【剑桥哲学指南系列】生物学哲学 524页

thecambridgecompaniontoTHEPHILOSOPHYOFBIOLOGYThephilosophyofbiologyisoneofthemostexcitingnewareasinthefieldofphilosophyandonethatisattractingmuchattentionfromworkingscientists.ThisCompanion,editedbytwoofthefoundersofthefield,includesnewlycommissionedessaysbyseniorscholarsandbyup-and-comingyoungerscholarswhocollectivelyexaminethemainareasofthesubject–thenatureofevolutionarytheory,classification,teleologyandfunction,ecology,andtheprob-lematicrelationshipbetweenbiologyandreligion,amongothertopics.Up-to-dateandcomprehensiveinitscoverage,thisuniquevolumewillbeofinterestnotonlytoprofessionalphilosophersbutalsotostudentsinthehumanitiesandresearchersinthelifesciencesandrelatedareasofinquiry.DavidL.HullisanemeritusprofessorofphilosophyatNorthwesternUniversity.Theauthorofnumerousbooksandarticlesontopicsinsystematics,evolutionarytheory,philosophyofbiology,andnaturalizedepistemology,heisarecipientofaGuggenheimFoundationfellowshipandisaFellowoftheAmericanAcademyofArtsandSciences.MichaelRuseisprofessorofphilosophyatFloridaStateUniversity.Heistheauthorofmanybooksonevolutionarybiology,includingCanaDarwinianBeaChristian?andDarwinismandItsDiscontents,bothpublishedbyCam-bridgeUniversityPress.AFellowoftheRoyalSocietyofCanadaandtheAmericanAssociationfortheAdvancementofScience,hehasappearedontelevisionandradio,andhecontributesregularlytopopularmediasuchastheNewYorkTimes,theWashingtonPost,andPlayboymagazine.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nvolumesintheseriesofcambridgecompanions:ABELARDEditedbyjeffreye.browerandkevinguilfoyADORNOEditedbytomhunnAQUINASEditedbynormankretzmannandeleonorestumpHANNAHARENDTEditedbydanavillaARISTOTLEEditedbyjonathanbarnesAUGUSTINEEditedbyeleonorestumpandnormankretzmannBACONEditedbymarkkupeltonenSIMONEDEBEAUVOIREditedbyclaudiacardDARWINEditedbyjonathanhodgeandgregoryradickDESCARTESEditedbyjohncottinghamDUNSSCOTUSEditedbythomaswilliamsEARLYGREEKPHILOSOPHYEditedbya.a.longFEMINISMINPHILOSOPHYEditedbymirandafrickerandjenniferhornsbyFREUDEditedbyjeromeneuGADAMEREditedbyrobertj.dostalGALILEOEditedbypetermachamerGERMANIDEALISMEditedbykarlameriksGREEKANDROMANPHILOSOPHYEditedbydavidsedleyHABERMASEditedbystephenk.whiteHEGELEditedbyfrederickbeiserHEIDEGGER,2ndEditionEditedbycharlesb.guignonHOBBESEditedbytomsorellHUMEEditedbydavidfatenortonHUSSERLEditedbybarrysmithanddavidwoodruffsmithWILLIAMJAMESEditedbyruthannaputnamKANTEditedbypaulguyer(ContinuedafterIndex)CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTothememoryofErnstMayr,19042005CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTheCambridgeCompaniontoTHEPHILOSOPHYOFBIOLOGYEditedbyDavidL.HullNorthwesternUniversityMichaelRuseFloridaStateUniversityCambridgeCollectionsOnline©CambridgeUniversityPress,2008\ncambridgeuniversitypressCambridge,NewYork,Melbourne,Madrid,CapeTown,Singapore,Sa˜oPaulo,DelhiCambridgeUniversityPress32AvenueoftheAmericas,NewYork,NY10013-2473,USAwww.cambridge.orgInformationonthistitle:www.cambridge.org/9780521851282ªCambridgeUniversityPress2007Thispublicationisincopyright.Subjecttostatutoryexceptionandtotheprovisionsofrelevantcollectivelicensingagreements,noreproductionofanypartmaytakeplacewithoutthewrittenpermissionofCambridgeUniversityPress.Firstpublished2007PrintedintheUnitedStatesofAmericaAcatalogrecordforthispublicationisavailablefromtheBritishLibrary.LibraryofCongressCataloginginPublicationDataTheCambridgecompaniontothephilosophyofbiology/editedbyDavidHullandMichaelRusep.cm.Includesbibliographicalreferencesandindex.ISBN-13:978-0-521-85128-2(hardback)ISBN-13:978-0-521-61671-3(pbk.)1.Biology–Philosophy.I.Hull,DavidL.II.Ruse,Michael.III.Title.QH331.C2852007570.1–dc222006025898ISBN978-0-521-85128-2hardbackISBN978-0-521-61671-3paperbackCambridgeUniversityPresshasnoresponsibilityforthepersistenceoraccuracyofURLsforexternalorthird-partyInternetWebsitesreferredtointhispublicationanddoesnotguaranteethatanycontentonsuchWebsitesis,orwillremain,accurateorappropriate.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\ncontentsContributorspagexiPrefacexix1Adaptationtimlewens12PopulationGeneticsrobertal.millsteinandroberta.skipperjr.223UnitsandLevelsofSelectionelisabetha.lloyd444What’sWrongwiththeEmergentistStatisticalInterpretationofNaturalSelectionandRandomDrift?robertn.brandonandgrantramsey665Genepaule.griffithsandkarolastotz856InformationinBiologypetergodfrey-smith1037Reductionism(andAntireductionism)inBiologyalexanderrosenberg1208MechanismsandModelslindleydarden139viiCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nviiiContents9Teleologyandreariew16010Macroevolution,Minimalism,andtheRadiationoftheAnimalskimsterelny18211PhilosophyandPhylogenetics:HistoricalandCurrentConnectionsmaureenkearney21112HumanEvolution:TheThreeGrandChallengesofHumanBiologyfranciscoj.ayala23313VarietiesofEvolutionaryPsychologydavidj.buller25514Neurobiologyvaleriegrayhardcastle27515BiologicalExplanationsofHumanSexuality:TheGeneticBasisofSexualOrientationchristopherhorvath29116GameTheoryinEvolutionaryBiologyzacharyernst30417WhatIsan‘Embryo’andHowDoWeKnow?janemaienschein32418EvolutionaryDevelopmentalBiologymanfredd.laubichler34219MolecularandSystemsBiologyandBioethicsjasonscottrobert36120Ecologygregorym.mikkelson37221FromEcologicalDiversitytoBiodiversitysahotrasarkar388CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nContentsix22BiologyandReligionrobertt.pennock41023TheMoralGrammarofNarrativesinHistoryofBiology:TheCaseofHaeckelandNaziBiologyrobertj.richards429ReferenceList453Index497CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nCambridgeCollectionsOnline©CambridgeUniversityPress,2008\ncontributorsandreariewisanassociateprofessorofphilosophyattheUniversityofMissouri–Columbia.Hehaswrittenonvarioustopicsinthephilosophyofbiologyincludingteleology,innateness,fitness,andthestructureofnaturalselectionexplanations.franciscoj.ayalaisUniversityProfessorandDonaldBrenProfessorofBiologicalSciencesattheUniversityofCalifornia,Irvine.On12June2002,PresidentGeorgeW.BushawardedhimtheNationalMedalofScienceattheWhiteHouse.From1994to2001,AyalawasamemberoftheU.S.President’sCommitteeofAdvisorsonScienceandTechnology.HehasbeenpresidentandchairmanoftheboardoftheAmericanAssociationfortheAdvancementofScience(1993–96)andpresidentofSigmaXi,theScientificResearchSocietyoftheUnitedStates(2004–05).robertn.brandonisprofessorofphilosophyatDukeUniversity.HeistheauthorofAdaptationandEnvironmentandConceptsandMethodsinEvolutionaryBiology(CambridgeStudiesinPhilosophyandBiology).davidj.bullerisPresidentialResearchProfessorintheDepartmentofPhilosophyatNorthernIllinoisUniversity.HeistheauthorofAdaptingMinds:EvolutionaryPsychologyandthePersistentQuestforHumanNature(MITPress/Bradford,2005)andtheeditorofFunction,Selection,andDesign(SUNYPress,1999).lindleydardenisprofessorofphilosophyintheCommitteeforPhilosophyandtheSciencesandinthePrograminBehavior,Evolution,Ecology,andSystematicsattheUniversityofMaryland,xiCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nxiiContributorsCollegePark.SheistheauthorofReasoninginBiologicalDiscoveries:Mechanisms,InterfieldRelations,andAnomalyResolution(CambridgeUniversityPress,2006)andTheoryChangeinScience:StrategiesfromMendelianGenetics(OxfordUniversityPress,1991).SheservedaspresidentoftheInternationalSocietyforHistory,Philosophy,andSocialStudiesofBiologyin2001–03.SheandCarlF.CravercoeditedtheJune2005‘‘MechanismsinBiology’’issueofStudiesinHistoryandPhilosophyofBiologicalandBiomedicalSciences.zacharyernstisamemberoftheDepartmentofPhilosophyattheUniversityofMissouri–Columbia.Hisresearchisonevolu-tionarygametheory,theapplicationofgametheorytoevolutionarybiology,formallogic,andautomatedtheoremproving.petergodfrey-smithisprofessorofphilosophyatHarvardUniversity.HehasdegreesfromSydneyUniversityandtheUniversityofCalifornia,SanDiego,andworksmainlyinthephilosophyofbiologyandphilosophyofmind.HeistheauthorofComplexityandtheFunctionofMindinNature(1996)andTheoryandReality:AnIntroductiontothePhilosophyofScience(2003).paule.griffithsisARCFederationFellowandProfessorofPhilosophyattheUniversityofQueensland.Heistheauthor,withKimSterelny,ofthetextbookSexandDeath:AnIntroductiontothePhilosophyofBiology(Chicago,1999).From2002to2005heandKarolaStotzwereprincipalinvestigatorsoftheNationalScienceFoundation–fundedproject‘‘RepresentingGenes’’attheUniversityofPittsburgh.ThisinterdisciplinaryprojectusedWeb-basedsurveymethodstoexaminedifferingunderstandingsofthegeneincontemporarybioscience.valeriegrayhardcastleisprofessorofscienceandtechnol-ogyinsocietyandassociatedeanoftheCollegeofLiberalArtsandHumanSciencesatVirginiaTech.Herareaofresearchinterestliesattheintersectionofpsychology,psychiatry,neuroscience,andphilosophy.Shehaspublishednumerousarticlesandbooksontherelationbetweenneuroscientificdataandpsychologicaltheories,mostrecentlyTheMythofPain(MITPress)andUnderstandingBrainActivity(forthcoming).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nContributorsxiiichristopherhorvathisassociateprofessorofphilosophyandbiologicalsciencesatIllinoisStateUniversityinNormal.HeholdsaPh.D.inphilosophyfromDukeUniversity.ProfessorHorvathhaspublishedpapersonphylogeneticsystematics,evolutionarypsychology,homosexuality,theevolutionofhumansexuality,andgenderstudies.maureenkearneyisassociatecuratorintheDepartmentofZoologyandheadoftheDivisionofAmphibiansandReptilesattheFieldMuseumofNaturalHistory.SheisalsoamemberoftheCommitteeonEvolutionaryBiologyandlecturerintheBiologicalSciencesCollegiateDivisionattheUniversityofChicago.Herresearchfocusesontheevolution,comparativeanatomy,anddevelopmentofreptiles,usingaphylogeneticperspective.Sheisalsointerestedinthetheoryandmethodsofphylogeneticanalysis.manfredd.laubichlerisatheoreticalbiologistandhistorianofbiology.Heworksonconceptualproblemsofevolutionarydevelopmentalbiology,thedevelopmentandevolutionofsocialinsects,andthehistoryoftheoreticalanddevelopmentalbiology.HeisassistantprofessoroftheoreticalbiologyandhistoryofbiologyintheSchoolofLifeSciencesatArizonaStateUniversityandamemberoftheCentersforBiologyandSocietyandSocialDynamicsandComplexity.HeisassociateeditorofBiologicalTheoryandtheJournalofExperimentalZoology,PartB:MolecularandDevelop-mentalEvolution.TogetherwithJaneMaienscheinhedirectstheEmbryoProject,aninternationalnetworkforthestudyofthehistoryofdevelopmentalbiology.HeiscoeditorofFromEmbry-ologytoEvoDevo(withJaneMaienschein,forthcomingfromMITPress),ModelingBiology(withGerdMu¨ller,forthcomingfromMITPress),andHochsitzdesWissens:DasAllgemeinealswissenschaft-licherWert(withHans-Jo¨rgRheinbergerandPeterHammerstein,2006,DiaphanesVerlag).timlewensisalecturerintheDepartmentofHistoryandPhilosophyofScienceattheUniversityofCambridge,whereheisalsoaFellowofClareCollege.Heistheauthoroftwobooks–OrganismsandArtifacts:DesigninNatureandElsewhere(MITPress)andDarwin(Routledge).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nxivContributorselisabetha.lloydisArnoldandMaxineTanisChairofHistoryandPhilosophyofScienceandProfessorofBiologyatIndianaUniversity,Bloomington.SheistheauthorofTheCaseoftheFemaleOrgasm:BiasintheScienceofEvolution(HarvardUniversityPress,2005)andTheStructureandConfirmationofEvolutionaryTheory(GreenwoodPress,1988;PrincetonUniversityPress,1994),aswellascoeditorofKeywordsinEvolutionaryBiology(HarvardUniversityPress,1992)withEvelynFoxKeller.janemaienscheinisRegents’ProfessorandParentsAssociationProfessoratArizonaStateUniversity,wheresheisdirectoroftheCenterforBiologyandSociety.President-electoftheHistoryofScienceSociety,shespecializesinthehistoryandphilosophyofbiologyandthewaythatbiology,bioethics,andbiopolicyplayoutinsociety.HermostrecentbookisWhoseViewofLife?Embryos,Cloning,andStemCells(HarvardUniversityPress).gregorym.mikkelsonhaspublishedworksofpoetry,philosophy,andscience.These,alongwithmanyofhisotheractivitiesandambitions,couldfairlybedescribedas‘‘naturalistic’’inseveralsensesofthatword.Hismaingoalinlifeistofindharmonybetweenthewildandthetame.robertal.millsteinspecializesinthephilosophyofscienceandthehistoryandphilosophyofbiology.Herresearchhasfocusedonconceptualandepistemologicalissueswithinevolutionarytheoryandhasaddressedquestionssuchas,Isevolutionindeter-ministic?Whatisthemostappropriateinterpretationofprobabilityforevolutionarytheory?Whatrole,ifany,doescausalityplayinevolutionarytheory?Isnaturalselectionamechanisminanyofthesensesrecentlypropoundedbyphilosophersofscience?Cannaturalselectionbedistinguishedfromrandomdrift?MillsteiniscurrentlyanassociateprofessorintheDepartmentofPhilosophyatCaliforniaStateUniversity,EastBay(formerlyHayward).robertt.pennockisprofessorofhistoryandphilosophyofscienceatMichiganStateUniversity,whereheisonthefacultyoftheLymanBriggsSchoolofScience,thePhilosophyDepartment,theDepartmentofComputerScience,andtheEcology,EvolutionaryBiologyandBehaviorgraduateprogram.HisresearchinterestsareinCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nContributorsxvphilosophyofbiologyandintherelationshipofepistemicandethicalvaluesinscience.HisbookTowerofBabel:TheEvidenceagainsttheNewCreationismhasbeenreviewedinmorethanfiftypublications;theNewYorkReviewofBookscalledit‘‘thebestbookoncreationisminallitsguises.’’Dr.Pennockalsodoesscientificresearchonexperimentalevolutionandevolutionarycomputation,someofwhichwasfeaturedinacoverstoryinDiscovermagazine.PennockspeaksregularlyaroundtheUnitedStatesonissuesofscienceandvaluesandwasnamedanationalDistinguishedLecturerbySigmaXi,theScientificResearchSociety.grantramseyisadoctoralcandidateintheDepartmentofPhilosophyatDukeUniversity.Hisprimaryinterestsincludethephilosophyofevolutionarytheoryandtherelationshipbetweencultureandbiology.Amonghispublicationsarearticlesontheconceptofbiologicalfitness,theevolutionofculture,animalinnovation,andplantecology.robertj.richardsistheMorrisFishbeinProfessoroftheHistoryofScienceattheUniversityofChicago,whereheisprofessorinthedepartmentsofPhilosophy,History,andPsychol-ogyanddirectoroftheFishbeinCenterfortheHistoryofScienceandMedicine.HeistheauthorofDarwinandtheEmergenceofEvolutionaryTheoriesofMindandBehavior(UniversityofChicagoPress,1987),TheMeaningofEvolution:TheMorphologicalConstructionandIdeologicalReconstructionofDarwinsTheory(UniversityofChicagoPress,1992),andTheRomanticConceptionofLife:ScienceandPhilosophyintheAgeofGoethe(2002).HehasjustcompletedTheTragicSenseofLife:ErnestHaeckelandtheBattleoverEvolutionaryThought.jasonscottrobertisassistantprofessoroflifesciencesintheSchoolofLifeSciencesatArizonaStateUniversity(ASU).HeteachesintheBioethics,Policy,andLawProgramwithintheCenterforBiologyandSocietyandisalsoaffiliatedwiththeConsortiumforScience,Policy,andOutcomes.PriortojoiningthefacultyatASU,RobertwasassistantprofessorandCanadianInstitutesofHealthResearchNewInvestigatorintheDepartmentofPhilosophyatDalhousieUniversity.Hehaspublishedmanyarticlesinthephilosophyofbiologyandbioethics,andhisfirstbookisCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nxviContributorsEmbryology,Epigenesis,andEvolution:TakingDevelopmentSeriously(CambridgeUniversityPress,2004).alexanderrosenbergjoinedtheDukefacultyin2000.PreviouslyhewasprofessorofphilosophyatSyracuseUniversityandtheUniversityofCalifornia,Riverside,andDirectoroftheHonorsProgramattheUniversityofGeorgia.HehasbeenavisitingprofessorandFellowattheUniversityofMinnesota,aswellastheUniversityofCalifornia,SantaCruz,andOxfordUniversity.HehasheldfellowshipsfromtheNationalScienceFoundation,theAmericanCouncilofLearnedSocieties,theNationalHumanitiesCenter,andtheJohnSimonGuggenheimFoundation.In1993RosenbergreceivedtheLakatosAwardinthephilosophyofscience.HewasthePhiBetaKappa–RomanellLecturerfor2006–07.Rosenbergistheauthoroftenbooks,thelatestofwhichisDarwinianReductionismorHowtoStopWorryingandLoveMolecularBiology(UniversityofChicagoPress),andapproximately170papersinthephilosophyofbiology;thephilosophyofcognitive,behavioral,andsocialscience(especiallyeconomics);andcausa-tion.RosenbergisalsocodirectorofDuke’sCenterforthePhilosophyofBiology.sahotrasarkarteachesintheSectionofIntegrativeBiologyandtheDepartmentofPhilosophyattheUniversityofTexasatAustin.HeistheauthorofBiodiversityandEnvironmentalPhilosophy:AnIntroduction(CambridgeStudiesinPhilosophyandBiology)andGeneticsandReductionism(CambridgeStudiesinPhilosophyandBiology).roberta.skipperjr.worksinthehistoryandphilosophyofbiologyandsciencemoregenerally.Inthehistoryandphilosophyofbiology,Skipperworksontheoryassessmentinpopulationgeneticsandconceptualproblemsatthefoundationsofevolutionarygeneticsmoregenerally.Withinphilosophyofscience,heusesevolutionarybiologytodrivehisresearchonscientificexplanation,confirmation,andtheorychange.HereceivedaPh.D.inhistoryandphilosophyofsciencefromtheUniversityofMaryland,CollegePark,andispresentlyassistantprofessorofphilosophyattheUniversityofCincinnati.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nContributorsxviikimsterelnyisprofessorofphilosophyatVictoriaUniversityinNewZealandandattheAustralianNationalUniversityinCanberra.Heistheauthorofmanybooksonthephilosophyofscience,includingSexandDeath,cowrittenwithPaulGriffiths.HeisawinneroftheLakatosPrizeforthebestbookoftheyearinthephilosophyofscience(ThoughtinaHostileWorld:TheEvolutionofHumanCognition,publishedin2003).karolastotzwastrainedinhumanbiologyandsocialsciencesattheUniversityofMainzandreceivedherPh.D.inphilosophyfromtheUniversityofGhent.SheiscurrentlyapostdoctoralFellowintheCognitiveScienceProgramatIndianaUniversity,Blooming-ton.From2002to2005sheandPaulGriffithswereprincipalinvestigatorsoftheNationalScienceFoundation–fundedproject‘‘RepresentingGenes’’attheUniversityofPittsburgh.ThisinterdisciplinaryprojectusedWeb-basedsurveymethodstoexaminedifferingunderstandingsofthegeneincontemporarybioscience.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nprefaceThephilosophyofbiologyisoneofthemostvigorousandexcitingareasinmodernphilosophy.Therearemanyactiveresearchersandtheirstudents,thereisaplethoraofnewideasandsuggestions,therearegood-quality,dedicatedoutletsforthework–notablythejournalBiologyandPhilosophy–andthereareorganizations–notablytheInternationalSocietyfortheHistory,Philosophy,andSocialStudiesofBiology–thatembraceandencouragetheworkersinthefield.Thisheadyandforward-lookingcommunityisofrecentvintage.Althoughinthehistoryofphilosophysomeoftheverygreatestthinkers–AristotleintheancientworldandImmanuelKantinthemodernworld–hadthingsofgreatimportancetosayaboutthelifesciences,formuchofthepasttwohundredyearsbiologywasillservedandlittleregardedbyphilosophers.Thiswillseemstrange.Afterall,inthenineteenthcenturycamethegreatevolutionarytheoryofCharlesDarwin,expoundedinhisOntheOriginofSpeciesbyMeansofNaturalSelectionin1859,andinthetwentiethcenturytherewasthecomingofmolecularbiology,asrepresentedbythedoublehelixdiscoveredin1953bytheAmericanJamesWatsonandtheEnglishmanFrancisCrick.Generally,however,itwasphysicsthatcaughttheattentionofphilosophers.Biologywasoftenneglectedorworse,beingputtousebypeoplewithdeeplyantiscientificmetaphysicalagendas,especiallythevitalistsatthebeginningofthetwentiethcentury.Conditionsstartedtochangeaboutfortyyearsago,inthe1960s,atfirstslowlyandthenwithincreasingspeedasthetopicattractedattention.Anumberofyoungphilosophersofsciencetooknoteoftheexcitingdevelopmentsinthebiologicalsciences,notjustxixCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nxxPrefacemolecularadvancesbutalsothewaysinwhichtraditionalareas(notablyevolutionarybiology)werenowdevelopingnewideasandattractingfirst-classminds.Thesephilosopherssensedthattherewasanimportantpartofsciencethatwasneglected,andtheymovedtounderstandandworkontheconceptualproblemsthatarealwaysraisedbyempiricalstudies.Atthesametime,anumberofbiologistsstartedtoturnseriouslytophilosophyforhelpinarticulatingissuesintheirscience,realizingthatgoodempiricalworkdemandssoundphilosophicalbasestothetheoriesandmodelsthatguideresearch.Spurringandstimulatingeachother,thephilosophersandbiologistsworkedonsuchissuesasthenatureofevolutionaryexplanationandtheroleplayedbyDarwin’smechanismofnaturalselection;theextenttowhichbiologyisanautonomousscienceandwhetherissuesliketeleologyandhis-toricitymarkthelifesciencesassomethingirreduciblydifferentfromthephysicalsciences,orwhetherultimatelythesearemattersthatdropawayinamaturescience;problemsofclassification,bothaboutthebasicunitsofdivision(notablyspecies)andabouttheproperwaytoconceptualizelife’shistory(thecomingofphyloge-neticsystematics,orcladism,wasamajorissuehere);therela-tionshipbetweentheolderMendeliangeneticsandthenewermoleculargenetics,andwhetherthiswasacaseoftheoryabsorp-tion(reduction)oroftheorychange(replacement);aswellasrelatedtopics.Perhapsreflectingtheinterestsoftheearlyresearchers,perhapsreflectingthefactthatitissimplyofgreatphilosophicalinterestinitsownright,evolutionarytheorytendedtodominatediscussions,andindeedyouwillsensethatthistendencypersiststothisday.Inthepastfourdecades,biologyitselfhascontinuedtoadvanceinmanyexcitingways.Thealready-mentionedrevolutioninclassifi-cation,systematics,broughtonfierydebatesaboutthenatureandintentionsofworkersinthefield,andever-increasingsophisticationasmoleculartechniquesbecamereadilyavailable,backedbytheincreasingpowerofcomputerprograms.Therewasthedevelopmentoftheevolutionaryapproachtosocialbehavior,sociobiology,withhighlycontroversialattemptstoexpandthesciencefromotherani-malstoushumans.Thereweredebatesaboutthehistoryoflifeasrevealedthroughthefossilrecord,andwhetherpureDarwinismisadequateasanexplanation,orwhethernewapproaches(particularlyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPrefacexxithetheoryofjerkychange,punctuatedequilibrium)demandnewtheoreticalapproaches.Mostrecently,therehasbeentherevivalofembryologyandconsiderationofthewaysinwhichthiscanbeimprovedandextendedthroughmolecularfindings.Thefieldofevolutionarydevelopment,or‘‘evo-devo,’’hasattractedsomeofthebestbiologicalmindsofthegeneration.Philosophyhasrespondedtothesedevelopmentsinthebiologicalsciences.Theyoffernewchallengestothosewhoareinterestedindeepconceptualissuesofscience,andatthesametimetheyoffernewinsightsintosomeoftheperennialquestionsofphilosophyitself,aboutknowledgeandtruthandaboutactionandmorality.ItisthisresponsethatweaseditorshavetriedtocaptureinthisCompanion.Wehavetriedtogivethereaderasenseoftheexcitingworkthattodaycharacterizesthedisciplineorsubdisciplineofthephilosophyofbiology.Wehavenottriedsimplytogiveasurveyoratextbookintroduction.Therearealreadygoodworksofthisilk,somewrittenbycontributorstothisCompanion.Ratherwehavetriedtogiveasenseoftheissuesthatengagetoday’sphilosophersofbiologyandanunderstandingofhowtheseissuesaretackled.Wehaveaskedourcontributorstowriteinawayandatalevelthatanonexpertwouldfindinterestingandunderstandable,butatthesametimewehaveemphasizedthatthisshouldnotbeattheexpenseoftryingtotacklecomplexproblemsandshowingbyexampleofworkinactionratherthansimplythroughoverallsurveys.WedidnotsetouttoimposeformaldivisionsonthesubjectandwewouldliketothinkthatallofthecontributionstotheCom-panionarefreestanding,inthatonecouldreadanyoneinisolationfromtheothers.However,thereaderwillsensethatthereisakindofinformalflowtotheissues,withonetopicleadingnaturallytoanother,andwehavetriedtoreflectthisintheorderinwhichthecontributionsarepresented.WestartwithDarwin’stheoryofevo-lutionthroughnaturalselection,orratherwithitsmodern-daysuc-cessor.ForDarwin,asfortoday’sevolutionists,theimportantpointaboutselectionisthatitexplainsnotjustchangebutchangeofaparticularkind,namely,inthedirectionofadaptiveadvantage–itexplainstheeyeandthehandandalloftheotherorganicfeaturesthatwereattheheartofthetraditionalargumentfromdesignfortheexistenceofGod.ItisthistopicofadaptationthatisthefocusofTimLewens’scontribution,asheteasesapartthemeaningoftheterminCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nxxiiPrefacemodernscienceandtriestoassessitssignificanceassomethinginneedofexplanation.Fromadaptation,withthecontributionbyRobertaL.MillsteinandRobertA.SkipperJr.,wemovetothestructureandnatureofmodernevolutionarythinking.AfterDarwin,themostimportantadvanceinsuchthinkingaccompaniedthediscoveryanddevelop-mentofgenetics,thetheoryofheredity.Thistransformedourideasaboutthepast,especiallyaftertheso-calledpopulationgeneticistsembeddednaturalselectioninatheoreticalframeworkbasedonMendeliangenesinteractingingroups.MillsteinandSkipperdiscusshowthishappenedandthengoontolookatsomeconsequences,especiallywhethertheimportanceofselectionisnowdiminishedbyotherfactors,notablySewallWright’snotionofgeneticdrift,essen-tiallyamatterofrandomchangecausedbycontingentfactorsover-whelmingthesystematiceffectsofdifferentialreproduction.FromherewemovenaturallyandsmoothlytoElisabethA.Lloyd’sessay,whichtakesupatopicthatplaguedDarwinhimselfandwhichhasraisedmuchdiscussioninrecentyears,aboutthelevelorlevelsatwhichnaturalselectionmaybesaidtoact,inparticularwhetheritisalwayssomethingbetweenindividuals,perhapsevenbetween‘‘selfishgenes,’’orwhetheritcananddoesactsignificantlyathigherlevelsbetweengroups.Then,completingthispartoftheCompanion,wehaveananalysisbyRobertN.BrandonandGrantRamseyofsomeofthemostinterestingissuesthatemergefrommodernevo-lutionarythinking,namely,thosethatconcernthestatisticalnatureofthetheoryandhowitworkswithandtriestoexplaintheactionsandeffectsnotofindividualsworkingalonebutofindividualsingroupsandofthecumulativeresults.Nextcomesabatchofarticlesdealingwithsomeoftheissuesinthephilosophyofbiologythatrelatetotraditionalquestionsinthephilosophyofscience.Muchtimehasbeenspentonquestionsaboutwhethernewtheoriesareabsorbedintooldtheories(reduction)orwhethertheysimplypushthemasideandreplacethem.TheessaybyPaulE.GriffithsandKarolaStotzopensthewayfordiscussionofthistopicinbiologywithrespecttothenatureofgeneticsandhowconceptsofthegenehavechangedthroughtheyears,especiallyinthelightofthecomingofmoleculartechniquesandtheories.ComplementingthispieceisthecontributionbyPeterGodfrey-Smith,takingupthetopicofbiologicalinformationandofhowandCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPrefacexxiiiinwhatsensesomethinglikethegene(orinitsmodernguise,theDNAmolecule)canbesaidtocarrysuchinformation.Afterthis,AlexanderRosenberggivesusamoregeneraldiscussionofreductioninbiologyandthewaysinwhichonecanandshouldexpectcon-tinuouschangebetweentheoriesofdifferenttimesandlevels.FromherewemovetoLindleyDarden,wholooksatexplanationsinbiologyandathowmodelsareanimportantfactorintryingtoexplicatemechanisms.Andre´Ariewthenbringsthiskindofdis-cussiontoaconclusionbysurveyinganddiscussingaperennialquestionforthephilosopherofbiology:whetherinsomesenseevolutionaryunderstandingisteleologicalorforward-lookinginawayalientothephysicalsciences,andwhetherinsomesensethisrepresentsanonreduciblefactorthatentersintoallexplanationsinthelifesciences.ThankstopeoplelikethelateStephenJayGould,therehasbeenmuchwrittenrecentlybybiologistsabouttherelationshipsbetweenmicroprocessesofchangeandtheoverallmacronatureoflife’shis-tory.Canthelatterbeexplainedintermsoftheformer?ThisisthetopicofKimSterelny’scontribution,whichfocusesonanotionthathecalls‘‘minimalism,’’namely,thethesisthatsmall-scalechangescanexplainall.Thewaysinwhichthistopictiesinwithsomeoftheearliercontributionsaboutreductionhardlyneedstressing.ThesystematistMaureenKearneythendiscussestheperennialproblemofbiologicalclassificationandtheextenttowhichitcanandshouldreflectlife’shistory.Hercontributionreflectssomeofthe(earlier-mentioned)majordebatesthathaveoccurredinthisarea,particu-larlyaboutthenatureofbiologicalspeciesandwhethertheyaretobeconsideredasindividualsorasgroups,andthenaboutthetechniquesandsignificanceoftheapparentlyall-conqueringtaxonomicapproachofthecladistswiththeirinteresting(andcontroversial)assumptionsaboutsuchnotionsassimplicityor(astheycallit)parsimony.Biologicallyspeaking,humansmayormaynotbethemostimportantofallorganisms.Theyarecertainlytheorganismsofmostinteresttoushumans!ThegeneticistFranciscoJ.Ayalagivesimportantbackgroundtotheproblemsfacingresearchersintotheevolutionofhumans.Inthelightofthecompletedmappingofthehumangenome,hestressesthatwestillfaceissuesabouthowwesenseandfeel,howthemindemergesanditsconnectiontotheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nxxivPrefacematerialbody,andhowandwhyhumansemergedfromtheapeline.DavidJ.Bullerthenrowsintocontroversialwaterssurroundingthenewscienceofhumansociobiologyor(asitisnowoftencalled)evolutionarypsychology.Ishumannaturetobeexplainedasafunctionofthegenesassiftedbynaturalselection,oristhisapproachaltogethertoosimplistic?Inherdiscussionofneurobiology,ValerieGrayHardcastletakesupinmoredetailsomeoftheissuesraisedbyAyala.Shestresseshowinattemptstounderstandthefunctioningofthebrainanditsrelationshiptomind,methodologicalandmeta-physicalquestionsarise,andoftenitisnoteasytoteaseoutthequestionsthatshouldbeaskedandtherelevanceofempiricalfindingsthatresultfromresearch.Next,goingfromthegeneraltothemoreparticular,ChristopherHorvathturnstothemuch-discussedtopicofhumansexualorien-tation.Hestressesthewayinwhichwhenwegettocontroversialaspectsofhumannatureitisnoteasytodisentanglestronglyheldsocialandmoralbeliefsfrommoreobjectivescientificfindingsandtheories.Horvath’scontributionillustratesthewayinwhichmodernphilosophyofscience,philosophyofbiologyinparticular,hasmovedfromparticipatinginpurelytheoreticaldiscussionstoaddressingissuesofimmediatesocietalinterestandconcern–andsotomoraltheoryandbehavior.Foralmostallofthetwentiethcentury,thanksparticularlytothedevastatingcritiqueinG.E.Moore’sPrincipiaEthica(1903),attemptstorelatebiologytoethicswereregardedasthephilosophicalequivalentofabadsmell–notjustwrong,butinsomesenseunclean.Mattershavechangeddramaticallyinrecentyears,thankstothedevelopmentofsuchareasassociobiologybutalsothankstoformalworkbyeconomistsandphilosophersandothers.Verysignificanthasbeenworkonthetopicofgametheoryandhowitappliestohumanevolution.Inhiscontribution,ZacharyErnstintroducesustothistopic.Formuchofthenineteenthcenturyandwellintothetwentiethcentury,embryologywasanimportantpartofthebiologicalscene.Thenwiththecomingofgeneticsandthemovetomoleculartopics,itratherdroppedfromsight,beingregardedassomewhatofadescriptivetopicoflittletheoreticalinterest.Inthepasttwoorthreedecades,asmentioned,conditionshavechangeddramaticallyasmolecularbiologistsandevolutionistshavejoinedforcestolookatdevelopment.Todayevolutionarydevelopment,orevo-devo,isoneCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPrefacexxvofthehottestareasofbiologicalresearch.Thehistorianandphilo-sopherofscienceJaneMaienscheinopensourclusterofarticlesonthistopic,lookingatthechangingmeaningofthenotionofanembryo,howitevolvedfromrathercrudenotionsintheeighteenthcenturytoavarietyofsophisticatedconceptstoday.ShestressesinafashionakintothatofHorvaththat,indiscussionsofthisnature,itisoftendifficulttodistinguishclaimsoffactfromdeeplyheldmoral,social,andreligiousbeliefs.ManfredD.Laubichlerfollowswithamoredetaileddiscussionofevo-devoandstresseshowitopensupquestionsofgreatphilosophicalinterest.Hehimselffocusesontheproblemofinnovationandofhownewfeaturesgetproducedandintroducedintopopulations.IsthisachallengetotraditionalDarwinianexplanationsthatemphasizetheall-sufficiencyofnaturalselection?JasonScottRobertconcludesthispartoftheCompanionbylookingathowmolecularbiologyhasbroughtnewinsights.Hetalksof‘‘systemsbiology,’’theareaofstudythatlooksathowthegenesgettranslatedandusedtobuildthefunctioningorganism,referringexplicitlytoethicalissuesthatareraisedandshowingthatatthispointphilosophyofbiologytouchesonandinrespectsblendswiththesortsofconcernsthatoccupybioethicists.Intheearlyyearsofcontemporaryphilosophyofbiology,ecologywascuriouslyandunfortunatelyoverlooked.Althoughenvir-onmentalistswereraisingimportantissues,mostofwhichcalledoutfordetailedanalyticscrutiny,philosophersofbiologywereunre-sponsive.Thingshavenowchanged,andthereisincreasinginterestintheissuesraisedbytheinteractionsoforganismsonadaily(ratherthanhistoricalandevolutionary)basis.GregoryM.Mikkelsongivesusabackgroundtoecologyandthephilosophicalissuesthatitraises.Heshowshowtherearequestionsaboutthenotionofhierarchy,fromindividualsuptogroupsandthentowholesystems;abouttherela-tionshipofecologytoevolutionaryquestions;abouttheverynotionoflaw-governedexplanationsinecology(isitmoreofasubjectdealingwiththeuniqueandthecontingent?);andabouttheextenttowhichecologistsshouldbenaturalists,observingwhatishappeninginthewild,andtheextenttowhichtheyshouldbeexperimenters,manip-ulatingsituationsandtryingtopredictoutcomes.SahotraSarkarfollowsbylookingatconceptsofecologicaldiversity,andhowintheirnewguiseof‘‘biodiversity’’manyfascinatingphilosophicalquestionsaboutunderstandingandmeasurementgetraised.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nxxviPrefaceComingtotheendoftheCompanion,wehavenextRobertT.Pennock,whotakesupthequestionofbiologyandreligion.Pennockhasbeenmuchinvolvedinthefightagainstcreationists,particularlyintheirnewincarnationasso-calledintelligentdesigntheorists.Hiscontributionreflectsthisactivity,butheaimstoputhisdiscussioninthebroadercontextofbiologyandreligiongenerally,andthusheisledintoissuessuchasthemeaningofmoralityinapost-Darwinianworld,aswellastheautonomyofreligionitself.Isitjustanadaptationlikeanyotheranddoesthishaveimplicationsforitstruthvalue?Pennockraisestheissueofbiology,religion,andsexualorientation,anditisinterestingtocomparehisthinkingwiththeearliercon-tributionofChristopherHorvathonthesametopic.Finally,throughanexaminationofthethinkinganduseoflanguageofthenineteenth-centurybiologistErnstHaeckel,thehistorianandphilosopherofscienceRobertJ.Richardstakesonquestionsaboutunderstandingsciencebythestudyofthehistoryofthesubject.WethinkthisaparticularlyappropriatecontributionwithwhichtoendtheCompa-nion.Therelationshipbetweenthehistoryofscienceandthephilo-sophyofsciencehasnotalwaysbeenasintimateandfruitfulasonemightdesire,butthephilosophyofbiologyhasbeenexceptionalinthewaysinwhichithas(asmanyofourcontributionsshow)drawnonthehistoryofbiologyforinsightsabouttheconceptualnatureofpresent-daybiology.Richards’scontributionshowsthevalueofthispracticeandwhyitshouldbecherishedandencouragedasamodeofinquirybythenextgenerationofphilosophersofbiology.Wehopethatyouenjoythecollection.Goodphilosophy,likegoodscience,isneverfinished.Ifyouarelucky,youhavemorequestionsattheendofthedaythanatthebeginning.Weinviteyoutoengagewithourcontributorsandtoaddtotheexcitingadvancesinthephilosophyofbiology.Weinviteyoualsototakeamomentandthinkaboutthemantowhosememorythisvolumeisdedicated.ErnstMayr,whodiedin2005inhishundred-and-firstyear,wasaGerman-borntaxonomistwhoemigratedtoAmericaandbecameoneofthemostimportantandinfluentialevolutionistsofthetwentiethcentury.Hewasalwaysinterestedinphilosophyandwastheleaderamongthosementionedearlierwhoworkedfromthesideofsciencetodevelopthenewlyinvigoratedfieldofthephilosophyofbiology.Thisvolumeinyourhandsstartedin2001asthefirstconferenceatFloridaStateUniversitysponsoredbythelegacyofWilliamH.andLucyleT.Werkmeister.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPrefacexxviiMayr,veryold,stillspenthiswintersinFloridaandwasinvitedtotheconference.Heaccepted,buttheorganizerswerewarnedthathewouldstayonlyashortwhile,thatmuchattentionmustbepaidtotheneedsofoldage,andthathewouldbetoofatiguedtoattendanysocialevents.Hearrivedandwithinminuteswasonhisfeetmakingpoints,levelingobjections,andstressingthecorrect(thatis,Mayrian)wayofseeingthings.OntheSaturday,hestartedatnineo’clockinthemorningandinsistedonjoiningtheparticipantsinanafternoontriptoWakullaSprings,alocalbeautyspotofunspoiledriverwithalligators,fish,andmanybirds(thesiteoftheoriginalTarzanmoviesaswellastheclassicsciencefictionmovieTheCreaturefromtheBlackLagoon).Mayrstoodintheprowoftheboat,identifyingbirdsandexplainingthedifferencessincehislasttriptothespotin1931.Atteno’clockthatnight,theexhaustedorganizersfinallyhadtoshuffleMayroutofthedoorandinsistthathegohometobed!ErnstMayrwasagreatscientist,butmorethanthis,hewasamensch.Fortheeditorsespecially,whofrequentlywerethefocusofMayr’sscolding,itisarealprivilegetoacknowledgeourdebtandourlove.Moreimmediately,theeditorsthankJasonZinser,whowasourassistantonthisproject,andAlanCasselman,whoworkedonthebibliography.Sadly,becauseofhistoo-earlydeath,wecannotthankTerryMooreatthePress,whorespondedwithsuchenthusiasmtotheideaofthisvolume,butwedorememberhimwithgratitude.BeatriceRehl,whotookovertheproject,hassupportedandhelpedusinanexemplaryway.Wearegratefultoourproductioneditor,JanisBolster,andourindexer,LinMariaRiotto.Finally,theeditorswanttothankeachother.Wehavebeenfellowphilosophersnowforfortyyearsandgoodfriendsfromthebeginning.Neveracrosswordhasbeenexchangedbetweenus.Deliberately,wedidnotourselvescontributetothevolume,wantingrathertocommissionandpromotetheworkofothers.Itisthefieldordisciplineofthephilosophyofbiologythatisourmaincreation,andinasmuchasthisvolumeshowsthatwehavesucceeded,itisbecauseeachofusknowshowmuchweowetotheother.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nCambridgeCollectionsOnline©CambridgeUniversityPress,2008\ntimlewens1AdaptationEyeshavelongfascinatedthosewhostudythenaturalworld.CleanthesthenaturaltheologianprotagonistofHumesDialoguesConcerningNaturalReligioninviteshisinterlocutortoconsider,anatomizetheeye:Surveyitsstructureandcontrivance;andtellme,fromyourownfeeling,iftheideaofacontriverdoesnotimme-diatelyflowinuponyouwithaforcelikethatofsensation(1990,65).Darwin,too,countedtheeyeamongwhathecalledorgansofextremeperfection.Placinghimselfsquarelywithinthetraditionthatrunsfromnaturaltheology,throughDarwin,toacertainstyleofmodernbiology,MaynardSmithwritesthatthemaintaskofanytheoryofevolutionistoexplainadaptivecomplexity,thatis,toexplainthesamesetoffactsthatPaleyusedasevidenceofacreator(1969,82).Morerecentlystill,Dawkins(1986)isimpressed,also,withaforcelikethatofsensation,byhowwellsuitedhowwelladapted,thatistheeyeistoitspurpose.LikePaley,hethinkseyesarebetterpiecesofworkthanwatches,althoughunlikePaleyheregardstheirartificerasblind.Anessayonadaptationcouldfillvolumes.Onemightbeginbyaskinghowadaptationistobeexplained.Immediatelywewouldneedtoanswerthepriorquestionofwhattheproperdefinitionofadaptationis,andwewouldalsohavetogetclearonthenatureofthediversecandidateprocessesnaturalselection,self-organisation,macromutation,development,divinedesignsometimestabledaspotentialexplanations.Wemightgoontoaskinwhatsensesadaptationsarepurposive,andwhethertheyallsharesomesingleultimatepurpose,suchastheproliferationofanorganismsgenes.Oncethenatureofadaptationispinneddown,wecouldmoveontoconsiderthequestionsofwhetheradaptationisubiquitousorrare,1CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n2timlewensandwhethertheremightbeimportantnonadaptivephenomenainthebiologicalworldthatanexclusiveconcernwithadaptationmightleadustooverlook.Inshort,athoroughstudyofadaptationwouldneedtoaddressmostofthetopicscoveredinthisCompanionteleology,theunitsofselection,development,andothers.Here,then,Iwillrestrictmyselftobriefdiscussionsoffourquestions.Howshouldwedefineadaptation,howshouldweexplainadapta-tion,howcanwediscoveradaptation,andhowimportantisadaptation?1.definingadaptationIntheanalyticaltableofcontentsofhislandmarkworkAdaptationandNaturalSelection,GeorgeWilliamsclaimsthatevolutionaryadaptationisaspecialandonerousconceptthatshouldnotbeusedunnecessarily,andaneffectshouldnotbecalledafunctionunlessitisclearlyproducedbydesignandnotbychance...Naturalselectionistheonlyacceptableexplanationforthegenesisandmaintenanceofadaptation(1966,vii).Iwanttotakesometimeinthefirsttwosectionsofthisessaytopickthesecommentsapart.What,precisely,isthespecialandonerousconceptofadaptation?Asapreliminary,weshouldtakeElliottSobers(1984,196)adviceanddistinguishproductsfromprocesses.Consideranexample:mar-riagesproducemarriages.Thissoundspeculiar,untilwerememberthatmarriagecanrefereithertotheprocessofgettinghitchedortotheblissfulunionthatistheproductofthatprocess.Similarly,adaptationcanrefereithertotheprocessbywhichorganismsbecomewellsuitedtotheirenvironments,oritcanrefertotheorganictraitsthataretheendresultsofthisprocess.UnlessIstip-ulateotherwise,Iwillbetalkingaboutadaptationasaproductinthisessay.Broadlyspeaking,therearethreequitedifferentstylesofdefinitionoftheadaptationconcept.First,wecouldgivearoughindicationofwhatadaptationmeansbypointingtosomeofitsinstancesthingsliketheeye,orthewing.Suchdefinitionbyexample,certainlywhentheexamplesarefew,tellsuslittleabouthowweshouldapplytheconcept.Atthispoint,asecondstyleofdefinitionmayappear.Adaptationisaconceptusedinmodernbiology,yetmodernbiolo-gistssometimesdefinetheterminaninformalwaythatechoesCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation3naturaltheologysconceptionoforganismsasdesignedobjects.Williamsgivesjustsuchadefinitioninthequotationwejustsaw:Aneffectshouldnotbecalledafunctionunlessitisclearlypro-ducedbydesignandnotbychance.Thisdistinctionbetweenwhatanobjectseffectsareandwhatitsfunctionsaremakesclearsensewhenwearetalkingabouttoolsdesignedbyagents.Ascrewdrivermaybegoodatleveringlidsfrompainttins,butthatisnotwhatthescrewdriverisforthatisnotitsfunctionbecausethescrewdriverwasnotdesignedtoliftlidsfrompainttins.Williamssdefinitionexpresseshisviewthatadaptationsaretraitsthatareforsomething.ForWilliams,therefore,thequestionofwhethersometraitisanadaptationshoulddependonitsdesignhistory.ButWilliamsisnocreationist:thedesignhistoryinquestionistheevolutionaryhistoryofthetrait.Williamsscommentexplainswhymanybiologistsdrawadis-tinctionbetweenadaptivetraitsandadaptations.AdaptivetraitsaugmentfitnessinsomewayoranotherwemightalsouseMayrs(1986)termandsaythattheyhavethepropertyofadaptedness.Theadaptednessofatraitisnotsufficientforthetraittobeanadapta-tion,becausethetrait,likethescrewdriver,maynothavetherightkindofhistory.RichardDawkinsalsodefinesadaptationintermsofgooddesign,andhedefinesdesign,inturn,asthatwhichgivesonlytheappear-anceofintelligence:Wemaysaythatalivingbodyororganiswelldesignedifithasattributesthatanintelligentandknowledgeableengineermighthavebuiltinordertoachievesomesensiblepurpose,suchasflying,swimming,seeing,eating,reproducing,ormoregenerallypromotingthesurvivalandreproductionoftheorganismsgenes(1986,21).ForDawkins,asforMaynardSmith,thewaytodefineadaptationisintermsofwhatanaturaltheologianmighthavecounted,mistakenly,asevidenceofintelligence.ItishardtosquareWilliamssclaimthatadaptationisaspecialandonerousconceptformodernevolutionarybiologywithallthesecovertusesofwhatappeartobenaturaltheologicalnotionsinthedefinitionofthatconcept.Ifadaptationcouldbedefinedonlyassomethingthatthesuperstitiouswouldtakeasevidenceforadesigner,thenthebestthingformodernbiologytodowouldbetoeliminatetheadaptationconceptaltogetheronthegroundsthatitispartofanaturaltheologicalworldviewwenolongershare.BecauseCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n4timlewensfew,ifany,biologistsorphilosopherscouldtoleratetheeliminationoftheadaptationconceptfrombiology,amovetoadefinitioninthethird,moreformal,styleiswidelypreferred.2.historicaldefinitionsofadaptationFormaldefinitionsofadaptationtendtodivideintohistoricalandnonhistoricalvarieties.Aformaldefinitionthatisendorsedbymanyphilosophers(althoughnotbysomanybiologists)isSobers(1984,208):AisanadaptationfortaskTinpopulationPifandonlyifAbecameprev-alentinPbecausetherewasselectionforA,wheretheselectiveadvantageofAwasduetothefactthatAhelpedperformtaskT.OneofthereasonswhyadefinitionlikethisisattractiveisthatitpromisestotidyupWilliamssclaimthatadaptationsaretheresultofdesignratherthanchance.Whatisrequired,ifthisclaimistobemaderespectable,issomeevolutionaryprocessthatcanplaytheroleofdesign.Soberachievesthisbydefiningadaptationastheproductofanaturalselectionprocess,aprocessthatcanbedistinguishedfromthemerechanceappearanceinapopulationofthetraitinquestion.Sobersdefinitionleadstosomeawkwardresults,especiallyifassessedbyitssuccessingroundingthenotionthatadaptationsareproducedbydesign.First,atraitcanbeanadaptationforsometaskevenwhenthefirstoccurrenceofthetraitisanentirelyfortuitousaffairthathasnothingtodowithselection.Thisisaconsequenceofthedefinitionofselectionforaproperty.Supposeapairofwingsarises,fullyformedandfullyfunctional,inapopulationofflightlessfoxes.Thesewingshelptheirprodigiouslyluckybearertocatchchickensmoreeffectivelythanotherfoxes,andasaresulttheflyingfoxisfarfitterthanitsfellows.Babyfoxesinheritthewingsoftheirparents,andwingssoonbecomeprevalentinthepopulation.Inthis(intentionallyabsurd)scenario,thereisselectionforflyinginvirtueofthefactthatwingsincreasetheirfrequencyinthepopulationbecausetheyallowflying.HencewingsareadaptationsforflyingbySobersdefinition,eventhoughthemetaphorsofselectiondesign-ing,building,orshapingthetraitarehardtoapply.ThisishardlyafatalobjectiontoSober:modernbiologycangetbyperfectlywellCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation5withanadaptationconceptthatjarssomeofourintuitionsaboutwhenitisappropriatetospeakofdesignorshaping.Evenso,wewillseethatforsomebiologists,adaptationsareunderstoodastraitsthathavebeen(insomesense)shaped,built,ormodifiedbyselec-tion,notmerelytraitswhosefrequencyhasincreasedbecauseofselection.SobersdefinitionhelpsustomakesomesenseofWilliamssclaimthatadaptationsarenotproductsofchance,butindoingsoitcausesproblemsforWilliamssfollow-upassertionthatselectionistheonlypermissibleexplanationofadaptation(aclaimthatRichardDawkins[1996]alsomakes).Itmakesthatsecondclaimtrue,butvacuouslyso.ItishardtoportrayDarwinsintellectualbreak-throughastherealisationthatadaptationisbestexplainedbynat-uralselection,ifadaptationissimplydefinedasaproductofaselectionprocess.Fisher(1985,120)makesthepointforcefully:Definingthestateofadaptationintermsofitscontributiontocurrentfitness,ratherthanoriginbynaturalselection,isessentialifnaturalselectionistobeconsideredanexplanationofadaptation.Fishersargumentcanberesisted.WecankeepholdofSobersdefinitionofadaptationwhilerephrasingourunderstandingofDarwinsbreakthroughinmoreparticularistterms:Darwinrealisedthatnaturalselectioncouldexplaintheorganisationofeyes,wings,instinctivebehaviours,andmanyotherspecifictraits.Noneoftheseclaimsisvacuouslytrue,evenifthegeneralclaimthatnaturalselectionistheonlyexplanationofadaptationis.WemightalsoconsiderreplacingWilliamssgeneralassertionwiththerephrasedclaimthatselectionistheonlypermissibleexplanationofadapt-edness,whereadaptednessisdefinednonhistoricallyintermsofacontributiontofitness.Whetherthisrevisedclaimistruewouldrequirefurtherassessment,butitseemsclearthatitisnottriviallytrue.Thereareotherproblemsthathavedrivensomebiologists(e.g.,ReeveandSherman,1993)toprefernonhistoricaldefinitions.Consideratraitthatbecomesprevalentinapopulationbychance,butthatissubsequentlymaintainedatahighfrequencyinthepopula-tionbecauseofitssuperiorfitnesscomparedwithalternatives.Sobersdefinitiondeniesthatthetraitisanadaptation,forselectionhasnotmadeitprevalent,eventhoughmaintainingselectiondoessubsequentlyexplainwhyitremainsprevalent.Conversely,SoberCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n6timlewensmighthavetoaccordthestatusofadaptationtotraitsthathavespreadthroughapopulationinvirtueofsomeeffect,butthathavenothadthateffectforseveralgenerations.Wemighthavetosaythatthehumanappendixisanadaptationfordigestion.Yetthesekindsoftraitsaremoreusuallythoughtofasvestiges,notadaptations.TheseproblemsarenotfataltoSobereithertheobvioussolutionistokeepahistoricaldefinition,butonethatlooksonlytoquiterecentselectionhistory,includingselectionthatmaintainsthefrequencyofatraitinvirtueofoneofitseffects.Soberhimselfconsiderssomeanalogousmovestoweakentheoriginaldefinitionwhileretainingitshistoricalelement(1984,198).AlthoughareviseddefinitionofthisformupholdsaconceptualdistinctionbetweenbeinganadaptationforEandpromotingfitnessbyE-ing,veryfewactualtraits,solongastheyareinherited,willfallintothelattercategorywithoutalsofallingintotheformer.Therevisedhistoricaldefinitionhelpstogroundafunction/effectdistinctionthatnon-historicalaccountswillhavetroublemaintaining,butthesatisfac-tionofthisdesideratummayseemlikeaphilosophicalindulgencewhenviewedfromtheperspectiveofbiologicalinquiry,especiallyonceweseehowrarelytheconceptualdistinctionwillmakeanypracticaldifference.ThebiologistsEndlerandMcClellanprefertouseadaptationtoindicatecurrentcontributiontofitnessonjustthesegrounds:Itisimportanttodistinguishbetweentraitsthatwerealwaysselectedforonefunction(adaptations)fromthosewhichwereoriginallyselectedforanotherfunctionandbychancecanbeusedinanewway(exaptationsforthenewfunction).Weuseadaptationinbothsensesbecauseassoonasanewfunctionforatraitoccurs,naturalselectionwillaffectthattraitinanewwayandchangetheallelefrequenciesthatgeneratethattrait.(1988,409)Thiscommentislikelytomislead,becausethehistoricaldefi-nitionofadaptationpreferredbymanyphilosophersisnotthesameasthatofGouldandVrba(1982),whosedistinctionbetweenadap-tationandexaptationEndlerandMcClellanarereferringtohere.GouldandVrbasdefinitionofadaptation,likeWilliamss,appealsnotjusttoselectionforsomeproperty,buttoastrongernotionofshaping,orstructuralmodification,consonantwiththeeverydayconceptofdesign.Anadaptation,forGouldandVrba,wasbuiltbyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation7naturalselectionforthefunctionitnowperforms(53).Exapta-tions,ontheotherhand,havenotbeenshapedbyselectionforthetaskstheynowperform.Adefinitionofadaptationintermsofrecent(maintaining)selectionwillmakealmostallexaptationsforsomefunctionadaptationsforthatsamefunction.Somephilosophershavequestionedthecoherenceoftheadaptation/exaptationdis-tinction(e.g.,Dennett1995,281);however,providingwecanmakesenseofthecontrastbetweenbeingshapedforafunctionandbeingselectedforafunction,andprovidingourdefinitionofadaptationappealstoshaping,thisscepticismispremature.3.nonhistoricaldefinitionsofadaptationReeveandShermanhavearticulatedthemostthoroughdefenceofanonhistoricaldefinitionofadaptation:Anadaptationisaphenotypicvariantthatresultsinthehighestfitnessamongaspecifiedsetofvariantsinagivenenvironment.