselylinkedtotheobservationsthatwouldverifytheclaim.Influencedbyverificationism,philosophicalbehavioristsarguedthattheCartesianaccountofthemindasthe“ghostinthemachine”(touseRyle’s(1949)memorableimage)wasprofoundlymistaken.Ifordinarymentalstatetermssuchas“belief,”“desire,”and“pain”aretobemeaningful,theyargued,theycan’trefertounobservableeventstakingplaceinsideaperson(or,worsestill,notlocatedinspaceatall).Rather,themeaningofsentencesinvokingthesetermsmustbeanalyzedintermsofconditionalsentencesspecifyinghowsomeonewouldbehaveundervariouscircumstances.So,forexample,aphilo-sophicalbehavioristmightsuggestthatthemeaningof(1)Johnbelievesthatsnowiswhitecouldbecapturedbysomethinglikethefollowing:(2)IfyouaskJohn,“Issnowwhite?”hewillrespondaffirmatively.Perhapsthemostseriousdifficultyforphilosophicalbehavioristswasthattheirmeaninganalysestypicallyturnedouttobeeitherobviouslymistakenorcircular–invokingonementaltermintheanalysisofanother.So,forexample,contraryto(2),eventhoughJohnbelievesthatsnowiswhite,hemaynotrespondaffirmativelyunlessheispayingattention,wantstoletyouknowwhathethinks,believesthatthiscanbedonebyrespondingaffirmatively,etc.Whilephilosophicalbehavioristsweregraduallybecomingconvincedthatthereisnowayaroundthiscircularityproblem,averysimilarproblemwasconfrontingphilosophersseekingverificationistaccountsofthemeaningofscientificterms.Verificationismrequiresthatthemeaningofatheoreticaltermmustbespecifiableintermsofobservables.Butwhenphilosophersactuallytriedtoprovidesuchdefinitions,theyalwaysseemedtorequireadditionaltheoreticalterms(Hempel1964).Thereactiontothisprobleminthephilosophyofsciencewastoexploreaquitedifferentaccountofhowtheoreticaltermsgettheirmeaning.Ratherthanbeingdefinedexclusivelyintermsofobservables,thisnewaccountproposed,aclusteroftheoreticaltermsmightgettheirmeaningcollectivelybybeingembed-dedwithinanempiricaltheory.Themeaningofanygiventheoreticaltermliesinitstheory-specifiedinterconnectionswithotherterms,bothobservationaland236\nFolkPsychologytheoretical.PerhapsthemostinfluentialstatementofthisviewistobefoundintheworkofDavidLewis(1970,1972).AccordingtoLewis,themeaningoftheoret-icaltermsisgivenbywhathecallsa“functionaldefinition.”Theoreticalentitiesare“definedastheoccupantsofthecausalrolesspecifiedbythetheory...;astheentities,whateverthosemaybe,thatbearcertaincausalrelationstooneanotherandtothereferentsoftheO[bservational]-terms”(1972:211;firstandlastemphasesadded).BuildingonanideafirstsuggestedbyWilfridSellars(1956),Lewiswentontoproposethatordinarytermsformentalorpsychologicalstatescouldgettheirmeaninginanentirelyanalogousway.Ifwe“thinkofcommonsensepsychologyasaterm-introducingscientifictheory,thoughoneinventedbeforetherewasanysuchinstitutionasprofessionalscience,”thenthe“functionaldefinition”accountofthemeaningoftheoreticaltermsinsciencecanbeappliedstraightfor-wardlytothementalstatetermsusedincommon-sensepsychology(Lewis1972:212).Andthis,Lewisproposed,istherightwaytothinkaboutcommon-sensepsychology:Imagineourancestorsfirstspeakingonlyofexternalthings,stimuli,andresponses...untilsomegeniusinventedthetheoryofmentalstates,withitsnewlyintroducedT[heoretical]terms,toexplaintheregularitiesamongstimuliandresponses.Butthatdidnothappen.Ourcommonsensepsychologywasneveranewlyinventedterm-introducingscientifictheory–notevenofprehistoricfolk-science.Thestorythatmentaltermswereintroducedastheoreticaltermsisamyth.Itis,infact,Sellars’myth....Andthoughitisamyth,itmaybeagoodmythorabadone.Itisagoodmythifournamesofmentalstatesdoinfactmeanjustwhattheywouldmeanifthemythweretrue.Iadopttheworkinghypothesisthatitisagoodmyth.(Ibid.:212–13)1InthethreedecadessinceLewisandothersdevelopedthisaccount,ithasbecomethemostwidelyacceptedviewaboutthemeaningofmentalstateterms.Sincetheaccountmaintainsthatthemeaningsofmentalstatetermsare2givenbyfunctionaldefinitions,theviewisoftenknownasfunctionalism.Wecannowseeonereasonwhyphilosophersofmindhavebeenconcernedtounder-standtheexactnatureofcommon-sense(orfolk)psychology.Accordingtofunctionalism,folkpsychologyisthetheorythatgivesordinarymentalstatetermstheirmeaning.Asecondreasonforphilosophers’preoccupationwithfolkpsychologycanbeexplainedmorequickly.Thecrucialpointisthat,accordingtoaccountssuchasLewis’s,folkpsychologyisanempiricaltheorywhichissupposedtoexplain“theregularitybetweenstimuliandresponses”tobefoundinhuman(andperhapsanimal)behavior.And,ofcourse,ifcommon-sensepsychologyisanempiricaltheory,itispossiblethat,likeanyempiricaltheory,itmightturnouttobemistaken.Wemightdiscoverthatthestatesandprocessesinterveningbetweenstimuliandresponsesarenotwelldescribedbythefolktheorythatfixesthemeaningofmentalstateterms.Thepossibilitythatcommon-sensepsychology237\nStephenP.StichandShaunNicholsmightturnouttobemistakenisgrantedbyjustabouteveryonewhotakesfunctionalismseriously.However,forthelastseveraldecadesanumberofprom-inentphilosophersofmindhavebeenarguingthatthisismorethanamerepossibility.Rather,theymaintain,agrowingbodyoftheoryandempiricalfindingsinthecognitiveandneurosciencesstronglysuggestthatcommon-sensepsychologyismistaken,andnotjustonsmallpoints.AsPaulChurchland,anenthusiasticsupporterofthisview,putsit:FP[folkpsychology]suffersexplanatoryfailuresonanepicscale...ithasbeenstagnantforatleasttwenty-fivecenturies,and...itscategoriesappear(sofar)tobeincommensurablewithororthogonaltothecategoriesofthebackgroundphysicalscienceswhoselongtermclaimtoexplainhumanbehaviorseemsundeniable.Anytheorythatmeetsthisdescriptionmustbeallowedaseriouscandidateforoutrightelimination.(1981:212)Churchlanddoesnotstopatdiscarding(or“eliminating”)folkpsychologicaltheory.Heandother“eliminativists”havealsosuggestedthatbecausefolkpsy-chologyissuchaseriouslydefectivetheory,weshouldalsoconcludethatthetheoreticaltermsembeddedinfolkpsychologydon’treallyrefertoanything.Beliefs,desires,andotherpositsoffolkpsychology,theyargue,areentirelycomparabletophlogiston,theether,andotherpositsofempiricaltheoriesthatturnedouttobeseriouslymistaken;likephlogiston,theether,andtherest,theydonotexist.Obviously,theseareenormouslyprovocativeclaims.Debatingtheirplausibilityhasbeenhighontheagendaofphilosophersofmindeversincethey3werefirstsuggested.Sincetheeliminativists’centralthesisisthatfolkpsychologyisamassivelymistakentheory,philosophersofmindconcernedtoevaluatethatthesiswillobviouslyneedaclearandaccurateaccountofwhatfolkpsychologyisandwhatitclaims.10.2WhatisFolkPsychology?TwoPossibleAnswersFunctionalists,aswehaveseen,maintainthatthemeaningofordinarymentalstatetermsisdeterminedbytheroletheyplayinacommon-sensepsychologicaltheory.Butwhat,exactly,isthistheory?Inthephilosophicalandcognitive4scienceliteraturetherearetwoquitedifferentapproachestothisquestion.ForLewis,andformanyofthosewhohavefollowedhislead,common-senseorfolkpsychologyiscloselytiedtotheclaimsaboutmentalstatesthatalmosteveryonewouldagreewithandtaketobeobvious.Collectalltheplatitudesyoucanthinkofregardingthecausalrelationsofmentalstates,sensorystimuli,andmotorresponses....Addalsotheplatitudestotheeffectthatonementalstatefallsunderanother–“toothacheisakindofpain”andthe238\nFolkPsychologylike.Perhapsthereareplatitudesofotherformsaswell.Includeonlyplatitudesthatarecommonknowledgeamongus–everyoneknowsthem,everyoneknowsthateveryoneelseknowsthem,andsoon.Forthemeaningsofourwordsarecommonknowledge,andIamgoingtoclaimthatnamesofmentalstatesderivetheirmeaningfromtheseplatitudes.(1972:212;emphasisadded)So,onthisapproach,folkpsychologyisjustacollectionofplatitudes,orperhaps,sincethatsetofplatitudesisboundtobelargeandungainly,wemightthinkoffolkpsychologyasasetofgeneralizationsthatsystematizestheplatitudesinaperspicuousway.Asystematizationofthatsortmightalsomakeitmorenaturaltodescribefolkpsychologyasatheory.We’llcallthistheplatitudeaccountoffolkpsychology.Thesecondapproachtoansweringthequestionfocusesonaclusterofskillsthathavebeenofconsiderableinteresttobothphilosophersandpsychologists.Inmanycasespeopleareremarkablygoodatpredictingthebehaviorofotherpeople.Askedtopredictwhatamotoristwilldoassheapproachestheredlight,almosteveryonesaysthatshewillstop,andfortunatelyourpredictionsareusuallycorrect.Wearealsooftenremarkablygoodatattributingmental5statestootherpeople–atsayingwhattheyperceive,think,believe,want,fear,andsoon,andatpredictingfuturementalstatesandexplainingbehaviorinterms6ofpastmentalstates.Inrecentdiscussions,thewhimsicallabelmindreadinghasoftenbeenusedforthisclusterofskills,andsincethemid-1980sdevelopmentalandcognitivepsychologistshavegeneratedalargeliteratureaimedatexploringtheemergenceofmindreadingandexplainingthecognitivemechanismsthatunderlieit.Themostwidelyacceptedviewaboutthecognitivemechanismsunderlyingmindreading(anduntilthemid-1980stheonlyview)isthatpeoplehavearichbodyofmentallyrepresentedinformationaboutthemind,andthatthisinforma-tionplaysacentralroleinguidingthementalmechanismsthatgenerateourattributions,predictions,andexplanations.Someofthepsychologistswhodefendthisviewmaintainthattheinformationexploitedinmindreadinghasmuchthesamestructureasascientifictheory,andthatitisacquired,stored,andusedinmuchthesamewaythatothercommon-senseandscientifictheoriesare.Thesepsychologistsoftenrefertotheirviewasthetheorytheory(GopnikandWellman1994;GopnikandMeltzoff1997).Othersarguethatmuchoftheinformationutilizedinmindreadingisinnateandisstoredinmental“modules”whereitcanonlyinteractinverylimitedwayswiththeinformationstoredinothercompon-entsofthemind(SchollandLeslie1999).Sincemodularitytheoristsandtheorytheoristsagreethatmindreadingdependsonarichbodyofinformationabouthowthemindworks,we’llusetheterminformation-richtheoriesasalabelforbothofthem.Thesetheoriessuggestanotherwaytospecifythetheorythat(iffunctionalistsareright)fixesthemeaningofmentalstateterms–itisthetheory(orbodyofinformation)thatunderliesmindreading.We’llcallthisthemindreadingaccountoffolkpsychology.239\nStephenP.StichandShaunNicholsLet’sask,now,howtheplatitudeaccountoffolkpsychologyandthemindreadingaccountarerelated.Howisthementallyrepresentedinformationaboutthemindpositedbyinformation-richtheoriesofmindreadingrelatedtothecollectionofplatitudeswhich,accordingtoLewis,determinesthemeaningofmentalstateterms?Onepossibilityisthattheplatitudes(orsomesystematizationofthem)isnearenoughidenticalwiththeinformationthatguidesmindreading–thatmindreadinginvokeslittleornoinformationaboutthemindbeyondthecommon-senseinformationthateveryonecanreadilyagreeto.Ifthisweretrue,thentheplatitudeaccountoffolkpsychologyandthemindreadingaccountwouldconverge.But,alongwithmostcognitivescientistswhohavestudiedmindreading,webelievethatthisconvergenceisveryunlikely.Onereasonforourskepticismisthecomparisonwithothercomplexskillsthatcognitivescientistshaveexplored.Injustabouteverycase,fromfacerecognition(Young1998)todecision-making(Gigerenzeretal.1999)tocommon-sensephysics(McCloskey1983;Hayes1985),ithasbeenfoundthatthemindusesinformationandprinciplesthataresimplynotaccessibletointrospection.IntheseareasourmindsuseagreatdealofinformationthatpeoplecannotrecognizeorassenttointhewaythatoneissupposedtorecognizeandassenttoLewisianplatitudes.Asecondreasonforourskepticismisthatinmanymindreadingtaskspeopleappeartoattributementalstatesonthebasisofcuesthattheyarenotawaretheyareusing.Forexample,Ekmanhasshownthatthereisawiderangeof“deceptioncues”thatleadustobelievethatatargetdoesnotbelievewhatheissaying.Theseinclude“achangeintheexpressionontheface,amovementofthebody,aninflectiontothevoice,aswallowinginthethroat,averydeeporshallowbreath,longpausesbetweenwords,aslipofthetongue,amicrofacialexpression,agesturalslip”(1985:43).Inmostcases,peoplearequiteunawareofthefactthattheyareusingthesecues.So,whilethereisstillmuchtobelearnedaboutmentalmechanismsunderlyingmindreading,wethinkitisverylikelythattheinforma-tionaboutthemindthatthosemechanismsexploitissubstantiallyricherthantheinformationcontainedinLewisianplatitudes.Ifwearerightaboutthis,thenthosewhothinkthatthefunctionalistaccountofthemeaningofordinarymentalstatetermsisontherighttrackwillhavetoconfrontaquitecrucialquestion:whichaccountoffolkpsychologypicksoutthetheorythatactuallydeterminesthemeaningofmentalstateterms?Isthemean-ingofthesetermsfixedbythetheorywecanarticulatebycollectingandsystema-tizingplatitudes,orisitfixedbythemuchrichertheorythatwecandiscoveronlybystudyingthesortofinformationexploitedbythemechanismsunderlyingmindreading?Wedon’tthinkthereisanyreallydefinitiveanswertothisquestion.Itwould,ofcourse,beenormouslyusefuliftherewereawell-motivatedandwidelyac-ceptedgeneraltheoryofmeaningtowhichwemightappeal.But,notoriously,thereisnosuchtheory.Meaningisatopiconwhichdisagreementsaboundevenaboutthemostfundamentalquestions,andtherearemanyphilosopherswhothinkthattheentirefunctionalistapproachtospecifyingthemeaningofmental240\nFolkPsychology7statetermsisutterlywrongheaded.Havingsaidallthis,however,weareinclinedtothinkthatthosewhoaresympathetictothefunctionalistapproachshouldpreferthemindreadingaccountoffolkpsychologyovertheplatitudeaccount.Foronthemindreadingaccount,folkpsychologyisthetheorythatpeopleactu-allyuseinrecognizingandattributingmentalstates,indrawinginferencesaboutmentalstates,andingeneratingpredictionsandexplanationsonthebasisofmentalstateattributions.Itishardtoseewhysomeonewhothinks,asfunction-alistsdo,thatmentalstatetermsgettheirmeaningbybeingembeddedinatheorywouldwanttofocusontheplatitude-basedtheorywhoseprinciplespeoplecaneasilyacknowledge,ratherthantherichertheorythatisactuallyguid-ingpeoplewhentheythinkandtalkaboutthemind.10.3TheChallengefromSimulationTheoryLet’stakeamomenttotakestockofwhereweare.Insection10.1weexplainedwhyfolkpsychologyhasplayedsuchanimportantroleinrecentphilosophyofmind:functionalistsmaintainthatfolkpsychologyisthetheorythatimplicitlydefinesordinarymentalstateterms,andeliminativists(whotypicallyagreewithfunctionalistsaboutthemeaningofmentalstateterms)arguethatfolkpsycho-logyisaseriouslymistakentheory,andthatboththetheoryandthementalstatesthatitpositsshouldberejected.Insection10.2wedistinguishedtwodifferentaccountsoffolkpsychology,andweargued,albeittentatively,thatfunctionalistsshouldpreferthemindreadingaccountonwhichfolkpsychologyistherichbodyofinformationortheorythatunderliespeople’sskillinattributingmentalstatesandinpredictingandexplainingbehavior.Inthissection,weturnourattentiontoanimportantnewchallengethathasemergedtoallofthis.Sincethemid-1980sanumberofphilosophersandpsychologistshavebeenarguingthatitisamistaketothinkthatmindreadinginvokesarichbodyofinformationaboutthemind.Rather,theymaintain,mindreadingcanbeexplainedasakindofmentalsimulationthatrequireslittleornoinformationabouthowthemindworks(Gordon1986;Heal1986;Goldman1989;Harris1992)Ifthesesimulationtheoristsareright,andifweacceptthemindreadingaccountoffolkpsychology,thenthereisnosuchthingasfolkpsychology.Thatwouldbebadnewsforfunction-alists.Itwouldalsobebadnewsforeliminativists,sinceifthereisnosuchthingasfolkpsychology,thentheircoreargument–whichclaimsthatfolkpsychologyisaseriouslymistakentheory–hasgoneseriouslyamiss.Howcoulditbethatthementalmechanismsunderlyingmindreadingdonotrequirearichbodyofinformation?Simulationtheoristsoftenbegintheiranswerbyusingananalogy.Supposeyouwanttopredicthowaparticularairplanewillbehaveincertainwindconditions.Onewaytoproceedwouldbetoderiveapredictionfromaeronauticaltheoryalongwithadetaileddescriptionoftheplane.Another,quitedifferent,strategywouldbetobuildamodeloftheplane,putit241\nStephenP.StichandShaunNicholsinawindtunnelthatreproducesthosewindconditions,andthensimplyobservehowthemodelbehaves.Thesecondstrategy,unlikethefirst,doesnotrequirearichbodyoftheory.Simulationtheoristsmaintainthatsomethinglikethissecondstrategycanbeusedtoexplainpeople’smindreadingskills.Forifyouaretryingtopredictwhatanotherperson’smindwilldo,andifthatperson’smindissimilartoyours,thenyoumightbeabletousecomponentsofyourownmindasmodelsofthesimilarcomponentsinthemindoftheotherperson(whomwe’llcallthe“target”).Hereisaquicksketchofhowtheprocessmightwork.Supposethatyouwanttopredictwhatthetargetwilldecidetodoaboutsomeimportantmatter.Thetarget’smind,we’llassume,willmakethedecisionbyutilizingadecision-makingor“practicalreasoning”systemwhichtakeshisrelevantbeliefsanddesiresasinputand(somehoworother)comesupwithadecisionaboutwhattodo.Thelighterlinesinfigure10.1areasketchofthesortofcognitivearchitecturethatmightunderliethenormalprocessofdecision-making.Nowsupposethatyourmindcanmomentarilytakeyourdecisionmakingsystem“off-line”sothatyoudonotactuallyactonthedecisionsthatitproduces.Supposefurtherthatinthisoff-linemodeyourmindcanprovideyourdecision-makingsystemwithsomehypotheticalor“pretend”beliefsanddesires–beliefsanddesiresthatyoumaynotactuallyhavebutthatthetargetdoes.Yourmindcouldthensimplysitbackandletyourdecision-makingsystemgenerateadecision.Ifyourdecision-makingsystemissimilartothetarget’s,andifthehypotheticalbeliefsanddesiresthatyou’vefedintotheoff-linesystemareclosetotheonesthatthetargethas,thenthedecisionthatyourdecision-makingsystemgenerateswillbesimilaroriden-ticaltotheonethatthetarget’sdecision-makingsystemwillproduce.Ifthatoff-linedecisionisnowsentontothepartofyourmindthatgeneratespredictionsaboutwhatotherpeoplewilldo,youwillpredictthatthatisthedecisionthetargetwillmake,andthereisagoodchancethatyourpredictionwillbecorrect.Allofthishappens,accordingtosimulationtheorists,withlittleornoconsciousawarenessonyourpart.Moreover,andthisofcourseisthecrucialpoint,theprocessdoesnotutilizeanytheoryorrichbodyofinformationabouthowthedecision-makingsystemworks.Rather,youhavesimplyusedyourowndecision-makingsystemtosimulatethedecisionthatthetargetwillactuallymake.Thedarklinesinfigure10.1sketchthesortofcognitivearchitecturethatmightunderliethiskindofsimulation-basedprediction.Theprocesswehavejustdescribedtakesthedecision-makingsystemoff-lineandusessimulationtopredictdecisions.Butmuchthesamesortofprocessmightbeusedtotaketheinferencemechanismorothercomponentsofthemindoff-line,andthustomakepredictionsaboutothersortsofmentalprocesses.Someofthemoreenthusiasticdefendersofsimulationtheoryhavesuggestedthatallmindreadingskillscouldbeaccomplishedbysomethinglikethisprocessofsimulation,andthusthatweneednotsupposethatfolkpsychologicaltheoryplaysanyimportantroleinmindreading.Ifthisisright,thenbothfunctionalism8andeliminativismareintrouble.242\nFolkPsychologyPerceptualprocessesInferenceBody-monitoringmechanismssystemBeliefsDesiresDecision-making(practicalreasoning)systemBehavior-predictingPretendbeliefandexplainingsystemanddesiregeneratorActioncontrolsystemsBEHAVIORFigure10.110.4ThreeAccountsofMindreading:Information-rich,Simulation-basedandHybridSimulationtheoristsandadvocatesofinformation-richaccountsofmindreadingoffer9competingempiricaltheoriesaboutthementalprocessesunderlyingmindreading,andmuchoftheliteratureonthetopichasbeencastasawinner-takes-alldebate10betweenthesetwogroups.Inrecentyears,however,therehasbeenagrow-ingawarenessthatmindreadingisacomplex,multifacetedphenomenonandthatsomeaspectsofmindreadingmightbesubservedbyinformation-poorsimulation-likeprocesses,whileothersaresubservedbyinformation-richpro-cesses.Thishybridapproachisonethatwehaveadvocatedforanumberofyears(StichandNichols1995;Nicholsetal.1996;NicholsandStich,forthcoming),andinthissectionwewillgiveabriefsketchofthecaseinfavorofthehybrid11approach.Wewillbeginbyfocusingononeimportantaspectofmindread-ingforwhichinformation-richexplanationsareparticularlyimplausibleandasimulation-styleaccountisverylikelytobetrue.Wewillthentakeuptwootheraspectsofmindreadingwhere,wethink,information-richexplanationsareclearlytobepreferredtosimulation-basedexplanations.243\nStephenP.StichandShaunNichols10.4.1Inferenceprediction:amindreadingskillsubservedbysimulationOnestrikingfactaboutthemindreadingskillsofnormaladultsisthatweareremarkablygoodatpredictingtheinferencesoftargets,eventheirobviouslynon-demonstrativeinferences.Suppose,forexample,thatFredcomestobelievethatthePresidentoftheUnitedStateshasresigned,afterhearingabriefreportontheradio.WhodoesFredthinkwillbecomePresident?WequicklygeneratethepredictionthatFredthinkstheVice-PresidentwillbecomePresident.Weknowperfectlywell,andso,wepresume,doesFred,thattherearelotsofwaysinwhichhisinferencecouldbemistaken.TheVice-Presidentcouldbeassassinated;theVice-PresidentmightresignbeforebeingsworninasPresident;ascandalmightleadtotheremovaloftheVice-President;theremightbeacoup.ItiseasytogeneratestoriesonwhichtheVice-PresidentwouldnotbecomethenewPresid-ent.YetwepredictFred’snon-demonstrativeinferencewithouthesitation.Andinmostcaseslikethis,ourpredictionsarecorrect.Anyadequatetheoryofmindreadingneedstoaccommodatethesefacts.Advocatesofinformation-richapproachestomindreadinghavebeennotablysilentaboutinferenceprediction.Indeed,sofaraswehavebeenabletodeter-mine,noleadingadvocateofthatapproachhaseventriedtoofferanexplanationofthefactthatwearestrikinglygoodatpredictingtheinferencesthatotherpeoplemake.Andweareinclinedtothinkthatthereasonforthisomissionisprettyclear.Forathorough-goingadvocateoftheinformation-richapproach,theonlyavailableexplanationofourinferencepredictionskillsismoreinformation.Ifwearegoodatpredictinghowotherpeoplewillreason,thatmustbebecausewehavesomehowacquiredaremarkablygoodtheoryabouthowpeoplereason.Butthataccountseemsratherprofligate.Toseewhy,considertheanalogybetweenpredictinginferencesandpredictingthegrammaticalintuitionsofsomeonewhospeaksthesamelanguagethatwedo.Toexplainoursuccessatthislattertask,anadvocateoftheinformation-richapproachwouldhavetosaythatwehaveatheoryabouttheprocessessubservinggrammaticalintuitionproductioninotherpeople.But,asHarris(1992)pointedout,thatseemsratherfar-fetched.Amuchsimplerhypothesisisthatwerelyonourownmechanismsforgeneratinglin-guisticintuitions,andhavingdeterminedourownintuitionsaboutaparticularsentence,weattributethemtothetarget.Harris’sargumentfromsimplicity,asweshallcallit,playedanimportantroleinconvincingusthatacomprehensivetheoryofmindreadingwouldhavetoinvokemanydifferentsortsofprocess,andthatsimulationprocesseswouldbeamongthem.However,wedon’tthinkthattheargumentfromsimplicityistheonlyreasontopreferasimulation-basedaccountofinferencepredictionoveraninformation-richaccount.Indeed,iftheargumentfromsimplicityweretheonlyoneavailable,aresolutedefenderoftheinformation-richapproachmightsimplydiginherheelsandnotethatthesystemsproducedbyMotherNatureareoften244\nFolkPsychologyfarfromsimple.Therearelotsofexamplesofredundancyandapparentlyunne-cessarycomplexityinbiologicalsystems.So,theinformation-richtheoristmightargue,themerefactthatatheory-basedaccountofinferencepredictionwouldbelesssimplethanasimulation-styleaccountishardlyaknock-downargumentagainstit.Thereis,however,anothersortofargumentthatcanbemountedagainstaninformation-richapproachtoinferenceprediction.Wethinkitisaparticularlyimportantargumentsinceitcanbegeneralizedtoanumberofothermindreadingskills,andthusitcanserveasavaluableheuristicinhelpingustodecidewhichaspectsofmindreadingareplausiblytreatedassimulation-based.Thissecondargument,whichwewillcalltheargumentfromaccuracy,beginswiththeobservationthatinferencepredictionisremarkablyaccurateoverawiderangeofcases,includingcasesthatarequitedifferentfromanythingthatmostmindreadersarelikelytohaveencounteredbefore.Thereis,forexample,arichliteratureinthe“heuristicsandbiases”traditionincognitivesocialpsychologychroniclingthewaysinwhichpeoplemakewhatappeartobeverybadinferences12onawiderangeofproblemsrequiringdeductiveandinductivereasoning.Inallofthisliterature,however,thereisnosuggestionthatpeoplearebadatpredictingotherpeople’sinferences,whetherthoseinferencesaregoodorbad.Thiscon-trastssharplywiththeliteratureondesire-attributionthatwediscussbelow,whereitisoftenremarkedhowsurprisingandunpredictablepeople’sdesiresanddecisionsare.Althoughithasn’tbeenstudiedsystematically,wethinkitisquitelikelythatpeopletypicallypredictotherswillmakejustthosebadinferencesthattheywouldmakethemselves,evenonproblemsthatarequitedifferentfromanytheyhaveencounteredbefore.Ifthatisindeedthecase,itposesaproblemforinformation-richaccounts:Howdoordinarymindreadersmanagetoendupwithsuchanaccuratetheoryabouthowpeopledrawinferences–atheorywhichsupportscorrectpredictionsevenaboutquiteunfamiliarsortsofinferences?Theproblemismademoreacutebythefactthatthereareothersortsofmindread-ingtasksonwhichpeopledoverybadly.Whydopeopleacquiretherighttheoryaboutinferenceandthewrongtheoryaboutothermentalprocesses?Asimulation-basedaccountofinferenceprediction,bycontrast,hasareadyexplana-tionofouraccuracy.Onthesimulationaccount,weareusingthesameinferencemechanismforbothmakingandpredictinginferences,soitistobeexpectedthatwewouldpredictthatotherpeoplemakethesameinferenceswedo.Obviously,theargumentfromaccuracyisatwo-edgedsword.Inthosedo-mainswhereweareparticularlygoodatpredictingorattributingmentalstatesinunfamiliarcases,theargumentsuggeststhatthemindreadingprocessisunlikelytobesubservedbyaninformation-richprocess.Butinthosecaseswherewearebadatpredictingorattributingmentalstates,theargumentsuggeststhattheprocessisunlikelytobesubservedbyasimulationprocess.Werecognizethattherearevariousmovesthatmightbemadeinresponsetotheargumentfromaccuracy,andthuswedonottreattheargumentasdefinitive.Wedo,however,thinkthattheargumentjustifiesastronginitialpresumptionthataccuratemind-readingprocessesaresubservedbysimulation-likeprocessesandthatinaccurate245\nStephenP.StichandShaunNicholsonesarenot.Andifthisisright,thenthereisastrongpresumptioninfavorofthehypothesisthatinferencepredictionissimulationbased.10.4.2Desire-attribution:amindreadingskillthatcannotbeexplainedbysimulationAnotherquitecentralaspectofmindreadingisthecapacitytoattributedesirestootherpeople.Withoutthatcapacitywewouldnotknowwhatotherpeoplewant,andwewouldbeseverelyimpairedintryingtopredictorexplaintheiractions.Thereareanumberofprocessesthatcangiverisetobeliefsaboutatarget’sdesires.Insomecasesweuseinformationaboutthetarget’sverbalandnon-verbalbehavior(includingtheirfacialexpressions)todeterminewhattheywant.Inothercasesweattributedesiresonthebasisofwhatotherpeoplesayaboutthetarget.Andinalllikelihoodavarietyofothercuesandsourcesofdataarealsousedinthedesire-attributionprocess.Itisourcontentionthatthesedesire-attributionskillsdonotdependonsimulation,butratheraresubservedbyinformation-richprocesses.Wehavetwoquitedifferentreasonsforthisclaim.First,desire-attributionexhibitsapatternofsystematicinaccuracyandthatsupportsatleastaninitialpresumptionthattheprocessisnotsimulation-based.Oneverystrikingexamplecomesfromwhatisperhapsthemostfamousseriesofexperimentsinallofsocialpsychology.Milgram(1963)hada“teacher”subjectflipswitchesthatweresupposedtodelivershockstoanothersubject,the“learner”(whowasactuallyanaccomplice).Foreachmistakethelearnermade,theteacherwasinstructedtodeliverprogressivelystrongershocks,includingonelabeled“Danger:SevereShock”andculminatinginaswitchlabeled“450-volt,XXX.”Iftheteachersubjectexpressedreservationstotheexperimentalassistant,hewascalmlytoldtocontinuetheexperiment.Theresultoftheexperimentwasastonish-ing.Aclearmajorityofthesubjectsadministeredalltheshocks.Peopleoftenfindtheseresultshardtobelieve.Indeed,theMilgramfindingsaresocounterintuitivethatinaverbalre-enactmentoftheexperiment,peoplestilldidn’tpredicttheresults(Bierbrauer1973,discussedinNisbettandRoss1980:121).OneplausibleinterpretationofthesefindingsisthatintheMilgramexperimenttheinstructionsfromtheexperimentergeneratedadesiretocomply,which,inmostcases,over-whelmedthesubject’sdesirenottoharmthepersontheybelievedtobeonthereceivingendoftheelectricshockapparatus.ThefactthatpeoplefindtheresultssurprisingandthatBierbrauer’ssubjectsdidnotpredictthemindicatesanimport-antlimitationinourcapacitytodeterminethedesiresofothers.Thereisalargeliteratureincognitivesocialpsychologydetailingmanyothercasesinwhichdesiresandpreferencesareaffectedinremarkableandunexpectedwaysbythecircumstancessubjectsencounterandtheenvironmentinwhichtheyareembedded.Theimportantpoint,forpresentpurposes,isthatpeopletypicallyfindtheseresultssurprisingandoccasionallyquiteunsettling,andthefactthattheyaresurprised(evenafterseeingorgettingadetaileddescriptionofthe246\nFolkPsychologyexperimentalsituation)indicatesthatthementalmechanismstheyareusingtopredictthesubjects’desiresandpreferencesaresystematicallyinaccurate.Thoughthisisnottheplaceforanextendedsurveyofthemanyexamplesintheliterature,13wecannotresistmentioningoneofourfavorites.LoewensteinandAdler(1995)lookedattheabilityofsubjectstopredicttheirownpreferenceswhenthosepreferencesareinfluencedbyasurprisingandlittle-knowneffect.TheeffectthatLoewensteinandAdlerexploitistheendowmenteffect,arobustandrapidlyappearingtendencyforpeopletosetasignificantlyhighervalueforanobjectiftheyactuallyownitthantheywouldiftheydidnotownit(Thaler1980).HereishowLoewensteinandAdlerdescribethephenomenon:Inthetypicaldemonstrationoftheendowmenteffect...onegroupofsubjects(sellers)areendowedwithanobjectandaregiventheoptionoftradingitforvariousamountsofcash;anothergroup(choosers)arenotgiventheobjectbutaregivenaseriesofchoicesbetweengettingtheobjectorgettingvariousamountsofcash.Althoughtheobjectivewealthpositionofthetwogroupsisidentical,asarethechoicestheyface,endowedsubjectsholdoutforsignificantlymoremoneythanthosewhoarenotendowed.(1995:929–30).Inanexperimentdesignedtotestwhether“unendowed”subjectscouldpredictthevaluetheywouldsetiftheywereactuallytoowntheobjectinquestion,theexperimenterfirstallowedsubjects(whoweremembersofauniversityclass)toexamineamugengravedwiththeschoollogo.Aformwasthendistributedtoapproximatelyhalfofthesubjects,chosenatrandom,onwhichtheywereasked“toimaginethattheypossessedthemugondisplayandtopredictwhethertheywouldbewillingtoexchangethemugforvariousamountsofmoney”(ibid.:931).Whenthesubjectswhoreceivedtheformhadfinishedfillingitout,allthesubjectswerepresentedwithamugandgivenasecondformwithinstructionsanalogoustothoseonthepredictionform.Butonthesecondformitwasmadeclearthattheyactuallycouldexchangethemugforcash,andthatthechoicestheymadeonthissecondformwoulddeterminehowmuchmoneytheymightget.“Subjectsweretoldthattheywouldreceivetheoptionthattheyhadcircledononeofthelines–whichlinehadbeendeterminedinadvancebytheexperi-menter”(ibid.).Theresultsshowedthatsubjectswhohadcompletedthefirstformsubstantiallyunderpredictedtheamountofmoneyforwhichtheywouldbewillingtoexchangethemug.Inonegroupofsubjects,themeanpredictedexchangepricewas$3.73,whilethemeanactualexchangepriceforsubjects(thesamesubjectswhomadetheprediction)was$5.40.Moreover,thereseemedtobean“anchoringeffect”inthisexperimentwhichdepressedtheactualexchangeprice,sincethemeanactualexchangepriceforsubjectswhodidnotmakeapredictionabouttheirownsellingpricewasevenhigher,at$6.46.Hereagainwefindthatpeoplearesystematicallyinaccurateatpredictingtheeffectofthesitu-ationondesires,andinthiscasethedesirestheyfailtopredictaretheirown.Ifthesedesirepredictionsweresubservedbyasimulationprocess,itwouldbe247\nStephenP.StichandShaunNicholssomethingofamysterywhythepredictionsaresystematicallyinaccurate.Butif,aswebelieve,theyaresubservedbyaninformation-richprocess,theinaccuracycanbereadilyexplained.Thetheoryorbodyofinformationthatguidesthepredictionsimplydoesnothaveaccurateinformationabouttherathersurprisingmentalprocessesthatgiverisetothesedesires.Oursecondreasonforthinkingthatthementalmechanismssubservingdesire-attributionuseinformation-richprocessesratherthansimulationisthatitishardtoseehowtheworkdonebythesemechanismscouldbeaccomplishedbysimu-lation.Indeed,sofarasweknow,simulationtheoristshavemadeonlyoneproposalabouthowsomeofthesedesiredetectiontasksmightbecarriedout,anditissingularlyimplausible.Theproposal,endorsedbybothGordon(1986)andGoldman(1989),beginswiththefactthatsimulationprocessesliketheonesketchedinfigure10.1canbeusedtomakebehaviorpredictions,andgoesontosuggestthattheymightalsobeusedtogeneratebeliefsaboutthedesiresandbeliefsthatgiverisetoobservedbehaviorbyexploitingsomethingakintothestrategyofanalysis-by-synthesis(originallydevelopedbyHalleandStevens(1962)forphonemerecognition).Inusingtheprocessinfigure10.1topredictbeha-vior,hypotheticalor“pretend”beliefsanddesiresarefedintothemindreader’sdecision-makingsystem(beingused“off-line”ofcourse),andthemindreaderpredictsthatthetargetwoulddowhatthemindreaderwoulddecidetodo,giventhosebeliefsanddesires.Inananalysis-by-synthesisaccountofthegenerationofbeliefsaboutdesiresandbeliefs,theprocessis,ineffect,runbackwards.Itstartswithabehavioralepisodethathasalreadyoccurredandproceedsbytryingtofindhypotheticalbeliefsanddesireswhich,whenfedintothemindreader’sdecisionmechanism,willproduceadecisiontoperformthebehaviorwewanttoexplain.Anobviousproblemwiththisstrategyisthatitwillgeneratetoomanycandid-ates,sincetypicallythereareendlesslymanypossiblesetsofbeliefsanddesiresthatmightleadthemindreadertodecidetoperformthebehaviorinquestion.Gordoniswellawareoftheproblem,andheseemstothinkhehasasolution:NomatterhowlongIgoontestinghypotheses,Iwillnothavetriedoutallcandidateexplanationsofthe[target’s]behavior.PerhapssomeoftheunexaminedcandidateswouldhavedoneatleastaswellastheoneIsettlefor,ifIsettle:perhapsindefinitelymanyofthemwouldhave.Butthesewouldbe“farfetched,”Isayintuitively.ThereinIexhibitmyinertialbias.Theless“fetching”(or“stretching,”asactorssay)Ihavetodototracktheother’sbehavior,thebetter.Itendtofeignonlywhennecessary,onlywhensomethingintheother’sbehaviordoesn’tfit....Thisinertialbiasmaybethoughtofasa“leasteffort”principle:the“principleofleastpretending.”Itexplainswhy,otherthingsbeingequal,Iwillpreferthelessradicaldeparturefromthe“real”world–i.e.fromwhatImyselftaketobetheworld.(Gordon1986:164)Unfortunately,itisnotatallclearwhatGordonhasinmindbyaninertialbiasagainst“fetching.”Themostobviousinterpretationisthatattributionsaremore“far-fetched”thefurthertheyare,onsomeintuitivescale,fromone’sownmental248\nFolkPsychologystates.Butifthat’swhatGordonintends,itseemsclearthatthesuggestionwon’twork.Forinmanycasesweexplainbehaviorbyappealingtodesiresorbeliefs(orboth)thatareveryfarfromourown.Imight,forexample,explainthecatchasingthemousebyappealingtothecat’sdesiretoeatthemouse.Butthereareindefinitelymanydesiresthatwouldleadmetochaseamousethatareintuitivelymuchclosertomyactualdesiresthanthedesiretoeatamouse.Simulationtheoristshaveofferednootherproposalfornarrowingdowntheendlesssetofcandidatebeliefsanddesiresthattheanalysis-by-synthesisstrategywouldgen-erate,andwithoutsomeplausiblesolutiontothisproblemthestrategylooksquitehopeless.Soitisnotsurprisingthataccountsofthissorthavelargelydisappearedfromthesimulationtheoryliteratureoverthelastdecade.Andthat,perhaps,reflectsatleastatacitacknowledgement,onthepartofsimulationthe-orists,thatdesire-attributioncanonlybeexplainedbyappealingtoinformation-richprocesses.10.4.3Discrepantbelief-attribution:anothermindreadingskillthatcannotbeexplainedbysimulationYetanotherimportantaspectofmindreadingisthecapacitytoattributebeliefsthatweourselvesdonothold–discrepantbeliefs,astheyaresometimescalled.Thereareanumberofprocessessubservingdiscrepantbelief-attribution,somerelyingonbeliefsaboutthetarget’sperceptualstates,othersexploitinginforma-tionaboutthetarget’sverbalbehavior,andstillothersrelyingoninformationaboutthetarget’snon-verbalbehavior.Allofthese,wesuspect,aresubservedbyinformation-richmechanisms,ratherthanbyamechanismthatusessimulation.Ourreasonsarelargelyparalleltotheonesweofferedfordesire-attribution.First,thereisabundantevidencethatthediscrepantbelief-attributionsystemexhibitssystematicinaccuraciesofthesortwewouldexpectfromaninformation-richsystemthatisnotquiterichenoughanddoesnotcontaininformationabouttheprocessgeneratingcertaincategoriesofdiscrepantbeliefs.Second,thereisnoplausiblewayinwhichprototypicalsimulationmechanismscoulddowhatthediscrepantbelief-attributionsystemdoes.Onedisquietingexampleofasystematicfailureindiscrepantbelief-attributioncomesfromthestudyofbelief-perseverance.Inthepsychologylaboratory,andineverydaylife,itsometimeshappensthatpeoplearepresentedwithfairlypersuas-iveevidence(e.g.testresults)indicatingthattheyhavesomehithertounex-pectedtrait.Inlightofthatevidencepeopletypicallyformthebeliefthattheydohavethetrait.Whatwillhappentothatbeliefif,shortlyafterthis,peoplearepresentedwithaconvincingcasediscreditingthefirstbodyofevidence?Suppose,forexample,theyareconvincedthatthetestresultstheyreliedonwereactuallysomeoneelse’s,orthatnorealtestwasconductedatall.Mostpeopleexpectthattheunderminedbeliefwillsimplybediscarded.Andthatviewwassharedbyagenerationofsocialpsychologistswhodupedsubjectsintobelievingallsortsof249\nStephenP.StichandShaunNicholsthingsaboutthemselves,oftenbyadministeringriggedpsychologicaltests,ob-servedtheirreactions,andthen“debriefed”thesubjectsbyexplainingtheruse.Theassumptionwasthatnoenduringharmcouldbedonebecauseoncetherusewasexplainedtheinducedbeliefwouldbediscarded.Butinawidelydiscussedseriesofexperiments,Rossandhisco-workershavedemonstratedthatthisissimplynotthecase.Onceasubjecthasbeenconvincedthatshehasatrait,showingherthattheevidencethatconvincedherwascompletelyphonydoesnotsucceedineliminatingthebelief(NisbettandRoss1980:175–9).Ifthetraitinquestionisbeinginclinedtosuicide,orbeing“latentlyhomosexual,”beliefperseverancecanleadtoseriousproblems.Thepartofthediscrepantbelief-attributionsystemthatledbothpsychologistsandeveryoneelsetoexpectthatthesediscrepantbeliefswouldbediscardedafterdebriefingapparentlyhasinaccur-ateinformationabouttheprocessofbelief-perseveranceandthusitleadstosystematicallymistakenbelief-attributions.Anotherexample,withimportantimplicationsforpublicpolicy,isprovidedbytheworkofLoftus(1979)andothersontheeffectof“post-eventinterventions”onwhatpeoplebelieveabouteventstheyhavewitnessed.Inoneexperimentsubjectswereshownafilmofanautoaccident.Ashorttimelatertheywereaskedaseriesofquestionsabouttheaccident.Forsomesubjects,oneofthequestionswas:“Howfastwasthewhitesportscartravelingwhenitpassedthebarnwhiletravelingalongthecountryroad?”Othersubjectswereasked:“Howfastwasthewhitesportscartravelingwhiletravelingalongthecountryroad?”Oneweeklaterallthesubjectswereaskedwhethertheyhadseenabarn.Thoughtherewasnobarninthefilmthatthesubjectshadseen,subjectswhowereaskedthequestionthatmentionedthebarnwerefivetimesmorelikelytobelievethattheyhadseenone.Inanotherexperiment,conductedintrainstationsandothernaturalisticsettings,Loftusandherstudentsstageda“robbery”inwhichamaleconfederatepulledanobjectfromabagthattwofemalestudentshadtemporarilyleftunat-tendedandstuffeditunderhiscoat.Amomentlater,oneofthewomennoticedthatherbaghadbeentamperedwithandshouted,“OhmyGod,mytaperecorderismissing.”Shewentontolamentthatherbosshadloanedittoherandthatitwasveryexpensive.Bystanders,mostofwhomwerequitecooperative,wereaskedfortheirphonenumbersincaseanaccountoftheincidentwasneededforinsurancepurposes.Aweeklater,an“insuranceagent”calledtheeyewitnessesandaskedaboutdetailsofthetheft.Amongthequestionsaskedwas“Didyouseethetaperecorder?”Morethanhalfoftheeyewitnessesrememberedhavingseenit,andnearlyallofthesecoulddescribeitindetail–thisdespitethefactthattherewasnotaperecorder.Onthebasisofthisandotherexperiments,Loftusconcludesthatevencasualmentionofobjectsthatwerenotpresentorofeventsthatdidnottakeplace(forexample,inthecourseofpolicequestioning)cansignificantlyincreasethelikelihoodthattheobjectsoreventswillbeincorpor-atedintopeople’sbeliefsaboutwhattheyobserved.AcentralthemeinLoftus’sworkisthatthelegalsystemshouldbemuchmorecautiousaboutrelyingoneyewitnesstestimony.Andamajorreasonwhythelegalsystemisnotascautious250\nFolkPsychologyasitshouldbeisthatourinformation-drivendiscrepantbelief-attributionsystemlacksinformationaboutthepost-eventprocessesofbelief-formationthatLoftushasdemonstrated.Asinthecaseofdesire-attribution,weseenoplausiblewayinwhichtheworkdonebythementalmechanismssubservingdiscrepantbelief-attributioncouldbeaccomplishedbysimulation.Hereagain,theonlyproposalthatsimulationthe-oristshaveofferedistheanalysis-by-synthesisaccount,andthatstrategywon’tworkanybetterforbelief-attributionthanitdoesfordesire-attribution.10.5ConclusionIntheprevioussectionwesketchedsomeofthereasonsforacceptingahybridaccountofmindreadinginwhichsomeaspectsofthatskillareexplainedbyappealtoinformation-richprocesses,whileotheraspectsareexplainedbysimula-tion.Thoughweonlylookedatahandfulofmindreadingskills,wehavearguedelsewhere(NicholsandStich,forthcoming)thatmuchthesamepatterncanbefoundmoregenerally.Mindreadingisacomplexandmultifacetedphenomenon,manyfacetsofwhicharebestexplainedbyaninformation-richapproach,whilemanyotherfacetsarebestexplainedbysimulation.Ifthisiscorrect,itpresentsbothfunctionalistsandeliminativistswithsomeratherawkwardchoices.Func-tionalists,aswehaveseen,holdthatthemeaningofordinarymentalstatetermsisdeterminedbyfolkpsychology,andeliminativiststypicallyagree.Insection10.2wearguedthatfunctionalismismostplausibleiffolkpsychologyistakentobetheinformation-richtheorythatsubservesmindreading.Butnowitappearsthatonlypartsofmindreadingrelyonaninformation-richtheory.Shouldfunc-tionalistsinsistthatthetheoryunderlyingtheseaspectsofmindreadingfixesthemeaningofmentalstateterms,orshouldtheyretreattotheplatitudeaccountoffolkpsychology?Weareinclinedtothinkthatwhicheveroptionfunctionalistsadopt,theirtheorywillbelessattractivethanitwasbeforeitbecameclearthattheplatitudeapproachandthemindreadingapproachwoulddiverge,andthatonlypartofmindreadingreliesonfolkpsychology.Notes1ThoughwewillfocusonLewis’sinfluentialexposition,manyotherphilosophersdevelopedsimilarviews,includingPutnam(1960),FodorandChihara(1965),andArmstrong(1968).2Thoughbeware.Inthephilosophyofmind,theterm“functionalism”hasbeenusedforavarietyofviews.Someofthembearaclearfamilyresemblancetotheonewe’vejustsketched,whileothersdonot.Forgoodoverviews,seeLycan(1994)andBlock(1994).251\nStephenP.StichandShaunNichols3Foranoverviewofthesedebates,seeStich(1996:ch.1),andchapter2inthisvolume.4ThedistinctionwasfirstnotedinStichandRavenscroft(1994).5Thoughnotalways,aswe’llseeinsection10.4.6Eliminativists,ofcourse,wouldnotagreethatwedoagoodjobatattributingandpredictingmentalstatesoratexplainingbehaviorintermsofpastmentalstates,sincetheymaintainthatthementalstatesweareattributingdonotexist.Buttheywouldnotdenythatthereisanimpressivedegreeofagreementinwhatpeoplesayaboutotherpeople’smentalstates,andthatthatagreementneedstobeexplained.7See,forexample,FodorandLePore(1992).Forausefuloverviewofmanyofthedisputesaboutthetheoryofmeaning,seeDevitt(1996).8RobertGordonisthemostaviddefenderoftheviewthatallmindreadingskillscanbeexplainedbysimulation.Hereisacharacteristicpassage:Itis...uncannythatfolkpsychologyhasn’tchangedverymuchoverthemillennia....Churchlandthinksthisasignthatfolkpsychologyisabadtheory;butitcouldbeasignthatitisnotheoryatall,not,atleast,intheacceptedsenseof(roughly)asystemoflawsimplicitlydefiningasetofterms.Instead,itmightbejustthecapacityforpracticalreasoning,supplementedbyaspecialuseofachildishandprimitivecapacityforpretendplay.(1986:71)Ofcourse,aneliminativistmightobjectthatthesimulationtheoristbegsthequestionsincethesimulationaccountofdecisionpredictionpresupposestheexistenceofbeliefs,desiresandotherpositsoffolkpsychology,whileeliminativistsholdthatthesecommon-sensementalstatesdonotexist.Constructingaplausiblereplytothisobjectionisleftasanexerciseforthereader.9ThoughHeal(1998)hasarguedthatthereisoneinterpretationofsimulationtheoryonwhichitistrueapriori.Foracritique,seeNicholsandStich(1998).10ManyoftheimportantpapersinthisliteraturearecollectedinDaviesandStone(1995a,1995b).11Wehavealsoarguedthatsomeimportantaspectsofmindreadingaresubservedbypro-cessesthatcan’tbecomfortablycategorizedaseitherinformation-richorsimulation-like.Butsincespaceislimited,wewillnottrytomakeacaseforthathere.SeeNicholsandStich(forthcoming).12Amongthebest-knownexperimentsofthiskindarethoseillustratingtheso-calledconjunctionfallacy.Inonequitefamousexperiment,KahnemanandTversky(1982)presentedsubjectswiththefollowingtask.Lindais31yearsold,single,outspoken,andverybright.Shemajoredinphilosophy.Asastudent,shewasdeeplyconcernedwithissuesofdiscrimina-tionandsocialjustice,andalsoparticipatedinanti-nucleardemonstrations.Pleaserankthefollowingstatementsbytheirprobability,using1forthemostprobableand8fortheleastprobable.(a)Lindaisateacherinelementaryschool.(b)LindaworksinabookstoreandtakesYogaclasses.(c)Lindaisactiveinthefeministmovement.252\nFolkPsychology(d)Lindaisapsychiatricsocialworker.(e)LindaisamemberoftheLeagueofWomenVoters.(f)Lindaisabankteller.(g)Lindaisaninsurancesalesperson.(h)Lindaisabanktellerandisactiveinthefeministmovement.Inagroupofnaivesubjectswithnobackgroundinprobabilityandstatistics,89percentjudgedthatstatement(h)wasmoreprobablethanstatement(f)despitetheobviousfactthatonecannotbeafeministbanktellerunlessoneisabankteller.Whenthesamequestionwaspresentedtostatisticallysophisticatedsubjects–gradu-atestudentsinthedecisionscienceprogramoftheStanfordBusinessSchool–85percentgavethesameanswer!Resultsofthissort,inwhichsubjectsjudgethatacompoundeventorstateofaffairsismoreprobablethanoneofthecomponentsofthecompound,havebeenfoundrepeatedlysinceKahnemanandTversky’spioneer-ingstudies,andtheyareremarkablyrobust.Forusefulreviewsofresearchintheheuristicsandbiasestradition,seeKahnemanetal.(1982),NisbettandRoss(1980),Baron(2001),andSamuelsetal.(2003).13Foranexcellentreviewoftheliterature,seeRossandNisbett(1991).ReferencesArmstrong,D.(1968).AMaterialistTheoryoftheMind.NewYork:HumanitiesPress.Baron,J.(2001).ThinkingandDeciding,3rdedn.Cambridge:CambridgeUniversityPress.Bierbrauer,G.(1973).EffectofSet,Perspective,andTemporalFactorsinAttribution,unpublisheddoctoraldissertation,StanfordUniversity.Block,N.(1994).“Functionalism.”InS.Guttenplan(ed.),ACompaniontothePhilosophyofMind.Oxford:Blackwell:323–32.Churchland,P.(1981).“EliminativeMaterialismandPropositionalAttitudes.”JournalofPhilosophy,78:67–90.ReprintedinW.Lycan(ed.),MindandCognition.Oxford:Blackwell(1990):206–23.PagereferenceistotheLycanvolume.Davies,M.andStone,T.(1995a).FolkPsychology.Oxford:Blackwell.——(1995b).MentalSimulation.Oxford:Blackwell.Devitt,M.(1996).ComingtoOurSenses:ANaturalisticProgramforSemanticLocalism.Cambridge:CambridgeUniversityPress.Ekman,P.(1985).TellingLies:CluestoDeceitintheMarketplace,Politics,andMarriage.NewYork:W.W.NortonandCo.Fodor,J.andC.Chihara(1965).“OperationalismandOrdinaryLanguage.”AmericanPhilosophicalQuarterly,2(4).ReprintedinJ.Fodor,Representations.Cambridge,MA:MITPress(1981):35–62.Fodor,J.andLePore,E.(1992).Holism:AShopper’sGuide.Oxford:Blackwell.Gigerenzer,G.,Todd,P.andtheABCResearchGroup(1999).SimpleHeuristicsthatMakeUsSmart.Oxford:OxfordUniversityPress.Goldman,A.(1989).“InterpretationPsychologized.”MindandLanguage,4:161–85.Gopnik,A.andMeltzoff,A.(1997).Words,ThoughtsandTheories.Cambridge,MA:MITPress.253\nStephenP.StichandShaunNicholsGopnik,A.andWellman,H.(1994).“TheTheory-Theory.”InL.HirschfeldandS.Gelman(eds.),MappingtheMind:DomainSpecificityinCognitionandCulture.NewYork:CambridgeUniversityPress:257–93.Gordon,R.(1986).“FolkPsychologyasSimulation.”MindandLanguage,1:158–70.ReprintedinDaviesandStone(1995a).PagereferenceistotheDaviesandStonevolume.Halle,M.andStevens,K.(1962).“SpeechRecognition:AModelandaProgramforResearch.”InJ.FodorandJ.Katz(eds.),TheStructureofLanguage:ReadingsinthePhilosophyofLanguage.EnglewoodCliffs,NJ:Prentice-Hall.Harris,P.(1992).“FromSimulationtoFolkPsychology:TheCaseforDevelopment.”MindandLanguage,7:120–44.Hayes,P.(1985).“TheSecondNaivePhysicsManifesto.”InJ.HobbsandR.Moore(eds.),FormalTheoriesoftheCommonsenseWorld.Norwood,NJ:Ablex:1–36.Heal,J.(1986).“ReplicationandFunctionalism.”InJ.Butterfield(ed.),Language,MindandLogic.Cambridge:CambridgeUniversityPress:135–50.——(1998).“Co-cognitionandOff-lineSimulation:TwoWaysofUnderstandingtheSimulationApproach.”MindandLanguage,13:477–98.Hempel,C.(1964).“TheTheoretician’sDilemma:AStudyintheLogicofTheoryConstruction.”InC.Hempel,AspectsofScientificExplanation.NewYork:TheFreePress:173–226.Kahneman,D.andTversky,A.(1982).“ThePsychologyofPreferences.”ScientificAmer-ican,246(1):160–73.Kahneman,D.,Slovic,P.,andTversky,A.(eds.)(1982).JudgmentUnderUncertainty:HeuristicsandBiases.Cambridge:CambridgeUniversityPress.Lewis,D.(1970).“HowtoDefineTheoreticalTerms.”JournalofPhilosophy,67:17–25.——(1972).“PsychophysicalandTheoreticalIdentifications.”AustralasianJournalofPhilosophy,50:249–58.ReprintedinN.Block(ed.),ReadingsinthePhilosophyofPsychology,vol.I.Cambridge,MA:HarvardUniversityPress:207–15.PagereferencesaretotheBlockvolume.Loewenstein,G.andAdler,D.(1995).“ABiasinthePredictionofTastes.”TheEconomicJournal:TheQuarterlyJournaloftheRoyalEconomicSociety,105:929–37.Loftus,E.(1979).EyewitnessTestimony.Cambridge,MA:HarvardUniversityPress.Lycan,W.(1994).“Functionalism.”InS.Guttenplan(ed.),ACompaniontothePhilosophyofMind.Oxford:Blackwell:317–23.McCloskey,M.(1983).“IntuitivePhysics.”ScientificAmerican,248(4):122–9.Milgram,S.(1963).“BehavioralStudyofObedience.”JournalofAbnormalandSocialPsychology,67:371–8.Nichols,S.andStich,S.(1998).“RethinkingCo-cognition:AReplytoHeal.”MindandLanguage,13:499–512.——(forthcoming).Mindreading.Oxford:OxfordUniversityPress.Nichols,S.,Stich,S.,Leslie,A.,andKlein,D.(1996).“VarietiesofOff-lineSimulation.”InP.CarruthersandP.Smith(eds.),TheoriesofTheoriesofMind.Cambridge:Cam-bridgeUniversityPress:39–74.Nisbett,R.andRoss,L.(1980).HumanInference.EnglewoodCliffs,NJ:Prentice-Hall.Putnam,H.(1960).“MindsandMachines.”InS.Hook(ed.),DimensionsofMind.NewYork:NewYorkUniversityPress:138–64.Ross,L.andNisbett,R.(1991).ThePersonandtheSituation:PerspectivesofSocialPsycho-logy.Philadelphia:TempleUniversityPress.254\nFolkPsychologyRyle,G.(1949).TheConceptofMind.London:Hutchinson.Samuels,R.,Stich,S.,andFaucher,L.(2003).“ReasoningandRationality.”InI.Niiniluoto,M.Sintonen,andJ.Wolenski(eds.),HandbookofEpistemology.Dordrecht:Kluwer:1–50.Scholl,B.andLeslie,A.(1999).“Modularity,Development,and‘TheoryofMind’”.MindandLanguage,14:131–53.Sellars,W.(1956).“EmpiricismandthePhilosophyofMind.”InH.FeiglandM.Scriven(eds.),TheFoundationsofScienceandtheConceptsofPsychologyandPsychoanalysis:MinnesotaStudiesinthePhilosophyofScience,vol.1.Minneapolis:UniversityofMinne-sotaPress:253–329.Stich,S.(1996).DeconstructingtheMind.Oxford:OxfordUniversityPress.Stich,S.andNichols,S.(1995).“SecondThoughtsonSimulation.”InDaviesandStone(1995b):86–108.Stich,S.andRavenscroft,I.(1994).“WhatisFolkPsychology?”Cognition,50:447–68.ReprintedinStich(1996).Thaler,R.(1980).“TowardaPositiveTheoryofConsumerChoice.”JournalofEconomicBehaviorandOrganization,1:39–60.Young,A.(1998).FaceandMind.OxfordUniversityPress.255\nTheBlackwellGuidetoPhilosophyofMindRobertA.WilsonEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter11IndividualismRobertA.Wilson11.1IntroductionMuchdiscussionhasbeengeneratedinthephilosophyofmindoverthelast25yearsorsoonthegeneralissueoftherelationshipbetweenthenatureofthemindoftheindividualandthecharacteroftheworldinwhichthatindividual,andhencehermind,exists.Thebasicissuehereissometimesglossedintermsofwhetherpsychologicalormentalstatesare“inthehead,”buttotheuninitiatedthatislikelytosoundlikeapuzzlingissuetodebate:ofcoursementalstatesareinthehead!(butseeRowlands1999;Wilson2000a,2001).Sooneofourfirsttasksistoarticulateaversionoftheissuethatmakesmoreperspicuouswhyithasbeenatopicofsomecontentionforsolong,andthatbeginstoconveysomethingofitsimportanceforarangeofdiverseissues,suchasthemethodologyofcognitivescience,thepossibilityofself-knowledge,andthenatureofintentionalrepresentation.Considerthequestionofwhetherthecharacterofanagent’senvironmentplayssomecrucialroleindeterminingorfixingthenatureofthatagent’smind.Anaturalthought,onesharedbythosewhodisagreeabouttheanswertothequestionabove,wouldbethatagentscausallyinteractwiththeirworld,gatheringinformationaboutitthroughtheirsenses,andsothenatureoftheirminds,inparticularwhattheirthoughtsareabout,areinpartdeterminedbythecharacteroftheirworld.Thus,theworldisacausaldeterminantofone’sthoughts,andthusone’smind.Thatis,theworldisacontributingcausetothecontentofone’smind,towhatoneperceivesandthinksabout.Thisisjusttosaythatthecontentofone’smindisnotcausallyisolatedfromone’senvironment.Separatingindividualistsandanti-individualistsinthephilosophyofmindisthequestionofwhetherthereissomedeepersenseinwhichthenatureofthemindisdeterminedbythecharacteroftheindividual’sworld.Wecanapproachthisissuebyextendingthebriefdiscussionaboveoftheideathatthecontentofthemindisinpartcausallydeterminedbytheagent’s256\nIndividualismenvironmenttoexploretheconditionsunderwhichadifferenceintheworldimpliesadifferenceinthemind.Individualistsholdthatthisissojustincasethatdifferenceintheworldmakessomecorrespondingchangetowhatoccursinsidetheboundaryoftheindividual;anti-individualistsdenythis,thusallowingforthepossibilitythatindividualswhoareidenticalwithrespecttoalloftheirintrinsicfeaturescouldnonethelesshavepsychologicalormentalstateswithdifferentcontents.And,assumingthatmentalstateswithdifferentcontentsareipsofactodifferenttypesorkindsofstate,thisimpliesthatanindividual’sintrinsicprop-ertiesdonotdetermineorfixthatindividual’smentalstates.Thisprovidesuswithanotherway,amorepreciseway,ofspecifyingthediffer-encebetweenindividualism(orinternalism)anditsdenial,anti-individualism(orexternalism),aboutthemind.Individualistsclaim,andexternalistsdeny,thatwhatoccursinsidetheboundaryofanindividualmetaphysicallydeterminesthenatureofthatindividual’smentalstates.Theindividualisticdeterminationthesis,unlikethecausaldeterminationthesis,expressesaviewaboutthenatureoressenceofmentalstates,andpointstoawayinwhich,despitetheircausaldeterminationbystatesoftheworld,mentalstatesareautonomousorindepend-entofthecharacteroftheworldbeyondtheindividual.Whatindividualismimpliesisthattwoindividualswhoareidenticalinalltheirintrinsicrespectsmusthavethesamepsychologicalstates.Thisimplication,andindeedthedebateoverindividualism,isoftenmademorevividthroughthefantasyofdoppelgangers,molecule-for-moleculeidenticalindividuals,andthecorrespondingfantasyofTwinEarth.Iturntothesedualfantasiesviaasketchofthehistoryofthedebateoverindividualism.11.2GettingtoTwinEarth:What’sintheHead?HilaryPutnam’s“TheMeaningof‘Meaning’”(1975)introducedbothfantasiesinthecontextofadiscussionofthemeaningofnaturallanguageterms.Putnamwasconcernedtoshowthat“meaning”doesnotandcannotjointlysatisfytwothesesthatitwasoftentakentosatisfybythenprevalentviewsofnaturallan-guagereference:theclaimthatthemeaningofatermiswhatdeterminesitsreference,andtheclaimthatmeaningsare“inthehead,”wherethisphraseshouldbeunderstoodasmakingaclaimofthetypeidentifiedaboveaboutthemetaphysicaldeterminationofmeanings.Thesethesestypifieddescriptivetheoriesofreference,prominentsinceFregeandRussellfirstformulatedthem,accordingtowhichthereferenceofatermisfixedormetaphysicallydeterminedbythedescriptionsthataspeakerattachestothatterm.Totakeaclassicexample,supposethatIthinkofAristotleasagreat,deadphilosopherwhowroteanumberofimportantphilosophicalworks,suchastheNicomacheanEthics,andwhowasastudentofPlatoandteacherofAlexandertheGreat.Then,onadescriptivistviewofreference,thereferenceofmyterm“Aristotle”isjustthe257\nRobertA.WilsonthingintheworldthatsatisfiesthevariousdescriptionsthatIattachtothatterm:itisthethingintheworldthatisagreatphilosopher,isdead,wroteanumberofimportantphilosophicalworks(e.g.,NicomacheanEthics),wasastudentofPlato,andwasateacherofAlexandertheGreat.SuchdescriptivistviewsofthereferenceofpropernameswerethecriticalfocusofSaulKripke’sinfluentialNamingandNecessity(1980),whileinhisattackonthisclusterofviewsandtheirpresuppositions,Putnamfocusedonnaturalkindterms,suchas“water”and“tiger.”BothKripkeandPutnamintendedtheircritiquesandthesubsequentalternativetheoryofnaturallanguagereference,thecausaltheoryofreference,tobequitegeneralandtoprovideanalternativewaytothinkabouttherelationshipbetweenlanguageandtheworld.ButletusstayclosetoPutnam’sargumentanddrawoutitsconnectiontoindividualism.Consideranordinaryindividual,Oscar,wholivesonEarthandinteractswithwaterinthewaysthatmostofusdo:hedrinksit,washeswithit,andseesitfallingfromtheskyasrain.Oscar,whohasnospecialchemicalknowledgeaboutthenatureofwater,willassociatearangeofdescriptionswithhisterm“water”:itisaliquidthatonecandrink,thatisusedtowash,andthatfallsfromtheskyasrain.Onadescriptiveviewofreference,thesedescriptionsdeterminethereferenceofOscar’sterm“water.”Thatis,Oscar’sterm“water”referstowhat-everitisintheworldthatsatisfiesthesetofdescriptionsheattachestotheterm.Andsincethosedescriptionsare“inthehead,”naturallanguagereferenceonthisviewisindividualistic.Butnow,tocontinuePutnam’sargument,imagineamolecule-for-moleculedoppelgangerofOscar,Oscar*,wholivesonaplanetjustlikeEarthinallrespectsbutone:thesubstancethatpeopledrink,washwith,andseefallingfromtheskyisnotwater(i.e.,H2O),butasubstancewithadifferentchemicalstruc-ture,XYZ.Callthisplanet“TwinEarth.”Thissubstance,whosechemicalcom-positionwemightdenotewith“XYZ,”iscalled“water”onTwinEarth,andOscar*,asadoppelgangerortwinofOscar,hasthesamebeliefsaboutitasOscarhasaboutwateronEarth.(RecallthatOscar,andthusOscar*ashistwin,havenospecialknowledgeofthechemicalstructureofwater.)TwinEarthhaswhatwemightcall“twin-water”or“twater”onit,notwater,anditistwaterthatOscar*interactswith,notwater–afterall,thereisnowateronTwinEarth.GiventhatOscar’sterm“water”referstoorisaboutwater,thenOscar*’sterm“water”referstoorisabouttwater.Thatis,theyhavenaturallanguagetermsthatdifferintheirmeaning,assumingthatreferenceisatleastoneaspectofmeaning.But,byhypothesis,OscarandOscar*aredoppelgangers,andsoareidenticalinalltheirintrinsicproperties,andsoareidenticalwithrespecttowhat’s“inthehead.”Thus,Putnamargues,themeaningofthenaturallanguagetermsthatOscarusesarenotmetaphysicallydeterminedbywhatisinOscar’shead.Putnam’stargetwasatraditionofthinkingaboutlanguagewhichwas,intermsthatPutnamappropriatedfromRudolphCarnap’sTheLogicalConstructionoftheWorld(1928),methodologicallysolipsistic:ittreatedthemeaningsofnaturallan-guagetermsandlanguagemoregenerallyinwaysthatdidnotsupposethatthe258\nIndividualismworldbeyondtheindividuallanguageuserexists.SincePutnam’schiefpointwasoneaboutnaturallanguagetermsandtherelationshipoftheirsemanticstowhat’sinsidethehead,oneneedsatleasttoextendhisreasoningfromlanguagetothoughttoarriveatapositionthatdeniesindividualismabouttheminditself.Butgiventhetraditiontowhichhewasopposed,suchanextensionmightbethoughttoberelativelytrivial,sinceineffectthoseinthetraditionofmethodo-logicalsolipsism–fromBrentano,toRussell,toHusserl,toCarnap–conceivedofnaturallanguagesandtheiruseinpsychologicalterms.Theintroductionoftheterm“individualism”itselfcanbefoundinTylerBurge’s“IndividualismandtheMental”(1979),whereBurgedevelopedaseriesofthoughtexperimentsinmanywaysparalleltoPutnam’sTwinEarththoughtexperiment.Burgeidentifiedindividualismasanoverallconceptionofthemindprevalentinmodernphilosophicalthinking,atleastsinceDescartesinthemid-seventeenthcentury,andarguedthatourcommon-sensepsychologicalframeworkforexplainingbehavior,ourfolkpsychology,wasnotindividualistic.Importantly,Burgewasexplicitinmakingacaseagainstindividualismthatdidnotturnonperhapscontroversialclaimsaboutthesemanticsofnaturalkindterms–hedevelopedhiscaseagainstindividualismusingagentswiththoughtsaboutarthritis,sofas,andcontracts–andsohisargumentdidnotpresupposeanytypeofscientificessentialismaboutnaturalkinds.LikePutnam’sargument,however,Burge’sargumentdoespresup-posesomeviewsaboutnaturallanguageunderstanding.Themostcentraloftheseisthatwecananddohaveincompleteunderstandingofmanyofthethingsthatwehavethoughtsaboutandforwhichwehavenaturallanguageterms.Giventhat,itispossibleforanindividualtohavethoughtsthatturnonthisincompleteunderstanding,suchasthethoughtthatonehasarthritisinone’sthighmuscle.Arthritisisadiseaseonlyofthejoints,oraswemightputit,“arthritis”inourspeechcommunityappliesonlytoadiseaseofthejoints.Consideranindividual,Bert,withthebeliefthathewouldexpressbysaying“Ihavearthritisinmythigh.”Intheactualworld,thisisabeliefaboutarthritis;itisjustthatBerthasanincompleteorpartiallymistakenviewofthenatureofarthritis,andsoexpressesafalsebeliefwiththecorrespondingsentence.ButnowimagineBertaslivinginadifferentspeechcommunity,oneinwhichtheterm“arthritis”doesapplytoadiseaseofboththejointsandofotherpartsofthebody,includingthethigh.Inthatspeechcommunity,Bert’sthoughtwouldnotinvolvethesortofincompleteunderstandingthatitinvolvesintheactualworld;infact,histhoughtinsuchaworldwouldbetrue.Giventhedifferencesinthetwospeechcommunities,itseemsthatanindividualwiththoughtsaboutwhathecalls“arthritis”willhavedifferentthoughtsinthetwocommunities:intheactualworld,Berthasthoughtsaboutarthritis,whileinthecounterfactualworldhehasthoughtsaboutsomeotherdisease–whatwemightrefertoas“tharthritis,”todistinguishitfromthediseasethatwehaveintheactualworld.Inprinciple,wecouldsupposethatBerthimselfisidenticalacrossthetwocontexts–thatis,heisidenticalinallintrinsicrespects.YetweattributethoughtswithdifferentcontentstoBert,andseemtodososolelybecauseofthedifferences259\nRobertA.Wilsoninthelanguagecommunityinwhichheislocated.Thus,thecontentofone’sthoughtsisnotmetaphysicallydeterminedbytheintrinsicpropertiesoftheindi-vidual.Andagaintakingadifferenceinthecontentoftwothoughtstoimplyadifferencebetweenthethoughtsthemselves,thisimpliesthatthoughtsarenotindividuatedindividualistically.Onecontrastthatissometimes(e.g.,Segal2000:chs.2–3)drawnbetweentheanti-individualisticviewsofPutnamandBurgeistocharacterizePutnam’sviewasaformofphysicalexternalismandBurge’sviewasaformofsocialexternalism:accordingtoPutnam,itisthecharacterofthephysicalworld(e.g.,thenatureofwateritself)that,inpart,metaphysicallydeterminesthecontentofone’smind,whileaccordingtoBurgeitisthecharacterofthesocialworld(e.g.,thenatureofone’slinguisticcommunity)thatdoesso.Whilethisdifferencemayserveasausefulreminderofonewayinwhichthesetwoviewsdiffer,weshouldalsokeepinmindthe“social”aspecttoPutnam’sviewofnaturallanguageaswell:hislinguisticdivisionoflabor.Importanttobothviewsistheideathatlanguageusersandpsychologicalbeingsdependandrelyononeanotherinwaysthatarereflectedinoureveryday,common-sensewaysofthinkingaboutlanguageandthought.Thusthereisasocialaspecttothenatureofmeaningandthoughtonbothviews,andthisisinpartwhatjustifiestheappropriatenessofthelabelanti-individualismforeachofthem.11.3TheCognitiveScienceGesturePhilosopherswhoseethemselvesascontributingtocognitivesciencehaveoccu-piedthemostactivearenainwhichthedebatebetweenindividualistsandexternalistshasbeenplayedout.AtaroundthetimethatindividualismwascomingunderattackfromPutnamandBurge,itwasalsobeingdefendedasaviewofthemindparticularlyaptforagenuinelyscientificapproachtounder-standingthemind,especiallyofthetypebeingarticulatedwithinthenascentinterdisciplinaryfieldofcognitivescience.Forthoseofferingthisdefense,therewassomethingsuspiciouslyunnaturalisticaboutthePutnam–Burgearguments,aswellassomethingabouttheirconclusionsthatseemedanti-scientific,andpartofthedefenseofindividualismandthecorrespondingattackonexternalismturnedonwhatIwillcallthecognitivesciencegesture:theclaimthat,ascontemporaryempiricalworkoncognitionindicated,anytrulyscientificunderstandingofthemindwouldneedtobeindividualistic.PickinguponPutnam’suseof“methodologicalsolipsism”,JerryFodordefendedmethodologicalsolipsismasthedoctrinethatpsychologyoughttoconcernitselfonlywithnarrowpsychologicalstates,wherethesearestatesthatdonotpre-suppose“theexistenceofanyindividualotherthanthesubjecttowhomthatstateisascribed”(Fodor1980:244).Fodorsawmethodologicalsolipsismasthepreferredwaytothinkofpsychologicalstates,givenespeciallytheChomskyan260\nIndividualismrevolutioninlinguisticsandtheaccompanyingcomputationalrevolutioninpsy-chology.Ifmentalstatesweretransitionsgovernedbycomputationalrules,thenthetaskofthecognitivescienceswouldbetospecifythoserules;insofarasmentalstateswerecomputational,broaderconsiderationsaboutthephysicalorsocialworldsinwhichanindividualislocatedseemirrelevanttothatindividual’spsychologicalnature.StephenStich’s(1978)principleofautonomyprovidesanalternativewaytoarticulateanindividualisticviewofcognitivescience,variationsonwhichhavebecomethestandardwaystoformulateindividualism.Theprinciplesaysthat“thestatesandprocessesthatoughttobeofconcerntothepsychologistarethosethatsuperveneonthecurrent,internal,physicalstateoftheorganism”(Stich1983:164–5).Thenotionofsupervenienceprovidesamoreprecisewaytospecifythetypeofmetaphysicaldeterminationthatweintroducedearlier.Asetofproperties,S(thesuperveningproperties),supervenesonsomeothersetofproperties,B(thebaseproperties),justifanythingthatisidenticalwithrespecttotheBpropertiesmustalsobeidenticalwithrespecttotheSproperties.Inpartbecauseoftheprominenceofsupervenienceinformulatingversionsofphysicalism,togetherwiththeperceivedlinkbetweenphysicalismandindividualism(moreofwhichinamoment),butalsoinpartbecauseoftheemphasisondoppelgangersinthePutnamandBurgearguments,ithasbecomemosttypicaltoexpressindividualismanditsdenialintermsofoneoranothersupervenienceformulation.CommontobothFodorandStich’sviewsofcognitivescienceistheideathatanindividual’spsychologicalstatesshouldbebracketedofffromthemere,beyond-the-headenvironmentsthatindividualsfindthemselvesin.UnlikePutnamandBurgeinthepapersdiscussedabove,FodorandStichhavefocusedontherelevanceofindividualismforexplanatorypracticeinpsychology,usingtheirrespectiveprinciplestoargueforsubstantiveconclusionsaboutthescopeandmethodologyofpsychologyandthecognitivesciences.Fodorcontrastedasolipsisticpsychologywithwhathecalledanaturalisticpsychology,arguingthatsincethelatter(amongstwhichheincludedJ.J.Gibson’sapproachtopercep-tion,learningtheory,andthenaturalismofWilliamJames)wasunlikelytoproveareliableresearchstrategyinpsychology,methodologicalsolipsismprovidedtheonlyfruitfulresearchstrategyforunderstandingcognition(seealsoFodor1987).Sticharguedforasyntacticorcomputationaltheoryofmindwhichmadenoessentialuseofthenotionofintentionalityormentalcontentatall,andsousedtheprincipleofautonomyindefenseofaneliminativistviewaboutcontent(seealsoStich1983).AlthoughIthinkthatthecognitivesciencegestureisagesture(ratherthanasolidargumentthatappealstoempiricalpractice),itisnotanemptygesture.WhileFodor’sandStich’sargumentshavenotwonwidespreadacceptanceineitherthephilosophicalorcognitivesciencecommunities,theyhavestruckachordwiththoseworkingincognitivescience,perhapsnotsurprisinglysincethedominantresearchtraditionsincognitivesciencehavebeenatleastimplicitlyindividual-istic.Relativelyexplicitstatementsofacommitmenttoanindividualisticviewof261\nRobertA.WilsonaspectsofcognitivescienceincludeChomsky’s(1986,1995,2000)deploymentofthedistinctionbetweentwoconceptionsoflanguage(the“I”-languageandthe“E”-language,for“internal”and“external”,respectively),Jackendoff’s(1991)related,generaldistinctionbetween“psychological”and“philosophical”concep-tionsofthemind,andCosmidesandTooby’s(1994)emphasisontheconstruc-tivenatureofourinternal,evolutionary-specializedcognitivemodules.Partoftheattractionofindividualismforpracticingcognitivescientistsisitsperceivedconnectiontotherepresentationaltheoryofmind,whichholdsthatweinteractwiththeworldperceptuallyandbehaviorallythroughinternalmentalrepresentationsofhowtheworldis(astheeffectsofperceiving)orhowtheworldshouldbe(asinstructionstoact).Jackendoffexpressessuchaviewwhenhesays:Whateverthenatureofrealreality,thewayrealitycanlooktousisdeterminedandconstrainedbythenatureofourinternalmentalrepresentations....Physicalstimuli(photons,soundwaves,pressureontheskin,chemicalsintheair,etc.)actmechan-icallyonsensoryneurons.Thesensoryneurons,actingastransducersinPylyshyn’s(1984)sense,setupperipherallevelsofrepresentationsuchasretinalarraysandwhateveracousticanalysistheearderives.Inturn,theperipheralrepresentationsstimulatetheconstructionofmorecentrallevelsofrepresentation,leadingeventu-allytotheconstructionofrepresentationsincentralformatssuchasthe3Dlevelmodel.(1991:159–61)Providedthattheappropriate,internal,representationalstatesoftheorganismremainfixed,theorganism’smoreperipheralcausalinvolvementwithitsenviron-mentisirrelevanttocognition,sincetheonlywayinwhichsuchcausalinvolve-mentcanmattertocognitionisbyalteringtheinternalmentalstatesthatrepresentthatenvironment.11.4Functionalism,Physicalism,andIndividualismFormanyphilosophersinterestedinthecognitivesciences,individualismhasbeenattractivebecauseofaperceivedconnectionbetweenthatviewandbothphysicalismandfunctionalisminthephilosophyofmind,bothofwhichhavebeenwidelyacceptedsincethe1980s.Physicalism(ormaterialism)isaviewthathasbeenexpressedinvariousways,perhapsthemostcommonofwhichisintermsofthenotionofsupervenience:allfacts,properties,processes,events,andthingssuperveneonthephysicalfacts,properties,processes,events,andthings,astheyarepositedinelementaryphysics.Thisontologicalformulationofphysicalism(concernedwithwhatexists)isoftenaccompaniedbyanexplanatorythesis,whichstatesthatphysicalexplanationsare,insomesense,theultimateexplanationsforanyphenomenonwhatsoever.262\nIndividualismIndividualismhasbeenthoughttobelinkedtophysicalism,sinceitimplies,viathesupervenienceformulation,thatthereisnopsychologicaldifferencewithoutacorrespondingdifferenceintheintrinsic,physicalstatesoftheindividual.Thoserejectingindividualismhavesometimesbeencharged(e.g.,byBlock1986andFodor1987:ch.2)withendorsingsomeformofdualismaboutthemind,ormakingamysteryofmentalcausationbyignoringormisconstruingtheroleofcausalpowersinpsychologicaltaxonomy.Connectingthisupwiththemethodo-logicalformulationsthathavehadinfluenceincognitivescienceitself,individual-ismhasbeenclaimedtobeaminimalconstraintonarrivingatpsychologicalexplanationsthatlocatethemindsuitablyinthephysicalworld,apsychologythattaxonomizesitsentitiesbytheircausalpowers.(Wehaveseen,however,thatindividualiststhemselvesdisagreeaboutwhatthisimpliesaboutpsychology.)Functionalismistheviewthatpsychologicalstatesandprocessesshouldbeindividuatedbytheircausalorfunctionalroles–thatis,bytheirplacewithintheoverallcausaleconomyoftheorganism–andithasbeencommontosupposethatthesefunctionalorcausalrolesareindividualistic.Certainly,thesecausalrolescanbeunderstoodindifferentways,butthetwo(complementary)waysmostprevalentincognitivescience–intermsofthenotionofcomputation(e.g.,Fodor1980;Pylyshyn1984),andintermsoftheideaofanalyticaldecomposi-tion(e.g.,Dennett1978;Cummins1983)–lendthemselvestoanindividualisticreading.Computationalprocesses,operatingsolelyonthesyntacticpropertiesofmentalstates,havebeenplausiblythoughttobeindividualistic;anditisnaturaltothinkofanalyticaldecompositionasbeginningwithapsychologicalcapacity(e.g.,memory,depthperception,reasoning)andseekingtheintrinsicpropertiesoftheorganisminvirtueofwhichitinstantiatesthatcapacity.Despitetheirprimafacieplausibility,however,neitheroftheseconnections–betweenphysicalismandindividualism,andbetweenfunctionalismandindividu-alism–isunproblematic,andinfactIthinkthatuponcloserexaminationneitherpurportedinferenceholds.Theseclaimscanbeexploredmorefullybyexaminingexplicitargumentsforindividualismthatspecifytheseconnectionsmoreprecisely.11.5TheAppealtoCausalPowersAnargumentforindividualismthathasbeenwidelydiscussedderivesfromchap-ter2ofFodor’sPsychosemantics(1987).Althoughaseriesofrelatedcriticisms(vanGulick1989;Egan1991;Wilson1992,1995:ch.2)seemtomedecisiveinshowingtheargumenttobefatallyflawed,theargumentitselftapsintoanintuition,orperhapsaclusterofintuitions,runningdeepinthephilosophicalcommunity.Theargumentitselfiseasytostate.Taxonomyorindividuationinthesciencesingeneralsatisfiesageneralizedversionofindividualism:sciencestaxonomizetheentitiestheypositanddiscoverbytheircausalpowers.Psychologyandthecognitivesciencesshouldbenoexceptionhere.Butthecausalpowersof263\nRobertA.Wilsonanythingsuperveneonthatthing’sintrinsic,physicalproperties.Thus,scientifictaxonomy,andsopsychologicaltaxonomy,mustbeindividualistic.Onewaytoelicittheproblemwiththisargumentistoaskwhatitisthatmakesthefirstpremise(aboutscientifictaxonomyingeneral)true.Giventhenaturalisticturnsupposedlyembracedbythoseworkingincontemporaryphilosophicalpsycho-logy,onewouldthinkthatthesupportherecomesfromanexaminationofactualtaxonomicpracticeacrossthesciences.However,onceonedoesturntolookatthesepractices,itiseasytofindavarietyofsciencesthatdon’ttaxonomize“bycausalpowers;”rather,theyindividuatetheirkindsrelationally,whereoftenenoughitistheactualrelationsthatdeterminekindmembership.Examplesoftencitedhereincludespeciesinevolutionarybiology,whichareindividuatedphylogenetically(andsohistorically),andcontinentsingeology,whosecausalpowersareprettymuchirrelevanttotheiridentityascontinents(seeBurge1986a).Theproblemisparticularlyacuteinthecontextofthisargumentforindividualism,sinceafurtherpremiseintheargumentstatesthatathing’scausalpowerssuperveneonthatthing’sintrinsicproperties,andsoonecannotsimplystipulatethatindividuationinthesesciencesis“bycausalpowers”insomeextendedsenseof“causalpowers.”(Ifonedoesthat,then“causalpowers”nolongersosupervene.)Theintuitionthatpersistsdespiteanacknowledgmentthattheargumentitselfisflawedinsomethinglikethewayidentifiedaboveisthatindividualismdoesarticulateaconstraintfortheexplanationofcognitionthatsciencesmoregener-allysatisfy,onethatwouldmakeforaphysicalisticallyrespectablepsychology(e.g.,seeWalsh1999).Myviewisthatthisintuitionitselfseriouslyunderes-timatesthediversityintaxonomicandexplanatorypracticeacrossthesciences(seeWilson2000b),anditsimplyneedstobegivenup.Attemptstorevitalizethissortofargumentforindividualismproceedbymakingthesortsofaprioriassumptionsaboutthenatureofscientifictaxonomiesandexplanationsthatarereminiscentofthegeneralized,rationalreconstructionsofscientificpracticethatgovernedlogicalpositivistviewsofscience,andthisshouldsoundalarmbellsforanyself-professedcontemporarynaturalisticphilosopherofmind.11.6ExternalismandMetaphysicsWhat,then,ofthemoregeneral,putativeconnectionbetweenphysicalismandindividualism?Ifthedenialofindividualismcouldbeshowntoentailthedenialofaplausiblygeneralversionofphysicalism,thenIthinkthatexternalismwoulditselfbeinrealtrouble.Butliketheindividualist’sappealtocausalpowersandscientifictaxonomy,Isuspectthatthemovefromthegeneralintuitionsthatmotivatesuchanargumenttotheargumentitselfwillitselfproveproblematic.Forexample,externalistscanrespectthephysicalistslogan“nopsychologicaldifferencewithoutaphysicaldifference”becausetherelevantphysicaldifferencesliebeyondtheboundaryoftheindividual;attemptstorefinethisslogan(e.g.,no264\nIndividualismpsychologicaldifferencewithoutahere-and-nowphysicaldifference)arelikelyeithertobegthequestionagainsttheexternalistorinvokeaconstrualofphysicalismthatisatleastascontroversialasindividualismitself.Whatistrueisthatexternaliststhemselveshavenotbeenasattentivetothemetaphysicalnotionsatthecoreofcontemporarymaterialismastheycouldhavebeen,andwhentheyhavesoattendedtheyhavesometimessoundedopposedtophysicalism.ThemostprominentcasehereisBurge’s(1979)originaldiscussionoftheimplicationsofindividualismforrelatedviewsaboutthemind,whereheclaimedthattherejectionofindividualismimpliedtherejectionofbothtype-typeandtoken-tokenidentitytheoriesofthemind,thesebeingtwoofthemajorformsofmaterialism.Tomymind,themostunder-discussedofthesenotionsisthatofrealization.Althoughithasbeencommontoexpressmaterialismasentailingthatallmentalstatesarerealizedasphysicalstates,andtotaketherelevantphysicalstatestobestatesofthebrain,therehasbeenlittlegeneraldiscussionofthepropertiesofthisrelationofrealization,orofthepropertiesofrealizerstates(seeShoemaker2000,Gillett2002,though).Thiscreatesaproblemforexternalists,sincethestandardviewofrealizationsmugglesinanindividualisticbias.Onthisstandardview,realizersareheldtobebothmetaphysicallysufficientforthestatestheyrealizeandphysicallyconstitutiveoftheindividualswiththerealizedproperties.Denyingthesecondoftheseconjuncts,asIthinkanexternalistshould,createsspacefortheideathatmentalstateshaveawiderealization,anoptionthatIhaveattemptedelsewheretodefendinthecontextofamoregeneraldiscussionofrealization(Wilson2001).11.7TheDebateOverMarr’sTheoryofVisionIhavealreadysaidthatindividualismreceivesprimafaciesupportfromthecom-putationalandrepresentationaltheoriesofmind,andthusfromthecognitivesciencecommunityinwhichthosetheorieshavebeeninfluential.ButIhavealsoindicatedthatIthinkthattheclaimthatatrulyexplanatorycognitivesciencewillbeindividualistichasanepistemicbasismorelikeagesturethanaproof.OnewaytosubstantiatethissecondviewinlightofthefirstistoturntoexaminethecontinuingphilosophicaldebateoverwhetherDavidMarr’scelebratedtheoryofearlyvisionisindividualistic.Apartfromtheintrinsicinterestofthedebateitself,ourexaminationherewillalsohelptoelicitsomeofthebroaderissuesaboutthemindtowhichtheindividualismissueiscentral,includingthenatureofcompu-tationandrepresentation.Inthefinalsectionof“IndividualismandtheMental,”Burgehadsuggestedthathisthoughtexperimentandtheconclusionderivedfromit–thatmentalcontentandthusmentalstateswithcontentwerenotindividualistic–hadimpli-cationsforcomputationalexplanationsofcognition.Theseimplicationswere265\nRobertA.Wilsontwofold.First,purelycomputationalaccountsofthemind,construedindividual-istically,wereinadequate;andsecond,insofarassuchexplanationsdidappealtoanotionofmentalcontent,theywouldfailtobeindividualistic.ItisthelatteroftheseideasthatBurgepursuedin“IndividualismandPsychology”(1986a),inwhichheargued,strikingly,thatMarr’stheoryofvisionwasnotindividualistic.Thiswasthefirstattempttoexploreawidelyrespectedviewwithincognitivesciencevis-à-vistheindividualismissue,anditwasacrucialturningpointinmovingbeyondthecognitivesciencegesturetowardastyleofargumentthatreallydoesutilizeempiricalpracticeincognitivescienceitself.Ashasoftenbeenpointedout,whatiscalled“Marr’stheoryofvision”isanaccountofarangeofprocessesinearlyor“low-level”visionthatwasdevelopedbyMarrandcolleagues,suchasEllenHildrethandTomasPoggio,attheMassa-chussettsInstituteofTechnologyfromthemid-1970s.Theseprocessesincludestereopsis,theperceptionofmotion,andshapeandsurfaceperception,andtheapproachisexplicitlycomputational.Marr’sVision:AComputationalInvestiga-tionintotheHumanRepresentationandProcessingofVisualInformation(1982),publishedposthumouslyafterMarr’stragicearlydeathin1980,becametheparadigmexpressionoftheapproach,particularlyforphilosophers,somethingfacilitatedbyMarr’scomfortableblendofcomputationaldetailwithbroad-brushed,programmaticstatementsoftheperspectiveandimplicationsofhisapproachtounderstandingvision.SincethepublicationofMarr’sbook,workonhistheoryofvisionhascontinued,beingextendedtocovertheprocessesconstitutinglow-levelvisionmoreextensively(e.g.,seeHildrethandUllman1989).Interestingly,byandlarge,thephilosophicalliteratureonindividualismthatappealstoMarr’stheoryhasbeencontenttorelyalmostexclusivelyonhisVisionininterpretingthetheory.CriticaltothecomputationaltheorythatMarradvocatesisarecognitionofthedifferentlevelsatwhichonecan–indeed,forMarr,must–studyvision.AccordingtoMarr,therearethreelevelsofanalysistopursueinstudyinganinformation-processingdevice.First,thereisthelevelofthecomputationaltheory(hereafter,thecomputationallevel),whichspecifiesthegoalofthecomputa-tion,andatwhichthedeviceitselfischaracterizedinabstract,formaltermsas“mappingfromonekindofinformationtoanother”(1982:24).Secondisthelevelofrepresentationandalgorithm(hereafter,thealgorithmiclevel),whichselectsa“representationfortheinputandoutputandthealgorithmtobeusedtotransformoneintotheother”(ibid.:24–5).Andthirdisthelevelofhardwareimplementation(hereafter,theimplementationallevel),whichtellsushowtherepresentationandalgorithmarerealizedphysicallyintheactualdevice.Philosophicaldiscussions,likeMarr’sowndiscussions,havebeenfocusedonthecomputationalandalgorithmiclevelsforvision,whatMarrhimself(ibid.:23)characterizes,respectively,asthe“whatandwhy”and“how”questionsaboutvision.Aswewillsee,thereisparticularcontroversyoverwhatthecomputationallevelinvolves.Inadditiontotheoften-invokedtrichotomyoflevelsatwhichaninformational-processinganalysisproceeds,therearetwofurtherinterestingdimensionstoMarr’sapproachtovisionthathavebeensomewhatneglectedin266\nIndividualismthephilosophicalliterature.TheseaddsomecomplexitynotonlytoMarr’stheory,butalsototheissueofhow“computation”and“representation”aretobeunderstoodinit.Thefirstistheideathatvisualcomputationsareperformedsequentiallyinstagesofcomputationalinference.Marrstatesthattheoverallgoalofthetheoryofvisionis“tounderstandhowdescriptionsoftheworldmayefficientlyandreliablybeobtainedfromimagesofit”(ibid.:99).Marrviewstheinferencesfrominten-sitychangesintheretinalimagetofull-blownthree-dimensionaldescriptionsasproceedingviatheconstructionofaseriesofpreliminaryrepresentations:theraw1primalsketch,thefullprimalsketch,andthe2/2-Dsketch.Callthisthetemporaldimensiontovisualcomputation.Thesecondisthatvisualprocessingissubjecttomodulardesign,andsoparticularaspectsoftheconstructionof3-Dimages–stereopsis,depth,motion,etc.–canbeinvestigatedinprincipleindependently.Callthisthemodulardimensiontovisualcomputation.Arecognitionofthetemporalandmodulardimensionstovisualcomputationcomplicatesanydiscussionofwhat“the”computationalandalgorithmiclevelsfor“the”processofvisionare.Minimally,inidentifyingeachofMarr’sthreelevels,weneedfirsttofixatleastthemodulardimensiontovisioninordertoanalyzeagivenvisualprocess;andtofixatleastthetemporaldimensioninordertoanalyzeagivenvisualcomputation.Burge’sargumentthatMarr’stheoryisnotindividualisticisexplicitlyandfullypresentedinthefollowingpassage:(1)Thetheoryisintentional.(2)Theintentionalprimitivesofthetheoryandtheinformationtheycarryareindividuatedbyreferencetocontingentlyexistingphysicalitemsorconditionsbywhichtheyarenormallycausedandtowhichtheynormallyapply.(3)Soifthesephysicalconditionsand,possibly,attendantphysicallawswereregularlydifferent,theinformationconveyedtothesubjectandtheintentionalcontentofhisorhervisualrepresentationswouldbedifferent.(4)Itisnotincoher-enttoconceiveofrelevantlydifferent(say,optical)lawsregularlycausingthesamenon-intentionally,individualisticallyindividuatedphysicalregularitiesinthesubject’seyesandnervoussystem....(5)Insuchacase(by(3))theindividual’svisualrepresentationswouldcarrydifferentinformationandhavedifferentrepresentationalcontent,thoughtheperson’swholenon-intentionalphysicalhistory...mightremainthesame.(6)Assumingthatsomeperceptualstatesareidentifiedinthetheoryintermsoftheirinformationalorintentionalcontent,itfollowsthatindividualismisnottrueforthetheoryofvision.(1986a:34)ThesecondandthirdpremisemakespecificclaimsaboutMarr’stheoryofvision,whilethefirstpremise,togetherwith(4)and(5),indicatetheaffinitybetweenthisargumentandtheTwinEarth-styledargumentofBurge’sthatwediscussedearlier.Burgehimselfconcentratesondefending(2)–(4),largelybyanappealtothewaysinwhichMarrappearstorelyon“thestructureoftherealworld”inarticulatingboththecomputationalandalgorithmiclevelsforvision.MarrcertainlydoesmakeanumberofappealstothisstructurethroughoutVision.Forexample,hesays267\nRobertA.WilsonThepurposeoftheserepresentationsistoprovideusefuldescriptionsofaspectsoftherealworld.Thestructureoftherealworldthereforeplaysanimportantroleindeterminingboththenatureoftherepresentationsthatareusedandthenatureoftheprocessesthatderiveandmaintainthem.Animportantpartofthetheoreticalanalysisistomakeexplicitthephysicalconstraintsandassumptionsthathavebeenusedinthedesignoftherepresentationsandprocesses.(1982:43;cf.alsopp.68,103–5,265–6)AndMarrdoesclaimthattherepresentationalprimitivesinearlyvision–suchas“blobs,lines,edges,groups,andsoforth”–thatheposits“correspondtorealphysicalchangesontheviewedsurface”(ibid.:44).Together,thesesortsofcommenthavebeentakentosupport(2)and(3)inparticular.MuchofthecontroversyoverhowtointerpretMarr’stheoryturnsonwhetherthisisthecorrectwaytounderstandhisappealstothe“structureoftherealworld.”Thereareatleasttwogeneralalternativestoviewingsuchcommentsasclaimingtheimportanceofthebeyond-the-headworldforthecomputationaltaxonomyofvisualstates.Thefirstistoseethemasgivingtherealworldaroletoplayonlyinconstruct-ingwhatMarrcallsthecomputationaltheory.Sincevisionisaprocessforextract-inginformationfromtheworldinordertoallowtheorganismtoacteffectivelyinthatworld,clearlyweneedtoknowsomethingofthestructureoftheworldinouraccountofwhatvisionisfor,whatitisthatvisiondoes,whatfunctionvisionisdesignedtoperform.Ifthisiscorrect,thenitseemspossibletoarguethatonedoesnotneedtolookbeyondtheheadinconstructingthetheoryoftherepres-entationandalgorithm.Asitisatthislevelthatvisualstatesaretaxonomizedquatheobjectsofcomputationalmechanisms,Marr’sreferencestothe“realworld”donotcommithimtoanexternalistviewofthetaxonomyofvisualstatesandprocesses.Thesecondistotakethesecommentstosuggestmerelyaheuristicroleforthestructureoftherealworld,notonlyindevelopingacomputationaltaxonomybutinthecomputationaltheoryofvisionmoregenerally.Thatis,turningtothebeyond-the-headworldismerelyausefulshort-cutforunderstandinghowvisionworksandthenatureofvisualstatesandcomputations,eitherbyprovidingimportantbackgroundinformationthatallowsustounderstandtherepresenta-tionalprimitivesandthustheearlieststagesofthevisualcomputation,orbyservingasinterpretativelensesthatallowustoconstructamodelofcomputa-tionalprocessesintermsthataremeaningful.Again,aswiththepreviousoption,thebeyond-the-headworldplaysonlyaperipheralrolewithincomputationalvision,evenifMarrattimesreferstoitprominentlyinoutlininghistheory.IndividualistshaveobjectedtoBurge’sargumentintwoprincipalways.First,Segal(1989)andMatthews(1988)havebothineffectdenied(2),withSegalarguingthattheseintentionalprimitives(suchasedgesandgeneralizedcones)arebetterinterpretedwithinthecontextofMarr’stheoryasindividuatedbytheirnarrowcontent.Second,Egan(1991,1992,1995,1999)hasmorestrikingly268\nIndividualismdenied(1),arguingthat,quacomputationaltheory,Marr’stheoryisnotinten-tionalatall.Bothobjectionsareworthexploringindetail,particularlyinsofarastheyhighlightissuesthatremaincontentiousincontemporarydiscussions.Infact,thediscussionofMarr’stheoryraisesmorefoundationalquestionsthanitsolvesaboutthenatureofthemindandhowweshouldinvestigateit.Segalpointsoutthattherearetwogeneralinterpretationsavailablewhenoneseekstoascribeintentionalcontentstothevisualstatesoftwoindividuals.First,onecouldfollowBurgeandinterpretthecontentofagivenvisualstateintermsofwhatnormallycausesit.Thus,ifitisacrackinasurfacethatplaysthisrole,thenthecontentofthecorrespondingvisualstateis“crack;”ifitisashadowintheenvironmentthatdoesso,thenthecontentofthevisualstateis“shadow.”Thiscouldbesoeveninthecaseofdoppelgangers,andsothevisualstatessoindividuatedarenotindividualistic.Butsecondandalternatively,onecouldofferamoreliberalinterpretationofthecontentofthevisualstatesinthetwocases,onethatwasneutralastothecauseofthestate,andtowhichwemightgivethename“crackdow”toindicatethisneutrality.Thiscontentwouldbesharedbydoppelgangers,andsowouldbeindividualistic.ThecrucialpartofSegal’sargumentishiscaseforpreferringthesecondoftheseinterpretations,anditisherethatonewouldexpecttofindanappealtothespecificsofMarr’stheoryofvision.WhilesomeofSegal’sargumentsheredosoappeal,healsointroducesanumberofquitegeneralconsiderationsthathavelittletodowithMarr’stheoryinparticular.Forexample,hepointstothesecondinterpretationashaving“economyonitsside”(1989:206),thusappealingtoconsiderationsofsimplicity,andsays:Thebesttheoreticaldescriptionwillalwaysbeoneinwhichtherepresentationsfailtospecifytheirextensionsatalevelthatdistinguishesthetwosortsofdistalcause.Itwillalwaysbebettertosupposethattheextensionincludesbothsortsofthing.(ibid.:207;myemphasis)Why“always”?Segaltalksgenerallyofthe“basiccanonsofgoodexplanation”(ibid.)insupportofhiscaseagainstexternalism,butaswiththeappealstothenatureofscientificexplanationthatturnedontheideathatscientifictaxonomyandthusexplanationindividuatesby“causalpowers,”hereweshouldbesuspi-ciousofthelevelofgenerality(andcorrespondinglackofsubstantivedetail)atwhichscientificpracticeisdepicted.LikeBurge’sownappealtotheobjectivityofperceptualrepresentationinformulatingageneralargumentforexternalism(1986a:section3;1986b),thesesortsofaprioriappealsseemtometorepresentgesturallapsesentwinedwiththemoreinteresting,substantive,empiricalargumentsoverindividualisminpsychology.WhenSegaldoesdrawmoreexplicitlyonfeaturesofMarr’stheory,heextractsthreegeneralpointsthatarerelevantforhisargumentthatthetheoryisindividu-alistic:eachattributionofarepresentationrequiresa“bottom-upaccount”(1989:194),a“top-downmotivation”(ibid.:195)andis“checkedagainstbehavioral269\nRobertA.Wilsonevidence”(ibid.:197).Together,thesethreepointsimplythatpositingrepresen-tationsinMarr’stheorydoesnotcomecheaply,andindeedistightlyconstrainedbyoveralltaskdemandsandmethods.Thefirstsuggeststhatanyhigher-levelrepresentationspositedbythetheorymustbederivedfromlower-levelinputrepresentations;thesecondthatallpositedrepresentationsderivetheirmotiva-tionfromtheirroleintheoverallperceptualprocess;andthethirdthat“inten-tionalcontentsareinferredfromdiscriminativebehavior”(ibid.:197).Segalusesthefirstassumptiontoarguethatsincethecontentoftheearliestrepresentations–“uptoandincludingzero-crossings”(ibid.:199)–indoppelgangersarethesame,thereisaprimafaciecasethatdownstream,higher-levelrepresentationsmustbethesame,unlessatop-downmotivationcanbegivenforpositingadifference.Butsinceweareconsideringdoppelgangers,thereisnobehavioralevidencethatcouldbeusedtodiagnosearepresentationaldiffer-encebetweenthetwo(Segal’sthirdpoint),andsonotop-downmotivationavailable.Ashesays,“[t]herewouldjustbenotheoreticalpointininvokingthetwocontents[ofthetwins],whereonewoulddo.Fortherewouldbenothe-oreticalpurposeservedbydistinguishingbetweenthecontents”(ibid.:206).Howmightanexternalistresistthischallengingargument?Threedifferenttackssuggestthemselves,eachofwhichgrantslesstoSegalthanthatwhichprecedesit.First,onecouldgrantthethreepointsthatSegalextractsfromhisreadingofMarr,togetherwithhisclaimthatthelowestlevelsofrepresentationareindividu-alistic,butquestionthesignificanceofthis.HereonecouldagreethatthegrayarrayswithwhichMarr’stheorybeginsdo,inasense,representlightintensityvalues,andthatzero-crossingsdo,inthatsamesense,representasuddenchangeinthelightintensity.Butthesearebothmerelyrepresentationsofsomestateoftheretina,notoftheworld,anditshouldbenosurprisethatsuchintra-organismicrepresentationshavenarrowcontent.Moreover,thedepthoftheintentionalityor“aboutness”ofsuchrepresentationsmightbecalledintoquestionpreciselybe-causetheydon’tinvolveanycausalrelationthatextendsbeyondthehead;theymightbethoughttoberepresentationalinmuchthewaythatmygrowlingstomachrepresentsmycurrentstateofhunger.However,oncewemovetodownstreamprocesses,processesthatarelateroninthetemporaldimensiontovisualprocessing,genuinelyrobustrepresentationalprimitivescomeintoplay,primitivessuchas“edge”and“generalizedcone.”Andthecontentsofstatesdeployingtheseprimitives,onemightclaim,asrepresentationsofastateoftheworld,metaphysicallydependonwhattheycorrespondtointheworld,andsoarenotindividualistic.TheplausibilityofthisresponsetoSegalturnsonboththestrengthofthedistinctionbetweenaweakerandastrongersenseof“representa-tion”inMarr’stheory,andtheclaimthatweneedthestrongersensetohavestatesthatarerepresentationalinsomephilosophicallyinterestingsense.Secondly,andmoreradically,onecouldallowthatalloftherepresentationalprimitivespositedinthetheoryrepresentinthesamesense,butchallengetheclaimthatthecontentofanyofthecorrespondingstatesisnarrow:itiswide270\nIndividualismcontentallthewayout,ifyoulike.Theideathattherepresentationalcontentofstatesdeployinggrayarraysandzero-crossingsisinfactwidemightitselftakeitscuefromSegal’ssecondpoint–thatrepresentationsrequireatop-downmotiva-tion–foritisbyreflectingonthepointoftheoverallprocessofconstructingreliable,three-dimensionalimagesofathree-dimensionalvisualworldthatwecanseethatevenearlyretinalrepresentationsmustberepresentationsofstatesandconditionsintheworld.ThisviewwouldofnecessitygobeyondMarr’stheoryitself,whichisexplicitlyconcernedonlywiththecomputationalproblemofhowweinferthree-dimensionalimagesfromimpoverishedretinalinformation,butwouldbe,Ithink,verymuchinthespiritofwhatwecanthinkofasaGibsonianaspecttoMarr’stheory(cf.Shapiro1993).Thirdly,andleastcompromisingly,onecouldrejectoneormoreofSegal’sthreepointsaboutMarr’stheoryor,rather,thesignificancethatSegalattachestothesepoints.Temporallylaterrepresentationsarederivedfromearlierrepresentations,butthisitselfdoesn’ttellusanythingabouthowtoindividuatethecontentsofeither.Likewise,thatMarrhimselfbeginswithlow-levelrepresentationsoftheretinalimagetellsuslittleaboutwhethersuchrepresentationsarenarroworwide.Top-downmotivationsareneededtojustifythepostulationofrepresentations,butsincethereisarangeofmotivationswithinMarr’stheoryconcerningtheoverallpointoftheprocessofthree-dimensionalvision,thisalsogivesuslittleguidanceaboutwhetherthecontentofsuchrepresentationsisnarroworwide.Behavioralevidencedoesplayaroleindiagnosingthecontentofparticularrepresentations,butsinceMarrisnotabehaviorist,behavioraldiscriminationdoesnotprovidealitmustestforrepresentationaldifference(Shapiro1993:498–503).Thisthirdresponseseemsthemostplausibletodevelopindetail,butitalsoseemstometheonethatimpliesthatthereislikelytobenodefinitiveanswertothequestionofwhetherMarr’stheoryemployseitheranarroworawidenotionofcontent,orbothorneither.AlthoughMarrwasnotconcernedatallhimselfwiththeissueoftheintentionalnatureoftheprimitivesofthistheory,thedepthofhismethodologicalcommentsandasideshasleftuswithanembarrassmentofricheswhenitcomestopossibleinterpretationsofhistheory.ThisisnotsimplyanindeterminacyaboutwhatMarrmeantorintended,butonewithinthecom-putationalapproachtovisionitself,and,Ithink,withincomputationalpsycho-logymoregenerally.Withthatinmind,IshallturnnowtoEgan’sclaimthatthetheoryisnotintentionalatall,aminorityviewofMarr’stheorythathasnot,Ibelieve,receiveditsdue(cf.critiquesofEganbyButler1996and1998andShapiro1997;seealsoChomsky1995:55,fn.25).AttheheartofEgan’sviewofMarrisaparticularviewofthenatureofMarr’scomputationallevelofdescription.CommentatorsonMarrhavealmostunivers-allytakenthistocorrespondtowhatothershavecalledthe“knowledgelevel”(Newell1980)orthe“semanticlevel”(Pylyshyn1984)ofdescription,i.e.,asofferinganintentionalcharacterizationofthecomputationalmechanismsgovern-ingvisionandothercognitiveprocesses.RatherthanignoringMarr’scomputa-tionallevel,assome(e.g.,Shapiro1997)haveclaimedshedoes(supposedlyin271\nRobertA.WilsonordertofocusexclusivelyonMarr’salgorithmiclevelofdescription),Eganrejectsthisdominantunderstandingofthecomputationallevel,arguinginsteadthatwhatmakesitacomputationallevelisthatitspecifiesthefunctiontobecom-putedbyagivenalgorithminprecise,mathematicalterms.Thatis,whilethislevelofdescriptionisfunctional,whatmakesitthefirststageinconstructingacomputationaltheoryisthatitoffersafunction-theoreticcharacterizationofthecomputation,andthusabstractsawayfromallotherfunctionalcharacterizations.Thus,whilevisionmighthaveallsortsoffunctionsthatcanbespecifiedinlan-guagerelativelyclosetothatofcommonsense(e.g.,it’sforextractinginforma-tionfromtheworld,forperceivinganobjectiveworld,forguidingbehavior),noneofthese,inEgan’sview,formsapartofMarr’scomputationallevelofdescription.Giventhisview,thecaseforMarr’stheorybeingindividualisticbecausecomputationalfollowsreadily:Acomputationaltheoryprescindsfromtheactualenvironmentbecauseitaimstoprovideanabstract,andhencecompletelygeneral,descriptionofamechanismthataffordsabasisforpredictingandexplainingitsbehaviorinanyenvironment,eveninenvironmentswherewhatthedeviceisdoingcannotcomfortablybedescribedascognition.Whenthecomputationalcharacterizationisaccompaniedbyanappropri-ateintentionalinterpretation,wecanseehowamechanismthatcomputesaparticu-larmathematicalfunctioncan,inaparticularcontext,subserveacognitivefunctionsuchasvision.(1995:191).AccordingtoEgan,whileanintentionalinterpretationlinksthecomputationaltheorytoourcommon-sense-basedunderstandingofcognitivefunctions,itformsnopartofthecomputationaltheoryitself.Egan’sviewnaturallyraisesquestionsnotonlyaboutwhatMarrmeantbythecomputationallevelofdescriptionbut,moregenerally,aboutthenatureofcomputationalapproachestocognition.TherearecertainlyplacesinwhichMarrdoestalkofthecomputationallevelassimplybeingahigh-levelfunctionalcharacterizationofwhatvisionisfor,andthusprimarilyasorientingtheresearchertoposecertaingeneralquestions.Forexample,oneofhistablesoffersthefollowingsummaryquestionsthatthetheoryanswersatthecomputationallevel:“Whatisthegoalofthecomputation,whyisitappropriate,andwhatisthelogicofthestrategybywhichitcanbecarriedout?”(1982:25,fig.1–4).ThosedefendingtheclaimthatMarr’stheoryisexternalisthavetypicallyrestedwiththisbroadandsomewhatlooseunderstand-ingofthecomputationallevelofthetheory(see,e.g.,Burge1986a:28;Shapiro1993:499–500;1997:134).Theproblemwiththisbroadunderstandingofthecomputationallevel,andthusofcomputationalapproachestocognition,isthatwhileitbuildsabridgebetweencomputationalpsychologyandmorefolksywaysofthinkingaboutcog-nition,itcreatesagapwithinthecomputationalapproachbetweenthecomputa-tionalandalgorithmiclevels.Forexample,ifwesupposethatthecomputationallevelspecifiessimplythatsomevisualstateshavethefunctionofrepresenting272\nIndividualismedges,othersthefunctionofrepresentingshapes,etc.,thereisnothingaboutsuchdescriptionsthatguidesusinconstructingalgorithmsthatgeneratethestate-to-statetransitionsattheheartofcomputationalapproachestovision.Moreinformalelaborationsofwhatvisionisfor,orofwhatitevolvedtodo,dolittlebythemselvestobridgethisgap.Thepointhereisthatcomputationalspecificationsthemselvesareaveryspecialkindoffunctionalcharacterization,atleastwhentheyaretobecompletedorimplementedinautomatic,algorithmicprocesses.Minimally,proponentsofthebroadinterpretationofcomputationalapproachestocognitionneedeithertocon-struethecomputationallevelasencompassingbutgoingbeyondthefunction-theoreticcharacterizationsofcognitivecapacitiesthatEganidentifies,ortheymustallocatethosecharacterizationstothealgorithmiclevel.Thelatteroptionsimplyexacerbatesthe“gap”problemidentifiedabove.Buttheformeroptionseemstometolumptogetheravarietyofquitedifferentthingsundertheheadingof“thecomputationallevel,”andsubsequentlyfailstorecognizetheconstraintsthatcomputationalassumptionsbringintheirwake.Thetemporalandmodular-itydimensionstoMarr’stheoryexacerbatetheproblemhere.Thereisalargeissuelurkinghereconcerninghowfunctionalismshouldbeunderstoodwithincomputationalapproachestocognition,andcorrespondinglyhowencompassingsuchapproachesreallyare.Functionalismhasusuallybeenunderstoodasofferingawaytoreconcileourfolkpsychology,ourmanifestimage(Sellars1962)ofthemind,withthedevelopingsciencesofthemind,evenifthatreconciliationinvolvesrevisingfolkpsychologyalongindividualisticlines(e.g.,factoringitintoanarrowfolkpsychologyviathenotionofnarrowcon-tent).Andcomputationalismhasbeentakentobeonewayofspecifyingwhattherelevantfunctionalrolesare:theyare“computationalroles.”ButifEganisrightaboutMarr’sunderstandingofthenotionofcomputationasafunction-theoreticnotion,andweaccepttheviewthatthisunderstandingissharedincomputationalapproachestocognitionmoregenerally,thenthecorrespondingversionoffunc-tionalismaboutthemindmustbecorrespondinglyfunction-theoretic:itmustnotonly“prescindfromtheactualenvironment,”assheclaimsthecomputa-tionallevelmustdo,butalsofromthesortofinternalcausalrolethatfunction-alistshaveoftenappealedto.Cognitivemechanisms,onthisview,takemathematicallycharacterizableinputstodelivermathematicallycharacterizableoutputs,andquacomputationaldevices,thatisall.Anyprospectsfortheconsilienceofour“twoimages”mustlieelsewhere.Inarguingforthenon-intentionalcharacterofMarr’stheoryofvision,Eganpresentsanausterepictureoftheheartofcomputationalpsychology,onethataccordswiththeindividualisticorientationofcomputationalcognitivescienceasithastraditionallybeendeveloped(cf.Chomsky1995),evenifcomputationalpsychologistshavesometimes(e.g.,Pylyshyn1984)attemptedtoplacetheirtheorieswithinmoreencompassingcontexts.Oneproblemwithsuchaviewofcomputation,asShapiro(1997:149)pointsout,isthatacomputationaltheoryofXtellsusverylittleaboutthenatureofX,includinginformationsufficientto273\nRobertA.WilsonindividuateXas(say)avisualprocessatall.WhileEgan(1999)seemswillingtoacceptthisconclusion,placingthissortofconcernoutsideofcomputationaltheoryproper,thisresponsehighlightsagapbetweencomputationaltheory,austerelyconstrued,andthemyriadoftheories–representational,functional,orecologicalinnature–withwhichsuchatheorymustbeintegratedforittoconstituteacomplete,mechanisticaccountofanygivencognitiveprocess.Themoreausteretheaccountofcomputation,thelargerthisgapbecomes,andthelessacomputationaltheorycontributestoourunderstandingofcognition.OnemightwellthinkthatEgan’sviewofcomputationaltheoryinpsychologyerrsonthesideofbeingtooaustereinthisrespect.11.8ExploitativeRepresentationandWideComputationalismAsabeginningonanalternativewayofthinkingaboutcomputationandrepres-entation,consideraninterestingdifferencebetweenindividualisticandexternalistinterpretationsofMarr’stheorythatconcernswhatitisthatMarriancomputa-tionalsystemshavebuiltintothem.Individualistsaboutcomputation,suchasEganandSegal,holdthattheyincorporatevariousinnateassumptionsaboutwhattheworldislike.Thisisbecausetheprocessofvisioninvolvesrecovering3-Dinformationfroma2-Dretinalimage,aprocessthatwithoutfurtherinputwouldbeunderdetermined.Theonlywaytosolvethisunderdeterminationprob-lemistomakeinnateassumptionsabouttheworld.ThebestknownoftheseisUllman’srigidityassumption,whichsaysthat“anysetofelementsundergoingatwo-dimensionaltransformationhasauniqueinterpretationasarigidbodymov-inginspaceandhenceshouldbeinterpretedassuchabodyinmotion”(1979:146).Theclaimthatindividualistsmakeisthatassumptionslikethisarepartofthecomputationalsystemsthatdrivecognitiveprocessing.ThisisthestandardwaytounderstandMarr’sapproachtovision.ExternalistslikeShapirohaveconstruedthismatterdifferently.Althoughcertainassumptionsmustbetrueoftheworldinorderforourcomputationalmechanismstosolvetheunderdeterminationproblem,thesearesimplyassumptionsthatareexploited(Shapiro1997:135,143;cf.Rowlands1999)byourcomputationalmechanisms,ratherthaninnateinourcognitivearchitecture.Thatis,theassump-tionsconcerntherelationshipsbetweenfeaturesoftheexternalworld,orbetweenpropertiesoftheinternal,visualarrayandpropertiesoftheexternalworld,butthoseassumptionsarenotthemselvesencodedintheorganism.Tobringoutthecontrastbetweenthesetwoviews,considerafewsimpleexamples.Anodometerkeepstrackofhowmanymilesacarhastraveled,anditdoessobycountingthenumberofwheelrotationsandbeingbuiltsoastodisplayanumberproportionaltothisnumber.Onewayinwhichitcoulddothiswouldbefortheassumptionthat1rotation=xmeterstobepartofitscalculationalmachinery;anotherwayofachievingtheendwouldbeforittobebuiltsoas274\nIndividualismsimplytorecordxmetersforeveryrotation,thusexploitingthefactthat1rotation=xmeters.Inthefirstcaseitencodesarepresentationalassumption,andusesthistocomputeitsoutput;inthesecond,itcontainsnosuchencodingbutinsteadusesanexistingrelationshipbetweenitsstructureandthestructureoftheworld.Ineithercase,ifitfindsitselfinanenvironmentinwhichtherelationshipbetweenrotationstodistancetraveledisadjusted(e.g.,largerwheels,orbeingdrivenonatreadmill),itwillnotfunctionasitissupposedto,andwillmisrepre-sentthedistancetraveled.Considertwodifferentstrategiesforlearninghowtohitabaseballthatis−2fallingverticallytotheground.Sincetheballacceleratesat9.8ms,thereisatimelagbetweenswingingandhitting.Onecouldeitherassumethattheballisfalling(say,ataspecificrateofacceleration),andthenusethisassumptiontocalculatewhenoneshouldswing;alternatively,onecouldsimplyaimacertaindistancebelowwhereoneperceivestheballatthetimeofswinging(say,twofeet).Inthislattercaseonewouldbeexploitingtherelationshipbetweenaccel-eration,time,anddistancewithouthavingtoencodethatrelationshipintheassumptionsonebringstobearonthetask.Thefactthattherearethesetwodifferentstrategiesforaccomplishingthesameendshould,minimally,makeuswaryofacceptingtheclaimthatinnateassumptionsaretheonlywaythatacomputationalsystemcouldsolvetheunderdeterminationproblem.ButIalsowanttodeveloptheideathatourperceptualsysteminparticularandourcognitivesystemsmoregenerallytypicallyexploitratherthanencodeinformationabouttheworldandourrelationshiptoit,aswellassaysomethingaboutwhereMarrhimselfseemstostandonthisissue(seealsoWilson,forthcoming).AnassumptionthatEganmakesandthatiswidelysharedinthephilosophicalliteraturesbothonindividualismandcomputationisthatatleastthealgorithmiclevelofdescriptionwithincomputationalpsychologyisindividualistic.Theideaherehas,Ithink,seemedsoobviousthatithasseldombeenspelledout:algorithmsoperateonthesyntacticorformalpropertiesofsymbols,andtheseareintrinsictotheorganismsinstantiatingthesymbols.WemightchallengethisneitherbydisputinghowmuchisbuiltintoMarr’scomputationallevel,norbysquabblingoverthelinebetweenMarr’scomputationalandalgorithmiclevels,but,rather,byarguingthatcomputationsthemselvescanextendbeyondtheheadoftheorgan-ismandinvolvetherelationsbetweenindividualsandtheirenvironments.Thisposition,whichholdsthatatleastsomeofthecomputationalsystemsthatdrivecognition,especiallyhumancognition,reachbeyondthelimitsoftheorganismicboundary,Ihaveelsewhere(1994;1995:ch.3)calledwidecomputationalism,anditsapplicationtoMarr’stheoryofvisionmarksadeparturefromtheparametersgoverningthestandardindividualist-externalistdebateoverthattheory.Widecomputationalismconstitutesonewayofthinkingaboutthewayinwhichcogni-tion,evenconsideredcomputationally,is“embedded”or“situated”initsnature(cf.alsoHutchins1995;McClamrock1995),anditprovidesaframeworkwithinwhichanexploitativeconceptionofrepresentationcanbepursued.275\nRobertA.WilsonThebasicideaofwidecomputationalismissimple.Traditionally,thesortsofcomputationthatgoverncognitionhavebeenthoughttobeginandendattheskull.Butwhythinkthattheskullconstitutesamagicboundarybeyondwhichtruecomputationendsandmerecausationbegins?Giventhatwearecreaturesembeddedininformationallyrichandcomplexenvironments,thecomputationsthatoccurinsidetheheadareanimportantpartbutarenotexhaustiveofthecorrespondingcomputationalsystems.Thisperspectiveopensupthepossibilityofexploringcomputationalunitsthatincludethebrainaswellasaspectsofthebrain’sbeyond-the-headenvironment.Widecomputationalsystemsthusinvolvemindsthatliterallyextendbeyondtheconfinesoftheskullintotheworld.Onewaytobringoutthenatureofthedeparturemadebywidecom-putationalismwithintheindividualismdebatedrawsonadistinctionbetweenalocationalandataxonomicconceptionofpsychologicalstates(seealsoWilson2000a;cf.Rowlands1999:chs.2–3).Individualistsandexternalistsareusuallypresentedasdisagreeingoverhowtotaxonomizeorindividuatepsychologicalstates,butbothtypically(thoughnotalways)presumethattherelevantstatesarewhatwemightcalllocationallyindividualistic:theyarelocatedwithintheorganismicenvelope.Whatindividualistsandexternaliststypicallydisagreeaboutiswhetherinadditiontobeinglocationallyindividualistic,psychologicalstatesmustalsobetaxonomicallyindividualistic.Widecomputationalism,however,rejectsthisassumptionoflocationalindividualismbyclaimingthatsomeofthe“relevantstates”–someofthosethatconstitutetherelevantcomputationalsystem–arelocatednotintheindividual’sheadbutinherenvironment.Theintuitiveideabehindwidecomputationalismiseasyenoughtograsp,buttherearetwocontroversialclaimscentraltodefendingwidecomputationalismasaviablemodelforthinkingaboutandstudyingcognitiveprocessing.Thefirstisthatitissometimesappropriatetoofferaformalorcomputationalcharacteriza-tionofanorganism’senvironment,andtoviewpartsofthebrainoftheorganism,computationallycharacterized,togetherwiththisenvironmentsocharacterized,asconstitutingaunifiedcomputationalsystem.Withoutthisbeingtrue,itisdifficulttoseewidecomputationalismasacoherentview.Thesecondisthatthisresultingmind–worldcomputationalsystemitself,andnotjustthepartofitinsidethehead,isgenuinelycognitive.Withoutthissecondclaim,widecom-putationalismwouldatbestpresentazanywayofcarvingupthecomputationalworld,onewithoutobviousimplicationsforhowweshouldthinkaboutrealcognitioninrealheads.Ratherthanattemptingtorespondtoeachoftheseproblemsinthespaceavailable,IshallturntotheissueofhowthisgeneralperspectiveonrepresentationandcomputationsitswithMarr’stheoryofvision.Aswehaveseen,Marrhimselfconstruesthetaskofatheoryofvisiontobetoshowhowweextractvisualinformationfrom“arraysofimageintensityvaluesasdetectedbythephotoreceptorsintheretina”(1982:31).Thus,aswehavealreadynoted,theproblemofvisionbeginswithretinalimages,notwithpropertiesoftheworldbeyondthoseimages,and“thetrueheartofvisualperceptionistheinferencefromthestructureofanimageaboutthestructureoftherealworld276\nIndividualismoutside”(ibid.:68;myemphasis).Marrgoesontocharacterizearangeofphysicalconstraintsthatholdtrueoftheworldthat“makethisinferencepossi-ble”(ibid.),buthemakesitclearthat“theconstraintsareusedbyturningthemintoanassumptionthatmayormaynotbeinternallyverifiable”(ibid.:104).ForallofMarr’stalkoftheimportanceoffactsaboutthebeyond-the-headworldforconstructingthecomputationallevelinatheoryofvision,thisisrepresentativeofhowheconceivesofthatrelevance(e.g.,ibid.:43,68,99,103–5,265–6).Itseemstomeclearthat,intermsthatIintroducedearlierinthissection,Marrhimselfadoptsanencodingviewofcomputationandrepresentation,ratherthananexploitativeviewofthetwo.Thevisualsystemis,accordingtoMarr,alocationallyindividualisticsystem.WhateverMarr’sownviewshere,theobviouswaytodefendawidecomputa-tionalinterpretationofhistheoryistoresisthisinferencefrom“xisaphysicalconstraintholdingintheworld”to“xisanassumptionthatisencodedinthebrain.”Thisis,inessence,whatIhavepreviouslyproposedoneshoulddointhecaseofthemultiplespatialchannelstheoryofformperceptionpioneeredbyCampbellandRobson(1968).LikeMarr’stheoryofvision,whichinpartbuildsonthiswork(seeesp.Marr1982:61–4),thistheoryhasusuallybeenunderstoodaspostulatingalocationallyindividualisticcomputationalsystem,onethatbeginswithchannelsearlyinthevisualpathwaythataredifferentiallysensitivetofourparameters:orientation,spatialfrequency,contrast,andspatialphase.Mysugges-tion(Wilson1994;1995:ch.3)wastotakeseriouslytheclaimthatanyvisualscene(intheworld)canbedecomposedintothesefourproperties,andsoseethecomputationalsystemitselfasextendingintotheworld,withthecausalrelation-shipbetweenstimulusandvisualchannelsitselfmodeledbytransitionrules.Ratherthansimplyhavingthesepropertiesencodedindistinctvisualchannelsinthenervoussystem,viewthein-the-headpartoftheformperceptionsystemasexploitingformalpropertiesintheworldbeyondthehead.WithrespecttoMarr’stheory,thereisarespectinwhichthiswidecomputationalinterpretationiseasytodefend,andanotherinwhichitisdifficulttodefend.ThefirstoftheseisthatMarr’s“assumptions,”suchasthespatialcoincidenceassumption(1982:70)andthe“fundamentalassumptionofstereopsis”(ibid.:114),typicallybeginasphysicalconstraintsthatreflectthestructureoftheworld;intheaboveexamples,theybeginastheconstraintofspatiallocalization(ibid.:68–9)andthreematchingconstraints(ibid.:112–14).Thus,thestrategyistoarguethattheconstraintsthemselves,ratherthantheirderivativeencoding,playaroleindefiningthecomputationalsystem,ratherthansimplyfillingaheuristicroleinallowingustoofferacomputationalcharacterizationofalocationallyindividualisticcognitivesystem.ThecorrespondingrespectinwhichawidecomputationalinterpretationofMarr’stheoryisdifficulttodefendisthattheseconstraintsthemselvesdonotspecifywhatthecomputationalprimitivesare.OnepossibilitywouldsimplybetoattributetheprimitivesthatMarrascribestotheimagetofeaturesofthescenesperceivedthemselves,butthiswouldbetooquick.Forexample,Marrconsiders277\nRobertA.Wilsonzero-crossingstobestepsinacomputationthatrepresentsharpchangesinintensityintheimage,andwhilewecouldtakethemtorepresentintensitychangesinthestimuliintheworld,zero-crossingsthemselvesarelocatedsome-whereearlyinthein-the-headpartofthevisualsystem,probablyclosetotheretina.Abetterstrategy,Ithink,wouldbetodeflatetheinterpretationoftheretinalimageandlook“upstream”fromittoidentifyricherexternalstructuresintheworld,structureswhichsatisfythephysicalconstraintsthatMarrpostulates.Thatis,oneshouldextendthetemporaldimensiontoMarr’stheorysothattheearlieststagesinbasicvisualprocessesbeginintheworld,notinthehead.Sincethestudyofvisionhasbeenlargelyconductedwithinanoverarchingindividual-isticframework,thisstrategywouldrequirerecastingthetheoryofvisionitselfsothatitrangesoveraprocessthatcausallyextendsbeyondtheretinalimage(seealsoRowlands1999:ch.5).11.9NarrowContentandMarr’sTheoryConsidertheveryfirstmoveinSegal’sargumentfortheconclusionthatMarr’stheoryofvisionisindividualistic,theclaimthattherearetwogeneralinterpreta-tionsavailablewhenoneseekstoascribeintentionalcontentstothevisualstatesoftwoindividuals:one“restrictive”(Burge’s)andone“liberal”(Segal’s).Some-thinglikethesetwogeneralalternativeswasimplicitinthebasicTwinEarthcaseswithwhichwe–andthedebateoverindividualism–began;theideathattwinsmustsharesomeintentionalstateaboutwaterysubstances(oraboutarthritis-likediseases,inBurge’sstandardcase)isthebasisforattemptstoarticulateanotionofnarrowcontent,i.e.,intentionalcontentthatdoessuperveneontheintrinsic,physicalpropertiesoftheindividual.Ihaveelsewhere(Wilson1995:ch.9)expressedmyskepticismaboutsuchattempts,andhereIwanttotiethisskepticismtotheinnocuous-lookingfirststepinSegal’sinterpretation.ThisfirststepinSegal’sinterpretation,thepresuppositionofaliberalinterpre-tationforMarr’stheory,andacorrespondingviewoftheoriginalTwinEarthcasesingeneral,arethemselvesquestionable.Notefirstthattherepresentationsthatwemight,inordertomaketheirdisjunctivecontentperspicuous,label“crackdow”or“waterortwater,”dorepresenttheirreliable,environmentalcauses:“crackdow”isreliablycausedbycracksorshadows,andhasthecontentcrackorshadow;similarlyfor“waterortwater.”Butthenthisdisjunctivecontentisaspeciesofwide,notnarrowcontent,asEgan(1995:195)haspointedout.Inshort,althoughbeingsharedbytwinsisnecessary,itisnotsufficientformentalcontenttobenarrow.Topressfurther,ifthecontentofone’svisualstateistobeindividualistic,itmustbesharedbydoppelgangersnomatterhowdifferenttheirenvironments.Thus,thecaseof“twins”ismerelyaheuristicforthinkingaboutapotentiallyinfinitenumberofindividuals.Butthenthefocusonacontentsharedbytwo278\nIndividualismindividuals,andthusonacontentthatisneutralbetweentwoenvironmentalcauses,representsamisleadingsimplificationinsofarasthecontentneededwon’tsimplybe“crackdow,”butsomethingmorewildlydisjunctive,sincethereisapotentiallyinfinitenumberofenvironmentsthatmightproducethesameintrinsic,physicalstateoftheindividual’svisualsystemas(say)cracksdointheactualworld(seealsoEgan1991:200,fn.35).Itisnotthatwecan’tsimplymakeupanameforthecontentofsuchastate(wecan:callit“X”),butthatitisdifficulttoviewastatesoindividuatedasbeingaboutanything.Andifbeingaboutsomethingisattheheartofbeingintentional,thenthiscallsintoquestionthestatusofsuchnarrowlyindividuatedstatesasintentionalstates.Segal(1991:490)hasclaimedthatthenarrowcontentof“crackdow,”orbyimplication“waterortwater,”neednotbedisjunctive,justsimplymoreencom-passingthan,respectively,crackorwater(seealsoSegal2000).Butcastingtheabovepointsintermsofdisjunctivecontentsimplymakesvividthegeneralproblemsthat(1)theindividuationofstatesintermsoftheircontentstillproceedsviareferencetowhatdoesorwouldcausethemtobetokened;and(2)onceoneprescindsfromaconceptionofthecognitivesystemasembeddedinandinteractingwiththeactualworldinthinkingabouthowtotaxonomizeitsstates,itbecomesdifficulttodelineateclearlythosestatesasintentionalstateswithsomedefinitecontent.Asitissometimesput,narrowcontentbecomesinexpressible.Tworesponsesmightbemadetothissecondobjection.First,onemightconcedethat,strictlyspeaking,narrowcontentisinexpress-ible,butthenpointoutwaysof“sneakinguponit”(Fodor1987:52).Onemightdosobytalkingofhowonecan“anchor”narrowcontenttowidecontent(ibid.:50–3);orofhowtospecifytherealizationconditionsforaproposition(Loar1988).Butthesesuggestions,despitetheircurrency,seemtomelittlemorethanwhistlinginthedark,andtheconcessiononwhichtheyrest,fatal.Allofthewaysof“sneakingupon”narrowcontentinvolveusingwidecontentsinsomeway.Yetifwidecontentissuchaproblematicnotion(becauseitisnotindividualistic),thensurelytheproblemspreadstoanynotion,suchassnuck-up-onnarrowcontent,forwhoseintelligibilitythenotionofwidecontentiscrucial.Moreover,ifnarrowcontentreallyisinexpressible,thentheideathatitisthisnotionthatiscentraltopsychologicalexplanationasitisactuallypracticed,andthisnotionthatdoesorwillfeatureinthenaturalkindsandlawsofthecognitivesciences,cannotreasonablybesustained.ExceptinDouglasAdamsesquespoofsofscience,therearenoscienceswhosecentralexplanatoryconstructsareinexpressible.Moreover,thisviewwouldmaketheclaimthatonearrivesatthenotionofnarrowcontentviaanexaminationofactualexplanatorypracticeinthecognitivesciencesextremelyimplausible,sinceifnarrowcontentisinexpressible,thenonewon’tbeabletofinditexpressedinanyexistingpsychologicaltheory.Inshort,andintermsthatIintroducedearlier,theideathatsnuck-up-onnarrowcontentiswhatcognitivescienceneedsorusesrepresentsareversiontothecognitivesciencegesture.Secondly,itmightbeclaimedthatalthoughitistruethatitisdifficultforcommon-sensefolktocomeupwithlabelsforintentionalcontents,thoseinthe279\nRobertA.Wilsonrelevantcognitivesciencescananddoallthetime,andweshoulddefertothem.Forexample,onemightclaimthatmanyifnotalloftherepresentationalprim-itivesinMarr’stheory,suchasblob,edge,andline,havenarrowcontents.Theseconcepts,likemanyscientificterms,aretechnicaland,assuch,maybearnoobviousrelationshiptotheconceptsandtermsofcommonsense,buttheystillallowustoseehownarrowcontentcanbeexpressed.Onemightthinkthatthisresponsehasthesamequestion-beggingfeeltoitasdoestheclaimthatourfolkpsychologicalstatesarethemselvesnarrow.However,theunderdeterminationofphilosophicalviewsbythedataofthescientifictheories,suchasMarr’s,thattheyinterpretremainsaproblemforbothindividualistsandexternalistsalikehere.Asmydiscussionofexploitativerepresentationandwidecomputationperhapssug-gests,myownviewisthatweneedtoreinvigoratethewaysinwhichthecompu-tationalandrepresentationaltheoriesofmindhaveusuallybeenconstruedwithincognitivescience.Ifthiscanbedoneinmorethanagesturalmanner,thentheissueofthe(in)expressibilityofnarrowcontentwillbelargelymoot.11.10IndividualismandtheProblemofSelf-knowledgeThusfar,Ihaveconcentratedondiscussionsofindividualismandexternalismincontemporaryphilosophyofmindwithaprimaryaffinitytocognitivescience.Itistestimonytothecentralityofindividualismandexternalismforphilosophymoregenerally–quiteapartfromtheirrelevancetoempiricalcognitivescience–thatthereisavarietyofdiscussionsthatexploretherelationshipbetweenthesepositionsandtraditionalissuesinthephilosophyofmindandphilosophymoregenerally.Themostinterestingoftheseseemtometoclusteraroundthreerelatedepistemologicalissues:self-knowledge,aprioriknowledge,andskepticism.Basictoself-knowledgeisknowledgeofone’sownmind,andtraditionallythisknowledgehasbeenthoughttoinvolvesomeformofprivilegedaccesstoone’sownmentalstates.Thisfirst-personprivilegedaccesshasoftenbeenunderstoodintermsofoneormoredistinctivepropertiesthattheresultingsecond-ordermentalstateshave.Thesestates,suchasmybeliefthatIbelievethattheEarthgoesaroundthesun,havebeenclaimedtobeinfallible(i.e.,incapableofbeingfalseormistaken),whichwouldimplythatsimplyhavingthesecond-orderbeliefguaranteesthatonehasthefirst-orderbeliefthatisitsobject;orincorrigible(i.e.,evenifmistaken,incapableofbeingcorrectedbyanyoneotherthanthepersonwhohasthem),whichwouldatleastimplythattheyhaveaformofepistemicsecuritythatothertypesofmentalstatelack.Ineithercase,thereisanasymmetrybetweenknowledgeofone’sownmindandknowledgeofthemindsofothers,aswellasknowledgeofotherthingsintheworld.Indicativeofthedepthoftheseasymmetriesinmodernphilosophyisthefactthatanintroductiontoepistemo-logy,particularlyonewithahistoricalslant,thatreflectsonskepticism,willlikelyintroducetheproblemofothermindsandtheproblemofourknowledgeoftheexternal280\nIndividualismworld,butnotthecorrespondingproblemofself-knowledge.Skepticismaboutone’sownmindhasseemedtobeprecludedbytheverynatureofself-knowledge.Althoughthecontrastbetweenfirst-andthird-personknowledgeofmentalstateshassoftenedinrecentphilosophyofmind,itremainspartofourcommon-senseconceptionofthemindthatthewaysinwhichIknowaboutmyownmentallifearedistinctivefromthewaysinwhichIknowaboutthatofothers(cf.Siewert1998).Thus,notunreasonably,theideaoffirst-personepistemicprivilegesurvives.Knowledgeaboutone’sself,abouttheconditionorstateofone’smindorbody,oftenenoughseemstobesimplyamatterofintrospection,ofinward-directedreflectionorattention,ratherthanrequiringthecollectionofevidencethroughobservationorexperiment.Isimplyfeelmyskinitching,oruponattend-ingnoticethatmytoesaresquashedupinmyshoes;tofindoutwhetheryourskinisitchingorwhetheryourtoesaresquashedupinyourshoes,Iobserveyourbodyanditsbehavior(includingwhatyousay),andthendrawaninferencefromthatobservationtoaconclusionaboutyourbodilystate.Self-knowledgeisdirect,whileknowledgeofothersisinferentialormediatedinsomeway,basedonobservationandotherformsofevidence.Sinceone’sownmentalstatesaretypicallytheobjectoffirst-personthoughts,weareacquaintedwithourownmindsinawaythatwearenotacquaintedwiththemindsofothers.Individualisticconceptionsofthemindhaveseemedwell-suitedtomakingsenseoffirst-personprivilegedaccessandthesubsequentasymmetrybetweenself-knowledgeandknowledgeofthementalstatesofothers.Ifmentalstatesareindividuatedinabstractionfromthebeyond-the-individualenvironment,thenthereseemstobenoprobleminunderstandinghowtheprocessofintrospection,turningourmind’seyeinwards(touseacommonmetaphor),revealsthecontentofthosestates.ToinvoketheCartesianfantasyinawaythatbringsouttheasymmetrybetweenself-knowledgeandotherformsofknowledge,eveniftherewereanevildemonwhodeceivedmeabouttheexistenceofanexternalworld–includingtheexistenceofotherpeoplewithmentalstateslikemine–theonethingthatIcouldbesureaboutwouldbethatIamhavingexperienceswithacertaincontent.Asitissometimesput,evenifIcouldbedeceivedaboutwhetherthereisreallyatreeinfrontofmeandthusaboutwhetherIamactuallyseeingatree,IcannotbedeceivedaboutwhetheritseemstomethatIamseeingatree.Thusindividu-alismseemstofacilitateasortofepistemicsecurityforfirst-personknowledgeofone’sownmentalstatesthatthecorrespondingthird-personknowledgelacks.Externalism,bycontrast,posesaprimafacieproblemforeventhemoremod-estformsoffirst-personprivilegedaccess,andhasevenbeenthoughttocallintoquestionthepossibilityofanyformofself-knowledge.Forexternalismclaimsthatwhatmentalstatesareismetaphysicallydetermined,inpart,bythenatureoftheworldbeyondtheboundaryofthesubjectofthosestates.Thusitwouldseemthatinordertoknowwhatoneisthinking,i.e.,toknowthecontentofone’smentalstates,onewouldhavetoknowsomethingabouttheworldbeyondone’sself.Butthiswouldbetoassimilateourfirst-personknowledgeofourownmindstoourknowledgeofotherthings,andsodenyanyprivilegedaccessthat281\nRobertA.Wilsonself-knowledgemightbethoughttohave.ItimpliesthatinordertoknowmyownmindIneedtoknowaboutperhapsdifficult-to-discernfactsaboutthenatureofthephysicalorsocialworldinwhichIlive,andsoitalsosuggeststhatinarangeofordinarycaseswherewemightunreflectivelyattributeself-knowledge,Idon’tactuallyhaveself-knowledgeatall.Wecanexpresstheproblemhereinanotherwaythatabstractsfromthediffer-encesbetweenbothspecificaccountsofprivilegedaccessandspecificversionsofexternalism.Whetheritbeinfallible,incorrigible,self-intimating,introspective,orapriori,knowledgeofone’sownmentalstateshasaspecialcharacter.Knowingone’sownmentalstatesinvolves,interalia,knowingtheircontents.Now,accord-ingtoexternalism,thecontentsofasubject’smentalstatesaremetaphysicallydetermined,inpart,byfactsaboutherphysicalorsocialenvironment.Knowledgeofthesefacts,however,doesnothavethisspecialcharacter.Butthenhowisthespecialcharacterofself-knowledgecompatiblewiththenon-specialcharacterofworldlyknowledge,giventhedependenceoftheformeronthelatter(seealsoLudlowandMartin1998:1)?Othershavestatedtheproblemmoredramatically.Forexample,Davidsonpresentsitas“atransposedimageofCartesianskepticism”(1987:94),accordingtowhich“[o]urbeliefsabouttheexternalvieware...directedontotheworld,butwedon’tknowwhatwebelieve”(ibid.),claimingthusthatexternalismseemstoimplythatwedon’thaveself-knowledgeatall;Heilpointsoutthat“ifexternalismweretrue,onecouldnotdiscoverastate’sinten-tionalpropertiesmerelybyinspectingthatstate”(1988:137),goingontoconnectthisupwithDavidson’sfocusona“nastierskeptic,onewhoquestionsthepresumptionthatwethinkwhatwethinkwethink”(ibid.).Theproblemcanbeschematizedasasupposedlyinconsistenttriadofpropositions(cf.alsoMcKinsey1991,whosetriaddiffers;seebelow).LetP=thecontentsofourmentalstates,E=factsabouttheenvironment,andlet“byintrospection”standinforthedistinctivecharacterofself-knowledge:1WeknowPbyintrospection.(Self-knowledge)2ParemetaphysicallydeterminedinpartbyE.(Externalism)3Earenotknownbyintrospection.(CommonSense)Theclaimisthatoneofthesethreepropositionsmustbegivenup.IfwerejectSelf-knowledge,thenwegiveupontheideathatwehaveprivilegedaccesstoourownminds;ifwerejectExternalism,thenwegiveuponanindependentlyplausibleviewofthemind;andifwerejectCommonSense,thenwemakeastrangeandimplausibleclaimaboutourknowledgeofthephysicalorsocialworld.Whenitisstatedsostarkly,Ithinkthattherightresponsetothe“problemofself-knowledge”istoarguethatallthreepropositionsaretrue,andsoconsistent,andthusthatthereisnoproblemofself-knowledgeforanexternalisttosolve.Theirconsistencyturnsonthefactthat(1)and(3),whichmakeepistemologicalclaims,areconnectedonlyby(2),whichmakesametaphysicalclaim.Asacounterexampletothechargeofformalinconsistency,consideraninstanceofthe282\nIndividualismargumentwhereP=thestateofbeinginpain,andE=aparticular,complicatedstateofthecentralnervoussystem.Thereisnoinconsistencyintheseinstancesof(1)–(3):wedoknowthatweareinpainbyintrospection;thatstateismetaphysicallydeterminedbysomeparticularstateofourcentralnervoussystem;butwedon’tknowaboutthatstatebyintrospection.(Or,toputitmorecarefully:wedon’taboutitquastateofourcentralnervoussystembyintrospection.)Thesameistrueofouroriginaltriad,aswellasofvariationsonthosepropositionswhichsubstitutesomeotherdistinctivefeatureofself-knowledgefor“byintrospection.”Ifthisisthecorrectwaytorepresentthesupposedproblemforexternalists,andthebasisforanadequateresponsetothatproblem,thentwofeaturesoftheproblemareworthnoting.Thefirstisthatattheheartoftheproblemisnotanexternalistviewoftheminditselfbut,rather,anythesisofmetaphysicaldetermination,wherethedeterminingstateisnotsomethingthatisknowninthespecialwaythatmentalstatesareknown.Sincenotallofanorganism’sinternal,individualisticallyindividuatedstatesaresoknown,thereisavariationontheproblemofself-knowledgethatindividualistsmustface,ifitisarealproblem.Thus,evenifonerejectsthewayofdissolvingtheproblemposedabove,aversionoftheproblemofself-knowledgeremainsforbothexternalistsandindividualiststosolve.Thisimpliesthatexternalistsdonot,despiteinitialappearances,faceaspecialproblemconcerningself-knowledge.Thesecondisthattheproblemandresponsesocharacterizedhaveaffinitieswithafamilyofproblem–responsepairs,includingonthe“problem”sideMoore’sopenquestionargumentandtheparadoxofanalysis,andwhoseclosestrelativeperhapsisastandardobjectiontothemind–brainidentitytheory.Pain,itwasclaimed,couldn’tbeidenticaltoC-fiberfiring,sinceonecanknowthatoneisinpainbutnotknowthatone’sC-fiberswerefiring.Andthenow-standardresponseisthatsuchanobjection,inattemptingtoderiveanontologicalconclu-sionfromepistemologicalpremises,commitsafallacy.Now,asapurportedlyinconsistenttriad,ratherthananargumentthatdrawssuchaconclusion,theproblemofself-knowledgeitselfdoesnotsufferfromthisspecificproblem,althoughtherejectionofexternalismasaresponsetotheproblemwouldbesubjecttojustthisobjection.However,thebroadaffinityhereisworthkeepinginmind.Howadequateonefindstheproposeddissolutionoftheproblemofself-knowledgeislikelytocorrelatewithhowadequateonefindsthistypeofresponsetothistypeofobjectionmoregenerally.Proponentsoftheproblemofself-knowledgeshouldobjecttotheclaimthat(1)–(3)adequatelyexpressesthedilemma.Inparticular,theyshould(andinfactdo)reject(2)asamemberofthetriad.Rather,theproblemofself-knowledgeisconstitutedbythefollowingtriad(cf.McKinsey1991):AWehaveaprioriknowledgeofP.(Self-Knowledge*)BWehaveaprioriknowledgethatPentailsE.(KnowledgeofExternalism)CWecannotknowEapriori.(CommonSense*)283\nRobertA.Wilson(A)–(C)areinconsistent.Butincontrastto(1)–(3),thisconstrualoftheproblemofself-knowledgecanbechallengedateverypoint.First,isanexternalistcommittedto(B)?Foranaffirmativeanswer,twopriorquestionsneedtobeansweredaffirmatively:accordingtoexternalism,(i)doweknowthatPentailsE?and(ii)doesPentailE?Take(ii):doesexternalismclaimthat,forexample,havingamentalstatewiththecontent“arthritisoccursinthethigh”entailthatarthritisdoesactuallyoccurinthethigh?Onereasontothinknotisthat,aswehaveseen,externalismincorporatestheideathatthereisasocialdivisionoflaborinboththoughtandlanguage,whichallowsforintention-alityevenin“vacuouscases”:wecanthinkPnotbecauseP,butbecauseothersthinkP.Given,however,thatexternalismclaimsthatthereisadeep,individuativerelationbetweenthenatureofanindividual’smentalstatesandhowtheworldbeyondtheindividualis,somesuchentailmentbetweenPandEseemsplausible.ThissuggeststhatEneedstobeconstruedinamorenuancedway,encompassingperhapsvariousdisjunctswhichtogethermustbetrueiftheexternalist’sviewofthemindiscorrect.Forexample,itmightbeclaimedthathavingthethoughtthatarthritisoccursinthethighentailseitherthatarthritisdoesoccurinthethighorthatonelivesinalinguisticcommunityofacertaincharacter;perhapsmore(ormorecomplicated)disjunctsneedtobeaddedhere(cf.Brown1995).Butthenitseemslessplausiblethat“we,”i.e.,eachofusordinaryfolk,know(2)soconstrued,letaloneknowthisapriori.Afterall,fewofushavereflectedsystematicallyonwhatthecontentsofourthoughtsimplyabouttheworld;indeed,manyofthosewhohavethusreflected–individualists–haveconcludedthattheytellusnothingaboutthecharacteroftheworld.Thisinturninvitestheresponsethattoformaninconsistenttriadwith(A)and(C),(B)needonlyclaimthatwecanhavesuchknowledge,andifexternalismistrue,andatleastsomepeoplebelieveitandwhatitentails,thenthatissufficienttogeneratetheinconsistency.Thisseemstometobeastrangewaytodeveloptheproblemofself-knowledge,sinceitnowsoundslikeaproblemthatariseschieflyfortheself-knowledgeofthoseversedintheexternalismliterature,ratherthanself-knowledgeperse.Buttherealproblemhere,andthesecondproblemwiththisconstrualofthetriad,isthatthetriadnowincludesaquestionablereadingof(C).Fornow(C),evenifitisadictateofcommonsense(andmodalized,as(3)isnot,thisseemsdoubtful),seemsfalse,sincealthoughitisusualforordinaryfolktoknowaboutwhatismentionedinEthroughempiricalmeans,andsotheydon’tusuallyknowEapriori,inlightofthisreadingof(B),itseemsatleastpossiblethatsome-onecouldknowaboutEinthisfashion.Combinedwiththereminderthatthisisnotusuallyhowwecometoknowfactsabouttheempiricalworld,thisconces-sionseemsfairlyinnocuous,andpreservestheintuitionthatself-knowledgeisepistemicallyprivileged.Wecanseehowthisconstrualofthetriadunderminesitsstatusasaproblemforexternalismbyturningto(A):doweknowthecontentsofourthoughtsa284\nIndividualismpriori?McKinseyconceivesofaprioriknowledgeasknowledge“obtainedinde-pendentlyofempiricalinvestigation”(1991:175),andreliesonintrospectionandreasoningasparadigmprocessesthroughwhichwegainsuchknowledge.Externalistsshouldbewaryofthisclaimifitistakentoimplythatself-knowledgecanbegainedcompletelyindependentlyofempiricalinvestigationoftheworld;whattheycanallow,andperhapsallthatisneededfor(A),isthatweknowthecontentsofourmentalstatesonparticularoccasionswithoutempiricallyinvesti-gatingtheworldonthoseoccasions.Onthisreading,(A)ismadetruebytheexistenceofintrospection,while(B)’struthturnsonourabilitytofollowtheargumentsfortheexternalistnatureofcontentandsointentionalmentalstates.While(C)mayseemtrueifwethinkonlyofintrospectionorreasoningaloneasmeansofsecuringaprioriknowledge(inthesenseabove),itbecomesmoredubiousonceweconsiderintrospectionandreasoningtogether.Sinceitisunusualforustobothintrospectourownmentalstatesandengageinsophisticatedphilosophicalreasoningusingthere-sultsofsuchintrospectionaspremises,thecircumstancesunderwhich(C)willbefalsifiedarethemselvesunusual;but(C)nonethelessis,strictlyspeaking,false.NoteIshouldliketothankGabrielSegal,FrancesEgan,andLawrenceShapiroforreadinganearlierversionofthisreview.ReferencesBlock,N.(1986).“AdvertisementforaSemanticsforPsychology.”InP.French,T.UehlingJr.,andH.Wettstein(eds.),MidwestStudiesinPhilosophy,vol.10(PhilosophyofMind).Minneapolis:UniversityofMinnesotaPress.Brown,J.(1995).“TheIncompatibilityofAnti-IndividualismandPrivilegedAccess.”Analysis,55:149–56.ReprintedinLudlowandMartin(1998).Burge,T.(1979).“IndividualismandtheMental.”InP.French,T.UehlingJr.,andH.Wettstein(eds.),MidwestStudiesinPhilosophy,vol.4(Metaphysics).Minneapolis:UniversityofMinnesotaPress.——(1986a).“IndividualismandPsychology.”PhilosophicalReview,95:3–45.——(1986b).“CartesianErrorandtheObjectivityofPerception.”InP.PettitandJ.McDowell(eds.),Subject,Thought,andContext.Oxford:OxfordUniversityPress.AlsoinGrimmandMerrill(eds.),ContentsofThought.Tucson,AZ:UniversityofArizonaPress(1988).Butler,K.(1996).“IndividualismandMarr’sComputationalTheoryofVision.”MindandLanguage,11:313–37.——(1998).“Content,Computation,andIndividuation.”Synthese,114:277–92.Campbell,F.W.andRobson,J.G.(1968).“ApplicationofFourierAnalysistotheVisibilityofGratings.”JournalofPhysiology,197:151–66.285\nRobertA.WilsonCarnap,R.(1928).TheLogicalConstructionoftheWorld,trans.byR.George,1967.Berkeley,CA:UniversityofCaliforniaPress.Chomsky,N.(1986).KnowledgeofLanguage.NewYork:Praeger.——(1995).“LanguageandNature.”Mind,104:1–61.——(2000).NewHorizonsintheStudyofLanguageandMind.NewYork:CambridgeUniversityPress.Cosmides,L.andTooby,J.(1994).“Foreword”toS.Baron-Cohen,Mindblindness.Cam-bridge,MA:MITPress.Cummins,R.C.(1983).TheNatureofPsychologicalExplanation.Cambridge,MA:MITPress.Davidson,D.(1987).“KnowingOne’sOwnMind.”ProceedingsoftheAmericanPhilo-sophicalAssociation.ReprintedinLudlowandMartin(1998).Dennett,D.C.(1978).“ArtificialIntelligenceasPhilosophyandasPsychology.”InM.Ringle(ed.),PhilosophicalPerspectivesonArtificialIntelligence.NewYork:HumanitiesPressandHarvesterPress.ReprintedinD.C.Dennett,Brainstorms.Cambridge,MA:MITPress.Egan,F.(1991).“MustPsychologybeIndividualistic?”PhilosophicalReview,100:179–203.——(1992).“Individualism,Computation,andPerceptualContent.”Mind,101:443–59.——(1995).“ComputationandContent.”PhilosophicalReview,104:181–203.——(1999).“InDefenseofNarrowMindedness.”MindandLanguage,14:177–94.Fodor,J.A.(1980).“MethodologicalSolipsismConsideredasaResearchStrategyinCognitivePsychology.”BehavioralandBrainSciences,3:63–73.ReprintedinJ.A.Fodor,Representations.Sussex:HarvesterPress(1981).——(1987).Psychosemantics.Cambridge,MA:MITPress.Gillett,C.(2002).“TheDimensionsofRealization:ACritiqueoftheStandardView.”Analysis(October).Heil,J.(1988).“PrivilegedAccess.”Mind,97:238–51.ReprintedinLudlowandMartin(1998).Hildreth,E.andUllman,S.(1989).“TheComputationalStudyofVision.”InM.Posner(ed.),FoundationsofCognitiveScience.Cambridge,MA:MITPress.Hutchins,E.(1995).CognitionintheWild.Cambridge,MA:MITPress.Jackendoff,R.(1991).“TheProblemofReality.”ReprintedinhisLanguagesoftheMind.Cambridge,MA:MITPress(1992).Kripke,S.(1980).NamingandNecessity.Cambridge,MA:HarvardUniversityPress.Loar,B.(1988).“SocialContentandPsychologicalContent.”InR.GrimmandD.Merrill(eds.),ContentsofThought.Tucson,AZ:UniversityofArizonaPress.Ludlow,P.andMartin,N.(eds.)(1998).ExternalismandSelf-Knowledge.PaloAlto,CA:CSLIPublications.Marr,D.(1982).Vision:AComputationalInvestigationintotheHumanRepresentationandProcessingofVisualInformation.SanFrancisco,CA:W.H.Freeman.Matthews,R.(1988).“CommentsonBurge.”InR.GrimmandD.Merrill(eds.),Con-tentsofThought.Tuscon,AZ:UniversityofArizonaPress.McClamrock,R.(1995).ExistentialCognition:ComputationalMindsintheWorld.Chi-cago:UniversityofChicagoPress.McKinsey,M.(1991).“Anti-IndividualismandPrivilegedAccess.”Analysis,51:9–16.286\nIndividualismNewell,A.(1980).“PhysicalSymbolSystem.”CognitiveScience,4:135–83.Putnam,H.(1975).“TheMeaningof‘Meaning’.”InK.Gunderson(ed.),Language,MindandKnowledge.Minneapolis,MN:UniversityofMinnesotaPress.ReprintedinH.Putnam,Mind,Language,andReality.NewYork:CambridgeUniversityPress(1975).Pylyshyn,Z.(1984).ComputationandCognition.Cambridge,MA:MITPress.Rowlands,M.(1999).TheBodyinMind.NewYork:CambridgeUniversityPress.Segal,G.(1989).“SeeingWhatisNotThere.”PhilosophicalReview,98:189–214.——(1991).“DefenceofaReasonableIndividualism.”Mind,100:485–94.——(2000).ASlimBookAboutNarrowContent.Cambridge,MA:MITPress.Sellars,W.(1962).“PhilosophyandtheScientificImageofMan.”InR.Colodny(ed.),FrontiersofScienceandPhilosophy.Pittsburgh:UniversityofPittsburghPress.ReprintedinW.Sellars,Science,Perception,andReality.Atascadero,CA:RidgeviewPublishingCompany(1963).Shapiro,L.(1993).“Content,Kinds,andIndividualisminMarr’sTheoryofVision.”PhilosophicalReview,102:489–513.——(1997).“AClearerVision.”PhilosophyofScience,64:131–53.Shoemaker,S.(2000).“RealizationandMentalCausation.”ReprintedinC.GillettandB.Loewer(eds.),PhysicalismanditDiscontents.NewYork:CambridgeUniversityPress.Siewert,C.(1998).TheSignificanceofConsciousness.Princeton,NJ:PrincetonUniversityPress.Stich,S.(1978).“AutonomousPsychologyandtheBelief-DesireThesis.”Monist,61:573–91.——(1983).FromFolkPsychologytoCognitiveScience.Cambridge,MA:MITPress.Ullman,S.(1979).TheInterpretationofVisualMotion.Cambridge,MA:MITPress.VanGulick,R.(1989).“MetaphysicalArgumentsforInternalismandWhyTheyDon’tWork.”InS.Silvers(ed.),Rerepresentation.Dordrecht,TheNetherlands:Kluwer.Walsh,D.M.(1999).“AlternativeIndividualism.”PhilosophyofScience,66:628–48.Wilson,R.A.(1992).“Individualism,CausalPowers,andExplanation.”PhilosophicalStudies,68:103–39.——(1994).“WideComputationalism.”Mind,103:351–72.——(1995).CartesianPsychologyandPhysicalMinds:IndividualismandtheSciencesoftheMind.NewYork:CambridgeUniversityPress.——(2000a).“TheMindBeyondItself.”InD.Sperber(ed.),Metarepresentation.NewYork:OxfordUniversityPress.——(2000b).“SomeProblemsfor‘AlternativeIndividualism’.”PhilosophyofScience,67:671–9.——(2001).“TwoViewsofRealization.”PhilosophicalStudies,104:1–31.——(forthcoming).TheIndividualintheFragileSciencesI:Cognition.287\nTheBlackwellGuidetoPhilosophyofMindPaulE.GriffithsEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter12EmotionsPaulE.Griffiths12.1BruteFeelingsorRationalJudgments?12.1.1ThefeelingtheoryofemotionsUntiltheearlytwentiethcenturyitwastakenforgrantedthatemotionsarefeelings:subjectivestatesofexperience.Darwincarriedoutextensiveempiricalinvestigationsofthephysiologicalandbehavioralcomponentsofemotionbutneverregardedtheseasanythingotherthan“expressions”ofthefeelingsexperiencedbypeopleandanimals.FollowingHerbertSpencer,Darwindefinedemotionsassensationscausedbystatesofaffairsoutsidethebody,intendingbythistodifferentiatethemfromsensationssuchashungerandpain.FeelingisalsocentraltoWilliamJames’stheoryofemotion.AlthoughJamesmadetheradicalsuggestionthatemotionfeelingsarecausedbythephysiologicalchangesassociatedwithemotion,ratherthancausingthosephysiologicalchanges,hestillidentifiestheemotionwiththefeeling,notthephysiologicalchangesortheearlierneuralprocessesthatcausethem.Naturallyenough,thebehavioristswerethefirsttoquestionthefeelingtheory.JohnB.Watsonarguedthatadultemotionalbehaviorswereconditionedresponsesbasedonthreeunconditionedreactionsininfantsthathetermedfear,rage,andpleasure.Later,undertheinfluenceofbehaviorismandtheverificationtheoryofmeaning,philosophicalbehavioristssuchasGilbertRyleclaimedthatacorrectanalysisofthemeaningsofemotionwordsinvolvesnoreferencetosubjectivestatesofexperience.Inthe1960sphilosophersenthusiasticallyembracedthe“cognitiverevolution”inpsychology,linguistics,andthenewfieldofartificialintelligence.Therejectionofbehaviorismwasthusnotaccompaniedbyarevivalofthefeelingtheory.Instead,aconsensusemergedinthephilosophyofemotionintheearly1960sthatemotionsaredefinedbythecognitionstheyinvolve.Thisconsensushaspersistedtothepresentday.Somephilosophershaveallowedfeelingsaroleinemotion,288\nEmotionsbutneveronethatdeterminestheidentityoftheemotion.Emotionfeelingsmerelyaddthe“heat”to“hotcognition.”PatriciaGreenspan(1988),forexample,arguesthatemotionsarefeelingsofcomfortordiscomfortdirectedtowardanevaluativethoughtaboutanexternal(orimaginary)stimulus.Itistheevaluativethoughtthatdefinestheemotion.Differentnegativeemotions,suchasangerandfear,aredifferentiatedonlybythedifferentevaluativethoughtstheyinvolve.Philosophershavegenerallyheldittobeaconceptualtruththatemotionsderivetheiridentitiesfromthethoughtsassociatedwiththem,butpsychologicalre-searchonthe“cognitivelabeling”ofstatesofarousalhasbeencitedasevidencethatempiricalfindingsconvergeonthesameconclusionasconceptualanalysis.Themostfrequentlycitedstudyshowedthatsubjectscouldbeinducedtodescribethesensationsproducedbyadrenalineinjectionsaseithereuphoriaorangerundertheinfluenceofcontextualcuesprovidedbytheexperimenters(SchachterandSinger1962).12.1.2PropositionalattitudetheoriesSincetheearly1960sthecognitivistorpropositionalattitudeschoolhasdom-inatedthephilosophyofemotion(Griffiths1989;Deigh1994).Thebasiccom-mitmentsofthisschoolaretwofold.First,emotionsaredifferentiatedfromoneanotherbythecognitivestatesthattheyinvolve.Secondly,thecognitivestatesinvolvedinemotioncanbeunderstoodintermsofapropositionalattitudetheoryofmentalcontent.Mentalstatesareattitudes,suchasbelief,desire,hope,andintention,topropositions.Thesimplestpropositionalattitudetheoryidentifiesemotionswithevaluativejudgments(Solomon1976).Apersonisangryiftheyhavetheattitudeofbelieftothepropositionthattheyhavebeenwronged.Otherprominentvarietiesofpropositionalattitudetheoryarebelief/desiretheories,hybridfeelingtheories,and“seeingas”theories.Belief/desiretheoriesanalyzeemotionsascombinationsofbeliefsanddesires(Marks1982).Hope,forex-ample,isanalyzedasthebeliefthatsomestateofaffairsispossibleandthedesirethatitbeactual.Hybridfeelingtheories,suchasthatofGreenspandiscussedabove,analyzeemotionsascombinationsofpropositionalattitudesandfeelings.Thefeelingcomponentisusedtodifferentiatecoldcognitionfromhot(emo-tional)cognitionandinsometheoriestodistinguishpositivefromnegativeemo-tions.Thespecificidentityoftheemotionisgivenbythepropositionalattitudecomponent.Finally,theincreasinglypopular“seeingas”approacharguesthatasubject’sbeliefsanddesiresaboutanobjectarenotsufficienttoconstituteanemotionunlessthesubject“sees”theobjectintherightway.Atypicalanecdoteinvolvesamountainclimberwhoissaidtoretainthesamebeliefsanddesiresasshefluctuatesbetweenseeingaclimbasterrifyingandseeingitasexhilarating.Earlierversionsofthisapproachwereinclinedtotreat“seeingas”asaprimitiveconcept,followingsomeaspectsofthelaterworkofWittgenstein(Lyons1980).Contemporaryversionsanalyze“seeingas”intermsofattentionalphenomenain289\nPaulE.Griffithscognition.Emotionsarebiasesincognitionthatdirectattentionatsomesourcesofinformationratherthanothersorleadtoahigherweightingforoneconsidera-tionthanforanotherandthusleadtoactionsthatwouldnothaveeventuatedintheabsenceoftheemotion(Calhoun1984;DeSousa1987).Themainconcernofthepropositionalattitudeschoolinthephilosophyofemotionhasbeenwithwhetheremotionsare“rational,”meaningthatanemo-tionalresponsecanbejudgedrightorwronginrelationtothestimulusthatelicitsit.Thefeelingtheoryofemotioniscondemnedforplacingemotionsoutsidetherealmofrationalevaluation.Thisisseenaspartofawiderandinvidioustendencytoseparatetherealmofthemoralfromtherealmoftherational.Thesimplestjudgmentalisttheorybringsemotionsbackintothedomainofreasonbyidentifyingthemwithbeliefs.Anemotionisrationaliftheevaluativebeliefscomposingitarejustifiedbytheevidenceavailabletothesubject.Morecomplexpropositionalattitudetheoriesgivemorecomplexaccountsoftherationalityofemotions.Belief/desiretheoriesfacethedifficultythatformalaccountsofrationality,suchasdecisiontheory,areconfinedtoevaluatingthesuitabilityofmeanstoendsandtaketheends(desires)asgiven.Sothesetheoriesmustprovideanaccountofwhatitisrationaltodesire.Hybridfeelingtheoriescanevaluatetherationalityofhavingoneemotionratherthananotherinthesamewaysasthetheoriesjustmentioned,sincetheidentityofanemotionisdeterminedsolelybyitspropositionalattitudecomponent.Whetherthestateisanemotioninthefirstplace,however,reliesonthefeelingcomponentandsohybridfeelingtheoriesmustgivesomeaccountofwhenitisrationaltotakeone’scognitionhotratherthancold.“Seeingas”theoriesfacetheirowndifficulties,suchasgivinganon-circularaccountofwhatitistoperceiveinanangryorlovingmanner,buttheyhavesomepromisingresourcestobringtobearontherationalityquestion.Thecognitivebiasesthatconstituteemotionsonthistheorycanbeevaluatedfortheirheuristicvalueingeneratingtruebelief,successfulaction,andsoforth,andjudgedrationaliftheyaresuccessfulintheserespects.12.2EvolutionaryTheoriesofEmotion12.2.1DarwinandtheemotionsFacialexpressionshavebeenthesubjectofcarefulinvestigationinanatomyforcenturies,generallywiththeaimofassistingpaintingandsculpture.Thistradi-tionprovidedawealthofanatomicaldataforCharlesDarwin’sTheExpressionoftheEmotionsinManandAnimals(1872).DarwinhadbeencollectingdataontheemotionssincetheMandNnotebooksofthelate1830s,andheoriginallyintendedtoincludethismaterialinTheDescentofMan(1871).Thetwobooksarethereforeintimatelyrelated.InDescent...Darwinaimedtoshowevolution-arycontinuitybetweenanimalsocialbehaviorandhumanmoralityandbetween290\nEmotionstheaestheticsenseofanimalsandofhumans.InExpression...heaimedtoshowevolutionarycontinuityinthefacialexpressionsofhumansandanimalsandthus,byimplication,evolutionarycontinuityintheemotionsunderlyingthoseexpres-sions.Thefundamentalaimofbothbookswastoshowthatineveryrespecthumansdifferfromanimalsonlyindegreeandthusthathumansmighthaveevolvedfromsimplerprecursors.InthePrefacetoExpression...Darwinexpli-citlytargetedSirCharlesBell’sclaimthatthemusclesofthehumanfacewerecreatedbyGodtoexpresshumanemotions.Withthisinmind,hearguedthatmanymovementsthatnowexpressemotionwerevestigesofpreviouswaysoflife:“Withmankindsomeexpressions,suchasthebristlingofthehairundertheinfluenceofextremeterror,ortheuncoveringoftheteethunderthatoffuriousrage,canhardlybeunderstood,exceptinthebeliefthatmanonceexistedinamuchlowerandanimal-likecondition”(1872:12).Darwinarguedthatexpressionsofemotioncouldbeunderstoodthroughthreecomplementaryevolutionaryprinciples.Themostimportantofthesewasthe“principleofserviceableassociatedhabits,”whichisastraightforwardapplicationofDarwin’stheoryofinstinctstothecaseofemotion.Darwinbelievedthatinstinctivebehaviorsderivefromhabitsacquiredbypsychologicalreinforcement.Theconsistentacquisitionofthesamehabitformanygenerationscausesittobecomeahereditary,orinstinctive,behaviorbytheinheritanceofacquiredchar-acteristics,inwhichDarwinwasafirmbeliever.Mostofthedistinctivebehaviorsassociatedwithparticularemotions,suchastheerectionofthehairinfear,reflectlong-sincevanishedlifestylesinwhichthosebehaviorswererewardedandrein-forcedineachgenerationuntiltheywerefinallyincorporatedintothehereditarymaterialasinstincts.Darwinsupplementsthisprinciplewithtwoothers,the“principleofantithesis”andthe“principleofdirectaction.”Hisantithesisprin-ciplepostulatesanintrinsictendencyforoppositestatesoffeelingtoproduceoppositebehaviors.Darwinremarksofasubmissivedogthat:Notoneofthemovements,soclearlyexpressiveofaffection,areoftheleastdirectservicetotheanimal.Theyareexplicable,asfarasIcansee,solelyfromtheirbeingincompleteoppositionorantithesistotheattitudeexpressiveofanger.(1872:51).Darwinexplainsthebehaviorsleftoveraftertheapplicationofthesetwoprin-ciplesastheresultsofthe“directaction”ofthenervoussystem.Excessnerveenergybuiltupinanemotionalepisodeisreleasedinbehaviorssuchassweatingandtremblingfornootherreasonthanthatitmustgosomewhereandthatthesechannelsarephysiologicallyavailableforitsrelease.12.2.2TheemotionsinclassicalethologyTheconceptofinstinctivebehaviorhadlittlecurrencyinthe1920sand1930swhenbehaviorismwasthedominantschoolincomparativepsychology.Itwas291\nPaulE.Griffithsrevivedbythefoundersofclassicalethology,whosawthemselvesasthedirectheirsofDarwin’sworkonmentalevolution(Lorenz1965).TheiraccountoftheevolutionofemotionalexpressionretainsDarwin’sprinciples,butreinterpretsthemtofitthetheoryofevolutionasitemergedduringthe1930sinthe“modernsynthesis”ofDarwinismandMendeliangenetics.Theprincipleofserviceableassociatedhabitsistransformedintotheethologicalconceptsof“ritualization”and“derivedactivity”(Tinbergen1952).Derivedactivitiesarebehaviorsthatoriginallyevolvedforonepurposebutwerelaterselectedforanotherpurpose.Ritualizedbehaviorsarederivedactivitiesthatoriginallyevolvedtofulfillsomepracticalfunctionbutwhichwerelaterselectedtofunctionassignals.Thus,althoughpiloerectioninfearandragedoesnotmakeahumanbeingappearlargertoanopponent,itdoescommunicatetheiremotionalstate.Derivedactivitiesrequireaspecialpatternofevolutionaryexplanation.Theycan-notbeunderstoodpurelyintermsofthefunctiontheycurrentlyperformandtheselectionpressuresthatcurrentlymaintaintheminthepopulation.Thisispar-ticularlyobviousinthecaseofsignals.Havingone’shairsstandonendisnotintrinsicallybetterasasignaloffearthansmilingorlaughing.Thisparticularbehaviorwasselectedasasignalonlybecauseitwasalreadyassociatedwithcertainemotionalstatesinthedistantpast.Itwasassociatedwiththosestatesnotbecauseitwasasignal,butbecauseitmadetheanimalappearlarger.TheconceptofritualizationallowedethologytoreconstructDarwin’sprincipleofserviceableassociatedhabitswhilstavoidinghiscommitmenttotheinheritanceofacquiredcharacteristics.MostofDarwin’sdescriptionsofthepay-offstotheorganismthatcausecertainemotionalbehaviorstobecomehabitualareequallyplausibleasdescriptionsoftheselectiveadvantagethatledtotheevolutionofthosebehaviorsbynaturalselection.Darwin’sothertwoprinciplesareequallyopentoreinterpre-tation.Theprincipleofantithesisisexplainedbytheselectivevalueofunambigu-oussignals.Itisasimportantforadogtosignalthatitwantstoavoidconflictasitisforittosignalaggression.Hencetherecanbeselectionofbehaviorsmerelybecausetheylookdifferentfromthebehaviorsthatsignalaggression.Theprin-cipleofdirectactionwastransformedintotheethologicalconceptofadisplace-mentactivity.TheearlyethologistssharedDarwin’sviewthatinstinctivemotivationscauseabuild-upofmentalenergythatmustbereleasedinsomebehaviororother.Anexamplecommonlygivenisthatofanangrycatthatisunwillingtoattackandbeginstowashitself.NikoTinbergenremarks:“Ithinkitisprobablethatdisplacementsdoserveafunctionasoutlets,throughasafetyvalve,ofdangeroussurplusimpulses”(1952:23).ThiswholesalereinterpretationofDarwin’sthreeprinciplesworkssosmoothlyandallowstheretentionofsomuchofthedetailofhisworkthattheearlyethologistsseemalmostunawareofthedifferencesbetweenDarwin’stheoryandtheirown.AswellasmodernizingDarwin’saccountofemotionalexpressions,classicalethologyofferedanaccountoftheemotionsthemselves,anaccountencapsulatedinOskarHeinroth’sepigram,“Iregardanimalsasveryemotionalpeoplewithverylittleintelligence”(inLorenz1966:180).Lorenzandhisearlyfollowers292\nEmotionsbelievedthatanimalbehaviorisorganizedaroundadefinitenumberofinnatebehaviorsequencesthatareperformedasaunitinthepresenceofasuitablereleasingstimulus.Incontrasttoearlierinstincttheorists,Lorenzdeniedthatanimalsaremotivatedtoseektheactualevolutionarygoalsofanimalbehavior–nutrition,shelter,procreation,andsoforth.Instead,animalsaremotivatedtoperformspecificinnatebehaviors,suchasgatheringnestmaterialsorweavinganest,behaviorsthatunbeknownsttothemwillleadtotheirobtainingshelterandotherfitness-enhancinggoals(Lorenz1957[1937]).Emotionsarethepsycho-logicalaccompanimentstotheperformanceoftheseinnatebehaviorpatterns.Thus,forexample,thebirdinsertingatwigintothenestwithastereotyped,species-specificmovementoftheneckexperiencesasatisfyingemotion(ibid.:138).Theearlierbehaviorsthathaveplaceditinapositiontoperformthissatisfyingmovementwillbereinforcedbythisandperformedmorefrequentlyinfuture.Conversely,awildturkey’sperformanceofitsaerialpredatorresponseisaccompaniedbyanegativeemotionthatwillcauseittoavoidinfuturethecircumstancesassociatedwithperformanceofthatbehaviorpattern.OneofthemostdistinctivetenetsofLorenz’stheoryofemotionisthatanimalshavemanymorekindsofemotionthanhumans(ibid.:163).AccordingtoLorenz,perform-anceofapleasurableinnatebehavior,suchascatchingpreyorproducingaterritorialdisplay,isfrequentlyprecededby“appetitivebehavior”inwhichtheanimalactivelyseeksoutthe“releaser”thatwilldischargetheinnatebehaviorpattern.Inhumanbeings,innatebehaviorsequencesbecomeincreasinglyves-tigialandappetitivebehaviorsbecomeelaboratedintointelligent,goal-directedbehaviors.Whereasabirdbuildsanestbecauseinacertainhormonalstateitfindsitrewardingtogathertwigsand,quiteseparately,rewardingtostamptwigsthathavebeengatheredintoplace,ahumanbuildsashelterasagoal-directedbehaviorsothatitcanobtainthesingle,rewardingfeelingofbeing“athome.”Thelossofsomanyhighlyspecificinnatebehaviorsinhumansmeansthelossofmanyhighlyspecificemotions.Insteadofanemotionalresponsetoaerialthreatsofpredationandaseparateemotionalresponsetoterrestrialthreatsofpredation,thereisasingleemotionoffear(ibid.).Similarly,whilstanotherprimatemighthavesepar-ateemotionstoaccompanydominantthreatanddefensive,subordinatethreat,humanshaveasingleemotionofanger.TheemotiontheoryofLorenzandhisearlyfollowersdidnotsurvivetherejectioninthe1960softhewholeclassicalethologicaltheoryofmotivation–theso-called“hydraulicmodel.”However,theideathatemotionfeelingsplayacriticalroleinsomekindofinternalconditioningprocessisanimportantpartofmanycontemporarytheories,suchasthatofAntonioDamasiodiscussedbelow.AnotherideathathasremainedpopularalmostwithoutinterruptionsinceLorenz’sworkisthatemotionsareaphylogeneticallyancientformofbehaviorcontrolsomepartsofwhichhavebeenretainedinhumansdespitethelaterevolutionofintelligentbehavior.Finally,someoftheargumentsusedagainstLorenz’stheorybytheethologistRobertHindehavesuggestedaradicallynewwaytolookatemotion,asdiscussedbelow.293\nPaulE.Griffiths12.2.3Ekmanand“basicemotions”Untilthe1970stherewasafairlysolidconsensusinpsychologyandanthropo-logythathumanemotionsvarywidelyacrosscultures.Instarkcontrasttotheviewsoftheircontemporariesintheanimalbehaviorcommunity,manyscientistsinthesefieldsbelievedthatculturallyspecificemotionalstatesweresignaledinaculturallyspecificcodeoffacialexpressionsandgesturesacquiredbytheindi-vidualduringtheirupbringing.Thisculturalisttraditionwasdisplacedinthelate1960sbyapowerfulrevivalintheDarwinianapproachwithinpsychologyitself.Today,theworkofPaulEkman(1972)andhiscollaboratorshasproducedanequallysolidconsensusthatcertain“basicemotions”arefoundinallhumancultures.OnefamousexperimentusedsubjectsfromtheForelanguagegroupinNewGuineawithaminimumofpriorcontactwithwesternersandtheirculturalproducts.Thesesubjectsweregiventhreephotographs,eachshowingaface,andtoldastorywhichwasdesignedtoinvolveonlyoneemotion.Theywereaskedtopickthephotographshowingthepersoninthestory.Thisdesignhastheadvant-agethatnotranslationofthenamesofemotionsisneeded.Thesubjectswereverysuccessfulinpickingthephotographoftheappropriateemotionalexpres-sion.TheNewGuineansubjectswerealsoaskedtoactoutthefacialbehaviorofthepeopledescribedinthestories.VideotapesoftheirresponseswereshowntoUScollegestudents.ThestudentsweregenerallyaccurateintheirjudgmentsoftheemotionintendedbytheNewGuineans.Ataroundthesametime,humanethologistsdemonstratedtheearlyemergenceofsomeoftheseexpressionsinhumaninfants(Eibl-Eibesfeldt1973)andprimatologistsreassertedthehomologybetweenhumanfacialexpressionsandthoseofnon-humanprimates(Chevalier-Skolnikoff1973).Thewidelyaccepted“basicemotions”arefear,anger,surprise,sadness,joy,anddisgust,whereeachterminthislistreferstoabrief,involuntaryresponsewithadistinctivefacialexpression.Ekman(1984)seesfacialexpressionsascomponentsofaffectprograms.Eachbasicemotioncorrespondstoanaffectprogramstoredsomewhereinthebrain.Whenactivated,thisprogramcoordinatesacomplexofactionsthatincludefacialexpression,autonomicnervoussystemchanges,expressivevocalchanges,andmuscular-skeletalresponsessuchasflinchingororienting.Theconceptofanaffectprograminheritsmanyofthefeaturesoftheearlieretho-logicalconceptofaninnatebehaviorsequence.Bothconceptssuggestthatcertainapparentlycomplexbehaviorsarereallyatomicunitsofbehaviorthatunfoldinthesame,stereotypedsequencewhenevertheyaretriggeredbyasuitablereleasingstimulus.Ekmancallsthemechanismthatreleasesaffectpro-gramsthe“automaticappraisalmechanism.”Thisisaspecializedneuralsystemthatappliesitsowndistinctiverulesforstimulusevaluationtoalimitedsetofdataderivedfromtheearlieststagesoftheprocessingofperceptualinformation.Consideredtogether,theappraisalmechanismsandaffectprogramsforma294\nEmotionscognitivemoduleinthesensefavoredbymorerecentevolutionarypsychologists(Barkowetal.1992).Theaffectprogramtheorywasaccompaniedbyatheoryoftheevolutionoftheemotionsystem.Thesystemwasinterpretedasanancientformofcognitionthathadoriginallyoperatedonitsownandhadlaterbeensupplementedbyhighercognitivefunctions.ThisviewwassupportedbytheneuroscientistPaulD.MacLean’s(1952)theoryofthe“triunebrain,”accordingtowhichtheemotionsarelocatedinthe“paleomammalian”portionsofthebrainwhilehighercognitivefunctionsarerealizedinmorerecentlyevolved,“neomammalian”structures.Thesurvivaloftheseancientformsofbehaviorcontrolinprimateswasexplainedbytheirvalueasfail-saferesponsesensuringthatvitalbehaviorsareperformedwhen-evernecessaryevenifthatmeanstheyareperformedtoooften.Thisviewoftheemotionsystemasacollectionofprimitivebutreliablefail-safemechanismsremainsinfluentialincontemporaryneuroscience(Panksepp1998).Ekman’saccountofbasicemotionswasaradicaldeparturenotonlybecauseoftheearlieremphasisonculturalvariationinemotion,butalsobecauseitreintro-ducedatypologicalaccountoftheemotionsthemselves.Underlyingemotionalbehavior,thereareadeterminatenumberofdiscreteemotions.InthisrespectEkman’sworkshowstheinfluenceofSilvanS.Tomkins,whoseargumentsforreintroducingemotionsasthe“primarymotivators”helpedtorehabilitateemo-tionasatopicinmainstreampsychology(Tomkins1962).Oneofthemostpersistentlinesofcriticismofbasicemotionstheoryhasbeenfromtheoristswhobelievethatemotionalstatesdonotfallintodiscretetypesbutaredistributedmoreorlesscontinuouslyalonganumberofaxessuchaspleasureandarousal(RussellandFehr1987;Russell1997).12.2.4SociobiologyandtheemotionsSociobiologybroughtanewperspectivetobearontheevolutionofemotioninthe1970sand1980s.Itmovedthefocusofinvestigationfromthebasicemo-tionstothemoralandquasi-moralemotionsinvolvedinhumansocialinterac-tion.Emotionssuchastrust,loyalty,guilt,andshameplayanobviousroleinmediatingthecompetitivesocialinteractionsthatwerethefocusofmostresearchinhumansociobiology.Numeroussociobiologistsmadebriefcommentstotheeffectthatthemoralemotionsmusthaveevolvedaspsychologicalmechanismstoimplementevolutionarystablestrategiesofsocialinteraction(Weinrich1980).RobertA.Frank(1988)suggestedthatthemoralemotionsevolvedassolutionsto“commitmentproblems.”Acommitmentproblemariseswhenthewinningstrategyinanevolutionaryinteractioninvolvesmakingabindingbutconditionalcommitmenttodosomethingthatwouldbeagainstone’sowninterestsiftheconditionwereevermet.Ifsuchacommitmentistobecredible,somespecialmechanismisneededwhichwouldcausetheorganismtoactagainstitsown295\nPaulE.Griffithsinterests.Franksuggeststhatemotionssuchasrageandvengefulnessevolvedtoalloworganismstoengageincredibledeterrence,threateningself-destructiveaggressiontodeteramorepowerfulaggressor.Conversely,emotionssuchasloveandguiltevolvedtoalloworganismstoengageinreciprocalaltruisminsituationswherenoretaliationispossibleifonepartnerfailstoreciprocate.Game-theoreticaccountsofemotionsuchasFrank’shavehadaconsiderableinfluenceonrecentmoralpsychology(Gibbard1990).12.2.5NarrowevolutionarypsychologyandtheemotionsTheterm“evolutionarypsychology”isfrequentlyusedinanarrowsensetorefertothespecificapproachchampionedbyJohnToobyandLedaCosmides(Barkowetal.1992).Themindisacollectionofhighlyspecialized,domain-specificcognitivedevices,ormodules,eachadaptedtoaspecificecologicalprobleminourevolutionarypast.Liketheemotionsystem,thesemodulesoperateonspecifickindsofdatausingalgorithmsthatdifferfromthoseusedbyothermodules.Hence,evolutionarypsychologyendorsestheaffectprogramtheoryofbasicemo-tions,butwantstogofurther,bothbyaddingtothecomplexityoftheknownaffectprogramsandbyfindingmodularmechanismsunderlyingotheremotionalbehaviors(ToobyandCosmides1990;CosmidesandTooby2000).DavidBusshasarguedfortheexistenceofamoduleforsexualjealousy–oneoftheaddi-tionalmodulespredictedbyToobyandCosmides.Bussarguesthatsexualjeal-ousyhassimpleperceptualelicitorssuchasunusualscents,changedsexualbehavior,excessiveeyecontact,andviolationofrulesgoverningpersonalspace(2000:45).Thejealousymoduleusesspecial-purposealgorithmsand,likethebasicemo-tions,itfunctionsasafail-safemechanism“designedtosoundthealarmnotjustwhenaninfidelityhasbeendiscovered,butalsowhenthecircumstancesmakeitslightlymorelikely”(ibid.:224).Themoduleproducesvariousformsofviolenceagainstfemalesexualpartners,including,underconditionsinwhichthisbehaviorwouldhavebeenadaptiveinancestralenvironments,murder.Anoticeablecon-trastbetweentheserecenttheoriesandmoretraditionalaccountsoftheevolutionofemotionistheabsenceoftheideathatemotionsrepresentamoreprimitiveformofbehavioralcontrolthatcanbecontrastedtorational,plannedaction.Theemotionsareseenasjustanothercognitivemodulereflectingdetailsoftheenvir-onmentofevolutionaryadaptedness.Theideathatthereareafewbasicemotionsandthatthesearecomponentsofmorecomplexemotionalresponseshasbeencriticizedbysomeevolutionarypsychologists,since,theyargue,everyemotionisaspecificmoduledesignedtosolveauniqueevolutionaryproblem,andsoallemotionsareequallybasic.EvenPaulEkmanhasallegedlyfallenvictimtothe“standardsocialsciencemodel”andfailedtoappreciatethatallaspectsofouremotionallivesareequallyopentoevolutionaryexplanation(GaulinandMcBurney2001:265–7).296\nEmotions12.2.6ThetransactionaltheoryofemotionEvolutionarytheoryhasalsobeenusedtodefendthetransactionalviewofemo-tion,accordingtowhichemotionsare“moves”madeinsocialinteractionsbetweenorganisms.Theveryideathatemotionalbehavioristheexpressionofdiscrete,underlyingemotionsiscalledintoquestionbytransactionaltheoristsonevolutionarygroundsandonthebasisofanimalmodelsofemotion.Thetransac-tionalviewcanbetracedbacktotheworkofRobertA.Hinde,animportantfigureinthedevelopmentoftheethologicaltraditioninanimalbehaviorresearch.Fromthemid-1950sHinde(1956)arguedthatLorenz’sandTinbergen’sclas-sicalmodelofanimalmotivationintermsofaction-specificdriveshadoutliveditsusefulness.Bythelate1960sanalysesofanimalbehaviorintermsofpostulatedunderlyingmechanismshadbeenreplacedbyadaptivemodelsoftheroleofthebehaviorsthemselvesininteractionsbetweenanimalsandbetweenanimalsandtheirenvironments.Behaviorsthathadpreviouslybeentreatedastheexpressionofinstinctivedriveswerenowtreatedassignalsoftheanimal’slikelyfuturebehaviororofitsmotivationalstate.Buttheapplicationofevolutionarygametheorytoemotionalbehaviorpredictsthatitwillbedesignedtomanipulatetheexpectationsofotherorganismsratherthantotransparently“express”thetruemotivationalstateoftheorganism.Ratherthanexpressingtheanimal’sunder-lyingmotivation,anemotionalbehaviorsendsasignalabouttheanimal’smotiva-tionthatiscredibleandtheacceptanceofwhichbyotherorganismswouldbeadvantageous:“[threat]signalsmakesenseonlyifthethreateningindividualisattemptingtobluff,deceiveormanipulatetherival...orelseisuncertainaboutwhattodonextbecausewhatheshoulddodependsinpartonthebehavioroftheother”(Hinde1985b:989).Theseideasaboutanimalcommunicationwerecommonplacebythe1980s,butHindeusedthemtoquestionwhetherthefolkpsychologyofhumanemotionisagoodstartingpointforstudyingtheanimalbehaviorthatappearshomologoustoemotionalbehaviorinhumans.Folkpsy-chologyleadsustoexpectthatananimalengagedinanaggressiveterritorialdisplayis“feelingangry.”Italsosuggeststhatitisthebasicstimulussituation–anintrusionintotheterritory–thatproducesangerandthatnotdisplayingangerinvolvesamentalefforttocontrolorsuppressit,somethingthatisdifficultandmayonlypartiallysucceed.Finally,folkpsychologysuggeststhatitisthesamestate–theanger–thatmotivatesanattackperformedbytheanimalontheintruderimmediatelyafterthedisplay.Hindesuggestedthatwhilesomeemo-tionalbehaviorinanimalsmeetstheseexpectations,muchdoesnot.Territorialdisplayswere,heargued,asignofambivalentmotivation–notsomuchanexpressionofaggressionaspartoftheprocessthatdetermineswhethertheanimalbecomesaggressive.Mostimportantly,thesocialcontextandthelikelyeffectofthebehaviordonotmerelydeterminewhethertheanimalwillexpressorsuppressits“truefeelings”butactuallydeterminewhatemotiontheanimalhas.297\nPaulE.GriffithsAlthoughHinde(1985a,1985b)conductedhisdiscussionmainlyintermsofnon-humananimals,heclearlythoughtthattheseideaswereapplicabletohumanemotion,andthatthestudyofanimalbehaviorcouldbeusedtoloosenthegripofamodelofemotionbuiltintofolk-psychologicaldiscourseandtoallowtheconsiderationofalternatives.Hinde’sideashaveattractedtheattentionofpsy-chologistsinterestedintheroleofsocialcognitionintheproductionandmodu-lationofemotion.ThebestknownoftheseisprobablyAlanFridlund(1994,1997),whohasdevelopedhisideasasacritiqueofEkman’smodelofbasicemotions.FridlundarguesthatfacialexpressionsofemotionareunlikelytobeobligateresponsestosimplestimulisituationsinthewayEkmansuggests,be-causesuchobligatecommunicationofinformationwouldoftennotbeintheinterestsoftheorganism.Ifhumanbeingsareabletodetermineoneanother’smotivationfromfacialinformation,Fridlundargues,thismustbetheresultofan“armsrace”inwhichsignalingorganismsstruggletohidetheirmotivationwhilstrecipientsstruggletodiscoverit.Fridlund’sargumentiscertainlyinlinewiththefundamentalorientationofthegame-theoreticliteratureonanimalcommunica-tion.However,evolutionarytheoryisnotoriousforitsinabilitytopredictthecourseofevolutionarychangeanditwouldbeamistaketogivethistheoreticalargumentmuchweightincomparisontoempiricalstudiesofthereliability,orlackthereof,withwhichpeoplerecognizeoneanother’semotions.Transactionaltheoristshavetriedtomeetthischallengewithempiricalstudiesoftheimport-anceofcontextintheinterpretationoffacialexpression(RussellandFernández-Dols1997).Theyarguethatobserversreademotionalsignificanceintofacesinthelightoftheirunderstandingofthesocialinteractioninwhichthefaceoccurs.Whileitisclearthatcontextisimportantandthatpeopleareoftenunawareofitsrole,italsoseemsundeniablethatpeople,likeotherprimates,doderivesomeinformationaboutthemotivationandactiontendenciesofotherorganismsfromfacialbehavioritself.Thismaybetheresultofan“armsrace”inwhichsignalrecipientshaveoutcompetedsignalsenders,butisprobablyinlargepartduetothefactthat,asHinderecognized,communicativeinteractionsarenotpurelycompetitive.Evolutionary“games”rangefromzero-sumgamestogamesofalmostpurecoordination,andtheevolutionarygamesthathaveshapedfacialexpressionslieatvariouspointsonthatcontinuum.Fridlund’sownempiricalworkhasconcentratedontheroleofsocialcontextintheproductionofemotionalbehavior.Heandothertransactionaltheoristshavedocumentedaudienceeffectsontheproductionofthebasicemotionsandhavearguedthatthisisinconsistentwiththeaffectprogramtheory.Forexample,smilingismorestronglypredictedbythekindofsocialinteractiontakingplaceatsomepointintimethanbythedegreeofsubjectivesatisfactionfeltbythesmil-ingperson.InaseriesofingeniousexperimentsFridlundhasalsotriedtoshowthatsolitarydisplaysoffacialbehaviorarepredictedbythepresenceofan“audi-enceinthehead”–potentialsocialinteractantswhoarethefocusofthesolitaryperson’sthoughts(Fridlundetal.1990).Fridlundframestheseresultsasarefutationofbasicemotiontheory,butitisnotclearthattheresultssupportthis298\nEmotionsinterpretation.ItistruethatEkmanhasarguedthatthe“displayrules”thatmodulateemotionalbehaviorsaccordingtosocialcontextareacquired,culturallyspecific,anddonotinterferewiththeactualinternalworkingoftheautomaticappraisalmechanismandtheaffectprograms(seebelow).Butthereisnothingtopreventanaffectprogramtheoristfrombuildingaudienceeffectsintotheevolved“emotionmodule”itself.Emotionalbehaviorexhibitsaudienceeffectsinmanyorganismsinwhichitseemsmuchmorelikelythattheyarepartoftheevolvedemotionsystemitselfthanthattheyareacquiredbehaviors–organismssuchasdomesticchickens(Marler1997).Inhisdefinitivereviewoftheanimalcommunicationliterature,MarkHauserhasalsoarguedthatFridlundandEkman’sviewsareconsistent.HehassuggestedthatFridlund’sargumentsbearonquestionsaboutthebiologicalfunctionofemotionalbehavior,whilstEkman’saffectprogrammodelisconcernedwiththemechanismsthatproducethatbehavior(Hauser1996:495–6).Thisundoubtedlyexplainssomepartoftheirdisagreement.Insomeplaces,however,Fridlunddoesseemtobediscussingthenatureoftheunderlyingemotionalprocessesandnotmerelytheirbiologicalfunction.Thebroader,transactionalperspectiveonemo-tioncertainlyinvolvesachallengetostandardideasaboutthepsychologicalpro-cessesunderlyingemotionalbehavior.Anangrypersonhasperceivedthatawronghasbeendonetothemandismotivatedtorightthatwrongortoobtainredressforit.Tobehaveangrilybecauseofthesocialeffectsofthatbehavioristobeangryinsincerely.This,however,ispreciselywhattransactionaltheoriesofemo-tionpropose:emotionsare“nonverbalstrategiesofidentityrealignmentandrelationshipreconfiguration”(Parkinson1995:295).Whilethissoundssuper-ficiallylikethebetter-knownideathatemotionsare“socialconstructions”(learntsocialroles),theevolutionaryrationalefortheemotionsview,andtheexistenceofaudienceeffectsinnon-humananimals,warnagainstanyfacileidentificationoftheviewthatemotionsaresocialtransactionswiththeviewthattheyarelearntorhighlyvariableacrosscultures.Indeed,thetransactionalviewmayseemlessparadoxicaltomanypeopleoncetheideathatemotionsarestrategic,socialbehaviorsisseparatedfromtheideathattheyarelearntbehaviorsorthattheyareintentionalactions.12.3TheUniversalityofEmotion12.3.1WhyitmattersEmotionsarewidelybelievedtobeacriticalelementofmoralagencyandofaestheticresponse.Theclaimthatallhealthypeopledisplay,recognize,andrespondtothesameemotionshasbeenusedtosupporttheviewthatmoralandaestheticjudgmentscanhaveuniversalvalidity.Conversely,ifhumanemotionsareasdiverseastheconceptsofemotionembodiedindifferentlanguagesandif299\nPaulE.Griffithshumanscanonlyunderstandtheexpressiverepertoireoftheirownculturalgroup,thiswouldseemtosupportculturalrelativismaboutethicsandaesthetics.12.3.2Ekman’s“neuroculturaltheory”Ekmanandhiscollaboratorshavehandledculturaldifferencesintheexpressionofbasicemotionswiththeconceptofadisplayrule,aconceptexemplifiedinanotheroftheirwell-knownexperiments(Ekman1971,1972).Neutralandstress-inducingfilmswereshownto25Americanand25Japanesecollegestudentswhilsttheywerealoneinaroom.Therepertoireoffacialbehaviorsshownduringthestressphasebythetwosetsofsubjectwasverysimilar.However,whenanexperimenterwasintroducedintotheroomandallowedtoaskquestionsaboutthesubject’semotionsasthestressfilmwasshownagain,thefacialbehavioroftheJapanesedivergedradicallyfromthatoftheAmericans.Videotapesshowedthemomentaryoccurrenceofnegativeemotionalexpressionsandtheirreplace-mentwithpolitesmiles.Thisexemplifiesanimportantfeatureofthedisplayruleconceptualizationofculturaldifferences:theevolvedexpressionsremainintactbutinteractwithculturallyspecificbehaviorstodeterminetheobservablepatternoffacialaction.Attemptstodisguiseemotionsaresubjectto“leakage”fromtheoperationoftheinvoluntaryemotionalresponse.Suchattemptstosuppressemo-tionalbehaviorcanonlyoperatebysimultaneouslyusingthemusclesinvolvedintheexpressionforsomeotherpurpose.Theycannotinterferewiththeactualoperationoftheemotionsystem.Ihavediscussedabovethepossibilitythatsocialcontextmightplayaroleintheactualoperationofthissystem.Thisis,ofcourse,entirelyconsistentwiththefurtheroperationofdisplayrulesofthekindexempli-fiedintheexperimentjustoutlined.12.3.3SocialconstructionismaboutemotionsCulturalrelativismaboutemotionswasrevivedinthe1980saspartofabroaderinterestinthesocialconstructionofmentalphenomena.Thisledtothefirstrealinvolvementbyanalyticphilosophersinthedebateoveruniversality,sincethenewargumentsforsocialconstructionismwereasmuchconceptualasempirical(Solomon1984;Harré1986).Oneinfluentialargumentstartsfromthewidelyacceptedideathatanemotioninvolvesacognitiveevaluationofthestimulus.Inthatcase,itisargued,culturaldifferencesinhowstimuliarerepresentedwillleadtoculturaldifferencesinemotion.Iftwoculturesthinkdifferentlyaboutdanger,then,sincefearinvolvesanevaluationofastimulusasdangerous,fearinthesetwocultureswillbeadifferentemotion.AdherentsofEkman’sbasicemotionstheoryareunimpressedbythisargumentsincetheydefineemotionsbytheirbehavioralandphysiologicalcharacteristicsandallowthatthereisagreatdealofvariationinwhattriggersthesameemotionindifferentcultures.Social300\nEmotionsconstructionistsalsodefinethedomainofemotioninawaythatmakesbasicemotionsresearchlessrelevant.Thesixorsevenbasicemotionsseemtorequireminimalcognitiveevaluationofthestimulus.Socialconstructionistsoftenrefusetoregardthesephysiologicalresponsesasemotionsinthemselves,reservingthattermforthebroadercognitivestateofapersoninvolvedinasocialsituationinwhichtheymightbedescribedas,forexample,angryorjealous.Itisthusunclearwhetherthedebatebetweentheconstructionistsandtheiruniversalistopponentsismorethanmerelysemantic.Onesidehasapreferencefortractable,reductiveexplanations,eveniftheseareoflimitedscope,andtheotherisconcernedthatsciencemayneglectthesocialandculturalaspectsofhumanemotion.12.3.4ConceptualconfusionsinthedebatesoveruniversalityEkman’sworkandsubsequentdiscussionhavehelpedtoclarifysomeoftheissuesabouttheuniversalityofemotion.Theaffectprogramshavethesameoutputacrosscultures,buttheydonothavethesameinput.Therearesomeuniversalelicitorsofaffectprogramsinchildhood,suchasunexpectedloudnoises,whichelicitfear.Therearealsosystematicbiasesintheconditioningofaffectprogramresponsesthatcouldleadtoaconvergenceintheelicitingconditionsforadultresponses(Öhman1993).Thegeneralpicture,however,isthataffectprogramscometobeassociatedwithwhateverstimulilocallyfulfillabroadfunctionalrole,sothatthefearaffectprogramcomestobeassociatedwithwhateverlocallyconstitutesathreat,thedisgustresponsewithwhateverlocallyappearsnoxiousorunclean,andsoforth.Theuniversalityofbasicemotionsdoesnot,therefore,implythattherearenoculturaldifferencesinwhatleadstoemotion.Furtherclarificationresultsfromdistinguishingthequestionofwhetheremotionsarepan-cultural(foundinallcultures)fromthequestionofwhetheremotionsaremonomorphic(foundinallhealthyindividuals).Thetypesofevidencenormallygatheredbyuniversalistsaredesignedtoshowthatemotionsarepan-culturalandhavelittlebearingonthequestionofmonomorphicity.Emotionsmightbepan-culturalbutstillbelikebloodtypeoreyecolor,withseveraldifferenttypesofindividualineachpopulation.Modelsoftheevolutionofsocialemotionstypic-allypredictthatcompetingtypeswillbemaintainedinthesamepopulationthroughcompetition.Itissurprisingthattheissueofwhetheremotionshaveevolvedisstillsostronglylinkedtotheissueofwhetherthereisasingle,uni-versal,humanemotionalnature.Thedebateoveruniversalitycouldalsobeclarifiedbyabandoningthelastvestigesofthetraditionaldichotomybetweenlearntandinnatebehaviors.Somecriticsoftheaffectprogramtheoryhavearguedthatabiologicalperspectiveonemotionisinappropriatemerelybecausetheemergenceandmaintenanceofemotionalresponsesdependsuponenvironmentalfactors(Ratner1989).Conversely,evid-encethatemotionsarepan-culturalandthuslikelytobetheproductsofevolu-tionisstillthoughttoimplythattheseemotionsaregeneticallydeterminedand301\nPaulE.Griffithsresistanttomodificationbyenvironmentalchanges.Theseinferencesignorethefactsthattheenvironmentplaysarichandconstructiveroleinthedevelopmentofeventhemoststereotypicallybiologicaltraits,suchasbodilymorphologyorsexualbehavior.Evolvedemotions,liketherestofevolvedpsychology,willlikelymakeuseofmanyreliablefeaturesoftheenvironmentofthedevelopingchildinordertoconstructandmaintainthemselves.Theywillbeopentoculturalandindi-vidualvariationasaresultofchangesinthesefeatures,aswellasthroughgeneticvariation.Narrowevolutionarypsychologistshaveembracedthisideaandsuggestedthatpsychologicaldifferencesbetweenculturesmayrepresentdifferentoptionsavailablewithinaflexibleprogramfordevelopmentdesignedbyevolution.Thisidea,however,doesnotallowthatenvironmentalchangesmayproduceemotionalphenotypesthathavenospecificevolutionaryhistoryandsodonotformpartoftheevolvedprogramfordevelopment.Togetaroundthisdifficulty,IhavesuggestedthatquestionsofuniversalitycanoftenbeusefullyreframedintermsoftheDarwinianconceptofhomology(Griffiths1997:135).Twoemotionalresponsesarehomologousiftheyaremodifiedformsofaresponseinacommonancestorofthoseindividualsorcultures.Usingtheconceptofhomologyavoidssteriledisputesabouthowsimilartworesponsesmustbetocountas“thesame”response.Iftworesponsesarehomologous,theyshareanevolutionaryhistory,andnomatterhowfartheyhavedivergedsincethen,thatsharedhistorycanbebroughttobearinexplainingthecommonfeaturesthattheyhaveretained.12.4TheEmotionsinCognitiveScience12.4.1TheresurgenceofthefeelingtheoryRecentworkincognitiveneurosciencehasshednewlightontherelationshipbetweenemotionandcognitionandledtoarevivalofthefeelingtheoryofemotion.AntonioDamasiohasarguedthatpracticalreasoningisdependentonthecapacitytoexperienceemotion.Patientswithbilaterallesionstotheprefrontalcortexshowbothreducedemotionalityandadiminishedabilitytoallocatecognitiveresourcesinsuchawayastosolverealworldproblems.Theydonot,however,havedeficitsinabstractreasoningability.Damasio(1994)interpretsthesefindingsasshowingthatemotionplaysanessentialroleinlabelingbothdataandgoalsfortheirrelevancetothetaskinhand.Thesesuggestionshavearousedinterestincognitivescientistswhohaveseenin“affectivecomputing”apossiblesolutiontothe“frameproblem”:theproblemofchoosingallandonlytherelevantdatawithoutassessingalltheavailabledataforpossiblerelevance(Picard1997).Damasio’stheorybearsaresemblancetosomeofthephilo-sophical“seeingas”theoriesthatidentifyemotionswithheuristicbiasesincog-nition.Incontrasttothosetheories,however,Damasioseesemotionsthemselvesasfeelings.Thisisimportant,sinceifemotionsfunctionedcognitively,thenhis302\nEmotionsproposalwouldbethatcognitiveprioritiesareassignedbycalculatingwhatismostrelevantandimportant.Thiswouldnotbeasolutiontotheframeprob-lem,butaninstanceofthatproblem.Damasioavoidsthistrapbyusingemotionfeelingstoprioritizecognition.Hedescribesaclassof“primaryemotions”thatbearastrongaffinitytoEkman’sbasicemotions.Damasioenvisagesemotionaldevelopmentasaprocessinwhichthefeelingsassociatedwiththebasicemotionsbecomeattachedtoparticularcognitivestatesgivingrisetocognition/feelingcompositesthathelabels“secondaryemotions.”Damasiohassofargivenonlyasuggestiveoutlineofhistheoryanditremainstobeseenwhetherthissketchcanbedevelopedintoaworkablemodelofcognitiveprocesses.AttemptstoexpandonDamasio’sideastodateresembletraditionalbehaviorconditioningwiththoughtstakingtheplaceofbehaviorsandemotionfeelingsactingasreinforcers.Thelimitationsofconditioningmodelsasexplanationsofcomplexcognitiveperformancesarewellknown.12.4.2Neurologicalsupportfortwin-pathwaymodelsofemotionOneofthemostheatedcontroversiesinemotiontheoryinthe1980sconcernedRobertZajonc’s“affectiveprimacythesis”(1980).Zajoncshowedthatsubjectscouldacquirepreferencesforsubliminalstimuliwhileshowingnoabilitytorecog-nizethosestimuliwhentheywerepresentedforlongerperiods.Hearguedthat,inthenormalcase,twoseparatepathwaysledtoemotionalresponsesandparadigmaticcognitiveresponsessuchasconsciousawarenessandrecall.Zajonc’sclaimswerecontroversialbecauseofthewidespreadviewthatanemotionessen-tiallyinvolvesan“evaluation”ofthestimulus,somethingthatwastakentobeaparadigmaticallycognitiveprocess(Lazarus1982;Lazarusetal.1984;Zajonc1984).Zajonc’sconceptoftwinpathwaystocognitionandemotionhasobvioussimilaritiestoEkman’sproposalthatan“automaticappraisalmechanism”isasso-ciatedwiththebasicemotionsandoperatesindependentlyoftheformationofconsciousorreportablejudgmentsaboutthestimulussituation.Inmorerecentyears,JosephLeDoux’s(1996)detailedmappingoftheneuralpathwaysinvolvedinfearconditioninghasconfirmedsomethinglikeZajonc’stwin-pathwaymodelforfear.Informationaboutthestimulusactivatesmanyaspectsofemotionalresponseviaafast,“lowroad”throughsub-corticalstructures,amongstwhichtheamlygdalaisparticularlyimportant.Aslower,“highroad”activatescorticalstructuresandisessentialforlonger-term,planned,andoftenconsciousresponsestothesamestimulus.LeDoux’sfindingssuggestthatatleastforcertainbasicemotionstheideathatanemotioninvolvesacognitiveevaluationofthestimulusneedstobereplacedwiththeideathatitinvolvestwoevaluations,whichcanconflictandwhichhavecomplimentarybutindependentcognitivefunctions.Twin-pathwaymodelsalsoprovidesomesupportforthemanyevolutionaryaccountsthatseethebasicemotionsas“quickanddirty”solutionstocommonsurvivalproblems.303\nPaulE.Griffiths12.5IsEmotionaNaturalKind?Damasiohasdefinedanemotionas“aspecificallycausedtransitionoftheorgan-ismstate”(1999:282).Confrontedbysimilardefinitions,Fridlundhasremarked:“Here,thelogicalquestioniswhatisn’temotion.Emotionhas,infact,replacedBergson’sélanvitalandFreud’slibidoastheenergeticbasisofallhumanlife”(1994:185).Formanytheorists,emotionhasindeedbecomesynonymouswithmotivationasawhole.Damasioiswellawareofthissituationandisself-consciouslyusingafamiliartermforhisownpurposesinordertofacilitatecom-municationinwhatheseesasaperiodofconceptualupheaval(Damasio1999:341).Giventheextremedifficultyof,forexample,distinguishingbetweenmoodandemotionordecidingwhether(some?)desiresareemotionsexceptinthelightofanactualtheoryoftheemotions,adoptingDamasio’sbroaddefinitionasastartingpointforinquiryhassomethingtorecommendit.Ihavearguedelse-where,however,thatthescientificinvestigationofthedomainofaffectivephe-nomenahasbeenhinderedbyacontinuedbeliefthat“theemotions”areaunitarykindofpsychologicalstate(Griffiths1997).Scienceaimstogroupphe-nomenainto“naturalkinds”:categoriesaboutwhichtherearemany,reliablegeneralizationstobediscovered.Thefolk-psychologicaldomainofemotionissodiversethatitisunlikelythatallthepsychologicalstatesinthatdomainformanaturalkind.Hencetherewillbefewifanyreliablegeneralizationsaboutemotionor,inotherwords,notheoryofemotioningeneral.Scientificprogresswouldbeservedbydividingupthedomainandinvestigatinggroupsofphenomenathatarelikelytoformnaturalkinds,ashasoccurredinresearchintomemory.New,morespecificconceptswillberequiredtoreplacetheemotionconceptandacentralroleforphilosophersofemotionistofacilitatethiskindofconceptualrevision.Mostphilosophersofemotionseenoseriousproblemwiththecategoryofemotion,althoughtheyadmitthatitisvagueandcoversadiverserangeofphenomena.Theirconcerniswiththeword“emotion”ineverydaylanguageandtheconceptthatliesbehindit.Philosophicalanalysesoftheemotionconceptareinreasonableagreementwiththoseproducedbypsychologistsstudyingtheuseoftheterm“emotion”inwesterncultures(FehrandRussell1984).Thereareclearparadigmsofemotion,suchaslove,happiness,anger,fear,andsadness,andmostphilosophersdefineemotionsoastoincludethese.Theirdefinitionsdis-agreeoverthesamecasesthatproducedisagreementbetweensubjectsinempiricalstudies,casessuchaspride,hope,lust,pain,andhunger.Philosophicaldefinitionsincludefeaturesthatpsychologistshavearguedarepartoftheprototypeoftheemotionconcept.Emotionsaredirectedontoexternalstatesofaffairs,arerelat-ivelyshort-lived,andhaveanevaluativeaspecttothem,suchthattheirobjectsarejudgedtobeeitherattractiveoraversive.Mostdefinitionsalsoprovidearoleforemotionfeelings.Hencephilosophers,likeordinaryspeakers,canachieveareasonablelevelofagreementaboutwhatcountsasanemotion,asopposedtoa304\nEmotionsmood,adesire,oranintention.Whetherthepsychologicalstatesgroupedtogetherinthiswayformasingle,productiveobjectofscientificinvestigationandwhetherotherculturesconceptualizeemotioninthesamewayremainstobeseen.12.6ConclusionThephilosophicalpsychologyofemotionisathrivingfield,withalargenumberofbooksandarticlesappearingeachyear.Thereisatrendtowardcloserintegra-tionwiththesciencesofthemind,anintegrationofthekindfamiliarfromthephilosophicalpsychologyofcognition,perception,andaction.Theevolutionarypsychologyofemotionhasreceivedphilosophicalattentioninrecentyears(Griffiths1997;Horst1998;Evans2001),ashasthepotentialofemotiontochallengeviewsincognitivesciencederivedfromthestudyofcognition(Delancey2001;Evans,inpress).Theemotiontheoriesproposedbyneuroscientistsonthebasisofrecentadvancesinaffectiveneurosciencehavealsobeenexposedtophilosoph-icalscrutiny(Prinz,forthcoming).Moretraditionalphilosophicalwork,orientedtowardsissuesinethicsandaesthetics,hasalsobeguntodrawontheclaimsofaffectiveneuroscience,perhapsbecauseDamasio’sclaimthatemotionandration-alityareinseparableresonatessostronglywitholderphilosophicalviews(Blackburn1998;Nussbaum2001).ReferencesBarkow,J.H.,Cosmides,L.,andTooby,J.(eds.)(1992).TheAdaptedMind:Evolution-aryPsychologyandtheGenerationofCulture.Oxford:OxfordUniversityPress.Blackburn,S.(1998).RulingPassions:ATheoryofPracticalReasoning.OxfordandNewYork:OxfordUniversityPress.Buss,D.M.(2000).TheDangerousPassion:WhyJealousyisasEssentialasLoveandSex.NewYork:SimonandSchuster.Calhoun,C.(1984).“CognitiveEmotions?”InC.CalhounandR.C.Solomon(eds.),WhatisanEmotion:ClassicReadingsinPhilosophicalPsychology.NewYork:OxfordUniversityPress.Chevalier-Skolnikoff,S.(1973).“FacialExpressionofEmotioninNon-humanPrimates.”InP.Ekman(ed.),DarwinandFacialExpression:ACenturyofResearchinReview.NewYorkandLondon:AcademicPress:11–89.Cosmides,L.andTooby,J.(2000).“EvolutionaryPsychologyandtheEmotions.”InM.LewisandJ.M.Haviland-Jones(eds.),HandbookoftheEmotions,2ndedn.NewYorkandLondon:GuildfordPress:91–115.Damasio,A.R.(1994).DescartesError:Emotion,ReasonandtheHumanBrain.NewYork:Grosset/Putnam.——(1999).TheFeelingofWhatHappens:BodyandEmotionintheMakingofConscious-ness.NewYork:HarcourtBrace.305\nPaulE.GriffithsDarwin,C.(1872).TheExpressionsofEmotionsinManandAnimals,1stedn.NewYork:PhilosophicalLibrary.——(1981/1871).TheDescentofManandSelectioninRelationtoSex,Facsimileofthe1stedn.Princeton:PrincetonUniversityPress.DeSousa,R.(1987).TheRationalityofEmotions.Cambridge,MA:MITPress.Deigh,J.(1994).“CognitivismintheTheoryofEmotions.”Ethics,104:824–54.Delancey,C.(2001).PassionateEngines:WhatEmotionsRevealAboutMindandArtificialIntelligence.NewYorkandOxford:OxfordUniversityPress.Eibl-Eibesfeldt,I.(1973).“ExpressiveBehaviouroftheDeafandBlindBorn.”InM.vonCranachandI.Vine(eds.),SocialCommunicationandMovement.LondonandNewYork:AcademicPress:163–94.Ekman,P.(1971).“UniversalsandCulturalDifferencesinFacialExpressionsofEmo-tion.”InJ.K.Cole(ed.),NebraskaSymposiumonMotivation4.Lincoln,Nebraska:UniversityofNebraskaPress:207–83.——(1972).EmotionsintheHumanFace.NewYork:PergamonPress.——(1984).“ExpressionsandtheNatureofEmotions.”InK.SchererandP.Ekman(eds.),ApproachestoEmotions.Hillsdale,NJ:Erlbaum.Evans,D.(2001).Emotion:TheScienceofSentiment.Oxford:OxfordUniversityPress.——(InPress).RethinkingEmotion:AStudyintheFoundationsofMind.Cambridge,MA:MITPress.Fehr,B.,andRussell,J.A.(1984).“ConceptofEmotionViewedfromaPrototypePerspective.”JournalofExperimentalPsychology:General,113;464–86.Frank,R.H.(1988).PassionsWithinReason:TheStrategicRoleoftheEmotions.NewYork:Norton.Fridlund,A.(1994).HumanFacialExpression:AnEvolutionaryView.SanDiego:Aca-demicPress.——(1997).“TheNewEthologyofHumanFacialExpressions.”InJ.A.RussellandJ.M.Fernández-Dols(eds.),ThePsychologyofFacialExpressions.Cambridge:CambridgeUniversityPress:103–29.Fridlund,A.J.,Schaut,J.A.,Sabini,J.P.,Shenker,J.I.,Hedlund,L.E.,andKnauer,M.J.(1990).“AudienceEffectsonSolitaryFacesDuringImagery:DisplayingtothePeopleinYourHead.”JournalofNonverbalBehaviour,14(2):113–37.Gaulin,S.J.C.andMcBurney,D.H.(2001).Psychology:AnEvolutionaryApproach.UpperSaddleRiver,NJ:PrenticeHall.Gibbard,A.(1990).WiseChoices,AptFeelings:ATheoryofNormativeJudgment.Cam-bridge,MA:HarvardUniversityPress.Greenspan,P.(1988).EmotionsandReasons:AnInquiryintoEmotionalJustification.NewYork:Routledge.Griffiths,P.E.(1989).“TheDegenerationoftheCognitiveTheoryofEmotion.”Philo-sophicalPsychology,2(3):297–313.——(1997).WhatEmotionsReallyAre:TheProblemofPsychologicalCategories.Chicago:UniversityofChicagoPress.Harré,R.(1986).“AnOutlineoftheSocialConstructionistViewpoint.”InHarré(ed.),TheSocialConstructionofEmotion.Oxford:OxfordUniversityPress:2–14.Hauser,M.D.(1996).TheEvolutionofCommunication.Cambridge,MA:MITPress.Hinde,R.A.(1956).“EthologicalModelsandtheConceptof‘Drive’.”BritishJournalforthePhilosophyofScience,6:321–31.306\nEmotions——(1985a).“ExpressionandNegotiation.”InG.Zivin(ed.),TheDevelopmentofExpressiveBehavior.NewYork:AcademicPress:103–16.——(1985b).“Was‘TheExpressionofEmotions’aMisleadingPhrase?”AnimalBehaviour,33:985–92.Horst,S.(1998).“OurAnimalBodies.”InP.A.FrenchandH.K.Wettstein(eds.),ThePhilosophyofEmotion.NotreDame,IN:UniversityofNotreDamePress.Lazarus,R.S.(1982).“ThoughtsontheRelationsBetweenEmotionandCognition.”AmericanPsychologist,37:1019–24.Lazarus,R.S.,Coyne,J.C.,andFolkman,S.(1984).“Cognition,EmotionandMotivation:DoctoringHumptyDumpty.”InK.SchererandP.Ekman(eds.),ApproachestoEmotions.Hillsdale,NJ:Erlbaum:221–37.LeDoux,J.(1996).TheEmotionalBrain:TheMysteriousUnderpinningsofEmotionalLife.NewYork:SimonandSchuster.Lorenz,K.(1957[1937]).“TheNatureofInstinct.”InC.H.Schiller(ed.),InstinctiveBehavior:TheDevelopmentofaModernConcept.NewYork:InternationalUniversitiesPress:129–75.——(1965).Prefaceto“TheExpressionoftheEmotionsinManandAnimals”byCharlesDarwin.Chicago:UniversityofChicago.——(1966).OnAggression,trans.byM.K.Wilson.NewYork:Harcourt,BraceandWorld.Lyons,W.(1980).Emotion.Cambridge:CambridgeUniversityPress.MacLean,P.D.(1952).“SomePsychiatricImplicationsofPhysiologicalStudiesonFrontotemporalPortionsoftheLimbicSystem(VisceralBrain).”ElectroencephlographyandClinicalNeurophysiology,4:407–18.Marks,J.(1982).“ATheoryofEmotions.”PhilosophicalStudies,42:227–42.Marler,P.(1997).“AnimalSoundsandHumanFaces:DotheyhaveAnythinginCom-mon?”InJ.A.RussellandJ.M.Férnandez-Dols(eds.),ThePsychologyofFacialExpres-sion.Cambridge:CambridgeUniversityPress:133–226.Nussbaum,M.C.(2001).UpheavalsofThought:TheIntelligenceofEmotions.CambridgeandNewYork:CambridgeUniversityPress.Öhman,A.(1993).“StimulusPrepotencyandFear:DataandTheory.”InN.BirbaumerandA.Öhman(eds.),TheOrganizationofEmotion:Cognitive,ClinicalandPsychologi-calPerspectives.Toronto:Hogrefe.Panksepp,J.(1998).AffectiveNeuroscience:TheFoundationsofHumanandAnimalEmo-tions.OxfordandNewYork:OxfordUniversityPress.Parkinson,B.(1995).IdeasandRealitiesofEmotion.LondonandNewYork:Routledge.Picard,R.(1997).AffectiveComputing.Cambridge,MA:MITPress.Prinz,J.(forthcoming).EmotionalPerception.Oxford:OxfordUniversityPress.Ratner,C.(1989).“ASocialConstructionistCritiqueoftheNaturalisticTheoryofEmo-tion.”JournalofMindandBehaviour,10(3):211–30.Russell,J.A.(1997).“ReadingEmotionfromandintoFaces:ResurrectingaDimensional-ContextualPerspective.”InJ.A.RussellandJ.M.Fernández-Dols(eds.),ThePsychologyofFacialExpression.Cambridge:CambridgeUniversityPress:295–320.Russell,J.A.andFehr,B.(1987).“RelativityinthePerceptionofEmotioninFacialExpressions.”JournalofExperimentalPsychology(General),116:223–37.Russell,J.A.andFernández-Dols,J.M.(1997).ThePsychologyofFacialExpression.Cambridge:CambridgeUniversityPress.307\nPaulE.GriffithsSchachter,S.andSinger,J.E.(1962).“Cognitive,SocialandPhysiologicalDeterminantsofEmotionalState.”PsychologicalReview,69:379–99.Solomon,R.(1976).ThePassions.NewYork:Doubleday.——(1984).“GettingAngry:TheJamesianTheoryofEmotioninAnthropology.”InR.A.SchwederandR.A.LeVine(eds.),CultureTheory:EssaysonMind,SelfandEmotion.Cambridge:CambridgeUniversityPress:238–54.Tinbergen,N.(1952).“DerivedActivities:TheirCausation,BiologicalSignificance,OriginandEmancipationDuringEvolution.”QuarterlyReviewofBiology,27(1):1–32.Tomkins,S.S.(1962).Affect,ImageryandConsciousness.NewYork:Springer.Tooby,J.andCosmides,L.(1990).“ThePastExplainsthePresent:EmotionalAdapta-tionsandtheStructureofAncestralEnvironments.”EthologyandSociobiology,11:375–424.Weinrich,J.D.(1980).“TowardsaSociobiologicalTheoryofEmotions.”InR.PlutchikandH.Kellerman(eds.),Emotion:Theory,ResearchandExperience,vol.1:TheoriesofEmotion.NewYork:AcademicPress:113–40.Zajonc,R.B.(1980).“FeelingandThinking:PreferencesNeednoInference.”AmericanPsychologist,35:151–75.——(1984).“OnthePrimacyofAffect.”InK.SchererandP.Ekman(eds.),ApproachestoEmotion.Hillsdale,NJ:LawrenceErlbaumAssociates:259–70.308\nTheBlackwellGuidetoPhilosophyofMindArtificialIntelligenceEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter13ArtificialIntelligenceandtheManyFacesofReasonAndyClark13.1PullingaThreadIshallfocusthisdiscussionononesmallthreadintheincreasinglycomplexweaveofartificialintelligence(AI)andphilosophyofmind:theattempttoexplainhowrationalthoughtismechanicallypossible.Thisis,historically,thecrucialplacewhereAImeetsphilosophyofmind.Butitis,Ishallargue,aplaceinflux.Forourconceptionsofwhatrationalthoughtandreasonare,andofwhatkindsofmechanismmightexplainthem,areinastateoftransition.Togetasenseofthisseachange,Ishallcompareseveralvisionsandapproaches,startingwithwhatmightbetermedtheTuring–Fodorconceptionofmechanicalreason,proceedingthroughconnectionismwithitsskill-basedmodelofreason,thenmovingtoissuesarisingfromrobotics,neuroscientificstudiesofemotionandreason,andworkon“ecologicalrationality.”Asweshallsee,thereisprobablybothmore,andless,tohumanrationalitythanoriginallymettheeye.First,though,thebasic(andIdomeanbasic)story.13.2TheCoreIdea,ClassicallyMorphedOnecoreidea,commontoalltheapproachesI’llconsiderinthischapter,isthatsometimesformcandodutyformeaning.Thisissurelythecentralinsightuponwhichallattemptstogiveamechanicalaccountofreasonarebased.Broadlyunderstood,itisthissametrickthatisatworkinlogic,intheTuringMachine,insymbolicAI,inconnectionistAI,andevenin“anti-representationalist”robotics.Thetrickistoorganizeandorchestratesomesetofnon-semanticallyspecifiablepropertiesorfeaturessothatadevicethusbuilt,inasuitableenviron-ment,canendupdisplaying“semanticgoodbehavior.”Theterm“semanticgood309\nAndyClarkbehavior”covers,intentionally,awidevarietyofthings.Itcoversthecapacitytocarryoutdeductiveinferences,tomakegoodguesses,tobehaveappropriatelyuponreceiptofaninputorstimulus,andsoon.Anythingthat(crudelyput)lookslikeitknowswhatitisdoingisexhibitingsemanticgoodbehavior:casesincludethelogicianwhoinfers−Afrom(−AvB,−B),thepersonwhochoosestotakeoutanumbrellabecausetheybelieveitwillrainanddesiretostaydry,thedogwhochoosesthefoodratherthanthetoxin,therobotthatrecoversitsbalanceandkeepsonwalkingafteronelegisdamaged.There’salotofsemanticgoodbehavioraround,andweunderstandsomeofitawholelotbetterthantherest.Where,though,doesreasoncomeintothepicture?Reason-governedbehavioris,arguablyatleast,aspecialsubsetofwhatIamcallingsemanticgoodbehavior.ItisJerryFodor’sview,forexample,thatitwasnotuntiltheworkofTuringthatwebegantohaveasenseofhowrationality(whichI’llassumetomeanreason-governedbehavior)couldbemechanicallypossible(foranicecapsulestatement,seeFodor1998:204–5).Formallogicshowedusthattruthpreservationcouldbeinsuredsimplybyattendingtoform,notmeaning.BfollowsfromAandBregardlessofwhatAmeansandwhatBmeans,andifyourkeeptorulesdefinedovertheshapesofsymbolsandconnect-ivesyouwillneverinferafalsehoodfromtruepremises,evenifyouhavenoideawhateitherthepremisesortheconclusionsareabout.Turing,asFodornotes,showedthatforallsuchformally(“byshape”)specifiableroutines,awell-programmedmachinecouldreplacethehuman.Itisataboutthispointthatwhatwasinitiallyjustanassertionofphysicalistfaith(thatsomehoworother,semanticgoodbehaviorhasalwaysandeverywhereanexplanatorilysufficientmaterialbase)morphsintoagenuineresearchprogramtargetingreason-governedbehavior.Theidea,rapidlyenshrinedintheresearchprogramofclassical,symbolicAI,wasthatreasoncouldbemechanicallyexplainedastheoperationofappropriatecomputationalprocessesonsymbols,wheresymbolsarenon-semanticallyindivisibleitems(itemstypedbyform,shape,voltage,whatever)andcomputationalprocessesaremechanical,automaticpro-cessesthatrecognize,write,andamendsymbolsinaccordancewithrules(whichthemselves,uptoacertainpoint,canbeexpressedassymbols).Insuchsystems,asHaugelandfamouslyremarks,“ifyoutakecareofthesyntax[thenon-semanticfeaturesandproperties]thesemanticswilltakecareofitself”(1981:23).Thecoreidea,asviewedthroughthelensofbothTuring’sremarkableachievementsandthenfurtherdevelopmentsinclassicalAI,thusbegantolookbothmoreconcrete,andlessgeneral.Itbecametheidea,inFodor’swords,that“some,atleast,ofwhatmakesmindsrationalistheirabilitytoperformcomputationsonthoughts;whenthoughts...areassumedtobesyntacticallystructured,andwhere‘computation’meansformaloperationsinthemannerofTuring”(1998:205).Thegeneralideaofusingform(broadlyconstrued)tododutyformeaningthusgentlymorphedintotheTuringMachine-dominatedvisionofreading,writ-ing,andtransposingsymbols:avisionwhichfoundfullexpressioninearlywork310\nArtificialIntelligenceinAI.HereweencounterNewellandSimon’s(1976)depictionofintelligenceasgroundedintheoperationsofso-calledphysicalsymbolsystems:systemsinwhichnon-semanticallyidentifiableentitiesactasthevehiclesofspecificcontents(thusbecoming“symbols”)andaresubjecttoavarietyoffamiliaroperations(typicallycopying,combining,creating,anddestroyingthesymbols,accordingtoinstruc-tions).Forexample,thestory-understandingprogramofSchank(1975)usedaspecialevent-descriptionlanguagetoencodethekindofbackgroundknowledgeneededtorespondsensiblytoquestionsaboutsimplestories,thusdevelopingasymbolicdatabasetohelpit“fillin”themissingdetails.Consideredasstoriesabouthowrational,reason-guidedthoughtismechan-icallypossible,theclassicalapproachthusdisplaysasatisfyingdirectness.Itex-plainssemanticallysensiblethoughttransitions(“theyenjoyedthemeal,sotheyprobablyleftatip;”“it’sraining,Ihatetherain,soI’lltakeanumbrella”)byimaginingthateachparticipatingthoughthasaninnersymbolicecho,andthattheseinnerechoessharerelevantaspectsofthestructureofthethought.Asaresult,syntax-sensitiveprocessescanregulateprocessesofinference(thought-to-thoughttransitions)inwaysthatrespectsemanticrelationsbetweenthethoughts.13.3TheCoreIdea,Non-classicallyMorphedTheideathatreason-guidedthoughttransitionsaregroundedinsyntacticallydrivenoperationsoninnersymbolstringshasafamouscompetitor.Thecompet-ingidea,favoredby(many)researchersworkingwithartificialneuralnetworks,isthatreason-guidedthoughttransitionsaregroundedinthevector-to-vectortrans-formationssupportedbyaparallelwebofsimpleprocessingelements.Aproperexpressionofthefulldetailsofthiscontrastisbeyondthescopeofthischapter(seeClark1989,1993formybestattempts).Butwecanatleastnoteoneespeciallyrelevantpointof(Ithink)genuinecontrast.ItconcernswhatI’llcallthe“besttargets”ofthetwoapproaches.Forclassical(TuringMachine-like)AI,thebesttargetsarerationalinferencesthatcanbedisplayedandmodeledinsententialspace.By“sententialspace”Imeananabstractspacepopulatedbymeaning-carryingstructures(interpretedsyntacticitems)thatsharethelogicalformofsentences:sequentialstringsofmeaningfulelements,inwhichdifferentkindsofsyntacticitemreliablystandfordifferentthings,andinwhichtheoverallmeaningisafunctionoftheitems(tokens)andtheirsequentialorder,includingthemodifyingeffectsofothertokens(e.g.the“not”in“itisnotraining”).Rationalinferencesthatcanbesatisfyinglyreconstructedinsententialspacein-cludeallofFodor’sfavoriteexamples(aboutchoosingtotaketheumbrella,etc.),allcasesofdeductiveinferencedefinedoversententialexpressions,andallcasesofabductiveinference(basically,goodguessing)inwhichthelinkbetweenpremisesandconclusionscanbemadebythecreativeretrievalordeploymentofadditionalsentences(asinSchank’sstory-understandingprogrammentionedearlier).311\nAndyClarkThebesttargetsfortheartificialneuralnetworkapproach,bycontrast,arevariousspeciesofreasonable“inference”inwhichtheinputsarebroadlyspeakingperceptualandtheoutputsare(often)broadlyspeakingmotoric.Reasonableinferencesofthiskindareimplicitin,forexample,thecat’srapidassessmentoftheload-bearingcapacityofabranch,leadingtoaswiftandelegantleaptoamoresecurerestingpoint,orthehandwritingexpert’srapidintuitiveconvictionthatthesignatureisaforgery,aconvictiontypicallyachievedinadvanceoftheconsciousisolationofspecifictell-talesigns.Thisisnottosay,however,thattheconnectionistapproachislimitedtotheperceptuo-motordomain.Rather,thepointisthatitstakeonrationalinference(and,morebroadly,onrationalchoice)isstructurallycontinuouswithitstakeonperceptuo-motorskill.Reasoningandinferencearereconstructed,onalllevels,as(roughlyspeaking)processesofpattern-completionandpattern-evolutioncarriedoutbycascadesofvector-to-vectortransformationsbetweenpopulationsofsim-pleprocessingunits.Forexample,anetworkexposedtoaninputdepictingthevisualfeaturesofared-spottedyounghumanfacemaylearntoproduceasoutputapatternofactivitycorrespondingtoadiagnosisofmeasles.Thisdiagnosismaylead,viaasimilarmechanism,toaprescriptionofpenicillin.Thevector-to-vectortransformationsinvolvedareperfectlycontinuous(onthismodel)withthosebywhichweperformmorebasicactsofrecognitionandcontrol,aswhenwerecog-nizeafamiliarfaceorcoordinatevisualproprioceptiveinputsinwalking.Suchpattern-completingprocesses,carriedoutinnetworksofsimpleprocessingunitsconnectedbynumericallyweightedlinks,areprimafaciequiteunlikethesententialAImodelsinwhichamedicaljudgment(forexample)mightdependontheconsultationofastoredsetofrulesandprinciples.Oneimportantsourceofthedifferenceliesinthewaytheconnectionistsystemtypicallyacquirestheconnec-tionweightsthatactbothasknowledge-storeandprocessing-engine.Suchweightingsareacquiredbyexposingthesystemtoawiderangeofexemplars(traininginstances):aregimewhichleads,courtesyofthespeciallearningrulesdeployed,tothedevelopmentofaprototype-dominatedknowledgebase(seeChurchland1989).Whatthismeansinpracticeisthatthesystemlearnsto“thinkabout”adomainintermsofthemostsalientfeaturesofabodyofexemplarcases,andthatitsresponses,judgments,andactionsareguidedbytheperceivedsimi-larityofthecurrentcasetothepatternsoffeaturesandresponsesmostcharacter-isticoftheexemplars.Andwhatthismeans,inturn,isthatwhatsuchasystemknowsisseldom,ifever,neatlyexpressibleasasetofsentences,rules,orpro-positionsaboutthedomain.Makingtheexpertmedicaljudgment,onthismodel,hasmoreincommonwithknowinghowtorideabicyclethanwithcon-sultingasetofrulesinasymbolicdatabase.Awell-tunedconnectionistnetworkmaythusissuejudgmentsthatarerationallyappropriatebutthatnonethelessresistquasi-deductivesententialreconstructionastheconclusionofanargumentthattakessymbolicexpressionsasitspremises.Suchappropriateresponsesandjudgmentsare,onthisview,thefundamentofreason,andofrationality.Lingua-formargumentandinferenceisdepictedasjustaspecialcaseofthisgeneral312\nArtificialIntelligenceprototype-basedreasoningcapacity,differentonlyinthatthetargetandtrainingdomainhereinvolvethesymbolstringsofpublicspeechandtext.Connectionismandclassicismthusdiffer(atleastinthecharacteristicincarnationsIamconsidering)intheirvisionsofreasonitself.Thelatterdepictsreasonas,atroot,symbol-guidedstatetransitionsinquasi-linguisticspace.Theformerdepictsreasonas,atroot,thedevelopmentofprototype-styleknowledgeguidingvector-to-vectortransformationsinthesamekindsof(typically)non-sententialspacethatalsounderlieperceptuo-motorresponse.Beneaththiscontrast,however,liesasignificantagreement.Bothcampsagreethatrationalthoughtsandactionsinvolvetheuseofinnerresourcestorepresentsalientstatesofaffairs,andtheuseoftransformativeoperations(keyedtonon-semanticfeaturesofthoseinternalrepresentations)designedtoyieldfurtherrepresentations(inacascadeofvector-to-vectortransformationsintheconnectionistcase)and,ultimately,action.13.4Robotics:BeyondtheCore?Isitperhapspossibletoexplainreasonedactionwithoutappealtoinner,form-basedvehiclesofmeaningatall?Mightinternalrepresentationsbetoolswecanlivewithout?Considerthehumblehouse-fly.Marr(1982:32–3,reportedbyMcClamrock1995:85)notesthattheflygetsbywithoutinanysenseencodingtheknowledgethattheactionofflyingrequiresthecommandtoflapyourwings.Instead,thefly’sfeet,whennotincontactwithground,automaticallyactivatethewings.Thedecisiontojumpthusautomaticallyresults(viaabolitionoffootcontact)intheflappingofwings.Nowimaginesuchcircuitrymultiplied.Supposethe“decisiontojump”isitselfby-passedbye.g.directlywiringa“loomingshadow”detectortotheneuralcommandforjumping.Andimaginethattheloomingshadowdetectorisitselfnothingbutadumbroutinethatusestherawoutputsofvisualcellstocomputesomesimple,perceptualinvariant.Finally,imagineifyouwillawholesimplecreature,madeupofafairlylargenumberofsuchbasic,automaticroutines,butwiththeroutinesthemselvesorchestrated–byexactlythesamekindoftricks–sothattheyturneachotheronandoffat(generallyspeaking)ecologicallyappropri-atemoments.Forexample,a“consumefood”routinemaybeoverriddenbythe“somethinglooming-so-jump”routine,whichinturncausesthe“flapwings”routine,andsoon.Whatyouhaveimaginedis,coarselybutnotinaccurately,thekindof“subsumptionarchitecture”favoredbyrobotistssuchasRodneyBrooks(1991),andresponsibleforsuchprovocativetitlesofarticlesas“IntelligenceWithoutRepresentation”andslogans(nowco-optedasmovietitles!)suchas“Fast,Cheap,andOutofControl.”Itisnotatallobvious,however,thatsuchastorycould(eveninprinciple)besimplyscaled-upsoastogiveus“rationalitywithoutrepresentation.”Forone313\nAndyClarkthing,itisnotobviouswhenweshouldsayofsomecomplexinnerstatethatitconstitutesatleastsomekindofrepresentationofevents,orstatesofaffairs.Thehouse-flywing-flappingroutinelookslikeasimplereflex,yetevenherethereisroomforsomeonetosuggestthat,giventheevolutionaryhistoryofthereflexcircuit,certainstatesofthatcircuit(theonesactivatedbythebreakingoffoot-surfacecontact)representthefactthatthefeethaveleftthesurface.WhatBrooksandothersarereallysuggesting,itoftenseems,israthertheabsenceofacertaintypeofinternalrepresentation,viz.thebroadlylinguaformrepresentationsfavoredbyclassicalAI.Amorefundamentaldifficulty,however,(whichgoeswellbeyondthevague-nessoftheterm“internalrepresentation”)concernsthekindsofbehaviorthatcanplausiblybeexplainedbyanycomplexofreflex-likemechanisms.Theprob-lematiccaseshereareobviouslydeliberativereasonandabstractthought.Thekindsofbehaviorthatmightbeinvolvedincludeplanningnextyear’sfamilyvacation,thinkingaboutUSguncontrolissues(e.g.“shouldgunmanufacturersbeheldresponsibleforproducingmoregunsthantheknownlegalmarketre-quires?”),usingmentalimagestocountthenumberofwindowsinyourSpanishapartmentwhilerelaxingontheRiverThames,andsoon.Thesecasesarebynomeansallofapiece.Buttheyshareatleastonecommoncharacteristic:theyareall“representationhungry”(touseatermfromClarkandToribio1994)inquiteastrongsense.Allthesecases,onthefaceofit,requirethebraintouseinternalstand-insforexternalstatesofaffairs,wherea“stand-in,”inthisstrongsense(seeClarkandGrush1999)isanitemdesignednotjusttocarryinformationaboutsomestateofaffairs(inthewaythat,e.g.,theinnercircuitmightcarryinforma-tionaboutthebreakingoffoot-surfacecontactinthefly)buttoallowthesystemtokeyitsbehaviortofeaturesofspecificstatesofaffairsevenintheabsenceofdirectphysicalconnection.Asystemwhichmustcoordinateitsactivitywiththedistal(thewindowsinmySpanishapartment)andthenon-existent(themonsterinthetool-shed)isthusagoodcandidatefortheuseof(strong)internalrepres-entations:innerstatesthataremeanttoactasfull-bloodedstand-ins,notjustasambientinformation-carriers.(Forsomeexcellentdiscussionofthetopicsofconnectionanddisconnection,seeSmith1996.)Bycontrast,nearlyall(butseeStein1994andBeer2000)thecasestypicallyinvokedtoshowrepresentation-freeadaptiveresponsearecasesinwhichtherelevantbehavioriscontinuouslydrivenby,andmodifiedby,ambientinputfromthestateofaffairstowhichthebehavioriskeyed.Rationalbehavioris,insomesense,behaviorthatisguidedby,orsensitiveto,reasons.Intuitively,thisseemstoinvolvesomecapacitytostepbackandassesstheoptions;toforeseetheconsequences,andtoactaccordingly.Butthisvisionofrationality(“deliberativerationality”)placesrationalactionsquarelyinthe“representation-hungry”box.Forfutureconsequences,clearly,cannotdirectlyguidecurrentaction(inthewaythat,say,anambientlightsourcemaydirectlyguideaphoto-sensitiverobot).Suchconsequenceswillbeeffectiveonlytotheextentthatthesystemusessomethingelsetostand-inforthoseconsequences314\nArtificialIntelligenceduringtheprocessofreasoning.Andthat,atleastonthefaceofit,requirestheuseofinternalrepresentationsinsomefairlyrobustsense.13.5EmotionsandReasonAmechanicalexplanationofourcapacitiestodisplayreason-guidedbehaviorcannot,itseems,affordtodispensewiththemostbasicnotionofinnerstand-inscapableofdirectingbehaviorandinferenceintheabsenceoftheeventsandstatesofaffairsconcerned.Workinconnectionismandreal-worldroboticsisbestviewed(Ibelieve)asexpandingourconceptionsofthepossiblenatureofsuchstand-ins,andashighlightingthemanywaysinwhichbodilyandenvironmentalstructures,motion,andactiveinterventionmayallservetotransformtheproblemsthatthebrainneedstosolve.Theuseofpenandpaper,forexample,maygreatlyaltertheproblemsthatthebrainneedstosolvewhenconfrontingcomplexarithmeticaltasks,whenplanningalong-termstrategy,andevenwhenreasoningaboutguncontrol.Butsuchtransformationsdonotby-passtheneedforinternalstructure-sensitiveoperationsdefinedoverinnercontent-bearingvehicles:rather,theyreshapetheproblemsthatsuchaninnereconomyneedstosolve.Thestressonreason-sensitivethoughtandinferencecan,however,blindustothecrucialimportanceofafurtherdimensionofhumancognition.Forhumanreasonistightly,perhapsinextricably,interwovenwithhumanemotion.DoingjusticetothissignificantinteractionisoneofthetwomajorchallengesforthenextgenerationofAImodels.Emotionswerelongregarded(atleastinabroadlyKantiantradition)astheenemyofreason.Andwecertainlydospeakof(forexample)judgmentsbeingcloudedbyenvy,actsasbeingdrivenbyshort-livedburstsoffuryandpassionratherthanbyreasonedreflection,andsoon.Itisbecomingincreasinglyclear,however,thatthenormalcontributionsofemotiontorationalresponsearefarfromdetrimental.Theyare,infact,bestseenaspartofthemechanismofreasonitself.Consider,totakeafamousexample,thecaseofPhineasGage.Gagewasanineteenth-centuryrailwayworkerwhosebrainwasdamagedwhenanironrodwasdriventhroughhisskullinanexplosion.Despiteextensivedamagetoprefrontalcortex,theinjuryleftGage’slanguage,motorskills,andbasicreasoningabilitiesintact.Itseemedasifhehadescapedallcognitivecompromise.Oversubsequentyears,however,thisprovedsadlyincorrect.Gage’spersonalandprofessionallifetooknoticeableturnsfortheworse.Helostjobs,gotintofights,failedtoplanforthefutureandtoabidebynormalconventionsofsocialconduct,becameadifferentandmarkedlylesssuccessfulperson.Theexplanation,accordingtoDamasioetal.(1994)wasthatthedamagetoprefrontalcortexhadinterferedwithasystemof(whattheytermed)“somaticmarkers”–brainstatesthattietheimage/traceofaneventtoakindofgutreaction(aversionorattraction,accord-ingtotheoutcome).Thismarkersystemoperatesautomatically(innormal315\nAndyClarksubjects)influencingbothon-the-spotresponseandthearrayofoptionsthatweinitiallygenerateforfurtherconsiderationandreflection.Itisactivealso–andcrucially–whenweimagineaneventorpossibleaction,yieldingapositiveornegativeaffectivesignalthatmanifestsitselfin(amongotherthings)galvanicskinresponse.Gage,itishypothesized,wouldhavelackedsuchresponses,andwouldnothavehadhisreasoninganddeliberationsconstrainedbytheautomaticoption-pruningandchoice-influencingoperationsofthesomaticmarkersystemgraduallyacquiredduringhislifetime’sexperienceofsocialandprofessionalaction.Contemporarystudiesseemtoconfirmandclarifythisbroadpicture.E.V.R.(apatientdisplayingsimilarventromedialfrontaldamage)sharesGage’spro-file.ThoughscoringwellonstandardIQandreasoningtests,E.V.R.likewiselostcontrolofhisprofessionalandsociallife.Inaninterestingseriesofexperi-ments(Becheraetal.1997)normalcontrolsandprefrontallylesionedpatientsplayedacardgameinvolving(unbeknownsttothesubjects)twowinningdecksandtwolosingdecks.Subjectscouldchoosewhichdeck(A,B,C,orD)toselectcardsfrom.Afteralittleplay,thenormalcontrolsfixonthebetterdecks(smallerimmediaterewards,butfewersecurepenaltiesandmorereliableinthelongterm)andrapidlyshowaheightenedgalvanicskinresponsewhenreachingforthe“bad”decks.Thisskinresponse,interestingly,appearsbeforethesubjectscouldarticulateanyreasonsforpreferringthebetterdecks.E.V.R.,bycontrast,showsnosuchskinresponse.Andthisabsenceofsomaticcuesseemstointerferewithhiscapacitytochoosethebetterdecksevenoncehisconsciousmindhasfigureditallout–hewillknowthatAandBarelosingdecks,yetcontinuetofavorthemduringplay.Thereisobviouslymuchtodiscusshere.Arethesecasesbestunderstood,asChurchlandsuggests,asarisingfrom“theinabilityofemotionstoaffect[thepatient’s]reasonanddecision-making”(1998:241)?Orisitacaseofinappropri-ateemotionalinvolvement–thetriumphofshort-termrewardoverdeferred(butgreater)gratification?Perhapsthesearenotreallyincompatible:eitherwayitisthelackoftheon-the-spotunconsciousnegativeresponses(evidencedbytheflatgalvanicskinresponses)thatopensthedoortocognitiveerror.Humanreason,itseemsfairtoconclude,isnotbestconceivedastheoperationofanemotionlesslogicengineoccasionallylockedintocombatwithemotionaloutbursts.Instead,trulyrationalbehavior(inhumans)istheresultofacomplexanditeratedseriesofinteractionsinwhichdeliberativereasonandsubtle(oftenquiteunconscious)affect-ladenresponsesconspiretoguideactionandchoice.Emotionalelements(atleastassuggestedbythesomaticmarkerhypothesis)function,infact,tohelprationalchoiceoperateacrosstemporaldisconnections.Somaticmarkersthusplayaroledeeplyanalogoustointernalrepresentations(broadlyconstrued);theyallowustoreasonprojectively,onthebasisofpastexperience.Whatcouldbemoreappropriatelydeemedpartofthemechanismofreasonitselfthansomethingthatallowsustoimaginativelyprobethefuture,usingthehard-wonknowledgeofalifetime’schoicesandexperiencesallneatlydistilledintoanetworkofautomaticaffectivereverberations?316\nArtificialIntelligence13.6GlobalReasoningAfurthersourceofcomplicationconcernswhatFodor(1983:111)calls“globalpropertiesofbeliefsystems.”AI,accordingtoFodor,confrontsaspecialproblemhereabouts.FortheTuringMachinemodelofrationalinference(recallsection13.2above)issaidtobeirredeemablylocal.Itisgreatatexplaininghowthethought(syntacticallytokened)thatitisraininggiveswaytothethoughtthatanumbrellaisindicated.Itisgreat,too,atexplaining(givenafewclassicalassump-tions–seeFodorandPylyshyn1988)whythespaceofpossiblethoughts(foranindividual)exhibitsacertainkindofclosureunderrecombination–thepropertyof“systematicity,”whereinthosewhocanthinkaRbtypicallyalsothinkbRa,andsoon.ButwherecurrentAI-basedmodelscrashandburn,Fodorinsists,iswhenconfrontingvariousformsofmoregloballysensitiveinference.Forexample,casesofabductiveinferenceinwhichthebestexplanationforsomeeventmightbehiddenanywhereintheentireknowledgebaseofthesystem:aknowledgebasedeemedtoolargebyfartosuccumbtoanyprocessofexhaustivesearch.Fodorrejectsclassicalattemptstogetaroundthisproblembytheuseofheuristicsandsimplifyingassumptions(suchastheuseof“frames”–seeMinsky1975;Fodor1983:116)arguingthatthissimplyrelocatestheproblemasaproblemof“executivecontrol”–viz.howtofindtherightframes(orwhatever)attherighttime.Sinceeventhedecisiontotaketheumbrellaagainsttherainispotentiallysensitivetocountervailinginformationcomingfromanywhereintheknowledgebase,Fodorisactuallyleftwithamodelofmechanicalrationalitywhich(asfarasIcansee)canhavenothingtosayaboutanygenuinebutnon-deductivecaseofreasoningwhatsoever.TheFodor–Turingmodelofrationalmechanismworksbest,asFodorfrequentlyseemstoadmit,onlyinthedomainof“informationallyencapsulatedsystems”–typically,perceptualsystemsthatprocessarestrictedrangeofinputsignalsinawayallegedlyinsensitivetoallformsoftop-downknowledge-driveninference.Hardlytheseatofreason,onecannothelpbutfeel.Givethispessimisticscenario–enshrinedinFodor’s“firstlawofthenon-existenceofcognitivescience:themoreglobal...acognitiveprocessis,thelessanybodyunderstandsit.Veryglobalprocesses...aren’tunderstoodatall”(1983:107)–itisnotsurprisingtofindsometheorists(Churchland1989:178;Clark1993:111)arguingforconnectionistapproachesasonesolutiontothisproblemof“globallysensitivereason.”SuchapproachesareindependentlyrejectedbyFodorforfailingtoaccountforsystematicityandlocalsyntax-sensitiveinference.Butitnowseemstome(thoughthisisalongstory–seeClark,2002)thattheproblemofglobalabductiveinferencereallydoesaffectconnectionistapproachestoo.Veryroughly,itemergesthereinasaproblemofroutingandsearching:aquestionofhowtouseinformation,whichcouldbedrawnfromanywhereintheknowledgebase,tosculptandredirecttheflowofprocessingitself,ensuringthattherightinputprobesareprocessedbytherightneuralsub-populationsattherighttimes.317\nAndyClarkChurchland(1989)andClark(1993)depictthisproblemassolved(intheconnectionistsetting)because“relevantaspectsofthecreature’stotalinformationareautomaticallyaccessedbythecodedstimulithemselves”(Churchland1989:187).Andcertainly,inputprobeswill(recallsection13.3above)automaticallyactivatetheprototypesthatbestfittheprobe,alongwhateverstimulusdimensionsarerepresented.Butthisisatbestafirststepintheprocessofrationalresponsive-ness.Forhavingfoundthesebestsyntacticfits(forthisisstill,ultimately,aform-drivenprocess),itisnecessarytoseeifcruciallyimportantinformationisstoredelsewhere,unaccessedbecauseofalackofsurfacematchingtheprobe.Anditisthisstepwhich,Ithink,doesmostoftheworkinthetypesofcasewithwhichFodoris(properly)concerned.Thegoodnews,whichImakemuchofinClark(2002)butcannotpursuehere,isthatthissecondstepnowlookspotentiallycomputationallytractable,thankstoanoddcombinationofneuro-connectionistresearchandaninnova-tive“second-order”searchproceduredevelopedforuseontheworldwideweb(Kleinberg1997).Theideaistocombineafirstpass(dumb,pattern-matching,syntax-based)searchwithafollow-upsearchbasedonthepatternsofconnectionsintoandawayfromtheelementsidentifiedonthefirstpass.Butthepoint,forpresentpurposes,issimplytoacknowledgethespecialproblemsthattrulyglob-allysensitiveprocessingcurrentlypresentstoallexistingmodelsoftheneuralcomputationsunderlyinghumanreason.13.7FastandFrugalHeuristicsItmightreasonablybeobjected,however,thatthiswholevisionofhumanrationalityiswildlyinflated.Veryoften,wedon’tmanagetoaccesstherelevantitemsofknowledge;veryoften,wedon’tchoosethatwhichmakesushappiest,ormostsuccessful;weeven(goon,admitit)makeerrorsinsimplelogic.Whatisnonethelesssurprisingisthatweveryoftendoaswellaswedo.Theexplanation,accordingtorecenttheoriesof“ecologicalrationality,”isour(brain’s)useofsimple,short-cutstrategiesdesignedtoyieldgoodresultsgiventhespecificcon-straintsandopportunitiesthatcharacterizethetypicalcontextsofhumanlearningandhumanevolution.Aquickexampleistheso-called“recognitionheuristic.”Ifyouaskmewhichcityhasthelargerpopulation,SanDiegoorSanAntonio,ImaywellassumeSanDiego,simplybecauseIhaveheardofSanDiego.ShouldIrecognizebothnames,Imightdeployadifferentfastandfrugalheuristic,check-ingforothercues.MaybeIthinkagoodcueis“haveIheardoftheirsym-phony?”andsoon.ThepointisthatIdon’ttryanyharderthanthat.Theremaybemultiplesmallcuesandindicators,whichIcouldtryto“factorin.”Butdoingso,accordingtoanimpressivebodyofresearch(seee.g.Chaseetal.1998)islikelytobebothtime-consumingand(here’sthecruncher)unproductive.I’ll318\nArtificialIntelligenceprobablychooseworsebytryingtoreplacethefastandfrugalheuristicwithsomethingslowerand(apparently)wiser.Itisnotyetclearhow(exactly)thisimportantbodyofresearchshouldimpactourvisionofjustwhatyouneedtoexplaininordertoexplainhowrationalityismechanicallypossible.Alikelyalliancemightseefansofroboticsandartificiallife-basedapproaches(section13.4)usingrelativelysimpleneuralnetworkcontrollers(section13.3)tolearnfastandfrugalheuristicsthatmaximallyexploitlocalopportunitiesandstructures.Thesomaticmarkermechanism(section13.5)mightbeconceivedas,inasense,implementingjustanotherkindoffastandfrugalheuristicenablingcurrentdecision-makingtoprofitcheaplyfrompastexperience.Undersuchanonslaught,itispossiblethatmuchoftheworryaboutglobalabductiveinference(section13.6)simplydissolves.Myownview,asstatedabove,isthatsomethingofthepuzzleremains.ButthesolutionIfavor(seeClark,2002)canitselfbeseenasaspecialinstanceofafastandfrugalheuristic:acheapprocedurethatreplacesglobalcontent-basedsearchwithsomethingelse(thesecondpass,connectivitypattern-basedsearch,mentionedearlier).13.8Conclusions:MovingTargetsandMultipleTechnologiesRationality,wehavenowseen,involvesawholelotmore,andawholelotless,thanoriginallymettheeye.Itinvolvesawholelotmorethanlocal,syntax-basedinferencedefinedovertractablesetsofquasi-sententialencodings.EvenFodoradmitsthis–oratleast,headmitsthatitisnotyetobvioushowtoexplainglobalabductiveinferenceusingsuchresources.Italsoinvolvesawholelotmorethan(asitwere)thedispassionatedeploymentofinformationintheserviceofgoals.Forhumanreasonseemstodependonadelicateinterplayinwhichemotionalresponses(oftenunconsciousones)helpsiftouroptionsandbiasourchoicesinwaysthatenhanceourcapacitiesoffluent,reasoned,rationalresponse.Theseemotionalsystems,Ihaveargued,areusefullyseenasakindofwonderfullydistilledstoreofhard-wonknowledgeconcerningalifetime’sexperiencesofchoos-ingandacting.Butrationalitymayalsoinvolvesignificantlylessthanwetendtothink.Perhapshumanrationality(andIamtakingthatasourconstanttarget)isessentiallyaquick-and-dirtycompromiseforgedintheheatofourecologicalsurround.Fastandfrugalheuristics,gearedtomakingthemostofthecheapestcuesthatallowustogetby,maybeascloseasnatureusuallygetstothespaceofreasons.Workinroboticsandconnectionismfurthercontributestothisvisionoflessasmore,asfeaturesofbodyandworldareexploitedtopressmaximalbenefitfrombasiccapacitiesofon-board,prototype-basedreasoning.Eventhebugbearofglobalabductivereason,itwashinted,justmightsuccumbtosomewilycombinationoffastandfrugalheuristicsandsimplesyntacticsearch.319\nAndyClarkWherethendoesthisleavethereputedlyfundamentalquestion“howisration-alitymechanicallypossible?”Itleavesit,Ithink,atanimportantcrossroads,uncertainlypoisedbetweentheoldandthenew.If(asIbelieve)theresearchprogramsdescribedinsections13.4–13.8areeachtacklingimportantaspectsoftheproblem,thentheproblemofrationalitybecomes,precisely,theproblemofexplainingtheproduction,insocial,environmental,andemotionalcontext,ofbroadlyappropriateadaptiveresponse.Rationality(orasmuchofitaswehumanstypicallyenjoy)iswhatyougetwhenthiswholemedleyoffactorsaretunedandinteranimatedinacertainway.Figuringoutthiscomplexecologicalbalancingactjustisfiguringouthowrationalityismechanicallypossible.ReferencesBechera,A.,Damasio,H.,Tranel,D.,andDamasio,A.R.(1997).“DecidingAdvant-ageouslyBeforeKnowingtheAdvantageousStrategy.”Science,275:1293–5.Beer,R.D.(2000).“DynamicalApproachestoCognitiveScience.”TrendsinCognitiveSciences,4(3):91–9.Brooks,R.(1991).“IntelligenceWithoutRepresentation.”ArtificialIntelligence,47:139–59.Chase,V.,Hertwig,R.,andGigerenzer,G.(1998).“VisionsofRationality.”TrendsinCognitiveSciences,2(6):206–14.Churchland,P.M.(1989).TheNeurocomputationalPerspective.Cambridge:MIT/Brad-fordBooks.Churchland,P.S.(1998).“FeelingReasons.”InP.M.ChurchlandandP.S.Churchland(eds.),OnTheContrary.Cambridge,MA:MITPress:231–54.Clark,A.(1989).Microcognition:Philosophy,CognitiveScienceandParallelDistributedProcessing.Cambridge,MA:MITPress.——(1993).AssociativeEngines:Connectionism,ConceptsandRepresentationalChange.Cambridge,MA:MITPress.——(1996).“Connectionism,MoralCognitionandCollaborativeProblemSolving.”InL.May,M.Friedman,andA.Clark(eds.),MindsandMorals.Cambridge,MA:MITPress:109–28.——(2002).“LocalAssociationsandGlobalReason:Fodor’sFrameProblemandSecond-OrderSearch.”CognitiveScienceQuarterly.Clark,A.andGrush,R.(1999).“TowardsaCognitiveRobotics.”AdaptiveBehavior,7(1):5–16.Clark,A.andToribio,J.(1994).“DoingWithoutRepresenting?”Synthese,101:401–31.Clark,A.andThornton,C.(1997).“TradingSpaces:ConnectionismandtheLimitsofUninformedLearning.”BehavioralandBrainSciences,20(1):57–67.Damasio,H.,Grabowski,T.,Frank,R.,Galaburda,A.M.,andDamasio,A.R.(1994).“TheReturnofPhineasGage:CluesabouttheBrainfromtheSkullofaFamousPatient.”Science,264:1102–5.Fodor,J.(1983).TheModularityofMind.Cambridge,MA:MITPress.——(1998).InCriticalCondition:PolemicalEssaysonCognitiveScienceandthePhilosophyofMind.Cambridge,MA:MITPress.320\nArtificialIntelligenceFodor,J.andLepore,E.(1993).“ReplytoChurchland.”PhilosophyandPhenomenologicalResearch,53:679–82.Fodor,J.andPylyshyn,Z.(1988).“ConnectionismandCognitiveArchitecture:ACriticalAnalysis.”Cognition,28:3–71.Haugeland,J.(1981).“SemanticEngines:AnIntroductiontoMindDesign.”InJ.Haugeland(ed.),MindDesign:Philosophy,Psychology,ArtificialIntelligence.Cambridge,MA:MITPress:1–34.Kleinberg,J.(1997).“AuthoritativeSourcesinaHyperlinkedEnvironment.”IBMResearchReport(RJ10076).AversionalsoappearsinH.Karloff(ed.),Proceedingsofthe9thACM-SIAMSymposiumonDiscreetAlgorithms(1998),andanextendedversioninJournaloftheACM46(1999).Marr,D.(1982).Vision.SanFrancisco,CA:W.H.Freeman.McClamrock,R.(1995).ExistentialCognition.Chicago,IL:ChicagoUniversityPress.Minsky,M.(1975).“AFrameworkForRepresentingKnowledge.”InP.Winston(ed.),ThePsychologyofComputerVision.NewYork:McGraw-Hill.Newell,A.andSimon,H.(1976).“ComputerScienceasEmpiricalInquiry:SymbolsandSearch.”CommunicationsoftheAssociationforComputingMachinery,19:113–26.Schank,R.(1975).“UsingKnowledgetoUnderstand.”TINLAP:75.Smith,B.C.(1996).OntheOriginofObjects.Cambridge,MA:MITPress.Stein,L.A.(1994).“ImaginationandSituatedCognition.”JournalofExperimentalandTheoreticalArtificialIntelligence,6:393–407.321\nTheBlackwellGuidetoPhilosophyofMindJohnBickleEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter14PhilosophyofMindandtheNeurosciencesJohnBickleNearlytwodecadeshavepassedsincePatriciaChurchlandexclaimed,withhercharacteristicverve,that“nothingismoreobviousthanthatphilosophersofmindcouldprofitfromknowingatleastsomethingofwhatthereistoknowabouthowthebrainworks”(1986:4).Neurosciencehassincedevelopedexponentially.Wearenowontheothersideof“theDecadeoftheBrain.”Weknowmuchabouttheneuralmachinerythatgeneratescognition,perception,andaction.Ourknowledgespanseverylevel,fromthebiophysicsofmembranechan-nelstothelarge-scaledynamicsofmassivelyparallelneuronalnetworks.Onemighthavethoughtthat“philosophyofneuroscience”wouldnowdominatephilosophyofmind.Onemighthavethoughtthatphilosopherswouldfeelashamedtoargueabout,e.g.,consciousness,cognitiverepresentation,theepistemologyofperception,andevensomenormativeissues,whenignorantofrelevantandavail-ableinformationfromneuralscience.Onewouldbewrong.Forthemostpart,mainstreamphilosophyofmindremainsindifferent.(Howmuchneurosciencedoyoufindinthiscollection?)Whywouldotherwiserational,intelligentthinkersignorethe“obvious”?Partoftheanswerisn’tcomplicated.Historically,andespeciallyinitspresentform,neuroscienceisareductiveenterprise.And“reductionism”isn’tpopularincon-temporaryphilosophy.Inthesamebook,Churchlandassertedthat“oftenasnotopposingsidesinadebateonreductionismgorightbyeachotherbecausetheyhavenotagreeduponwhattheydisagreeabout”(1986:278).Thisassessmentstillholds.Reductionremainsdeeplymisunderstoodbyphilosophers,includingitsmethodologicalimplicationsforthe“special,”potentiallyreducedsciences.Oneprincipalgoalofthischapteristoclarifythesenseandmethodologicalimportofthekindof“reductionism”thatinspirescontemporaryneuroscience.Otherfactorsmake“reductionist”enterprisesunattractivetocontemporaryphilosophers.Jobsecurity,forinstance.OnlyphilosopherswithVillageAtheisttemperamentstakepleasureinseeing“mind”usurpedbyscience.Thisconcepthasbeensocentraltophilosophyforsolong.Andif“mind”getswrestedaway322\nPhilosophyofMindandtheNeurosciencesbyareductivescience,joiningtheranksof“divinepurpose,”“naturalworld,”and“livingbeing,”whatwillbeleftforphilosopherstoruminateabout?Obviously,theseremarksdon’taddressargumentsthatmotivatethedimviewaboutreductionism,whichhavegrownincreasinglysophisticatedoflate.Nordotheyprovideanti-reductionistswithanyempiricalreasonsforpause.Thesearemytasksinwhatfollows.OverthenextfoursectionsIwilldefendthefollowingclaims.•the“putuporshutupchallenge”topsychoneuralreductionismhasalreadybeenmet,andresidualworriesaboutexamplesfromrecentsciencerevealwidespreadmisconceptionsaboutreductionthatstillpervadephilosophy(andcognitivepsychology);•recentworkatthelevelofsingle-cellneurophysiologyisyieldingresultsdirectlyrelevanttophilosophicalconcerns,evenaboutconsciousness;•philosophersarenottheonlytheoristsseekingtoaddressthe“qualitative”and“subjective”aspectsofconsciousness;increasingly,hard-coreneuroscientistsareraisingquestionsaboutthesefeaturesandaddressingtheminingeniousyetstraightforwardlyempiricalways.Qualiaandsubjectivity:they’renotjustforphilosophersanymore.Iwillcloseonasomewhattangentialissuebyarguingthatthemuch-ballyhooed“interdisciplinarity”betweenphilosophers,psychologists,andneuroscientistsremainsmostlyamythinpractice.Everybodyremainsconvincedthatevery-bodyelseisignorantoftheimportantcontributionsfromone’sownarea.Andconsensusisrightaboutthis,thoughwithpropertrainingphilosopherscouldmakeauniquecontributiontowardchangingthis.14.1RealReductioninRealNeuroscienceAssessingexistingtheoriesofscientificreductionanddevelopinganalternativeis1ahugetaskinthephilosophyofscience,farbeyondthescopeofthischapter.ButtwofeaturesrequireexplicitmentiontofendofftheverbaldisputesthatChurchlandwarnedabout.First,scientificreductionisinter-theoreticreduction.Itisarelationbetweenscientifictheories,notentities,properties,orevents.Scien-tificreductionsmightyieldcross-theoreticontologicalconsequences,buttheseconsequencesaresecondarytoanddependentupontheprimaryinter-theoreticrelation.Secondly,theconceptofinter-theoreticunificationliesattheheartofscientificreduction.Whenreductionsobtain,thereducingtheoryfullyexplainsthereducedtheory’sdata,whichareusuallystillexpressedinthelatter’ster-minologyandframework.(Thatthisconditionholdsinprincipleandnotalwaysinpracticeshouldgowithoutsaying,butoftencan’t.)Thatcontemporaryneuroscienceaspirestoreducepsychologyisnicelyex-pressedinapairofquotesfromprominenttextbooks.GordonShepherdwrites:323\nJohnBickleManycognitivepsychologists...believethattheoriesaboutlearningandmemoryshouldbeself-consistentandself-sufficient,withoutrecoursetoneuralmechanisms....Formostneurobiologists,thisviewisoutdated,andoneofthegoalsofmodernresearchistojointhetwolevelsintoacoherentframework.(1994:619;myemphases).Theemphasizedphrasesreflectthetwofeaturesofscientificreductionstressedabove,itsprimarilyinter-theoreticcharacterandunificationistgoal.Kandeletal.carrythereductionistbannerdownonemorelevel:Thegoalofneuralscienceistounderstandthemind,howweperceive,move,think,andremember.Inthepreviouseditionsofthisbookwestressedthatimportantaspectsofbehaviorcouldbeexaminedatthelevelofindividualnervecells....[T]heapproach...wasforthemostpartframedincell-biologicalterms.Nowitisalsopossibletoaddressthesequestionsdirectlyonthemolecularlevel.(1991:xii;myemphasis)Theyurgedthisreorientationintheearly1990s,whenthe“molecularrevolu-tion”wasjustbeginningtosweepthroughneuroscience.FiveminutesperusalofSocietyforNeuroscienceAbstractsfromtheearly1990supthroughthepresentrevealshowmoreprevalentmoleculartheoriesandexperimentalmethodolo-gieshavebecome.Reductionismisaliveandthrivingincurrentmainstreamneuroscience.However,researchgoalsareonething,whileaccomplishedresultsareanother.Iscurrentneurobiologyactuallydevelopingtheoriestowhichgenuinelycognitivepsychologicaltheoriesreduce?I’vetermedthisquestion“theputuporshutupchallenge”forpsychoneuralreduction,andhavearguedforanaffirmativeanswer(Bickle1995;1998:ch.5).Myargumentinvolvestwoplanks:1Currentpsychologicaltheoriesofassociativelearningappealtoresources(rep-resentationsandcomputationsovertheircontents)thatmeetthestandard,widelyaccepted“markofthegenuinelycognitive.”2Thesepsychologicaltheoriesreducetoneurobiologicaltheoriesabouttheneuronalcircuitriesintheappropriatebrainregionsandthecellularandmolecularmechanismsofsomeformsofsynapticplasticity(themechanismsbywhichtheefficiencyofelectrochemicaltransmissionbetweenneuronsincreasesordecreasesovertime).Theneurobiologicalreductionofgenuinelycognitivepsychologicaltheoriesisalreadyanaccomplishedfact.Thecaseforthefirstplankisinterestingandwidelyunknownamongbothphilosophersandcognitivepsychologists(Rescorla1988);butI’vetoldittwiceinprint(citedinthepreviousparagraph)andwon’trepeatthedetailshere.Sufficeittosaythatowingtoadvancesinexperimentaltechnology,ingeniousexperi-mentaldesign,andaquantitativemodelyieldingcounterintuitivepredictionsthatwereverifiedempirically,associativelearningtheory,sincethe1970s,324\nPhilosophyofMindandtheNeurosciencesemphasizestheinformationthatonestimulusgivesaboutanother....Thesethe-oriesemphasizetheimportanceofadiscrepancybetweentheactualstateoftheworldandtheorganism’srepresentationofthatstate.Theyseelearningasaprocessbywhichthesetwoarebroughtintoline....AusefulshorthandisthatorganismsadjusttheirPavlovianassociationsonlywhentheyare“surprised.”(Rescorla1988:152–3)Thisapproachiscompletelygeneral.Learningtheoristsappliedittoexoticasso-ciativephenomenasuchastheblockingeffectandbehaviorallysilentlearning,butalsotoclassicalconditioning.Inthepaperwheretheyfirstarticulatedonesuchtheoryinprecise,quantifiedfashion,RescorlaandWagnerstateitin“explic-itlycognitivistterms”:Organismsonlylearnwheneventsviolatetheirexpectations.Certainexpectationsarebuiltupabouttheeventsfollowingastimuluscomplex:expectationsinitiatedbythatcomplexanditscomponentstimuliarethenonlymodifiedwhenconsequenteventsdisagreewiththecompositeexpectation.(1972:75)Furtherdevelopmentandempiricaltestingoftheirmodelquicklyfollowed,andbythelate1970sitdominatedthefield(Dickinson1980).Whataboutthecaseformysecondplank?GoingbacktoRamonyCajal,andfirstdevelopedexplicitlybyHebb(1949),neuroscientistshavemaintainedthatlearningandmemoryinvolvechangesincentralnervoussystem(CNS)circuits.Sincethemid-twentiethcentury,thesitethathasattractedthemostattentionisthesynapse,thetinycleftbetweenneuronswherethetransmissionofelec-trochemicalactivitytakesplace.IntheCNSthistransmissionprimarilyisbywayofchemicalneurotransmittersreleasedbythepresynapticneuronintothesynapticcleft,whichthenbindwithmembrane-boundproteins(receptors)onthepostsynapticneuron.Thisbindinginitiatesachainofbiochemicaleventsthatopenion-selectivemembranechannels,resultingineitherdepolarization(excitatorypostsynapticpotentials,orEPSPs)orhyperpolarization(inhibitorypostsynapticpotentials,orIPSPs)atthatpatchofpostsynapticmembrane.Alargenumberofpresynaptic,postsynaptic,andintra-cleftbiochemicalfactorsaffecttheefficacyofsynaptictransmission.Thesefactorsareplastic:changeableatthebehestofahuge2varietyofendogeneousandexternalbiochemicalevents.Abundantandwidelyvariedexperimentalevidencesupportssynapticplasticity3asaprincipalmechanismoflearningandmemory.Drawingonavarietyofexperimentalmethodologies,animalpreparations(bothvertebrateandinverte-brate),anatomicalregions,andbehavioraltasks,ageneralmodelofthesynapticbasisoflearningandlong-termmemoryhasemerged(Shepherd1994:648–9).Thebasiccell-biologicalconceptislong-termpotentiation(LTP)(seefigure14.1).Anactionpotential,spreadingdownthepresynapticaxonmembranetoitster-2+2+minalbulb,opensvoltage-gatedcalciumion(Ca)channels.Caflowsintothepresynapticterminal(alongitsconcentrationandelectricgradient).Thisinfluxproducesabiochemicalcascadethatresultsintheincreasedbindingofvesicles325\nJohnBickleNucleusofpresynapticneuronDNAstructuralchanges?transcriptionfactorslearning?LTP?intracellularpathwaysAANOintracellularpathways2+CaGLUGLUSynapticcleft++KKNMDAAMPAreceptorsreceptors++NaNa2+CaAANOintracellularpathwaysphosphoproteinsstructuralchanges?transcriptionfactorslearning?LTP?immediateNucleusofpostsynapticneuronearlygenesDNAFigure14.1SimplifiedillustrationofthecurrenttheoryofLTP-inducedsynapticplasticity.Seetextforexplanationandabbreviations.(AdaptedfromShepherd1994:648,figure29.18.)326\nPhilosophyofMindandtheNeurosciencescontainingtheneurotransmitterglutamate(GLU)toactivezonesonthepre-4synapticmembrane,andsubsequentglutamatereleaseintothesynapticcleft.Theglutamatebindstotwotypesofpostsynapticreceptor.OnetypeisionotrophicAMPA(α-amino-3-hydroxy-5-methyl-4-isoxazoleproprionicacid)receptors,which++opendirectsodiumion(Na)channels,resultingintheinfluxofNa(alongbothitsconcentrationandelectricgradients)andsubsequentEPSPs.TheothertypeisNMDA(N-methyl-D-aspartate)receptors.Atornearrestingmembranepoten-tial,glutamatebindingyieldsnoionicinfluxbecauseNMDAreceptorsareblockedbymagnesium.However,whenthemembraneissufficientlydepolarized(i.e.,byglutamatebindingatnearbyAMPAreceptors),themagnesiumblockpopsoff2+andglutamatebindingtoNMDAreceptorsopenspostsynapticCachannels.2+Postsynaptically,Caactsasasecondmessenger.Itactivates:2+•acascadeofCabindingproteinsandproteinkinasesthatbreakdownandthenreconstructthecytoskeletonofthepostsynapticneuronintoadifferentconfiguration,yieldingchangesinreceptornumbersandlocations;•phosphoproteinsand(probably)othertranscriptionfactorsthatinturnactiv-ateimmediateearlygenesinthenucleusofthepostsynapticneuron,produc-inglong-termchangesinreceptorandotherproteinsynthesis;•nitricoxidesynthesis,whichservesasaretrogradetransmitterbackonthepresynapticneurontoenhancesubsequentglutamaterelease.Inaddition,postsynapticactivationofNMDAreceptorsgeneratesproductionofarachidonicacid(AA),whichalsoappearstoactasaretrogradetransmitter.Presynaptically,AAinitiatesacascadeofproteinkinaseswhichinteractultimatelywithgenetictranscriptionfactors,yieldinglong-termchangesinproteinproduc-5tion,cellstructure,andfunction.HowdoesthetheoryofLTP-inducedsynapticplasticityfigureintoreductionsofcognitivistlearningtheories(suchasmodernassociativelearningtheory)?ThekeyiswhatHawkinsandKandel(1984)calledthe“cell-biologicalalphabetoflearning”andIcalled“combinatorialreduction”(Bickle1995;1998;ch.5).Thecell-biologicalandmolecularmechanismsprovidethe“letters,”andtheirsequencesandcombinations(the“words”)madeavailablebyincreasinglycomplexneuralanatomiesandphysiologiesexplainallthebehavioraldataaddressedbythecognitivepsychologicaltheory.Forexample,behavioralchangesresultingfromclassicalconditioningareexplainedbystimulus-pairedincreasesinpresynapticneurotransmitterrelease.(ThisisoneofHawkinsandKandel’s“letters.”)Initially,theneutralconditionedstimulus(CS)elicitsweakneurotransmitterrelease(abovebaselinerates)incentralpathwaysleadingfromthestimulatedsensoryreceptors.Thebehaviorallysignificantunconditionedstimulus(US)elicitsstrongrelease.ActivityintheUSpathwayactivatesinterneuronsthatsynapseonthepresynapticterminalbulbsintheCSpathway.Theseinterneuronsreleasetheneurotransmitterserotonin,whichbindstoreceptorsonthepresynapticCSpathwayterminals.Thisinitiatesabiochemicalcascadeintheseterminalsthatinhibitspotassiumion327\nJohnBickle+(K)efflux,broadeningtheactionpotentialsinitiatedbytheCSandeliciting2+2+increasedCainflux.TheadditionalCafacilitatesincreasedbindingofvesiclestoterminalmembrane,neurotransmitterrelease,andpostsynapticresponse.Inthisway,theCS–USpairingsenabletheweakerCStoaccessthesamesecond6messengersystemelicitedbythesalientUS.CS-inducedactivitythenreplacestheUSinactivatingtheunconditionedresponsemotorpathways.Theultimateresultisactivityinmotorneuronsthatproducetheappropriatemusclecontractionsagainsttheskeletalframethatgeneratethebehavioraldynamicsovertime.Appealingonlytopresynapticmechanisms,HawkinsandKandel(1984)ex-plainsomehigher-ordercognitivefeaturesofassociativelearningbysequencesandcombinationsofthecell-biological“letters.”Alloftheircircuitryassumptionswerebasedonknownanatomyandphysiology.Theydemonstratehowthebeha-vioraldynamicsoftheblockingeffect,stimulusspecificityandgeneralization,extinctionandspontaneousrecovery,second-order(S-S)conditioning,andUSpre-exposurecanbeexplaineddirectlybybiologicallyplausiblesequencesandcombinationsofthecell-biological“letters.”Theultimateoutcomeofthesesequencesandcombinationsovertimeischangesinmotorneuronactivitydriv-ingbehavioralresponse.Thesebehavioraldataweretheonesthatprompted“cognitivist”modelsofassociativelearning(Dickinson1980).TheadditionalmolecularresourcesprovidedbymorerecentdiscoveriesaboutLTP-inducedsynapticplasticityincreasethescopeofneurobiological“combinatorialreductions”tonumeroustypesoflearningandmemory(Bickle1998:ch.5).Morerecently,cognitivepsychologicaltreatmentsof“declarativelong-termmemory”andthe“consolidationswitch”fromshort-termtolong-termmemoryhavebeenaddedtothisgroup(SquireandKandel1999:ch.7).Theresultingcellandmolecularbiologicalexplanationsdomorethanjustcap-turethebehavioraldataqualitatively.Forexample,Hawkins(1989)developedaquantitativemodelofthepresynapticfeaturesusedinhisandKandel’searlierreductions.Thismodelenabledhimtomimicthesecell-biological“letters”inananatomicallyplausiblecomputersimulation.Hawkinsshowedthattheactionpotentialratecurvesovertimeinsimulatedmotorneuronsmatchedexactlythelearningcurves,behavioraldynamics,andchangingpatternsofreinforce-mentpredictedbytheRescorla–Wagnerequations.Hisquantitativemeasure,thefiringratesovertimeinthesimulatedmotorneurons,wascomputedbyparametersandchangeablesynapticweightvaluesacrosssimulatedsensory,facilitator,andmotorneurons.Allvalueswerechosentomimicknownbiologicalfeatures.Henceevenwhentheneurophysiological“letters”arelimitedtothepresynapticcell-biologicalmechanismsofHawkins’sandKandel’searlyaccount,simulatedmotorneuronactivitygeneratedbytheirsequencesandcombinationsinincreasinglycomplexneuralanatomiescaptureexactlythebehavioraldynamics7andpredictionsofthecognitive-psychologicalaccount.Thiscaseisjustoneexampleofareductionofagenuinelycognitivepsy-chologicaltoacell-biological/molecularneuroscientifictheory.ThereareotherexamplesthatdrawuponnewerdetailsofthecurrenttheoryofLTP-induced328\nPhilosophyofMindandtheNeurosciences8synapticplasticity.Thesecondplankofmyargumenthasthusbeenaccom-plishedforavarietyofgenuinelycognitivepsychologicaltheories.Psychoneuralreductionofthegenuinelycognitiveisalreadyanaccomplishedscientificfact.14.2Neurofunctions?PsychologistsMauriceSchoutenandH.LoorendeJong(1999)havechallengedmyargumentforthesecondplank.Theircriticismsdeservediscussionhereforatleasttworeasons.First,theyexpresspopularandscientificallymotivatedanti-reductionistthemes.Secondly,theyaddressdirectlymyempiricalcasestudy.Theirargumentsthusserveasgoodtemplatesforresponsiblecounterstoanempiricalargumentforpsychoneuralreduction.However,theirargumentsalsocontainimportantflaws,andpointingtheseouthelpstoclarifygeneralthemesofthereductionismimplicitincurrentmainstreamneuroscience.ThroughouttheircriticismsSchoutenanddeJong(1999)stresstwopoints:•theneedtospecifyfunctionsincomprehensivescientificexplanations;•theinaccessibilityoffunctionsfromtheoriesofphysicalmechanismsalone.Appliedspecificallytopsychoneuralinter-theoreticrelations,theyclaimthatbrainfunctionscannotbediscoveredby“purelybottom-up”theorizing,evenbyanapproachthatspecifiescomplexsequencesandcombinationsofcell-biologicalandmolecularprocesses.Theirfirstargumentcontainstwopremises.First,higher-leveldispositionsaremultiplysupervenientonphysicalsubstratesandmech-anisms.Inotherwords,numeroushigher-leveldispositionssuperveneononeandthesamephysicalsubstrate.Manyreaderswillrecognizemultiplesupervenienceasthereverseofthemorefamiliarnotionofmultiplerealizability.Multiplesuperveniencehasappearedincreasinglyinanti-reductionistarguments(Kincaid1988;Endicott1994).Secondly,givenmultiplesupervenience,ahigher-leveltheorytypicallyisrequiredinagivencasetodistinguishthecausallyrelevantlower-leveltraitsfromthecausallyirrelevantones.Onlysomelower-leveltraitsarecausallyrelevantforagivenevent(outofthemyriadthatoccuratthetime).Eschewinghigher-leveltheorieswillproducealossofobjectiveinformationabouttheparticulardispositionaltraitsofthephysicalsubstratumthatarerelevantforagivenexplanation.Sothe“purelybottom-up”methodologythatSchoutenanddeJongassumetobecharacteristicofcombinatorialreductionandthecell-9biological/molecular“alphabet”approach“won’twork.”Thereisavarietyofproblemswiththisargument.Thefirstisasimplemisun-derstandingofreductionism’smethodologicalcommitments.Themethodologypracticedincurrentneuroscience(andanalyzedseparatelybyHawkinsandKandelandbyme)doesnot“eschew”higher-leveltheories.Mostreductionistsnowexplicitlyembracecoevolutionaryresearchideology(firstespousedbyHooker1981).329\nJohnBickleSomeevenrecognizehigher-leveltheorizingasmethodologicallyindispensable,bothpriortoandafteranaccomplishedreduction(Bickle1996;1998:ch.4).Coevolutionitselfisamethodologicalrecommendation,notaconstraintorimpositionontheorychoice.Itisdesignednottoruleoutcertainhigher-leveltheories(i.e.,thoselackingreductivepotential),butrathertokeepafloatnascenttheoreticalsuggestions,togivethemachancetodisplaytheirexplanatorypowerandempiricalveracity.Historically,andevennow,itisphysiologicaltheoriesthatfacethestrongestresistanceinmainstreampsychology,socialscience,andphilo-10sophy.Furthermore,sinceanadequatepsychoneuralreductionismmustcoherewithcross-leveltheoryrelationsandmethodologyacrosstheboardinscience,psychoneuralreductionistsmustacknowledgethathigher-levelgeneralizationscanhave“strongepistemicwarrant”(Horgan1993)beforeandafterinter-theoreticreductionsobtain.Thehistoryofscienceoffersmanycasesofsuc-cessfultheoriesthatdevelopedforalongtimewithonlycursoryacknowledgmentoftheoriesaboveandbelow.Eventheverylogicoftheinter-theoreticreductionrelationspeakstotheneedtoacknowledgehigher-leveltheories.Reductionisatwo-placerelationbetween(developed)theoriesandsorequiresdeveloped,epistemicallywarrantedhigher-levelinstances.Thespecialsciencesmustcontinuetoprovidetheoriesevenastheirreducersdevelop,ifinter-theoreticreductionsaretoobtain.Finally,noticethattheroleascribedtohigher-leveltheoriesinSchouten’sanddeJong’sfirstargument,thatofdistinguishingcausallyrelevantfromcausallyirrelevantlower-leveldispositionaltraitsforparticularexplananda,isconsistentwithascribingtothemanessentialbutneverthelesspurelymethodolo-gicalrole.Theycanbeineliminableforguidinglower-leveltheorydevelopmentwithoutcommittingustoananti-reductionistconclusion.SchoutenanddeJongalsocriticizemyappealtothelearningandmemory–LTPlinkasanaccomplishedpsychoneuralreduction.Theirmistakesarecommonenoughtowarrantdiscussionhere.Theyfirstpointouthowhigher-levelneuropsychologicalresearchpromptedinitialphysiologicalinvestigationsofthemammalianhippocampus,whereLTPwasfirstdiscovered.Theirhistoryiscor-rect.Butthisonlyshowsthathigher-leveltheorizingismethodologicallyimport-antforneuroscienceandwejustscoutedreasonswhyreductionistsshouldnotdenythat.Thehistoricaldetailsdon’tjustifyanythingmorethanamethodo-logicalroleforhigher-leveltheorizing.SchoutenanddeJongalsoclaimthatsincethemid-1980s,“theempiricalsupportforthe‘LTPasmemorysubstrate’hypothesishascomemainlyfromtheuseofpharmacologicalagents...thatappeartoantagonizeNMDA[receptor]activityandtoimpairspatiallearning,”andthat“[i]nthistypeofresearch,spatiallearningisoperationallydefinedasperformanceinawatermaze”(1999:247;myemphasis).Eveninapapertar-getedforphilosophersandcognitivepsychologists,this“statementoffact”aboutneuroscientificresearchisnaive.Itwasn’teventrueintheearly1990s.Searchingfortitlewordsorkeywordsofabstractsofpresentationsatthe2000SocietyforNeuroscienceAnnualMeetingusingeither“LTP”or“synapticplasticity”yieldedmorethan200presentations(www.sfn.org).Onlyfourofthesealsocontained330\nPhilosophyofMindandtheNeurosciences“watermaze”inthetitleorasakeyword.Thecomplete200+abstractsindicatethevastnumberandvarietyofmolecularmanipulationsandbehavioralparadigmsnowemployedtostudyLTP,andofthenaivityofSchouten’sanddeJong’sassertion.Thisproblemisfarmorethanjustonefactualerroraboutneurosciencemadebynon-neuroscientists.SchoutenanddeJonginturnraisesomemethodologicalandinterpretiveproblemsspecifictoNMDAreceptorantagonistsandwatermazetasks,implyingthattheseproblemsconstituteageneralchallengefortheclaimedlearningandmemory–LTPreduction.Theirproblemsprovidenosuchthingbecauseofthewidevarietyofmolecularmanipulations,behavioraltasks,andrecentgeneticknockoutandtransgenicmanipulations(inmammals)thatprovideevidenceforthereduction(SquireandKandel1999:ch.7).Infact,themostconvincingrecentexperimentalworkdemonstratingthatLTPisacellular/molecularmechanismforlearningandmemoryinvolvesneitherpharmacologicalmanipulationsnortheMorriswatermaze.Instead,itcomesfromtransgenicadultmicemanipulatedtooverexpressagenewhoseproteinproductblocksthecata-lyticsubunitofproteinkinaseAinthehippocampus.Itemploysadualfear-conditioningbehavioraltestinvolvingenvironmentalcues,aneutralCS,andafootshockUS(seeSquireandKandel1999:149–53).Thereisagenerallessonforpsychoneuralanti-reductionistsinSchouten’sanddeJong’serror:hoorayforconsideringempiricalwork,butfirstmasterthescopeandvarietyofscientificinvestigationsbeingpursuedonthattopic.SchoutenanddeJongalsoraiseamoregeneralinterpretiveworryabouttheneuroscientificevidenceforthelearningandmemory–LTPinductionlink.Theypointoutthatitisimportanttoseparateinfluencesonlearningandmemoryfromthoseonothersystemsthatmightbecontributingtothebehavior.ManysystemsaresusceptibletoNMDAreceptorantagonists,includingsensory,motor,motiva-tional,andattentional.Allofthesesystemsareinvolvedinthewatermazetask.PerhapsLTPisamechanismprimarilyforplasticityinoneoftheseothersystems?Perhapsitis.Butthatisnotnewstoneuroscientists.Infact,it’sthereasonwhyneuroscientistsaresocarefulintheirexperimentaldesign.(Incidentally,specificmethodologicalworriesaboutspecificpharmacologicalagentshavepromptedneuroscientistsstudyingthecellularandmolecularmechanismsoflearningandmemorytoshifttheirexperimentalprotocolstogeneticknockoutandtransgenicpreparations;see,e.g.,SquireandKandel1999:119–24,151–3.)Philosophersandcognitivepsychologistsshouldnotskipoverthe“Methods”sectionofneurosciencepapers.Thisiswhereneuroscientistsrevealtheircontrolsfortheexperimentalvariableatissue–learning,vision,attention,movement,whatever–toavoidconfoundingfactorsthatcanwreckaninterpretation.Obviously,neuroscientistswilllistentoanybody’sfruitfulcriticismsofthespecificcontrolstheyemploy.Butphilosophersandpsychologistsreallyaren’trequiredtoinformthemaboutaneedtocontrolforpossibleconfoundinginfluencesasobviousastheonesSchoutenanddeJongpointout.Itmightsurprisephilosopherstosee11howsubtlethecontrolsarethatneuroscientistsroutinelyemploy.331\nJohnBickleFinally,SchoutenanddeJongarguethatfunctionaltheoriesaremorethanjustmethodologicallyessentialorimportant.Foronlywithsuchtheoriescanweanswer“‘why’and‘whatfor’questions,”questionsabout“what[themechanism]issupposedtodo,”about“therequisitenormativedimension”(1999:255–6).Theyinsistthatthesequestionsrequire“amoreontologicalinterpretationoffunctions”(ibid.:256).Thisemphasistiesinwiththeirplumbforteleofunctions.Asystem’steleofunctionsdependuponitsselective(evolutionary)history.Ateleofunctionaltheoryspecifieskindsthatunifydistinctphysicalsystemsbyrefer-encetothegoalstheyholdincommonviatheirselectivehistories.Onlyanappropriatefunctionaltheorycanaccountforthese“objectivepropertiesofreal-ity”(ibid.:256).Teleofunctionsareatpresentapopularnotioninthephilosophyofbiology,psychology,mind,andlanguage(Millikan1984;Post1991).Theyarecentraltothestrongestscientificallyinspiredanti-reductionistargumentaround.WhatIamabouttosayshouldnotbetakentobemy“definitiveresponse”(nosuchbeastyetexists).Butthereisalotthatisproblematicaboutthisnotionandargument.Whydoweneeda“unifyingspecification”ofthesedistinctphysicalsystemsonceweunderstandhoweachworksindividually–thatis,oncesuchafunctionalaccounthasperformeditsessential(butexclusively)methodologicalrole?Whatdoesthisunificationaddtoourontology?Whythinkthatanswersto“why”and“whatfor”questionsareontologicallycommitting,beyondthevarietyofphysicalmechanismsatwork?Noticealsothatformostofthepurposesassignedtophysicalmechanismsinpro-teleofunctionaldiscussions,the“higher-leveltheoriz-ing”istrivialandobvious.Considerthefavoriteexample:theheart’steleofunctionistopumpblood.Doesitreallytakemuch“high-leveltheorizing”toreachthisinsight?Theexampleisillustrative:the“teleofunction”ofmostsystemsisusuallyobvious,especiallywhenweunderstandtheirphysicalmechanisms.(Pleasenotethatthisisnottosaythatthetaskofunveilingtheirselectivehistoriesisusuallytrivialorobvious–itisn’t,asthedifficultyofrealevolutionarybiologyandecologyattest.)Thereisalsoanothersciencebesidesmainstreamevolutionarybiologycon-cernedwithexplainingwhyatraitexistsinagivensystem.Thatscienceismolecu-largenetics,anditsaspirationsareruthlesslyreductive.Somemoleculargeneticistseventhinkofevolutionarytheoryasservinganessentialbutexclusivelymethodo-logicalrole.Forexample,molecularbiologistJamesShapirohasstatedrecently:Mostofthebasicconceptsinconventionalevolutionarytheorypredate1953whenvirtuallynothingwasknownaboutDNA.Inthefirsthalfofthe20thcentury,mathematicaltreatmentsoftheevolutionaryprocesswereelaboratedusingtermssuchasgenes,alleles,dominance,penetrance,mutation,epistasis,fitness,andselec-tion....Althoughmoleculargeneticistsstillusemuchoftheoldlanguage...theyactuallyoperateinadistinctconceptualuniverse.Theconceptualuniverseofmoleculargeneticsisasdifferentfromclassicalgeneticsandevolutionarytheoryasquantumphysicsisfromclassicalmechanics.(1999:23;myemphasis)332\nPhilosophyofMindandtheNeurosciencesBuildingontheinitialinsightsofNobellaureateBarbaraMcClintock,Shapiro’spictureisofgeneticvariationresultingfromahostofcellularbiochemicalevents.“Mostevolutioniststry(unrealistically)tomodeltheactionofthesecellularfunctionstoresembletherandommutationaleventsofconventionalevolutionarytheory”(ibid.:28).Instead,internal(tothegenome)“signaltransductionnet-works”regulatethetimingandlocationofgeneticchanges,includingsimultan-eouschangesatmultipleloci.The“why”and“whatfor”questionsareaddressableatthelevelofDNAbiochemistryregulationbythevarietyofsignaltransductionnetworks,themselvesunderstoodincreasinglyinmolecularterms.The“environ-ment”doesnothingmorethan(occasionally)kick-starttheseinternalnetworks.Inthislight“evolutionmustbeviewedafreshattheendofthe20thcentury”(ibid.:23;myemphasis).Moleculargeneticsisignoredbythephilosophersofbiologywhohavebeenmostactiveindevelopingtheteleofunctionconcept.Theydrawinspirationfromascientifictheorywhich,accordingtosomemolecu-largeneticists,mustnowberethoughtcompletely.Perhapsthisignoranceofthenewdawningofmoleculargeneticsisthegreatmistakethattheteleofunctionconceptrestsupon.Finally,inthecontextofbothphilosophyofbiologyandmind,an“ontologicalinterpretation”of(teleo-)functionsmustbeseenforwhatitis.Theresultingaccountisadualismoftheclassicalpropertyoreventvariety.Thisinterpretationimpliesthattherearepropertiesoreventsnotexplainablebyphysicalmech-anisms.Nothing(atpresent)isobjectionableinandofitselfaboutsuchaview.Ourbestbiologyandpsychologymightcommitusintheendtonon-physicalpropertiesorevents.Butthosewhoseektodefendaphysicalisminanymeaning-fulsensecan’thelpthemselvessocavalierlyto(teleo-)functionsinterpretedontologically.Italsoseemsextremelycavaliertoignorethereductionistsympa-12thiesofcontemporaryneuroscienceandmolecularbiology.Ifanyareasconsti-tutethe“crowningglory”ofcurrentmainstreambiology,itisthesetwo.That’ssubjecttochange,ofcourse,butindifferencetothembycurrentphilosophersofbiologyandmindisperverse.14.3ConsciousnessandCellularNeuroscienceConsciousnessisonepsychologicalphenomenonthatmanythinktobefarre-movedfromreductionisticneuroscience.Ignoredfornearlytheentiretwentiethcenturybymainstreamsciencesofmind,ithasroaredbackrecentlyinbothsci-enceandphilosophy.Followinginitswakehavebeenexplicitrevivalsofdualism(Jackson1982;Nagel1989),“newmysterian”worriesaboutour(human)capacitytosolvetheconsciousness-brainproblem(McGinn1989),andcallsto“revolu-tionize”physics(Chalmers1996;Penrose1994).Evenphysicalistssympathetictoneuroscienceassumethatexplainingconsciousnessrequires“exotic,”“whole-brain”resources:sophisticatedbrain-imagingtechniques,massivelyparallelneural333\nJohnBicklenetworks,andmathematicalanalysisoftheirglobalactivity.Thesharedideahasbeenthatthetechniquesoftraditionalneurophysiologyarenotuptothetask,evenifneuroscienceultimatelyis.OnenotableexceptionisperceptualneurophysiologistWilliamNewsome.Inanexchangeaboutthe“singleunitapproach”ofmainstreamneuroscience,heexclaimsthat“wehavenotyetbeguntoexhaustitsusefulness....[E]xcitingtome...istherecenttrendtowardapplyingthesingleunitapproachinbehavinganimalstrainedtoperformsimplecognitivetasks”(inGazzaniga1997:57).Newsomementionstasksinvolvingperception,attention,learning,memory,andmotorplanning.Resultsfromhisownlabcanbeinterpretedinawaythatmakesthemrelevanttoarecentphilosophicalcontroversyaboutconsciousness.Phenomenalexternalismholdsthattheenvironmentexternaltoanindividual’sreceptorsurfacesdetermines(“individuates”)thequalitativecontents(“qualia”)ofhissensoryexperiences(Dretske1996;Lycan1996).Partofitsmotivationistherecentstampedetoward“representational”theoriesofqualiacoupledwiththedominanceofrepresentational-contentexternalisminphilosophygenerally.Argumentsforthelatterappealtoaphilosopher’spopularfantasy:TwinEarth.Itiscommontousethesethoughtexperimentstodefendexternalismaboutlin-guisticmeaningandcognitivecontent.FredDretskeinsiststhatnothingpreventsonewhoacceptsthemforcontentexternalismfromacceptingthesameargumentsforphenomenalexternalism:“Justaswedistinguishandidentifybeliefsbywhattheyarebeliefsabout,andwhattheyarebeliefsaboutintermsofwhattheystandintheappropriaterelationto,sowemustdistinguishandidentifyexperiencesin13termsofwhattheyareexperiencesof”(1996:145).TheradicalnatureofthisviewisapparentinDretske’ssloganesquephrase:“Theexperiencesthemselvesareinthehead...butnothinginthehead...needhavethequalitiesthatdistin-guishtheseexperiences”(ibid.:144–5).Althoughtheyarephysicalduplicates,andtherebyneurophysiologicalduplicates,Fred’sandTwinFred’sconscioussensoryexperiencesmighthavedifferentqualiaowingtodifferencesintheirexternalenvironments.However,aninterpretationofNewsome’s“single-unit”resultsutilizingmicrostimulationofvisualareaMTinrhesusmonkeys(seefigure14.2)demon-stratestheempiricalimplausibilityofexternalistintuitionsaboutqualitativecon-tent.AreaMT(middletemporalcortex)isthegatewaytothe“dorsal”(“parietal,”“where”)visualprocessingstream(seefigure14.3).Bothlesionstudiesandelectrophysiologicalrecordingshaverevealeditsroleinvisualjudgmentsofmotiondirection.MostMTneuronsaredirectionselective,firingoptimallytoavisualstimuluswithmotioninasingledirection.Likeothercorticalareas,MThasacolumnarorganization,withneuronsinagivencolumnsharingsimilarreceptivefieldsandpreferredmotionselectivity.Thesefeaturesvaryfromcolumntocolumn,soMTasawholerepresentsallmotiondirectionsateachpointinthevisualfield(Albrightetal.1984).Newsomeandhiscolleaguesdevelopedatechniqueforquantifyingthestrengthofamotionstimulus(Salzmanetal.1992;seefigure14.4).Sequencesofdotsare334\nPhilosophyofMindandtheNeurosciencesPosteriorparietalcortexFrontaleyefieldsV3DLPCV1V4V2MTITMSTTEOFigure14.2Anatomicalorganizationofprimate(macaque)visualsystem.Abbrevi-ations:MT(middletemporalcortex);MST(middlesuperiortemporal);IT(inferiortemporalcortex);DLPC(dorsallateralprefrontalcortex).plottedonacomputerscreen.Thestrengthofamotionstimulus,expressedintermsofapercentagecorrelation,reflectstheprobabilitythatagivendotwillbereplottedatafixedspatialandtemporalinterval.Forexample,ina“50percentcorrelationverticalstimulus,”halfthedotsarereplottedatafixedupwardinterval(providingtheillusionofverticalmotion),whiletheotherhalfarereplottedrandomly.Newsome’sgroupalsodevelopedabehavioralparadigmfordetermin-ingjudgmentsofmotiondirection.Theircontrolsareelaboratebutthebasicideaisstraightforward.Themonkeyfixatesonanilluminatedcentralpoint,andmain-tainsfixationwhilepresentedwithavisualmotionstimulusofaparticularstrength(seefigure14.5).Boththefixationpointandthemotionstimulusareextin-guished,andtargetlights(LEDs)appearintheperiphery.OneLEDislocatedinthedirectionofthemotionstimulus.Theotherislocatedintheoppositeperi-phery.Themonkeyindicatesitsjudgmentofmotiondirectionbysaccading(mov-ingitseyesrapidly)totheappropriateLED.Monkeysareonlyrewardedwhentheysaccadecorrectly(i.e.,totheLEDinthedirectionofthemotionstimulus).ByfirstlocatinganMTcell’sreceptivefield(theportionofthevisualfieldinwhichstimulielicitaresponse)andpreferredmotionselectivity,experimenterscanpresentthemotionstimulustoonlythatregionofthevisualfield.Theycanthencomparethemonkey’sreportaboutthemotiondirectionacrossstimulusstrengthswhenelectricalmicrostimulationisappliedtothatcellduringstimuluspresentationandwhenitisnot.ThetargetLEDinthecell’spreferredmotiondirectionisdubbedthePrefLED,andthetargetintheoppositedirectionisdubbedtheNullLED.Themonkey’ssaccadeconstitutesareportofapparent(perceived)motiondirection.ThismeasureofmotionstrengthandthebehavioralparadigmenableNewsome’sgrouptoplottheproportionofthemonkeys’reportsofapparentmotioninanMTneuron’spreferreddirectionasafunctionofmotionstimulusstrength(see335\nJohnBickleFEFAITMSTVIPPITMTV4Dorsal(“where”)Ventral(“what”)streamstreamV3V2V1LGNRetinaFigure14.3Flowchartofthemajorstructures,corticalanalyzerareas,circuitries,andprocessingstreamsinthemammalianvisualsystem.Abbreviations:asinfigure14.2,exceptLGN(lateralgeniculatenucleusofthedorsalthalamus);PIT(posteriorinferiortemporalcortex);AIT(anteriorinferiortemporalcortex);VIP(ventralintraparietalarea);FEF(frontaleyefields).figure14.6).Ifmicrostimulationtodirection-selectiveMTneuronsadds“signal”totheneuronalprocessesunderlyingvisualjudgmentofmotiondirection,thenitwillbiasthemonkeys’reportstowardthatneuron’spreferreddirection.Graphic-ally,thiswillresultinaleftwardshiftofthepsychometricfunction(seeagainfigure14.6).TheseareexactlytheresultsNewsomeandhiscolleaguesobserved,underavarietyofstimulusstrengthsandmicrostimulationfrequencies(Salzmanetal.1992;Murasugietal.1993).Atnearlyeverypercentagecorrelation,microstimulationofadirection-selectiveMTcellbiasedsignificantlythemon-keys’saccadestothePrefLED.Thisbiasoccurredeveninthepresenceofstrong336\nPhilosophyofMindandtheNeurosciencesNoverticalmotion50%verticalmotion100%verticalmotioncorrelationcorrelationcorrelationFigure14.4Quantitativemeasureofstrengthofmotiondirectionstimulus.Actualdisplayscontainedmanymoredotsthanareillustratedhere.(AdaptedfromSalzmanetal.1992:2333,figure1.)motionstimuliintheoppositedirection(e.g.,>−50percentcorrelation).Recallalsothatmonkeysareonlyrewardedwhentheyreportthestimulus’smotiondirectioncorrectly.Theyneverreceivearewardforthesecontinuallyincorrectchoices.Increasingmicrostimulationfrequencyincreasedtheproportionofap-parentmotionreportsintheneuron’spreferreddirection,evenunderconditionsofstrongermotionstimuli(percentcorrelation)intheoppositedirection.Theseresultsleadnaturallytothequestion:whatdoesthemonkeyseeinmicrostimulationtrials?Isthemonkeyconsciouslyawareofmotionintheneu-ron’spreferreddirection,evenwhenthemotionstimulusisintheoppositedirection?Newsomeandhiscolleaguesadmitthattheirresultscan’tanswersuchquestionsconclusively.Buttheyalsodon’tshrinkfromofferingsomesuggestions:[A]plausiblehypothesisisthatmicrostimulationevokesasubjectivesensationofmotionlikethatexperiencedduringthemotionaftereffect,orwaterfallillusion....Motionthereforeappearstobeaqualitythatcanbecomputedindependentlywithinthebrainand“assigned”topatternedobjectsintheenvironment.(Salzmanetal.1992:2352;myemphases).Theyareclaimingthatmotionqualiaaregeneratedinternallybyneuralactivityand“attached”torepresentationsofexternalobjects.Happily,our“internalassignments”tendtomatchupwellwithexternalevents.Naturalselectionwascruelertocreatureswhose“assignments”weremorehaphazard.Butundertherightconditions,ourinternallygeneratedqualiaandtheexternaleventscanbedissociated.ThatiswhathappensinNewsome’smicrostimulationstudies.ThegeneralideaatworkhereiswhatneuroscientistRodolfoLlinásandneurophilosopherPatriciaChurchlandcallendogenesis.Astheyputit,“[t]hecruxhereisthatsensoryexperienceisnotcreatedbyincomingsignalsfromtheworldbutbyintrinsic,continuingprocessesofthebrain”(LlinásandChurchland1996:x).Incomingsignalsfromreceptorskeyedtoexternalparametersfunctionto“trellis,shape,andotherwisesculpttheintrinsicactivitytoyieldasurvival-facilitating,me-in-the-worldrepresentationalscheme”(ibid.).Naturalselection–adequacyfor14exploitinganenvironmentalniche,nottruth–determinesascheme’s“success.”337\nJohnBickleFixationpointPrefLEDNullLEDreceptivefieldstimulusaperatureFixationpointVisualstimulusElectricalstimulationTargetLEDsEyepositionTimeT1T2T31secFigure14.5Newsome’sexperimentalparadigminvolvingelectricalmicrostimulationofindividualneuronsinareaMT.(AdaptedfromSalzmanetal.1992:2334,figure2.)Newsome’sexperimentalevidenceandinterpretation,alongwiththegeneralconceptofendogenesis,countstronglyagainstphenomenalexternalism.Noticefirstthatamonkeyina“microstimulation+(strong)nulldirectionstimulus”trial,comparedtoa“nomicrostimulation+(strong)preferreddirectionstimu-lus”trial,isanempiricalanalogueofaTwinEarthcase.Thetwobrainstatesare(closeto)identicalinthetwocases,atleastfromMTandfurtherupthedorsalstream(thesitesthatmatterforvisualmotiondetectionandjudgment).Yettheenvironmentalstimuliaredifferent.Inthefirstcase,motioninthenulldirec-tioncorrelateswiththatbrainstate(becauseofthemicrostimulation).Inthesecondcase,motioninthe(opposite)preferreddirectioncorrelateswithit.If338\nPhilosophyofMindandtheNeurosciences1.00.80.60.4Proportionpreferreddirection0.2−40−200204060CorrelationpercentFigure14.6Aschematicpsychometricfunctionplottingproportionsofdecisionsinamotion-selectiveMTneuron’spreferreddirectionasafunctionofmotionsignalstrength(dotsandsolidline).Theleftwardshiftofthefunctionispredictedfollowingmicrostimulationifmicrostimulationaddssignaltotheneuronalprocessesunderlyingvisualjudgmentofmotiondirection(dottedline).(AdaptedfromSalzmanetal.1992:2335,figure3.)phenomenalexternalismistrue,themotionqualiashoulddiffer.Andyetthemonkeysreportthesamedirectionofapparentmotioninthetwocases(bywayoftheirtrainedsaccadestothePrefLED).InaccordancewithNewsomeandhiscolleagues’interpretationquotedabove,thissuggeststhatthemotionqualiaaresimilarinthetwocases,notdifferent.Thereisalsoevidencethatthiseffectisnotspecifictorhesusmonkeys.AsNewsomeandhiscolleaguesremark,“ithasrecentlybeenreportedthatcrudemotionperceptscanbeelicitedwithelectricalstimulationofhumanparietal-occipitalcortex”(Salzmanetal.1992:2352).Norisitspecifictomotion.Themeasureofstimulusstrength,thebehavioralparadigm,andthemicrostimulationtechniquegeneralizetoothertypesofvisualstimuli,includingorientation,color,andstereoscopicdisparity.Morerecently,Newsomeandhiscolleagueshavereportedsimilarmicrostimulationresultsforstereoscopicdepth(DeAngelisetal.1998).Withregardtothequalitative339\nJohnBickleSpikingfrequency0DegreeofstimulusdimensionFigure14.7Tuningcurveforastimulusdimension-selectiveneuron(e.g.,color,orientation)displayingastandardGaussianresponse.Dimensiondegreeonthex-axisunderneaththehighestpointofthecurvereflectstheneuron’spreferredstimulusdimension.contentofconsciousvisualexperiences,whatmattersiswhatgoeson“inthehead”(thebrain).Theintuitionsdrivingphenomenalexternalismappeartobeempir-icallyimplausible.Anditisgoodoldsingle-unitneurophysiologythatprovidestheempiricalevidenceforthisphilosophicalconclusionaboutconsciousqualit-ativecontent.Considerasecondexampleof“single-neuron”neurophysiologyyieldingre-sultsthatareapplicabletophilosophicalconcernsaboutconsciousness.McAdamsandMaunsell(1999)studiedtheeffectsofexplicitconsciousattentiononactivityofsingleneuronsinmacaque(visual)areasV4andV1(seeagainfigure14.2above).V1(primaryvisualcortex)receivesretinotopicinputsviathelateralgeniculatenucleusofthedorsalthalamus.V4,furtherupinextrastriatecortex,isthegatewaytothe“ventral”(“temporal,”“what”)visualprocessingstream(seeagainfigure14.3above).V4containsbothorientation-andcolor-selectiveneurons.Mosthaveapreferredorientationorcolorthatelicitsmaximalactiv-ity.Similarstimulielicitlessactivity,anddissimilaroneselicitnone(overbaselineresponserate)(seefigure14.7).Psychologistshaveknownforalongtimethatexplicitconsciousattentionyieldsimprovedsensoryperformance.Measuresincludeimproveddetectionthresh-oldsandquickerdiscrimination.Atthelevelofindividualsensoryneurons,explicitconsciousattentioncouldalterneuronalresponsetoaccountforthese15behavioralimprovementsinoneoftwoways.First,itcouldincreasetheampli-tudeofneurons’activity(seefigure14.8A).Theneurons’stimulusselectivityremainsthesame,asreflectedinthesimilarwidthsofthetwotuningcurves.Frequencyofactionpotentialsgeneratedtostimuliincreases,asreflectedintheheightofthetuningcurveatvirtuallyallstimulusdimensions.(Thiseffectis340\nPhilosophyofMindandtheNeurosciencesABFigure14.8Twopossibleeffectsofexplicitconsciousattentiontolocationofavisualneuron’sreceptivefieldonitsactivityprofile.A:Multiplicativescalingofneuron’sresponse(relativelyconstantincreaseinactivityratetoavarietyofstimulusdimensiondegrees)withoutincreasedstimulusselectivity.B:Increasedstimulusselectivity,reflectedbyasharpeningoftheneuron’stuningcurve.referredtoas“multiplicativescaling.”)Strongerneuronalresponsestypicallyhaveabettersignal-to-noiseratio,whichcouldexplainimprovedbehavioraldetectionthresholdsandspeed.However,thisroleforexplicitconsciousattentionwouldbedeflationaryforconsciophiles,whoinsistthatconsciousnessis“special”or“unique”atleastinitsmodeofneuralrealization.Thisresultwouldrendertheeffectsofexplicitconsciousattentionsimilarto,e.g.,simplyincreasingthesali-enceofthevisualstimulus.Consciousattentionwouldserveasaninternal,endogenousmechanismforjust“turningupthegain”onindividualneurons.Ontheotherhand,consciousattentionmighthaveamorerobustanduniqueeffect.Perhapsitaltersthestimulusselectivityofindividualneurons,causingactivityintheseneuronstosignalmorepreciselytheattributesoftheattendedstimulus.Asharpeningofneuronaltuningcurvesunderconditionsofexplicitconsciousattentionwouldreflectthiseffect(seefigure14.8B).Asharpertuningcurvewouldprovideamorefine-grainedrepresentationofthestimulusdimension,whichcouldimprovedetectionthresholdandspeed.Consciophilescouldbeheartenedbythisresult,sinceincreasingneurons’stimulusselectivityisnotacommonneurophysiologicaldynamic.Totestthesecompetingexplanations,McAdamsandMaunsell(1999)developedadelayedmatching-to-sampletaskconjoinedwithsingle-cellrecordingsinV4andV1(seefigure14.9).TheyfirstdeterminedreceptivefieldsandstimulusselectivityofV4andV1neuronstoberecordedfromduringsessions.Thedashedovalinallframesoffigure14.9representsthelocationoftherecordedneuron’sreceptivefield.Priortoatesttrial,themonkeyhadbeencuedastowhichlocationtoattend:theonewithintheneuron’sreceptivefieldortheone341\nJohnBickleUnattendedAttendedTestDelaySampleTimeFixationFigure14.9SchematicillustrationofMcAdamsandMaunsell’sdelayedmatch-to-sampletask.Seetextforexplanation.(Adaptedfromtheir1999:433,figure1.)locateddiametricallyoppositeit.Atrialbeganwhenthemonkeyfixatedacentraldotanddepressedabutton.Samplestimuli–orientationbarsoracolorpatch–appearedonascreen500millisecondslater.Onestimulusoccupiedtheneuron’sentirereceptivefield,theothertheoppositelocation.Thesamplesoccupiedthescreenfor500milliseconds,andthendisappeared.Thedelayperiodlasted500milliseconds,afterwhichteststimuliappeared.Themonkeyhadtoindicatewhethertheteststimulusatthecuedlocationmatchedthesamplebyeitherreleasingthebuttonwithin500millisecondsifthestimulimatched,orbycon-tinuingtodepressthebuttonforatleast750millisecondsiftheydidnot.Inthecaseillustratedinfigure14.9,forexample,themonkeymustcontinuetodepressthebuttonifcuedtoattendtotheorientationlocation,sincethesampleandtestorientationbarsdonotmatch.Butthemonkeymustreleasethebuttonifithadbeencuedtoattendtothecolorlocation,sincethesampleandtestcolorpatchesmatch(thoughthisisnotapparentintheblack-and-whitefigure).Monkeyswererewardedonlyiftheyreportedcorrectlymatchornon-matchinthecuedloca-tion.Matchesandnon-matchesatthetwolocationswereuncorrelated,sothemonkeycouldgainnoadvantagebyattendingtothewronglocation.342\nPhilosophyofMindandtheNeurosciencesMonkeysmaintainedfixationonthecentralpointthroughoutallphasesofatrial.Thisinsuredthatvisualinputtothecellremainedconstant.Whenpresent-ingthesamevisualstimulustothecuedlocationinsampleandtestphases,anydifferencesinrecordedneuronactivitycouldbeattributedtodifferencesinthemonkey’sattentionalstate.SinceallV4recordingsweremadefromorientation-selectiveneurons,the“Attended”modeoccurredwhentheanimalperformedanorientation-matchingtask(seefigure14.9).Therecordedneuronwasthenrespondingtothestimulusrelevantforthematchingtask.The“Unattended”modeoccurredwhentheanimalperformedthecolor-matchingtask,sincetherecordedneuronwasstillrespondingtotheorientationstimulus,butthatstimu-luswasirrelevanttothematchingtaskathand.Anychangestotheneuron’sfiringrateinAttendedcomparedtoUnattendedmodereflecttheneuronaleffectsofexplicitconsciousattentiontothelocationoftheneuron’sreceptivefield.Experimentalresultswithmorethan200orientation-selectiveV4neuronsand124V1neuronsclearlysupportedthemultiplicativescalinghypothesis(figure14.8Aabove).(SeeMcAdamsandMaunsell1999:figs2,4,5,6,7,and10.)Forbothindividualcellsandaverageswithinpopulations,amplitudeofAttendedresponses(frequencyofactionpotentials)comparedtoUnattendedresponsestothesameorientationstimuluswas(statistically)significantlygreater.Explicitcon-sciousattentiontothelocationofasensoryneuron’sreceptivefieldenhancesitsactionpotentialfrequencytoitsfavoreddegreeoftherelevantstimulusdimen-sionandtootherssimilartoit.However,thestandarddeviationtotheentirerangeofstimulusdimensiondegreesremainedconstantacrossAttendedandUnattendedmodes.Thismeansthatthetwotuningcurveshavenearlyidenticalwidths.Henceexplicitconsciousattentiondoesnotaffectaneuron’sstimulusselectivity.Finally,theAttendedandUnattendedtuningcurveshadnearlyiden-ticalasymptotevalues.Thismeansthatexplicitconsciousattentionhasnoeffectonaneuron’sresponseto“unpreferred”degreesofastimulusdimension.Com-biningtheseresultsyieldsaclearconclusion.Directingexplicitconsciousattentiontothelocationofasensoryneuron’sreceptivefieldsimplyincreasestheneuron’sresponsetopreferredandsimilarstimuli.Itonly“turnsupthegain”withoutsharpeningtheneuron’sstimulusselectivity.McAdamsandMaunsellpointoutthatexplicitconsciousattentionthereforehasthesameeffectonsingleneuronactivityasdoproceduresasmundaneasmanipulatingstimulussaliencyandcontrast:Thephenomenologicalsimilaritybetweentheeffectsofattentionandtheeffectsofstimulusmanipulationsraisesthepossibilitythatattentioninvolvesneuralmechanismsthataresimilartothoseusedinprocessingascendingsignalsfromtheretinas,andthatcorticalneuronstreatretinalandattentionalinputsequivalently.(1999:439)Theirresultssupportthe“deflationary”viewofconsciousnessmentionedabove.Concerningitseffectsonsingleneurons,explicitconsciousattentionisjustanother“gainincreaser.”343\nJohnBickleItmightsoundmysterioustoattributecausaleffectstoexplicitconsciousatten-tionatthelevelofsingleneuronactivity.Singleneuronsarebiochemicallycom-plicatedionchannelsandpumps.Doesexplicitconsciousattentionalterchannelproteins’shapesandelectricmembranegradients?Iscellularneurosciencerevital-izingdualism?Ofcoursenot.Extensiveexcitatoryprojectionsfromhigherneuralregionsinthevisualstreamsandcross-columnarprojectionswithinacorticalregionprovideastraightforwardphysicalexplanationofendogenouslygeneratedsingle-neurondynamics(Gilbertetal.2000).Despitethis,evenphysicalistconsciophilesshouldbetroubledbyMcAdams’sandMaunsell’sresults.Althoughtheyhavegrowncomfortablewiththeeventualphysicalexplanationofconscious-ness,theystillholdoutforthespecial,uniquenatureofitsneuralrealizationandeffects.Somehow,consciousnessmustdosomethingmoreinthebrainthanjustwhatincreasingstimulussaliencyandcontrastaccomplish.McAdams’sandMaunsell’sresultsdenyconsciophileseventhis.Thisconsequencebyitselfisphilosophicallyinteresting.Thatitwasgarnishedby“single-cell”neurophysiologyshowsfurtherthepotentialofreductionisticneuroscience,evenforphilosophicalconcernsaboutconsciousness.14.4ReductionistNeuroscienceand“HardProblems”Thereareneuroscientistswhothinkofthebrainas“justanotherorgan.”How-ever,manypursueneuroscienceto“knowthyself”andareunashamedtoexpressthisattitude.Forexample,intheIntroductiontohisinfluentialtextbook,neuro-biologistGordonShepherddescribessomereasonsforstudyingneurobiology.Twoareespeciallyrevealing:Aswegrowolder,weexperiencethefullrichnessofhumanbehavior–theabilitytothinkandfeel,torememberandcreate–andwewonder,ifwehaveanywonderatall,howthebrainmakesthispossible.(1994:3)Whatistheneurobiologicalbasisofracism–thefearandhatredofpeoplewhoaredifferent?Doterrorismandcrimegetbuiltintoourbraincircuits?Whydohumanbeingsseembentonself-destructionthroughenvironmentalpollutionandthedevelopmentofweaponsofannihilation?Whydowehavethisinourbrains,andhowcanwecontrolit?Inallofscienceandmedicine,neurobiologyistheonlyfieldthatcanultimatelyaddressthesecriticalissues.(ibid.)Thesearen’ttherantingsofsomeleft-fieldcrank;theyarefromtheeditoroftheJournalofNeuroscience.NoraretheyidiosyncratictoShepherd.Similarcitationscouldbeexpandedmany-fold.Mostneuroscientistsaren’tphilosophicalphilistines.Thisstillwon’tsatisfysomephilosophers.Manyremainjealousguardiansofthe“qualitative”and“subjective”aspectsofmind.Theyseemtothinkthatonlythey(alongwithperhapsahandfulofpsychologists)grappleseriouslywith“whatitis344\nPhilosophyofMindandtheNeuroscienceslike”tobeaconscious,mindfulhumanbeing.Theyimplythatthesefeaturesofmindarebeyondneuroscientists’professionalinterestandreach.Buttheyarewrongevenaboutthis.ConsiderthefollowingquotefromWilliamNewsome.Thetaskhereferstoisthemotiondirectiontaskdiscussedintheprevioussection.Ibelievethenatureofinternalexperiencemattersforourunderstandingofnervoussystemfunction....EvenifIcouldexplainamonkey’sbehavioronourtaskinitsentirety(inneuralterms),IwouldnotbesatisfiedunlessIknewwhethermicrostimulationinMTactuallycausesthemonkeytoseemotion.Ifwecloseupshopbeforeansweringthisquestionandunderstandingitsimplications,wehaveminedsilverandleftthegoldlyinginthetailings.(inGazzaniga1997:65–6;myemphases)YetNewsomeasksfornospecialdisciplineormethodologytoaddress“hardproblems”aboutconsciousness.Therearenoshortcutsaroundabroadlyempirical,reductionistpath:“Forthetimebeing...Isuspectwemustfeelourwaytowardstheseambitiousgoalsfromthebottomup,lettingthenewlightobtainedateach16levelofinquiryhintatthequestionstobeaskedatthenextlevel”(ibid.:67).Thezealousguardiansof“hardproblems”inthephilosophyofmindshouldlightenup.Theyaren’ttheonlyonesrespectfulorinpursuitofthefullgloryofmind.Iftheneuroscientiststhemselvesaretobetrusted,theseproblemsarenotbeyondtheprofessionalinterestsorreachofneuroscience.Newsome,forexample,concludes:“ThoughIamsensitivetotheissueof‘hard’limitstoourunderstanding,theoverallendeavorofcognitiveneuroscienceisgrand.Itisworththededicationofascientificcareer,anditcertainlybeatscloninganothergene!”(inGazzaniga1997:68).Italsobeatsconcoctingyetanothervariantonwornphilosophers’fantasies,liketheTwinsandMarytheutopianneuroscientist(tonamejusttwo).14.5TowardGenuinelyInterdisciplinaryPhilosophyandNeuroscienceOneofthecelebratedthemesoflatetwentieth-century“analytic”philosophyisthecontinuitybetweenthesciencesandphilosophy.WitnessQuine:Ontologicalquestionsareonaparwithquestionsofnaturalscience...thisdiffer-enceisonlyoneofdegree...that...turnsuponourvaguelypragmaticinclinationtoadjustonestrandofthefabricofscienceratherthananotherinaccommodatingsomeparticularrecalcitrantexperience.(1949:45)OrWilfridSellars:“Itisthe‘eyeonthewhole’whichdistinguishesthephilo-sophicalenterprise.Otherwise,thereislittletodistinguishthephilosopherfromthepersistentlyreflectivespecialist”(1962:39).OrHansReichenbach:“Toputitbriefly:thisbookiswrittenwiththeintentionofshowingthatphilosophyhas345\nJohnBickleproceededfromspeculationtoscience”(1957:vii).Thenextgenerationofphilosophersofmindtooktheseclaimstoheart.HilaryPutnam(1960)foundinspirationforhisearlyfunctionalismincomputabilitytheory.JerryFodor(1975)foundevidenceforhisincognitivepsychologyandChomskianlinguistics.DanielDennett(1978)foundthe“intentionalstance”lurkinginartificialintelligence.PaulandPatriciaChurchland(1985;1986)foundanalternativeaccountofthestructureandkinematicsofcognitionemergingfromtheneurosciences.Thisinterdisciplinaryturninphilosophywasthevanguardofanentireintellec-tualtrend.Interdisciplinaryprogramsbeganspringingupthroughoutthesci-ences.Itisnoaccidentthatphilosophyofmindsawsomuchofthisimpact.Cognitivescience,especially“cognitiveneuroscience”oflate,isthemostvisible(andwell-funded)exampleofself-proclaimed“interdisciplinarity.”PsychologistStephenKosslyn’scharacterizationistypical:[C]ognitiveneuroscienceisaninterdisciplinarymeldingofstudiesofthebrain,ofbehaviorandcognition,andofcomputationalsystemsthathavepropertiesofthebrainandthatcanproducebehaviorandcognition.Idon’tthinkofcognitiveneuroscienceastheintersectionoftheseareas,ofthepointsofoverlap,butratherastheirunion:Itisnotjustthateachapproachconstrainstheothers,butratherthateachapproachprovidesinsightsintodifferentaspectsofthesamephenomena.(inGazzaniga1997:158–9)Yetonediscoversadifferentattitudeamongcognitiveneuroscientistswhenthekidglovesareoffanddecorumpermitsgripestobeaired.Fewrejecttheinterdis-ciplinaryidealinprinciple.Butinpractice,almosteverybodyisconvincedthatthoseinotherdisciplinesremainignorantofthecontributionsofone’sown.Findingpublishedevidenceofthisattitudeisnoteasy.Scientificwritingtendstokeepsuchattitudessubterranean,andthephilosophersinvolvedwantsomuchtobetakenseriouslybythescientiststhattheyexpressitonlyrarely.However,abookeditedbyneuroscientistMichaelGazzaniga(1997)providesthenecessaryformat,andmanyreaderswillbesurprisedtoseehowdeeplythisattituderuns.Thebookcontains“interviews”withtenprominentcognitiveneuroscientistsfromthevarietyofdisciplinesmakinguptheendeavor.PublishedoriginallyintheJournalofCognitiveNeurosciences,theinterviewswereemailcorrespondences,andwereeditedonlyminimally.Theideawastomimictheafter-hoursconversationsthatexcite,invigorate,andsometimesevenmotivate.(ThequotefromKosslynjustabovecomesfromhisinterview.)Onethemethatemergesisthat“interdisciplinarity,”whilecommendableinprinciple,isstillamythinpractice.RadiologistMarcusRaichle,whoseworkwassoinstrumentalindevelopingpositronemissiontomography(PET)andfunctionalmagneticresonanceimaging(fMRI)technologiesandanalysis,labelsthe“simplisticbehavioralmethods”and“indiscriminateuseofsoftwarepackagestoanalyzedata”asthe“Achillesheel”ofmanyfunctionalimagingexperiments(inGazzaniga1997:33).Psychologist346\nPhilosophyofMindandtheNeurosciencesRandyGallistelclaimsthatneurobiologicalapproachestothecellularmechanismsofmemoryarehamperedbyoutdatedideasaboutthescopeofandcrucialparametersforevenassociativelearning:“[C]urrentresearchon[theneuralbasisof]memoryisbasedonafundamentallyerroneousconceptionofwhattheelementsofmemoryformationare”(inibid.:75).Computationallycomplexandrealisticmodelsofmemoryarebeingdevelopedinhumancognitivepsychologyandethology,butthesearebeingignoredbyneuroscientists.PsychologistEndelTulvingcarriesthisgripeastepfurther.Neuroscientistsstudyingthemechanismsofmemoryhaveignoredone-halfofthephenomenonentirely:retrieval.Cognitivepsychologists“discoveredretrievalandfiguredouthowtoseparateitanalyticallyandexperiment-allyfromstorageinthe1960s.”Thesediscoveriesrevolutionizedmemoryresearchincognitivepsychologyinthe1970s,but“thatrevolutionhasnotyetreachedbrainscientists”(inibid.:95–6).LinguistStevenPinkerinsiststhatneithertheimportanceofanevolutionaryperspectiveonlanguagenorevenafamiliaritywith“mainstreamevolutionarybiology”hasreachedChomskianpsycholinguists.Moregenerally,“thevastmajorityofcognitivescientistsandneuroscientistshavenotreallythoughtabouttheevolutionofthebrain”(inibid.:113–14).NeuropsychologistAlfonsoCaramazzaclaimsthatoutdatedviewsfromgeneralphilosophyofscienceaboutpredictabilityhaveimpededacceptanceofthenew“cognitiveneuro-psychological”approachtolanguage,despitethevarietyofnewdeficitstheapproachcontinuestoreveal(inibid.:142–3).Granted,thesearetheattitudesofonlyahandfulofresearchers.Buttheyarefromprominentones.OneleavesGazzaniga’sinterviewswiththefeelingthatinvestigatorsatthelowerlevelsremainweddedtobehavioralmethodologiesandcognitivetheoriesandconceptsthathavebeenoutofdateforthreedecadesinthedisciplinesfromwhichtheyaredrawn.Similarjudgmentsabouthigher-levelpractitioners’knowledgeofcellularandmolecularmechanismsfromlower-levelinvestigatorsarealsocommon.Recall,forexample,thequotefromShepherdnear17thebeginningofsection14.1above.Thesearehardlytheattitudesonewouldexpectinanendeavorthatconsidersitselfthecuttingedgeofinterdisciplinaryscience.Theproblemisthateachdisciplinecomprisingcognitiveneuroscienceisdiffi-cult.Theendeavorcallsforacommunitywillingtoteachandlearntherelevantportionsofvoluminousdetailgatheredinindividualdisciplines.Researcherswill-ingtoconferwiththoseworkingatotherlevelsareanecessaryfirstcomponent,buteventuallycognitiveneuroscienceneedsresearcherstrainedinthemethodsandfactualdetailsofavarietyoflevels.Itneedstransdisciplinaryresearchers.Thisisadauntingjobdescription.Butitdoesofferhopeforphilosopherswant-ingtocontributetorealneuroscience,ratherthanjustreflectingonthediscip-line.Thinkerswithgraduatetraininginbothphilosophy’s“synopticvision”andneuroscience’sfactualandexperimentaldetailswouldbeequippedideallyforthistask.Thephilosophyprofessionhasbeenslowtorecognizethispotentialniche,butthereissomehopethatafewgraduateprograms,publishingcompanies,and18fundingagenciesaretakingstepstofillit.347\nJohnBickleNotesSpecialthankstoMaricaBernstein,whocreatedoradaptedthefiguresandcommentedonearlierdraftsofthispaper,andRobertRichardson,whosecommentsonthepenultimatedraftledtonumerousclarifications.1InBickle(1998:chs2and3),Iprovidesuchanassessmentandalternative.InlaterchaptersIextendthisgeneralaccounttospecialfeaturesofpsychoneuralreductions.Foranassessmentofthegeneraltheory,seeRichardson(1999).Foracriticalresponsetomyattempttodistinguish“newwave”from“classical”reductionism,seeEndicott(1998).Forsomeempiricalandconceptualargumentsagainstmyextensionofthegeneraltheorytopsychoneuralcases,seeSchoutenanddeJong(1999)andmydiscussioninsection14.2below.2Goodoverviewsofsynaptictransmissionareavailableinanypassableneurobiologyorphysiologicalpsychologytext.Shepherd(1994:chs6,7,and8)isparticularlygood.Forthosewholearnedtheirelementaryneurosciencetwentyyearsagoandhaven’tkeptup,however,beforewarned:thestoryhaschanged!Theimportanceofmetabotrophicreceptors,secondmessengers,retrogradetransmission,andthebio-chemicaleffectsongeneexpressioninbothpre-andpostsynapticneuronsyieldaverydifferentpictureofsynaptictransmissionandplasticity.I’llintroducesomeofthiscomplexityinthesubsequentdiscussion.3See,e.g.,tables29.1and29.3inShepherd(1994)foralistofhistoricalexperimentalsupport.Theselistsonlyincluderesultspriortothemid-1970s.Bothlistshavegrownconsiderablysincethen.4Iwillleaveagreatdealoftheknownbiochemistryoutofmydiscussion.See,e.g.,Shepherd(1994:ch.6)foragoodintroductiontothat.5Recently,ourunderstandingofthemoleculargeneticsandbiochemistryofLTPinductionhasincreaseddramatically.See,e.g.,SquireandKandel(1999:chs6and7)foragoodintroductiontosomeofthesenewdetails.(Incidentally,thisincludesworkforwhichEricKandelsharedthe2000NobelPrizeforMedicine.)6SeeKandeletal.(1991:ch.65)forthefullmoleculardetailsofthiscell-biological“letter.”SquireandKandel(1999:ch.3)includemorerecentdiscoveries.7InBickle(1998:ch.5,sec.2),Ishowhowfeaturesofthiscasemeetalltheconditionsonmygeneralaccountofinter-theoreticreductiondevelopedearlierinthatbook(inch.3).8InBickle(1998:ch.5,sec.2),Isketchanother:thereductionofacognitivetheoryofhierarchicallystructuredmemorystoragetothemechanismsofLTPinmammaliansensorycortex.ThekeyneuroscientificevidenceiselectrophysiologicalandcomputersimulationresultsbyneurobiologistGaryLynch,computerscientistRichardGranger,andtheircolleagues(Lynchetal.1988;Grangeretal.1989).9Forexample,SchoutenanddeJongclaimthat“Bickle’sideawasthatthereductiveapproachmustbeconductedinapurelybottom-upfashioninthesensethatitshunsreferencetohigher-levelfunctions”(1999:253;myemphases).10Ifahigher-leveltheorypostulatesentitiesorprocessesthatareintensionorareflat-outinconsistentwiththoseofavailablelower-leveltheories,thatissufficientreasontorejecttheformer.Butthis“constraint”ispartofgeneralscientificmethodology.We348\nPhilosophyofMindandtheNeurosciencesdon’tneedaspecial“coevolution”principletoruleouthigher-leveltheoriesofthissort(seeBickle1996;1998:ch.4).11Anautobiographicalnoteisinorderhere.Irememberasaphilosophygraduatestudentbeingfrustratedinneurosciencegraduateandlabseminarsbythetopicsthatdominateddiscussion.Wehadreadpapersfromthethen-currentneuroscientificliterature,filledwithrichtheoreticalideasandimplications–andspenttheseminarmeetingtalkingabout,e.g.,thefilmspeedinthecameraandthediameteroftheelectrodetips.WhatIdidn’trealizethenwashowmuchgraduatetraininginascienceisintheartofexperimentaldesign.Atalentfor“abstractcriticalreasoning”isnosubstituteforapprenticeshipwithagoodexperimenter.12Thesetwosciencesarethemselvesbecomingunifiedunderdevelopmentalbiology.Inlightofthesharedmolecularmechanismsofsynapticplasticityandneurondevelop-ment,EricKandelinvitedusto“conceiveoflearningas...alate...stageofneuronaldifferentiation”(1979:76).Thatwasaquarterofacenturyago,andsincethenourknowledgeofthesharedmolecularbasisoflearningandneuraldevelopmenthasincreased(see,e.g.,Shepherd1994:ch.9).Learningasalatestageofneurondifferentiation,espousedbyaleadingmainstreamneuroscientist:couldadisciplinebeanymore“maddog”reductionist?13Themodalityinthefinalclauseofthisquotationisdeceptive.Intheessay,Dretskeiscarefultopointoutthatheisurgingtheavailability,notthetruth,ofphenomenalexternalism.Lycan(1996)isabitbolder.14SeetheessaysinLlinásandChurchland(1996),especiallytheessaybyLlinásandParé(ch.1),forneurobiologicalevidenceforendogenesis.15Forthoseworriedthatthistalkofcausaleffectsofexplicitconsciousattentiononsingleneuronactivitybordersonthemysterious,becomforted.Aneuralexplanationoftheseeffectsisunderactivedevelopment.Seemybriefdiscussionfiveparagraphsbelow.16NotethatNewsome’s“bottom-up”methodologyalsodoesnot“eschewhigher-leveltheories”inthefashioncriticizedbySchoutenanddeJong(1999).