(1993,9)Whyinsistthatanadaptationmustbethefittestofasetofvar-iants?Whichvariantsgointothatset?Toanswerthesequestions,weneedtolookattheprimarygoalofReeveandShermansdefi-nition,whichistodevelopaconceptsuitableforansweringques-tionsaboutwhattheycallphenotypeexistence.Theydistinguishthesekindsofquestionsfromthoseaboutevolutionaryhistory.Studentsofphenotypeexistenceaskwhycertaintraitspredominateoverconceivableothersinnature,irrespectiveoftheprecisehis-toricalpathwaysleadingtotheirpredominance,andtheninferevolutionarycausationbasedoncurrentutility.Practitionersofevolutionaryhistory,ontheotherhand,seektoinfertheoriginsandphylogenetictrajectoriesofphenotypicattributes,andhowtheircurrentutilityrelatestothepresumedfunctionsintheirbearersancestors(2).Therearetwoslightlydifferentrationalesforappealingtoarangeofconceivablevariantsindefiningadaptation.Thefirsthastodowithestablishingtheselectivehistoryofthetrait(i.e.,evolutionarycausation).Thesecondhastodowithestablishingthetraitscausalcontributiontosurvivalandreproduction(i.e.,itscurrentutility).Beginningwiththefirstrationale,ReeveandShermanwanttheclaimthatatraitisanadaptationtobeevidencefor,ratherthan(asitCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n8timlewensisforSober)synonymouswith,anyfurtherclaimaboutevolutionarycausation.Ifatraitisfitterthanallthehypotheticalalternativetraitsweareconsidering,thenthechancesareitalsooutcompetedtheactualalternativetraitsinthepopulation.Itisthereforelikelythatselectionexplainsitspresence.Thatiswhyweshouldexerciserestraintinthehypotheticalalternativesweincludeinthecon-sideredsettheyneednotincludeallandonlyactualcompetitors,buttheyshouldreflectlikelycompetitors:Asuitablechoicerequiresonlythatthesetcontainsphenotypesthatmightplausiblyarise(ReeveandSherman1993,10).ReeveandShermansinsis-tencethatatraitberecognisedasanadaptationonlyifitisthefittestofthephenotypesetisnotobligatorygiventhegoalsofthiskindofevolutionaryresearch:atraitcanmakeasignificantcontributiontofitnessincludingthegreatestcontributionamongactualvariantspresentandpastevenwhensomeotherplausibletraitmighthavebeenbetterstill.Thesecondreasonforappealingtoarangeofconceivablevariantsindefiningadaptationhasconsiderablemetaphysicalinterest,especiallyasacasestudyintheproblemsofcausation.Onthenonhistoricalapproach,toaskwhetherhumaneyesareadaptationsistoaskwhethertheymakeacausalcontributiontofitness,andifso,whatthatcontributionis.Itmightseemthatthereisnoneedtospecifyasetofalternativepossibleeyesinordertoanswerthisquestion;weneedonlyconsiderthequestionWhatwouldwebelikewithouteyes?Theproblemisthatthisquestionhasmanyplausibleanswers.Wemightsaythatvisionissoimportantthatifwehadnoeyes,wewouldhavesomeotherkindofsensoryapparatusinstead.Ifwesaythis,wewillsaythateyesarenotadaptationsforprovidingsensoryinformation,forwewoulddojustaswellinthatrespectwithouteyes.Alternatively,wemightsaythatifwehadnoeyes,wewouldbedead,asaresultofinfectioninouremptyeyesockets.Ifwesaythis,wewillsaythateyesareadaptationsforpreventinginfectionreachingtheeyesockets.Bothanswersseemsilly,butsuchsillinessseemstoresultfromasking,withoutcon-straint,Whatwouldwebelikewithouteyes?Theseproblemsabouthowtosaywhatthecausalcontributionofsomepartistoawholearenotspecifictobiology.Considermylaserprinter.Ourinclinationistosaythattheinkcartridgecontributestotheworkingsofthewholebydispensingink.ButwhatallowsustoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation9saythis?Afterall,itisnottruethattheonlythingthatwouldbedifferentifmyprinterhadnocartridgeisthatitwoulddispensenoink.Paperwouldnotpassthroughtheprinter,either.Ifweunder-standcounterfactualconditionalsinthemannerofDavidLewis(1973a),thencounterfactualsaremadetruebystatesofaffairsatthenearestpossibleworldswherethecounterfactualsantecedentsaretrue.Roughlyspeaking,apossibleworldisawaythingsmighthavebeen.ThestatementWereBeckhamtohavegotthepenalty,EnglandwouldhavewonEuro2004istruejustincasethosenearestworlds(i.e.,theworldsmostsimilartowaythingsactuallyare)whereBeckhamgetsthepenaltyarealsoworldswhereEnglandwinsEuro2004.Nowthenearestworldatwhichmyprinterhasnocartridgeis,presumably,onewhereIhaveremovedthecartridgetoshakeit,orsomesuch.Atthisworld,theprinterwillnotfunctionatall.Arewetosay,then,thatthefunctionoftheinkcartridgeistoenablepapertopassthroughtheprinter?Comparingactualeyeswithaclearlyspecifiedsetofalternativetraitsseemstobeagoodwayofcircumventingtheseproblemsforcausalanalysis.Onemightwonder,though,exactlywhattheroleisofspecifyingalternativesinthedeterminationofatraitscausalcontribution.Thismethodissomewhatatoddswithcontemporarycounterfactualviewsofcausation(e.g.,Lewis1973b).Accordingtothesetheories,causationisindeedboundupwithwaystheworldmighthavebeen,butwedetermine,say,thecausalimpactofabricksflightthroughawindownotbyspecifyingalternativeflightpaths,butbyspecifyingwhichactualevent,orwhichactualfact,weareinterestedinunderstandingcausally,andaskingwhatwouldhavebeenthecasehadthateventnotoccurred,orhadthatfactnotobtained.Onthisview,alternativeflightpathsfollowfromaspecificationofthefactoreventofinteresttous;thespecificationofalternativesisnotapreliminarytocausalanalysisofsomefactorevent.Generalisingthismethodtotheanalysisofsystems,thefirststeptodeterminingacausalcontributionofaparttoasystemisnotthespecificationofalternativepossibilitiesforwhatthepartmighthavebeenlike;rather,itistospecifywhataspectofthepartweareinterestedin.Theeffectofthisistomoveusawayfromaskingbluntquestionsaboutthecausalcontributionsofparts(organs,forexample,inbiology),andtowardsaskingquestionsaboutthecontributionsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n10timlewensoftraitsunderstoodasfinelyspecifiedfactsaboutsystemicorgani-sation.Inthecaseoftheprinter,wecanask,forexample,whattheeffectofthecartridgeshavinginkatsuchandsuchdensitymightbe.Immediatelywedismisstheworldwheretheprinterhasnoinkcartridgeasirrelevanttoansweringourcausalquestion;thenearestworldwheretheinkdensityisdifferentisnotonewherethereisnoinkcartridgeatall.Sooneoftherolesforspecifyingarangeofalternativetraitsistodrawout,throughcomparison,theaspectofthetraitunderconsiderationthatweareinterestedin.Notethatweneednotsupposethatanyactualprintercartridgehasexistedwithinkatadifferentdensityforanappealtosuchcounterfactualcir-cumstancestohavelegitimacyincausalreasoning.Thatiswhy,totheextentthatReeveandShermansphenotypesetissupposedtofocusattentiononspecificaspectsofactualtraitsbypositingwaystheworldmighthavebeenifthetraithadbeendifferentinthoserespects,theyarequiterightnottoinsistthatmembershipofthesetberestrictedtoactualtraits.4.explainingadaptationDoesnaturalselectionexplainadaptation?Wehavealreadyseenhowthisquestionrunsintothedefinitionalproblemsoftheearliersections.Ifadaptationshouldbedefinedasaproductofselection,thentheclaimthatselectionexplainsadaptationissecuredbydefinitionalone.Letusask,instead,whetherselectionexplainsadaptednessunderstoodascontributiontofitnesswhilewith-holdingjudgementonthequestionofwhetherweshouldoptforanonhistoricaldefinitionofadaptationitself.Ourfirstjobistodistinguishthequestionofwhetherselectionexplainsthespreadoftraitsfromthequestionofwhetherselectionexplainstheoriginoftraits.InthefirstsectionIgavetheexampleofawingthatarisesfullyformedthroughmacromutation,andthatthenspreadsthroughapopulation.Inthisscenario,selectionexplainswhythetraitincreasesitsfrequency,anditalsoexplainstheincreaseinadaptednessofthepopulation,butitdoesnotexplaintheappearanceofthefirstwingtoken.Ifselectionneverexplainstheoriginoftraittypes,thenDarwinsinnovationisnotasgreatasitseems.PeoplelikePaleywerepuzzledathowsuchthingsaseyescouldcometoexistatall;theresponsethatonceoneeyeexists,eyesCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation11willtendtobecomeprevalentisnotenoughtosatisfythedemandforanexplanationofthefirstappearanceofsuchanextraordinarything.Andnotethatifonerepliesthateyesarereallyratherlikelytoarisethroughmacromutationafterall,orthatoneshouldsimplylooktochancetoexplainthefirstinstanceofthetrait,thenthereisasenseinwhichonehasnottrulyofferedaselectiveexplanationfortheexistenceofeyes;rather,onehasappealedtosomethinglikeagenerallawofgenerationoforganicform,ortoblindchance.Selectioncanexplaintheoriginofnoveltraittypesbycausingtokensofothertraitstospread(Endler1986).Selectioniscreative.Considerthefollowingcase,borrowedfromNeander(1995).Imaginethattherearethreegeneticplans,P1,P2,andP3.P3givesusafullyfunctionaleye.P2yieldsaslightlyinferioreye,andP1isevenworse.ApopulationexistsinwhichallmembershaveP1,barafewwithP2.AsP2increasesinfrequencyinthepopulation,thenthechancesofavariantarisingwithP3mayincreasetoo.Incasethisisnotclear,thinkofanexample;ifIcanpersuademorepeopletobuylotterytickets,thenIincreasethechancesthatsomeonewillwin.Simi-larly,ifthenumberofgeneticlociwhereafavourablemutationmightoccurincreasesasaresultofselection,thenselectioncanexplainthefirstappearanceofafavourablemutationofthattype.Stickingwiththelotteryexample,althoughincreasingthenumberofticketsexplainswhysomeonewins,itdoesnotexplainwhyEmmawins,ratherthanNicole,whohasalsoboughtaticket;increasingthenumberofticketsmakesEmmanomorelikelytowinthanNicole.Similarlyifselection,byincreasingthenumberoforganismswithP2,explainswhyP3eventuallyarises,itdoesnotfollowfromthisthatselectionexplainswhySamfirstacquirestheP3mutation,ratherthanSuzy,whoisalsodescendedfromaP2individual.Theclaimthatselectioniscreativedoesnotstraight-forwardlyresolvetherelatedquestion(addressedbySober1995andNeander1995,amongothers)ofwhetherselectionexplainswhyindividualorganismshavethetraitstheydo.Nowthatwehaveseenhowselectionexplainsadaptedness,wecanalsoseethatselectiondoesnotexplainadaptednessalone(Lewens2004).Thereareanumberofassumptionshiddeninourearlierscenarioabouttheemergenceofeyes.IfincreasingthenumberofP2variantsistoincreasethechanceofaP3variantarising,itmustbethecasethattheP3planismorelikelytoariseCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n12timlewensfromtheP2planthanfromtheP1plan,forsupposethatwhileP2isfitterthanP1,andP3isfitterthanP2,P3ismutationallyclosertoP1thantoP2.Ifthatisthecase,thenselectionwillmaketheemer-genceoftheP3planlesslikelyasitincreasesthefrequencyoftheP2plan.Whatisrequiredforselectiontohaveitscumulativecharacter,wherebyitleadstoprogressiveadaptation,isforfitnessorderingsofphenotypestomapontothemutationaldistancesbetweentheplansthatcodeforthosephenotypes.That,inturn,isarequirementontheorganisationofindividualorganisms;indeed,thefamiliarpointthatselectioncancausepopulationstogetstuckonthelocaloptimaoffitnesslandscapesisanotherwayofmakingthepointthatorganismsmaybeconstructedinsuchawaythatmutationsforhighlyfunctionalversionsofsomeorganmaysometimesbemorelikelytoarisefromformsoflowfunctionthanfromformsofintermediatefunction.Selectiondoesnotexplainadaptednessalone,becauseselectionhasthecharacterthatmakesitcumulativeonlywhenitactsonsystemswiththerightkindoforganisation.Thisisnottheplacetosaywhatthoseorganisationalconditionsmightbe,butitistheplacetopointoutthatthismakesorganicorganisationanelementoftheexplanationofadaptednessitself(Lewens2004,Walsh2000).Wealsohavereasontobelievethatdriftorsamplingerror,wherebypopulations(especiallysmallpopulations)changeinwaysthatdonotreflectthefitnessesofthetraitsinthepopulation,canhelptoexplainadaptedness.Driftcanpreventpopulationsfromgettingstuckonlocaloptima,andinthatsensetheexistenceofsamplingerror,wherebylower-fitnessformssometimesreplacehigher-fitnessforms,willoccasionallyincreasethechancesthatthepopulationwillarriveatanadaptivepeakthatishigherstill.Sodrift,also,canexplainadaptedness.5.discoveringadaptationLetusmovenowfromthemetaphysicsofadaptationtoitsepiste-mology.Howcanwedeterminewhetheratraitisanadaptation?Onepopularapproachtothisquestiononceagainlooksbacktonaturaltheologyforinspiration(Lauder1996givesausefulsurveyofthesemoves).Itusedtobethoughtthatthegooddesignofanorganismindicatedintelligentdesign;ifgooddesignisinsteadbestCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation13explainedbynaturalselection,thengooddesignisevidencefortheactionnotofthehandofthedivineartificer,butthehandofselection.ThusPinkerandBloom(1990,707)writethatevolu-tionarytheoryoffersclearcriteriaforwhenatraitshouldbeattrib-utedtonaturalselection:complexdesignforsomefunction,andtheabsenceofalternativeprocessescapableofexplainingsuchcomplexity.Isthisagoodinference?Partlyitdepends,onceagain,onthestrengthwereadintotheclaimthatatraitshouldbeattributedtonaturalselection.WehavealreadyseenthatifatraitpromotesfitnessinsomepopulationthrougheffectE,itishighlylikelythattherehasbeenselectionforE.ThisrequiresonlythattherehasbeensomeheritablevariationinpastgenerationswithrespecttotheperformanceofE.Yetsomeintendbyadaptationnotmerelytraitthathasbeenselectedforitsfunctionbutthefarstrongertraitthathasbeenbuiltforitsfunction.HenceGouldandLewontinsworrythattheusefulnessofatraitinsomerespectdoesnotentail,norevenmakeprobable,theclaimthatthetraitisanadaptationforthateffect:maletyrannosaursmayhaveusedtheirdiminutivefrontlegstotitillatefemalepartners,butthiswillnotexplainwhytheygotsosmall(1979,581).WecanalsoexplainwhyLewontin(1984)isconcernedaboutthepotentiallymisleadinglabeladaptation.IfwestressparallelsbetweennaturalselectionandnaturaltheologywearelikelytothinkofselectionasakintoaBlindWatchmakerselectionasacraftsmanshapesorganicformtomeetenvironmentalproblems.Butadaptednessisnotalwaysproducedbyselectionactinginthisshapingmode.Onealternativeisthatmembersofapopulationslowlymigrateuntiltheyfindahabitatinwhichsomepreexisting,unchanged,traitenhancesfitness.Thisisanaturalselectionexpla-nationofsorts,fortherelationaltraitofbeinginabetterhabitatincreasesitsfrequencyasthemembersofthispopulationchangetheirenvironments.Butitisnonethelessmisleadingtosaythatthefactofadaptednessindicatesselectivedesigninthisscenario,andwewouldbewrongtoinferselectiveshaping,althoughnotselectionfor,fromthefunctionalityofthetrait.Thereisasecond,andmorecontentious,roleforappealstogoodengineeringdesigninestablishingtheexistenceofadaptation.Inthefirstmode,weknowthatthetraitpromotesfitness,andweinferCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n14timlewensthatselectionexplainsthisstateofaffairs.Inthesecondmode,weinferfromthestructureofthetraitbothalikelyfitnesscontributionandselectionforthatcontribution.ThisisthewaythatDennettcharacterisestheinferencefromthefossilremainsofArchaeopteryxtothejointclaimsthatArchaeopteryxwasabletofly,andthatitswingswereselectedforflight:Ananalysisoftheclawcurvature,supplementedbyaerodynamicanalysisofwing-structure,makesitquiteplainthatthecreaturewaswell-designedforflight(1995,233).Itisatthispointthattheobjectionthatadaptationisthypothesesarejust-sostories(GouldandLewontin1979)ismostlikelytorearitshead.Theproblemofjust-sostoriesisaproblemofunder-determinationitissimplytooeasytomakeevolutionarysenseofatraitbyshowinghowitmighthavecontributedtofitness.Con-siderthecrestoftheCorythosauruscasuaris,anexamplebeautifullydiscussedbyTurner(2000).Corythosauruswasaduckbilleddino-saur(ahadrosaur),whosefossilremainssuggestthatitsskullboreacuriouscrest,shapedlikeaCorinthiansoldiershelmet.Thesecrestsarehollow,andairinhaledthroughthenostrilswouldhavepassedthroughthecavitiesofthecrestonthewaytothelungs.Wasthecrestanadaptation?Ifso,forwhat?Turnerlistsavarietyofresponsesthathavebeentabledatonetimeoranotherforthefunctionsofhadrosaurcrests:theywereweapons;theywereanchorsforashorttrunk;theyhousedanolfactorysystemthatdetectedpredators.Ifwefocusonlyondatathatrelatetotheroughstructureofthehadrosaurcrest,thenitseemsthereareverymanydesignhypothesesthatwillmakesenseofthestructure.Somewillberuledoutthecrestcouldnothavebeenasnorkel,fortherearenoholesinthecrestthatwouldpermitairtobedrawnintothelungs.Butmanyothersremain,includingthedelightfulhypothesisthatCorythosaurus’screstwasmoreSCUBAthansnorkelashort-termairtankthatwouldhaveenabledthedinosaurtoforageunderwater.Itisnotonlyadaptationisthypothesesthatfacetheproblemofunderdetermination.Maybehadrosaurcrestsarenotadaptationsatallperhapsacranialcrestisasideeffectofselectionworkingonsomeotheraspectofhadrosauranatomy,withnospecialfunctionofitsown.Ifwearetoassertthisweneedevidence;otherwisethisisajust-sostory,too,albeitnotanadaptationistone.Nonetheless,thehistoricalnatureofsomebiologicalhypothesesdoespresentthem,asaclass,withespeciallyacuteproblemsofunderdeterminationCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation15comparedwiththoseofnonhistoricalsciences.Ifwehavetwoincompatiblehypothesesabouthadrosaurcrests,weshouldtryhardtofindsomedatathatmightdiscriminatebetweenthem.Theprob-lemforbiologyisthatsometimeswecansaywhatdataweneed,butthesedataareinaccessible.Arehadrosaurcrestsairtanks,pred-atordetectors,orsomethingelse?Wecouldtellquiteeasilyifwecouldobservehadrosaursinaction,butthatavenueis(andalwayshasbeen)closedofftous.FortheremainderofthissectionIwillfollowTurnerindis-cussingthediscoveryofadaptationinthemoregeneralcontextofinferencetothebestexplanation.Withinscienceandwithout,weofteninferthatapropositionistrueonthegroundsthatifitweretrue,itwouldbethebestexplanationofourdata(Lipton2004).SuchinferencesfeaturethroughouttheOriginofSpecies,forexample.Timeandagain,Darwinsaysthathistheoryshouldbeacceptedbecauseitoffersabetterexplanationofdiversesetsoffactsthandoitscompetitors.Tosupportinferencetothebestexplanationisnottosaymerelythatoncewehaveahypothesisthatwouldmakeourdataprobable,weshouldbelievethathypothesis.Thiswouldentitleus,forexample,toinferanycombinationofpastecologicalcircumstances,nomatterhowoutlandish,justsolongastheymakelikelythesparsestructuraldatathatwehavetohandrelatingtofossilremains.Foranexplanationtobegood,henceforittocommandourassent,itmustmeetfurtherconstraints,andwecanusethesefurthercon-straintstoilluminatesomeofthewayswemightdiscriminatebetweenalternativeadaptivehypotheses,includinghypothesesaboutCorythosaurusscrest.Ifanexplanatoryhypothesismakesthedataveryprobable,wesaythatthehypothesishashighlikelihood.Immediately,itwillbeclearthatlikelihooddoesnothandleunderdeterminationproblemswell.Ifseveralincompatiblehypothesesallentailthedata,thentheyallhaveequallikelihoods,solikelihoodcannotbeusedtochooseamongthem.Andlikelihoodcertainlyisnottheonlythingwelookforinagoodexplanation(althoughitisimportant).SupposeourdatumisthefactthatSamhaswontheNationalLottery.ThehypothesisthatSamenteredthelotteryfairandsquaremakesthisdatumexcep-tionallyimprobable.ThehypothesisthatSamhasabrotherwhoriggedthemachinesmakesourdatumfarmoreprobable.ButwedoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n16timlewensnotassumethateverylotterywinnerisacheat,eventhoughsuchhypotheseshavefargreaterlikelihoodsthanthehypothesesthattheseindividualsarelegitimatewinners.IsthereevidencethatSamhasabrother?Isthereanyreasontothinkhisbrotherhascontroloverthelotterymachine?Withoutfurtherevidenceinfavouroftheseassumptionsthefactthattheyconstituteanexplanatoryhypothesiswithahighlikelihooddoesnotcountstronglyinfavourofthetruthofthathypothesis.Similarly,unlesswehaveevidenceinfavouroftheassumptionslaidoutinsomeselectionhypothesis,thehighlikelihoodofthathypothesisdoesnotcountstronglyinitsfavour.Thefirstlesson,then,isthataswellasofferingadaptivehypothesesthatmakeevolutionarysenseofthetraitsweareinterestedin,weneedtotesttheassumptionsofthosehypothesesdirectlybeforeweputfaithinthem.Isthereanyreasontothinkthathadrosaursforagedinwater,soSCUBAapparatuswouldhavebeenusefultothem?Perhapsthecrestisamusicalinstrumentorresonator(ahypothesismuchfavouredtoday),butistherereasontothinkthathadrosaurhearingwasgoodenoughforthenoisesthecrestproducedtohavebeenaudible?Didhadrosaursliveingroups,inwhichauditorysig-nalswerethereforevaluable?Theanswerswegivetothesequestionscanhelptorulesomehypothesesin,andothersout.Asecondwaytobolsterourexplanatoryinferencesistoseewhethertheyconformtoanygeneralpattern.SupposewehaveobservedlotteryfixinginmanycasesthatresembleSamsownvictory.ThiscountsasfurtherevidenceinfavouroftheclaimthatSamsvictorywasalsofixed.Similarly,wecanbolsterselectivehypothesesbylookingtoseewhetherthehypothesisofadaptationinquestionconformstoanygeneralpattern.Thisiswheretheso-calledcomparativemethodcanbepowerful.ConsiderDarwinsexampleofthebaldheadofthevulture.Anengineeringanalysisofwhatabaldheadmightbegoodfor,coupledwithknowledgeofthefeedinghabitsofthevulture,couldleadustothehypothesisthatthebaldheadisanadaptationforwallowinginputridity(1964,197).This,afterall,makesevolutionarysenseofthetraitsappearance.Darwinpointsout,however,thatweshouldbeverycautiousindrawinganysuchinference,whenweseethattheskinontheheadoftheclean-feedingmaleturkeyislikewisenaked(ibid.).Supposeweconclude,onthebasisofthefactthatmaleturkeyshavebaldheads,thatsomecommonancestorofturkeysandvulturesCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation17hadabaldhead,andthatbothturkeysandvultureshaveinheritedthistrait.Thatwouldunderminetheclaimthatthebaldheadofthevulturehasbeenshapedbyselectiontoenableittowallowinputridity,providingwethinkthecommonancestorwasclean-feedingastheturkeyis.Butitisconsistentwiththisscenariothatthebaldheadofthevulturecontributestofitnessbyenablingthevulturetowallowinputridity,andeventhattherehasbeenmaintainingselectionforthiseffectintherecentpastthatexplainsthecontinuedpresenceofthevulturesbaldhead.SoappealsofthiskindtothetraitsofrelatedspeciesintherefutationofadaptationclaimsmakesbestsenseifoneunderstandsadaptationneitherinthemannerofReeveandSherman,norinthemannerofSober,butasGouldandVrbaunderstanditasatraitshapedbyselectionforitsfunction.Forathirdwaytostrengthenouradaptiveinferences,letusreturntothelotterycase.SupposewethinkthatSamcheatedinwinningthelottery.Thisclaimwillleadustomakecertainpredictionsaboutwhatelseweshouldobserve.OnepredictionthatissuesfromourhypothesismightbethatsomeoneSamknowswillhavebeenintheLondonarea(wherethelotterymachinesarelocated)ontheweek-endwhenthelotterywasheld.Ifthispredictionturnsouttoberight,thenitprovidesonlytheweakestevidenceinfavourofourhypothesis.Ontheotherhand,wemightmakeamoredetailedpredictionofthesametype:amemberofSamsfamilywillhavebeenlurkinginthetelevisionstudiowherethemachinesarekeptattheprecisetimeofthelottery.IfCCTVevidenceshowsthispre-dictiontobetrue,thenwewillbefarmoreconfidentofourhypothesis.Ingeneral,weshouldbemoreconfidentofanexplana-toryhypothesisifitsfine-grainedpredictionsturnouttruethanifitsmorehand-wavingpredictionsturnouttrue.Here,then,isafinalway,championedinaseriesofpapersbyOrzackandSober(e.g.,OrzackandSober1994),forustoincreaseourconfidenceinthetruthofplausibleadaptiveexplanations.Weshoulddemandthatourhypothesesyieldrigorouslyquantifiedpredictions.Suchpredictionsmightincludedetailedengineeringspecificationsofwhatstructuresometraitoughttohave,ontheassumptionthatithasacertainfunction.This,roughlyspeaking,isRudwicks(1964)ParadigmMethodfordeterminingthefunctionoffossils.ButOrzackandSoberpointoutthatwecanalsomakequantitativepredictionsatthelevelofthepopulation,byspecifyingtheexactfrequencyofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n18timlewenstraitdistributionsweshouldexpecttofindundersomefunctionalhypothesis.6.adaptationandadaptationismAlthoughIhaveconcentratedformostofthisessayonadaptation,Iwanttoclosewithsomewordsaboutadaptationism.Theproblemswehavelookedatsofarconcernwhatadaptationsare,howweshouldexplaintheirappearance,andhowwecantellwhenwehavefoundone.Thevarietiesofadaptationismallassert,inonewayoranother,thatadaptationisofspecialimportanceforbiology.Need-lesstosay,thatisanambiguousclaim,andvariouswritershavemadeeffortstodistinguishthevarietiesofadaptationism,producingestimatesfromtwoformsthroughtoseven(e.g.,Godfrey-Smith2001,Lewens2002,Lewensforthcoming).Itwillserveourpurposeswellenoughheretodistinguishfourtypesofadaptationism:Empiricaladaptationismmosttraitsareadaptations.Natureis,insomesense,welldesigned.Methodologicaladaptationismregardlessoftheactuallevelofdesignqualityinnature,ortheprevalenceofadaptations,thebestwaytoinvestigatenatureistoassumethatalltraitsareadaptations.Explanatoryadaptationismtheproperbusiness,orthemostimportantorinterestingbusiness,ofevolutionarybiologyistheexplanationofadaptation.Epistemologicaladaptationismbiologistshavemethodsthataregoodenoughtoestablishfairlydecisivelythetruthandfalsehoodofmosthypothesesaboutadaptation.Iincludethefourthinmylistofformsofadaptationismonlyforhistoricalreasons.Oneofthefavouritecriticismsofantiadapta-tionistshasbeenthathypothesesofadaptationarefrequentlyjust-sostories.Wesawinthelastsectionhowevolutionaryhistoricalhypothesesofallkindsnotjustthehypothesesaboutadaptationcanbesubjecttounderdeterminationproblems,andwelookedatwaystoenrichourinferentialpracticeswhentryingtoovercometheseproblems.Evenso,antiadaptationistshavehistoricallybeenCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation19scepticalofthemethodsbywhichwetesthypothesesofadaptation;conversely,astrandofadaptationismhasincorporatedagreateroptimismaboutsuchmethods.Inowsetthisfourthformofadaptationismaside.Letmesayalittleabouttheotherthreekindsofadaptationism,beginningwithempiricaladaptationism.Iphrasedthisastheviewthatnatureiswelldesigned.Butwhatdoesthatmean?Thereareseveraloptionsforhowtoexplainthis.Atthestrongendofthespectrumwemighttakeadaptationismtobetheviewthatalltraitsarethebestonespossible.Evenherewerunintoproblems(whatisthemeaningofpossible?),butwearelikelytodismissthisversionofthehypothesisasfalse,evenobviouslyso.Afterall,muchoftheevidenceagainstintelligentdesignusestheimperfectionoforganictraitsfortheirapparentpurposesasevidence.Attheweakendofthespectrumistheviewthatnaturalselectionhasbeeninvolvedinsomewayinthehistoryofmosttraits.Thisthreatenstoturnouttriviallytrue.AnefforttogiveaninterestingreadingtothehypothesisofadaptationismcomesfromSoberandOrzack:Naturalselectionhasbeentheonlyimportantcauseofmostofthepheno-typictraitsfoundinmostspecies(Sober1998,72).Whatthismeansisthatthefittestavailablephenotypesarealwaystheonlyonespresentinpopulations.Takeapopulationofmoths,inwhichsomearewellcamouflaged,othersarepoorlycamouflaged,andnonehas(orhaseverhad)chemicalsecretionsthatkillbirdsinstantly.Ifthepoorlycamouflagedmothsenduptheonlyonesinthepopulation,thenthisresultcountsagainstadaptationism,forabetteravailablevariantdidnotreachfixation.Ifmothswithdeadlychemicalsfailtoreachfixation,thenthisdoesnotcountagainstadaptationism,forsuchmothswereneveravailabletobeselectedfrom.OnOrzackandSobersview,theadaptationisthypothesiscouldbetrueeveniftherangeofavailablevariationforselectiontoactonturnsouttobeveryhighlyconstrained.Indeed,adaptationismwouldbetrueatapossibleworldwherethelawsofnaturedictatethatonlytwokindsofentityexist,bothmakecopiesofthemselves,andthetypethatismorefecundreplacesthelessfecundtype.Yettheremaybenothingmuchlikegooddesignatthisworldnoorgansofextremeperfection.Thismakesclearthedifficultyoffindingareadingoftheadaptationisthypothesisthatsatisfiesourintuitionthatifadaptationismistrue,thennaturalselectionisCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n20timlewensapowerfulforce.CertainlyinthedullworldIjustdescribedselec-tionseemstohaveverylittlepower.Adaptationismissometimesbetterunderstoodasaheuristicratherthananempiricalhypothesis.Itisarecommendationforhowtogoaboutinvestigatingnatureinafruitfulmanner.Onceagain,wecouldcataloguemanyvariantformsofmethodologicaladaptation-ismrangingfromtheboringlysensibleadvicethatsometimesitisusefultotesthypothesesoftheformXisanadaptationforE,throughtotheimplausibleinsistencethatnothingthatisofvalueinbiologycanbediscoveredunlessonetestshypothesisoftheformXisanadaptationforE.Whatisimportanttonoteisthatversionsofmethodologicaladaptationismcanbeusefulevenifonethinksthatadaptationisnotubiquitous.Theadaptationistheuristicisusefulwhenthefailuretoestablishanadaptationisthypothesissuggeststousthatselectionisnotresponsiblefortheaspectoforganicformweareinvestigating.Explanatoryadaptationismtellsusthat,insomesenseoranother,theproperbusinessofevolutionarybiologyisthestudyofadapta-tions(forthisreasonIhaveelsewherecalleditDisciplinaryAdap-tationism[Lewensforthcoming],butforthesakeofclarityIwillstickwithGodfrey-Smithstermshere).JohnMaynardSmithisanadaptationistofthiskind,asisRichardDawkins,whowritesthatlargequantitiesofevolutionarychangemaybenon-adaptive,inwhichcasethesealternativetheoriesmaywellbeimportantinpartsofevolution,butonlyintheboringparts(Dawkins1986,303).SterelnyandGriffiths(1999,228)worrythattheremaybesomekindofvacuityinexplanatoryadaptationism.Theproblemstems,onceagain,fromthewidelyaccepteddefinitionofadaptationastraitselectedforsomefunction.Ifexplanatoryadaptationismincludestheclaimthatadaptationisalwaysexplainedbyselection,thenthisaspectofthepositionis,indeed,trivialunderthisdefinition.Butthisdoesnotmakethegeneralclaimthatbiologistsoughttobusythemselveswithadaptationsvacuous.Thestandarddefinitionofadaptationleavesopenavarietyofquestionsthatonemighttrytoanswer,noneofwhichistrivial,andnotallofwhichareaboutadaptation.Whicharetheadaptations?Whichtraitsareadaptationsforwhat?Whichtraitsarefunctionless?Whichorgan-ismsaredescendedfromwhichothers?Anexhortationtofocusonquestionsofthefirsttwotypesisnotvacuous.