(Seesection14.2above.)17SeealsothePrefaceandIntroductiontoKandeletal.(1991).Whiletheauthorsdon’tsingleouthigher-leveltheoristsforbeingignorantofadvancesincellularandmolecularneuroscience,itisclearfromcontentthattheyareaprincipaltarget.18ExamplesincludeWashingtonUniversity’s“Philosophy–Neuroscience–Psychology”program,OxfordUniversity’s“Philosophy,Psychology,Physiology”program,PatriciaChurchland’sMacArthurFoundation“Genius”grant,theMcDonnellProjectinPhilosophyandtheNeurosciencesawardedrecentlytoKathleenAkins,andKluwerAcademicPublisher’snewjournal,BrainandMind:ATransdisciplinaryJournalofNeuroscienceandNeurophilosophy.ReferencesAlbright,T.D.,Desimone,R.,andGtoss,C.G.(1984).“ColumnarOrganizationofDirectionallySelectiveCellsinVisualAreaMTofMacaques.”JournalofNeurophysiology,51:15–31.349\nJohnBickleBickle,J.(1995).“PsychoneuralReductionfortheGenuinelyCognitive:SomeAccom-plishedResults.”PhilosophicalPsychology,8(3):265–85.——(1996).“NewWavePsychoneuralReductionandtheMethodologicalCaveats.”Philo-sophyandPhenomenologicalResearch,56(1):57–78.——(1998).PsychoneuralReduction:TheNewWave.Cambridge,MA:MITPress.Chalmers,D.(1996).TheConsciousMind.Oxford:OxfordUniversityPress.Churchland,P.M.(1985).“SomeReductiveStrategiesinCognitiveNeurobiology.”ReprintedinNeurocomputationalPerspective.Cambridge,MA:MITPress(1989):77–110.Churchland,P.S.(1986).Neurophilosophy.Cambridge,MA:MITPress.DeAngelis,G.C.,Cumming,B.G.,andNewsome,W.T.(1998).“CorticalAreaMTandthePerceptionofStereoscopicDepth.”Nature,394:677–80.Dennett,D.(1978).Brainstorms.Montgomery,VT:BradfordBooks.Dickinson,A.(1980).ContemporaryAnimalLearningTheory.Cambridge:CambridgeUniversityPress.Dretske,F.(1996).“PhenomenalExternalism.”InE.Villaneuva(ed.),Perception.Atascadero,CA:Ridgeview:143–57.Endicott,R.(1994).“ConstructivalPlasticity.”PhilosophicalStudies,74(1):51–75.——(1998).“CollapseoftheNewWave.”JournalofPhilosophy,95(2):53–72.Fodor,J.A.(1975).TheLanguageofThought.NewYork:ThomasCrowell.Gazzaniga,M.(ed.)(1997).ConversationsintheCognitiveNeurosciences.Cambridge,MA:MITPress.Gilbert,C.,Ito,M.,Kupadia,M.,andWestheimer,G.(2000).“InteractionsBetweenAttention,ContextandLearninginPrimaryVisualCortex.”VisionResearch,40(10–20):1217–26.Granger,R.,Ambros-Ingerson,J.,andLynch,G.(1989).“DerivationofEncodingChar-acteristicsofLayerIICerebralCortex.”JournalofCognitiveNeurosciences,1:61–87.Hawkins,R.D.(1989).“ASimpleCircuitModelforHigher-orderFeaturesofClassicalConditioning.”InJ.H.ByrneandW.O.Berry(eds.),NeuralModelsofPlasticity:ExperimentalandTheoreticalApproaches.SanDiego:AcademicPress:74–93.Hawkins,R.D.andKandel,E.R.(1984).“IsThereaCell-biologicalAlphabetforSimpleFormsofLearning?”PsychologicalReview,91:375–91.Hebb,D.O.(1949).TheOrganizationofBehavior.NewYork:Wiley.Hooker,C.A.(1981).“TowardaGeneralTheoryofReduction.”PartI:“HistoricalandScientificSetting.”PartII:“IdentityinReduction.”PartIII:“Cross-categorialReduc-tion.”Dialogue,21:38–59,201–36,496–529.Horgan,T.(1993).“NonreductiveMaterialismandtheExplanatoryAutonomyofPsy-chology.”InS.WagnerandR.Werner(eds.),Naturalism:ACriticalAppraisal.NotreDame:UniversityofNotreDamePress.Jackson,F.(1982).“EpiphenomenalQualia.”PhilosophicalQuarterly,32:127–36.Kandel,E.R.(1979).“CellularInsightsintoBehaviorandLearning.”HarveyLectures,73:19–92.Kandel,E.R.,Schwartz,J.H.,andJessell,T.M.(eds.)(1991).PrinciplesofNeuralScience,3rdedn.NewYork:Elsevier.Kincaid,H.(1988).“SupervenienceandExplanation.”Synthese,77:251–81.Llinás,R.andChurchland,P.S.(eds.)(1996).TheMind–BrainContinuum.Cambridge,MA:MITPress.350\nPhilosophyofMindandtheNeurosciencesLycan,W.(1996).ConsciousnessandExperience.Cambridge,MA:MITPress.Lynch,G.,Granger,R.,Larson,J.,andBaudry,M.(1988).“CorticalEncodingofMemory:HypothesesDerivedfromAnalysisandSimulationofPhysiologicalLearningRulesinAnatomicalStructures.”InL.Nadel,L.Cooper,P.Culicover,andR.M.Harnish(eds.),NeuralConnections,MentalComputations.Cambridge,MA:MITPress:180–224.McAdams,C.J.andMaunsell,J.H.R.(1999).“EffectsofAttentiononOrientation-tuningFunctionsofSingleNeuronsinMacaqueCorticalAreaV4.”JournalofNeuro-science,19(1):431–41.McGinn,C.(1989).“CanWeSolvetheMind–BodyProblem?”Mind,98:349–66.Millikan,R.(1984).Language,Thought,andOtherBiologicalCategories.Cambridge,MA:MITPress.Murasugi,C.M.,Salzman,C.D.,andNewsome,W.T.(1993).“MicrostimulationinVisualAreaMT:EffectsOvaryingPulseAmplitudeandFrequency.”JournalofNeuro-science,13(4):1719–29.Nagel,T.(1989).TheViewfromNowhere.Oxford:OxfordUniversityPress.Penrose,R.(1994).ShadowsoftheMind.Oxford:OxfordUniversityPress.Post,J.F.(1991).Metaphysics:AContemporaryIntroduction.NewYork:ParagonHouse.Putnam,H.(1960).“MindsandMachines.”InS.Hook(ed.),DimensionsofMind.NewYork:Collier.Quine,W.V.O.(1949).“TwoDogmasofEmpiricism.”ReprintedinFromaLogicalPointofView.Cambridge,MA:HarvardUniversityPress.Reichenbach,H.(1957).TheRiseofScientificPhilosophy.Berkeley,CA:UniversityofCaliforniaPress.Rescorla,R.A.(1988).“PavlovianConditioning:It’sNotWhatYouThinkItIs.”Amer-icanPsychologist,43:151–60.Rescorla,R.A.andWagner,A.R.(1972).“ATheoryofPavlovianConditioning:Vari-ationsintheEffectivenessofReinforcementandNonreinforcement.”InA.H.BlackandW.F.Prokasy(eds.),ClassicalConditioningII:CurrentResearchandTheory.NewYork:Appleton-Century-Crofts:64–99.Richardson,R.C.(1999).“CognitiveScienceandNeuroscience:NewWaveReductionism.”PhilosophicalPsychology,12(3):297–308.Salzman,C.D.,Murasugi,C.M.,Britten,K.R.,andNewsome,W.T.(1992).“MicrostimulationinVisualAreaMT:EffectsonDirectionDiscriminationPerform-ance.”JournalofNeuroscience,12(6):2331–55.Schouten,M.anddeJong,H.L.(1999).“Reduction,Elimination,andLevels:TheCaseoftheLTP-learningLink.”PhilosophicalPsychology,12(3):237–62.Sellars,W.(1962).“PhilosophyandtheScientificImageofMan.”ReprintedinScience,Perception,andReality.London:RoutledgeandKeganPaul.Shapiro,J.A.(1999).“GenomeSystemArchitectureandNaturalGeneticEngineeringinEvolution.”InL.H.Caporale(ed.),MolecularStrategiesinBiologicalEvolution.NewYork:NewYorkAcademyofSciences:23–35.Shepherd,G.(1994).Neurobiology,3rdedn.Oxford:OxfordUniversityPress.Squire,L.andKandel,E.(1999).Memory:FromMindtoMolecules.NewYork:ScientificAmericanLibrary.351\nTheBlackwellGuidetoPhilosophyofMindEricT.OlsonEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter15PersonalIdentityEricT.Olson15.1TheProblemsofPersonalIdentityItishardtosaywhatpersonalidentityis.Discussionsthatgounderthatheadingaremostoftenaboutsomeofthefollowingquestions.WhoamI?Tomostpeople,thephrase“personalidentity”suggestswhatwemightcallone’sindividualidentity.Youridentityinthissenseconsistsroughlyofthoseattributesthatmakeyouuniqueasanindividualanddifferentfromothers.Oritisthewayyouseeordefineyourself,whichmaybedifferentfromthewayyoureallyare.Persistence.Whenpsychologiststalkaboutpersonalidentity,theyusuallymeanitinthe“WhoamI?”sense.Philosophersgenerallymeansomethingquitediffer-ent.Mostoftentheymeanwhatittakesforapersontopersistfromonetimetoanother–forthesamepersontoexistatdifferenttimes.Theyareaskingforourpersistenceconditions.Whatsortsofadventurecouldyoupossiblysurvive?Whatsortofthingwouldnecessarilybringyourexistencetoanend?Whatdetermineswhichfuturebeing,orwhichpastone,isyou?Youpointtoagirlinanoldphotographandsaythatsheisyou.Whatmakesyouthatone–ratherthan,say,oneoftheothers?Whatisitaboutthewaysherelatestoyouasyouarenowthatmakesheryou?Historically,thisquestionoftenarisesoutofthehopethatwemightcontinuetoexistafterwedie.Whetherthisisinanysensepossibledependsonwhetherbiologicaldeathisthesortofthingthatonecouldsurvive.Imaginethatafteryourdeaththerereallywillbesomeone,inthenextworldorinthisone,relatedtoyouincertainways.What,ifanything,wouldmakethatpersonyou–ratherthanme,say,oranewpersonwhodidn’texistbefore?Howwouldhehavetorelatetoyouasyouarenowinordertobeyou?Evidence.Howdowefindoutwhoiswho?Whatevidencedoweappealtoindecidingwhetherthepersonherenowistheonewhowashereyesterday?Whatoughtwetodowhendifferentkindsofevidencesupportopposingverdicts?One352\nPersonalIdentitysourceofevidenceismemory:ifyoucanrememberdoingsomething,oratleastseemtorememberit,itwasprobablyyouwhodidit.Anothersourceisphysicalcontinuity:ifthepersonwhodiditlooksjustlikeyou,or,evenbetter,ifsheisinsomesensephysicallyorspatio-temporallycontinuouswithyou,thatisreasontothinksheisyou.Inthe1950sand’60sphilosophersdebatedaboutwhichofthesecriteriaismorefundamental:whethermemorycanbetakenasevidenceofidentityallbyitself,forinstance,orwhetheritcountsasevidenceonlyinsofarasitcanbecheckedagainstthird-person,“bodily”evidence.ThisisnotthesameasthePersistenceQuestion,thoughthetwoaresometimesconfused.Whatittakesforyoutopersistthroughtimeisonething;howwefindoutwhetheryouhaveisanother.Ifthecriminalhadfingerprintsjustlikeyours,thecourtsmaycon-cludethatheisyou.Butevenifitisconclusiveevidence,havingyourfingerprintsisnotwhatitisforsomepastorfuturebeingtobeyou.Population.IfwethinkofthePersistenceQuestionashavingtodowithwhichofthecharactersintroducedatthebeginningofastoryhavesurvivedtobecomethecharactersattheendofit,wecanalsoaskhowmanycharactersareonthestageatanyonetime.Whatdetermineshowmanyofustherearenow,orwhereonepersonleavesoffandthenextonebegins?Youmaythinkthatthenumberofpeople(orpersons–Itakethesetermstobesynonymous)issimplythenumberofhumananimals–membersoftheprimatespeciesHomosapiens,perhapsdiscount-ingthoseinadefectivestatethatdon’tcountaspeople.Butthisisdisputed.Surgeonssometimescutthenervebandsconnectingone’scerebralhemispheres(commissurotomy),resultinginsuchpeculiarbehaviorassimultaneouslypullingone’strousersupwithonehandanddownwiththeother.Doesthisgiveustwopeople–twothinking,consciousbeings?(Seee.g.Nagel1971.Puccetti1973arguesthattherearetwopeoplewithintheskinofeverynormalhumanbeing.)Couldahumanbeingwithsplitpersonalityliterallybethehomeoftwo,orthree,orsevendifferentthinkingbeings(Wilkes1988:127f.;Olson2003)?Thisissometimescalledtheproblemof“synchronicidentity,”asopposedtothe“diachronicidentity”ofthePersistenceQuestion(andthe“counterfactualidentity”ofthe“HowcouldIhavebeen?”Questionbelow).Iavoidthesephrasesbecausetheysuggestthatidentitycomesintwokinds,synchronicanddiachronic,andinvitetheabsurdquestionofwhetherthisandthatmightbesynchronicallyidenticalbutdiachronicallydistinctorviceversa.Thereisonlyonerelationofnumericalidentity.Therearesimplytwokindsofsituationwherequestionsabouttheidentityanddiversityofpeopleandotherconcretethingsarise:synchronicsituationsinvolvingjustonetimeanddiachroniconesinvolvingseveraltimes.Personhood.Whatisittobeaperson?Whatfeaturesmakesomethingaperson,asopposedtoanon-person?Atwhatpointinyourdevelopmentfromafertilizedeggdidtherecometobeaperson?WhatwouldittakeforachimpanzeeoraMartianoranelectroniccomputertobeaperson,iftheycouldeverbe?Somephilosophersseemtothinkthatallquestionsaboutpersonalidentityreducetothisone.Whenweaskwhatittakesforapersontopersistthroughtime,orwhatdetermineswhetherwehaveonepersonortwoatanyonetime,353\nEricT.Olsontheysaythatweareinquiringintoourconceptofaperson(e.g.Perry1975:7ff.;Wilkes1988:viif.).Ithinkthisisamistake.Theusualdefinitionsof“person”tellusnothing,forinstance,aboutwhetherIshouldgoalongwithmybrainifthatorganweretransplanted.Suppose,asLockethought,thatapersonis“athink-ingintelligentbeing,thathasreasonandreflection,andcanconsideritselfasitself,thesamethinkingthing,indifferenttimesandplaces”(1975:335).Iamapersononthisaccount,andsoisthebeingthatwouldgetmytransplantedbrain.Butthatdoesn’ttelluswhetherheandIwouldbetwopeopleorone.Whatarewe?Whatsortofthings,metaphysicallyspeaking,areyouandIandotherhumanpeople?Arewematerialorimmaterial?Arewesubstances,attri-butes,events,orsomethingdifferentstill?Arewemadeofmatter,orofthoughtsandexperiences,orofnothingatall?Herearesomepossibleanswerstothisadmittedlyrathervaguequestion.Wearehumananimals.Surprisingly,mostphilosophers,bothpastandpresent,rejectthisanswer.Iwillsaymoreaboutitlater.Historically,themostcommonansweristhatwearepartless,immaterialsouls(or,alternatively,compoundthingsmadeupofanimmaterialsoulandamaterialbody:seeSwinburne1984).Humesaidthateachofusappearstobe“abundleorcollectionofdifferentperceptions,whichsucceedeachotherwithaninconceivablerapidity,andareinaperpetualfluxandmovement”(1888:252;seealsoQuinton1962;Rovane1998:212).Amoderndescendantofthisviewsaysthatyouareasortofcomputerprogram,awhollyabstractthingthatcouldinprinciplebestoredonmagnetictape(acommonideainsciencefiction).Perhapsthemostpopularviewnowadaysisthatwearematerialobjects“consti-tutedby”humananimals:youaremadeofthesamematterasacertainanimal,butyouandtheanimalaredifferentthingsbecausewhatittakesforyoutopersistisdifferent(Wiggins1967:48;Shoemaker1984:112–14;Baker2000).Thereiseventheparadoxicalviewthatwedon’treallyexistatall.Theexistenceofhumanpeopleisametaphysicalillusion.Parmenides,Spinoza,Hume,andHegel(asIreadthem),andmorerecentlyRussell(1985:50)andUnger(1979),alldeniedtheirownexistence.AndwefindtheviewinIndianBuddhism.Whatmatters?Whatisthepracticalimportanceoffactsaboutouridentityandpersistence?Imaginethatsurgeonsaregoingtoputyourbrainintomyhead.Willtheresultingperson(whowillthinkheisyou)beresponsibleformyactions,orforyours?Orboth?Orneither?Towhosebankaccountwillhebeentitled?Supposehewillbeinterriblepainaftertheoperationunlessoneofuspaysalargesuminadvance.Ifwewerebothentirelyselfish,whichofusoughttopay?YoumightthinkthattheanswertothesequestionsturnsentirelyonwhethertheresultingpersonwillbeyouorI.Onlyyoucanberesponsibleforyouractions.Theonlyonewhosefuturewelfareyoucan’tignoreisyourself.Youhaveaspecial,selfishinterestinyourownfuture,andnooneelse’s.Butmanyphilo-sophersdenythis.Theysaythatsomeoneelsecouldberesponsibleforyouractions.Youcouldhaveaselfishreasontocareaboutsomeoneelse’swell-being.Icare,oroughtrationallytocare,aboutwhathappenstoOlsontomorrownotbecauseheisme,butbecauseheis“psychologicallycontinuous”withme,orrelatesto354\nPersonalIdentitymeinsomeotherwaythatdoesn’timplynumericalidentity.Ifsomeoneelsewerepsychologicallycontinuouswithmetomorrow,Ioughttotransfermyselfishcon-cerntohim.(SeeShoemaker1970:284;Parfit1971,1984:215;Martin1998.)HowcouldIhavebeen?HowdifferentcouldIhavebeenfromthewayIactuallyam?WhichofmypropertiesdoIhaveessentially,andwhichonlyaccid-entallyorcontingently?Forinstance,couldIhavehaddifferentparents?Thatis,couldsomeonebornofdifferentparentshavebeenme,orwouldithavetohavebeensomeoneelse?CouldI–thisveryphilosopher–haveceasedtoexistinthewombbeforeIacquiredanymentalfeatures?Aretherepossibleworldsjustliketheactualoneexceptforwhoiswho–wherepeoplehave“changedplaces”sothatwhatisinfactyourcareerismycareerandviceversa?Whetherthesearebestdescribedasquestionsaboutpersonalidentityisdebatable.(Theycertainlyaren’taboutwhetherbeingsinotherworldsareidenticalwithpeopleintheactualworld:seevanInwagen1985.)Buttheyareoftendiscussedinconnectionwiththeothers.Thatcompletesoursurvey.Thesequestionsarealldifferent,andshouldbekeptapart.IwishIcouldsaywhatcommonfeaturemakesthesequestions,andthemalone,problemsofpersonalidentity.ButasfarasIcanseethereisnone,apartfromthename.Thereisnooneproblemofpersonalidentity,butonlyanumberoflooselyrelatedproblems.IwillfocusinthischapteronthePersistenceQuestion–notbecauseitisthemostimportant(ifany,thatisthe“Whatarewe?”Question),butbecauseithasdominatedthephilosophicaldebateonpersonalidentitysinceLocke.ButIwilltouchonseveraloftheothers.15.2UnderstandingthePersistenceQuestionIdentityandchangearenotoriouslyhardtopics,andevenexpertsoftengetthePersistenceQuestionwrong.WehavealreadymentionedthetendencytoconflateitwiththeEvidenceQuestion.Herearetwofurthercaveats.First,itisaboutnumericalidentity.Tosaythatthisandthatarenumericallyidenticalistosaythattheyareonething,ratherthantwo.Ifwepointtoyounow,andthenpointtoordescribesomeoneorsomethingthatexistsatanothertime–acertainagedman,say–thequestioniswhetherwearepointingtoonethingtwice,orpointingoncetoeachoftwothings.Youarenumericallyidenticalwithacertainfuturebeinginthatapictureofhimtakenthenandapictureofyoutakennowwouldbetwopicturesofonething.Numericalidentityisn’tthesameasqualitativeidentity.Thingsarequalitativelyidenticalwhentheyareexactlysimilar.Apastorfuturepersonneedn’tbeexactlylikeyouarenowinordertobeyou–thatis,tobenumericallyidenticalwithyou.Youdon’tremainqualitativelythesamethroughoutyourlife:youchangeinsize,appearance,andinmanyotherways.Nordoessomeone’sbeingexactlylikeyou355\nEricT.Olsonarenowguaranteethatsheisyou.Somewhereintheuniversesomeoneelsemaybejustlikeyouarenow,downtothelastatomandquirkofpersonality.None-theless,youandshewouldn’tbeoneandthesame.(Youwouldn’tbeintwoplacesatonce.)Twopeople,ortwocatsortwotoasters,couldbequalitativelyidentical.Nothingcanchangeitsnumericalidentity.Wesometimessaythingslike“IfIlostallmymemories,Iwouldn’tbemeanylonger,”or“Iwouldn’tbethesameperson,”oreven“Iwouldbesomeoneelse.”Iftheseclaimswereaboutnumer-icalidentity,theywouldbeself-contradictory.Nothingcanliterallybeonethingatonetimeandanother,numericallydifferentthinglateron.IfIsaythatafteracertainadventureIshallbeadifferentperson,orthatIamnotthepersonIoncewas,Imustmeanthatthatfutureorpastpersonisnumericallyidenticalwithmebutqualitativelydifferentinsomeimportantway.Otherwiseitwouldn’tbeIbutsomeoneelsewhowasthatwaythen.Peoplewhosaythesethingsareusuallytalkingaboutsomeone’sindividualidentity,inthe“WhoamI?”sense.PerhapsIcouldcontinuetoexistwithoutbeingthesamepersonasIamnowbycastingoffmycurrentidentityandacquiringanother–thatis,bychangingmycharacterorthewayIseemyself.Ishouldbelikeasenatorwho,onbeingelectedpresident,isnolongerthesameelectedofficialassheoncewas,havingexchangedherfirstelectedofficeforanother.Inbothcaseswehavenumericallythesamebeingthroughout.Itisunfortunatethatthewords“identity”and“same”areusedtomeansomanydifferentthings:numericalidentity,qualitativeidentity,individualpsycho-logicalidentity,andmore.Tomakemattersworse,somephilosophersspeakof“surviving,”and“survivingas”or“becoming”someone,inawaythatdoesn’timplynumericalidentity,sothatIcould“survive”acertainadventureeventhoughIwon’texistafterwards.Confusionisinevitable.WhenIaskwhetheryouwouldsurvivesomething,Imeanwhetheryouwouldexistbothbeforeandafterit.Hereisadifferentmisunderstanding.ThePersistenceQuestionisalmostalwaysstatedlikethis:(1)Underwhatpossiblecircumstancesisapersonexistingatonetimeidenticalwith(orthesamepersonas)apersonexistingatanothertime?Wehaveapersonexistingatonetime,andapersonexistingatanothertime,andthequestioniswhatisnecessaryandsufficientfor“them”tobeonepersonratherthantwo.Thisisthewrongquestiontoask.Wemaywanttoknowwhetheryouwereeveranembryoorafetus,orwhetheryoucouldsurvivethecompletedestructionofyourmentalfeaturesasahumanvegetable.Theseareclearlyquestionsaboutwhatittakesforustopersist,andanyaccountofouridentitythroughtimeoughttoanswerthem.(Theiranswersmayhaveimportantethicalimplications.)However,mostanswerstothePersonhoodQuestion–Locke’sanswerquotedearlier,forinstance–agreethatyoucan’tbeapersonwithouthavingcertainmentalfeatures.Andtheexpertssaythatearly-termfetusesandhumanbeingsinapersistentvegetativestatehavenomentalfeatures.Ifso,theyaren’tstrictlypeople.Thus,if356\nPersonalIdentitythePersistenceQuestionwerewhatittakesforapastorfuturepersontobeyou,someonewhoaskedwhetherwewereeverfetusesorcouldcometobevegetableswouldn’tbeaskingaboutouridentitythroughtime.Butobviouslyshewouldbe.Atypicalanswertoquestion(1)illustratesthetrouble:“Necessarily,apersonwhoexistsatonetimeisidenticalwithapersonwhoexistsatanothertimeifandonlyiftheformerpersoncan,attheformertime,rememberanexperiencethelatterpersonhadatthelattertime,orviceversa.”WemightcallthistheLockeanView,thoughitprobablyisn’tquitewhatLockebelieved.Itsaysthatapastorfuturepersonisyoujustincaseyoucannowrememberanexperienceshehadthen,orshecanthenrememberanexperienceyouarehavingnow.Itisn’tveryplausible,butnevermind.Thepointisthis.TheLockeanViewmightseemtoruleoutyourbecomingavegetable,sinceavegetablecan’trememberanything.Thatis,itmightseemtoimplythatifyouweretolapseintoapersistentvegetativestate,theresultingvegetablewouldn’tbeyou.Youwouldhaveeitherceasedtoexistorpassedontothenextworld.ButinfacttheLockeanViewimpliesnosuchthing.Thatisbecausewedon’thavehereapersonexistingatonetimeandapersonexistingatanothertime(assumingthatahumanvegetableisn’taperson).TheLockeanViewtellsuswhichpastorfuturepersonyouare,butnotwhichpastorfuturething.Ittellsuswhatittakesforonetopersistasaperson,butnotwhatittakesforonetopersistwithoutqualification.Soitsimplydoesn’tapplyhere.Forthesamereasonitsaysnothingaboutwhetheryouwereeveranembryo(Olson1997:22–6;Mackie1999:224–8).Soquestion(1)istoonarrow.Insteadweoughttoask:(2)Underwhatpossiblecircumstancesisapersonwhoexistsatonetimeidenticalwithsomethingthatexistsatanothertime(whetherornotitisapersonthen)?Why,then,dosomanyphilosophersask(1)ratherthan(2)?Becausetheyassumethateverypersonisapersonessentially:nothingthatisinfactapersoncouldpossiblyexistwithoutbeingaperson.(Bycontrast,somethingthatisinfactastudentcouldexistwithoutbeingastudent:nostudentisessentiallyastudent.)Ifthatistrue,thenwhateverisapersonatonetimemustbeapersonateveryothertimewhensheexists.Thisassumptionmakesquestions(1)and(2)equival-ent.Whetheritistrue,though,isaseriousissue(aninstanceofthe“HowcouldIhavebeen?”Question).Ifyouareapersonessentially,youcouldn’tpossiblyhavebeenanembryo,orcometobeavegetable(ifsuchthingsaren’tpeople).Theembryothatgaverisetoyouisn’tnumericallyidenticalwithyou.Youcameintoexistenceonlywhenitdevelopedcertainmentalcapacities.Theassumptionalsorulesoutourbeinganimals,fornoanimalisessentiallyaperson:everyhumananimalstartedoutasanunthinkingembryo,andmayendupasanunthinkingvegetable.Whetherweareanimalsorwereonceembryosarequestionsthatanaccountofpersonalidentityoughttoanswer,andnotmatterswecansettleinadvanceby357\nEricT.Olsonthewayweframetheissues.Sowehadbetternotassumeattheoutsetthatwearepeopleessentially.Askingquestion(1)prejudgestheissuebyfavoringsomeaccountsofwhatweareandwhatittakesforustopersistoverothers.(Inparticular,asking(1)effectivelyrulesouttheSomaticApproachdescribedinthenextsection.)Itislikeaskingwhichmancommittedthecrimebeforerulingoutthepossibilitythatitmighthavebeenawoman.15.3AccountsofOurIdentityThroughTimeTherearethreemainsortsofanswertothePersistenceQuestion.Thefirstsaysthatsomepsychologicalrelationiseithernecessaryorsufficient(orboth)foronetopersist.Youarethatfuturebeingthatinsomesenseinheritsitsmentalfeatures–personality,beliefs,memories,andsoon–fromyou.Youarethatpastbeingwhosementalfeaturesyouhaveinherited.IwillcallthisthePsychologicalApproach.MostphilosopherswritingonpersonalidentitysinceLockehaveendorsedsomeversionofit.TheLockeanViewisatypicalexample.Anotheransweristhatouridentitythroughtimeconsistsinsomebrutephys-icalrelation.Youarethatpastorfuturebeingthathasyourbody,orthatisthesameanimalasyouare,orthelike.Whetheryousurviveorperishhasnothingtodowithpsychologicalfacts.IwillcallthistheSomaticApproach.Itiscompara-tivelyunpopular,thoughIwilllaterdefendit.Youmaythinkthatthetruthliessomewherebetweenthetwo:weneedbothmentalandphysicalcontinuitytosurvive;orperhapseitherwouldsufficewithouttheother.ViewsofthissortareusuallyversionsofthePsychologicalApproach.Hereisatestcase:yourcerebrum–theupperbrainthoughttobechieflyresponsibleforyourmentalfeatures–istransplantedintomyhead.(Thisisphysicallypossible,thoughitwouldbeadelicatebusinessinpractice.)Twobeingsresult:thepersonwhoendsupwithyourcerebrumandyourmentalfeatures,andtheempty-headedbeingleftbehind,whichmaystillbealivebutwillhavenomentalfeatures.Ifpsychologicalfactsareatallrelevanttoourpersist-ence,youwillbetheonewhogetsyourcerebrum.Ifyouwouldbetheempty-headedvegetable,youridentityconsistsinsomethingnon-psychological.BoththePsychologicalandSomaticApproachesagreethatthereissomethingthatittakesforustopersist–thatouridentitythroughtimeconsistsinornecessarilyfollowsfromsomethingotherthanitself.Athirdviewdeniesthis.Mentalandphysicalcontinuityareevidenceforidentity,butdon’tguaranteeit,andaren’trequired.Nosortofcontinuityisabsolutelynecessaryorabsolutelysufficientforyoutosurvive.TheonlycorrectanswertothePersistenceQuestionisthatapersonherenowisidenticalwithapastorfuturebeingifandonlyiftheyareidentical.Therearenoinformative,non-trivialpersistenceconditionsforpeople.ThisissometimescalledtheSimpleView(Chisholm1976:108ff.;Swinburne1984;Lowe1996:41ff.;Merricks1998).Itisoftencombinedwith358\nPersonalIdentitytheviewthatweareimmaterialorhavenoparts,thoughitneedn’tbe.(Hybridviewsarealsopossible:mentalorphysicalcontinuitymaybenecessaryorsuffi-cientforsurvival,evenifnothingisbothnecessaryandsufficient.)TheSimpleViewispoorlyunderstood,anddeservesmoreattentionthanithasreceived.However,Imustpassoverit.AnotherviewIwillmentionandthenignoreisthatwedon’tpersistatall.Nopastorfuturebeingcouldeverbenumericallyidenticalwithyou.Strictlyspeaking,youaren’tthepersonwhobeganreadingthissentenceamomentago(Hume1888:253;Sider1996).Thisispresumablybecausenothing,oratleastnochangingthing,canexistattwodifferenttimes.15.4ThePsychologicalApproachThePsychologicalApproachmayappeartofollowtriviallyfromtheveryideaofaperson–fromtheanswertothePersonhoodQuestion(Baker2000:124).Nearlyeveryonewouldagreethattobeapersonisatleastinparttohavecertainmentalfeatures.Peoplearebydefinitionpsychologicalbeings.Mustn’ttheythere-forehavepsychologicalpersistenceconditions?Attheveryleast,can’tweruleoutaperson’ssurvivingthecompletelossofallhermentalfeatures?Mustn’tapersonwholosesallhermentalfeaturesnotmerelyceasetobeaperson,butceasetobealtogether?Thatwouldmakesomepsychologicalrelationnecessaryforapersontopersist.Butmattersaren’tsosimple.Consideraparallelargument.Tobeateenagerisbydefinitiontohaveacertainage.Mustn’tteenagersthereforehaveage-relatedpersistenceconditions?Attheveryleast,can’tweruleoutateenager’ssurvivingthelossofherteen-age?Clearlynot.Ioffermyselfaslivingproofthatonecansurviveone’s20thbirthday.Theparallelargumentreliesonthemistakenassump-tionthateveryteenagerisessentiallyateenager,oratleastthatonceyou’reateenageryoucan’tceasetobeonewithoutperishing.Theoriginalargumentreliesontheanalogousassumptionthateverypersonisessentiallyaperson,oratleastthatceasingtobeapersonmeansceasingtobe.Aswesawearlier,thatassumptionisfarfromobvious.SothePsychologicalApproachisn’tobviouslytrue,andmustbearguedfor.Themostcommonargumentsarebasedontheideathatyouwouldgoalongwithyourbrainorcerebrumifitweretransplantedintoadifferenthead,andthatthisissobecausethatorgancarrieswithityourmemoriesandothermentalfeatures.ButitisnotoriouslydifficulttogetfromthisintuitivebelieftoaspecificanswertothePersistenceQuestionthathasanyplausibility.Wemustfirstsaywhatmentalrelationouridentitythroughtimeistoconsistin.TheLockeanViewofsection15.2appealstomemory:apastorfuturebeingisyoujustincaseyoucannowrememberanexperienceshehadthenorviceversa.Thisfacestwowell-knownproblems,discoveredintheeighteenthcenturybyReidandButler(seetheexcerptsinPerry1975).359\nEricT.OlsonFirst,supposeayoungstudentisfinedforoverduelibrarybooks.Asamiddle-agedlawyer,sherememberspayingthefine.Stilllater,inherdotage,sheremem-bersherlawcareer,buthasentirelyforgottenpayingthefine,andeverythingelseshedidinheryouth.TheLockeanViewimpliesthattheyoungstudentisthemiddle-agedlawyer,thatthelawyeristheoldwoman,butthattheoldwomanisn’ttheyoungstudent:animpossibleresult.Ifxandyareoneandyandzareone,xandzcan’tbetwo.Identity,asthelogicianssay,istransitive,andLockeanmemorycontinuityisn’t.Secondly,itseemstobelongtotheveryideaofrememberinganexperiencethatyoucanrememberonlyyourownexperiences.Torememberpayingafine(or,ifyoulike,theexperienceofpaying)istorememberyourselfpaying.Thatmakestheclaimthatyouarethepersonwhoseexperiencesyoucanremembertrivialanduninformative(thoughitdoesn’taffecttheclaimthatmemoryconnec-tionsarenecessaryforidentity).Youcan’tknowwhethersomeonegenuinelyremembersapastexperiencewithoutalreadyknowingwhetherheistheonewhohadit.Weshouldhavetoknowwhowaswhobeforeapplyingthetheorythatissupposedtotelluswhoiswho.Oneresponsetothefirstproblemistoswitchfromdirecttoindirectmemoryconnections:theoldwomanistheyoungstudentbecauseshecanrecallexperi-encesthelawyerhadatatimewhenshe(thelawyer)rememberedthestudent’slife.Thesecondproblemistraditionallymetbyinventinganewconcept,“retrocognition”or“quasi-memory,”whichisjustlikememorybutwithouttheidentityrequirement(Penelhum1970:85ff.;Shoemaker1970).Thisinventionhasbeencriticized,thoughnot,Ithink,inawaythatmattershere(McDowell1997).Butneithersolutiongetsusveryfar,fortheLockeanViewfacestheobviousproblemthattherearemanytimesinmypastthatIcan’trememberatall,evenindirectly.Ican’tnowrecallanythingthathappenedtomewhileIwasasleeplastnight.Butifweknowanything,weknowthatwedon’tstopexistingwhenwefallasleep.Thebestwayforwardistoexplainmentalcontinuityintermsofcausaldepend-ence(Shoemaker1984:89ff.).Abeingatalatertimeispsychologicallyconnectedwithsomeonewhoexistsatanearliertimejustincasethelaterbeinghasthepsychologicalfeaturesshehasatthelatertimeinlargepartbecausetheearlierbeinghadthepsychologicalfeaturesshehadattheearliertime.IinheritedmycurrentloveofphilosophyfromayoungmancalledOlsonwhocametoloveitmanyyearsago:atypicalpsychologicalconnection.Andyouarepsychologicallycontinuouswithsomepastorfuturebeingifyourcurrentmentalfeaturesrelatetothoseshehasthenbyachainofpsychologicalconnections.Thenwecansaythatapersonwhoexistsatonetimeisidenticalwithsomethingexistingatanothertimejustincasetheformeris,attheformertime,psychologicallycon-tinuouswiththelatterassheisatthelattertime.Thisstillleavesimportantquestionsunanswered.Suppose,forinstance,thatwecouldelectronicallycopythementalcontentsofyourbrainontomine,therebyerasingthepreviouscontentsofbothbrains.Theresultingbeingwouldbe360\nPersonalIdentitymentallyverymuchlikeyouwereamomentbefore.Whetherthiswouldbeacaseofmentalcontinuitydependsonwhatsortofcausaldependenceisrelevant.Theresultingpersonwouldhaveinheritedyourmentalpropertiesinaway,butnotintheusualway.Isittherightway,sothatyoucouldliterallymovefromonehumananimaltoanothervia“brain-statetransfer”?AdvocatesofthePsycholo-gicalApproachdisagree(Unger1990:67–71;Shoemaker1997).15.5TheFissionProblemWhatevermentalcontinuitycomesdowntointheend,afarmoreseriousworryforthePsychologicalApproachisthatyoucouldapparentlybementallycontinu-ouswithtwopastorfuturepeople.Ifyourcerebrumweretransplanted,theresultingbeingwouldbementallycontinuouswithyou,andso,onthePsycho-logicalApproach,wouldbeyou.Nowthecerebrumhastwohemispheres,andifoneofthemisdestroyedtheresultingbeingisalsomentallycontinuouswiththeoriginalperson.HerethePsychologicalApproachagreeswithreal-lifejudgments:hemispherectomy(eventheremovalofthelefthemisphere,whichcontrolsspeech)isconsideredadrasticbutacceptabletreatmentforotherwise-inoperablebraintumors,andnotaformofmurder(Rigterink1980).Noonewhohasactuallyconfrontedsuchacasedoubtswhethertheresultingbeingistheoriginalperson.SothePsychologicalApproachimpliesthatifwedestroyedoneofyourcerebralhemispheresandtransplantedtheother,youwouldbetheonewhogotthetransplantedhemisphere.Butnowletthesurgeonstransplantbothhemispheres,eachintoadifferentemptyhead.CalltheresultingpeopleLeftyandRighty.Bothwillbementallycontinuouswithyou.Ifyouareidenticalwithanyfuturebeingwhoismentallycontinuouswithyou,itfollowsthatyouareLeftyandyouareRighty.ThatimpliesthatLeftyisRighty:twothingscan’tbenumericallyidenticalwithonething.ButLeftyandRightyareclearlytwo.Soyoucan’tbeidenticalwithboth.Wecanmakethesamepointinanotherway.SupposeLeftyishungryatatimewhenRightyisn’t.IfyouareLefty,youarehungry.IfyouareRighty,youaren’t.IfyouareLeftyandyouareRighty,youarebothhungryandnothungryatonce,whichisimpossible.ShortofgivingupthePsychologicalApproachaltogether,therewouldseemtobejusttwowaysofavoidingthiscontradiction.Oneistosaythat,despiteappear-ances,“you”werereallytwopeopleallalong–apositionwhimsicallycalledthedouble-occupancyview(Lewis1976;Noonan1989:122–48;Perry1972offersamorecomplicatedvariant).Therearetwodifferentbutexactlysimilarpeopleinthesameplaceandmadeofthesamematteratonce,doingthesamethingsandthinkingthesamethoughts.Thesurgeonsmerelyseparatethem.Thisisimplaus-ibleforanumberofreasons,notleastbecauseitmeansthatwecan’tknowhowmanypeopletherearenowuntilweknowwhathappenslater.(Theviewisusually361\nEricT.Olsoncombinedwith“four-dimensionalism,”thecontroversialmetaphysicalthesisthatallpersistingobjectsareextendedintimeandmadeupoftemporalparts.)Theotherwayoutistogiveuptheclaimthatmentalcontinuitybyitselfissufficientforyoutopersist.Youareidenticalwithapastorfuturebeingwhoismentallycontinuouswithyouasyouarenowonlyifnooneelseisthenmentallycontinuouswithyou:the“non-branchingview”(Wiggins1967:55;Shoemaker1984:85;Unger1990:265;Garrett1998).NeitherLeftynorRightyisyou.Ifbothyourcerebralhemispheresaretransplanted,thatistheendofyou–thoughyouwouldsurviveifonlyoneweretransplantedandtheotherdestroyed.Thistooishardtobelieve.Ifyoucouldsurvivewithhalfyourbrain,howcouldpreservingtheotherhalfmeanthatyoudon’tsurvive?(SeeNoonan1989:14–18,149–68.)Forthatmatter,youwouldperishifoneofyourhemispheresweretransplantedandtheotherleftinplace(thoughNozick’s1981variantwouldavoidthis).Andif“brain-statetransfer”givesusmentalcontinuity,youwouldceasetoexistifyourtotalbrainstatewerecopiedontoanotherbrainwithouterasingyours.Hereisanotherconsideration.Facedwiththeprospectofhavingoneofyourhemispherestransplanted,therewouldseemtobenoreasontopreferthattheotherbedestroyed.Onthecontrary:wouldn’tyouratherhavebothpreserved,eveniftheygointodifferentheads?Yetonthenon-branchingview,thatistopreferdeathovercontinuedexistence.ThisiswhatleadsParfitandotherstosaythatyoudon’treallywanttocontinueexisting.Insofarasyouarerational,anyway,youonlywanttheretobesomeonementallycontinuouswithyouinthefuture,whetherornotheisstrictlyyou.Moregenerally,factsaboutwhoisidenticalwithwhomhavenopracticalimportance.ButthenwehavetowonderwhetherwehadanyreasontoacceptthePsychologicalApproachinthefirstplace.Supposeyouwouldcareaboutthewelfareofyourtwofissionoffshootsinjustthewaythatyouordinarilycareaboutyourownwelfare,eventhoughneitherofthemwouldbeyou.Thenthefactthatyouwouldcareaboutwhathappenedtothepersonwhogotyourwholebrainintheoriginaltransplantcasedoesn’tsuggestthathewouldbeyou.Itissometimessaidthatfissionisn’taproblemforthePsychologicalApproachperse,butafflictsallanswerstothePersistenceQuestionapartfromtheSimpleView.IamnotpersuadedthatitarisesfortheversionoftheSomaticApproachthatsaysthatweareanimals(seesection15.7).Idoubtwhetheranythingthatcouldhappentoahumananimalwouldproducetwohumananimals,eitherofwhichweshouldbehappytoidentifywiththeoriginalwereitnotfortheexistenceoftheother.ButIcan’targueforthatclaimhere.15.6TheProblemoftheThinkingAnimalThePsychologicalApproachfacesasecondproblemthathasnothingtodowithfission.Itarisesbecausethatviewimpliesthatwearen’thumananimals.Nosort362\nPersonalIdentityofmentalcontinuityiseithernecessaryorsufficientforahumananimaltopersist.(Carter1989;Ayers1990:278–92;Snowdon1990;Olson1997:80f.,100–9.McDowell1997:237andWiggins1980:160,180apparentlydisagree.)Notnecessary:everyhumananimalstartsoutasanembryo,andmayendupinapersistentvegetativestate.Neitheranembryonorahumanvegetablehasanymentalfeaturesatall,andsoneitherismentallycontinuouswithanything.Soahumananimalcanpersistwithoutanysortofmentalcontinuity.Ifyouneedmentalcontinuitytopersist,youaren’tahumananimal.Notsufficient:ifyourcerebrumweretransplantedintoanotherhead,thentheonewhogotthatorgan,andnooneelse,wouldbementallycontinuouswithyouasyouwerebeforetheoperation.Butthesurgeonswouldn’ttherebymoveanyhumananimalfromoneheadtoanother.Theywouldsimplymoveanorganfromoneanimaltoanother.(Theempty-headedthingleftbehindwouldstillbeananimal,whileadetachedcerebrumisnomoreananimalthanafreshlyseveredarmisananimal.)Nomentalcontinuityofanysortsufficesforahumananimaltopersist.Ifitsufficesforyoutopersist,thenagainyouaren’tahumananimal.NoadvocateofthePsychologicalApproachdeniesthatyourelateinanintim-atewaytoacertainhumananimal–theoneyouseewhenyoulookinamirror.Andhumananimalscanthinkandhaveexperiences.Theimmatureandthebrain-damagedmaybeexceptions,butcertainlythosewithmaturenervoussystemsingoodworkingordercanthink.Sothereisathinkinghumananimalnowlocatedwhereyouare.Butsurelyyouarethethinkingthinglocatedwhereyouare.Itfollowsthatyouarethatanimal.Andsincetheanimalhasnon-psychologicalpersistenceconditions,thatcontradictsthePsychologicalApproach.Callthistheproblemofthethinkinganimal.Theproblemwouldn’tariseifthehumananimalassociatedwithyouwereunabletothink.Butthatisimplausible.Ithasahealthyhumanbrainingoodworkingorder.Itevenhasthesamesurroundingsandevolutionaryhistoryasyouhave.Whatcouldpreventitfromthinking?If“your”animalcan’tthink,thatmustbebecausenoanimalofanysortcouldeverthink.Strictlyspeaking,animalsmustbenomoreintelligentthantrees.Thatsuggeststhatthinkingthingsmustbeimmaterial:ifanymaterialthingcouldthink,itwouldbeananimal.ButfewfriendsofthePsychologicalApproachsaythatweareimmaterial.Anyonewhodeniesthatanimalscanthink,yetinsiststhatwe(whocanthink)arematerial,hadbetterhaveanexplanationforthisastonishingclaim.Shoemakerproposesthatanimalscan’tthinkbecausetheyhavethewrongpersistenceconditions(1984:92–7;1997;1999).Thenatureofmentalpropertiesentailsthatmentalcontinu-itymustsufficefortheirbearerstopersistthroughtime.Materialthingswiththerightpersistenceconditions,however,canthink.Buthehasfoundfewfollowers(Noûs2002).Ontheotherhand,ifhumananimalscanthink,butyouandIaren’tanimals,thenthereareatleasttwothinkingthingswhereverwethoughttherewasjustone.Thischapterwasco-writtenbyananimalandanon-animalphilosopher.IoughttowonderwhichoneIam.ImaythinkI’mthenon-animal.Buttheanimalhas363\nEricT.Olsonthesamereasonsforthinkingthatitisthenon-animalasIhaveforthinkingthatIam,yetismistaken.SohowdoIknowthatI’mnottheonemakingthemistake?IfIweretheanimal,I’dstillthinkIwasthenon-animal.SoevenifIamsomethingotherthanananimal,itishardtoseehowIcouldeverknowit.Forthatmatter,if“my”animalcanthink,itpresumablyhasthesamementalfeaturesasIhave.(Otherwiseweshouldexpectanexplanationforthedifference.)Thatoughttomakeitaperson.Peoplewouldthencomeintwokinds:animalpeopleandnon-animalones.Animalpeoplewouldhavenon-psychologicalpersistenceconditions.ButthePsychologicalApproachclaimedthatallpeoplepersistbyvirtueofmentalcontinuity.Alternatively,ifhumananimalsaren’tpeople,thenatmosthalfoftherational,intelligent,self-conscious,morallyre-sponsiblebeingswalkingtheeartharepeople.Beingaperson,perse,wouldhavenopracticalsignificance.Andwecouldneverknowwhetherwearepeople.Thatconflictswithmostaccountsofwhatitistobeaperson.Noonanproposesalinguistichypothesistosolvesomeoftheseproblems(1989:75f.;1998:316).First,notjustanyrational,self-consciousbeingisaperson,butonlyonewithpsychologicalpersistenceconditions.Sohumananimalsdon’tcountaspeople.Secondly,personalpronounssuchas“I”(andnamessuchas“Socrates”)alwaysrefertopeople.Thus,whentheanimalassociatedwithyousays“I,”itdoesn’trefertoitself.Rather,itreferstoyou,“its”person.Whenitsays“Iamaperson,”itisn’tsayingfalselythatitisaperson,buttrulythatyouare.Sotheanimalisn’tmistakenaboutwhichthingitis,andneitherareyou.Youcaninferthatyouareapersonfromthelinguisticfactsthatyouarewhateveryourefertowhenyousay“I,”andthat“I”alwaysreferstoaperson.Youcanknowthatyouaren’tananimalbecausepeoplebydefinitionhavepersistenceconditionsdifferentfromthoseofanimals.ThisproposalfacesdifficultiesthatIcan’tgointohere.Inanycase,itstillleavesuswithanuncomfortablesurplusofthinkingbeings,andmakespersonhoodatrivialproperty.Ofcourse,anotherwayroundtheproblemofthethinkinganimalistoacceptthatweareanimals,andgiveupthePsychologicalApproach.15.7TheSomaticApproachThePsychologicalApproachisattractivebecausewhenweimaginecaseswherementalandphysicalcontinuitycomeapart,itiseasytothinkthatwegoalongwiththeformer.Butanequallyattractiveideaisthatweareanimals.Thatiscertainlywhatweappeartobe.Whenyouseeyourselforanotherperson,youseeahumananimal.Andaswehaveseen,theapparentfactthathumananimalscanthinkprovidesastrongargumentforourbeinganimals.Ifweareanimals,though,thenwehavethepersistenceconditionsofanimals.Andanimalsappeartopersistthroughtimebyvirtueofsomesortofbrutephysicalcontinuity.Thus,themostnaturalaccountofwhatweareleadstotheSomaticApproach.364\nPersonalIdentityAfewphilosophersendorsetheSomaticApproachwithoutsayingthatweareanimals.Theysaythatweareourbodies(Thomson1997),orthatouridentitythroughtimeconsistsintheidentityofourbodies(Ayer1936:194).Theseareversionsoftheso-calledBodilyCriterionofpersonalidentity.Itisunclearhowtheyrelatetotheviewthatweareanimals.Itisoftensaidthatsomeonecouldhaveapartlyorwhollyinorganicbody.Butnoanimalcouldbepartlyorwhollyinorganic.Ifyoucutoffananimal’slimbandreplaceitwithaninorganicprosthesis,theanimalonlygetssmallerandhasaninorganicprosthesisattachedtoit(Olson1997:135).Ifthisisright,thenyoucouldbeidenticalwithyourbodywithoutbeingananimal.Somephilosopherssaythatananimal’sbodyisalwaysadifferentthingfromtheanimalitself:ananimalceasestoexistwhenitdies,butunlessitsdeathisparticularlyviolentitsbodycontinuestoexistasacorpse;orananimalcanhavedifferentbodiesatdifferenttimes(Campbell1994:166).Ifso,thennoonecouldbebothananimalandidenticalwithhisbody.ButIwon’tenterintothesecontroversies.IfindtheBodilyCriterionhardtounderstandbecauseitisuncleartomewhatitisforsomethingtobesomeone’sbody(vanInwagen1980;Olson1997:142–53).Ibelievethatthephrase“humanbody”or“one’sbody”isresponsibleformuchphilosophicalconfusion,andisbetteravoided.Inanycase,theviewthatweareanimalsistheclearestandmostplausibleversionoftheSomaticApproach,andIwilldevotetherestofthischaptertoit.Ourbeinganimalsdoesn’timplythatallpeopleareanimals.Itisconsistentwiththeexistenceofwhollyinorganicpeople:gods,angels,orrobots.Theclaimisthatwehumanpeopleareanimals.(AhumanpersonissomeonewhorelatestoahumananimalasyouandIdo:ifyouinsist,someonewithahumanbody.)Nordoesitimplythatallanimalsorevenallhumananimalsarepeople.Humanembryosandhumanbeingsinapersistentvegetativestatearehumanorganisms,butwemaynotwanttocallthempeople.Infacttheviewimpliesnothingaboutwhatitistobeaperson.Thus,theSomaticApproachgivespersistenceconditionsforsomepeoplebutnotforothers:forusbutnotforgodsorangels,ifsuchtherebe.Anditassignstosomenon-peoplethesamepersistenceconditionsitassignstosomepeople:humananimalssharetheirpersistenceconditionswithdogs.Thisleadssometoobjectthatitisn’taviewofpersonalidentityatall(Baker2000:124;seealsoLowe1989:115).Thereissometruthinthiscomplaint.TheSomaticApproachdoesn’tpurporttogivethepersistenceconditionsofallandonlypeople,orofpeopleassuch.Itevenimpliesthatweareonlytemporarilyandcontingentlypeople(ontheusualdefinitionsofthatterm).Butwhyisthatanobjection?Ifsomepeopleareanimals,thentherearenopersistenceconditionsthatnecessarilyapplytoallandonlypeople,anymorethantherearepersistenceconditionsthatnecessarilyapplytoallandonlystudentsorteenagers.Thatdoesn’tmeanthatbeingapersonisnomoreimportantapropertythanbeingastudent.Itmeansonlythatathing’sbeingapersonhasnothingmoretodowithitsidentitythroughtimethanitsbeingastudenthas.AndtheSomaticApproachisanaccountofpersonalidentityinthesenseofsayingwhatittakesforsomepeople365\nEricT.Olsontopersist,namelyourselves,andinthesenseofbeingincompetitionwithotherviews,suchasthePsychologicalApproach,whichgiveaccountsofpersonaliden-titystrictlysocalled.Othersobjecttotheideathatwearemerelyanimals.Surelywe’remorethanjustanimals?Butwhyshouldourbeinganimalsimplythatweare“merely”animals?Descarteswasaphilosopher,butnotmerelyaphilosopher:hewasalsoamathematicianandaFrenchman.Whycouldn’tsomethingbeaperson,agrandmother,asocialist,andmanyotherthings,aswellasananimal?Although“animal”canbeatermofabuse(itisn’tnicetocallsomeoneananimal),ourbeinganimalsinthemostliteralzoologicalsenseneedn’timplythatwearebrutish,orthatwearenodifferentfromotheranimals,orthatwehaveonly“animal”properties.Weareveryspecialanimals.Butweareanimalsallthesame.ItseemsclearthatourbeinganimalsisinconsistentwiththePsychologicalApproach:animalsdon’tpersistbyvirtueofmentalcontinuity.Whatitdoestakeforananimaltopersistislessclear.Alivingorganismissomethingwithalife:acomplexbiologicaleventthatmaintainsanorganism’sstructuredespitewholesalematerialturnover.ThisleadsLockeandotherstosaythatanorganismpersistsjustaslongasitslifecontinues(Locke1975:330f.;vanInwagen1990:142–58;Olson1997:131–40;Wilson1999:89–99).Thishasthesurprisingconsequencethatanorganismceasestoexistwhenitdiesandcannotberevived.Strictlyspeaking,thereisnosuchthingasadeadanimal;atanyratenothingcanbefirstalivinganimalandthenadeadanddecayingone.Othersarguethatalivinganimalcancontinuetoexistasacorpseafteritdies(Feldman1992:89–105;Carter1999;Mackie1999).AsIseeit,livingorganismsandcorpsesareprofoundlydifferent.Alivingthing,likeafountain,existsbyconstantlyassimilatingnewmatter,imposingitscharacteristicformonit,andexpellingtheremains(Miller1978:140f.).Acorpse,likeamarblestatue,maintainsitsformmerelybyvirtueoftheinherentstabilityofitsmaterials.Thechangesthattakeplacewhenanorganismdiesarefarmoredramaticthananythingthathappenssubsequentlytoitslifelessremains.Ihaveneverseenaplausibleaccountofwhatittakesforananimaltopersistthatallowedforalivinganimaltocontinuetoexistasadecayingcorpse.Butthesearedifficultmatters.15.8ConclusionIbelievethatthePsychologicalApproachowesmuchofitspopularitytothefactthatphilosopherstypicallybegintheirinquiriesintopersonalidentitybyaskingwhatittakesforustopersistthroughtime.(Aswesawinsection15.2,anotherfactoristhewaythisquestionisoftenput.)Butanequallyimportantquestioniswhatweare:whetherweareanimals,whatwemightbeifwearen’tanimals,andhowwerelatetothoseanimalsthatsomecallourbodies,forinstance.Thisquestion366\nPersonalIdentityisoftenignored,oraddressedonlyasanafterthought.Thatiswhyphilosophershavefailedtoappreciatetheproblemofthethinkinganimal.Perhapstheyoughtinsteadtobeginbyaskingwhatweare,andonlythenturntoouridentitythroughtimeandothermatters.Manywouldendupthinkingdifferently.ReferencesIntroductorydiscussionsaremarkedwithanasterisk.Ayer,A.J.(1936).Language,Truth,andLogic.London:Gollancz.Ayers,M.(1990).Locke,vol.2.London:Routledge.Baker,L.R.(2000).PersonsandBodies.Cambridge:CambridgeUniversityPress.Campbell,J.(1994).Past,Space,andSelf.Cambridge,MA:MITPress.Carter,W.R.(1989).“HowtoChangeYourMind.”CanadianJournalofPhilosophy,19:1–14.——(1999).“WillIBeaDeadPerson?”PhilosophyandPhenomenologicalResearch,59:167–72.Chisholm,R.(1976).PersonandObject.LaSalle,IL:OpenCourt.Feldman,F.(1992).ConfrontationswiththeReaper.NewYork:OxfordUniversityPress.*Garrett,B.(1998).“PersonalIdentity.”InE.Craig(ed.),TheRoutledgeEncyclopediaofPhilosophy.London:Routledge.Hume,D.(1888).TreatiseonHumanNature,ed.L.A.Selby-Bigge.Oxford:ClarendonPress.Originalwork1739.PartlyreprintedinPerry(1975).Lewis,D.(1976).“SurvivalandIdentity.”InA.Rorty(ed.),TheIdentitiesofPersons.Berkeley:California.ReprintedinhisPhilosophicalPapers,vol.I.NewYork:OxfordUniversityPress(1983).Locke,J.(1975).AnEssayConcerningHumanUnderstanding,ed.P.Nidditch.Oxford:ClarendonPress.Originalwork,2ndedn.,firstpublished1694.PartlyreprintedinPerry(1975).Lowe,E.J.(1989).KindsofBeing.Oxford:Blackwell.——(1996).SubjectsofExperience.Cambridge:CambridgeUniversityPress.Mackie,D.(1999).“PersonalIdentityandDeadPeople.”PhilosophicalStudies,95:219–42.Martin,R.(1998).SelfConcern.Cambridge:CambridgeUniversityPress.McDowell,J.(1997).“ReductionismandtheFirstPerson.”InJ.Dancy(ed.),ReadingParfit.Oxford:Blackwell.Merricks,T.(1998).“ThereAreNoCriteriaofIdentityOverTime.”Noûs,32:106–24.Miller,J.(1978).TheBodyinQuestion.NewYork:RandomHouse.Nagel,T.(1971).“BrainBisectionandtheUnityofConsciousness.”Synthèse,22:396–413.ReprintedinPerry(1975)andinNagel,MortalQuestions.Cambridge:CambridgeUniversityPress(1979).*Noonan,H.(1989).PersonalIdentity.London:Routledge.——(1998).“AnimalismVersusLockeanism:ACurrentControversy.”PhilosophicalQuar-terly,48:302–18.Nozick,R.(1981).PhilosophicalExplanations.Cambridge,MA:HarvardUniversityPress.Olson,E.(1997).TheHumanAnimal.NewYork:OxfordUniversityPress.——(2002).“WhatDoesFunctionalismTellUsAboutPersonalIdentity?”Noûs,36:682–98.367\nEricT.Olson——(2003).“WasJekyllHyde?”PhilosophyandPhenomenologicalResearch.——(forthcoming).“ThereIsNoBodilyCriterionofPersonalIdentity.”InF.MacBrideandC.Wright(eds),IdentityandModality.Oxford:OxfordUniversityPress.Parfit,D.(1971).“PersonalIdentity.”PhilosophicalReview,80:3–27.ReprintedinPerry(1975).——(1984).ReasonsandPersons.Oxford:OxfordUniversityPress.Penelhum,T.(1970).SurvivalandDisembodiedExistence.London:Routledge.Perry,J.(1972).“CantheSelfDivide?”JournalofPhilosophy,69:463–88.——(ed.)(1975).PersonalIdentity.Berkeley:UniversityofCaliforniaPress.Puccetti,R.(1973).“BrainBisectionandPersonalIdentity.”BritishJournalforthePhilo-sophyofScience,24:339–55.Quinton,A.(1962).“TheSoul.”JournalofPhilosophy,59:393–403.ReprintedinPerry(1975).Rigterink,R.(1980).“PuccettiandBrainBisection:AnAttemptatMentalDivision.”CanadianJournalofPhilosophy,10:429–52.Rovane,C.(1998).TheBoundsofAgency.Princeton:PrincetonUniversityPress.Russell,B.(1918).“ThePhilosophyofLogicalAtomism.”Monist,28:495–527;Monist,29:32–63,190–222.ReprintedinR.Marsh(ed.),LogicandKnowledge.London:AllenandUnwin(1956),andinD.Pears(ed.),ThePhilosophyofLogicalAtomism.LaSalle,IL:OpenCourt(1985);pagenumbersfromthelatter.Shoemaker,S.(1970).“PersonsandTheirPasts.”AmericanPhilosophicalQuarterly,7:269–85.*——(1984).“PersonalIdentity:AMaterialist’sAccount.”InS.ShoemakerandR.Swinburne,PersonalIdentity.Oxford:Blackwell.——(1997).“SelfandSubstance.”PhilosophicalPerspectives,11:283–319.——(1999).“Self,Body,andCoincidence.”ProceedingsoftheAristotelianSociety,Sup-plementaryvolume73:287–306.Sider,T.(1996).“AlltheWorld’saStage.”AustralasianJournalofPhilosophy:433–53.Snowdon,P.(1990).“Persons,Animals,andOurselves.”InChristopherGill(ed.),ThePersonandtheHumanMind.Oxford:ClarendonPress.*Swinburne,R.(1984).“PersonalIdentity:TheDualistTheory.”InS.ShoemakerandR.Swinburne,PersonalIdentity.Oxford:Blackwell.Thomson,J.J.(1997).“PeopleandTheirBodies.”InJ.Dancy(ed.),ReadingParfit.Oxford:Blackwell.Unger,P.(1979).“IdonotExist.”InG.F.MacDonald(ed.),PerceptionandIdentity.London:Macmillan.(ReprintedinM.Rea(ed.),MaterialConstitution,Lanham,MD:RowmanandLittlefield.)——(1990).Identity,Consciousness,andValue.NewYork:OxfordUniversityPress.VanInwagen,P.(1980).“PhilosophersandtheWords‘HumanBody’.”InvanInwagen(ed.),TimeandCause.Dordrecht:Reidel.——(1985).“PlantingaonTrans-WorldIdentity.”InJ.TomberlinandP.vanInwagen(eds.),AlvinPlantinga.Dordrecht:Reidel.——(1990).MaterialBeings.Ithaca:CornellUniversityPress.Wiggins,D.(1967).IdentityandSpatio-TemporalContinuity.Oxford:Blackwell.——(1980).SamenessandSubstance.Oxford:Blackwell.Wilkes,K.(1988).RealPeople.Oxford:ClarendonPress.Wilson,J.(1999).BiologicalIndividuality.Cambridge:CambridgeUniversityPress.368\nTheBlackwellGuidetoPhilosophyofMindFreedomoftheWillEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdChapter16FreedomoftheWillRandolphClarke1Wecommonlythinkthatwearefreeinmakingdecisionsandacting,andforseveralreasonsitisimportanttousthatwehavethisfreedom.Decidingoractingfreelyishavingavaluablevarietyofcontroloverwhatonedoes,thepossessionofwhich,wethink,ispartlyconstitutiveofhumandignity.Itiswidelythoughtthatonlywhenanagenthassuchcontroloverwhatshedoesareherdecisionsandotheractionsattributabletoherinsuchawaythatshemaybemorallyrespons-ibleforwhatshedoes,deservingofpraiseorblame,rewardorpunishment,dependingonthemoralqualitiesofherdecisionsandotheractions.Moreover,wewantittobethecasethatbyfreeexercisesofcontrol,wearemakingadifferencetowhathappensintheworld,includingwhatkindsofpersonwebecome.Andwhenwedeliberate,itgenerallyseemstousthatmorethanoneoptionisopentousandwearefreetopursueeachofthealternativesweareconsidering;ifthisimpressionissystematicallymistaken,weareroutinelysubjecttoanundesirableillusion.Doweinfacthavethefreedomthatwevalueintheserespects?Somehave2thoughtthatwedonotbecause,theyhold,ourworldisdeterministic.(Theworldisdeterministicifthelawsofnaturearesuchthathowtheworldisatanygivenpointintimefullynecessitateshowitisatanylaterpoint;weshalllookmorecloselyatdeterminismbelow.)Theviewthattherecanbenofreewillinadeterministicworldisknownasincompatibilism.Whilesomeincompatibilistsaffirmdeterminism,others,calledlibertarians,denydeterminismandaffirmfreewill.Anditisworthnotingthatanincompatibilistmightholdthattheworldisnotdeterministicandstillwedonothavefreewill.Manyphilosophersrejectincompatibilisminfavorofcompatibilism,theviewthatfreewillcanexisteveninadeterministicworld.Althoughsomecompatibilistsbelievethatourworldisdeterministic,othersholdthatitisnotorremainuncommittedonwhetheritis.369\nRandolphClarke16.1TheCompatibilityQuestionIsfreewillcompatiblewithdeterminism?Beforeaddressingthisquestion,weneedtoseewhatdeterminismis.Wemayunderstandittobeafeaturethattheworldmighthaveorlack,orasthethesisthatourworldactuallyhasthatfeature.3Understoodeitherway,whatisthefeatureinquestion?16.1.1DeterminismSometimesdeterminismissaidtoconsistinthefactthateveryeventhasacause.Butthisisnotright.Asweshallseebelow,theremaybenon-deterministiccausation;itmaybethatsomeeventsarecausedbutnotdetermined.Inthatcase,itmaybethateveryeventhasacauseandyettheworldisnot4deterministic.Determinismcanbewellcharacterizedintermsofhowitispossibleforworldstobeiftheyhavethesamelawsofnatureandarealikeatsomepointintime.Insuchterms,ourworldisdeterministic(inbothtemporaldirections)justincaseanypossibleworldthathasexactlythesamelawsofnatureasoursandthatisexactlylikeoursatanyonepointintimeisexactlylikeourworldateverypointintime.Aslightlymorelimited,future-directeddeterminismholdsinourworldjustincaseanypossibleworldthathasexactlythesamelawsofnatureasoursandthatisexactlylikeoursatanygivenpointintimeisexactlylikeourworldateverylaterpointintime.Ifhowtheworldisatanygivenpointintimecanbecompletelydescribedbyaproposition,andif,likewise,thelawsofnaturecanbecompletelystated,thenwemayofferequivalentcharacterizationsofdeterminismintermsofpropositionsand(broadly)logicalnecessity(truthineverypossibleworld).Thefuture-directedvarietyofdeterminismholdsinourworldjustincase,foranyproposi-tionpthatcompletelydescribeshowtheworldisatanypointintime,anytruepropositionqabout(evenpartof)howtheworldisatsomelaterpointintime,andanypropositionlthatcompletelystatesthelawsofnature,itislogicallynecessarythatif(pandl)thenq.Inthesymbolismofmodallogic,wewritethisstatementas:[(p&l)⊃q].16.1.2TheConsequenceArgumentInrecentyears,themostwidelydiscussedargumentsinsupportoftheviewthatfreewillisincompatiblewithdeterminismhavebeenversionsofwhatiscalledtheConsequenceArgument.Inthecontextofthisargument,weshalltakeitthatanagentactswithfreewilljustincaseshehasachoiceaboutwhethersheperformsthataction,orjustincaseitisuptoherwhatshedoes,orjustincasesheisable370\nFreedomoftheWilltodootherwisethanperformthataction.Informally,theargumentmaybestatedasfollows:Ifdeterminismistrue,thenouractsaretheconsequencesofthelawsofnatureandeventsintheremotepast.Butitisnotuptouswhatwentonbeforewewereborn,andneitherisituptouswhatthelawsofnatureare.Therefore,theconsequences5ofthesethings(includingourpresentacts)arenotuptous.Therearevariouswaysofmakingthisinformalargumentmoreprecise.Weshall6focushereononethathasreceivedagreatdealofattention.Thisversionoftheargumentemploysamodaloperator“N”which,whenattachedtoanysentencep,givesusasentencethatsaysthatpandnohumanagenthasoreverhadanychoiceaboutwhetherp.(Asitissometimesput,“Np”saysthatitispowernecessaryforallhumanagentsatalltimesthatp.)Forexample,where“P”abbreviatesasentenceexpressingthepropositionthattheEarthrevolvesaroundtheSun,“NP”saysthattheEarthrevolvesaroundtheSun,andnohumanagenthasoreverhadanychoiceaboutwhethertheEarthrevolvesaroundtheSun.Theargumentreliesonthefollowingtwoinferencerulesinvolvingpowernecessity:(α)pNp(β)N(p⊃q),NpNq.Rule(α)saysthatthepremisethatitis(broadly)logicallynecessarythatpentailsthatitispowernecessarythatp;ifitislogicallynecessarythatp,itfollowsthatpandnohumanagenthasoreverhadanychoiceaboutwhetherp.Rule(β)saysthatthetwopremisesthatitispowernecessarythatp⊃qandthatitispowernecessarythatpentailthatitispowernecessarythatq;ifp⊃qandnohumanagenthasoreverhadanychoiceaboutwhetherp⊃q,andifpandnohumanagenthasoreverhadanychoiceaboutwhetherp,thenitfollowsthatqandnohumanagenthasoreverhadanychoiceaboutwhetherq.Nowlet“H”abbreviateasentenceexpressingapropositionthatcompletelydescribeshowtheworldwasatsomepointintimepriortotheexistenceofanyhumanagents.Let“L”abbreviateasentenceexpressingapropositionthatcom-pletelystatesthelawsofnature.Andlet“A”standinforanysentenceexpressingatruepropositionabouthowtheworldisatsomepointintimelaterthanthatcoveredby“H”(e.g.,“A”maysaythatClarkeagreestowritethisessay).Nowsupposethatdeterminism(ofeithervariety)obtainsinourworld.Givenourearliercharacterizationsofdeterminism,itfollowsfromthissuppositionthat(1)[(H&L)⊃A].(1)islogicallyequivalentto371\nRandolphClarke(2)[H⊃(L⊃A)].Byanapplicationofrule(α)toline(2),weget(3)N[H⊃(L⊃A)].Nowtheargumentassertsasapremise(4)NH.Then,byanapplicationofrule(β)tolines(3)and(4),weget(5)N(L⊃A).Nowtheargumentassertsasasecondpremise(6)NL.Then,byanapplicationofrule(β)tolines(5)and(6),wegettheconclusion(7)NA.Theargument,ifsound,showsthatiftheworldisdeterministic,then,giventhatinfactClarkeagreestowritethischapter,ClarkesoagreesandnohumanagenthasoreverhadanychoiceaboutwhetherClarkesoagrees.(Thiswouldbenewstome,sinceIthinkIhadachoiceinthematter!)Andsince“A”maybereplacedwithanysentenceexpressingatruthabouthowtheworldisatanytimelaterthanthatcoveredby“H,”thesamewillgoforanyactionperformedbyanyhumanagent;iftheargumentissound,thenitshowsthatiftheworldisdeterministic,nohumanagenthasoreverhadanychoiceaboutwhetheranysuchactionisperformedbyanysuchagent.Determinism,theargumentpurportstoshow,altogetherprecludesfreewill,ourhavingachoiceaboutwhatwedo.16.1.3AssessingtheargumentTheargumentreliesontwopremises.Thefirst,line(4),says,roughly,thatwehavenochoiceaboutwhathappenedinthedistantpast(beforeanyofusexisted).Thesecond,line(6),says,againroughly,thatwehavenochoiceabout7whatthelawsofnatureare.Bothpremisesstrikemanyasevidentlytrue.But,dependingonhow“havingachoice”aboutsomethingisconstrued,thedenialofoneoranotherofthesepremisesmaybelessincrediblethanitfirstappears.Givenourcharacterizationsofdeterminism,iftheworldisdeterministic,thenifanyhumanagenthaddonesomethingthatshedidnotinfactdo,eitherthe372\nFreedomoftheWillworldwouldhavebeendifferentateveryearlierpointintime(andhence“H”wouldhavebeenfalse)orthelawsofnaturewouldhavebeendifferent(andhence“L”wouldhavebeenfalse).“Multiple-PastsCompatibilists”optforthefirstdisjunct,andtheyclaimthat,iftheworldisdeterministic,thenweareabletodothingssuchthat,werewetodothem,thepast(ateverypointintime)would8havebeendifferent.Inthissense,theyaccept,wemaybesaidtohaveachoiceaboutthedistantpast.Buttheydistinguishthisclaimfromastrongeronetowhichtheyarenotcommitted,viz.,thatweareabletodothingsthatwouldcausallyaffectthepast.Oncewedistinguishthesetwoclaimsandseeclearlytheonetowhichtheyarecommitted,theysuggest,theairofincredibilityabouttheirpositionshoulddissipate.“Local-MiracleCompatibilists”optforthesecondofthedisjunctsidentifiedabove,andtheyclaimthat,iftheworldisdeterministic,thenweareabletodo9thingssuchthat,werewetodothem,thelawsofnaturewouldbedifferent.(Ifanagenthaddonesomethingthatshedidnotinfactdo,theysay,thealternativeactionwouldhavebeenprecededbysomelaw-breakingevent(somemiracle)allowingforitsoccurrence.Themiraculouseventwouldhavebeenaviolationofsomeactuallawofnature,butnotofanylawofitsworld.Thatworldincludesthemiraculousevent,butotherwiseitspastresemblesours,andhenceitslawsdifferfromtheactuallaws.)Inthissense,thesecompatibilistsaccept,wemaybesaidtohaveachoiceaboutwhatthelawsofnatureare.Buttheydistinguishthisclaimfromastrongeronetowhichtheyarenotcommitted,viz.,thatweareabletoperformactionsthateitherwouldbeorwouldcauselaw-breakingevents.Whilethestrongerclaimmaybeincredible,theweakerclaimtowhichtheyarecommittedissaidtobemerelycontroversial.Manyfindeventheclaimstowhichthesecompatibilistsarecommittedincred-10ible.Inanycase,thepremisesoftheargumentremainpointsofcontention.Therehasbeenconsiderabledisagreementaswellabouttheinferencerule(β)on11whichtheargumentdepends.Whether(β)isavalidinferenceruledependsonhowtheoperator“N”isinterpreted,whichdependsinturnonhow“havingachoice”isconstrued.Supposethatweunderstand“havingachoice”alongthelinessuggestedbymultiple-pastsandlocal-miraclecompatibilists.Wewillsay,then,thatanagenthasachoiceaboutwhetherpjustincasesheisabletoperformsomeactionsuchthat,wereshetoperformthataction,itwouldnotbethecasethatp.“Np,”then,saysthatpandnohumanagentisableatanytimetoperformanyactionsuchthat,wereshetoperformthataction,itwouldnotbethecasethatp.Asithappens,thereareexamplesshowingthat(β),with“N”sointerpreted,isinvalid.12Hereisonesuchexample.Supposethatthereexistsjustonehumanagent,Sam.Samhasabitofradium,asubstancethatsometimesemitssubatomicparticles;whetherornotagivenbitofitemitsaparticleataparticulartimeisundetermined.Samdestroysthisbitofradiumbeforetimet,therebyensuringthattheradiumdoesnotemitaparticleatt,andthisistheonlywaythatSamcanensurethis.Samhasachoiceaboutwhetherhedestroystheradiumatt;heis373\nRandolphClarkeabletorefrainfromdoingso.Let“R”saythattheradiumdoesnotemitaparticleatt;let“S”saythatSamdestroystheradiumbeforet.Thenwehavethefollowinginstanceof(β):(1)N(R⊃S),(2)NR,therefore(3)NS.Theconditional“R⊃S”istrueifboththeantecedentandtheconsequentaretrue,andgiventheexamplebotharetrue.ThereissomethingthatSamcandosuchthat,werehetodoit,thisconditionalwouldbefalsejustincasethereissomethinghecandosuchthat,werehetodoit,R&~S.ButthereisnothingthatSamcandothatwouldensurethatR&~S.(Hecanrefrainfromdestroyingtheradium,butifhedoesso,theradiummightemitaparticleatt.)Hence,thefirstlineofthisinstanceof(β)istrue.Thesecondlineaswellistrue.Theradiumdoesnotemitaparticleatt.Andsinceitsemissionofparticlesisundetermined,Samcannotdoanythingthatwouldensurethattheradiumemitsaparticleatt.(Hecanrefrainfromdestroy-ingtheradium,butifhedoesso,itstillmightnotemitaparticleatt.)However,aswesupposed,Samisabletorefrainfromdestroyingtheradiumbeforet.Hencetheconclusion,line(3),isfalse.Thus,onthecurrentinterpreta-tionof“N,”wehaveacounterexampletorule(β),aninstanceofitinwhichthepremisesaretruebuttheconclusionisfalse.Rule(β),with“N”sointerpreted,isthusinvalid.Defendersoftheargumentforincompatibilismmightrespondbyofferinga13differentinferenceruletoreplace(β),orbyproposingadifferentinterpretationof“N”in(β).Alongthelatterlines,itiseasytoseethatasmallmodificationofourearlierconstrualof“N”willsufficetoleave(β)immunefromthepresentcounterexample.Letussaythatanagenthasachoiceaboutwhetherpjustincasesheisabletoperformsomeactionsuchthat,wereshetoperformthataction,itmightnotbethecasethatp.“Np”willnowsaythatpandnohumanagentisableatanytimetoperformanyactionsuchthat,wereshetoperform14thataction,itmightnotbethecasethatp.With“N”sointerpreted,bothpremisesoftheinstanceof(β)willbefalseintheradiumexample.(Samisabletorefrainfromdestroyingtheradiumbeforet;werehetodoso,itmightbethecasethatR&~S,andwerehetodoso,itmightbethecasethat~R.)Hencetheradiumexampleisnocounterexampleto(β)with“N”sointerpreted,nordoesitappearthattherecanbeanyothers.Withthisconstrualof“N,”(β)appearstobe15avalidinferencerule.Theargumentforincompatibilismthatwehaveconsideredisquitestrong.Withtheinterpretationof“N”suggestedinthepreviousparagraph,theinferencerule(β)onwhichtheargumentreliesappearsvalid.Thepremisesoftheargu-mentarequiteplausibleaswell,though,aswehaveseen,thereremainssomeroomtodoubtthatoneoranotherofthemistrue.374\nFreedomoftheWill16.2CompatibilistAccountsAlthoughsomecompatibilistsmaintainthat,eveniftheworldisdeterministic,itisgenerallythecasewhenweactthatwecouldhavedoneotherwise,othercompatibilistsallowthatdeterminismmayprecludesuchanability.Recallthatinvaluingfreewill,weareinterestedinatypeofcontrolthatwebelievetobeconnectedtoseveralthings:humandignity,moralresponsibility,makingadiffer-ence,andtheopennessofalternatives.Somecompatibilistsholdthatwemayhaveavarietyofcontrolthatsufficesforsomeofthesethingsbutnotforothers.Inparticular,someholdtheviewthatiftheworldisdeterministic,wemayalwayslacktheabilitytodootherwise,butweneverthelessgenerallyactwiththetypeofcontrolthatsufficesformoralresponsibility(andthatisthuspartlyconstitutiveofhumandignity).Thenamegiventothisviewbysomeofitsproponentsis16“semicompatibilism.”16.2.1FrankfurtcasesSemicompatibilistsrejectaviewconcerningresponsibilitythathaslongbeenwidelyheld,aviewthatwemayexpressinthefollowing“principleofalternatepossibilities”:(PAP)Anagentismorallyresponsibleforwhatshehasdoneonlyifshecouldhavedoneotherwise.SomeexamplespresentedbyHarryG.Frankfurt(1969)havebeenmostrespons-ibleforleadingmany,compatibilistsandincompatibilistsalike,torejectPAP.Frankfurtnotedthatanagentmightactincircumstancesthatconstitutesuffi-cientconditionsforherperformingacertainaction,andthatthusmakeitimposs-ibleforhertoactotherwise,butthatdonotactuallyproduceheraction.Whenanagentactsinsuchcircumstances,heargued,thefactthatshecouldnothavedoneotherwisedoesnotexcuseherfromresponsibility.HereisoneofthecasesthatFrankfurtofferedtoillustratetheseclaims:Supposesomeone–Black,letussay–wantsJonestoperformacertainaction.Blackispreparedtogotoconsiderablelengthstogethisway,buthepreferstoavoidshowinghishandunnecessarily.SohewaitsuntilJonesisabouttomakeuphismindwhattodo,andhedoesnothingunlessitiscleartohim(Blackisanexcellentjudgeofsuchthings)thatJonesisgoingtodecidetodosomethingotherthanwhathewantshimtodo.IfitdoesbecomeclearthatJonesisgoingtodecidetodosomethingelse,BlacktakeseffectivestepstoensurethatJonesdecidestodo,andthathedoesdo,whathewantshimtodo.WhateverJones’initialpreferencesandinclinations,then,Blackwillhavehisway....375\nRandolphClarkeNowsupposethatBlackneverhastoshowhishandbecauseJones,forreasonsofhisown,decidestoperformanddoesperformtheveryactionBlackwantshimtoperform.(1969:835–6)Here,Frankfurtclaimed,wehaveacaseinwhichconditionsobtain–Black’spresenceandhisreadinesstointervene–thatrenderitimpossibleforJonestodoanythingotherthanwhatheactuallydoes.ButJonesisunawareoftheseconditions,andtheyneverinfluenceintheleasthisdecisionoraction;JonesdecidesandactsjustashewouldhaveifBlackhadbeenabsent.Wewouldnot,andshouldnot,excuseJonesfromresponsibilityforwhathedoesinthisinstanceonthegroundsthathecouldnothaveactedotherwise.HencePAPisfalse.17DiscussionofthecaseagainstPAPhasbeenextensive.Here,givenlimitationsofspace,letussimplynoteacoupleofpointsconcerningthesignificanceofFrankfurt’sargument.First,itwouldnotfollow,justfromthefalsehoodofPAP,thatresponsibilityiscompatiblewithdeterminism.Fordeterminismmightpre-cluderesponsibilityevenifitdoesnotdosobyprecludingtheabilitytodo18otherwise.Butsecondly,ifPAPisfalse,theninevaluatingcompatibilistaccounts,weneedtobealerttowhattheypurporttobeaccountsof.Infact,mostrecentlyadvancedcompatibilistaccountsoffreedomofactionareputforwardasaccountsofwhat,withrespecttocontrol,isrequiredformoralresponsibil-ity.IfFrankfurtisright,thentheseaccountscannotbeshowntobemistakenjustbyshowing(ifitcanbeshown)thatdeterminismprecludestheabilitytodootherwise.16.2.2AhierarchicalaccountLetusturntosomeofthemostprominentrecentcompatibilistaccounts.Frank-furthimselfhasadvancedaviewemployingtheideaofahierarchyofattitudes,a19notionthathasbeenutilizedbyseveralothercompatibilistsaswell.Persons,Frankfurtpointsout,arecapablenotonlyofdesiringtoperform(ornottoperform)certainactions–ofhavingwhathecallsfirst-orderdesires–butalsoofreflectinguponandcriticallyevaluatingourownfirst-orderdesires.Givensuchreflectiveself-evaluation,wearecapableofformingsecond-orevenhigher-orderdesires,suchasdesirestohave(ornottohave)certainlower-orderdesires.Ofspecialinterestamongthesehigher-orderattitudesarewhatFrankfurtcallssecond-ordervolitions,desiresthatcertainfirst-orderdesiresbe(ornotbe)theonesthatmoveonetoact.Whenanagentwithconflictingfirst-orderdesiresformsasecond-ordervolitionthatacertainoneofthembetheonethatmoveshertoact,shemaythereby“identifyherself”withthatdesire.Afirst-orderdesirethateffectivelymovesanagenttoactionFrankfurtcallstheagent’swill.Frankfurtdistinguishesbetweenhavingafreewillandactingfreely.Aperson’swillisfree,onhisview,onlyif,376\nFreedomoftheWillwithregardtoanyofhisfirst-orderdesires,heisfreeeithertomakethatdesirehiswillortomakesomeotherfirst-orderdesirehiswillinstead.Whateverhiswill,then,thewillofthepersonwhosewillisfreecouldhavebeenotherwise;hecouldhavedoneotherwisethantoconstitutehiswillashedid.(1971:18–19)Incontrast,Frankfurtinitiallymaintained,itsufficesforactingfreelythatanagent“hasdonewhathewantedtodo,thathediditbecausehewantedtodoit,andthatthewillbywhichhewasmovedwhenhediditwashiswillbecauseitwasthewillhewanted”(ibid.:19).Anagentmayactfreely,onthisview,evenwhensheisunabletodootherwise,evenwhenshelacksfreewill.Anditisactingfreely,ratherthanhavingafreewill,thatisrequiredformoralresponsibility,accordingtoFrankfurt.AnumberofdifficultieshavebeenraisedforthisearlyversionofFrankfurt’saccountoffreeaction.First,asFrankfurthimselfnoted(ibid.:21),justastheremaybeconflictsamonganagent’sfirst-orderdesires,sotheremaybeconflictsatanyhigherlevelinthehierarchy.Whatarewetosayaboutfreedomofactionincasesofsuchhigher-orderconflict?Secondly,andmorefundamentally,higher-orderdesiresare,afterall,justdesires,anditisnotclearhowtheycanhaveanymoreauthoritythanfirst-orderdesireshavewithrespecttoanagent’sidentityor20freedom.AthirdproblemisthatFrankfurtplacednorequirementsonhow21higher-orderdesiresareformed.Itappearsthattheirformationinsomecasecouldbeduetofreedom-underminingcompulsion,orthatitcouldbeexternallycontrolledinawaythatwouldunderminetheagent’sfreedom;andthustheconditionssaidbyFrankfurttobesufficientforfreeactionmaynotinfact22suffice.Andfinally,anagentmay,onacertainoccasion,desirenottoactonacertaindesirebutnevertheless,throughperversity,weaknessofwill,orresigna-tion,freelyactonit,andonemayinsomeinstanceoffreeactionfailtoexerciseone’sreflectivecapacities.Hence,theconditionsinitiallyadvancedbyFrankfurtappearnottobenecessaryforfreedom.Frankfurthasmadeseveralrevisionstohisinitialaccountinordertoaddresssomeofthesedifficulties.Anearlyproposal(1976)wasthatbydecidingthatshewantstobemovedbyacertainfirst-orderdesire,anagentmayidentifywiththatdesire.Suchadecision,hesuggested,unlikeahigher-orderdesire,isnotcapableitselfofbeingsomethingwithwhichtheagentisnotidentified;theideahereseemstobethatadecisionofthissortcannotlackauthorityinidentifyingonewithacertainfirst-orderdesire,andthatbymeansofmakingsuchadecisionan23agentcanresolveanyconflictamongherhigher-orderdesires.Inalaterwork(1987),Frankfurtproposedthatadecisionfavoringacertainfirst-orderdesireeffectivelyidentifiestheagentwiththatdesireonlywhenitleavestheagent“wholehearted.”Mostrecently(1992),itisthisnotionofwholeheartednessthat24Frankfurthasemphasizedandfurtherarticulated.Itrequiresthat,ifthereisanyconflictamongtheagent’shigher-orderattitudes,theagentisunambivalent,fullyresolvedconcerningwhereshestandswithrespecttothisconflict.Moreover,awholeheartedagent,Frankfurtsays,hasnointerestinmakingchangestoher377\nRandolphClarkecommitments,andherlackofsuchaninterestisnotunreflectivebutderivesfromherunderstandingandevaluationofherpsychicstate.Arequirementofwholeheartednessaddressesthefirstoftheproblemsnotedabove,foritrulesoutcertaintypesofhigher-orderconflict.Buttherequirement,althoughperhapsappropriateforanaccountofidentification,istoostrongforanaccountofactingwiththetypeofcontrolthatisrequiredformoralrespons-ibility;ambivalenceisnottypicallyanexcusingcondition.Thisrequirementmaybethought,aswell,tosolvetheproblemoftheauthorityofhigher-orderdesires.Butthisclaimseemsdoubtful,particularlyinlightofthethirdproblem,thatconcerningthesourceofanagent’shigher-orderattitudes.Anagent’swhole-heartedlyendorsingthedesiresonwhichsheactswouldnotseemtorenderheractionfreeifherendorsement,aswellasherwholeheartedness,aretheresultofcompulsionormanipulation.Frankfurtfirmlydeniestherelevancetofreedomofanyfactsaboutthecausalhistoryofhigher-ordervolitions;what25matters,heinsists,isjustthestructureoftheagent’sattitudes.Wemaygrantthatinwholeheartedlyendorsingacertainfirst-orderdesire,anagent“takesresponsibility”(1975:121)forthatdesireandforactingonit.Butanagentmaytakeresponsibility,inthissense,withoutreallybeingresponsibleforwhatshedoes,andhencewithoutgenuinelydeservingpraiseorblamefortheensuing26action.16.2.3CapacityaccountsThelastofthedifficultiesidentifiedforFrankfurt’saccountwasthat,itseems,wemaysometimesactfreelyevenwhenwedonotexerciseourcapacitytoactinaccordwithandonthebasisofahigher-orderendorsement.TheproblemherestemsfromthefactthatFrankfurtrequiresameshbetweenone’seffectivefirst-orderdesireandacertainhigher-orderattitude.Whatwemaycallcapacityaccountsevadethisdifficulty.Onsuchviews,freeagencyrequiresthatonehaveageneralabilityorcapacitytoappreciatepracticalreasonsandtogovernone’sbehaviorbypracticalreasoning(andonseveralversions,itrequiresaswellacapacitytoreflectrationallyonone’sreasonsandtoinfluenceone’sreason-states–suchasone’sdesires–andhenceone’sbehaviorbymeansofsuchreflection);butitisheldthatonemayactfreelyonsomeoccasionevenifonedoesnoton27thatoccasionexercisethiscapacity.Thereisagreatvarietyofsuchviews;weshallconsiderhereaproblemthatfacesallofthem.Actingfreelyisactingwithacertaintypeofcontrol.Itrequires,itseems,notjustthatoneactwithacapacityfor(reflective)rationalself-governance,butalsothatonecontrolwhetherandhow,onagivenoccasion,thatcapacityisexercised.Andacompatibilistversionofacapacityaccountwillhavetoexplainhow,iftheworldisdeterministic,anagentmaycontrolwhetherandhowhercapacityforrationalself-governanceisexercised.378\nFreedomoftheWill16.2.4AresponsivenessviewThereasons-responsivenessviewadvancedbyJohnMartinFischerandMarkRavizza(1998)iscloselyrelatedtocapacityaccounts,anditoffersaresponsetotheproblemjustidentified.AccordingtoFischerandRavizza,thevarietyof28controlthatsufficesforresponsibilityiswhattheycallguidancecontrol.Guid-ancecontrolofagivenactionischaracterizednotintermsoftheagentorhercapacities,butintermsofthemechanism(orprocess)bywhichtheactionis29produced:thatmechanismmustbesufficientlyreasons-responsive,anditmustbetheagent’sownmechanism.Letustaketheserequirementsinturn.FischerandRavizza(1998:69)recognizetwoaspectsofreasons-responsiveness:receptivityandreactivity.Thefirstisamatterofappreciatingorrecognizingreasons,thesecondamatterofproducingcertaindecisionsandotheractionsonthebasisofone’srecognitionofreasons.Sinceresponsivenessisadispositionalormodalfeature,bothoftheseaspectsarecharacterizedintermsofhowthemech-anismthatproducesanagent’sactiononagivenoccasionwouldfunctioninvarioushypothetical(ornon-actual)situations.Thereceptivitythatisrequiredisanunderstandablepatternofrecognitionofreasons,minimallygroundedinreality.Thatis,theagentmust“notbesubstan-tiallydeludedaboutthenatureofreality”(ibid.:73),andtheremustbeavarietyofscenariosinwhich,withthemechanisminquestionoperating,theagentwouldexhibitapatternofreasons-recognitionindicatingthatshe“recognizeshowreasonsfittogether,seeswhyonereasonisstrongerthananother,andunderstandshowtheacceptanceofonereasonassufficientimpliesthatastrongerreasonmustalsobesufficient”(ibid.:71).(Forexample,ifBethhastoldalie,inorderforthemechanismthatproducedheractoflyingtohavebeensufficientlyreceptivetoreasons,theremustbevarioushypotheticalscenariosinwhichBethhasvariousreasonsnottolie,thesametypeofmechanismoperates,andinasuitablevarietyofthesescenarios,sherecognizesthesereasonsnottolie.)Fur-ther,themechanisminquestionmustbereceptivetomoralreasons,amongothers;andaswithreceptivitytoreasonsingeneral,thereceptivitytomoralreasonsmustexhibitanunderstandablepattern.Thereactivityrequirementisweaker;itissatisfiedifthereisatleastonescenarioinwhichtheagenthassufficientreasontoactotherwise,themechanisminquestionoperates,andtheagentactsotherwisebecauseofthatreasontodoso.Moralresponsibility,accordingtoFischerandRavizza,doesnotrequirethattherebeanysituationinwhich,withthemechanisminquestionoperating,theagentwouldactonmoralreasons.Anagentwhorecognizesbutsteadfastlyrefusestobemovedbymoralreasonsmaybeblameworthyforhermisdeeds30(ibid.:79–80).Turningtothesecondmainrequirement,amechanismistheagent’sown,accordingtoFischerandRavizza,justincasetheagenthas“takenresponsibility”379\nRandolphClarkeforactionsthatstemfromit.Takingresponsibility(foractionsstemmingfrommechanismsofcertaintypes),theyhold,isaprocessinwhichtheagentcomestoseeherselfasanagent–assomeonewhosechoicesandactionsareefficaciousintheworld;sheacceptsthatsheisanappropriatetargetofreactiveattitudes(suchasgratitudeandindignation)andofcertainpractices(suchastheissuingofrewardsandpunishments)insofarasheractionsareproducedbymechan-ismsofthosetypes;andtheseviewsofherselfareappropriatelybasedonthe31evidence.Whenanagent’sbehaviorisproducedbyherown,sufficientlyreasons-responsivemechanism,sheactswithguidancecontrol.And,itmaythenbeheld,sheactswithacapacityforrationalself-governanceandexercisescontroloverwhetherandhow,onthisoccasion,thatcapacityisexercised.However,sinceFischerandRavizzacharacterizeownershipintermsoftheagent’sattitudesaboutherself,theiraccountmaybevulnerabletoobjectionsof32thefollowingsort.Supposethat,fromthebeginningofhislife,themechanismsthathaveproducedtheactionsofacertainagent,Allen,haveoneveryoccasionbeeninfluencedbyacertainneuroscientist,Nina.WithoutdirectlyalteringwhatAllendesires,andwithoutrenderinghimlessrationalthananaverageoneofus,NinaroutinelyalterstherelativemotivationalstrengthsofAllen’sdesiressothatheiscausallydeterminedtochooseandperformtheactionsthatNinaselects;andwereitnotforNina’sinterventions,Allen’sdecisionsandotheractionswouldhavebeenquitedifferent.Moreover,Ninaisfondofreasons-responsiveness;invarioushypotheticalsituations,shewouldinfluenceAlleninsuchawaythathewoulddisplayanunderstandablepatternofreasons-recognition(includingtherecognitionofmoralreasons)andwouldatleastsometimes,whenthereissuffi-cientreasontoactotherwise,actotherwiseforthatreason.Andsupposethat,unawareofNina’sinterventions,AllenhascometoholdtheviewsofhimselfthatFischerandRavizzarequire,andthattheseviewsareappropriatelybased33ontheevidenceofwhichheisaware.Hehasunwittingly“takenresponsibility”foractionsstemmingfromatypeofmechanismcontrolledbysomeoneelse.Allenseemstomeettherequirementsofactingonhisown,sufficientlyreasons-responsivemechanism,butitdoesnotseemthatheactsfreelyorismorally34responsibleforwhathedoes.FischerandRavizzaclaimthatinacaseofthissort,whereanagenthasbeensubjecttorepeatedintervention,theagent“cannoteverhavedevelopedintoacoherentself.Thatis,undertheenvisagedcircumstances,thereisnoselforgenuineindividualatall”(1998:234–5,note28).Thereisnoresponsibleagenthere,theyimply,becausethereisnogenuineselforindividual.Thisreplyseemstoimposeanew,thirdrequirementforfreeaction,arequirementofgenuineselfhood;andweneedtobetoldwhatthisrequiremententails.Tounderscorethisneed,wemaynotethatitisnotatallclearwhy,inthepresentcase,Allencouldnotbesaidtohavedevelopedintoacoherentorgenuineself.Afterall,35Ninamaylikecoherenceasmuchasshelikesreasons-responsiveness.380\nFreedomoftheWill16.2.5TheremainingdisputeWehavethoughtthroughseveralcases(hereandinthenotes)thatmaybeacceptedbycompatibilistsandincompatibilistsalikeasshowingthatacertaincompatibilistaccountfailstodrawintherightplacethelinebetweeninfluencesthatdoandinfluencesthatdon’tundermineresponsibility.Thistypeofactivitycanyieldanegativeverdictaboutparticularcompatibilistaccounts,butitisunlikelytosettledefinitivelythegeneralquestionwhetherthecontrolthatisrequiredforresponsibilityiscompatiblewithdeterminism.Howeversuccessfulweareatthisactivity,compatibilistsmayreasonablycontinuetosearchforacompatibilistaccountagainstwhichthey,atleast,findnocounterexamples.Andshouldtheyproducesuchatheory,incompatibilistsmightstill,againwithoutunreasonableness,maintainthatallcasesinwhichitissupposedthatdeterminismholdsarecounterexamplestotheviewinquestion.Thegeneralquestionmightbemorefruitfullyaddressedbyseekingsomebasic3637principlesconcerningresponsibility,oratheoryofwhatitistoberesponsible.Workinthesedirectionsmight,ifnotsettlethedispute,atleastclarifythepointsofdisagreement.Butsinceproposedprinciplesoraproposedtheoryofrespons-ibilitywillthemselvesbecontroversial,adefinitiveresolutionofthequestionbeforeusdoesnotappearimminent.16.3LibertarianAccountsIfdecidingandactingfreelyareincompatiblewithdeterminism,theneithersuchfreedomisimpossibleorindeterminismwouldsomehowmakeitpossible.Howmightthelatterbeso?Recentincompatibilist(orlibertarian)accountsoffreeactionandfreewillofferthreedifferentanswerstothisquestion.Someholdthatfreedecisionsandotherfreeactionsmust(oratleastcan)havenocauseatall;othersholdthattheymustbenon-deterministicallycausedbycertainpriorevents;andathirdtypeholdsthatafreedecisionorotherfreeactionmustbecausedby38theagent,asubstance.Beforeexaminingrepresentativesofeachofthesetypesofview,letusbrieflyconsidertherelationbetweenlibertarianfreedomandthenatureofthemental.Historically,manylibertarianshavebeenmind–bodydualists,holdingeitherthatmindsareimmaterialsubstancesorthatmentalpropertiesandeventsareimma-terial.Butiffreewillandactionareincompatiblewithdeterminism,dualismappearstobeofnohelptothosewhowishtofindaplaceintheworldforfree-dom.Foronething,recallthatthecharacterizationsofdeterminismofferedabovearenotrestrictedtophysicalevents;iftheworldisdeterministic,then(assumingdualismistrue)immaterialmentaleventsareasfullydeterminedbypriorevents381\nRandolphClarkeasareanyotherevents.Secondly,evenifphysicaleventsalonewerefullydeter-mined,allmovementsofourbodieswouldbefullydetermined,andthen,iffreedomisincompatiblewithdeterminism,atmostwemightbeabletomakefreedecisionsthatcouldmakenodifferencetowhatbodilybehaviorweengagein.Finally,andmostimportantly,iffreedomanddeterminismareincompatible,thenadualiststillhastheproblemofexplaininghowindeterminismintherealmoftheimmaterialcanmakefreedompossible.Andifwecanexplainhowanundeterminedimmaterialdecisionisfree,thenitappearsthatwecanjustaswellexplainhowanundetermineddecisionthathappenstobeaphysicaleventisfree.Henceitisnotclearthatdualismconfersanyadvantagetolibertarians;andconversely,ifmaterialismisthebetterviewofthemind,thatappearstobeno39problemforlibertarians.16.3.1Non-causalviewsSomelibertarianaccountsrequirethatafreeactionhavenocauseatall;somerequirethatiteitherhavenocauseorbeonlynon-deterministicallycaused.Sincebothsuchviewsholdthattherearenopositivecausalrequirementsthatmustbesatisfiedinorderforanactiontobefree,wemaycallthem“non-causalviews.”CarlGinet(1990)hasadvancedoneofthemostsophisticatednon-causal40libertarianaccounts.Onhisview,everyactionisorbeginswithacausallysimplementalevent,i.e.,amentaleventwithnointernalcausalstructure.(Decisionsandvolitionsaresaidtobeexamplesofsuchbasicactions;avolitionisheldtobeanagent’swillingortryingtomakeacertainexertionofherbody.)Andwhatmakessomementaleventabasicaction,ratherthanachangethattheagentpassivelyundergoes,isnothowthateventiscausedbutratheritshavingacertainintrinsicfeature,an“actishphenomenalquality”(1990:13).Thisqualityisbestdescribed,Ginetsuggests,asitsseemingtotheagentasifshedirectlyproducesordeterminesthementaleventinquestion.(Non-basicactionsarethenheldtoconsistinanaction’sgenerating–e.g.causing–somefurtherevent,orinan41aggregateofactions.)Giventhatacertaineventisanaction,whatmoreisneededinorderforittobeafreeaction?Therearenofurtherpositiveconditionsthatmustbesatisfied,onGinet’sview;theadditionalrequirementsarewhollynegative.Theactionmustnotbecausallydetermined,andinperformingtheaction,theagentmustnotbesubjecttoirresistiblecompulsion(suchasanirresistiblecravinginducedbyaddictiontoadrug).Twoproblemsariseforthisview,andtheyconfrontallnon-causalaccounts.First,actingwithfreewillisexercisingacertainvarietyofcontroloverone’sbehavior,andnon-causalaccountsappeartolackanadequateaccountofinwhatthatcontrolconsists.Anobviouscandidateisthatitconsistsintheaction’sbeingcaused,inanappropriateway,bytheagent,orbycertaineventsinvolvingthe382\nFreedomoftheWill42agent(suchasherhavingcertainreasonsandacertainintention).AlthoughGinetholdsthateverybasicactionseemstotheagentasifsheisdirectlyproduc-43ingit,hemaintainsthatitisstrictlyfalsethatagentscausetheiractions.Asforthisactishphenomenalqualityitself,itseemsdoubtfulthathowamentaleventseemstotheindividualundergoingitcanconstitutethatindividual’sexerciseofcontroloverthatevent,ratherthanbea(moreorlessreliable)signofsuchcontrol.Thedoubtisreinforcedbythefactthat,onGinet’sview(1990:9),amentaleventwithanactishfeelcouldbebroughtaboutbyexternalbrainstimu-lation,intheabsenceofanyrelevantdesireorintentiononthepartofthe“agent.”Aneventsoproducedhardlyseemstobeanexerciseofactivecontrol,evenifitseemstotheindividualthatitis.Secondly,actingfreelyisactingwithacapacityforrationalself-governanceanddetermining,oneself,whetherandhowoneexercisesthatcapacityonagivenoccasion.Henceitmustbepossibleforafreeactiontobeanactionperformedforacertainreason,anactionforwhichthereisarationalexplanation.Obviouscandidatesforaccountsofthesephenomenarequirecausalconnectionsbetweenreason-states(suchasdesires)andactions:anagentactsforacertainreasononlyifthecorrespondingreason-state(ortheagent’spossessingthatstate)causes,inanappropriateway,theagent’sbehavior;andcitingareason-statecontributestoarationalexplanationofanactiononlyifthatreason-state(orthe44agent’spossessingit)caused,inanappropriateway,theaction.Non-causalviewsrejectsuchproposals,butitisdoubtfulthatthealternativestheyofferareadequate.Ginet(ibid.:143)offersthefollowingaccountofrationalexplanationthatcitesadesire.Suppose,forexample,thatCatewantstocheerupDaveandbelievesthatifshetellsajoke,thatwillcheerhimup;shethentellsajoke.OnGinet’sview,citingthedesiretocheerhimupexplainshertellingthejokejustincase:(a)priortohertellingthejoke,CatehadadesiretocheerupDave,and(b)concurrentlywithtellingthejoke,Caterememberedthatdesireandintendedof45heractoftellingthejokethatitsatisfy(orcontributetosatisfying)thatdesire.Note,first,thattheconcurrentintentionrequiredhereisasecond-orderattitude:anattitudeabout(amongotherthings)anotherofone’sownattitudes(acertaindesire).Butitseemsplainthatonecanactforacertainreason,andcitingadesirecanrationallyexplainone’saction,evenifonedoesnothavewhenoneactsanysuchsecond-orderintention.Cate,forexample,mightactonherdesiretocheerupDave(andcitingthatdesiremightrationallyexplainheraction)evenifheronlyintentionisanintentiontocheerhimupbytellingthejoke.Further,itisdoubtfulthatGinet’saccountprovidessufficientconditionsforrationalexplana-tion.ForsupposethatCatealsohadotherreasonsfortellingthejoke,reasonsthatcausallycontributedtoherdoingsoandofwhichshewasquiteawarewhenshetoldthejoke.Then,ifherdesiretocheerupDaveplayednoroleatallinbringingabout(causing)herbehavior,itisquestionable(atbest)whethershereallyactedonthatdesireandhencewhethercitingittrulyexplainswhatshe46did.383\nRandolphClarke16.3.2Non-deterministicevent-causalviewsBothoftheobjectionsraisedagainstnon-causalaccountssuggestthatonanadequatelibertarianview,freeactionswillbeheldtobecaused.Somelibertariansmaintainthatwhatisneededisanappealtonon-deterministiceventcausation.Whenoneeventbringsaboutanother,thatinstanceofcausationmaybe(onsomeviewsofcausation,itmustbe)governedbyacausallaw.Butcausallawsmaybeeitherdeterministicornon-deterministic.Statementsoftheformerimplythateventsofonetypealwayscauseeventsofasecondtype.Statementsofnon-deterministiclawsimplythateventsofonetypemightcauseeventsofasecondtype.Suchlawsmaybeprobabilistic,theirstatementsimplyingthateventsofthefirsttypeprobabilify(toacertaindegree)eventsofthesecondtype,orthatwhenthereoccursaneventofthefirsttype,thereisacertainprobabilitythatitwillcauseaneventofthesecondtype.Whenoneeventnon-deterministicallycausesanother,thefirstproducesthesecond,thoughtherewasachancethatitwould47notbringaboutthatsecondevent.Thesimplestevent-causallibertarianviewtakestherequirementsofagoodcompatibilistaccountandaddsthatcertainevents(suchastheagent’shavingcertainreasons)thatcausethedecisionorotheractionmustnon-deterministicallycauseit.Anagentmay,forexample,havecertainreasonsfavoringonealternativethatsheisconsideringandotherreasonsfavoringanother.Onthetypeofac-countinquestion,theagentmayfreelydecideinfavorofthefirstactionifthatdecisionisnon-deterministicallycausedbyherhavingthefirstsetofreasons,whilethereremainedachancethatshewouldinsteaddecideinfavorofthesecondalternative,wherehersodecidingwouldhavebeencausedbyherhavingthesecondsetofreasons.Whentheseconditionsaresatisfied,theactionisperformedforreasons,itis(aproponentwillsay)performedwithacertainvariety48ofcontrol,anditwasopentotheagenttodootherwise.Acommonobjectionagainstsuchaviewisthattheindeterminismthatitrequiresisdestructive,thatitwoulddiminishthecontrolwithwhichagentsact.Theobjectionisoftenpresentedintermsofanallegedproblemofluck.Suppose,forexample,thatacertainagent,Isabelle,hasbeendeliberatingaboutwhethertokeepapromiseornot.Shejudgesthatshe(morally)oughttokeepit,thoughsherecognizes(andistemptedtoacton)reasonsofself-interestnotto.Shedecidestokeepthepromise,andherdecisionisnon-deterministicallycausedbyherpriordeliberations,includinghermoraljudgment.Butuntilshemadeherdecision,therewasachancethatherdeliberativeprocesswouldterminateinadecisionnottokeepthepromise,adecisionnon-deterministicallycausedbyIsabelle’sreasonsofself-interest;everythingpriortothedecision,includingeverythingaboutIsabelle,mighthavebeenexactlythesameandyetshemighthavemadethealternativedecision.Hence,accordingtotheobjection,itisamatterofluckthatIsabellehasdecidedtodowhatshejudgedtobemorallyright.(Isabelle,itmightbesaid,hascounterpartsinotherpossibleworldswhoareexactlylikeheruptothemoment384\nFreedomoftheWillofdecisionbutwhodecidenottokeepthepromise;there,butforgoodluck,goesshe.)Totheextentthatsomeoccurrenceisamatterofluck,theobjectionstates,itisnotunderanyone’scontrol.Therequiredindeterminismisthussaid49todiminishIsabelle’scontroloverthemakingofherdecision.Motivatedpartlybyadesiretorespondtothisobjection,someproponentsofevent-causallibertarianaccountshavemodifiedthesimpleversionofsuchaview50thatweconsideredtwoparagraphsback.ThemostdetailedmodifiedviewisthatadvancedbyRobertKane(1996),whichdiffersfromthesimpleversionin51twomainrespects.First,ifadecisionsuchasIsabelle’sisfree,then,onKane’sview,thedecisionisimmediatelyprecededbyaneffortofwill,aneffortonthe52agent’sparttogetherendsorpurposessortedout.Insuchacaseofmoralconflict,theagentmakesanefforttoresisttemptationandtodecidetodowhatshehasjudgedshemorallyoughttodo.And,Kanerequires,suchaneffortis“indeterminate”inawayanalogoustothewayinwhich,accordingtothelawsofquantummechanics,thepositionormomentumofasubatomicparticlemaybeindeterminate.Indeed,itisduetosuchindeterminacyoftheeffort,Kaneholds,thatitwillbeundeterminedwhichdecisiontheagentmakes.Secondly,onKane’sview,whensuchadecisionisfree,theagentwill,bymakingthatdecision,makethereasonsforwhichshedecidesthereasonsshewantsmoretoactonthanshewantstoactonanyothers.InIsabelle’scase,shewillmakehermoralreasonstheonesshewantsmosttoactonbydecidingforthosereasons.Thefirstofthesemodifications,thatrequiringeffortsofwill,isheldtoaddresstheproblemofluckintwoways.TheproblemwasraisedabovebynotingthatIsabellehascounterpartsexactlylikeheruptothemomentofdecisionwhodecidenottokeepthepromise.Kaneclaimsthatwherethereisindeterminacy–asthereisonhisviewwiththeindeterminateeffortofwill–therecanbenoexactsamenessfromoneworldtoanother.Hence,onhisview,therewouldbenocounterpartofIsabellewhomakesexactlythesameeffortofwillandsoisexactlylikeheruptothemomentofdecisionbutdecidesotherwise.Andthus,hesuggests,theargumentfromluckisdefused.Buttheproblemisnotsoeasilydismissed.Itisnotclearwhytherecannotbeexactsamenessofoneworldtoanotherifthereisindeterminacy.Inphysics,theindeterminatepositionofaparticlemaybecharacterizedbyawavefunction(onespecifyingtheprobabilitiesoftheparticle’sbeingfound,uponobservation,invariousdeterminatepositions),andaparticleanditscounterpartmaybothbecorrectlycharacterizedbyexactlythesamewavefunction.Further,evenifthereisnosuchexactsameness,theproblemremains.ForitisstillthecasethatIsabelle’sdecisionresultsfromtheworkingoutofachancyprocess,aprocessthatmightinsteadhaveproducedadecisionnottokeepthepromise.Andtheobjectionmaystillberaisedthatthenherdecisionisamatterofluckandhencelessunderhercontrolthanitwouldhavebeenhadherdeliberationscausallydeterminedit.Thesecondwayinwhich,Kaneholds,therequiredeffortsofwillhelptoaddresstheproblemofluckconcernsthefactthattheyareactiveattemptsbythe385\nRandolphClarkeagenttodosomethinginparticular.Onhisview,whenanagentsuchasIsabelledecidestodowhatshehasjudgedshemorallyoughttodo,itisasaresultofherefforttomakethatverydecisionthatshemakesit.Shesucceeds,despitetheindeterminism,indoingsomethingthatshewastryingtodo.AndKanepointsoutthattypically,whenthisisso,theindeterminismthatisinvolveddoesnotundermineresponsibility(andhenceitdoesnotsodiminishcontrolthatthereisnotenoughforresponsibility).Hedrawsananalogywithacase(1999b:227)inwhichamanhitsaglasstabletopattemptingtoshatterit.Evenifitisundeter-minedwhetherhiseffortwillsucceed,Kanenotes,ifthemandoessucceed,hemaywellberesponsibleforbreakingthetabletop.Kane(1999a,1999b,and2000)hasrecentlyextendedthisstrategytocoverdecisionstodowhatoneistemptedtodoaswellasdecisionstodowhatone53believesoneoughttodo.InacasesuchasIsabelle’s,heproposes,thedecisionisprecededbytwo,simultaneouseffortsofwill,bothofwhichareindeterminate.Theagenttriestomakethemoraldecision,andatthesametimeshetriesto54maketheself-interesteddecision.Whicheverdecisionshemakes,then,shesuc-ceeds,despitetheindeterminism,atdoingsomethingthatshewastryingtodo.Hence,Kaneholds,whicheverdecisionshemakes,shemaybe,likethemanwhobreaksthetabletop,responsibleforwhatshedoes.Note,however,thatthemaninKane’sexampleactswiththecontrolthatsufficesforresponsibilityonlyifhisattempttobreakthetabletopisitselffree.Aneffort’sbringingaboutadecisioncancontributeinthesamewaytothedecision’sbeingfree,then,onlyiftheeffortitselfisfree.Hencewhatisneededisan55accountoftheagent’sfreedominmakingtheseefforts.AndKanefacesthefollowingdilemmainprovidingsuchanaccount.IftheaccountofthefreedomofaneffortofwillthatprecedesadecisionsuchasIsabelle’srequiresthatthiseffortitselfresultfromapriorfreeeffort,thenaviciousregresslooms.Ontheotherhand,iftheaccountofthefreedomofaneffortofwillneednotappealtoanypriorfreeeffortsofwill,thenitwouldseemthattheaccountofafreedecisionitselfcouldlikewisedispensewithsuchanappeal.Insum,itdoesnotappearthatanythingisgainedbytherequirementthatafreedecisionsuchasIsabelle’sbeprecededbyanindeterminateeffortofwill.NeitherdoesitseemthatthesecondmodificationfavoredbyKanehelpstoaddresstheproblemofluck.Theproblemconcernstheagent’scontroloverwhatshedoes,andcontrol,itseems,isacausalphenomenon,amatterofwhatcausesdecisionsandotheractions.Butanagent’swantingmoretoactoncertainreasonsis,onKane’sview,somethingthatisbroughtaboutbymakingadeci-sion,notsomethingthatbringsaboutthedecision.Henceitdoesnotseemtocontributeinanywaytotheagent’scontroloverhermakingthatdecision.Thesemodificationstothesimple,event-causalaccountdonotseemtohelpwiththeproblemofluck.Buthowbadistheproblemforthatsimplerview?First,itisclearthatIsabelle’sdecisionisnotentirelyamatterofluck.Foritiscaused(inanappropriateway,wemaysuppose)byherappreciationofherreasons,includingherjudgmentthatsheoughttokeepthepromise.Andits386\nFreedomoftheWillbeingcausedinthisway,compatibilistsshouldagree,constitutestheagent’s56makingthedecisionwithacertaindegreeofcontrol.Secondly,itmaybequestionedwhetherIsabelle’sdecisionisatallamatterofluck,inanordinarysense.Theterm“luck,”inordinaryusage,carriesconnota-tionsofsomething’sbeingoutofanagent’scontrol,butitisnotsoobviousthattheindeterminismrequiredbyanevent-causallibertarianviewyieldscontrol-diminishingluck.Toseethis,wemaydistinguishtwoimportantlydifferentkindsofcase:acaseinwhichthereisindeterminismbetweenabasicactionandanintendedresultthatisnotitselfanaction,andacase–forexample,Isabelle’s–inwhichtheindeterminismisinthecausationofabasicactionitself.Forthefirstsortofcase,supposethatyouthrowaballattemptingtohitatarget,whichyousucceedindoing.Theball’sstrikingthetargetisnotitselfanaction,andyouexercisecontroloverthiseventonlybywayofyourprioractionofthrowingtheball.Nowsupposethat,duetocertainpropertiesoftheballandthewind,theprocessbetweenyourreleasingtheballanditsstrikingthetargetisindeterministic.Indeterminismlocatedhereinhibitsyoursucceedingatbringingaboutanon-activeresultthatyouwere(freely,wemaysuppose)tryingtobringabout,andfor57thisreasonitclearlydoesdiminishthecontrolthatyouhaveovertheresult.ButtheindeterminisminIsabelle’scase–andtheindeterminismrequiredbythesimpleevent-causallibertarianview–islocateddifferently.Itislocatednotbetweenanactionandsomeintendedresultthatisnotitselfanaction,butratherinthedirectcausationofthedecision,whichisitselfanaction.Isabelleexercisescontroloverthatdecisionnotonly(sheneednotatall)bywayofherperform-anceofsomeprioraction.Henceindeterminismlocatedhereisnotaninhibitingfactorinthewaythatitisinthefirstsortofcase.IftheindeterminisminIsabelle’scaseneverthelessdiminishescontrol,thentheexplanationofwhyitdoessowillhavetobedifferentfromthatavailableinthefirstsortofcase.Butitisunclearwhatthisalternativeexplanationwouldbe,andhenceitisnotclearthattheindeterminisminIsabelle’scasedoesinfactyieldcontrol-diminishingluck.Theluckobjectionagainstevent-causallibertarianaccountsappearsincon-clusive.Butasecondobjectionremainstobeconsidered.Eveniftherequiredindeterminismdoesnotdiminishcontrol,itissometimesobjected,itaddsnoth-ingofvalue,itissuperfluous.Inordertoassessthisclaim,letusreturntothereasonswhyfreedomisimportanttous.Wevalueafreedomthatgroundsdignityandresponsibility,intheexerciseofwhichwemakeadifferencetothewaytheworldgoes,andonethataccordswiththeappearanceofopennessthatwefindindeliberating.Wecandistinguishtwoaspectsofthisfreedom:akindofleewayoropennessofalter-natives,andatypeofcontrolthatisexercisedinaction.Aswenoticedwhenconsidering(insection16.2.1)Frankfurt’sattackontheprincipleofalternatepossibilities,thefreedominwhichweareinterestedforsomeoftheabovethingsmayinvolveonebutnottheotheroftheseaspects.Inasimilarfashion,itmaybethatwhatisgainedwiththeindeterminismthatanevent-causallibertarianviewrequireshastodowithoneoftheseaspectsbutnottheother.387\nRandolphClarkeAnagent’sexerciseofcontrolinactingisherexerciseofapositivepowertodeterminewhatshedoes.Wehaveseenreasontothinkthatthisisamatteroftheaction’sbeingcaused(inanappropriateway)bytheagent,orbycertaineventsinvolvingtheagent–suchasherhavingcertainreasonsandacertainintention.Anevent-causallibertarianviewaddsnonewcausestothosethatcanberequiredbycompatibilistaccounts,andhencetheformerappearstoaddnothingtotheagent’spositivepowertodeterminewhatshedoes.Asfarasthisaspectoffreedomisconcerned,therequirementofindeterminismdoesindeedappear(atbest)superfluous.Butnotsowithregardtotheotheraspect,theopennessofmorethanonecourseofaction.IftheConsequenceArgument(consideredinsections16.1.2and16.1.3)iscorrect,thereisneveranysuchopennessinadeterministicworld.Theindeterminismrequiredbyanevent-causallibertarianaccountsufficestosecurethisleewayoropenness,andthismaybeimportanttousforseveralreasons.Someindividuals,atleast,mayfindthatwhentheydeliberate,theycannothelpbutpresumethatmorethanonecourseofactionisgenuinelyopentothem.Iftheworldisinfactdeterministic,theseindividualsaresubjecttoanunavoidableillusion(sincewecannotavoiddeliberating).Andtheymayreason-ablyjudgethatitwouldbeforthisreasonbetterifthingsareaspresentedintheevent-causallibertarianview.Similarly,someindividualsmayreasonablyjudgethatifthingsareaspresentedinthisview,thatisbetterwithregardtoourmakingadifference,inperformingouractions,tohowtheworldgoes.Eveniftheworldisdeterministic,thereisawayinwhich,inacting,wegenerallymakeadifference:hadwenotdonewhatwedid,thingswouldhavegonedifferently.Ifthingsareaspresentedinanevent-causallibertarianaccount,westillgenerallymakeadifferenceinthisway.Butwemaymakeadifferenceinasecondwayaswell:inactingwemayinitiate,bytheexerciseofactivecontrol,branchingsinaprobabilisticunfoldingofhistory.Theremayhavebeenarealchanceofthings’notgoingacertainway,andouractionsmaybetheeventsthatsetthingsgoingthatway.Onemayreasonablyjudgethatitisbettertobemakingadifferenceinthissecondaswellasinthefirstwaywithone’sactions.Sincewecannotbemakingadifferenceinthissecondwayiftheworldisdeterministic,someindi-vidualsmayhavereasontofindthattheindeterminismrequiredbyanevent-causallibertarianviewisnotsuperfluousbutaddssomethingofvalue.Isthereanythingtobegainedwithrespecttoresponsibility?Thatisnotclear.Ifresponsibilityisnotcompatiblewithdeterminism,thenwhatmoreisrequiredforitthanwhatisofferedbyagoodcompatibilistaccount?Theleewaysecuredbytheevent-causallibertarianviewdoesn’tseemtobetherequiredaddition;ifFrankfurtisright,itisn’trequiredatall.Theactualcausalprocessthatproducesadecisionorotheractiononthisviewisindeterministic,butitisnotclearthatthatmakesthecrucialdifference.Itisstill,asitisonacompatibilistaccount,aprocessinwhichallofthecausesofthedecisionorotheractionareevents,whichmaybebroughtaboutbyotherevents,leadingbacktotheBigBang.Aswassuggestedabove,itisnotclearthatonthisviewtheagentexercisesanygreater388\nFreedomoftheWillpositivepowersofcontrol.Andthatiswhatwouldseemtobeneededifthereistobeadifferentverdictregardingresponsibility.Ifresponsibilityisnotcompat-iblewithdeterminism,itmaynotbesecuredbyanevent-causallibertarianview,either.16.3.3Agent-causalaccountsIf,onanevent-causallibertarianview,agentsdonotexerciseanygreaterpositivepowersofcontrolthantheydooncompatibilistaccounts,whattypeoflibertarianviewwouldsecuregreatercontrol?Anumberoflibertarianshavemaintainedthatsuchaviewmustholdthatafreedecisionorotherfreeaction,whilenotcausally58determinedbyevents,iscausedbytheagent,andthatcausationbyanagentisdistinctfromanddoesnotconsistincausationbyevents(suchastheagent’s59havingcertainreasons).Anagent,itissaid,isacontinuantorsubstance,andhencenotthekindofthingthatcanitselfbeaneffect(thoughvariouseventsinitslifecanbe).Ontheseagent-causalaccounts,then,anagentisinastrictandliteralsenseanoriginatorofherfreeactions,anuncausedcauseofherbehavior.Thiscombinationofindeterminismandoriginationisthoughttocapturebestthekindoffreedomwedesirewithrespecttodignity,responsibility,difference-making,andtheappearanceofopenness.Twomainproblemsconfrontdefendersofagent-causalaccounts,oneconcern-ingthenotionofagentcausationandtheotherconcerningtherationalexplicabilityoffreedecisionsorotherfreeactionsonsuchviews.Alltheoristswhoacceptacausalconstrualofagents’controloverwhattheydo–andthisincludesmostcompatibilistsaswellasmanylibertarians–holdthat,inasense,agentscausetheirfreeactions.However,mostholdthatcausationbyanagentisjustcausationbycertaineventsinvolvingtheagent,suchastheagent’shavingcertainreasonsandacertainintention.But,aswehaveseen,theagentcausationpositedbyagent-causalaccountsisheldnottobethisatall.Itissaidbymostagent-causaltheoriststobefundamentallydifferentfromeventcausa-tion.Andthisraisesthequestionwhetheranyintelligibleaccountofitcanbegiven.Evensomeproponentsofagent-causalviewsseemdoubtfulaboutthis,60declaringagentcausationtobestrangeorevenmysterious.Moreover,evenifthenotionofagentcausationcanbemadeintelligible,thequestionremainswhetherthethingitself–causationbyasubstanceorcontinu-ant–ispossible.Anoftenrepeatedargumentsuggeststhatitisnot.Eachevent,includingeachaction,itissaid,occursatacertaintime.Andifanactioniscaused,theargumentcontinues,thensomepartofthataction’stotalcausemustbeanevent,somethingthatitselfoccursatacertaintime.Otherwisetherewouldbenowaytoaccountfortheaction’soccurringwhenitdid.Hence,ifanagentcausesanaction,theremustbesomethingtheagentdoes,orsomechangetheagentundergoes,thatcausesthataction.Sinceeithersomethingtheagentdoesorsomechangetheagentundergoeswouldbeanevent,itisconcluded,it389\nRandolphClarkecannotbethecase,asmostagent-causalaccountsmaintain,thatfreeactionsare61causedbyagentsandnotbyanyevents.Thesecondmainproblemforagent-causalviewsisthatfreeactionscanbeperformedforreasonsandcanberationallyexplicable,butif,asmostagent-causalistshold,freeactionshavenoeventcauses,itdoesnotappearthatsuchrationalfreeactionwouldbepossible.Earlierwesawthatplausibleaccountsofactingforcertainreasonsandofrationalexplanationappealtoanaction’sbeingcausedbytheagent’shavingcertainreasons,anditappearedthatnon-causalaccountsofthesephenomenawerenotadequate.Indenying,then,asmostagent-causalistsdo,thatfreeactionshaveanyeventcauses,thesetheoristsappeartoruleoutrationalfreeaction.Inresponsetothissecondproblem,Ihaveproposed(Clarke1993,1996)anagent-causalaccountonwhichafreeactioniscausedbytheagentandnon-deterministicallycausedbycertainagent-involvingevents,suchastheagent’shavingcertainreasons.Giventhisappealtoreasons-causation,theviewcanpro-videthesameaccountsofactingforreasonsandofrationalexplanationascanevent-causalviews.Andsincetheeventcausationthatispositedisrequiredtobenon-deterministic,theviewsecurestheopennessofalternatives,evenontheassumptionthatthisisincompatiblewithdeterminism.Finally,theagentcausa-tionitselfisstillheldtobedistinctfromandnottoconsistincausationbyanyevents,andsothisviewsecurestheoriginationoffreeactionsthatseemedan62appealingfeatureofmoretraditionalagent-causalaccounts.Thismodificationoftraditionalagent-causalviewsalsoaddressestheobjectiondescribedearliertothepossibilityofagentcausation.Thatobjectionconcludesthatitcannotbethecasethatfreeactionsarecausedbyagentsandnotbyanyevents;ifanagentcausesanaction,itissaid,thensomeeventinvolvingthatagentmustcausetheactionandaccountfortheaction’soccurringwhenitdoes.Ontheproposedview,someeventsinvolvingtheagentdocauseeachfreeactionandaccountfortheaction’soccurringwhenitdoes.Still,questionsremainconcerningtheintelligibilityandpossibilityofagentcausation.TimothyO’Connor(1995a,1996,2000)andI(1993,1996),thoughwedifferondetails,havebothsuggestedthatagentcausationmightbecharacter-izedalongthesamelinesaseventcausationifthelatterisgivenanon-reductiveaccount.Familiarreductiveaccountscharacterizeeventcausationintermsofconstantconjunctionorcounterfactualdependenceorprobabilityincrease,andifeventcausationissocharacterizable,thencertainlyagentcausationwouldhavetobefundamentallydifferent.Butifcausationisabasic,irreduciblefeatureoftheworld,thenwemightwithequalintelligibilitybeabletothinkofsubstancesaswellaseventsascauses.Evenifwecanunderstandtheideaofagentcausation,andeveniftheargu-mentforitsimpossibilityconsideredearlierisnoteffective,thereremainreasonstodoubtthatitispossibleforasubstancetocausesomething.Togivejustoneexample:evenifcausationcannotbereducedtoprobabilityincrease,itseemsplausiblethatanycausemustbethekindofthingthatcanaffecttheprobability390\nFreedomoftheWillofitseffectpriortotheoccurrenceofthateffect,evenwhenthecausedirectlybringsaboutthateffect.Eventsarethesortofthingthatcansoaffectprobabil-ities,andthisisdue,itseems,tothefactthattheyoccurattimes.Substancesdonotoccur(eventsinvolvingthemdo),andtheydonotappeartobethesortofthingthatcanaffectprobabilitiesintheindicatedway.Thisconsideration,althoughnotdecisive,seemstocountagainstthepossibilityofcausationbyasubstance.16.3.4TheexistencequestionEvenifoneoranotheroftheselibertarianviewscharacterizeswellthefreedomthatwevalue,andevenifwhatthataccountcharacterizesissomethingthatispossible,thequestionremainswhetherthereisgoodevidencethatwhatispositedbythataccountactuallyexists.Andtheanswerseemstobenegative.Libertarianaccountsrequire,first,thatdeterminismbefalse.Butmorethanthis,theyrequirethattherebeindeterminismofacertainsort(e.g.,withsomeeventsentirelyuncaused,ornon-deterministicallycaused,orcausedbyagentsandnotdeterministicallycausedbyevents)andthatthisindeterminismbeloc-atedinspecificplaces(generally,intheoccurrenceofdecisionsandotherac-tions).Whatisourevidencewithregardtotheserequirements’beingsatisfied?Itissometimesclaimedthatourexperiencewhenwemakedecisionsandactconstitutesevidencethatthereisindeterminismoftherequiredsortinthere-63quiredplace.Wecandistinguishtwopartsofthisclaim:one,thatindecidingandacting,thingsappeartoustobethewaythatoneoranotherlibertarianaccountsaystheyare,andtwo,thatthisappearanceisevidencethatthingsarein64factthatway.Somecompatibilistsdenythefirstpart.Butevenifthisfirstpartiscorrect,thesecondpartseemsdubious.Ifthingsaretobethewaytheyaresaidtobebysomelibertarianaccount,thenthelawsofnature–lawsofphysics,65chemistry,andbiology–mustbeacertainway.Anditisincrediblethathowthingsseemtousinmakingdecisionsandactinggivesusinsightintothelawsofnature.Ourevidencefortherequiredindeterminism,then,willhavetocomefromthestudyofnature,fromnaturalscience.Thescientificevidenceforquantummechanicsissometimessaidtoshowthatdeterminismisfalse.Quantumtheoryisindeedverywellconfirmed.However,thereisnothingapproachingaconsensusonhowtointerpretit,onwhatitshowsuswithrespecttohowthingsareintheworld.Indeterministicaswellasdeter-ministicinterpretationshavebeendeveloped,butitisfarfromclearwhetherany66oftheexistinginterpretationsiscorrect.Perhapsthebestthatcanbesaidhereisthat,giventhedemiseofclassicalmechanicsandelectromagnetictheory,thereisnogoodevidencethatdeterminismistrue.Theevidenceisevenlessdecisivewithrespecttowhetherthereisthekindofindeterminismlocatedinexactlytheplacesrequiredbyoneoranotherlibertarianaccount.Unlessthereisacompleteindependenceofmentaleventsfromphysical391\nRandolphClarkeevents,thenevenforfreedecisionstherehastobeindeterminismofaspecificsortatspecificjuncturesincertainbrainprocesses.Therearesomeinteresting67speculationsintheworksofsomelibertariansabouthowthismightbeso;butourcurrentunderstandingofthebraingivesusnoevidenceonewayortheotheraboutwhetheritisinfactso.Atbest,itseemswemustremain,forthetimebeing,agnosticaboutthismatter.Iflibertarianfreedomrequiresagentcausation,andifsuchathingispos-sible,thatisanotherrequirementaboutwhichwelackevidence.Indeed,itisnotclearthattherecouldbeanyempiricalevidencefororagainstthisaspectof68agent-causalviews.16.4ConclusionTheissuesofwhetherfreewilliscompatiblewithdeterminismandwhetherwehavefreewillhaveusuallybeentakentobeall-or-nothingmatters:foreachquestion,ithasbeenassumed,theanswerwillbeyesorno.Butourinterestinfreedomstemsfromourconcernforavarietyofthings.Thecontrolthatisrequiredforsomeofthesethings,orforsomeinterestingversionofsomeofthem,maybecompatiblewithdeterminism(andwithevent-causalindetermin-ism),whilewhatisrequiredforothersmaynotbe;wemayhavesomeofthesethings,orsomeinterestingversionofsomeofthem,butnotothers.WearenotcontrolledbyneuroscientistssuchasNina,andmostofusarequitefreefromcompulsionsandaddictions.Ourrecognitionofreasonsfitsintoquitecompre-hensiblepatterns,andwearenotradicallyoutoftouchwithreality.Whocandenythatwethereforehavecertainvaluablevarietiesofcontrol,givingusacertaindegreeofdignity.Eveniftheabilitytodootherwiseisnotcompatiblewithdeterminism,wehaveseenreasontothinkthatsuchanabilityisnotrequiredforresponsibility.Andevenifcertainaspectsofresponsibilityarestillunderminedbydeterminism(orbyevent-causalindeterminism),otheraspectsofitmaynotbe.Actionscanbeattributedtoagentsevenifdeterminismistrue,anditmaystillbeappropriatetoadoptcertainsortsofreactiveattitude(suchasresentment)towardandtoprotectourselvesfromoffendersevenifnooneeverdeservesoneroustreatmentinreturnforwrongdoing.Further,evenifdeterminismistrue,inactingwegenerallymakeadifference,inoneway,tohowtheworldgoes,evenifwedonotmakeadifferenceinanotherway.Indeliberatingandmakingdecisions,too,wemakeadifference,inonewayatleast,evenifweare,unfortunately,subjecttoanillusionwheneverwedeliberate.Ifinfactwehavesomebutnotallofthethingsforthesakeofwhichwevaluefreewill,thenthewayofwisdomistorecognizethisfactandacceptit.Todosoistoescapeanexcessivepessimism.Butitistorejectboththeviewthatsomedeflatedvarietyoffreedomisallthatweeverwantedinthefirstplaceaswellas392\nFreedomoftheWilltheobstinateconviction,intheabsenceofevidence,thatwehavethemost69robustfreedomthatwecanimagine.NotesIamgratefultoCharlesCross,JohnMartinFischer,RobertKane,AlfredMele,andBruceWallerforhelpfulcommentsonearlierdraftsofthischapter.1Makingadecisionisacting;itisperformingamentalaction.Idistinguishithereforemphasis.Amongouractions,decisionsseemtobeespeciallyimportantasdeliberateexercisesofouractivecontrol.2Ithasalsobeenarguedthatdivineforeknowledgewouldprecludeourhavingfreewill,andsomeoftheargumentsofferedforthisviewarestructurallysimilartosomethatareofferedfortheviewthatdeterminismisincompatiblewithfreewill.SeeFischer(1994:chs.1–6)foradiscussionthathighlightstheseparallels.Givenspaceconstraints,weshallfocushereontheallegedthreatofdeterminism.3AthoroughdiscussionofdeterminismcanbefoundinEarman(1986).Thoughpartsofthebookaresomewhattechnical,chapter2providesanexcellentandaccessibleintroductiontotheissue.AnothercarefuldiscussionmaybefoundinvanInwagen(1983:2–8,58–65).4Forfurtherdiscussionofthedistinctionbetweendeterminismanduniversalcausa-tion,seevanInwagen(1983:2–5)andEarman(1986:5–6).5VanInwagen(1983:16).6HereIfollowtheargumentsetoutinvanInwagen(1983:93–105).Otherargu-mentsforincompatibilism,allofwhichmayfairlybeviewedasversionsoftheConsequenceArgument,areadvancedinWiggins(1973),Lamb(1977),vanInwagen(1983:68–93),andGinet(1990:ch.5).Forgeneraldiscussionofthesearguments,seeFischer(1983,1988,and1994:chs.1–5),Flint(1987),Vihvelin(1988),Kapitan(1991),Hill(1992),andO’Connor(2000:ch.1).Discussionsofspecificaspectsoftheargumentsarereferencedinthefollowingnotes.7ConsiderwhatvanInwagensays.Using“P0”forour“H,”hewrites:ThepropositionthatP0isapropositionabouttheremotepast.Wecould,ifwelike,stipulatethatitisapropositionaboutthedistributionandmomentaofatomsandotherparticlesintheinchoate,presiderialnebulae.Therefore,surely,noonehasanychoiceaboutwhetherP0.ThepropositionthatLisaproposi-tionthat“records”thelawsofnature.Ifitisalawofnaturethatangularmomentumisconserved,thennoonehasanychoiceaboutwhetherangularmomentumisconserved,and,moregenerally,sinceitisalawofnaturethatL,noonehasanychoiceaboutwhetherL.