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nAdaptation21Thefactthatevolutionarybiologycontainsquestionsthatarenotaboutadaptationsavesexplanatoryadaptationismfromvacuity,butthisonlygeneratesnewproblemsfortheposition.Howcouldweargueagainstsomeonewhosaysthatanevolutionarybiologistsproperbusinessistodeterminegenealogicalrelationshipsamongspecies,regardlessofwhichtraitsmightbeadaptationsforwhat?Onemightreplybysayingthatexplaininghowselectionshapestraitsfortheirfunctionsisimportantbecauseitisnecessaryinthestruggleagainstcreationism.Thesewell-designedtraitsarejusttheonesthat,absentagoodselectionexplanation,willbeusedasevi-dencebyintelligentdesigntheorists.Butfirst,thiswouldonlyestablishadaptationasthemostimportantpartofbiologyifweagreedthatintelligentdesignwassothreateningthatitsdefeatshouldbethedisciplinesprimarygoal.Andevenifweweretoinflateintelligentdesigninthisway,itstillwouldnotunderminetheimportanceoftracinglinesofdescentbetweenspecies.Thisproject,too,iscentrallyimportantinunderminingcreationistarguments.Wearrivebackatthebeginning,then,withnaturaltheology.Explanatoryadaptationismismerelyastatementofexplanatoryinterests,intereststhatweshouldnotfeelcompelledtoshare,butwhosesalienceisderivedfromAnglo-Americanbiologysrootsinnaturaltheology.Thisconclusionisironic:farfromexpressingenmitybetweenmodernbiologyandnaturaltheology,explanatoryadaptationismistestimonytothefellowshipbetweenthetwotraditions.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nrobertal.millsteinandroberta.skipperjr.2PopulationGenetics1.introductionPopulationgeneticsisasubfieldofevolutionarybiologythataimstorepresentmathematicallythechangesinthegeneticvariationofpopulations(specifically,sexuallyreproducingpopulationswithMendelianheredity)overtime.Themathematicalmodelsofpopu-lationgeneticsprovideatheoreticalbasisforexperimentalstudiesoflaboratorypopulationsandstudiesofnaturalpopulations.Ourpri-maryfocusinthisessayisonpopulationgeneticstheoryitself,ratherthanitsapplications,althoughtowardstheendoftheessaywegivesomediscussionofthelatter.Populationgeneticsattemptstomeasuretheinfluenceofthecausesofevolution,namely,mutation,migration,naturalselection,andran-1domgeneticdrift,byunderstandingthewaythosecauseschangethegeneticsofpopulations.Buthowdoesitaccomplishthisgoal?WebegininSection2withabriefhistoricaloutlineoftheoriginsofpopulationgenetics.InSection3,wesketchthemodeltheoreticstructureofpopulationgenetics,providingtheflavorofthewaysinwhichpopu-lationgeneticstheorymightbeunderstoodasincorporatingcauses.InSections4and5wediscusstwospecificproblemsconcerningtherelationshipbetweenpopulationgeneticsandevolutionarycauses,namely,theproblemofconceptuallydistinguishingnaturalselectionfromrandomgeneticdrift,andtheproblemofinterpretingfitness.InSection6,webrieflydiscussthemethodologyandkeyepistemolo-gicalproblemsfacedbypopulationgeneticistsinuncoveringthecausesofevolution.Section7oftheessaycontainsconcludingremarks.Wearefocusingontheissueofcausalityinpopulationgeneticsbecausewetakethisissuetobeatthecoreofmanyofthe22CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics23contemporaryphilosophicaldebatesinpopulationgenetics.How-ever,itshouldbenotedthatpopulationgeneticsraisesotherphilosophicalissuesthatthisessaywillnotaddress.Togivetwoexamples,therehasbeenmuchdebateoverthequestionsofwhetherpopulationgeneticsdescribesanyscientificlaws(see,e.g.,Ruse1977;Beatty1995;Waters1998),andwhetherthemodelsofpopulationgeneticshavebeenreconciledwiththe‘‘semanticconception’’oftheories(see,e.g.,Beatty1981,Lloyd1988).2.originsofpopulationgeneticsandtheevolutionarysynthesisIntheearlypartofthetwentiethcentury,GregorMendel’sexperi-mentalworkonpeaplantswascommonlyperceivedtobeatoddswithDarwiniannaturalselection;theformer,itwasargued,wasevidencefordiscontinuousevolution(involvinglargechangesfromparenttooffspring),whereasthelatterrequiredcontinuousevolu-tion(involvingsmallgradualchangesfromparenttooffspring).However,nolaterthan1932,thefieldoftheoreticalpopulationgeneticsemergedasareconciliationbetweenMendelismandDarwinism.Mostbiologistsatthetimeacceptedthefactofevolu-tion,orDarwin’s‘‘descentwithmodification,’’butDarwin’sideaofnaturalselectionasacauseofthatmodificationwascontroversial.Indeed,lackingwereanygenerallyacceptedaccountsofgeneticvar-iationinpopulations(isthevariationcontinuousordiscontinuous?)orevolutionarychange(ischangegradualorsaltational?),andanunderstandingoftheappropriateuseofstatisticalmethodsforstudyingthese.Theemergenceoftheoreticalpopulationgenetics,whichaddressedalloftheseissues,istypicallyassociatedwiththeworkofR.A.Fisher,SewallWright,andJ.B.S.Haldane.Thefoun-dationalworksthatusheredintheoreticalpopulationgeneticsareFisher’s(1930)TheGeneticalTheoryofNaturalSelection,Wright’s(1931)‘‘EvolutioninMendelianPopulations,’’andHaldane’s(1932)TheCausesofEvolution.Whatfollowsisabriefsummaryoftheviewsofthesethreebiologists.Fisher(1890–1962)wasanEnglishbiologisttrainedinmathe-maticsatCambridgeUniversity.Inadditiontohisunsurpassedcontributionstostatistics,hisinitialcontributionstoevolutionarybiologypredatethoseoftheothertwotheorists.Fisher’saiminCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n24robertal.millsteinandroberta.skipperjr.TheGeneticalTheoryofNaturalSelectionwasformallyandmathematicallytodemonstratehowthe‘‘vague’’conceptofnaturalselection(asitwasthenconsidered)couldpossiblywork.AndhedoesthisbyconsideringthetheoryofnaturalselectionagainsttheprinciplesofMendelianinheritanceonananalogywiththemathe-maticaltechniquesofstatisticalmechanics.Fisher’sviewistypi-callyunderstoodasfollows:Evolutionisdrivenprimarilybynaturalselection,ormassselection,atlowlevelsactingontheaverageeffectsofsingleallelechanges(ofweakeffect)atsinglelociindependentofallotherloci.Wright(1895–1988)wasanAmericanbiologisttrainedbyWilliamE.CastleatHarvardUniversityinphysiologicalgenetics.Priortohisgroundbreakingresearchinevolutionarytheory,whichhecarriedoutattheUniversitiesofChicagoandWisconsin,Madison,heworkedasastaffscientistfortheU.S.DepartmentofAgriculture.Wright’sviewschangedgreatlyovertheyears,butwhatWrightwaslookingforinhis1931essay‘‘EvolutioninMendelianPopulations’’weretheidealconditionsforevolutiontooccur,givenspecificassumptionsabouttherelationshipbetweenMendelianheredityandtheadaptivevalueofgenecomplexes.Idealconditions,forWright,arethoseconditionsthatproducethefastestrateofevolu-tiontothehighest‘‘adaptivepeak.’’Wrightbelievedthattheseconditionsrequiredthatpopulationsbesubdividedandsemi-isolated,andthatselection,alongwithrandomgeneticdriftandmigration,operatedina‘‘shiftingbalance’’ofphases.Wright’sShiftingBalanceTheorycanbesummarized:Evolutionproceedsviaashiftingbalanceprocessthroughthreephases:PhaseI,Randomgeneticdriftcausessubpopulationssemi-isolatedwithintheglobalpopulationtolosefitness;PhaseII,Massselectiononcomplexgeneticinteractionsystemsraisesthefitnessofthosesubpopula-tions;PhaseIII,Interdemicselectionthenraisesthefitnessofthelargeorglobalpopulation.Haldane(1892–1964)wasanEnglishbiologisttrainedinmathe-matics,classics,andphilosophyatOxfordUniversity.Haldanebeganhisworkonevolutionaryproblemsin1922withtheoreticalormathematicalinquiriesintotheconsequencesofnaturalselectioninMendelianpopulations.Haldane’s1932TheCausesofEvolutionisanoriginalandimportantcontributiontotheoriginsoftheoreticalpopulationgeneticswithitscriticalanalysisofextantempiricalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics25workagainstthebackgroundofhisownandFisher’sandWright’stheoreticalwork.TheappendixtoCausesisacriticaldiscussionofhisownandFisher’sandWright’sachievementsregardingthereconciliationofMendelismwithDarwinism.HaldaneagreeswithportionsofbothFisher’sandWright’sevolutionarytheories.Forinstance,HaldanesupportedWright’semphasisonepistasisandmigration,andhesupportedFisher’sviewontheimportanceofnaturalselectionoverrandomgeneticdrift.Haldanefurtherthoughtthatnaturalselectionwouldproceedrapidlyinlargepopulations,anideaneitherFishernorWrightbelieved.ItiseasyandfairlycommontoviewHaldaneasapopularizerofpopulationgeneticsamongbiologistsgenerally.Butthisisamistake:MuchofHaldane’sworkinthe1920sadumbratesideasfoundintheworkofFisherandWright.ThemathematicalreconciliationofMendelismwithDarwinismachievedbyFisher,Wright,andHaldanebeganthehistoricalperiodofevolutionarybiologycalledthe‘‘EvolutionarySynthesis’’(alsoknownasthe‘‘ModernSynthesis’’).Theirtheoreticalachievementscom-binedwithearlyexperimentalworkbysuchluminariesasTheodosiusDobzhansky(1937)setthestagefortheintegrationofpreviouslydivergentfieldssuchaspaleontology,zoology,botany,systematics,andgenetics.Tobesure,therewasconsiderabledisagreementamongthearchitectsoftheSynthesis.FisherandWrightinparti-cularwereengagedinaninitiallyfriendlycontroversythatrapidlybecameheated,from1929untilFisher’sdeathin1962,overhowtointerprettheirmathematicaltheories.Bythe1950s,asStephenJayGould(1983)pointsout,theSynthesiswould‘‘harden,’’emphasiz-ingnaturalselectionasthemostsignificantevolutionarycause.Toalargeextent,theapplicationsofcontemporarypopulationgeneticsaredeeplyrootedintheachievementsoftheperiodbetween1918and1960.Indeed,thereis,forinstance,persistentcontroversyovertherelativesignificanceofFisher’sandWright’spopulationgeneticstheories(Skipper2002).Since1960,applicationofmolec-ulartechniquestoevolutionaryproblemshasledtorevisionsintheinterpretationofsomeofthebasicassumptionsofpopulationgeneticstheoryaswellasofevolutionatthemolecularlevel.Moreover,advancesinmicrobiologyanddevelopmentalbiologyhaveledtochallengestotheexplanatoryscopeofpopulation2genetics.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n26robertal.millsteinandroberta.skipperjr.3.contemporarypopulationgeneticsThemodelsofcontemporarypopulationgeneticsexemplifythegeneralizedreconciliationbetweenMendelismandDarwinismusingthenowwell-entrenchedstatisticalmethodsintroducedbythearchitectsofthefield.Assuch,populationgeneticsdefines‘‘evolution’’aschangeingenefrequencies,ormorestrictly,anychangeinthefrequencyofalleleswithinapopulationfromonegenerationtothenext.Differentlyput,populationgeneticsaimstoaccountforthedynamicsofgeneticvariationinpopulations.Anditdoessobyattemptingtouncoverthepatternsofthosedynamicsviathecausesofevolution,namely,mutation,migration,naturalselection,andrandomgeneticdrift.Ourgoalinthissectionistoprovidethereaderwithageneralsenseofwhatthemodelsofcon-temporypopulationgeneticsarelike;consequently,ourdiscussionmusttakeaslighttechnicalturn.Populationgeneticsbeginsitstaskbyspecifyingtheconditionsunderwhichgenefrequenciesremainunchangedfromonegenera-tiontothenext:theconditionsunderwhichevolutionisnotoccurring.Theseconditionsarecapturedbythefoundationalprin-cipleofpopulationgeneticscalledthe‘‘Hardy-WeinbergPrinciple.’’ThePrinciplebeginswithasetofassumptionsaboutthegeneticsystem,matingsystem,andpopulationstructure:Assumearan-domlyinterbreeding,large(mathematically,infinite)populationofdiploidorganismswithonegeneticlocusandtwoalleles.Infact,theseassumptionsarefundamentaltomoststandardpre-sentationsofpopulationgenetics.Giventhesebasicassumptions,theHardy-WeinbergPrinciplestatesthatintheabsenceofevolu-tionarycauses,thatis,mutation,migration,naturalselection,andrandomgeneticdrift,thegenefrequenciesofthepopulationwillremainunchangedfromonegenerationtothenext;thepopulation3willbein‘‘Hardy-Weinbergequilibrium.’’InHardy-Weinbergequilibrium,whenthetwoallelefrequenciesareequal,thedis-tributionofgenotypefrequencieswillmapontotheMendelian3:1phenotypicratio.ThemathematicalrelationbetweentheallelefrequenciesandthegenotypefrequenciesisAA:p2Aa:2pqaa:q2CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics2722wherep,2pq,andqarethefrequenciesofthegenotypesAA,Aa,andaainzygotesofanygeneration;pandqaretheallelefrequenciesofAandaingametesofthepreviousgeneration;andpþq¼1.Thechancethatallpossiblecombinationsofalleleswilloccurrandomly2is(pþq)¼1sothatwearriveatthefamousequationdescribingthe22Hardy-WeinbergPrinciple,pþ2pqþq¼1.Aslongasthebasicassumptionsholdintheabsenceoftheevolutionarycauses,theallelefrequenciespandqwillremainconstantandgenotypefre-quencieswillbeinaccordwiththeequation;inotherwords,thereisnoevolutionarychangeinapopulationinHardy-Weinbergequilibrium.Understandingevolutionaschangeingenefrequencies,then,isunderstandingthewaysinwhichpopulationsdeviatefromHardy-Weinbergequilibrium.Populationgeneticistsmaybeginwithassumptionsaboutthegeneticsystem,matingsystem,andpopu-lationstructure,andthenproceedtomodifythemathematicalrepresentationoftheHardy-WeinbergPrinciplebyaddingparam-etersformutation,migration,naturalselection,andrandomgeneticdrift.Considerasimplecase.First,considerthattheprecedingassumptionsconcerninggeneticsystem,matingsystem,andpopu-lationstructurehold.Second,assumewewanttounderstandhownaturalselectionmaycauseapopulationtodeviatefromHardy-Weinbergequilibrium;specifically,wewanttounderstandasimplecaseofviabilityselection.ThefrequencyofthegenotypesinourpopulationbeforeselectionisgivenbytheHardy-Weinbergequilib-22riumequation,pþ2pqþq¼1.Sincewewanttounderstandhownaturalselectioncausesadeviationfromthatequilibrium,wemodifytheequationtoincludeaparameterthatcapturesthe‘‘selectivepressure’’onthegenotypes,orinotherwords,theprob-abilityofsurvivorshipofthegenotype.Thisparameteriscalled‘‘fitness’’(w)andisusuallymeasuredrelativelysothatthefitnessofonegenotypeisexpressedrelativetoanothergenotype;thegenotypethatisthestandardofcomparisonisassignedafitnessvalueof1.Giventhefitnessparameter,ifthefrequenciesofthegenotypes22AA,Aa,aabeforeselectionarep,2pq,andq,respectively,thenthe2frequenciesofthegenotypesafterselectionarepwAA,2pqwAa,and2qwaabyincorporatingthefitnessesofthegenotypesinthecom-putationoftheirpostselectionfrequencies.Indeed,thesumoftheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n28robertal.millsteinandroberta.skipperjr.frequenciesofthegenotypesafterselectionequalstheaveragefit-nessforthepopulation,thatis,p2wþ2pqwþq2w¼w.AndweAAAaaahaveasthemathematicalrelationbetweentheallelefrequenciesandthegenotypefrequenciesp2w2pqwq2wAAAaaaAA:Aa:aa:wwwWemaythencomputethepostselectionfrequenciesofAanda,whicharedesignatedasp0andq0,respectively:p2wþpqwpqwþq2wp0¼AAAaq0¼AaaawwFromtheseequations,theoutcomeofselectioncanbededuced:Forinstance,ifp>p0,wherep¼[(p2þ2pq)/2],thenselectionisactingtodecreasethefrequencyofalleleAinthenextgeneration.Thepreviousexampleisnotintendedtoprovideaprimeronthestatisticalmethodsofpopulationgeneticsletaloneanunder-standingofthem.Rather,itisintendedtoprovidetheflavorofthewayinwhichevolutionaschangeingenefrequenciesisapproachedusingaversionofthosetools:Startingfromamathematicalstate-mentaboutthedistributionofallelefrequenciesintheabsenceofevolutionarycauses,onemayunderstandthewaysinwhichthosecauseschangethatdistributionbymodifyingthemathematicalstatementwithparametersmeasuringtheinfluenceofthosecauses.Indeed,mutation,migration,multiplemodesofselection,andran-domgeneticdriftaretreatedinmoreorlessthesameways,thatis,bymodifyingandextendingthebasicmathematicalstatementoftheHardy-WeinbergPrinciple.Moreover,thebasictoolsroughlyintroducedherecanbeexpandedtocoverevolutionforalternativeassumptionsregardingthegeneticsystem,matingsystem,andpopulationstructure.Further,thetheoreticalapparatuscanbemademorepowerfulandexpressivebyintroducingmodelsthatallowpopulationgeneticiststorepresenttheprobabilitiesofarangeofpossibleresults,ratherthansimplypredictingasingleresultasinthemodeldescribed.(Biologistscallmodelsthatpredictonespecificvaluedeterministicmodels;thisshouldnotbeconfusedwiththeLaplaceanorphilosophicalsenseofdeterminism,whichgenerallyreferstoapropertyoftheworldratherthanapropertyofamodel.ModelsthatprovideaprobabilitydistributionforarangeofresultsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics29arecalledstochasticmodels.Deterministicandstochasticmodelswillbediscussedfurtherinthenextsection.)4.populationgeneticstheoryandevolutionarycausesAsMichaelRusehasdocumented,CharlesDarwinconstruednat-uralselectionasacause(or,moreprecisely,averacausa)inordertoconformtothepredominantphilosophiesofscienceofhistime(Ruse1979,chap.7).Contemporarypopulationgenetics,aswehaveseen,incorporatesnotjustnaturalselection,butalsomutation,migration,andrandomgeneticdrift.Isnaturalselectionstillcon-struedasacause?Andaretheotherphenomenacausesaswell?Aswehaveseen,populationgeneticistsdefine‘‘evolution’’as‘‘changeingenefrequencies.’’Forselection,drift,mutation,andmigrationtobecausesofevolution,theymustbeabletobringaboutsuchchanges–atleasttheoretically,ifnotinrealityaswell.Unfortunately,eversinceDavidHume’sskepticalchallengetocause-effectrelationships,philosophershavebeenunabletoagreeonadefinitionof‘‘cause,’’orevenwhetherwecanlegitimatelyinfertheexistenceofcausesatall.Assuming,however,thattherearesuchthingsascausesandthatwecandevelopasatisfactoryaccountofcausation,itseemsfairtosaythatinthecontextofpopulationgenetics,mutation,migration,selection,anddriftarecausesofevolution.Forexample,itiseasytoseehowmutationwithinapopulationwillleadtoachangeingenefrequencieswithinthatpopulation.Similarly,migrationintoorawayfromapopulationalsoyieldsachangeingenefrequenciesinthepopulation.(Selectionanddriftwillbediscussedfurtherlater.)Theimplicationsofthepopulationgeneticist’sconstrualofevo-lutionarethreefold.First,thecommonlyheldnotionthatevolutionandnaturalselectionarethesameisfalse.Second,withfourpos-siblecausestoconsider,theequationscanbecomequitecomplex.Thisisbecause,unlikeinthesimplifiedpreviousscenarios,thesecausescanactincombination,asisimpliedbyRichardLewontin’ssuggestionthatpopulationgenetictheoryisadescriptivetheorythatprovidesthemappingofcausalprocessesasgeneticoutcomes.Itsays,‘ifmutationratesaresuchandsuch,ifthematingpatternissuchaone,iftherearefivegenesaffectingCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n30robertal.millsteinandroberta.skipperjr.thecharacterwiththefollowingnormsofreaction,thenthetrajectoryofthepopulationintime,ortheequilibriumstate,orthesteadystatedistribution4ofgenefrequencieswillbesuchandsuch.’(Lewontin1985,10)Third,eventhoughthecausescanactinconcert,theyarecon-sideredtobedistinctcauses.Themostdifficultcaseofdistin-guishingbetweenthecausesofevolutionisthatencounteredindistinguishingselectionfromdrift.Thus,wepayspecialattentiontothatcasehere.However,ourdiscussionherewillofnecessityberelativelybrief;forfurtherdiscussion,seeMillstein(2002)andear-lierworks(Beatty1984,Hodge1987).Theproblemindistinguishingselectionfromdriftarisesatleastinpartasaresultofambiguitiesinthemodelsofpopulationgenetics.Thiswillrequireanexplorationofthreedifferentaspectsofpopulationgenetics.Wewillarguethatitisamistaketochar-acterizeselectionanddriftintermsofthefirsttwooftheseaspects;thepropercharacterizationofselectionanddriftderivesfromthethirdaspect.Considerfirstthemodelofnaturalselectiondiscussedintheprevioussection.Althoughthefitnessvalue(thewintheequations)isgenerallyunderstoodtobeaprobability,namely,theprobabilityofsurvivorshipofthegenotype,theequationsthemselveswillnotgeneratearangeofpossiblefuturegenotypefrequencies.Rather,theywillgenerateonespecificgenotypefrequencyforeachofthegenotypes.Thatis,themodelofnaturalselectionis‘‘determinis-tic,’’inthesensedescribedpreviously.Ontheotherhand,accordingtothestandardpresentation,mod-elsbecomestochasticandgeneratearangeofpossiblegenotypefrequencies,when–and,accordingtosomeauthors,onlywhen(see,e.g.,Brandon2005;butcf.Millstein2005)–theassumptionofinfi-nitepopulationsizeisrelaxed.Tounderstandthis,imagineanurnfilledwithredandgreenballswhereballsaresampledwithoutrespecttocolor.Ifalargesampleofballsweretaken,wewouldexpectthefrequenciesofcoloredballsinthesampletobeveryclosetothefrequenciesintheurn.Ontheotherhand,ifweonlytakeasmallsampleofcoloredballs,oursamplemayverywellhavedif-ferentproportionsofcoloredballsthantheurndoes.Inthesameway,ifourpopulation(the‘‘sample’’thatistakenwitheachgen-eration)isinfinite,thenweexpect(withaveryhighprobability)thatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics31descendantgenerationswillhavegenotypefrequenciesveryclosetothoseoftheparentgenerations.However,ifthepopulationsizeisfinite,thenthesamplemaynotberepresentative;thatis,thege-notypefrequenciesofthedescendantgenerationmaydivergecon-siderablyfromthatoftheparents.Butinwhichdirectionwilltheydiverge?Forexample,willthefrequencyofAAsincreaseordecrease?Andbyhowmuch?Wecannotsayforcertain;wecanonlypredicttheprobabilityofvariousdivergences,justaswewouldnotbeabletosayforcertainwhetherasmallsampleofballswouldhaveasmallerorgreater(orequal)percentageofgreenballsascomparedtotheurn,onlytheprobabilitiesofdrawingvariousnumbersofgreenballs.So,theintroductionoffinitepopulationsizeyieldsastochasticmodel,thatis,amodelthatgeneratesaprobabilitydistributionoffutureoutcomes.Ifoneweretotrytounderstandwhatselectionanddriftarefromaliteralinterpretationofthesemodels,onemightbetemptedtoconcludethatnaturalselectionistheachievementofthepredictionsofthemodels.Thatis,onemightbetemptedtoconcludethatnaturalselectionoccurswhengenotypefrequenciesareexactlythosethatthefitnessvaluesleadustoexpect.Onemightbefurthertemptedtolabeltheintroductionoffinitepopulationsizeintothemodelsastheintroductionofdrift;again,readingliterallyfromthemodels,driftthenbecomesanydeviationfromtheexpectationsofselection.Onthisview,selectionisdeterministic,butdriftisstochastic(inthesensesdescribed).Thisis,infact,onewayofdistinguishingselectionfromdrift,but,asweshallarguelater,itisnotaverygoodway.Thepointtonoticenowisthatonthisinterpretationofpopulationgenetics,selectionanddriftaredistinguishedbytheoutcomesthatareproduced(agreementwithfitnesspredictionsanddivergencefromfitnessdifferences,respectively).Nowletusconsiderasecondaspectofpopulationgenetics.Indiscussingwhetherselectionordriftpredominatesinaparticularpopulation,biologistswillsometimesrelyonthefollowing‘‘ruleofthumb’’:naturalselectionhasprevailedif4Nes1,whereasran-domdrifthasprevailedif4Nes1,whereNeistheeffectivepopulationsize(i.e.,thenumberofindividualsinapopulationwhocontributeoffspringtothenextgeneration)andsistheselectioncoefficient(Futuyma1986,173).Inotherwords,whentheeffectivepopulationislargeand/ortheselectioncoefficientishigh,selectionCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n32robertal.millsteinandroberta.skipperjr.tendstoprevail.Whentheeffectivepopulationsizeissmalland/ortheselectioncoefficientislow,randomdrifttendstoprevail.Ifyouweretotakethissecondaspectofpopulationgeneticsonitsface,youprobablywouldcometoaverydifferentconclusionthanbefore.Namely,youwouldconcludethatnaturalselectionandrandomdriftarenotentirelydistinct;instead,itwouldappearthatselectionanddriftareonacontinuum.Withalowselectioncoef-ficientandasmallpopulationsize,youhavedrift,butincreasetheselectioncoefficientand/orthepopulationsizeandeventuallyyouwillhaveselection.Withanintermediateselectioncoefficientandanintermediatepopulationsize,however,itisunclearonthisviewwhetherthepopulationisundergoingselectionordrift.Althoughitmightnotappearso,theconclusionthatthereisacontinuumbetweenselectionanddriftisalsoreachedbyaconsiderationofoutcomes.Thequestionis,whichcontributesmoretothegenotypefrequencyproduced–theachievementoffitnessexpectationsorthedeviationfromthem?Theideaisthatwhentherearealowselectioncoefficientandsmallpopulationsize,theeffectsofdrift(theeffectsofsampling)swamptheeffectsofselection,butwhenthereareahighselectioncoefficientandlargepopulationsize,theeffectsofselectionswamptheeffectsofdrift.Thus,theliteralreadingofthesetwoaspectsofpopulationgeneticshasledtoaconundrum;twodifferentaspectsofthemodelsofpopulationgeneticsyielddifferentconclusionsaboutwhetherdriftandselectionaredistinctconcepts.Onthefirstview,theyaredistinctconcepts;selectionistheachievementoffitnessexpecta-tionsanddriftisanydeviationfromthatexpectation.Onthesecondview,thetwoconceptsarenotdistinct;rather,thereisacontinuumbetweendriftandselection.Thereis,however,athirdalternative,whichtakesanaltogetherdifferentapproach.Thisistheapproachthatoneofushasendorsed(e.g.,Millstein2002).Ratherthanliterallyinterpretingthemodelsinisolation,wederiveourconceptsfromphenomenathatthemodelsareintendedtorepresent.Interestinglyenough,thepresentationofthephenomenonofnaturalselectioninpopulationgeneticstextbooksgenerallydoesnotdeviatemuchfromDarwin’sownpresentation.Inorderforselectiontooccurinapopulation,1)theremustbeheritablevaria-tionamongindividuals,2)thevariationmustconferadifferentialCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics33abilitytosurviveandreproduceinthegivenenvironment,and3)moreoffspringareproducedthancansurviveinthegivenenviron-ment(i.e.,thereis,touseDarwin’sphrase,a‘‘struggleforexis-tence’’).ButfromtheseconditionsDarwindidnotconclude,astheselectionequationsdiscussedseemtoimply,thatorganismshavingadvantageousvariationswouldnecessarilybetheoneswithgreaterreproductivesuccess.Instead,Darwinclaimed,‘‘ifvariationsusefultoanyorganicbeingdooccur,assuredlyindividualsthuschar-acterizedwillhavethebestchanceofbeingpreservedinthestruggleforlife’’([1859]1964,127;italicsadded;seealso,e.g.,pp.61,81).Inotherwords,weexpectthatthefittestorganismswillbethemostsuccessful,butthatdoesnotalwayshappen;perhaps,forexample,thefittestorganismfailstofindfoodoriscrushedbyafallingboulder.Infact,unlessonewerewhiggishlytoclaimthatDarwin,inacknowledgingthattheexpectedmaynothappen,hadanotionofdrift,oneisleftwiththeconclusionthatthephenomenonthatDarwincallednaturalselection–arguably,thesamephenomenonthatthemodelsareattemptingtorepresent–isnot‘‘deterministic’’atall,butrather‘‘stochastic.’’Whatphenomena,then,arethedriftmodelspurportedlyrepre-senting?Thereareatleastsevendifferentkindsofdriftphenomena(Millstein2002).Here,wementiononlytwo:indiscriminategametesamplingandindiscriminateparentsampling(seeBeatty1984).Gametesamplingistheprocessinwhichsome–butnotall–gametesaresuccessfullyunitedinzygotes,whereasparentsamplingistheprocessinwhichsome–butnotall–organismssuccessfullyreproduceandbecomeparents.Butthereisatemptingrejoinder,whichgoessomethinglikethefollowing:‘‘Butwhyisitthatsomegametesbecomejoinedtogetherinzygotesandothersdonot?Perhapsthesuccessfulgametesarefitterinsomeway;perhapsthespermswimfasterortheeggsaremorerobust.Andwhyisitthatsomeorganismssurvivetobecomeparentswhenothersdonot?Again,perhapstheyarejustfitter.’’Thisrejoinderputsitsfingeronthedifferencebetweenthephenomenonofselectionandthephe-nomenonofdrift.Ifsomegameteswerefitter,oriftheindividualswerefitter,thenwewouldnotbedescribingdriftatall;wewouldbedescribingselection(i.e.,discriminatesampling).Thepointbehinddiscussingdriftisthattheremaynotbeanyfitnessdifferences(althoughtheremaybephysicaldifferencesthatdonotconferanyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n34robertal.millsteinandroberta.skipperjr.fitnessbenefits),andyetsomegametesorindividualsmaystillbemoresuccessfulthanothers.TouseHartlandClark’sexample,imagineshellfishthat‘‘producevastnumbersofpelagiclarvaethatdriftaboutinthesea’’(1989,70).AlthoughHartlandClarkdonotelaborate,theimageisofvirtuallyidenticallarvae,subjecttothevagariesoftidesandpredators(i.e.,indiscriminatesampling).Thus,examinationofthephenomena,priortoanyrepresentationbythemodels,yieldsathirdwayofunderstandingthedifferencebetweenselectionanddrift.Selection,onthisview,isadiscriminatesamplingprocessinwhichphysicaldifferencesbetweenbiologicalentities(gametes,organisms,etc.)arecausallyrelevanttodiffer-encesinreproductivesuccess.Drift,ontheotherhand,isanindis-criminatesamplingprocessinwhichphysicaldifferencesbetweenbiologicalentitiesarecausallyirrelevanttodifferencesinrepro-5ductivesuccess.Notethatunlikeinthefirsttwoattemptstospelloutthedifferencebetweenselectionanddrift,thedistinctionismadebyidentifyingselectionanddriftasdifferenttypesofprocessesratherthandifferentoutcomes.Inotherwords,selectionanddriftaredifferentkindsofcausalprocesses.Inusingtheterm‘‘causalprocess,’’wemeantosuggestthatselectionanddriftarephysicalprocessesoccurringinnatureandinthelaboratory;furthermore,theyaretobedistinguishedfrompseudoprocessessuchasthemovementofashadow(Salmon1984).Finally,inusingtheterm‘‘causalprocess’’todescribeselectionanddrift,wemeantosuggestthatselectionanddriftconsistofaseriesofstatesoccurringthrough6time,wherethestatesaregeneratedcausally.Theoutcomesoftheseprocesses,ontheotherhand,refertoonestate(e.g.,thegeno-typefrequenciesofapopulation)ataparticularpointintime.Notonlyareselectionanddriftdifferentkindsofcausalpro-cesses,theyaredifferentkindsofcauses,bothofwhichcanleadtoevolution.Consideringselectionfirst,iftheindividualswhosevar-iationsconferonthemagreaterabilitytosurviveandreproducedoinfactreproduceingreaternumbersthanindividualswholackthesevariations,thenthegenefrequenciesofthesecondgenerationhavechangedfromthoseoftheprecedinggeneration.Nowthereexistsagreaterproportionofindividualswith‘‘advantageous’’variations;evolutionhasoccurred.Naturalselectionhascausedevolution.Butifthereisachangeintheproportionsoftypesfromonegenerationtothenext,butthatchangeisnotduetophysicaldifferencesbetweenCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics35individuals,thenitisdriftthathascausedevolution.Eachisadifferentcausebecauseeachisadifferentkindofcausalprocess.Thereareatleastthreeadvantagestothisviewovertheothertwo.First,wethinkitisamistaketointerpretmodelsliterally.Themodelsofpopulationgeneticsarehighlyidealizedmetamathematicalstructuresthat,atbest,areunderstoodasbearingasimilarityrela-tionshiptotherealworldsystemstheydescribe.Anddeterminingpreciselyhowtounderstandtheextentofsimilaritybetweenthemodelsandtherealworldisnoeasytask(Wimsatt1980a).Atanyrate,themodelsweredevelopedsubsequenttounderstandingthephenomenatheyaretryingtocapture.Itseemsbackward,therefore,totrytounderstandthephenomenaviathemodels.Second,confusionarisesindistinguishingselectionanddriftinlargepartbecausepopulationgeneticistssometimesspeakofselectionanddriftascausalprocesses(asinthequotefromLewontin),yetatothertimestheyspeakofselectionanddriftasoutcomes,oreffects(thus,e.g.,driftissometimesreferredtoasthe‘‘SewallWright7Effect’’).Amoment’sreflectionwillshow,however,thatpopulationgeneticistscannothaveitbothways.If,forexample,selectionisidentifiedwithitsoutcomes,thenselectionoccurswhenorganismshavingagreaterabilitytoreproduceascomparedtotheirconspecificsdoinfactenjoygreaterreproductivesuccess.However,thisisjustevolution,thatis,achangeingenefrequenciesfromonegenerationtothenext.Itwouldnotmakesenseforselectioninthissensetobeacauseofevolution;selection,consideredasanoutcome,isoneformofevolution.But,aswediscussedearlier,biologistscommonlycon-strueselectionasacauseofevolution.Thismakessenseontheviewofselectionasaprocess,butnotontheviewofselectionasanoutcome.Andthird,furtherconfusionarisesbecausetheoutcomesofthedifferentprocessesoftencannotbedistinguished.Toseethis,firstconsiderapopulationinwhichphysicaldifferencesbetweenorganismsdonotconferanydifferencesinsurvivalorreproductiveability(apopulationundergoingdrift,onouraccount),sothattherelativevaluesofdifferenttypesmayfluctuatefromgenerationtogeneration.Now,considerasecondpopulation,inwhichphysicaldifferencesbetweenorganismsdoconferdifferencesinsurvivalandreproductiveability(apopulationundergoingselection,onouraccount).SupposethattheenvironmentofthesecondpopulationisCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n36robertal.millsteinandroberta.skipperjr.fluctuating.Becauseofthefluctuatingenvironment,differenttypesmaybefavoredindifferentgenerations,producingafluctuationoftypesoverthegenerationsthatproducesapatternthatisindis-tinguishablefromthatofdrift.Ifselectionanddriftareunderstoodpurelyasoutcomes,theneitherboththepopulationsareundergoingdriftorboththepopulationsareundergoingselection.Weareskep-ticalthatanybiologistwould,whenpresentedwiththisscenario,actuallytakeeitherofthesepositions,becausethetwopopulationsarebiologicallyverydifferent.Characterizingdriftandselectionasprocessesinsteadofoutcomescaptures,ratherthanglossesover,thatdifference.Weshouldemphasizethattheclaimhereisthattheconceptsofnaturalselectionandrandomgeneticdriftcanbedistinguishedfromoneanother,andthatthatdistinctionshouldbebasedonthekindsofprocessestheyare,andnotonthekindsofoutcomes,asaliteralinterpretationofthemodelswouldsuggest.However,thatisnottosuggestthatselectionanddriftcanbeeasilydistinguishedempiri-cally.That,unfortunately,isamuchmorecomplicatedproblem,whichwepresentinamoregeneralfashioninthefollowing.5.causes,butnotallofthecausesEventhoughpopulationgeneticsmodelsthecausesofevolution,itisnotclearthatpopulationgeneticstellsacompletecausalstoryofevolution.Thatis,thequestionarisesastowhethertherearecausesinvolvedintheprocessofevolutionthatarenotcapturedbythemodelsofpopulationgenetics.Herewewillfocusonjustonearea:theconceptoffitness.Thereisanextensivebodyofliteratureontheconceptoffitness,especiallyonthepropensityinterpretationoffitness(Brandon1978,MillsandBeatty1979).However,althoughweacknowledgeourintellectualdebttothepropensityinterpreta-tion,wedonotintendourdiscussionheretobeadefenseofthisoranyotherinterpretationoffitness.Rather,inthissectionweseekonlytoexploresomeissuesofcausalitythattheconceptoffitnessraises.8Earlier,wearguedthatnaturalselectionisacauseofevolution.Yetnaturalselectionisacausalprocessintwosenses:itisitselfacause(ofevolution),anditismadeupofcauses.Itistothislattersense,thecausalitywithintheprocess,thatwenowturn.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics37Primafacie,itmakessensetoinvokefitnessasacausalconceptinacolloquialdescriptionofnaturalselection.Afterall,whatmakesoneorganismfitterthananotherorganisminagivenenvironmentareitsphysicalcharacteristics.Andthosephysicalcharacteristicscausetheorganismtohavesuperiorsurvivalandreproductivesuccessinthegivenenvironment.Thatis,theorganism’ssuperiorfitness,underauspiciouscircumstances,causesitssuperiorreproductivesuccess.Theidentificationof‘‘fitness’’asacauseofreproductivesuccessseemssotrivialastobetautological.However,asElliottSober(1984)haspointedout,thesituationisnotassimpleasthecolloquialstorymakesitappear.Afterall,inpopulationgenetics,fitness(thewintheequationsdescribed)doesnotrepresentjustonephysicaltrait.Thatis,thefitnessofafinchisnotjusttheshapeandsizeofitsbeak,notjustitsabilitytoavoidpredatorsanddisease,andnotjustitsabilitytofindamateandreproduce;itsfitnessisallofthosethingstogether.Soberarguesthatwhereasanyofthoseelementsindividuallymaycauseorpreventreproductivesuccess,theygenerallydonotallcauseorpreventreproductivesuccessinanygiveninstance.Thus,accordingtoSober,fitnessis‘‘causallyinert.’’Itshouldbenoted,however,thatpopulationgeneticsmodelstypicallysubdividefitnesstraitsintotwobroadcomponents:anorganism’sabilitytosurvive(viability)andanorganism’sabilitytoreproduce(fecundity).Yetthisgrossdistinctiondoesnotaddressthedifferentkindsofabilitiesthatfitnessencompasses.Considerfecundity.Inthemodelsofpopulationgenetics,fecundityencom-passesatleastthreedifferentkindsofabilities:1)anorganism’sabilitytoproducevariousnumbersofoffspring,2)anorganism’sabilitytoattractmates,and3)anorganism’sabilitytofightforpotentialmates.ForDarwin,ontheotherhand,thelattertwoabilitieswereinadistinctcategory.Infact,Darwinconsideredthesetraitstobethebasisofsexualselection,whichforDarwinwasadifferenttypeofcausefromnaturalselection.Thisisbecausenaturalselection,whichwouldinvolveviabilityaswellasfecundityinthefirstsense,wouldtendtoproduceorganismsthatareadaptedtotheirenviron-ments:fincheswithbeaksofacertainshapeandsize,forexample.Sexualselection,ontheotherhand,wouldprimarilyinvolvefecundityinthesecondorthirdsenses,whichwouldmeanthatthetraitsthatwereproducedwouldnottendtobethosethatwereadaptive.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n38robertal.millsteinandroberta.skipperjr.Instead,wefindsexualselectionyieldingtraitsliketheinefficient-but-beautifultailofapeacock(asaresultofthesecondsenseoffecundity)orantlersofamaledeer(asaresultofthethirdsenseoffecundity).Ifsexualselectionandnaturalselectiontrulyarediffer-entkindsofcauses,asDarwinthought,thentheyarecausesthatpopulationgeneticsmodelsdonotdistinguishbetween(inthesensethattheyarenottreatedanydifferently–althoughofcoursesexualselectionisexplicitlydiscussedinmostpopulationgeneticstext-books),andthisisinpartaresultofnotdistinguishingamongdif-ferenttypesoffecundity.Furthercomplicatingmattersisthefactthat,asdiscussed,populationgeneticsusesrelativefitness.Thatis,itisnotanindi-vidual’sabilitytosurviveandreproducethatappearsintheequa-tion,butrather,its–or,moreprecisely,itsgenotype’s–abilitytosurviveandreproduceascomparedtotheabilitiesofothersinthe9population.Ofcourse,relativefitnessesarecalculatedfromgeno-typefitnesses.Butthisisnotsimplecomputationalconvenience;therelativefitnessesarewhatreallymatterfortheevolutionofthepopulation.Iftherearenofitnessdifferencesamonggenotypes,therewillbenoselection,andapopulationinwhichonegenotypeistwiceasfitasitsconspecificwillevolvedifferentlythanoneinwhichitisthreetimesasfit.Andyet,isrelativefitnessacausalconcept?Ifitis,itdoesnotseemtobeacausalconceptthatadherestoanyoneindividual.So,interpretingthecausalityofrelativefitnessischallenging.Thus,itwouldappeareitherthatpopulationgeneticsfailstocapturefullytheactualcausesthatoperateinapopulationundergoingselection,orthatbiologistsandphilosophershaveyettoprovideanadequateinterpretationoffitness.Giventhenumerousalternativesintheliterature,thelatterwouldseemtobetheconsensusview.6.howdowefindthecauses?Thusfarwehaveinterpretedpopulationgeneticsasacausaltheoryfromthepointofviewofthetheory,itsstructure.Ourviewmaybeexpandedbyconsideringhowpopulationgeneticstheoriesarere-latedtorealpopulationsoforganisms.Whatwewanttoexplore,then,ishowpopulationgeneticistscredentialempiricalcausalclaimsmadeabouttheirmodels.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics39ElisabethLloyd(1988)hasconstructedasimple,broadframeworkfortheconfirmationofevolutionaryandecologicalmodels.Onherview,empiricalclaims,orhypotheses,thatsomemodelissimilartosomenaturalsystemareconfirmedbywayof(C1)fitbetweenmodelanddata,(C2)independentsupportforaspectsofthemodel,and(C3)varietyofevidence.Fitbetweenamodelanddataisjustevidencethatdemonstratesamatchingbetweenthemodelandthedata.Sincepopulationgeneticstheoriesareembeddedwithnumerousassumptions,forexample,thatmatingisrandomorthatpopulationsarelarge,anyindependentempiricalsupportforthoseassumptionswillincreasetheconfirmationalstandingofamodel.Mostimpor-tantisthatthereareavarietyoftypesofsupportforamodel,thatis,avarietyofinstancesoffitandavarietyofinstancesofindependentsupportforanyassumptions.OnLloyd’sview,standardstatisticaltechniquesforanalyzingdata,commonacrossthebiosciences,areusedtoanalyze(C1)–(C3).Oneofus(Skipper2004)hasarguedthataconstellationofexperimentalstrategiesformsthebasisofLloyd’sconfirmationframework.Theideahereisthatfit,independentsupport,andvarietyofevidencearealldrivenbyexperimentalmethodology.Thus,forinstance,fitbetweenmodelanddataisatbottomdrivenbythepracticalproceduresandtechniquesusedtocollectthedata.Thereareanumberofstrategies,including(E1)experimentalchecksandcalibration,(E2)reproducingartifactsknowntobepresent,(E3)intervention,(E4)independentconfirmationusingdifferentexperi-ments,(E5)eliminationoferror,(E6)usingtheresultstoarguefortheirvalidity,(E7)usinganindependentlywell-corroboratedtheoryofthephenomenontoexplaintheresults,(E8)usinganapparatusbasedonawell-corroboratedtheory,and(E9)usingstatisticalargu-ments(cf.Rudge1998).Andtheyareusedtojustifyexperimentalclaimsacrossthethreemainclassesofexperiment,thatis,naturalexperiments,orobservationsofevolutioninaction,fieldexperi-ments,controlledmanipulationsofpopulationsinthewild,andlaboratoryexperiments,highlycontrolledmanipulationsofpopu-10lationsinthelaboratory(Diamond1986).Asanillustration,considerthefamouscaseoftheScarletTigermoth,Panaxiadominula,perhaps(atmorethansixtyyears)thelongest-runningfieldstudyinecologicalgenetics,firststudiedbyR.A.FisherandthefamedecologicalgeneticistE.B.Ford(FisherandCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n40robertal.millsteinandroberta.skipperjr.Ford1947).FisherandFordcarriedoutafieldexperimentusingthenovel(atthetime)captureandreleaseprotocoltodeterminewhethernaturalselectionorrandomgeneticdriftcausedthefluctuationsinthemedionigrageneofthemoth,responsibleforaspecificwingcolorationphenotype.BycapturingmothsfromanOxfordshirefield,markingthemwithadabofpaint,andrecapturingthem,FisherandFordcould,overtime,trackthefluctuationsinthemedionigrageneviaitsphenotype,aparticularwingcolorationpattern.Fromdatacollectedbetween1939and1946,FisherandFordperformedsta-tisticalanalyses(E9)thatsupportedafitwiththeirselectionistmodel(C1)overarandomgeneticdriftmodel.NotethatFisherandFordinferredthatselectioncontrolledthefateofthemedionigrageneandnotthattheyhaddirectobservationalevidenceofselectionactingonthemothswiththisphenotype.Theirstatisticalargument(E9)wasthatthefluctuationsinthegenefromyeartoyearweretoogreattobeduetodriftand,sotheymusthavebeencausedbyselection.Overthelast10–15years,furtherfieldandlaboratoryexperi-mentshaverevealedflawsinFisherandFord’sexperimentalpro-cedures,flawsthathavebeencarriedthroughthesixtyyearsoffieldworkonPanaxia(e.g.,GoulsonandOwen1997).FisherandFord,aswellassubsequentexperimenters,failedtoaccountfortemperaturefluctuationsinthemoths’environmentduringtheirexperiment.TheinterpretationofFisherandFord’sresultshingesonperformingthisexperimentalcheck(E1),sincetemperaturefluctuationsaffecttheexpressionofthemedionigrageneinthemothduringthelarvalstage,turningthewingcolordarkerineitheroftheextremesofcoldorhot.Affectedmothswiththemedionigragenewouldlookmorelikethedominantform,f.dominula,and,thus,wouldbescoredassuch.Thecapture-mark-releasecensusdatawouldbeskewed,and,thus,thefitbetweentheselectionistmodelandthedatawouldbecalledintoquestion.Becausethecheckwasnotperformed,alargeportionoftheexperimentalresultsonPanaxiaareambiguous.Ironically,SewallWright(1948)pointedoutthatFisherandFord’s(1947)‘‘argumentbyelimination’’ofdrifttoselectionwasnotasstrongastheybelieved.ButWright’scritiques,nowvindicated,wereignored.TheexampleoftheScarletTigermothhighlightsthewayinwhichweviewthewaypopulationgeneticists(andecologicalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics41geneticists)credentialempiricalorexperimentalclaimsaboutevo-lutionarycauses.Theexamplealsohighlightsthewaysinwhichsuchclaimscangowrong.Theprecedinghasonlybeenasketch,butwethinkitisaplausibleone.What,ultimately,canbeclaimedaboutthestrengthofsuchclaims?Lewontin(2000b)makesplainthattheepistemologicallandscapeofpopulationgeneticsisacontinuumfroma‘‘maximalinferentialprogram’’anda‘‘minimaldeductivepro-gram.’’Theminimalprogramistantamounttotheoreticalpopulationgenetics,ortheprogramofprovidingthenetworkofrelationshipsbetweenevolutionarycausesandtheiroutcomesatthegeneticlevel.Themaximalprogramaimstogiveacorrectaccountofevolutionarycausesthathaveledtoanyandallobservedpatternsofgeneticvar-iationinnaturalpopulations.Themaximalprogramisepistemolo-gicallyunrealisticbecauseitrequiresthatscientistsknowtheapparentlyunknowable:allofthebiologicalandnaturalhistoricaldetailsofanyarbitrarilychosenspecies.Theminimalprogramisentirelyanalytic,having‘‘notrulyepistemologicalproblems,onlyquestionsofmethodologicalingenuity’’(Lewontin2000b,200).SomewherebetweenthetwoextremesiswhatLewontinthinksisepistemicrealityinpopulationgenetics.Indeed,Lewontinclaims,‘‘thebesttowhichpopulationgeneticistscanaspireisaformalstructurethatsetsthelimitsofallowableexplanationandasetofexistentiallymodifiedclaimsaboutwhathasactuallyhappenedintherealhistoryoforganisms’’(Lewontin2000b,213).Weagree.7.conclusionTheproblematicofpopulationgeneticsistoaccountforthedynamicsofgeneticvariationinnaturalpopulationsviathecausesofevolution,namely,mutation,migration,naturalselection,andrandomgeneticdrift.Thepresentessayhasverybrieflysurveyedthehistoricaldevelopmentofpopulationgenetics,thecurrentmodeltheoreticstructureofpopulationgenetics,keyconceptualproblemsinunderstandingimportantevolutionarycauses,andtheproblemofferretingoutthosecausesviatheoreticalandexperimentalwork.Wehaveheremanagedtopickawayonlyattheverytipoftheiceberg;thereismuchphilosophicalworktobedoneonthekeythemeofthisessay–populationgeneticsasacausaltheory–aswellasonotherproblemsofthefield.Moreover,itisimportanttonotethatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n42robertal.millsteinandroberta.skipperjr.whilepopulationgeneticsaimstounderstandthecausesofevolu-tion,itisgenerallyunderstoodasnotcontributingsignificantlyoratalltoourunderstandingofotherevolutionarycauses,suchasthecausesofspeciationandextinction.Thatis,thereismoretoevo-lutionarystudiesthanpopulationgenetics,inspiteofthedramaticproblemsandprogressofthefield.ACKNOWLEDGMENTSThankstoJohnBeatty,MichaelDietrich,JonHodge,DavidHull,AnyaPlutynski,andMichaelRuseforhelpfulcommentsanddiscussiononanearlierdraftofthisessay.Remaininginfelicitiesareourown.notes1.Biologistsrefervariouslyto‘‘randomgeneticdrift,’’‘‘geneticdrift,’’‘‘randomdrift,’’orsimply,‘‘drift.’’Theseareallthesamephenomen-on.2.See,inparticular,Chapter18(Laubichler)ofthisvolumeonthechallengeissuedfrom‘‘evo-devo.’’3.Thereissomequestionhereaboutwhatcountsasan‘‘assumption’’andwhatcountsasa‘‘cause,’’aswellasconcerningwhichassumptionsare‘‘basic.’’Hereweignorethesecomplexitiesandsimplyechothemostcommonwaythatpopulationgeneticsmodelsarepresented.4.Lewontin’spointhereisactuallytoshowthelimitationsofpopulationgenetics,anissuethatwillbediscussedlater;hisquotealsohintsatanevengreatercomplexitytothecausalityofpopulationgeneticsthanispresentedhere.5.JohnBeattyhassuggestedtous(personalcommunication)thatperhapsindiscriminatesamplingphenomenaoughttobeseenascausesofdriftratherthandriftitself.Onthisview,driftshouldbeseenastheanalogtoevolutionbynaturalselection,nottheanalogtonaturalselectionitself;theanalogtoselection(discriminatesampling)wouldbesimplyindiscriminatesampling.Herewemustacknowledgethatbiologistsdosometimesrefertoindiscriminatesamplingasdriftandatothertimesrefertotheeffects(oroutcomes)ofindiscriminatesamplingasdrift.So,ifwewanttopreventconfusion,wehavetodecidewhetherdriftreferstothecausalprocessortheoutcome.Sincebiologistsdosometimesspeakofdriftandselectionasalternatives,andsincetheyarebothinabroadsensetreatedascausesofevolutioninpopulationgeneticstextbooks(whereitiscommontolistselection,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPopulationGenetics43drift,mutation,andmigrationascauses),wearguethatitmakessensetohavedriftrefertothecausalprocess.Thatis,itmakessensetotreatdriftinthesamewaythatselectionistreated.However,itisprobablytruethatinsomesenseitdoesnotmatterwhetheryoucallthecausalprocess‘‘drift’’ortheoutcome‘‘drift,’’aslongasitisclearwhenoneisreferringtothecausalprocessandwhenoneisreferringtotheoutcome(Millstein2002).6.Eventhoughindiscriminatesamplingischaracterizedasaprocesswherebyphysicaldifferencesbetweenbiologicalentitiesarecausallyirrelevanttodifferencesinreproductivesuccess,itisstillacausalprocessinthissense.Forexample,ingametesampling,theunitingofgametestoformzygotesovertimeconsistsofmanystatesthataretheresultofunderlyingcauses.7.OnemighteasilyobjecttotheattributionofthisphenomenontoSewallWright;ourpointhereisonlytoemphasizehowcommonitisfordrifttobecalledaneffect.8.Evenhereaproblemofinterpretationpresentsitself:atwhatlevel(gene,organism,group,species)doesthiscauseoperate?ButseeChapter3(Lloyd),thisvolume.9.Usingthefitnessofgenotypesratherthanthefitnessofindividualsmaybeanotherwayinwhichpopulationgeneticsdoesnotcaptureacompletecausalpicture–assuming,thatis,thatthecausalstoriesofindividualsarepartoftheevolutionarystory,whichisnotentirelyclear.10.Diamondintroduceshisownexperimentalstrategiesaswell.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nelisabetha.lloyd3UnitsandLevelsofSelectionThetheoryofevolutionbynaturalselectionis,perhaps,thecrowningintellectualachievementofthebiologicalsciences.Thereis,however,considerabledebateaboutwhichentityorentitiesareselectedandwhatitisthatfitsthemforthatrole.InthischapterIaimtoclarifywhatisatissueinthesedebatesbyidentifyingseveraldistinct,thoughoftenconfused,concernsandthenidentifyinghowthedebatesonwhatconstitutetheunitsofselectiondependtoasignificantdegreeonwhichofthesedifferentquestionsathinkerregardsascentral.Chiefamongthesedistinctionsarereplicatorsversusinteractorsaswellaswhobenefitsfromaprocessofevolutionbyselection,thatis,whobenefitsinthelongrunfromaselectionprocessandwhogetsthebenefitofpossessingadaptationsthatresultfromaselectionprocess.BecauseRichardDawkinsistheprimarysourceofseveraloftheconfusionsaddressedinthisessay,Itreathisworkatsomelength.1.introductionFormorethantwenty-fiveyears,certainparticipantsinthe‘‘unitsofselection’’debateshavearguedthatmorethanoneissueisatstake.RichardDawkins(1978,1982a),forexample,introduced‘‘repli-cator’’and‘‘vehicle’’tostandfordifferentrolesintheevolutionaryprocess.Heproceededtoarguethattheunitsofselectiondebatesshouldnotbeaboutvehicles,astheyhadformerly,butaboutreplicators.DavidHull(1980)inhisinfluentialarticle‘‘IndividualityandSelection’’suggestedthatDawkins’s‘‘replicator’’subsumestwoquitedistinctfunctionalrolesandbrokethemupinto‘‘replicator’’and‘‘interactor.’’RobertBrandon(1982),arguingthattheforceof44CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection45Hull’sdistinctionhadbeenunderappreciated,analyzedtheunitsofselectioncontroversiesfurther,claimingthatthequestionaboutinteractorsshouldmoreaccuratelybecalledthe‘‘levelsofselec-tion’’debatetodistinguishitfromthedisputeaboutreplicators,whichheallowedtokeepthe‘‘unitsofdebate’’title.Thepurposeofthischapteristodelineatefurtherthevariousquestionspursuedundertherubricof‘‘unitsandlevelsofselection.’’Thisanalysisisnotmeanttoresolveanyoftheconflictsaboutwhichresearchquestionsaremostworthpursuing;moreover,Imakenoattempttodecidewhichofthequestionsorcombinationsofquestionsdiscussedoughttobeconsidered‘‘the’’unitsofselectionquestion.2.fourbasicquestionsWithrespecttothecontroversiesthatsurroundtheunitsandlevelsofselectionquestion,fourbasicquestionscanbedelineatedasdis-tinctandseparable.AswillbedemonstratedinSection3,thesequestionsareoftenusedincombinationtorepresenttheunitsofselectionproblem.Butletusbeginbyclarifyingterms(seeLloyd1992,2001).Thetermreplicator,originallyintroducedbyDawkinsbutsincemodifiedbyHull,isusedtorefertoanyentityofwhichcopiesaremade.Dawkins(1982a,47)classifiesreplicatorsusingtwoortho-gonaldistinctions.A‘‘germ-line’’replicator,asdistinctfroma‘‘dead-end’’replicator,is‘‘thepotentialancestorofanindefinitelylonglineofdescendantreplicators’’(1982a,47).Forinstance,DNAinachicken’seggisagerm-linereplicator,whereasthatinachicken’swingisadead-endreplicator.An‘‘active’’replicatoris‘‘areplicatorthathassomecausalinfluenceonitsownprobabilityofbeingpropagated,’’whereasa‘‘passive’’replicatorisnevertran-scribedandhasnophenotypicexpressionwhatsoever.Dawkins(1982a,47)isespeciallyinterestedinactivegerm-linereplicators,‘‘sinceadaptations‘for’theirpreservationareexpectedtofilltheworldandtocharacterizelivingorganisms.’’Dawkins(1982b,295)alsointroducedthetermvehicle,whichhedefinesas‘‘anyrelativelydiscreteentity...whichhousesreplica-tors,andwhichcanberegardedasamachineprogrammedtopre-serveandpropagatethereplicatorsthatrideinsideit.’’AccordingtoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n46elisabetha.lloydDawkins(1982a,62),mostreplicators’phenotypiceffectsarerepresentedinvehicles,whicharethemselvestheproximatetargetsofnaturalselection.Hull(1980,318),inhisintroductionoftheterminteractor,observesthatDawkins’stheoryhasreplicatorsinteractingwiththeirenvironmentsintwodistinctways:theyproducecopiesofthemselves,andtheyinfluencetheirownsurvivalandthesurvivaloftheircopiesthroughtheproductionofsecondaryproductsthatultimatelyhavephenotypicexpression.Hull(1980,318)suggestsinteractorfortheentitiesthatfunctioninthissecondprocess.Aninteractordenotesthatentitythatinteracts,asacohesivewhole,directlywithitsenvironmentinsuchawaythatreplicationisdif-ferential–inotherwords,anentityonwhichselectionactsdirectly.Theprocessofevolutionbynaturalselectionis‘‘aprocessinwhichthedifferentialextinctionandproliferationofinteractorscausethedifferentialperpetuationofthereplicatorsthatproducedthem’’(Hull1980,318;seealsoBrandon1982,317–18).Hullalsointroducedtheconceptof‘‘evolvers,’’whicharetheentitiesthatevolveasaresultofselectiononinteractors:theseareusuallywhatHull(1980,327)callslineages.Sofar,noonehasdirectlyclaimedthatevolversareunitsofselection.Theycanbeseen,however,tobeplayingaroleinconsideringthequestionofwhoownsanadaptationandwhobenefitsfromevolutionbyselection,whichwewillconsiderinSections2.3and2.4.2.1TheInteractorQuestionInitstraditionalguise,theinteractorquestionis,Whatunitsarebeingactivelyselectedinaprocessofnaturalselection?Assuch,thisquestionisinvolvedintheoldestformsoftheunitsofselectiondebates(Darwin1859,Haldane1932,Wright1945).Inhisclassicreviewarticle,Lewontin(1970,7)contraststhelevelsofselection,‘‘especiallyasregardstheirefficiencyascausersofevolutionarychange.’’Similarly,SlobodkinandRapaport(1974,184)assumedthataunitofselectionissomethingthat‘‘respondstoselectiveforcesasaunit–whetherornotthiscorrespondstoaspatiallylocalizeddeme,family,orpopulation.’’Questionsaboutinteractorsfocusonthedescriptionoftheselectionprocessitself,thatis,ontheinteractionbetweenanentity,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection47thatentity’straitsandenvironment,andhowthisinteractionproducesevolution;theydonotfocusontheoutcomeofthisprocess(seeWade1977;VrbaandGould1986).Theinteractionbetweensomeinteractoratacertainlevelanditsenvironmentisassumedtobemediatedby‘‘traits’’thataffecttheinteractor’sexpectedsur-vivalandreproductivesuccess.Here,theinteractorispossiblyatanylevelofbiologicalorganization,includingagroup,akin-group,anorganism,agamete,achromosome,oragene.Someportionoftheexpectedfitnessoftheinteractorisdirectlycorrelatedwiththevalueofthetraitinquestion.Theexpectedfitnessoftheinteractoriscommonlyexpressedintermsofgenotypicfitnessparameters,thatis,intermsofthefitnessofcombinationsofreplicators;hence,interactorsuccessismostoftenreflectedinandcountedthroughreplicatorsuccess.Severalmethodsareavailableforexpressingsuchacorrelationbetweeninteractortraitand(genotypicorgenic)fitness,includingpartialregression,variances,andcovariances.Infact,muchoftheinteractordebatehasbeenplayedoutthroughtheconstructionofmathematicalgeneticmodels–withtheexceptionofWade’s(1978,1980)andsomeofWilsonandColwell’s(1981)workonfemale-biasedsexratios(seeespeciallyGriesemerandWade1988).Thepointofbuildingsuchmodelsistodeterminewhatkindsofselection,operatingatwhichlevels,maybeeffectiveinproducingevolutionarychange.Itiswidelyheld,forinstance,thattheconditionsunderwhichgroupselectioncaneffectevolutionarychangearequitestringentandrare.Typically,groupselectionisseentorequiresmallgroupsize,lowmigrationrate,andextinctionofentiredemes.Somemodelers,how-ever,disagreethatthesestringentconditionsarenecessary.MatessiandJayakar(1976,384),forexample,showthatintheevolutionofaltruismbygroupselection,verysmallgroupsmaynotbenecessary,contraMaynardSmith(1964).WadeandMcCauley(1980,811)alsoarguethatsmalleffectivedemesizeisnotanecessaryprerequisitetotheoperationofgroupselection.Similarly,Boorman(1978,1909)showsthatstrongextinctionpressureondemesisnotnecessary.Andfinally,Uyenoyama(1979)developsagroupselectionmodelthatviolatesallthreeofthe‘‘necessary’’conditionusuallycited.Thatdifferentresearchersreachsuchdisparateconclusionsabouttheefficacyofgroupselectionoccurspartlybecausetheyareusingdifferentmodelswithdifferentparametervalues.Wade(1978)CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n48elisabetha.lloydhighlightedseveralassumptions,routinelyusedingroupselectionmodels,thatbiasedtheresultsofthesemodelsagainsttheefficacyofgroupselection.Forexample,henotedthatmanygroupselectionmodelsuseaspecificmechanismofmigration;itisassumedthatthemigratingindividualsmixcompletely,forminga‘‘migrantpool’’fromwhichmigrantsareassignedtopopulationsrandomly.Allpopulationsareassumedtocontributemigrantstoacommonpoolfromwhichcolonistsaredrawnatrandom.Underthisapproach,whichisusedinallmodelsofgroupselectionpriorto1978,smallsamplesizeisneededtogetalargegeneticvariancebetweenpopu-lations(Wade1978,110;seediscussioninOkasha2003).If,incontrast,migrationoccursbymeansoflargepopulations,higherheritabilityoftraitsandamorerepresentativesamplingoftheparentpopulationwillresult.Eachpropagateismadeupofindividualsderivedfromasinglepopulation,andthereisnomixingofcolonistsfromthedifferentpopulationsduringpropaguleforma-tion.OnthebasisofSlatkinandWade’s(1978,3531)analysis,muchmorebetween-populationgeneticvariancecanbemaintainedwiththepropagulemodel.Theyconcludethatbyusingpropagulepoolsastheassumptionaboutcolonization,onecangreatlyexpandthesetofparametervaluesforwhichgroupselectioncanbeeffective.Anotheraspectofthisdebatethathasreceivedagreatdealofconsiderationconcernsthemathematicaltoolsnecessaryforiden-tifyingwhenaparticularlevelofbiologicalorganizationmeetsthecriteriaforbeinganinteractor.ExamplesofsuggestedtechniqueswithinthephilosophicalcommunityincludeBandon’suseofSalmon’snotionofscreeningoffandtheworkbyWimsatt(1980,1981)andLloyd([1988]1994)ontheadditivityapproach(seeSarkar1994andGodfrey-Smith1992forcriticismsofthislastapproach,andGriesemerandWade1988,andOkasha2004afordefensesofit).Biologistshavealsosuggestedavarietyofstatisticaltechniquesforaddressingthisissue.See,forexample,theworkofArnoldandFristrup(1982),HeislerandDamuth(1987),andWade(1985),respectively.Overall,whilemanyofthesuggestedtechniqueshavehadstrengths,nooneapproachtothisaspectoftheinteractorquestionhasbeengenerallyaccepted,andindeeditremainsthesubjectofdebateinbiologicalcircles(Okasha2004b,c).DiscussionsoftheseissueswithinphilosophyhavebeenmutedoflateasaresultoftheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection49influenceofgenicpluralism,whichregardstheentireinteractordebateasamistake.Notethatthe‘‘interactorquestion’’doesnotinvolveattributingadaptationsorbenefitstotheinteractors,orindeed,toanycandidateunitofselection.Interactionataparticularlevelinvolvesonlythepresenceofatraitatthatlevelwithaspecialrelationtogenicorgenotypicexpectedsuccessthatisnotreducibletointeractionsatalowerlevel.Aclaimaboutinteractionindicatesonlythatthereisanevolutionarilysignificanteventoccurringatthelevelinquestion;itsaysnothingabouttheexistenceofadaptationsatthatlevel.Asweshallsee,themostcommonerrormadeininterpretingmanyoftheinteractor-basedapproachesisthatthepresenceofaninteractoratalevelistakentoimplythattheinteractorisalsoamanifestorofanadaptationatthatlevel.2.2TheReplicatorQuestionThefocusofdiscussionsaboutreplicatorsconcernsjustwhichorganicentitiesactuallymeetthedefinitionofreplicator.Answeringthisquestionobviouslyturnsonwhatonetakesthedefinitionofreplicatortobe.InthisconnectionHull’scontributionturnedouttobecentral.StartingfromDawkins’sview,Hull(1980,318)refinedandrestrictedthemeaningof‘‘replicator,’’whichhedefinedas‘‘anentitythatpassesonitsstructuredirectlyinreplication.’’ThetermsreplicatorandinteractorwillbeusedinHull’ssenseintherestofthisessay.Hull’sdefinitionofreplicatorcorrespondsmorecloselythanDawkins’stoalong-standingdebateingeneticsabouthowlargeorsmallafragmentofagenomeoughttocountasareplicatingunit–somethingthatiscopied,andthatcanbetreatedseparatelyinevo-lutionarytheory(seeespeciallyLewontin1970).ThisdebaterevolvescriticallyaroundtheissueoflinkagedisequilibriumandledLewontin,mostprominently,toadvocatetheusageofparametersreferringtotheentiregenomeratherthantoalleleandgenotypicfrequenciesingeneticalmodels.Thebasicpointisthatwithmuchlinkagedisequilibrium,individualgenescannotbeconsideredasreplicatorsbecausetheydonotbehaveasseparateunitsduringreproduction.Althoughthisdebateremainspertinenttothechoiceofstatespaceofgeneticalmodels,ithasbeeneclipsedbyconcernsaboutinteractorsinevolutionarygenetics.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n50elisabetha.lloydThisisnottosuggestthatthereplicatorquestionhasbeensolved.Workonthereplicatorquestionispartofarichandcontinuingresearchprogram;itissimplynolongeralargepartoftheunitsdebates.ThatthispartingofwaystookplaceislargelyduetothefactthatevolutionistsandphilosophersworkingontheunitsproblemstacitlyadoptedDawkins’ssuggestionthatthereplicator,whateveritturnedouttobe,becalledthe‘gene’(seeSection3.3).Thismoveneatlyremovesthereplicatorquestionfromconsideration.Exactlywhythismoveshouldhavemetwithnear-universalacceptanceistosomeextenthistorical.However,thefactthattheintellectualtools(largelymathematicalmodels)oftheparticipantsintheunitsdebateswerebettersuitedtodealingwithaspectsofthatdebateotherthanthereplicatorquestion,whichrequiresmainlybio-chemicalinvestigation,surelycontributedtothisoutcome.Thereisaveryimportantclassofexceptionstothisgeneralabandonmentofthereplicatorquestion.SusanOyama,PaulGriffiths,andRussellGrayhavebeenleadingthinkersinformulatingaradicalalternativetotheinteractor/replicatordichotomyknownasDevel-opmentalSystemsTheory(Oyama1985;GriffithsandGray1994,1997;Oyama,Griffiths,andGray2001).Heretheevolvingunitisunderstoodtobethedevelopingsystemasawhole,privilegingneitherthereplicatornortheinteractor.JamesGriesemer(2000)hasoriginatedaprofoundreconceptualizationoftheevolutionbyselectionprocessandhasrejectedtheroleofreplicatorasmis-conceived.Heproposesinitsplacetheroleof‘‘reproducer,’’whichfocusesonthematerialtransferenceofgeneticandothermatterfromgenerationtogeneration.Thereproducerplaysacentralrole,alongwithahierarchyofinteractors,inhismuch-awaitedbookontheevolutionaryprocess.2.3TheBeneficiaryQuestionWhobenefitsfromaprocessofevolutionbyselection?Therearetwopredominantinterpretationsofthisquestion:Whobenefitsulti-matelyinthelongterm,fromtheevolutionbyselectionprocess?Andwhogetsthebenefitofpossessingadaptationsasaresultofaselectionprocess?Takethefirstofthese,theissueoftheultimatebeneficiary.Therearetwoobviousanswerstothisquestion–twodifferentwaysofcharacterizingthelong-termsurvivorsandbeneficiariesofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection51theevolutionbyselectionprocess.Onemightsaythatthespeciesorlineages(Hull’sevolvers)aretheultimatebeneficiariesoftheevo-lutionaryprocess.Alternatively,onemightsaythatthelineagescharacterizedonthegeniclevel,thatis,thesurvivingalleles,aretherelevantlong-termbeneficiaries.Ihavenotlocatedanyauthorsholdingthefirstview,but,forDawkins,thelatterinterpretationistheprimaryfactaboutevolution.Toarriveatthisconclusion,Dawkinsaddstherequirementofagencytothenotionofbeneficiary(seeHampeandMorgan1988).ForDawkins,abeneficiary,bydefi-nition,doesnotsimplypassivelyaccruecreditinthelongterm;itmustfunctionastheinitiatorofacausalpathway.Underthisdefi-nition,thereplicatoriscausallyresponsibleforallofthevariouseffectsthatarisefurtherdownthebiochemicalorphenotypicpathway,irrespectiveofwhichentitiesmightreapthelong-termrewards.Asecondandquitedistinctversionofthebeneficiaryquestioninvolvesthenotionofadaptation.Theevolutionbyselectionpro-cessmaybesaidto‘‘benefit’’aparticularlevelofentityunderselection,throughproducingadaptationsatthatlevel(Williams1966,MaynardSmith1976,Vrba1984,Eldredge1985).Onthisapproach,thelevelofentityactivelyselected(theinteractor)bene-fitsfromevolutionbyselectionatthatlevelthroughitsacquisitionofadaptations.Itiscrucialtodistinguishthequestionconcerningthelevelatwhichadaptationsevolvefromthequestionabouttheidentityoftheultimatebeneficiariesofthatselectionprocess.Onecanthink–andDawkinsdoes–thatorganismshaveadaptationswithoutthinkingthatorganismsarethe‘‘ultimatebeneficiaries’’oftheselectionprocess.Thissenseof‘‘beneficiary’’thatconcernsadaptationswillbetreatedasaseparateissue,discussedinthenextsection.2.4TheManifestor-of-AdaptationQuestionAtwhatleveldoadaptationsoccur?Or,asSober(1984,204)putsthisquestion,‘‘Whenapopulationevolvesbynaturalselection,what,ifanything,istheentitythatdoestheadapting?’’Asmentionedpre-viously,thepresenceofadaptationsatagivenlevelissometimestakentobearequirementforsomethingtobeaunitofselection.Wright(1980),inanabsolutelycrucialobservation,distinguishedCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n52elisabetha.lloydgroupselectionfor‘‘groupadvantage’’fromgroupselectionperse.Inotherwords,heclaimedthatthecombinationoftheinteractorquestionwiththequestionofwhatentityhadadaptationshadcreatedagreatdealofconfusionintheunitsofselectiondebatesingeneral.Some,ifnotmost,ofthisconfusionisaresultofaveryimportantbutneglecteddualityinthemeaningof‘‘adaptation.’’Sometimes‘‘adaptation’’istakentosignifyanytraitatallthatisadirectresultofaselectionprocessatthatlevel.Inthisview,anytraitthatarisesdirectlyfromaselectionprocessisclaimedtobe,bydefinition,anadaptation.Sometimes,ontheotherhand,theterm‘‘adaptation’’isreservedfortraitsthatare‘‘goodfor’’theirowners,thatis,thosethatprovidea‘‘betterfit’’withtheenvironmentandthatintuitivelysatisfysomenotionof‘‘goodengineering.’’Thesetwomeaningsofadaptation,theselection-productandengineeringdefinitions,respectively,aredistinct,andinsomecases,incompatible.Williams,inhisextremelyinfluentialbookAdaptationandNaturalSelection(1966),advocatedanengineeringdefinitionofadaptation.Hebelievedthatitwaspossibletohaveevolutionarychangeresultfromdirectselectionfavoringatraitwithouthavingtoconsiderthatchangedtraitasanadaptation.Consider,forexample,hisdiscussionofWaddington’s(1956)geneticassimilationexperi-ments.Williams(1966,70–81)interpretstheresultsofWaddington’sexperimentsinwhichlatentgeneticvariabilitywasmadetoexpressitselfphenotypicallybecauseofanenvironmentalpressure(seetheluciddiscussioninSober1984,199–201).Williams(1966,75–78)considersthequestionofwhetherthebithoraxcondition(resultingfromdirectartificialselectiononthattrait)shouldbeseenasanadaptivetrait,andhisansweristhatitshouldnot.Williamsinsteadseesthebithoraxconditionas‘‘adisruption...ofdevelopment,’’afailureoftheorganismtorespond.Hence,Williamsdrivesawedgebetweenthenotionofatraitthatisadirectproductofaselectionprocessandatraitthatfitshisstrongerengineeringdefinitionofanadaptation(seeGouldandLewontin11979;Sober1984,201;cf.Dobzhansky1956).Insum,whenaskingwhetheragivenlevelofentitypossessesadaptations,itisnecessarytostatenotonlythelevelofselectioninquestionbutalsowhichnotionofadaptation–eitherselection-productorengineering–isbeingused.ThisdistinctionbetweenthetwomeaningsofadaptationalsoturnsouttobepivotalinCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection53thedebatesabouttheefficacyofhigherlevelsofselection,aswewillseeinSections3.1and3.2.2.5SummaryInthissection,fourdistinctquestionshavebeendescribedthatappearundertherubricof‘‘theunitsofselection’’problem:Whatistheinteractor?Whatisthereplicator?Whatisthebeneficiary?Andwhatentitymanifestsanyadaptationsresultingfromevolutionbyselection?Thereisaseriousambiguityinthemeaningof‘‘adapta-tion’’;whichmeaningisinplayhashaddeepconsequencesforboththegroupselectiondebatesandthespeciesselectiondebates.Commentingonthisanalysis,JohnMaynardSmith(2001,1497)wroteinEvolution,Lloyd(2001)argues,‘‘correctlyIbelieve,thatmuchoftheconfusionhasarisenbecausethesametermshavebeenusedwithdifferentmeaningsbydifferentauthors...[but]Ifearthattheconfusionsshementionswillnoteasilybeended.’’InSection3,thistaxonomyofquestionsisusedtosortoutsomeofthemostinfluentialpositionsinthreedebates:groupselection(3.1),speciesselection(3.2),andgenicselection(3.3).3.ananatomyofthedebates3.1GroupSelectionGeorgeWilliams’s(1966)famousnear-deathblowtogrouppanse-lectionismwas,oddlyenough,aboutbenefit.Hewasinterestedincasesinwhichtherewasselectionamonggroupsandthegroupsasawholebenefitedfromorganism-leveltraits(includingbehaviors)thatseemeddisadvantageoustotheorganism.(Similarly,forMaynardSmith[1964].)Williamsarguedthatthepresenceofabenefittothegroupwasnotsufficienttoestablishthepresenceofgroupselection.Hedidthisbyshowingthatagroupbenefitwasnotnecessarilyagroupadaptation.(Hence,Williamsishereusingthetermbenefittosignifythemanifestationofanadaptationatthegrouplevel.)Hisassumptionwasthatagenuinegroupselectionprocessresultsintheevolutionofagroup-leveltrait–arealadaptation–thatservesadesignpurposeforthegroup.Themereexistence,however,oftraitsthatbenefitthegroupisnotenoughCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n54elisabetha.lloydtoshowthattheyareadaptations;inordertobeanadaptation,underWilliams’s(1966)view,thetraitmustbeanengineeringadaptationthatevolvedbynaturalselection.Williamsarguedthatgroupben-efitsdonot,ingeneral,existbecausetheybenefitthegroup;thatis,theydonothavetheappropriatecausalhistory.ImplicitinWilliams’sdiscussionistheassumptionthatbeingaunitofselectionatthegrouplevelrequirestwothings:(1)havingthegroupasaninteractorand(2)havingagroup-levelengineering-typeadaptation.Thatis,Williamscombinestwodifferentquestions,theinteractorquestionandthemanifestor-of-adaptationquestion,andcallsthiscombinedsettheunitofselectionquestion.Theserequirementsfor‘‘groupselection’’makeperfectsensegiventhatWilliams’sprimetargetwasVeroWynne-Edwards,whopromotedaviewofgroupselectionthatincorporatedthissametwo-prongeddefinitionofaunitofselection.Thiscombinedrequirementofengineeringgroup-leveladaptationinadditiontotheexistenceofaninteractoratthegrouplevelisaverypopularversionofthenecessaryconditionsforbeingaunitofselectionwithinthegroupselectiondebates.DavidHull(1980,325)claimsthatthegroupselectionissuehingeson‘‘whetherentitiesmoreinclusivethanorganismsexhibitadaptations.’’JohnCassidy(1978,582)statesthattheunitofselectionisdeterminedby‘‘whoorwhatisbestunderstoodasthepossessorandbeneficiaryofthetrait.’’Similarly,Eldredge(1985,108)requiresadaptationsforanentitytocountasaunitofselection,asdoesVrba(1983,1984).MaynardSmith(1976,282)alsotiestheengineeringnotionofadaptationintotheversionoftheunitsofselectionquestionhewouldliketoconsider.Inanargumentseparatinggroupandkinselection,MaynardSmithconcludesthatgroupselectionisfavoredbysmallgroupsize,lowmigrationrates,andrapidextinctionofgroupsinfectedwithaselfishalleleandthat‘‘theultimatetestofthegroupselectionhypothesiswillbewhetherpopulationshavingthesecharacteristicstendtoshow‘self-sacrificing’or‘prudent’behaviormorecommonlythanthosewhichdonot.’’Thismeansthatthepresenceofgroupselectionortheeffectivenessofgroupselectionistobemeasuredbytheexistenceofnonadaptivebehaviorofindi-vidualorganismsalongwiththepresenceofacorrespondinggroup-leveladaptation.Therefore,MaynardSmithdoesrequireagroup-leveladaptationfromgroupstocountasunitsofselection.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection55AswithWilliams,itissignificantthatheassumestheengineeringnotionofadaptationratherthantheweakerselection-productnotion.AsMaynardSmith(1976,278)putsit,‘‘anexplanationintermsofgroupadvantageshouldalwaysbeexplicit,andalwayscallsforsomejustificationintermsofthefrequencyofgroupextinction.’’Incontrasttotheprecedingauthors,SewallWright(1929,1931)separatedtheinteractorandmanifestor-of-adaptationquestionsinhisgroupselectionmodels.Hedistinguishesbetweenwhathecalls‘‘intergroupselection,’’thatis,interdemicselectioninhisshiftingbalanceprocess,and‘‘groupselectionforgroupadvantage.’’Wright(1980,840)citesHaldane(1932)astheoriginatoroftheterm‘‘altruist’’todenoteaphenotype‘‘thatcontributestogroupadvan-tageattheexpenseofdisadvantagetoitself.’’Wright(1980,841)connectsthisdebatetoWynne-Edwards,whomhecharacterizesasassertingtheevolutionaryimportanceof‘‘groupselectionforgroupadvantage.’’HearguesthatHamilton’skinselectionmodelis‘‘verydifferent’’from‘‘groupselectionfortheuniformadvantageofagroup.’’Hamiltonhimselfconcurredinalittle-knownpaperfrom1975(Hamilton1996,vol.1,337).Wright(1980,841)takesMaynardSmith,Williams,andDawkinstotaskformistakenlythinkingthatbecausetheyhavesuccessfullycriticizedgroupselectionforgroupadvantage,theycanconcludethat‘‘naturalselectionispracticallywhollygenic.’’Wright(1980,841)arguesthat‘‘noneofthemdiscussedgroupselectionfororgan-ismicadvantagetoindividuals,thedynamicfactorintheshiftingbalanceprocess,althoughthisprocess,basedonirreversiblelocalpeak-shiftsisnotfragileatall,incontrastwiththefairlyobviousfragilityofgroupselectionforgroupadvantage,whichtheycon-sideredworthyofextensivediscussionbeforerejection.’’ThisisafaircriticismofMaynardSmith,Williams,andDawkins.AccordingtoWright,theproblemisthattheseauthorsfailedtodistinguishbetweentwoquestions:theinteractorquestionandthemanifestor-of-adaptationquestion.Wright’sinterdemicgroupselec-tionmodelinvolvesgroupsonlyasinteractors,notasmanifestorsofgroup-leveladaptations.Further,heisinterestedonlyintheeffectthegroupshaveonorganismicadaptednessandexpectedreproduc-tivesuccess.Morerecently,modelersfollowingSewallWright’sinterestinstructuredpopulationshavecreatedanewsetofgeneticalmodelsthatarealsocalled‘‘groupselection’’modelsandinwhichCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n56elisabetha.lloydthequestionsofgroupadaptationsandgroupbenefitplaylittleornorole.Foraperiodspanningtwodecades,however,MaynardSmith,Williams,andDawkinsdidnotacknowledgethatthepositiontheyattacked,namely,Wynne-Edwards’s,issignificantlydifferentfromotheravailableapproachestogroupselection,suchasthatofWright,Wade,Wilson,Uyenoyama,Feldman,andLewontin.Ultimately,however,bothWilliamsandMaynardSmithrecognizedthesig-nificanceofthedistinctionbetweentheinteractorquestionandthemanifestor-of-an-adaptationquestion.Asaresult,Williams(1985,7–8)wrote,‘‘Ifsomepopulationsofspeciesaredoingbetterthanothersatpersistenceandreproduction,andifsuchdifferencesarecausedinpartbygeneticdifferences,thisselectionatthepopulationlevelmustplayaroleintheevolutionofthespecies,’’whilecon-cludingthatgroupselection‘‘isunimportantfortheoriginandmaintenanceofadaptation.’’Shortlythereafter,MaynardSmith(1987,123)madeanextra-ordinaryconcession.Therehasbeensomesemanticconfusionaboutthephrase‘‘groupselec-tion,’’forwhichImaybepartlyresponsible.Forme,thedebateaboutlevelsofselectionwasinitiatedbyWynne-Edwards’book.Hearguedthattherearegroup-leveladaptations...whichinformindividualsofthesizeofthepopulationsothattheycanadjusttheirbreedingforthegoodofthepopu-lation.Hewasclearthatsuchadaptationscouldevolveonlyifpopulationswereunitsofselection....Perhapsunfortunately,hereferredtotheprocessas‘‘groupselection.’’Asaconsequence,formeandformanyotherswhoengagedinthisdebate,thephrasecametoimplythatgroupsweresuffi-cientlyisolatedfromoneanotherreproductivelytoactasunitsofevolution,andnotmerelythatselectionactedongroups.Theimportanceofthisdebatelayinthefactthatgroup-adaptationistthinkingwasatthattimewidespreadamongbiologists.Itwasthereforeimportanttoestablishthatthereisnoreasontoexpectgroupstoevolvetraitsensuringtheirownsurvivalunlesstheyaresufficientlyisolatedforliketobegetlike....WhenWilson(1975)introducedhistrait-groupmodel,Iwasforalongtimebewilderedbyhiswishtotreatitasacaseofgroupselectionanddoublysobythefactthathisoriginalmodel...hadinter-estingresultsonlywhenthemembersofthegroupweregeneticallyrelated,aprocessIhadbeencallingkinselectionfortenyears.Ithinkthatthesesemanticdifficultiesarenowlargelyover.(1987,123)CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection57Dawkins(1989a)alsoseemstohaverediscoveredtheevolu-tionaryefficacyofhigher-levelselectionprocessesinanarticleonartificiallife.Inthisarticle,heisprimarilyconcernedwithmodelingthecourseofselectionprocesses,andheoffersaspecies-levelselectioninterpretationforanaggregatespecies-leveltrait.Still,heseemsnottohaverecognizedtheconnectionbetweenthisevolu-tionarydynamicandthecontroversiessurroundinggroupselectionbecauseinhissecondeditionofTheSelfishGene(Dawkins1989b)hehadyettoacceptthedistinctionmadesoclearlybyWrightin1980.Thiswasinspiteofthefactthatby1987,theimportanceofdistinguishingbetweenevolutionbyselectionprocessesandanyengineeringadaptationsproducedbytheseprocesseshadbeenacknowledgedbytheworkersDawkinsclaimedtobefollowingmostclosely,WilliamsandMaynardSmith.ThemostrecentsignificantentryintothesedebatesisElliottSoberandDavidSloanWilson’sUntoOthers,whichtheypublishedin1998.InthisworkSoberandWilsondevelopacaseforgroupselectionbasedontheneedtoaccountfortheexistenceofbiologicalaltruism.Biologicalaltruismisanybehaviorthatbenefitsanotherorganismatsomecosttotheactor.Suchbehaviormustalwaysreducetheactor’sfitness,butitmay,asSoberandWilson(followingtheworkofHaldaneandWright)show,increasethefitnessofcer-taingroupswithinastructuredpopulation.WhilethebiologicalmodelinginUntoOtherswasnotnew,thebookdidcalltheissuesinvolvedinthegroupselectiondebatestotheattentionofthelargerphilosophiccommunity.3.2SpeciesSelectionAmbiguitiesaboutthedefinitionofaunitofselectionhavealsosnarledthedebateaboutselectionprocessesatthespecieslevel.Thecombiningoftheinteractorquestionandthemanifestor-of-adapta-tionquestion(intheengineeringsense)ledtotherejectionofresearchaimedatconsideringtheroleofspeciesasinteractors,simpliciter,inevolution.Onceitisunderstoodthatspecies-levelinteractorsmayormaynotpossessdesign-typeadaptations,itbecomespossibletodistinguishtworesearchquestions:Dospeciesfunctionasinteractors,playinganactiveandsignificantroleinevolutionbyselection?Anddoestheevolutionofspecies-levelCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n58elisabetha.lloydinteractorsproducespecies-levelengineeringadaptationsand,ifso,howoften?Formostofthehistoryofthespeciesselectiondebate,thesequestionshavebeenlumpedtogether;askingwhetherspeciescouldbeunitsofselectionmeantaskingwhethertheyfulfilledboththeinteractorandmanifestor-of-adaptationroles.Forexample,Vrba(1984)usedMaynardSmith’streatmentoftheevolutionofaltruismasatouchstoneinherdefinitionofspeciesselection.MaynardSmith(1976)arguedthatkinselectioncouldcausethespreadofaltruisticgenesbutthatitshouldnotbecalledgroupselection.Again,thiswasbecausethegroupswerenotconsideredtopossessdesign-typeadaptationsthemselves.Vrba(1984,319)agreedthatthespreadofaltruismshouldnotbeconsideredacaseofgroupselectionbecause‘‘thereisnogroupadaptationinvolved;altruismisnotemergentatthegrouplevel’’(MaynardSmithgivesdifferentreasonsforhisrejection).Thisamountstoassumingthattheremustbegroupbenefitinthesenseofadesign-typegroup-leveladaptationinordertosaythatgroupselectioncanoccur.Vrba’s(1983,388)viewwasthatevolutionbyselectionisnothappeningatagivenlevelunlessthereisabenefitorengineeringadaptationatthatlevel.Sheexplicitlyequatesunitsofselectionwiththeexistenceofaninter-actorplusadaptationatthatlevel.Furthermore,itseemsthatshehasadoptedthestrongerengineeringdefinitionofadaptation.Eldredge(1985,134,196)alsoarguesthatspeciesselectiondoesnothappenunlesstherearespecies-leveladaptations.Eldredge(1985,133)rejectscertaincasesashigher-levelselectionprocessesoverallbecause‘‘frequenciesofthepropertiesoflower-levelindivi-dualswhicharepartofahigh-levelindividualsimplydonotmakeconvincinghigher-leveladaptations.’’Vrba,Eldredge,andGouldalldefinedaunitofselectionasrequiringanemergent,adaptiveprop-erty.Thisamountstoaskingacombinationoftheinteractorandmanifestor-of-adaptationquestions.Butconsiderthelineagewidetraitofvariability.Treatingspeciesasinteractorshasalongtradition(Thoday1953,Dobzhansky1956,Lewontin1958).Ifspeciesareconceivedasinteractors(andnotnecessarilymanifestors-of-adaptations),thenthenotionofspeciesselectionisnotvulnerabletoWilliams’soriginalanti-group-selectionobjections.AsWilliams(1992,27)remarks,‘‘theanswertothesedifficultiesmustbefoundinLloyd’sideathathigherlevelsofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection59selectiondepend,notonemergentcharacters,butonanyandallemergentfitnesses.’’Theoldideawasthatlineageswithcertainpropertiesofbeingabletorespondtoenvironmentalstresseswouldbeselectedfor,thatthetraitofvariabilityitselfwouldbeselectedfor,andthatitwouldspreadinthepopulationofpopulations.Inotherwords,lineagesweretreatedasinteractors.Theearlierresearchersspokelooselyofadaptationswherebyadaptationsweretreatedintheweaksenseasequivalentsimplytotheoutcomeofselectionprocesses(atanylevel).Theywereexplicitlynotconcernedwiththeeffectofspeciesselectiononorganismicleveltraitsbutwiththeeffectonspecies-levelcharacterssuchasspecia-tionrates,lineage-levelsurvival,andextinctionratesofspecies.LloydandGould(1993)arguethatthissortofcaserepresentsaperfectlygoodformofspeciesselectioneventhoughsomebalkatthethoughtthatvariabilitywouldthenbeconsidered,underaweakdefinition,aspecies-leveladaptation(cf.Lloyd[1988]1994).Vrba(1989)alsoeventuallyrecognizedtheadvantagesofkeepingtheinteractorquestionseparatefromarequirementforanengi-neering-typeadaptation.Inhermorerecentreviewarticle,shehasdroppedherformerrequirementthatinorderforspeciestobeunitsofselection,theymustpossessspecies-leveladaptations.Ulti-mately,hercurrentdefinitionofspeciesselectionisinconformitywithasimpleinteractorinterpretationofaunitofselection(cf.DamuthandHeisler1988;Lloyd[1988]1994).Itiseasytoseehowthetwo-prongeddefinitionofaunitofselection–asinteractorandmanifestor-of-adaptation–heldswayforsolonginthespeciesselectiondebates.Afterall,itdominatedmuchofthegroupselectiondebatesuntiljustrecently.Someofthecon-fusionandconflictoverhigher-levelunitsofselectionarosebecauseofahistoricalcontingency–Wynne-Edwards’simplicitdefinitionofaunitofselectionandtheresponsesitprovoked.3.3GenicSelection:TheOriginatorsOnemayunderstandablythinkthatDawkinsisinterestedinthereplicatorquestionbecauseheclaimsthattheunitofselectionoughttobethereplicator.Thiswouldbeamistake.Dawkinsisinterestedprimarilyinaspecificontologicalissueaboutbenefit.Heisaskingaspecialversionofthebeneficiaryquestion,andhisCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n60elisabetha.lloydanswertothatquestiondictateshisanswerstotheotherthreequestionsflyingundertherubricofthe‘‘unitsofselection.’’Briefly,Dawkinsarguesthatbecausereplicatorsaretheonlyentitiesthat‘‘survive’’theevolutionaryprocess,theymustbethebeneficiaries.Whathappensintheprocessofevolutionbynaturalselectionhappensfortheirsake,fortheirbenefit.Hence,interactorsinteractforthereplicators’benefit,andadaptationsbelongtothereplicators.Replicatorsaretheonlyentitieswithrealagencyasinitiatorsofcausalchainsthatleadtothephenotypes;hence,theyaccruethecreditandaretherealunitsofselection.Dawkins’sversionoftheunitsofselectionquestionamountstoacombinationofthebeneficiaryquestionplusthemanifestor-of-adaptationquestion.Thereislittleevidencethathethinksheisansweringthepredominantinteractorquestion;rather,hearguesthatpeoplewhofocusoninteractorsarelaboringunderamis-understandingofevolutionarytheory.Onereasonhethinksthismightbethathetakesashisopponentsthosewhoholdacombi-nationoftheinteractorplusmanifestor-of-adaptationsdefinitionofaunitofselection(e.g.,Wynne-Edwards).Unfortunately,Dawkinsignoresthosewhoarepursuingtheinteractorquestionalone;theseresearchersarenotvulnerabletothecriticismsheposesagainstthecombinedinteractor-adaptationview.Dawkins(1982b,113–16)believesthatinteractors,whichhecalls‘‘vehicles,’’arenotrelevanttotheunitsofselectionproblem.Therealunitsofselection,heargues,shouldbereplicators,‘‘theunitsthatactuallysurviveorfailtosurvive.’’Organismsorgroupsas‘‘vehicles’’maybeseenastheunitoffunctionintheselectionprocess,buttheyshouldnot,heargues,beseenastheunitsofselectionbecausethecharacteristicstheyacquirearenotpassedon(Dawkins1982b,99).Here,DawkinsisfollowingWilliams’s(1966,109)line.Genotypeshavelimitedlivesandfailtoreproducethem-selvesbecausetheyaredestroyedineverygenerationbymeiosisandrecombinationinsexuallyreproducingspecies;theyareonlytem-porary.Hence,genesaretheonlyunitsthatsurviveintheselectionprocess.Thegene(replicator)istherealunitbecauseitisan‘‘indi-visiblefragment’’;itis‘‘potentiallyimmortal’’(Williams1966,23–24;Dawkins1982b,97).Theissue,forDawkins(1982b,82),iswhether,‘‘whenwetalkaboutaunitofselection,weoughttomeanavehicleatall,oraCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection61replicator.’’Heclearlydistinguishesthedisputehewouldliketogeneratefromthegroup-versus-organismicselectioncontroversy,whichhecharacterizesasadisagreement‘‘abouttherivalclaimsoftwosuggestedkindsofvehicles.’’Inhisview,replicatorselectionshouldbeseenasanalternativeframeworkforbothorganismicandgroupselectionmodels.TherearetwomistakesthatDawkinsisnotmaking.First,hedoesnotdenythatinteractorsareinvolvedintheevolutionaryprocess.Heemphasizesthatitisnotnecessary,underhisview,tobelievethatreplicatorsaredirectly‘‘visible’’toselectionforces.Dawkins(1982b,176)hasrecognizedfromthebeginningthathisquestioniscompletelydistinctfromtheinteractorquestion.Heremarks,infact,thatthedebateaboutgroupversusorganismicselectionis‘‘afactualdisputeaboutthelevelatwhichselectionismosteffectiveinnature,’’whereashisownpointis‘‘aboutwhatweoughttomeanwhenwetalkaboutaunitofselection.’’Dawkins(1982a,46–47)alsostatesthatgenesorotherreplicatorsdonot‘‘literallyfacethecuttingedgeofnaturalselection.Itistheirphenotypiceffectsthataretheproximalsubjectsofselection.’’Second,Dawkinsdoesnotspecifyhowlargeachunkofthege-nomehewillallowasareplicator;thereisnocommitmenttothenotionthatsinglegenesaretheonlypossiblereplicators.HearguesthatifLewontin,Franklin,Slatkin,andothersareright,hisviewwillnotbeaffected(seeSection2.2).Iflinkagedisequilibriumisverystrong,thenthe‘‘effectivereplicatorwillbeaverylargechunkofDNA’’(Dawkins1982b,89).WecanconcludefromthisthatDawkinsisnotinterestedinthereplicatorquestionatall;hisclaimhereisthathisframeworkcanaccommodateanyofitspossibleanswers.Onwhatbasis,then,doesDawkinsrejectthequestionaboutinteractors?Ithinktheanswerliesintheparticularquestioninwhichheismostinterested,namely,whatis‘‘thenatureoftheentityforwhosebenefitadaptationsmaybesaidtoexist?’’Onthefaceofit,itiscertainlyconceivablethatonemightidentifythebeneficiaryoftheadaptationsas–insomecases,anyway–theindividualorganismorgroupthatexhibitsthephenotypictraittakentobetheadaptation.Infact,Williams(1966)seemstohavedonejustthatinhisdiscussionofgroupselection.ButDawkins1982a,60)rejectsthismove,introducinganadditionalqualificationCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n62elisabetha.lloydtobefulfilledbyaunitofselection;itmustbe‘‘theunitthatactuallysurvivesorfailstosurvive.’’Becauseorganisms,groups,andevengenomesaredestroyedduringselectionandreproduction,theanswertothesurvivalquestionmustbethereplicator.Strictlyspeaking,thisisfalse;itiscopiesofthereplicatorsthatsurvive.Hethereforemustmeanreplicatorsinsomesenseofinformationandnotasbiologicalentities(seeHampeandMorgan1988).Butthereisstillaproblem.AlthoughDawkins(1982a,60)con-cludes,‘‘thereshouldbenocontroversyoverreplicatorsversusvehicles.Replicatorsurvivalandvehicleselectionaretwoaspectsofthesameprocess,’’hedoesnotjustleavethevehicleselectiondebatealone.Instead,hearguesthatwedonotneedtheconceptofdiscretevehiclesatall.Theimportantpointisthat,onDawkins’sanalysis,thefactthatreplicatorsaretheonlysurvivorsoftheevolution-by-selectionprocessautomaticallyanswersalsotheques-tionofwhoownstheadaptations.Heclaimsthatadaptationsmustbeseenasbeingdesignedforthegoodoftheactive-gene-linereplicatorforthesimplereasonthatreplicatorsaretheonlyentitiesaroundlongenoughtoenjoythemoverthecourseofnaturalselection.Dawkins(1982b,114)acknowledgesthatthephenotypeis‘‘theallimportantinstrumentofreplicatorpreservation,’’andthatgenes’phenotypiceffectsareorganizedintoorganisms(thattherebymightbenefitfromthemintheirlifetimes).Butbecauseonlytheactivegerm-linereplicatorssurvive,theyarethetruelocusofadaptations(Dawkins1982b,113;emphasisadded).Theotherthingsthatbenefitovertheshortterm(e.g.,organismswithadaptivetraits)aremerelythetoolsoftherealsurvivors,therealowners.Hence,Dawkinsrejectsthevehicleapproachpartlybecauseheidentifiesitwiththemanifestor-of-adaptationapproach,whichhehasansweredbydefinition,intermsofthelong-termbeneficiary.ThesecondkeyaspectofDawkins’sviewsoninteractorsisthatheseemstowanttodoawaywiththementirely.Dawkins(1982b,116)isawarethatthevehicleconceptis‘‘fundamentaltothepre-dominantorthodoxapproachtonaturalselection.’’Nevertheless,herejectsthisapproachinTheExtendedPhenotype,claimingthatthe‘‘mainpurposeofthisbookistodrawattentiontotheweak-nessesofthewholevehicleconcept’’(Dawkins1982b,115).Buthis‘‘vehicle’’approachisnotequivalentto‘‘theinteractorquestion’’;itencompassesamuchmorerestrictedapproach.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection63Inparticular,whenDawkins(1982b,5,55)arguesagainst‘‘thevehicleconcept,’’heisonlyarguingagainstthedesirabilityofseeingtheindividualorganismastheoneandonlypossiblevehicle.Histargetisexplicitlythosewhoholdwhathecallsthe‘‘CentralTheorem,’’whichsaysthatindividualorganismsshouldbeseenasmaximizingtheirowninclusivefitness.Dawkins’sargumentsareindeeddamagingtotheCentralTheorem,buttheyareinef-fectiveagainstotherapproachesthatdefineunitsofselectionasinteractors.OnewaytointerprettheCentralTheoremisthatitimpliesthattheindividualorganismisalwaysthebeneficiaryofanyselectionprocess.Dawkinsseemstomeanby‘‘beneficiary’’boththemani-festor-of-adaptationandthatwhichsurvivestoreaptherewardsoftheevolutionaryprocess.Heargues,rightlyandpersuasively,Ithink,thatitdoesnotmakesensealwaystoconsidertheindividualorganismtobethebeneficiaryofaselectionprocess.ButitiscrucialtoseethatDawkins(1982b,189)isnotarguingagainsttheimportanceoftheinteractorquestioningeneral,butratheragainstaparticulardefinitionofaunitofselection.Theviewheiscriticizingassumesthattheindividualorganismistheinter-actor,andthebeneficiary,andthemanifestor-of-adaptation.Con-siderhismainargumentagainsttheutilityofconsideringvehicles:theprimaryreasontoabandonthinkingaboutvehiclesisthatitconfusespeople.Butlookathisexamples;theirpointisthatitisinappropriatealwaystoaskhowanorganism’sbehaviorbenefitsthatorganism’sinclusivefitness.Weshouldaskinstead,saysDawkins(1982b,80),‘‘whoseinclusivefitnessthebehaviorisben-efiting.’’Hestatesthathispurposeinthisbookistoshowthat‘‘theoreticaldangersattendtheassumptionthatadaptationsareforthegoodof...theindividualorganism’’(Dawkins1982b,91).So,Dawkinsisquiteclearaboutwhathemeansbythe‘‘vehicleselectionapproach’’;italwaysassumesthattheorganismisthebeneficiaryofitsaccruedinclusivefitness.Dawkinsadvancespowerfulargumentsagainsttheassumptionthattheorganismisalwaystheinteractorcumbeneficiarycummanifestor-of-adapta-tions.Thisapproachisclearlynotequivalenttotheapproachtounitsofselectioncharacterizedastheinteractorapproach.Unfor-tunately,Dawkinsextendshisconclusionstotheseotherapproaches,whichhehas,infact,notaddressed.Dawkins’slackofconsiderationCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n64elisabetha.lloydoftheinteractordefinitionofaunitofselectionleadstotwograveproblemswithhisviews.Oneproblemisthathehasatendencytointerpretallgroupselectionistclaimsasbeingaboutbeneficiariesandmanifestors-of-adaptationsaswellasinteractors.Thisisaseriousmisreadingofauthorswhoarepursuingtheinteractorquestionalone.Consider,forexample,Dawkins’s(1982b,85)argumentthatgroupsshouldnotbeconsideredunitsofselection:Totheextentthatactivegerm-linereplicatorsbenefitfromthesurvivalofthegroupofindividualsinwhichtheysit,overandabovethe[effectsofindividualtraitsandaltruism],wemayexpecttoseeadaptationsforthepreservationofthegroup.Butalltheseadaptationswillexist,fundamen-tally,throughdifferentialreplicatorsurvival.Thebasicbeneficiaryofanyadaptationistheactivegerm-linereplicator.NoticethatDawkinsbeginsbyadmittingthatgroupscanfunctionasinteractors,andeventhatgroupselectionmayeffectivelyproducegroup-leveladaptations.Theargumentthatgroupsshouldnotbeconsideredrealunitsofselectionamountstotheclaimthatthegroupsarenottheultimatebeneficiaries.Tocounteracttheintuitionthatthegroupsdo,ofcourse,benefit,insomesense,fromtheadap-tations,Dawkinsusestheterms‘‘fundamentally’’and‘‘basic,’’thussignalingwhatheconsidersthemostimportantlevel.Evenifagroup-leveltraitisaffectingachangeingenefrequencies,‘‘itisstillgenesthatareregardedasthereplicatorswhichactuallysurvive(orfailtosurvive)asaconsequenceofthe(vehicle)selectionprocess’’(Dawkins1982b,115).Thus,thereplicatoristheunitofselection,becauseitisthebeneficiary,andtherealownerofalladaptationsthatexist.Sayingallthisdoesnot,however,addressthefactthatotherresearchersinvestigatinggroupselectionareaskingtheinteractorquestionandsometimesalsothemanifestor-of-adaptationquestion,ratherthanDawkins’sspecialversionofthe(ultimate)beneficiaryquestion.Hegivesnoadditionalreasontorejecttheseotherques-tionsaslegitimate;hesimplyreassertsthesuperiorityofhisownpreferredunitofselection.Insum,Dawkinshasidentifiedthreecriteriaasnecessaryforsomethingtobeaunitofselection:itmustbeareplicator;itmustbethemostbasicbeneficiaryoftheselectionprocess;anditisautomaticallytheultimatemanifestor-of-adapta-tionthroughbeingthebeneficiary.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nUnitsandLevelsofSelection65note1.NotethatWilliamssaysthat‘‘naturalselectionwouldproduceormaintainadaptationasamatterofdefinition’’(1966,25;cf.Mayr1976).ThiscommentconflictswiththeconclusionsWilliamsdrawsinhisdiscussionofWaddington;however,Williamslaterretractshisbithoraxanalysis(1985).Williamsiscommittedtoanengineeringdefinitionofadaptation(personalcommunication1989).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nrobertn.brandonandgrantramsey4WhatsWrongwiththeEmergentistStatisticalInterpretationofNaturalSelectionandRandomDrift?Population-leveltheoriesofevolutionthestockandtradeofpopulationgeneticsarestatisticaltheoriesparexcellence.Butwhataccountsforthestatisticalcharacterofpopulation-levelphenomena?Oneviewisthatthepopulation-levelstatisticsareaproductof,aregeneratedby,probabilitiesthatattachtotheindivi-dualsinthepopulation.Onthisconception,population-levelphe-nomenaareexplainedbyindividual-levelprobabilitiesandtheirpopulation-levelcombinations.Anotherview,whicharguablygoes1backtoFisher(1930)buthasbeendefendedrecently,isthatthepopulation-levelstatisticsaresuigeneris,thattheysomehowemergefromtheunderlyingdeterministicbehavioroftheindivi-dualscomposingthepopulation.Walsh,Lewens,andAriew(2002)labelthisthestatisticalinterpretation.Wearenotwillingtogivethemthatterm,sinceeveryonewilladmitthatthepopulation-leveltheoriesofevolutionarestatistical,sowewillcallthistheemer-gentiststatisticalinterpretation(ESI).Ourgoalsaretoshowthat(1)thisinterpretationisbasedongrossfactualerrorsconcerningthepracticeofevolutionarybiology,concerningbothwhatisdoneandwhatcanbedone;(2)itsadoptionwouldentailgivinguponmostoftheexplanatoryandpredictive(i.e.,scientific)projectsofevolu-tionarybiology;andfinally(3)arivalinterpretation,whichwewilllabelthepropensitystatisticalinterpretation(PSI),succeedsexactlywheretheemergentistinterpretationfails.66CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift671.propensityandemergentistinterpretationsThepropensityinterpretationoffitnesswasintroducedintothephilosophicalliteraturein1978(seeBrandon1978;alsoseeMillsandBeatty1979).Theprimemotivationwastomakeroomforanexplanatorytheoryofnaturalselection,whichistantamounttosolvingtheso-calledtautologyproblem.Thisproblemarisesfromacasualinspectionofthephrasesurvivalofthefittest,followedbythequestionofwhatdefinesthefittest.Iftheansweristhosethatreproducethemost,thenitseemsweareexplainingaphenomenon,differentialreproduction,intermsofitself,whichisnoexplanationatall.Brandonsapproachwastothinkoffitness(oradaptedness)asadisposition.Justasitisnotexplanatorilyemptytocitethewatersolubilityofsaltinexplainingthebehaviorofaparticularsampleofsaltwhenplacedinwater,sotooitisnotexplanatorilyemptytocitedifferencesinadaptednesstoacommonenvironmentwhenexplainingaparticularcaseofdifferentialreproduction.Ofcourse,inthecaseofwatersolubilitywewant,andindeedhave,adeeperexplanationofthatdispositionageneralexplanationgivenintermsofmolecularbonding.Thecaseoffitnessdiffersintwoimportantwaysfromthatofwatersolubility.First,ceterisparibus,water-solublesubstancesdissolvewhenplacedinwater,period.Thatis,althoughwequalifytheclaimwithaceterisparibusclausewewanttoexcludecasessuchasthatwhenthewaterisfrozen,oralreadysaturated,andsoontheclaimitselfisnotprobabilistic.Butwethinkchancecaninterveneinrealbiologicalpopulationssothathigherfitnessandhigherlevelsofreproductivesuccesscanbedissociated.Onthepropensityinterpretation,fitness(oradaptedness)isanexplicitly2probabilisticconcept.Thusitisaprobabilisticpropensity.Second,unlikeinthecaseofwatersolubility,thereisnogeneralunderlyingexplanationofdifferentialfitness(seeBrandon1978or1990,1325;alsoseeRosenberg1978,1985;andSober1984).Theunderlyingcausalbasisoffitnessdifferencescanbeuncoveredbydetailedstudyofparticularpopulationsinparticularselectiveenvironments,butitwillnotbegeneral.ItisourimpressionthatthisinterpretationoffitnessiswidelyacceptedinboththephilosophicalandbiologicalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n68robertn.brandonandgrantramseycommunities.Butsomeofitsbroaderimplicationsareprobablynotappreciated.Ifthepropensityinterpretationiscorrect,thenpopulation-levelprobabilitiesarederivablefromindividual-levelprobabilitiesinafamiliarway.Forexample,ifacoinandtossingdeviceyieldsaprobabilityofheadsof.5,thenwecancalculatetheprobabilityofvariousresultsinanensembleoftosses,sayfour,bythelawsofprobabilitytheory.Forthoseofyouwhohavetroublethinkingofcointossingasgenuinelystochastic,substitutethefollowingexample.Oxygen-15hasahalf-lifeoftwominutes.Takefouratomsofthatisotope.Whatistheprobabilitythatexactlytwoofthemwilldecayduringatwo-minutetimeinterval?Toanswerthatquestionwedotheexactsamecalculationasinthecoinexampleandgetthesameanswer.Theprobabilityofthatoutcomeis.375.Sothepro-pensityinterpretationoffitnessyieldsafamiliarandnaturalwayofunderstandingthepopulation-levelprobabilitiesthatareessentialtoevolutionarytheory.Theonlystickingpointhereisthat,takenliterally,thepropensityinterpretationoffitnessiscommittedtothefundamentalindeterminacyofthelives,deaths,andulti-matelyreproductivesuccessesofindividualorganisms(seeBrandonandCarson1996).Somepeoplefindthatadifficultontologicalcommitment.Soperhapsonemotivationfortheemergentiststatisticalinter-pretationisthatitisnotcommittedtotheindeterminacyofindi-viduallivesanddeaths.IndeeditseemsthattheinspirationforthisinterpretationistherelationbetweenstatisticalthermodynamicsandNewtonianmechanics.Here,supposedly,theunderlyingmechanicsofthemoleculesinagasaredeterministic,butatthemacrolevelwegettheexplicitlyprobabilisticsecondlawofther-3modynamics.Perhapsthenthereisananalogueoftheemergentiststatisticalinterpretationinphysics.Butwedonotthinkoneshouldbemuchimpressedbythat,sincenooneunderstandstherelation-shipbetweenmechanicsandstatisticalthermodynamics(Sklar1999).Lookingtophysicswillnothelpusunderstandtheemergenceofpopulation-levelprobabilities.Anotherpossiblemotivationfortheemergentistinterpretationisthatwearestuckwithit.Thatis,wecanclearlyseethattherearepopulation-levelprobabilitiesgoverningtheevolutionarytrajec-toriesofpopulations,butwehavenoaccesstotheindividual-levelCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift69probabilitiespostulatedbythepropensityinterpretationeitherbecausetheydonotexist,orbecausewehaveepistemiclimitations.Wewilldealwiththispossibilityinthenextsection.Finally,onemightfindtheemergentistinterpretationattractiveprimarilybecauseonethinksthatevolutionarytheorydealssolelywithpopulation-levelprobabilities,andthereforehasnoneedfor4individual-levelprobabilities.Soontologicalparsimonysuggests5wedowithoutthem.ThispointwillbedealtwithinSection3.2.theargumentfortheemergentiststatisticalinterpretationBasingfitnessontype-orpopulation-leveleffectshasprecedent,especiallyamongbiologists.Fisher(1930)tookthefitnessofatypetobetheobjectiverepresentationofthattypeinthenextgeneration.Similarly,astandardevolutionarybiologytextbook(Futuyma1986)definesfitnessastheaveragecontributionofonealleleorgenotypetothenextgenerationorsucceedinggenerations,comparedwiththatofotherallelesorgenotypes.Despitedrawingsupportfromanumberofbiologists,untilrecentlythispositionhasreceivedlittlesupportfromphilosophers.InourdiscussionoftheESI,wewillfocusonrecentdefensesofthisviewbyWalshetal.(2002)andMatthenandAriew(2002).Walshetal.askwhetherevolutionarytheoryisastatisticaltheoryoradynamicaltheory.Astatisticaltheoryisphenomenological,notcausal,andadynamicaltheoryisatheoryofforces,a`laNewtonianmechanics.Thustheirquestionismoreorlessequivalenttothis:isevolutionarytheorylikethekinetictheoryofgasesorNewtonianmechanics?Thisseemsanimpoverishedrangeofoptions.Whyshoulditberelevantlysimilartoeitherone?Arethosetheonlytwotypesofscientifictheories?Althoughtheypresentuswithafalsechoicethelogicoftheirargumentisclear:Sobers(1984)descriptionofevolutionarytheoryasatheoryofforcesis,theyclaim,wrong;6thereforetheemergentiststatisticalinterpretationiscorrect.Sobersdescriptionofevolutionarytheoryasatheoryofforceshasflaws,somemoreseriousthanothers.AsEndler(1986)haspointedout,thereareanumberofdisanalogiesbetweennaturalselection7andtheconceptofforceinphysics.ButthisisaquibblecomparedtothemostimportantproblemwithSobersanalogy,whichistheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n70robertn.brandonandgrantramseyfactthatselectionanddriftarenotopposingforces,butrathertwocopossibleoutcomesofthesameprocesstheprocessofsamplingfromapopulationwheretheprobabilitiesofbeingsampledforeachmemberofthepopulationdonotallequal1or0(seeBrandon2005).ThatisaseriousflawinSobersaccount,butitdoesnotmeanwehavetosettleforapurelyphenomenologicalaccount.WhydoWalshetal.thinkthatitdoes?Giventheimpoverishedrangeofoptionstheypresent,logicseemstoforcethischoiceonthem.Butwethinkthereismoretoitthanthat.InarelatedarticleMatthenandAriew(2002)presentanotherargumentforwhatamountstothesameconclusion.AgainSoberscomparisonofevolutionarytheorytoNewtonianmechanicsisthetarget.MatthenandAriew(2002,67)arguethatdifferentcompo-nentsoffitnessarenotcomparable,andinparticularthatthereisnothinglikevectoradditionthatwouldallowustocombinedif-ferentforcesofselection.Soalthoughweknowthat,everythingelsebeingequal,itisbesttoproducetheminoritysexinapopula-tionwithaskewedsexratio:wehavenowayofcalculatingwhetheragivensex-selectionstrategyinteractswithagivenparental-care-strategy,andhowthefitnessproducedbyvariantsofthesestrategiescombine.Thisinabilitytoaddtheforcesoffitnessisevenmorepronouncedwhenthesourcelawsareinunrelateddomains.Supposeacertainspeciesundertakesparentalcare,isresistanttomalaria,andissomewhatweakbutveryquick.Howdothesefitnessfactorsaddup?Wehavenoideaatall.Thetheoryofprobabilityhasnogeneralwaytodealwithsuchquestions.(2002,67)(Thislastsentenceofthisquoteisquiteodd.Whyshouldthetheoryofprobabilitytellushowdifferentcomponentsoffitnessinteractinbiology?Shouldthatnotbeamatterofbiology?)Theconclusionofthisisthefollowing:Thedisanalogyisthat,whileforceaffordsNewtonianmechanicsthemeanstocompareandadduptheconsequencesofthesediversecauses,fitnessdoesnotadduporresolve.Thisiswhypopulationgeneticistsareforcedtoestimatefitnessbymeasuringpopulationchange.(2002,68,emphasisadded)Thelogicofthisargumentis,wethink,clearenough.Itscon-clusionisfalse,andwewanttofocusonthat.Butletusbrieflyexaminethemajorpremise.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift71First,populationgeneticmodelsregularlydocombinedifferentfactorsofevolutionarychangeinstraightforwardways.Onecould,forinstance,writedownasimplemodelthattrackstheevolutionoftwoalleles,Aanda,inahaploidpopulationwithdiscretegenera-tions.HerethefrequencyofAingeneration2issimplytheproductofthefitnessofA,wA,anditsfrequencyingeneration1,p1.Thus(wherep2isthefrequencyofAingeneration2):p2¼wAp1andsimilarlyforchangeina(whereqiisthefrequencyofaingenerationi):q2¼waq1:Giventhisverysimplemodelwecaneasilyaddtheeffectsofmigrationandmutation(wherelisthemutationratefromAtoa,tisthemutationratefromatoA,andm1AistherateoflossofAduetoemigration,m2AthegaininAduetoimmigration,m1atherateoflossofaduetoemigration,andm2atherateofgainofaduetoimmigration):p2¼wAp1þp1ð1
lÞþð1
p1Þt
m1Aþm2Aq2¼waq1þq1ð1
tÞþð1
q1Þl
m1aþm2aThefrequencyofAingeneration2,p2,equalswAp1plusthemutationratefromatoA,minusthemutationratefromAtoa,minustheemigrationrateofA,plustheimmigrationrateofA.Mutation,migration,andselectionarefullycomparable.This,ofcourse,isnotadiscoverybyusbutissimplyelementarypopulationgenetics.Thus,ifMatthenandAriewsclaimwerethatdifferentfactorsofevolutionarychange,suchasselectionandmutation,arenotcomparable,theirclaimwouldbecontrarytostandardpracticeinpopulationgeneticsandwouldbewrong.Butthatisnottheirclaim;rathertheyclaimthatdifferentcom-ponentsoffitnessarenotcomparable.Againthisclaimseemstobecontradictedbystandardpopulationgenetics.AsMichod(1999,12)pointsout,Almostallmodelsofnaturalselectioninvolvesomekindoffitnessdecompositioninoneformoranother.Perhapstheexamplesmostfamiliartophilosophersaregroupselectionmodelsfortheevolutionofaltruism.InsuchmodelsdifferentcomponentsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n72robertn.brandonandgrantramseyoffitnessareseparatedinfitnessequations.Thefollowingequationsarerepresentative:ws¼1þbxwa¼1þbx