(1983:96)8Gallois(1977)andNarveson(1977)arerepresentativesofthisposition;theirpapersarefollowed,inthesamevolume,byresponsesfromvanInwagen.ThediscussionthereconcernsasomewhatdifferentversionoftheConsequenceArgument;Ihaveadaptedcertainclaimssothattheyapplytotheversionunderconsiderationhere.Iborrowthename“Multiple-PastsCompatibilists”(aswellas“Local-MiracleCompatibilists”–seethetextbelow)fromFischer(1994:ch.4).393\nRandolphClarke9Lewis(1981)isaproponentofthisposition.Again,hisdiscussionisdirectedatadifferentversionoftheConsequenceArgument,andIhavemadethenecessaryadaptationsinsomeofhisclaims.Lewis’sviewsarediscussedinHorgan(1985),Fischer(1988and1994:ch.4),andGinet(1990:111–17).10Theplausibilityofdenying(6)–NL–maydependinpartonwhatlawsofnatureare,inparticular,onwhethertheyinvolveanyirreduciblenecessitation.Fordefenseofcompatibilismbyappealtoanon-necessitarianviewoflaws,seeSwartz(1985:ch.10)andBerofsky(1987:esp.chs.8and9).11Therehasbeenextensivediscussionofthevalidityof(β).See,forexample,Slote(1982),Fischer(1983,1986b,and1994:ch.2),Widerker(1987),Vihvelin(1988),O’Connor(1993),Kapitan(1996),McKayandJohnson(1996),Carlson(2000),andCrispandWarfield(2000).12TheexampleisadaptedfromWiderker(1987:38–9).13Fortwosuchproposals,seeWiderker(1987)andO’Connor(1993).14AninferencerulewiththeoperatorunderstoodinthiswayisrecommendedbyMcKayandJohnson(1996).15Foranargumentthataruleofthissortisvalid,seeCarlson(2000:286–7).16SeeFischer(1994)andFischerandRavizza(1998).17See,forexample,Blumenfeld(1971),Naylor(1984),Stump(1990,1996,1999a,and1999b),Rowe(1991:82–6),Widerker(1991,1995a,1995b,and2000),Haji(1993and1998:ch.2),Lamb(1993),Zimmerman(1993),Fischer(1994:ch.7,1995,and1999:109–25),FischerandHoffman(1994),Ginet(1996),Hunt(1996and2000),Kane(1996:40–3and142–3),WiderkerandKatzoff(1996),Copp(1997),McKenna(1997),Wyma(1997),DellaRocca(1998),MeleandRobb(1998),Otsuka(1998),Goetz(1999),O’Connor(2000:18–22and81–4),Vihvelin(2000),andPereboom(2001:ch.1).18ForargumentsthatdeterminismprecludesresponsibilitythatdonotrelyonPAP,seevanInwagen(1983:161–88and1999).Fordiscussion,seeFischer(1982),Heinaman(1986),Warfield(1996),FischerandRavizza(1998:ch.6),Stump(2000),andStumpandFischer(2000).19Lehrer(1997:ch.4)presentsanotherrecenthierarchicalaccount.Forathoroughdiscussionofsuchviews,seeShatz(1986).20ThispointwasfirstraisedbyWatson(1975:218).Frankfurtacknowledgesitwhenhewrites:Themerefactthatonedesireoccupiesahigherlevelthananotherinthehierarchyseemsplainlyinsufficienttoendowitwithgreaterauthorityorwithanyconstitutivelegitimacy.Inotherwords,theassignmentofdesirestodiffer-enthierarchicallevelsdoesnotbyitselfprovideanexplanationofwhatitisforsomeonetobeidentifiedwithoneofhisowndesiresratherthanwithanother.(1987:166)21Hewrites:[A]personmaybecapriciousandirresponsibleinforminghissecond-ordervolitionsandgivenoseriousconsiderationtowhatisatstake.Second-ordervolitionsexpressevaluationsonlyinthesensethattheyarepreferences.Thereis394\nFreedomoftheWillnoessentialrestrictiononthekindofbasis,ifany,uponwhichtheyareformed.(1971:note6)Andfurther,“thequestionsofhow[anagent’s]actionsandhisidentificationswiththeirspringsarecausedareirrelevanttothequestionsofwhetherheperformstheactionsfreelyorismorallyresponsibleforperformingthem”(1975:122).Frankfurtdoesmaintainthat“itisonlyinvirtueofhisrationalcapacitiesthatapersoniscapableofbecomingcriticallyawareofhisownwillandofformingvolitionsofthesecondorder”(1971:17).Weshallconsiderbelowcompatibilistaccountsthatemphasizetherequirementofacapacityforpracticalreasoningandrationalaction.22FordiscussionofthisproblemfacedbyFrankfurt’saccount(andbyothersimilarviews),seeFischerandRavizza(1998:ch.7).Mele(1995:ch.9)arguesthatanadequatecompatibilistaccountmustplacesomerequirementsonthehistoryofanagent’sattitudes.23Notethatfreedominperformingcertainactionshasnowbeenaccountedforintermsofthemakingofcertaindecisions.ThoughFrankfurtsuggeststhatnodecisioncanbe“external”totheagent,plainlydecisionscanbeunfree.Hence,someaccountisneededofthefreedomofthedecisionsthatarenowappealedto.However,sinceFrankfurtlaterdropstheappealtodecisions,weneednotpursuethispoint.24Theambivalencethatisopposedtowholeheartedness,henotes,“cannotbeover-comevoluntaristically.Apersoncannotmakehimselfvolitionallydeterminate,andtherebycreateatruthwheretherewasnonebefore,merelybyan‘actofwill.’Inotherwords,hecannotmakehimselfwholeheartedjustbyapsychicmovementthatisfullyunderhisimmediatevoluntarycontrol”(1992:10).Anyrolefordecisionsinanagent’sconstitutingheridentity,then,isseverelydownplayed.Bratman(1996)faultsFrankfurtfordenyingthatdecisionhasacrucialroletoplayinidentification,andhedevelopsaviewthatcombinesdecisionandwholeheartedness(or,ashecallsit,satisfaction).Hisviewisnotadvancedasanaccountoffreeaction,butifitweretobeadaptedforthatpurpose,then,asnotedabove,somethingwouldhavetobesaidaboutthefreedomoftherequireddecisions.25See,forexample,Frankfurt(1987:note13).26Waller(1993)developsthisobjection.27Wallace(1994)andWolf(1990)advancecapacityaccounts.Mele(1995)offersacompatibilistviewthatappealstotheagent’scurrentrationalcapacitiesandtothehistoryofhermentalattitudes.28Guidancecontrolisheldtosufficeforthe“freedom-relevant”componentofmoralresponsibility.FischerandRavizza(1998:26)recognizethattheremaybeothertypesofrequirement(suchasanepistemicorknowledgerequirement)forrespons-ibility.29FischerandRavizzacalltherequiredtypeofresponsiveness“moderatereasons-responsiveness,”distinguishingit(ibid.:chs.2and3)fromaweakerandastrongervarietythattheydescribe.30Thisreactivityrequirementmaybetooweak.Consideranagent,Karla,whoroutinelyhasacompulsivedesiretodoacertaintypeofthing(e.g.,acompulsivedesiretosteal).Karlamaybeappropriatelyreceptivetoreasons;shemaybedisposedtorecog-nizeanunderstandablepatternofreasonsfornotstealing,includingmoralreasonstorefrain.AnditmaybethatKarla,likemanyakleptomaniac,wouldrefrainforsome395\nRandolphClarkegoodreason,forexample,iftherewereapoliceofficerwatchingher;henceshemaysatisfythereactivenessrequirement,evenifthereisnoothertypeofsituationinwhichshewouldbemovedbyreasonsnottosteal.Butwhen,withnopoliceofficerinthevicinity,shesteals,sheisbehavingcompulsively,movedbyacompulsivedesire,andsheisnotincontrolofwhatshedoesinthemannerthatisrequiredformoralresponsibility.Thereactivitytoreasonsthatisrequiredforresponsibility,then,appearstobegreaterthanthatrequiredonthisaccount.(AnobjectionofthistypeisraisedinMele(2000).)FischerandRavizzamightobjectthatinthesituationinwhichKarlarespondstothepresenceofthepoliceofficer,themechanismthatoperatesisnotthesameastheonethatoperateswhenhercompulsivedesiretostealmoveshertosteal,andhencethatthemechanismthatproduceshertheftsdoesnotcountassufficientlyreasons-responsiveontheirview.(Seetheirdiscussion(1998:74)ofacaseinwhichacertaintypeofreasongivesanagentmore“energyorfocus.”)Butiftheysorespond,thenweneedtoknowmoreabouthowtodistinguishmechanisms.Otherwise,themovehereappearsadhoc.ItmightalsobeobjectedthatinKarla’scase,thesecondrequirementforguidancecontrol–thatthemechanismbetheagent’sown–isnotmet.(Thisrequirementisdiscussedinthetextbelow.)Hereitcanbesaidbriefly,inresponse,thatsuchownershipissaidtobeamatteroftheagent’shavingcertainattitudesaboutherself,andthereappearstobenoreasonwhyKarlacouldnothavetherequiredattitudes.31OnFischerandRavizza’sview,then,anagentmusthavecertainbeliefsaboutherselfifsheistoactwiththefreedomrequisiteformoralresponsibility.GalenStrawson(1986)agrees,holdingthatbelievingthatoneisafreeagentisanecessaryconditionofbeingafreeagent.32ThecasepresentedinthetextsuggeststhatanagentmaysatisfyalltherequirementsofFischerandRavizza’sviewbutnotbemorallyresponsible.Adifferentkindofcase(describedbyAlfredMeleinconversation)suggeststhatanagentmaybemorallyresponsiblebutfailtosatisfytherequirementsofthisview.SupposethatSamocca-sionallyactsakratically:sometimeshejudgesonecourseofactionbestbut,becausehisdesiretodosomethingdifferentisstrongest(hasthegreatestmotivationalstrength),hedoessomethingdifferent.SeeingthatSamhasthisproblem,awell-meaninggroupofneuroscientistssurreptitiouslyimplantsinhisbrainacomputerchipthatfunctionsinthefollowingway:wheneverSamjudgesacertaincourseofactionbest,thechipensuresthathisdesiretopursuethatcourseofactionisstrongest.Allthatthechipdoes,then,istohelpSamovercomehisweaknessofwillandactashejudgesbest.Suchassistance,evenifSamisunawareofit,neednoteliminateSam’sresponsibilityforhisbehavior.ButitappearsthatitwouldonFischerandRavizza’sview.Atleastintheperiodimmediatelyfollowingtheimplantation,themechanismthatoperateswhenthechipcontributestotheproductionofSam’sbehaviorwouldbeadifferenttypeofmechanismfromanyforactionsproducedbywhichSamhastakenrespons-ibility,andsoitwouldappearnottobehisownmechanism.33ItmaybethoughtthatAllen’sunawarenessofNina’sinfluencerendershistakingresponsibilityforhisactionsnotappropriatelybasedontheevidence.However,asFischerandRavizzarecognize,torequirefullknowledgeofthemechanismsbywhichouractionsareproducedwouldbetorequiretoomuch,fortherearenumerouscausalinfluencesonourbehaviorofwhichweareroutinelyunaware.Theevidential396\nFreedomoftheWillrequirementmaybesatisfied,then,byanagent(suchasAllen)whoisunawareofcertainfeaturesofthemechanismbywhichhisactionisproduced.AsFischerandRavizzaputit:“whenonetakesresponsibilityforactingfromakindofmechanism,itisasifonetakesresponsibilityfortheentireiceberginvirtueofseeingthetipoftheiceberg”(1998:216–17).34NotethattheinfluencesofwhichAllenisunawarearethedeliberateinterventionsofanotherintelligentagent,whereasinfluencesofwhichwearetypicallyunawarecomefromunthinkingcauses.ButitisdoubtfulthatthisdifferencecanaccountforAllen’sunfreedom.Indeed,wemayimagineavariationofhiscaseinwhichsomeinanimateobjectplaysaroleparallel,inrelevantrespects,tothatofNina.Suppose,forexample,thatthroughoutAllen’slife,wheneverheacts,MraysemittedbyameteoriteinMongoliahappen(bycoincidence)tohavejusttheeffectonhimandhisbehaviorthatitwaspreviouslysupposedNina’sinterventionshave.Again,itisnotclearthatanyrequirementsofFischer’sandRavizza’sviewareviolated,butitseemsdoubtfulthatherewehaveanagentactingfreelyandonewhoisresponsibleforwhathedoes.35AsFischerandRavizzasay(inresponsetoasimilardefenseraisedbyFrankfurtagainstasimilarobjection):“Continuousmanipulationiscompatiblewithcontinuityandintelligibility.Whetheranagent’shistoryiscontinuousorepisodicinitscontentisquiteadifferentmatterfromwhetheritisinternallyorexternallygenerated”(1998:198–9).36Thediscussionscitedinnote18abovepursuethisstrategy.37TwocompatibilistswhotakethisapproachareWallace(1994)andScanlon(1998:ch.6).38Aswillbeexplainedbelow,viewsofthisthirdtypeholdthatcausationbyanagentdoesnotconsistincausationbyevents.39Foradissentingview,seeCoverandO’Leary-Hawthorne(1996).Theyarguethatacertaintypeoflibertarianview–anagent-causalview–fitsmorecomfortablywithdualistviewsofpersonsandthemental.40Non-causalaccountsarealsoadvancedbyMcCall(1994:ch.9),Goetz(1997),andMcCann(1998).41Ginet’saccountofnon-basicactionsandparticularlyofgenerationisrathercomplic-ated.Interestedreadersshouldexaminehis(1990:ch.1).42Theexpression“inanappropriateway”isincludedheretoruleoutwhatiscalled“deviant”or“wayward”causation.Proponentsofcausaltheoriesofactionholdthatactionsaredistinguishedbythefactthattheyarecausedbyagent-involvingeventsofcertaintypes.Butitisrecognizedthatabodilymovementmaybecausedbyeventsoftherightsortsandyetfailtobeanactionifthecausalpathwayisdeviantorwayward.Fordiscussionofthisproblemandproposedsolutions,seeDavidson(1973:153–4),Brand(1984:17–30),Bishop(1989:chs.4and5),andMeleandMoser(1994).43Velleman(1992:466,note14)consequentlyobjectsthat,onGinet’sview,theactishphenomenalqualitythateverybasicactionissaidtopossessismisleading,illusory.However,Ginettakeshisdescriptionoftheexperienceonehasinactingtobemetaphorical;theexperience,heholds,doesnotliterallyrepresenttotheagentthatsheisbringingabouttheeventinquestion.44SeeAudi(1986)forasophisticatedcausalaccountofactingforacertainreason.45Ginetclaims(1990:143)thatconditionsofthissortaresufficientforthetruthofanexplanationthatcitesadesire.Butheseemstoregardthem(oratleasthavingthe397\nRandolphClarkerelevantconcurrentintention)asnecessaryaswell.Forhemaintains(ibid.:145)thatadesirethattheagenthasthatisareasonforperformingacertainactionandofwhichtheagentisawarewhensheactswillfailtobeareasonforwhichtheagentactsifshedoesnothavetherelevantconcurrentintention.46ThisobjectionisdevelopedinMele(1992:250–5).47Foraccountsofnon-deterministiccausation,see,forexample,Lewis(1973[1986]:postscriptB),Tooley(1987:289–96),andEells(1991).48Relativelysimpleevent-causallibertarianviewsofthissortaresketchedbyWiggins(1973),Sorabji(1980:ch.2),andvanInwagen(1983:137–50).Asimilarview,thoughwiththeadditionalrequirementthatatleastsomefreedecisionsbe“self-subsuming”(self-explaining),isadvancedbyNozick(1981:294–316).49ArgumentsfromluckareadvancedbyHaji(1999)andMele(1999aand1999b).50Dennett(1978),Mele(1995:ch.12),andEkstrom(2000:ch.4)offerevent-causallibertarianviewsonwhichindeterminismisrequiredonlyatearlierstagesofthedeliberativeprocess.Ontheirviews,itisallowedthatsomeundeterminedeventsinthedeliberativeprocesscausallydetermineafreedecision.Forcriticaldiscussionofsuchviews,seeClarke(2000).51ForthisdiscussionofKane’sview,IdrawfromClarke(1999).52IassumeherethatIsabelle’sdecisioniswhatKanecallsa“self-formingaction,”anactionthatisnotcausallydeterminedbyanypriorevents,andhenceonethefreedomofwhichdoesnotderivefromthefreedomofearlierfreeactionsthatcausallydeter-mineit.Itmayneverthelessbethecase,onKane’sview,thatthefreedomofaself-formingactionderivesfromthefreedomofaneffortofwillthatnon-deterministicallycausesit.Thispointwillbediscussedlaterinthissection.53ThisrecentproposalcomesinresponsetoanobjectionraisedbyMele(1999a:98–9and1999b:279).54Thisdoublingofeffortsofwillintroducesatroublingirrationalityintotheaccountoffreedecision.Thereisalreadypresent,inacaseofmoralstruggle,anincoherenceintheagent’smotives;butthistypeofconflictisfamiliarandnoapparentthreattofreedom.However,tohavetheagentactivelytrying,atonetime,todotwoobvi-ouslyincompatiblethings–thingssuchthatitisobviouslyimpossiblethatshedoboth–raisesseriousquestionsabouttheagent’srationality.Thisadditionalincoher-encemaythusbemoreofathreatthananaidtofreedom.55Thetaskofprovidingsuchanaccountmightbedelayedbyholdingthattheseeffortsareindirectlyfree,derivingtheirfreedomfromthatofearlierfreeactions.Butthismaneuverwouldnotevadetheproblemraisedhere.Thequestionwouldremainwhytheaccountofthefreedomofthoseearlieractionscouldnotbeapplieddirectlytothedecisionthatresultsfromtheeffortofwill.56Infact,manycontemporarycompatibilists(see,forexample,Fischer(1999:129–30))holdthatthecontrolthatsufficesforresponsibilityiscompatiblewithnon-deterministicaswellasdeterministiccausationofdecisionsandotheractions.Iftheindeterminismrequiredbytheevent-causallibertarianaccountdiminishescontrol,thesecompatibilistsaccept,itdoesnotdosototheextentthatitunderminesresponsibility.Itisworthnotingaswellthatnon-deterministiccausationdoesnotconstitutewhathasbeencalleddeviantorwaywardcausation.Forthelatterconcernstherouteorpathwayofacausalprocess,andnon-deterministiccausationmayfollowthesamepathwayasdeterministiccausation.398\nFreedomoftheWill57Although,asKanepointsout(withtheexampleofthemanwhobreakstheglasstabletop),evenhereindeterminismneednotdiminishcontroltotheextentthattheagentisnotresponsibleforproducingtheresult.58Someagent-causaltheoristsholdnotthatafreeactioniscausedbyanagentbutthatanagent’scausingacertaineventisafreeaction.Thisdifferencewillnotbearonourconsiderationshere.59Inrecentyears,agent-causalaccountshavebeenadvancedbyChisholm(1966,1971,1976a,1976b,and1978),Taylor(1966and1992),Thorp(1980),Zimmerman(1984),Donagan(1987),Rowe(1991),Clarke(1993and1996),andO’Connor(1995a,1996,and2000).60See,forexample,Thorp(1980:106)andTaylor(1992:53).61ThisobjectionstemsfromBroad(1952:215).ItisraisedaswellbyGinet(1990:13–14).62Eventhough,onthistypeofagent-causalview,afreeactionisnon-deterministicallycausedbyeventsinvolvingtheagent,sincetheagentmakesafurthercausalcontribu-tiontowhatshedoesinadditiontothecontributionmadebythoseevents,itwouldseemthatsheexercisesgreaterpositivepowersofcontrolthanwhatcouldbeexer-cisedifallcauseswereevents.(Fordiscussionofthispoint,seeClarke(1996:27–30).)Hencethistypeofviewmayhaveastrongerdefenseagainsttheproblemofluckthanhavenon-deterministicevent-causalaccounts.Morewouldhavetobesaid,however,toestablishthatthisdefenseisthoroughlyadequate.63Campbell(1957:168–70)andO’Connor(1995a:196–7)appealtothisexperienceasevidenceforlibertarianfreewill.64See,forexample,Mele(1995:135–7).65Thisissoforovert,bodilyactionsregardlessoftherelationbetweenmindandbody,anditissofordecisionsandothermentalactionsbarringacompleteindependenceofmentaleventsfromphysical,chemical,andbiologicalevents.66Forabriefandaccessiblediscussionoftheseissuesastheybearontheoriesoffreewill,seeLoewer(1996).Inadditiontosurveyingsomeofthemoreprominentinterpretationsofquantummechanics,Loewerarguesthatlibertarianismrequiresthatsomeeventslackobjectiveprobabilities.Manylibertarianswouldrejectthatclaim.67See,forexample,Kane(1996:128–30and137–42)andthesourcescitedthere.68Foradissentingopinion,seePereboom(2001:ch.3),whoarguesthatwenowhaveevidenceagainsttheexistenceofagentcausation.69Forcarefuldiscussionoftheimplicationsofourlackingfreewill(orsomevaluablevarietyoffreedom),seeHonderich(1988:part3),Smilansky(2000),andPereboom(2001).ReferencesAudi,Robert(1986).“ActingforReasons.”ThePhilosophicalReview,95:511–46.Berofsky,Bernard(1987).FreedomfromNecessity:TheMetaphysicalBasisofResponsibility.NewYork:RoutledgeandKeganPaul.Bishop,John(1989).NaturalAgency:AnEssayontheCausalTheoryofAction.Cam-bridge:CambridgeUniversityPress.Blumenfeld,David(1971).“ThePrincipleofAlternatePossibilities.”JournalofPhilosophy,67:339–44.399\nRandolphClarkeBrand,Myles(1984).IntendingandActing:TowardaNaturalizedActionTheory.Cam-bridge,MA:BradfordBooks.Bratman,Michael(1996).“Identification,Decision,andTreatingasaReason.”PhilosophicalTopics,24(2):1–18.Broad,C.D.(1952).EthicsandtheHistoryofPhilosophy.London:RoutledgeandKeganPaul.Campbell,C.A.(1957).OnSelfhoodandGodhood.London:GeorgeAllenandUnwin.Carlson,Erik(2000).“IncompatibilismandtheTransferofPowerNecessity.”Noûs,34:277–90.Chisholm,RoderickM.(1966).“FreedomandAction.”InKeithLehrer(ed.),FreedomandDeterminism.NewYork:RandomHouse:11–44.——(1971).“ReflectionsonHumanAgency.”IdealisticStudies,1:33–46.——(1976a).“TheAgentasCause.”InMylesBrandandDouglasWalton(eds.),ActionTheory.Dordrecht:D.Reidel:199–211.——(1976b).PersonandObject:AMetaphysicalStudy.LaSalle,IL:OpenCourt.——(1978).“CommentsandReplies.”Philosophia,7:597–636.Clarke,Randolph(1993).“TowardaCredibleAgent-CausalAccountofFreeWill.”Noûs,27:191–203.ReprintedinO’Connor(ed.)(1995b):201–15.——(1996).“AgentCausationandEventCausationintheProductionofFreeAction.”PhilosophicalTopics,24(2):19–48.ReprintedinabbreviatedforminPereboom(ed.)(1997):273–300.——(1999).“FreeChoice,Effort,andWantingMore.”PhilosophicalExplorations,2:20–41.——(2000).“ModestLibertarianism.”PhilosophicalPerspectives,14:21–46.Copp,David(1997).“DefendingthePrincipleofAlternatePossibilities:BlameworthinessandMoralResponsibility.”Noûs,31:441–56.Cover,J.A.andO’Leary-Hawthorne,John(1996).“FreeAgencyandMaterialism.”InJordanandHoward-Snyder(eds.)(1996):47–71.Crisp,ThomasM.andWarfield,TedA.(2000).“TheIrrelevanceofIndeterministicCounterexamplestoPrincipleBeta.”PhilosophyandPhenomenologicalResearch,61:173–84.Davidson,Donald(1973).“FreedomtoAct.”InHonderich(ed.)(1973):139–56.ReprintedinDavidson,EssaysonActionsandEvents.Oxford:ClarendonPress(1980):63–81.DellaRocca,Michael(1998).“Frankfurt,FischerandFlickers.”Noûs,32:99–105.Dennett,DanielC.(1978).“OnGivingLibertariansWhatTheySayTheyWant.”InDennett,Brainstorms:PhilosophicalEssaysonMindandPsychology.Montgomery,VT:BradfordBooks:286–99.Donagan,Alan(1987).Choice:TheEssentialElementinHumanAction.London:RoutledgeandKeganPaul.Earman,John(1986).APrimeronDeterminism.Dordrecht:D.Reidel.Eells,Ellery(1991).ProbabilisticCausality.Cambridge:CambridgeUniversityPress.Ekstrom,LauraWaddell(2000).FreeWill:APhilosophicalStudy.Boulder,CO:WestviewPress.Fischer,JohnMartin(1982).“ResponsibilityandControl.”JournalofPhilosophy,79:24–40.ReprintedinFischer(ed.)(1986a):174–90.——(1983).“Incompatibilism.”PhilosophicalStudies,43:127–37.——(ed.)(1986a).MoralResponsibility.Ithaca:CornellUniversityPress.400\nFreedomoftheWill——(1986b).“PowerNecessity.”PhilosophicalTopics,14(2):77–91.——(1988).“FreedomandMiracles.”Noûs,22:235–52.——(1994).TheMetaphysicsofFreeWill:AnEssayonControl.Oxford:Blackwell.——(1995).“LibertarianismandAvoidability:AReplytoWiderker.”FaithandPhilosophy,12:119–25.——(1999).“RecentWorkonMoralResponsibility.”Ethics,110:93–139.Fischer,JohnMartinandHoffman,Paul(1994).“AlternativePossibilities:AReplytoLamb.”JournalofPhilosophy,91:321–6.Fischer,JohnMartinandRavizza,Mark(eds.)(1993).PerspectivesonMoralResponsibility.Ithaca:CornellUniversityPress.——(1998).ResponsibilityandControl:ATheoryofMoralResponsibility.Cambridge:CambridgeUniversityPress.Flint,ThomasP.(1987).“CompatibilismandtheArgumentfromUnavoidability.”Jour-nalofPhilosophy,84:423–40.Frankfurt,HarryG.(1969).“AlternatePossibilitiesandMoralResponsibility.”JournalofPhilosophy,66:828–39.ReprintedinFischer(ed.)(1986a):143–52;inFrankfurt(1988):1–10;andinPereboom(ed.)(1997):156–66.——(1971).“FreedomoftheWillandtheConceptofaPerson.”JournalofPhilosophy,68:5–20.ReprintedinFischer(ed.)(1986a):65–80;inFrankfurt(1988):11–25;inPereboom(ed.)(1997):167–83;andinWatson(ed.)(1982):81–95.——(1975).“ThreeConceptsofFreeAction.”ProceedingsoftheAristotelianSociety,Supplementaryvol.49:113–25.ReprintedinFischer(ed.)(1986a):113–23;andinFrankfurt(1988):47–57.——(1976).“IdentificationandExternality.”InAmélieO.Rorty(ed.),TheIdentitiesofPersons.Berkeley:UniversityofCaliforniaPress:239–51.ReprintedinFrankfurt(1988):58–68.——(1987).“IdentificationandWholeheartedness.”InFerdinandSchoeman(ed.),Responsibility,Character,andtheEmotions:NewEssaysinMoralPsychology.Cambridge:CambridgeUniversityPress:27–45.ReprintedinFischerandRavizza(eds.)(1993):170–87;andinFrankfurt(1988):159–76.——(1988).TheImportanceofWhatWeCareAbout.Cambridge:CambridgeUniversityPress.——(1992).“TheFaintestPassion.”ProceedingsandAddressesoftheAmericanPhilosophicalAssociation,66:5–16.Gallois,André(1977).“VanInwagenonFreeWillandDeterminism.”PhilosophicalStud-ies,32:99–105.Ginet,Carl(1990).OnAction.Cambridge:CambridgeUniversityPress.——(1996).“InDefenseofthePrincipleofAlternativePossibilities:WhyIDon’tFindFrankfurt’sArgumentConvincing.”PhilosophicalPerspectives,10:403–17.Goetz,Stewart(1997).“LibertarianChoice.”FaithandPhilosophy,14:195–211.——(1999).“StumpingforWiderker.”FaithandPhilosophy,16:83–9.Haji,Ishtiyaque(1993).“AlternativePossibilities,MoralObligation,andMoralRespons-ibility.”PhilosophicalPapers,22:41–50.——(1998).MoralAppraisability:Puzzles,Proposals,andPerplexities.NewYork:OxfordUniversityPress.——(1999).“IndeterminismandFrankfurt-typeExamples.”PhilosophicalExplorations,2:42–58.401\nRandolphClarkeHeinaman,Robert(1986).“IncompatibilismwithoutthePrincipleofAlternativePos-sibilities.”AustralasianJournalofPhilosophy,64:266–76.ReprintedinFischerandRavizza(eds.)(1993):296–309.Hill,ChristopherS.(1992).“VanInwagenontheConsequenceArgument.”Analysis,52:49–55.Honderich,Ted(ed.)(1973).EssaysonFreedomofAction.London:RoutledgeandKeganPaul.——(1988).ATheoryofDeterminism:TheMind,Neuroscience,andLife-Hopes.Oxford:ClarendonPress.Horgan,Terence(1985).“CompatibilismandtheConsequenceArgument.”PhilosophicalStudies,47:339–56.Hunt,David(1996).“FrankfurtCounterexamples:SomeCommentsontheWiderker–FischerDebate.”FaithandPhilosophy,13:395–401.——(2000).“MoralResponsibilityandUnavoidableAction.”PhilosophicalStudies,97:195–227.Jordan,JeffandHoward-Snyder,Daniel(eds.)(1996).Faith,Freedom,andRationality:PhilosophyofReligionToday.Lanham:RowmanandLittlefield.Kane,Robert(1996).TheSignificanceofFreeWill.NewYork:OxfordUniversityPress.——(1999a).“OnFreeWill,ResponsibilityandIndeterminism.”PhilosophicalExplorations,2:105–21.——(1999b).“Responsibility,Luck,andChance:ReflectionsonFreeWillandIndeter-minism.”JournalofPhilosophy,96:217–40.——(2000).“ResponsestoBernardBerofsky,JohnMartinFischerandGalenStrawson.”PhilosophyandPhenomenologicalResearch,60:157–67.Kapitan,Tomis(1991).“HowPowerfulAreWe?”AmericanPhilosophicalQuarterly,28:331–8.——(1996).“IncompatibilismandAmbiguityinthePracticalModalities.”Analysis,56:102–10.Lamb,JamesW.(1977).“OnaProofofIncompatibilism.”ThePhilosophicalReview,86:20–35.——(1993).“EvaluativeCompatibilismandthePrincipleofAlternatePossibilities.”Jour-nalofPhilosophy,90:517–27.Lehrer,Keith(1997).SelfTrust:AStudyofReason,KnowledgeandAutonomy.Oxford:ClarendonPress.Lewis,David(1973).“Causation.”JournalofPhilosophy,70:556–67.ReprintedwithpostscriptsinLewis(1986):159–213.——(1981).“AreWeFreetoBreaktheLaws?”Theoria,47:113–21.ReprintedinLewis(1986):291–8.——(1986).PhilosophicalPapers,vol.II.NewYork:OxfordUniversityPress.Loewer,Barry(1996).“FreedomfromPhysics:QuantumMechanicsandFreeWill.”PhilosophicalTopics,24(2):91–112.McCall,Storrs(1994).AModeloftheUniverse.Oxford:ClarendonPress.McCann,HughJ.(1998).TheWorksofAgency:OnHumanAction,Will,andFreedom.Ithaca:CornellUniversityPress.McKay,ThomasJ.andJohnson,David(1996).“AReconsiderationofanArgumentagainstCompatibilism.”PhilosophicalTopics,24(2):113–22.McKenna,Michael(1997).“AlternativePossibilitiesandtheFailureoftheCounterexampleStrategy.”JournalofSocialPhilosophy,28:71–85.402\nFreedomoftheWillMele,AlfredR.(1992).SpringsofAction:UnderstandingIntentionalBehavior.NewYork:OxfordUniversityPress.——(1995).AutonomousAgents:FromSelf-ControltoAutonomy.NewYork:OxfordUni-versityPress.——(1999a).“Kane,Luck,andtheSignificanceofFreeWill.”PhilosophicalExplorations,2:96–104.——(1999b).“UltimateResponsibilityandDumbLuck.”SocialPhilosophyandPolicy,16:274–93.——(2000).“ReactiveAttitudes,Reactivity,andOmissions.”PhilosophyandPhenomeno-logicalResearch,61:447–52.Mele,AlfredR.andMoser,Paul(1994).“IntentionalAction.”Noûs,28:39–68.Mele,AlfredR.andRobb,David(1998).“RescuingFrankfurt-StyleCases.”ThePhilo-sophicalReview,107:97–112.Narveson,Jan(1977).“CompatibilismDefended.”PhilosophicalStudies,32:83–7.Naylor,MargeryBedford(1984).“FrankfurtonthePrincipleofAlternatePossibilities.”PhilosophicalStudies,46:249–58.Nozick,Robert(1981).PhilosophicalExplanations.Cambridge,MA:BelknapPress.O’Connor,Timothy(1993).“OntheTransferofNecessity.”Noûs,27:204–18.——(1995a).“AgentCausation.”InO’Connor(ed.)(1995b):173–200.——(ed.)(1995b).Agents,Causes,andEvents:EssaysonIndeterminismandFreeWill.NewYork:OxfordUniversityPress.——(1996).“WhyAgentCausation?”PhilosophicalTopics,24(2):143–58.——(2000).PersonsandCauses:TheMetaphysicsofFreeWill.NewYork:OxfordUniversityPress.Otsuka,Michael(1998).“IncompatibilismandtheAvoidabilityofBlame.”Ethics,108:685–701.Pereboom,Derk(ed.)(1997).FreeWill.Indianapolis:Hackett.——(2001).LivingWithoutFreeWill.Cambridge:CambridgeUniversityPress.Rowe,WilliamL.(1991).ThomasReidonFreedomandMorality.Ithaca:CornellUniversityPress.Scanlon,T.M.(1998).WhatWeOwetoEachOther.Cambridge,MA:BelknapPress.Shatz,David(1986).“FreeWillandtheStructureofMotivation.”MidwestStudiesinPhilosophy,10:451–82.Slote,Michael(1982).“SelectiveNecessityandtheFree-WillProblem.”JournalofPhilosophy,79:5–24.Smilansky,Saul(2000).FreeWillandIllusion.Oxford:ClarendonPress.Sorabji,Richard(1980).Necessity,Cause,andBlame:PerspectivesonAristotle’sTheory.Ithaca:CornellUniversityPress.Strawson,Galen(1986).FreedomandBelief.Oxford:ClarendonPress.Stump,Eleonore(1990).“Intellect,Will,andthePrincipleofAlternatePossibilities.”InMichaelD.Beaty(ed.),ChristianTheismandtheProblemsofPhilosophy.NotreDame:UniversityofNotreDamePress:254–85.ReprintedinFischerandRavizza(eds.)(1993):237–62.——(1996).“LibertarianFreedomandthePrincipleofAlternativePossibilities.”InJordanandHoward-Snyder(eds.)(1996):73–88.——(1999a).“AlternativePossibilitiesandMoralResponsibility:TheFlickerofFreedom.”TheJournalofEthics,3:299–324.403\nRandolphClarke——(1999b).“Dust,Determinism,andFrankfurt:AReplytoGoetz.”FaithandPhilosophy,16:413–22.——(2000).“TheDirectArgumentforIncompatibilism.”PhilosophyandPhenomenologicalResearch,61:459–66.Stump,EleonoreandFischer,JohnMartin(2000).“TransferPrinciplesandMoralResponsibility.”PhilosophicalPerspectives,14:47–56.Swartz,Norman(1985).TheConceptofPhysicalLaw.Cambridge:CambridgeUniversityPress.Taylor,Richard(1966).ActionandPurpose.EnglewoodCliffs:Prentice-Hall.——(1992).Metaphysics,4thedn.EnglewoodCliffs:Prentice-Hall.Thorp,John(1980).FreeWill:ADefenceAgainstNeurophysiologicalDeterminism.London:RoutledgeandKeganPaul.Tooley,Michael(1987).Causation:ARealistApproach.Oxford:ClarendonPress.VanInwagen,Peter(1983).AnEssayonFreeWill.Oxford:ClarendonPress.——(1999).“MoralResponsibility,Determinism,andtheAbilitytoDoOtherwise.”TheJournalofEthics,3:341–50.Velleman,J.David(1992).“WhatHappensWhenSomeoneActs?”Mind,101:461–81.Vihvelin,Kadri(1988).“TheModalArgumentforIncompatibilism.”PhilosophicalStudies,53:227–44.——(2000).“Freedom,Foreknowledge,andthePrincipleofAlternatePossibilities.”CanadianJournalofPhilosophy,30:1–23.Wallace,R.Jay(1994).ResponsibilityandtheMoralSentiments.Cambridge,MA:HarvardUniversityPress.Waller,BruceN.(1993).“ResponsibilityandtheSelf-madeSelf.”Analysis,53:45–51.Warfield,TedA.(1996).“DeterminismandMoralResponsibilityareIncompatible.”PhilosophicalTopics,24(2):215–26.Watson,Gary(1975).“FreeAgency.”JournalofPhilosophy,72:205–20.ReprintedinFischer(ed.)(1986a):81–96;andinWatson(ed.)(1982):96–110.——ed.(1982).FreeWill.Oxford:OxfordUniversityPress.Widerker,David(1987).“OnanArgumentforIncompatibilism.”Analysis,47:37–41.——(1991).“Frankfurton‘OughtImpliesCan’andAlternativePossibilities.”Analysis,51:222–4.——(1995a).“LibertarianFreedomandtheAvoidabilityofDecisions.”FaithandPhilosophy,12:113–18.——(1995b).“LibertarianismandFrankfurt’sAttackonthePrincipleofAlternativePossib-ilities.”ThePhilosophicalReview,104:247–61.——(2000).“Frankfurt’sAttackonthePrincipleofAlternativePossibilities:AFurtherLook.”PhilosophicalPerspectives,14:181–202.Widerker,DavidandKatzoff,Charlotte(1996).“AvoidabilityandLibertarianism:AResponsetoFischer.”FaithandPhilosophy,13:415–21.Wiggins,David(1973).“TowardsaReasonableLibertarianism.”InHonderich(ed.)(1973):31–61.Wolf,Susan(1990).FreedomwithinReason.NewYork:OxfordUniversityPress.Wyma,Keith(1997).“MoralResponsibilityandLeewayforAction.”AmericanPhilosophicalQuarterly,34:57–70.Zimmerman,MichaelJ.(1984).AnEssayonHumanAction.NewYork:PeterLang.——(1993).“Obligation,ResponsibilityandAlternatePossibilities.”Analysis,53:51–3.404\nTheBlackwellGuidetoPhilosophyofMindIndexEditedbyStephenP.Stich,TedA.WarfieldCopyright©2003byBlackwellPublishingLtdIndexacaciatree/antelopeexample161–2anti-reductionism:conceptual13,18–19;Adams,Fred159,161,162ontological13–17;teleofunctions332Adler,D.247arachidonicacid327affectprograms294–5,301Aristotle14,231n4,257–8agency:causality389–92,397n42,Armstrong,David16,25,51,58,90399n62;ConsequenceArgumentartificialintelligence309,317,346372–4;environment256–7;freedomassociativelearning324–5,328ofwill378,399n62;interventionsassumptions274,275,277–8397n34;moraldecision-makingastronomers182–3299–300,386;moralresponsibilityatomicrepresentationalism:cognitive379–80,396n31,396n32;non-causalrepresentations172;compositionalityviews382–3;wholeheartedness377–8,181;Gödelnumerals183–7;395n24systematicityofthought176–8,aggression297179–80Aizawa,K.159,161,162Atomism:Conceptual202–5,206,207;Alexander,Samuel18content201–2;eliminativism40–1n83;Allen,Colin161Fodor203;individuation203–4alternatepossibilitiesprinciple(PAP)attention340–1,343–4375–6,394n18attitudes4–5,289,290;propositionalAMPAreceptors3274–5,55,57analyticity207–8,209n9attributetheory85,98n3angels365automaticappraisalmechanism303anger297autonomy10,75–6,215–16,261animals:behavior293,297;conditioning151–2;humananimals354,362–4;bachelorconcept191,192motivation297;thought363Baker,L.160Annas,J.35n32batrachians175–6antelope/acaciatreeexample161–2behavior:animals293,297;appetitiveanti-individualism256–7,260293;consciousness138n26;folkanti-materialism56psychology29;linguistic145–6;anti-physicalists79–81mentalstates11;rationality314–15405\nIndexbehaviorism:analytical49–50;criterial21;390–1;explanation225–6;feelings288;innateness293;instinctfunctionalism22;interactionism15;291–2,293;logical21,37n58;mentalevents217–18,226–8,230–1;Pavlovianconditioning151–2;nexus125;physicalstate77–8,82n14;philosophy49–50,236;psychologyphysicalism81;reference15748–50;ritual292;semantics309–10;centralnervoussystem325translational21cerebralhemispheres353;transplantedbelief:ascriptions58–9;desire4–6,361–255,58–9,242,289,290;discrepantChalmers,D.J.57,111,135n1,137–8n21,249–51;false4,32n5;Fodor317;138n26,229–30functional111–12;Hume128;chaostheory35–6n46intentionality3–4;knowledge191,Chomsky,Noam260–1,262192;perseverance249–50;post-eventChurchland,P.M.:connectionism317,interventions250–1;publicpolicy318;contentsimilarity210–11n17;implications250–1eliminativism28–9,60,61;folkBerkeley,George16psychology238;luminescence136n13;Bickle,John348n7,348n8neuroscience346;Quinean123;state-Bierbrauer,G.246spacesemantics200biogeography175–6Churchland,P.S.346;connectionismbiologicalsystems54201;consciousness119;emotionBlack,Max137n19316;neuroscience60–1,322,337,Blackburn,Simon99n13346Block,Ned22,23,38n66,56,137n17Clark,A.318,319body1,216,365;seealsodualism;Clarke,Randolph390mind–bodyproblemClassicalTheory:analyticity209n9;brain295,329concepts191,193,209n5,209n7;braindamage315–16primitiveconcepts209n5;psychologybrain-imagingtechniques333–4195,205;typicalityeffects194Bratman,Michael395n24classicism144,172–4,182,313Brentano,Franz3,55,57Clifford,W.K.16bridgelaws19,37–8n62,75,87co-consciousness97Broad,C.D.:emergentism18,19,125;coevolutionaryresearchideology329–30againstmechanism136n9;mentalcognition:computationalapproach272–3;states102;TheMindanditsPlaceindecision-making242,243;FodorNature102–3;vitalism110173–4;Pylyshyn173–4;systematicityBrooks,Rodney313172,178–80;unconsciousinferenceBurge,Tyler259,260,265–6,26732n11Buss,David296cognitivepsychology288–9,347cognitivescience261,346;emotionsCampbell,F.W.277302–3,305;individualism262;socialCaramazzo,Alfonso347psychology246–7carburetorexample203collapseprinciples125–6,138n25;Carnap,R.48–9,136n16,258–9quantummechanics127,138n27categorization191,198–9combinatorialism,functional172,183–7causality:agency389–92,397n42,commissurotomy353399n62;consciousness126;common-sensepsychology:seefolkcontingency24;events24,384–9,psychology406\nIndexcompatibilism:determinism375,381;ConsequenceArgument:agency372–4;Frankfurtcases375–6;hierarchicaldeterminism370–2;event-causalaccount376–8;local-miracle373;libertarianview388;inferencerule373multiple-pasts373;responsibilitycontent:atomism201–2;disjunctive398n56278–9;externalism334;identity200;CompatibilityQuestion370–4individualism265–6;intentionalcomplexsystems12279–80;meaning153;names156–7;compliance246narrow207,210n12,211n20,279;compositionality175,180–1,182similarity200,201;thought143,computationalapproach275–6;cognition146–7,165–6,219;wide210n12272–3;individualism276;Marr’scontingency24,73,74theoryofvision273–4;mentalstatescontinuity:identity353,354–5;mental263;psychology261361,362,363,364;psychological360computertechnology22Cosmides,Leda262,296conceivabilityarguments105–6,119counterfactuals97–8,100n15concepts4;acquisition192,204;“crackdow”example278atomism202–5;autonomy10;Classicalcreationism176Theory191,193,209n5,209n7;culture294definitions192–3;DualTheory197;Cummins,R.54,159–60explanatoryrole203;mental8–9,28–9;nativism204;phenomenal118;Damasio,Antonio293,302–3,304,315physical118;primitive209n5;Dancy,J.192properties7;PrototypeTheory195–8;Darwin,Charles175,288,290–1,302psychology208;structure206–7,Davidson,Donald:anomalousmonism210n13;theoryof190–118–19,59–60;intentionalstates217;connectionism:activationvalues38n63;skepticism282;supervenience33n19;Churchland201;classicism313;Swampman152inferences175,317;rationality312;DeAngelis,G.C.339robotics315,319deceptioncues240consciousness:attention340–1,343–4;decision-making242,243,319,369,attributetheory98n3;behavior386,393n1138n26;causalrole126;conceivabilitydeducibility137n17argument105–6;Descartes2;Democritus40–1n83epistemicgap107–8,118;experienceDennett,D.C.:artificialintelligence103–4;explanatoryargument104–5,346;beliefs/desires58–9;115;functionalism109,136n12,229;heterophenomenology111;Huxley19;identity113,114;psychologicalexplanation54;Quineanintentionality5,34n30;knowledge123;realpatternsofreality89;vitalismargument106–7;materialistsolution110104;mentalstates3,5;metaphysicsDescartes,René:causalinteractionism15;102–3,135n1;naturalworld102;cogito92;consciousness2;dualismneurophysiology323;phenomena13–15,47–8,85,98n2,124–5,231,102–3,122;physicalism65–6,119,235–6;epiphenomenalism15,86;God333–4;privacy85;problemswith14;mentalcausation230–1;mind28,103–4;qualia55–7;reductionism216;motion231n1;phantompain104,135n4;tokenreductions89;214–15;sensation143;substance14,truth12386,214;thought14,31n3407\nIndexdescription91,122,271–2intentionality225;mentalstates25,descriptivism257,25828–9,252n6;mind–bodyproblem13;desires:ascriptions58–9;attributionrejectionof61;representationalism246–9;belief4–6,55,58–9,242,187–8n1;type-Amaterialism109289,290;first-order377;predictionElizabethofBohemia,Princess215,216247–8;simulationtheory248–9emergentism:bridgelaws37–8n62;Broaddeterminism381–2;compatibilism375,18,19,125,127;emergentmaterialism381;ConsequenceArgument370–2;18,36n51;epiphenomenalism19;Earman393n3;freedomofwill96–7,neutral18370,392–3;libertarianaccounts391;emotions58;braindamage315–16;responsibility388–9,394n18cognitivescience302–3,305;culturedisjunctionproblem147–8,153–4,294;Darwin288;Ekman294–5,298;158–61ethology291–3;evaluativejudgmentsdisplacementactivity292289;evolutionarytheories290–1,295,disposition32n10296,297,302;experience288;facialdivineforeknowledge393n2expressions290–1,294,298;feelingdoppelgangers257,269,278–91,215,288,302–3;folkpsychologydoubleaspecttheory16–17297,298,304;gametheory296;double-occupancyview361–2mood304;moralagency299–300;Dretske,Fred:externalism334;historicalneuroculturaltheory300;philosophyinstantiationcondition164–5;288–9;rationality315–16;socialinformationtheory166n10;knowledgeconstructions299,300–1;sociobiology147;learningperiod149,151,158–9;295–6;transactionaltheory297–9;meaning153–6;misrepresentationtwin-pathwaymodels303;universality150;names157;Swampman162–3;301–2teleology54;thought/symbols145,endogenesis337,338146,162;uninstantiatedproperties157endowmenteffect247DualTheory197,199,206–7,210n12entities66–7,90,216,222–3dualism35n38;bundle85–6,92–5;environment256–7Descartes13–15,47–8,85,98n2,epiphenomenalism:Descartes15,86;124–5,231,235–6;modal118;dualismtype-E127–9;emergentismpredicate86,87–9;property36–7n54,19;intuition139n29;non-reductivism127;substance85–6,98n2,124–5,134;physicalism19127;type-D124–7,134;type-E124,epistemicgap:consciousness107–8,118;127–9,134;type-O133;seealsomaterialism112–13,114,119–21mind–bodyproblemepistemology74–5error193–4,318Earman,John393n3essentialism198–9Egan,F.268,271–4,275,278–9ethology291–3Ekman,Paul:automaticappraisalevaluativejudgments289mechanism303;deceptioncues240;event-causallibertarianview384–9,390displayrules299;emotions294–5,events31n4;anomalousmonism36n49;298;evolutionaryexplanation296;causalrelations24,384–9,390–1;neuroculturaltheory300;universalitymental2;non-physical2,78,80–1ofemotions301–2evolution176eliminativism:atomism40–1n83;entitiesevolutionarybiology264222–3;folkpsychology91,238,241;evolutionarypsychology296408\nIndexevolutionarytheoriesofemotions290–1,systematicity/compositionality175,295,296,297,302180–1;teleology54;TwinEarth159;existence354,391–2unicorns158;uninstantiatedpropertiesexperience20;consciousness103–4;157;vacuityoftheory164–5decision-making319;emotion288;folkpsychology:behavior29;Churchlandhumananimals363;mentalepisodes60–1;description91;eliminativism91,50;pain6,217;qualities227–8,238,241;emotion297,298,304;231n2;representation231n2functionalism237;individualism273;explanation54,154,163,225–6,296Lewis51,237;mentalstates24,60,explanatoryargument104–5,115,217241;mindreading239,251;philosophyexplanatoryrole,concepts203237–8;platitudes239,240;simulationextension14theory241;thought145externalism:assumptions274;contentform/meaning309–11334;Dretske334;functionalism27–8;four-dimensionalism362individualism257;metaphysics264–5;Frank,RobertA.295–6Newsome338;phenomenal334,338,Frankfurt,HarryG.375–8,387,340;reductionism28;self-knowledge394–5n21,394n20281–2,285;social260freedomofwill:agency378,399n62;extras,problemof70–1decision-making369;determinism96–7,eyewitnesstestimony250–1370,392–3;divineforeknowledge393n2;Frankfurt376–7;libertarianfacialexpressions290–1,294,298accounts381–2;mind–bodyproblemFarkas,Katalin99n13,99n1410;self-governance383feelings1,215,288,302–3;hybrid289,Frege,Gottlob257290Fregecases146Feigl,Herbert19,25Fridlund,Alan298–9,304fetuses,early-term356,363functionalcombinatorialism172,183–7Feyerabend,Paul40–1n83,60functionalmagneticresonanceimagingfinitestatemachines22346Fischer,JohnMartin379–80,393n2,functionalspecifier231–2n5395n29,396n30,396n32functionalism251n2;analytic21–2,23,fission,PersistenceQuestion361–224–5,109;Aristotle231n4;causalFodor,Jerry:asymmetricaldependencyrole22;consciousness109,136n12,160–1,165;atomism203;beliefs317;229;counterexamples56;externalismclassicism172–4;cognition173–4;27–8;Fodor346;folkpsychology237;cognitiverepresentations178–80;homuncular53–4;identitytheoryconceptnativism204;functional231–2n5;individualism273;machinecombinatorialism183;functionalism52–3,58;meaning237,240–1;mental346;historicalinstantiationconditionstates23–4,238–9;physicalism262–3;157–8,159,164–5,183;identitypropositionalattitudes57;Putnam22,theory52;language145–6,147;220,346;Pythagoreans37n61;learningperiod149;meaning150–3;realizability232n5;supervenience9;methodologicalsolipsism260–1;teleological58;zombies56names156–7;prototypes197;psychosemantics57;rationality310,Gage,Phineas315–16311,319;“SpecialSciences”87–8;Gallistel,Randy347Swampman163;symbols162;Gallois,André393n8409\nIndexgameconcept192idealism15–16,133gametheory296identity:bodilycriterion365;“gant”example158consciousness113,114;content200;Gazzaniga,M.334,345,346,347continuity354–5;counterfactual94,generalizations,psychological20096,100n15;deduced114;diachronicGettier,Edmund192353;empathyfailure95–6;memoryghostinthemachine99n8,236353;non-physical75;numerical93,Gibson,J.J.261355–6,361;persistence355–8;Ginet,Carl382–3,397–8n45,397n41,personal86,352–5;personhood397n43353–4;philosophy352;physicalstateglutamate327113,353;psychology352;synchronicGod/gods14,123,365353;time92,353;tokenidentityGödelnumerals183–7,188n1736n48,52–3,56,72,81n11,231n3;Godfrey-Smith,P.164twinsexample94–5;typeidentity56,gold19472,232n3;vague/partial93–4Goldman,A.248identitytheory:Fodor52;functionalismGordon,R.248–9,252n8231–2n5;instrumentalism58–60;grandmotherconcept196–7Kripke137n19;Lewis,C.I.40n74;Greenspan,Patricia289Lewis,D.24,27,51,76;mind50–2;Grice,H.P.146,147,148pain52;properties230;psychophysicalguidancecontrol379,395n2821,24–6,76–7;Putnam52imprinting149–50haecceitas93–4,96incompatibilism369,374,393n6Hare,R.M.33n19inconsistency282–3Harris,P.244indeterminism385,387,388Hart,W.D.57indication147,148,155Haugeland,J.310indiscernabilityclaims33–4n20Hauser,Mark299individualism:anti-individualism256–7;Hawkins,R.D.327,328Burge259;cognitivescience262;Hebb,D.O.325computationalapproach276;Heil,John228,282externalism257;folkpsychology273;Heinroth,Oskar292functionalism273;locational276;hemispherectomy361mentalcontent265–6;physicalismheterophenomenology111261,262–3;Putnam260;rejectionHildreth,Ellen266265;representationalprimitives270–1;Hill,Christopher118,137n21self-knowledge280;vision267–8Hinde,Robert293,297,298individuation203–4,263–4,334historicalinstantiationcondition152,inference:abductive317,319;153,157–8,159,164–5,183conjunctionfallacy252–3n12;homology302connectionism175,317;Consequencehomuncularfunctionalism53–4Argument373;logical174;predictionhope289244–5;rational311,312,317;Horgan,T.61systematicity173–8humananimals354,362–4information147,148,166n4,166n10Hume,David35n45,85–6,125,128,information-processing266–7354innatebehavior293Huxley,T.H.18,19inner-causethesis59410\nIndexinstinct291–2,293language:behavior145–6;communityofinstrumentalism58–60259–60;evolutionaryperspective347;intentionality:belief3–4;consciousness5,Fodor145–6,147;naturalization34n30;content279–80;eliminativism146–9;seealsonaturallanguage225;mentalstates5,32n8,32n9,languageofthought144–6,161–238–9n71,85,217–19,232n9;learning:associative324–5,328;physicalism65;problemswith57–8;centralnervoussystem325;neuronrepresentation270–1;teleology54–5differentiation349n12interactionism:causal15;Chalmerslearningperiod149,151,158–9138n26;dualismtype-D134;LeDoux,Joseph303microphysics124;physics125,Leibniz,G.W.16,39–40n73138n26;rejected126Leibniz’sLaw56interdisciplinarity323,345,347Leucippus40–1n83internalism257Levine,Joseph118,217interpretations138n25Lewes,G.H.17interventions397n34Lewis,C.I.37n58,40n74introspection283,285Lewis,David97;folkpsychology51,intuition92,112,137n21,139n29237;identitytheory24,27,51,76;invertedspectrum23pain39n72,51;theoryofmind58irrealism13,28–9libertarianaccounts:agent-causality389–92;determinism391;event-Jackendoff,R.262causality384–9;existencequestionJackson,Frank56,106–7391–2;freedomofwill381–2;non-jadeexample148,149–50,168n43causalviews382–3James,William16,20,132,261,288linguistics261jealousy296Llinás,Rodolfo337deJong,H.Looren329,330,331,332Loar,Brian118,137n21Locke,John50,354,356,357,359–60,Kahneman,D.252–3n12366Kandel,E.R.324,327,328,349n12Loewenstein,G.247Kane,Robert385–6,398n52Loftus,E.250–1Kant,Immanuel16logicalbehaviorism21,37n58Katz,J.209n9logicalpositivism34n23,48,87Kim,Jaegwon70Lorenz,K.292–3,297Kitcher,P.61Lormand,Eric200kittenexample210–11n17LTP:seepotentiation,long-termknowledge:aposteriori8;apriori8;luck385–7belief191,192;consciousness106–7;luminescence136n13disjunctionproblem147–8;DretskeLycan,WilliamG.57,60147;experiential9–10Kosslyn,Stephen346McAdams,C.J.340,341–4Kripke,Saul:God123;identitytheorymacaques340–4137n19;materialismtype-B114,McClintock,Barbara333115–18;names40n78,193,258;McGinn,C.32n7,119,139n33naturalkindterms40n78,193,202;Mach,Ernest20tableexample92–3;type/tokenmachinefunctionalism52–3,54,58identity56McKinsey,M.283–5411\nIndexMcLaughlin,B.36n51,175,183Mervis,Carolyn194,196MacLean,PaulD.295metaphysics16,102–3,135n1,264–5Maddell,Geoffrey94methodologicalsolipsism260–1Manfredi,P.161methodologyinneuroscience329–31manifestationcondition32n10mice/shrewsexample159–60Marr,David265,266,277–8,313microphysics124,133,137n17Marr’stheoryofvision265,266,267–8;microstimulationstudies337,338–9computationalapproach273–4;Eganmiddletemporalcortexarea334–7,271–4;perception276–7;Segal338–9268–71,278Milgram,S.246Martin,C.B.228,231Millikan,R.G.54match-to-sampletask341–4mind:dependence163;Descartes28,materialism:centralstate25–6;216;entities216;identitytheory50–2;consciousness104;emergent18,theoryof58;seealsodualism36n51;epistemicarguments108;mind–bodyproblem10–12,31n1,epistemicgap112–13,114,119–21;216–17;continentaldivide29,31;explanatoryargument104–5;false107,Descartes235–6;eliminativism13;108;reductive35n41;type-A108–12,logicalspaceofsolutions30;properties123,136n16;type-B112–19,123,7–8;reduction8–9;supervenience136n15;type-C119–229–10;seealsodualismMatthews,R.268mindreading:deceptioncues240;desireMaunsell,J.H.R.340,341–4attribution246–9;folkpsychologymeaning149–50;asymmetrical239,251;information-richaccountsdependency165;content153;Dretske243–4;simulationtheory241–2,153–6;Fodor150–3;form309–11;243–4,245–6,252n8functionalism237,240–1;informationmisrepresentation150166n4;natural147,148,151,153–4;modularitytheory239naturallanguageterms257;semanticmoleculargenetics332–3promiscuity161–2;verification48,monism:anomalous18–19,36n49,236–759–60;neutral13,20–1,35n45,131;mechanicalsystems13–14,54,310physicalism32,139;type-F117–18,Medin,D.198124,129–33,134,139n32memory325,328–9,353,359–60monkeys:macaques340–4;match-to-mentalepisodes/experience50sampletask341–4;middletemporalmentalparticletheory13cortexarea334–7,338–9mentalstates:attitudes289;attributeMoore,G.E.2theory85;behavior11;Broad102;moralagency299–300causality217–18,226–8,230–1;moralresponsibility379–80,396n31,computationalprocesses263;396n32consciousness3,5;continuity361,Morgan,C.Lloyd18,19362,363,364;eliminativism25,motion231n1,334–5,33728–9,252n6;folkpsychology24,motiondirectiontask34560,241;functionalism23–4,238–9;multiplespatialchannelstheory277intentionality5,32n8,32n9,38–9n71,85,217–19,232n9;pain221;physicalNagel,Ernest75state125;Putnam53;second-orderNagel,Thomas3,16,32n7,56,119,280;self-knowledge281136n7412\nIndexnames:causaltheoryofreference157;Pappas,G.60contents156–7;descriptivism258;parallelism15,16–17,86Dretske157;Fodor156–7;KripkeParfit,Derek99n13,36240n78,193,258;Putnam40n78,193Pavlovianconditioning151–2Narveson,Jan393n8Peacocke,Christopher209n9naturalkindterms40n78,88,114,193,Peirce,C.S.16202,258people,inorganic365naturallanguage144–5,168n35,257,perception5,276–7258–9PersistenceQuestion353;double-naturalselection337occupancyview361–2;fission361–2;naturalworld1,102identity355–8;Lockeanview357,necessitation108358;PsychologicalApproach358,necessity:aposteriori115;conceptual359–61,362–4,366–7;SimpleView8–9,70;metaphysical8,9–10,26,27,358–9;SomaticApproach358,364–670;nomological8,9,70persistentvegetativestate356–7,363neo-dualism57personhood353–4,356–7,364,366neuralnetworks200–1,311,312,322,phenomena:consciousness102–3,122;333–4mental2,65–6,235–6;physics12;neurobiologicalmechanisms60–1,324,reductionism135n4,136n13;states344133–4;straightforward/Gestalt89;neuroculturaltheory300teleology88neurons340–4,349n12phenomenalindividuals56–7neurophysiology61,323,334,340phenomenalism16neuroscience322–3,329–31,333phenomenology132neurotransmitters327–8philosophyofmind:artificialintelligenceNewell,A.311309;behaviorism49–50,236;emotionsNewsome,William334–7,338,345288–9;folkpsychology237–8;identityNisbett,R.250352;neuroscience322–3;physics66;NMDAreceptors327,330–1realism223–4non-causalviews,agency382–3physicalstate69;causalclosure77–8,non-mentalism1182n14;identity113,353;mentalnon-reductionism75,134state125;predicates8–9;qualitiesNoonan,H.3646–7,216–17Nordby,Knut136n8physicalism6–7;abstract/concreteentities66–7;causality81;O’Connor,Timothy390consciousness65–6,119,333–4;odometerexample274–5contingency73,74;counterexamplesOlson,EricT.36378–9;counterintuitive67–8;Ortony,A.198epiphenomenalism19;epistemology74–5;false71;functionalism262–3;pain:Armstrong51;experience6,217;individualism261,262–3;intentionalityidentitytheory52;introspection283;65;justifying76–8;logicalbehaviorismLewis,D.39n72,51;mentalstate221;21;mentalphenomena65–6;mentalphantom214–15;realizability221–2properties9;monism139n32;painpredicate224–5neuroscience333;non-reductionismpanprotopsychism117–18,131–2,13475;objections78–81;ontologypanpsychism16,35n41262–3;particulars71;predicate413\nIndexphysicalism(cont’d)reductionism99n8;unavoidabilitydualism86;properties71;rationality90–2;seealsocognitivepsychology;folk65;realizability68,72;reductionismpsychology;socialpsychology75;supervenience69–71,73–4,261psychoneuralinter-theoreticrelations329physics:closedsystems215;common-psychophysicalidentitytheory21,24–6,sensephysics198;entities90;76–7indeterminacy385;interactionism125,psychosemantics57138n26;phenomena12;philosophypublicpolicyimplications250–166;properties130Putnam,Hilary:functionalism22,220,Pinker,Steven347346;identitytheory52;individualismPlace,Ullin25,50260;jadeexample148;“TheMeaningplagueconcept199–200of‘Meaning’”257;mentalstates53;platetectonicsexample88methodologicalsolipsism260;namesplatitudes238–9,24040n78,193;naturalkindterms40n78,Plato35n33202,258–9;naturallanguage258–9;Poggio,Tomas266person/robot38n66;“PsychologicalPopper,Karl29Predicates”220;TwinEarthexamplepositronemissiontomography34657potentiation,long-term325–7,330–1,Pylyshyn,Zenon:classicism172–4;348n8cognition173–4;cognitivepredicates8–9,232n6representations178–80;functionalprediction244–5,247–8combinatorialism183;sensoryneuronsprincipleofalternatepossibilities375–6,262;systematicity/compositionality394n18180–1properties2,14;concepts7;dispositionalPythagoreans37n61228;dualism36–7n54,127;identitytheory230;mental7–8,9,10,13,16;qualia:consciousness55–7;individuationnarrow/wide57;non-mental34n24;334;materialismtype-A111,135n3;non-relational34n25;physical/mentalcausation226–8;motionphenomenal133;physicalism71;334–5,337;reification3physics130;protophenomenal123–4,qualities227–8,231n2;mental215,129–31;qualitative228;uninstantiated216–17,220;physical6–7,216–17157–8quantummechanics34n31,124,126,propositionalattitudes4–5,55,57391–2;collapsetheories127,138n27PrototypeTheory195–8,199,203,quarkexample130,203209n11quasi-memory360proximalprojections162Quine,W.V.123,193,195,207–8,psychologicalexplanation54345psychologicallaws34n21PsychologicalApproach,Persistenceradiumexample373–4Question358,359–61,362–4,366–7Raichles,Marcus346psychology:autonomy75–6;behaviorismrationality:behavior314–15;48–50;ClassicalTheory195,205;connectionism312;ecological318–19;cognitiverevolution288–9;emotions315–16;errors318;Fodorcomputationalrevolution261;concepts310,311,319;mechanical310,317,208;continuity360;evolutionary296;320;physicalism65;representationgeneralizations200;identity352;313–14414\nIndexRavizza,Mark379,380,395n29,Rosch,Eleanor194,196396n30,396n32Ross,L.250reactivity379,395–6n30Russell,Bertrand4–5,20,34n29,130,realism10,97–8,166n2,223–4257reality89,262Ryle,Gilbert21–2,32n10,48–9,realizability:functionalism232n5;136n16,236,288multiple22,68–9,88,217,221–3,232n5,329;pain221–2;physicalismSalzman,C.D.337,33968,72Schank,R.311reasons-responsiveness379–80Schopenhauer,Arthur16recognitionheuristic318–19Schouten,Maurice329,330,331,332reduction10,87,89,323–4,327Schrödingerevolution125,126reductionism88,89;conceptual13,Searle,John4,38n6520–8;consciousness104,135n4;eventSegal,G.268–71,278causality390;externalist28;naturalSejnowski,T.60–1kindterms88;neuroscience322–3,self35n45329–30;phenomenal135n4,136n13;self-formingaction398n52physicalism75;psychology99n8self-governance380,383reference,directtheory157self-knowledge:externalism281–2,285;Reichenbach,Hans345–6inconsistency282–3;individualism280;relativitylaws124introspection283,285;McKinseyRenaissancephilosophy16283–5;mentalstates281representation:abstraction190;Sellars,Wilfrid60,237,273,345algorithmiclevel266;cognitive172,semantics200–1,309–10178–80;combinatorialmental172;semicompatibilism375–6compositionality180–1;experiencesensation6,143231n2;functional111,121;sensoryneurons262implementationallevel266;sensoryperformance340–1intentionality270–1;internal314;serotonin327–8mental172;molecular172;sexualjealousy296phenomenal111;rationality313–14;Shapiro,James332–3reality262;Segal269–71Shapiro,L.274representationalprimitives270–1,280Shepherd,Gordon323–4,344,347representationalism138n23,187–8n1;Shoemaker,S.363seealsoatomicrepresentationalismShope,R.163,164Rescorla,R.A.325similarity200,201responsibility388–9,394n18,398n56;Simon,H.311seealsomoralresponsibilitySimpleView,PersistenceQuestion358–9retrocognition360simulationtheory:desiredetection248–9;Rey,Georges41n83folkpsychology241;mindreadingrhesusmonkeys:seemonkeys241–2,243–4,245–6,252n8ritual292singleunitapproach,neurophysiology334Robinson,W.S.57skepticism10,280–1,282robot/catexample155–6,160–1Skinner,B.99n8robots38n66,310,313,315,319,365Smart,J.J.C.25,50Robson,J.G.277Smolensky,Paul183Rorty,Richard40–1n83,60socialconstructions299,300–1415\nIndexsocialinteraction295taxonomy263–4socialpsychology249–50teleofunctions332,333sociobiology295–6teleology54–5,58,88,91solipsism260–1TensorProductTheory183SomaticApproach,PersistenceQuestionterritorialdisplays297358,364–6TheoryTheory198–200,202,210n13,somaticmarkers315–16,319239soul1,354thought:abstraction190;animalsspace,sentential311363;content143,146–7,165–6,sphericityexample228219;Descartes14,31n3;folkSpinoza,B.15,16–17psychology145;humananimalssplitpersonality353363–4;identical153;languageStalnaker,R.137n17of144–6,161–2;naturalworld1;Stampe,D.146private215;sensation143;state2,5;seealsomentalstates;physicalsymbols145,146,160,161,162;statessystematicity176–8,179–80;stereopsis277transitions311stereoscopicdepth339thoughtexperiments11,23,38n65,Stich,Stephen28–9,59,261136n7stimulus:conditioned/unconditioned327Tinbergen,Niko292,297Stoljar,D.132tokens36n48;identity36n48,52–3,56,story-understanding31172,81n11,231n3;reductionism88,Strawson,Galen57,396n3189;robust151Strawson,P.F.15Tomkins,SilvanS.295substance14,86,214Tooby,John262,296substancedualism85–6,98n2,124–5,127transactionaltheory297–9Summerfield,D.161“triunebrain”theory295supervenience33n19;Davidson33n19;truth108–9,121,123functionalism9;global70,71,73;Tulving,Endel347metaphysical10;mind–bodyproblemTuring,Alan499–10;multiple329;nomological10,TuringMachine309–11,31719;physicalism69–71,73–4,261;TuringTest49,50strong33–4n20,70Tversky,A.252–3n12sustainingmechanism,syndrome-basedTwinEarthexample334,338;204–5“crackdow”278;doppelgangers257,Swampman152,162–3278–9;Fodor159;meaning144;symbols145,146,160,161,162,Putnam57;water144,152–3,219,168n35,311258,278synapticplasticity324,325–7,328–9,twinsexample94–5,278–9330–1,348n2TwoFactorTheory210n12,211n20synaptictransmission348n2typeidentity56,72,231n3systematicity:artificialintelligence317;typicalityeffects194,196,206classicism182;cognitiverepresentations178–80;compositionality175,180–1;Ullman,S.274functionalcombinatorialism183–7;understanding259inference173–8;thought176–8,unicornexample157–8179–80universals90,99n10,301–2416\nIndexvacuityoftheory164–5Watson,Gary394n20VanGulick,R.54,119Watson,JohnB.288vanInwagen,Peter393n6,393n7Whitman,Walt1verification48,49,118,236–7wholeheartedness377–8,395n24vision:individualism267–8;informationWilson,E.O.12processing266–7;Marr265,266,Wilson,RobertA.265267–74,278Witmer,D.Gene71vitalism110Wittgenstein,L.40–1n83,143,192,289Woodward,J.61Wagner,A.R.325water:microphysics137n17;naturalkindZajonc,Robert303terms87–8,114,258;TwinEarthzero-crossings270,278144,152–3,219,258,278zombies56,105–6,119,228–30417