cwherewsisthefitnessofaselfishtypeandwathefitnessofanaltruist,cisthecostofaltruism,bisthebenefit,andxisthenumberofaltruistswithintheparticulargroup;crepresentsthecomponentofselectionduetotheselectivedisadvantageofaltruismwithinagroup,whiledifferencesinthevalueofx,andthereforeofbx,representthecomponentofselectionduetotheselectiveadvantageofgroupswithalargernumberofaltruists.Thus,inthiscaseatleast,differentcomponentsofselectionarecomparable,justasselection,mutation,andmigrationarefullycomparable.Andso,itwouldseem,themajorpremiseofMatthenandAriewsargumentissimplywrong.Wethinkitiswrong,butwearenotsurethattheprecedingexamplefullyaddressestheirpoint.Theirpoint,wethink,isthatwehavenogeneraltheorythatwouldallowustocomparedifferentcomponentsoffitness,thatis,nothinglikevectoradditioninNewtonianmechanics.Intheirexample,wehavenotheorythatallowsustocombinethecomponentsofselectionduetosex-ratiodifferencesandthoseduetoparental-caredifferences.Thisinter-pretationmayexplaintheiroddremarkaboutthefailureofprob-abilitytheorytoprovideaframeworkforsuchacomparison.Soourgroupselectionexamplewouldbeatypicalinthatinthiscasewedohaveanexplicittheoryofhowtocomparetheindividual-andgroup-selectioncomponents.Ifthisinterpretationoftheirremarksiscor-rect,thenweagreewiththem,butthentheconclusionoftheirargumentdoesnotfollow.Itishardlysurprisingthatwehavenogeneraltheorythatwouldallowustopredictthefitnessofeverypossiblecombinationofeverycharacterstate.Anysuchtheorywedevelopislikelytobelocalandposthoc.Inapopulationwherevarioussex-ratiostrategiesareextantasarevariousparental-carestrategieswecanmeasurethefitnessesoftheextantcombinations.Withsufficientstudywemaybeabletoofferanecologicalexplanationofthesefitnessvalues.Buttheresultinggeneralizationsdonotderivefromanygeneraltheory.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift73Thefitnessvaluesofvarioustypesareamongthebasicparametersofmodelsinpopulationgenetics.Theyarelikeotherbasicparameters,suchasmutationrates,migrationrates,andeffectivepopulationsize,inthattheyneedtobemeasuredempiricallyandcannotbepredictedfromsomegeneraltheory.Thereishereadifferencefrom,andasimilarityto,Newtonianmechanics.Thedifferenceisthelocalityoftheseparametervaluesthattheyapplytoparticularpopulationsinparticularenvironmentsandtheresultantneedtoremeasurethemtimeandagain(seeBrandon1994).ThesimilarityisthatthebasicparametersofNewtonianphysicsneedtobeempiricallymeasuredaswell,forexample,thevalueofG,thegravitationalconstant.ButhavewenotjustconcededMatthenandAriewspoint,namely,thatfitnessmustbemeasuredintermsofitsconsequences?No.First,wecandevelop,throughdetailedecologicalinvestiga-tions,localtheoriesoforganism-environmentinteractionsthatwouldallowustomeasurefitnessindirectly.Second,andmuchmoreimportantly,whenlookingattheeffectsoffitnesswedonot8havetolookatevolutionary,ortransgenerational,change.Wecan,andbiologistsoftendo,lookatsomethingelse.3.methodsfordetectingselectionJohnEndler(1986)inhiscomprehensiveoverviewofstudiesofnaturalselectioninthewilddescribestenmethodsfordetectingnaturalselection.Forpresentpurposeswedonotneedananalysisatthatfineagrain,althoughwewouldrecommendhisaccounttoanyphilosopherwhowouldpronounceonhowbiologistsmustmeasurenaturalselection.Asimplerclassificationresultsfromfirstdistin-guishingbetweenthosemethodsthatdetectselectionintermsofitseffectsversusthosethatdetectselectionintermsofitscauses.LetuslabelthesecondcategoryCF,forcausesoffitness.Thefirstcategoryneedstobefurthersubdivided.ThefirstsubdivisionmethodsthatdetectselectionintermsofevolutionaryconsequenceswillbelabeledEC.Wewilllabelthesecondsubdivisionmethodsthatdetectselectionbydirectmeasurementof(partsof)theprocessofnaturalselection,thatis,measurementofdifferentialsurvivorship,matingability,fertility,fecundity,andsoforthDM.StudiesusingmethodCFaredifficultinthattheyrequiredetailedknowledgeoforganism-environmentrelations.AswewillseetheyareCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n74robertn.brandonandgrantramseyrare.Butitisimportantthattheyarenotimpossible(Lewontin1978;Brandon1990,chap.1).Theyarenot.However,onthebasisofourknowledgeofevolutionarybiology,wewouldsaythatweareclosetobeingabletoapplymethodCF,usually,ifnotinevitably,throughrepeatedapplicationsofmethodDM.Wewillreturntothisshortly.MethodECisthemethodthatMatthenandAriew(2002)claim9biologistsmustuse.WhatiscommontoallthecaseswelumpunderECisthatpatternsofvariation,eitherextant(horizontal)orovertime(vertical),areusedtocomparemodelsofselectiontoanullmodelofnoselection.Themodelsmaybeinformalandqualitative(EndlersIIIIandsomecasesofV),ortheymaybeformalandquantitative(EndlersIVandsomecasesofV).Buttheessentialfeatureofallofthesecasesisthatthepastorpresentexistenceofselectionisinferredfromdatathateliminatethenull(no-selection)hypothesis.AfewexampleswillclarifyjusthowECworks.Ifoneobservesaconsistentcorrelationbetweensomeenviron-mentalfeatureandcharacterstate,thenonecanhypothesizethattheseenvironmentaldifferencesleadtodifferentselectiveenviron-mentsthatresultintheobserveddistributionofcharacterstates.Forexample,onemight,asKettlewell(1955,1956)did,observeacor-relationbetweenthedarknessoftreetrunks(duetoindustrialpol-lution)andthefrequencyofthedarkmorphinBistonbetularia.Theselectionhypothesisisthatinwoodsaffectedbypollutionthedarkmorphisselectivelyfavoredoverthelightform,andviceversa,in10nonpollutedwoods.Ofcourse,Kettlewelldidnotstopwiththishypothesis;hewentontodemonstrateexperimentally,usingmethodDM,thatselectionwasindeedoperatinginaccordancewiththeselectionhypothesis.Wethinkallwillagreethatthiswasagoodthing.Althoughtheobservedpatternsofvariationwerenotcon-sistentwiththenullhypothesis(whichinthiscasewouldbethatthedifferentcolormorphsweredistributedrandomlyaboutthedifferentareasKettlewellstudied),theyareconsistentwithstillotherhypotheses.Forinstance,thehypothesisthatairpollutionhasadevelopmentaleffectonmothsthatdarkenswingcolorisnoteliminatedbytheobservedpatternsofvariation.Thissortofprob-lemseemstobequitegeneralwhenthemodelsinquestionareinformalandqualitative.If,asintheprecedingcase,therearemultiplenonselectionhypotheses,thentheeliminationofoneofthemwillnotautomaticallysupportthehypothesisofselection.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift75Thisislessofaproblem,butstillaproblem,whenthemodelsarequantitative.ToillustratehowECworkswithsuchmodelswewilldescribetwoexamples,thefirstasimpletoyexample,andthesec-ondagenuinelyinterestingpieceofcontemporarybiologicalresearch.Considerasimpleexampleofheterozygotesuperiority.Supposethattherearetwoalleles,Aanda,atalocusandthatthelocusisinlinkageequilibriumwithallotherloci.ThefitnessofAaisnor-malizedto1,andselectioncoefficientsareassignedtothetwohomozygoticgenotypes.Asimplepopulationgeneticmodelshowshowapopulationsatisfyingtheprecedingwillsettleintoanequi-librium.TheequilibriumfrequenciesofAandacanbemathemat-icallyderivedgiventhevaluesoftheselectioncoefficientsassociatedwiththetwohomozygotes.Thatis,thefitnessvaluesmathematicallydeterminetheequilibriumfrequenciesofthetwoalleles.Buttherelevantequationsworkinbothdirections,sogiventheequilibriumfrequencieswecandeterminethefitnessvalues.Howcouldweusethisfacttoinfernotjusttheexistenceofselec-tion,butthequantitativestrengthofselection?Ifweobservedstableallelefrequenciesatthelocusoveranumberofgenerations,thenwecouldshowthatthelikelihoodofthenullhypothesis(inthiscasethattheallelesareselectivelyneutral)isconsiderablylowerthanthatoftheselectionhypothesis(seeBrandon2005andBrandonandNijhoutforthcoming).Andgiventhesupportoftheselectionhypothesiswecouldthengoontoinfer,inthemanneroutlined,thefitnessvaluesofthethreegenotypes.Ofcoursetheseinferencesarebasedontheassumptionsmentioned,soonesconfidenceintheinferencesshouldbeproportionaltoonesconfidenceinthetruth,orapproximatetruth,oftheassumptions.Considerthisslightlydifferentscenario.Inthiscasewehavenoaccesstoverticaldata,butwecanobservestrongselectionagainstoneofthetwohomozygotes(e.g.,weobservethenegativeeffectsofsicklecellanemiaassociatedwithaknownhomozygoticgenotypeatthehemoglobinlocus).Ifweassumethatthepopulationisatequilibrium,thenwecanagaineliminatethenullhypothesisandestimatethefitnessvaluesoftheothertwogenotypes.TheprecedingtoyexamplesaremeanttoillustrateclearlytheinferentialcharacteroftheECmethod.LetusnowturntotheworkofMartyKreitmanandothers,whichweconsidertobethemost11interestinguseofECmethod.KreitmanhasdevelopedanelegantCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n76robertn.brandonandgrantramseymethodfordeterminingwhereinthegenomeselectionisacting,withdriftbeingtheexplicitnullhypothesis.Becauseoftheredun-dancyofthegeneticcode,substitutionsinthethirdpositionofacodonoftenproducesynonymouscodons(i.e.,codonsthatcodeforthesameaminoacid).Giventhisfact,onecancomparethebehaviorofthefirsttwocodonswiththatofthethird.Ifselectionisactingattherelevantgenomicregion,thefirsttwopositions(substitutionsinwhichwillnotproducesynonymouscodons)shouldbehavediffer-entlythanthethird.Incontrast,ifselectionisnotactingontheregion,thenthefirsttwopositionsshouldbeasfreetodriftasthethird,andthusnodifferenceisexpectedinthebehaviorofthethirdposition.Thisisexcitingworkandiscertainlyapowerfulwayofinvestigatingtheselectionist/neutralistdebate.Butnoticethatthissortofworktellsusnothingaboutthewhyofselection.Itoffersnoecologicalexplanationofselection.AccordingtoDarwiniantheory,smalldifferencesinorganismscanresultindifferencesinvariousabilitiesandcapacities,suchastheabilitytosurvive,theabilitytoattractmates,fertility,orfecundity.AlthoughMatthenandAriew(2002)complainthatwehavenowaytocombineorcomparethesedifferentcapacities,na-turesurelydoes,sinceattheendoftheday,attheendofthegen-eration,theycombinetoproduceagivenlevelofreproductivesuccess.Andthiscapacityisexactlywhatthepropensityinter-pretationoffitnessdefines.Asempiricalbiologistsweshouldnotbesurprisedthatthereisnogeneraltheoryabouthowthesevariouscapacitiescombinetoproducefitness,butthereisageneralmethodtoinvestigatethis.Itiscalledfitnesscomponentanalysis(seeEndler1986,8486,fordiscussionandreferences).Basically,wewanttosamplethepopulationunderstudyatasmanylifehistorystagesaspossible.Theideaisthatdifferentcapacitieswillmanifestthem-selvesatdifferentstagesoflifehistoryandwecanthenseeempiricallyhowtheycombine.Theideal,thoughrarelyattainable,wouldbetoobserveeverymemberofthepopulationthroughoutitsentirelifetime.Inpractice,biologiststypicallylookattheeffectsofasmallnumberofthesecapacities,gettingadirectmeasureofsomecomponentoffitness.ThisisthemethodwearelabelingDM.LetusfirstdescribeacoupleoffamiliarstudiesthathaveusedtheDMmethod;thenwewanttoexploresomeimportantphilosophicaldifferencesbetweenitandtheECmethod.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift77ThemostfamousstudiesofnaturalselectioninthewildarethoseconductedbyH.B.D.Kettlewell(1955,1956).Aswehavealreadyseen,thecorrelationaldatasupported(perhapsonlyweakly)thehypothesisthatselectionwasresponsiblefortheincreaseinthemelanicformofBistonbetulariainwoodsdownwindoflargeindustrialareas.ButKettlewelldidaseriesofexperimentstosupportthathypothesismorestrongly,andinhisbestknownexperimentshemarkedindividualmoths,releasedthemintothewoods,andthenrecapturedthemseveraldayslater.Hethencomparedtherelativefrequencyofthetwocolormorphsintherecapturedgrouptothatinthereleasedgroup.Intheexperimentsconductedinwoodsdown-windofindustrialareas,heobservedanincreaseintherelativefrequencyofthedarkformintherecapturedgroupcomparedtothereleasedgroup.Onthebasisofmanyauxiliarystudies,heattributedthischangeinfrequencytoselection(bybirds),andthushadameasureofonecomponentoffitnessofthetwomorphsinthatenvironment.Asameasureoflifetimefitness,thesortoffitnessthatultimatelymattersforevolution,Kettlewellsstudyisincomplete.Ittellsusnothingabouthowthetwoformsperforminthelarvalstage;ittellsusnothingaboutanydifferencesthatmightexistbetweentheminmatingability,fertility,fecundity,andsoon.Itspowertoexplaintheexistingpatternsofvariationdependsonthetruthoftheassumptionthatthetwoformsaremoreorlessequivalentwithrespecttotheirothercomponentsoffitness.Butgiventhefitbetweentheobservedselectiondifferentialsandtheobservedpat-ternsofvariation,thatassumptionisnotatallunreasonable.And,mostimportantly,Kettlewellsworkisadirectdemonstrationoftheexistenceofselectionintheareashestudied,duringthelifehistorystagethathestudied.Lessfamiliar,butmorecomplete,arethestudiesoftheevolutionofheavymetaltoleranceingrassesconductedbyJanisAntonovicsandothers(Antonovics,Bradshaw,andTurner1971).Herefitnessesweremeasuredmoredirectlyandmorecompletely.AsisthatofKettlewellsmoths,thistooisastudyofadaptationtoanenviron-mentalperturbation.Miningactivitiesproducesoilswithhighlevelsofheavymetals.Thesemetalsaretypicallytoxictomostplants.Whenthiscontaminatedsoilispiledbythesideofamine,thereisoftenasharpboundarybetweenmetal-contaminatedsoilsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n78robertn.brandonandgrantramseyandnormalsoils.Thisdramaticdifferenceinafactorintheexternalenvironmentleadstoadramaticdifferenceinselectiveenviron-ments.Antonovicswasabletoshowthisbymonitoringthelivesoflargenumbersofindividuals,bothinthecontaminatedsoilandinnoncontaminatedsoil.Lookingatdifferencesinlifetimesurvivor-shipthestrengthofselectionwasmeasuredinbothselectiveenvironments.Geneticallytolerantplantswerestronglyselectedovernontoleranttypesinthecontaminatedsoil,andviceversainthenormalsoil.Surprisingly,selectionwassostrongthatgeneti-callydifferentiatedsubpopulationswereproducedoververyshortdistancesinspiteofconsiderablegeneflowbetweenthem.AntonovicssmeasuresoffitnessvaluesweremorecompletethanKettlewellsinthathewasabletolookatlifetimesurvivorship,ratherthansurvivorshipduringasmallportionofthelifecycle.Theyweremoredirectinthatdifferentialdeathswereactuallyobserved,ratherthaninferredfromdifferencesbetweenreleasedandrecapturedgroups.Butthisstillisameasureofacomponentoffitness,notcompletefitness.Forinstance,therewasnoattempttomeasurepotentialdifferencesinfertilityandfecundity.Butgiventhestrengthofselectionobserved,andtheremarkablegeneticdif-ferentiationassociatedwiththedifferentselectiveenvironments,wecanbereasonablyconfidentthatAntonovicssmeasuresoffitnessescapturedacrucialpartofthecausalstory.Aswesaidpreviously,theidealapplicationoftheDMmethodwouldbetoobserveeverystageofthelifehistoriesoftheorganismsinthepopulationunderstudy,andsotomeasureeverycomponentoffitnessandtothenseehow,inthisparticularsituation,thosecomponentscombinetoproduceoverallfitness.Thereisabsolutelynophilosophicalorconceptualdifficultyindoingthis.Thediffi-cultiesareofapracticalnature.AndsobiologistsusingtheDMmethodalmostalwaysmeasuresomecomponentorcomponentsof12fitness.Suchmeasuresprovidegood(butnotcomplete)explana-tionsofevolutionarychangetotheextentthatthemeasuredcomponentsoffitnessdominatetheunmeasuredcomponents.OnemightthinkthatthedifferencebetweentheECandDMmethodsisratherminor,thattheybothdetectfitnessvaluesintermsofeffects,andthattheonlyrealdifferenceisthattheECmethodlooksforeffectsoveranevolutionarytimescalewhiletheDMmethodlooksateffectsoftraitssuchasdifferentialmatingCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift79abilitythatareobservableoverthetimescaleofasinglegeneration.Thatisadifference,andanimportantdifference,butitisnottheonlydifference.Beforediscussingthelessobviousdifferencesbetweenthesetwomethods,letusbrieflycommentonthemajorconsequenceoftheprecedingdifference.Itisnotminor.SinceEClooksattrans-generationaleffects,itnecessarilyconfoundstheeffectsoftheeco-logicalprocessofselection(whichisbasicallywhatDMstudies)withmultipleeffectsofthegeneticsystem.Forinstance,incasesofheterozygotesuperiority,oncethepopulationreachesequilibriumthereisnoevolutionarychange.ThustherearenoECeffects.Withtheappropriateverticaldata(manygenerationswithnochangeatthatlocus)orhorizontaldata(theexactsamesystemfoundinanumberofrelatedspecies)theECmethodcouldeliminatethenullhypothesisofneutrality,butitcouldnotestimatetheselectiondifferentials.WithoutthosedatatheECmethodcannotevendif-ferentiateacaseofstrongselection,saythatbothhomozygotesarelethal,fromnoselection.Combiningtheresultsoftwoseparableprocessesresultsinalossofinformation.And,unfortunatelyfortheECapproach,thatinformationiscrucialtoevolutionaryexplanationandprediction.Wewillreturntothisshortly.Atleastasimportantphilosophicallyisalessobviousdifferencebetweenthetwomethods.Althoughwewillneedtoaddalittlenuancetothis,theECmethodismodelbasedandinferential;theDMmethodisnotmodelbasedandisappropriatelydescribedasmeasurement.Letusfocusonhowwecancometoknowfitnessvaluesbymeansofthesemethods.ItshouldbeclearfromourdiscussionofECexamplesthatdataaboutpatternsofvariationareusedtosupportamodelofselection,fromwhichfitnessvaluescanbeestimated.Withoutthemodel,therecouldbenoestimationoffitnessvalues,sincethedataaresimplypatternsofhorizontaland/orverticalvariation.Philosophicallyspeaking,theinferencetofitnessvaluesinthesecasesisabductive.NowwedonotwanttoclaimthatinusingtheDMmethodnoinferencesaremade.RememberthatKettlewellinferredthatthedifferencesinrelativefrequencyofthetwoformsintherecapturedclasscomparedtothereleasedgroupreflecteddifferentialpredation.He,infact,hadalotofevidencetobackupthatinference.WedonotCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n80robertn.brandonandgrantramseythink,however,weshoulddescribethatevidenceasamodel.Antonovicssmoredirectmeasurementsavoidedthatparticularinference.Wearenotparticularlyconcernedherewiththecurseofpost-Kuhnianphilosophyofsciencetheviewthatallobservationistheoryladen.Whateveronethinksaboutthat,wewouldhopeonewouldstillbeabletodistinguishmeasurementsofsomequantityinnaturefrommodel-basedinferences.SurelymuchDMworkisdonebecausetheinvestigatorisinterestedinsomehypothesisormodel.Andonecanalwaysdescribeanyparametermeasurementasahypothesistest(seeBrandon1994).Butthatpossibilitydoesnotmeanthatthatisthemostperspicuousdescription.TheDMmethodisappropriatelynamed;itisamethodofparametermeasurement,onethatismoreorlessdirect.4.applyingthemethodsofselectiondetectionWhenMatthenandAriewclaimthatpopulationgeneticistsareforcedtoestimatefitnessbymeasuringpopulationchange,wetakethistobebothadescriptiveandaprescriptiveclaim.IfbiologistsareforcedtousetheECmethod,thentheydouseit,andpresumablyuseitexclusively.Thatisdescriptive.Furthermoreiftheymustestimatefitnessthatway,thentheyoughttodoitthatway.Thatisprescriptive.Ourownprescriptiveviewsmaywellhavecomethroughinthelastsection,butfortherecord,letusbeexplicit.WethinkthatitisimportantthatitbepossibletoapplytheCFmethod,inprincipleatleast.ButwethinkthattheonlywaywewillhavethebiologicalknowledgerequiredtoapplythatmethodisthroughrepeatedapplicationsoftheDMmethod.InthiswaytheDMmethodhaspriorityovertheCF.WehavealsosaidthatsomereallyinterestingworkhasbeendonebyusingtheECmethod.Wereweincharge,wewouldcertainlyfundmoreofit.ButtheECmethodisreallyamethodoflastresort.InKrietmansstudiesitisusedbecauseheislookingovervastexpansesofevolutionarytimeandislookingatgenomicregionswherethefunctionisoftenunknown.ItwouldbeimpossibletoapplytheDMmethodhere.ButwecanimaginenosituationinwhichboththeDMandCFmethodswereapplicableandtheCFmethodpreferable.Inthissense,theDMmethodisabetterwayofdoingevolutionarybiology.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift81Prescriptivedisputesarenoteasilysettled,thoughwehopewehavegivensomegoodreasonsinsupportofourviews.ButthedescriptiveimplicationsofMatthenandAriewsclaimcanbeeasilydismissed.Therearerealdatahere.Table5.1inEndler(1986)listsallofthepublisheddemonstrationsofnaturalselectioninthewildthatEndlercouldfind.Surelyhemissedoneortwo,butthisisbyfarthemostcomprehensivesurvey13oftheliteratureinexistence.Endlerliststhestudiesbyspeciesandthenbythetraitsstudiedinthatspecies.Forinstance,inHomosapiensthereareentriesfortoothsize,birthweightandgestationtime,height,bodyshape,andhaemoglobinS.Foreachspeciestraitherecordsthemethod(s)ofdemonstration.WemappedEndlerstenmethodsontoourthreemethodsasfollows:EndlersIVareourEC,hisVIVIIIareourDM,andhisIXXequatetoourCF.WhenEndlerlistedmorethanonemethodwecountedmorethanoneofourmethodsifhislistedmethodscrossedourcategories(ashappenedonlyonce).Theresultsareasfollows:MethodNumberofstudiesEC1DM172CF2IthinkitisfairtoconcludefromthisthatMatthenandAriewsdescriptiveclaimisfalse.Andiftheirprescriptiveclaimistrue,thenevolutionarybiologistsarecertainlynotbehavingastheyought.5.conclusionsIfcaversweretoracefromonecaveentrancetoanother,thewinnerwouldsurelyowehersuccesstosuchcharacteristicsasherabilitytonavigate,herswiftnessatmakingverticalascentsanddescents,andherabilitytosqueezethroughnarrowapertures.Theabovegroundobserverwillrecognizethattheseskillsarenecessarybutwouldnotbeabletosayhowthecaversskillscombinedtoleadhertovictory.ThosewhoarguefortheESIapparentlythinkthatbiologistsareinthesameepistemologicalcul-de-sac.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n82robertn.brandonandgrantramseyThosewhoargueforthePSIwouldarguethatthecaseofthecaversisdisanalogoustonaturalselectionintwoways.First,unliketheabovegroundobserver,scientistsareabletoobservenotjusttheoutcomeofnaturalselection,butalsotheprocess.Thiswouldbelikebeingabletotrackthecaversandsee,forexample,howmanynavigationalerrorstheymakeorhowfasttheyareabletomakeascentsanddescents.Fromthesedatawecould(1)explainwhythewinnerwontheraceand(2)predicthowthecaverswouldfareindifferentcaves.Analogously,aswesawearlier,wecanuseCFtomeasurethevarietyoffactorsthatleadtothesuccessofanindivi-dualorganismortypeoforganism.Wecanusethesedatatoexplainthesuccessoforganismsandpredicthowtheywouldfareindifferentenvironments.Theseconddisanalogyisthatunlikethesingularcaverace,bio-logicalphenomenaarerepeatable.Evenifwecouldnotobservehowthecaversactunderground,wecouldracetheminanumberofdifferentcaves,somewithoutalternateroutes(toeliminatenavi-gationalerrors)andotherswithmany,somewithfewverticaldropsandotherswithmany.Throughthiscomparison,wecouldseewhichcaversfarewellinwhichkindsofcaves.Thiswouldallowustolearnwhichindividualcavers(orcavertype)dowellinwhichkindofcave.Wecouldusethesedatatopredictwhichcaverwouldwininaparticularcaveraceandtoexplainwhythewinnerwonandtheloserlost.Analogously,wecoulduseDMtoseehowdifferentorganismsfareindifferentenvironments.Wecouldevencloneorganismsandraisetheclonesinadiversityofenvironments.Thiswouldgiveusdatatounderstandhowdifferentcomponentsoftheorganismsfitnesscombinetoprovesuccessfulinaparticularenvironment.Aswesawintheprevioussection,DMiscommonlyemployedbybiologists.Insum,theargumentsfortheESIhavebeenthoroughlyrefuted.Thisisagoodthing.Itisincapableofexplainingdifferentialrepro-duction,akeypartoftheprocessofevolutionbynaturalselection.Itcanmerelyposittheexistenceofpopulation-levelstatisticaldis-tributionstheyemergeinmysteriousways.Anditcanmakenosenseofthewaybiologistsactuallymeasurefitnessinthewild.Incontrast,thePSIdoesthesethingseasilyandnaturally.TheESIdoeshavetheadvantageofallowingonetohangontoaphilosophicalprejudice,namely,thatphenomenaatthelevelofindividualCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNaturalSelectionandRandomDrift83organismsaredeterministic.But,aswehaveseen,hangingontothisparticularprejudiceisquitecostly.notes1.Walsh,Lewens,andAriew(2002),andMatthenandAriew(2002);seealsoSterelnyandKitcher(1988).2.In1978Brandondefinedadaptednessasfollows:foranorganismOinOEenvironmentEthereisarangeofpossibleoffspringnumbers,Q1,OEOEQ2,...,Qn,andforeachnumberthereisanassociatedprobability,OEP(QI).TheadaptednessofOinE,A(O,E),thenistheexpectedvalueOEOEofOsreproductivesuccessinE.Thatis,A(O,E)¼RP(QI)QI.Later,drawingontheworkofJohnGillespie(1973,1974,1977),itwasdiscoveredthatthisexpectedvalueneededtobediscountedbysomefunctionofthevarianceinoffspringnumber(seeBrandon1990,1820).Forfurtherdiscussionofthispoint,seeBeattyandFinsen(1989)andSober(2001).3.Weemphasizethewordsupposedlybecauseitseemstousthatourconfidencethattheunderlyingmechanicsisreallydeterministicshouldbemuchlowerthanourconfidenceinthesecondlaw.4.SterelneyandKitcher(1988,345)arguethatevolutionarytheory,likestatisticalmechanics,hasnouseforsuchafinegrainofdescription:theaimismakeclearthecentraltendenciesinthehistoryofevolvingpopulations.5.Wewillnotbeabletodealwiththegeneralissueofontologicalcommitmenthere,butletussimplyassertourviewthatparsimonyisnotanontologicalvirtue;ratheraccuracyis.Theworldeitherisorisnotasimpleplace.Ourjobistodescribeitasitis,notaswewishitwere.6.SeeStephens(2004)forarecentendorsementoftheNewtonianoption.7.ButseeBrandon(2005).8.MatthenandAriew(2002)donotfullyunderstandtheimplicationsoftheirposition.Astheydefineitpredictivefitness(asweshallcallit)isastatisticalmeasureofevolutionarychange,theexpectedrateofincrease(normalizedrelativetoothers)ofagene,atrait,oranorganismsrepresentationinfuturegenerations.Thusintheconclusionquotedwhentheyspeakofpopulationgeneticistsbeingforcedtoestimatefitnessbymeasuringpopulationchange,wemustinterpretpopulationchangeastransgenerationalchange.Theirconceptionoffitnessisnotnovel,itiscalledFisherianfitness.Itisunsuitedforexplanatorypurposes(seeBrandon1990,chap.1;Ramsey2006).UnfortunatelyforMatthenandAriew,theydorealizethattheyarecommittedtothis(seefootnote30,74).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n84robertn.brandonandgrantramsey9.Walshandcolleagues(2002)arelessexplicitonthispoint,butitseemstheyalsoarecommittedtothisview.10.TheviceversahypothesisisnotreallynecessaryherebutwaspartofKettlewellsexplicitexperimentalresearch.SeeBrandon(1999)andRudge(1999).11.SeeYangandBielawski(2000),BamshadandWooding(2003),andHamblin,Thompson,andDiRienzo(2002).12.SeeBrandon(1990,chap.5)foranaccountofideallycompleteadaptationexplanations.13.Ofcourse,itisnowseventeenyearsold.Foramorerecent(198497)listofstudiesofnaturalselectioninthewild,seeKingsolveretal.(2001).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\npaule.griffithsandkarolastotz5GeneThehistorianRaphaelFalkhasdescribedthegeneasaconceptintension(Falk2000)anideapulledthiswayandthatbythedifferingdemandsofdifferentkindsofbiologicalwork.Severalauthorshavesuggestedthatinthelightofcontemporarymolecularbiologygeneisnomorethanahandytermthatacquiresaprecisemeaningonlyinsomespecificscientificcontextinwhichitisused.HencethebestwaytoanswerthequestionWhatisagene?,andtheonlywaytoprovideatrulyphilosophicalanswertothatquestionistooutlinethediversityofconceptionsofthegeneandthereasonsforthisdiversity.Inthisessaywedrawontheextensiveliteratureinthehistoryofbiologytoexplainhowtheconcepthaschangedovertimeinresponsetothechangingdemandsofthebiosciences.InthissectionwehavedrawnprimarilyontheworkofRaphaelFalk(1986,1991,1995,2000,2001,2005,inpress),MichaelDietrich(2000a,2000b),RobertOlby(1974,1985),PetterPortin(1993),andMichaelMorange(1998).Whenourhistoricalclaimsarecommonplacesthatcanbefoundinseveralofthesesourceswedonotcitespecificworksintheirsupport.Wehavealsochosennottoexplainbasicgeneticterminology,asthiswouldhaveoccupiedmuchofthechapter.Morespecializedtermsareexplainedwhentheycannotbeavoided.Inthefinalpartoftheessayweoutlinesomeoftheconceptionsofthegenecurrenttoday.Theseedsofchangeareimplicitinmanyofthosecurrentconceptionsandthefutureofthegeneconceptappearssettobeatasturbulentasitspast.theinstrumentalgeneInthefirstthreedecadesofgeneticresearchthegenehadadualidentity(Falk1986,2005).Genes,orMendelianfactors,were85CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n86paule.griffithsandkarolastotzinterveningvariablesdefinedbytheMendelianpatternofinheri-tance.Fromthisperspective,thefactthatsometraitofanorganismcanberesolvedintooneormoreMendeliancharactersestablishesdefinitivelythattherearegenesforthosecharacters.Indeed,itseemsthatatleastsomeoftheearliestMendeliansdidnotclearlydistinguishbetweentheMendeliancharacteritselfandtheMendelianfactorunderlyingit.ThatdistinctionwasmadeclearbyWilhelmJohannsensintroductionofthetermsphenotypeandgenotypein1909.Butaswellasinterveningvariables,geneswerehypotheticalmaterialconstituentsofthecellwhosephysicaltransmissionfromparenttooffspringcausallyexplainedtheMendelianpatternofinheritance.InhisNobelPrizeacceptancespeechThomasHuntMorgan,thefatherofclassicalgenetics,noted,Thereisnotcon-sensusofopinionamongstgeneticistsastowhatgenesarewhethertheyarerealorpurelyfictitiousbecauseatthelevelatwhichgeneticexperimentslie,itdoesnotmaketheslightestdifferencewhetherthegeneisahypotheticalunit,orwhetherthegeneisamaterialparticle(1933,quotedinFalk1986,148).Inourview,oneoftheclearestthemesinthecentury-longevolutionoftheconceptofthegeneisthedialecticbetweenthesetwoconceptionsofthegene,astructuralconceptionanchoredfirstincytologyandlaterinbiochemistry,andafunctionalconceptionanchoredintheobserv-ableresultsofhybridizations,atfirstbetweenorganismsandlaterdirectlybetweenDNAmolecules.Recentscholarshiphasstressedthefactthatclassicalgeneticswasnotmerelyatheoryofheredity,butatleastasimportantlyanexperimentalpracticegeneticanalysisinwhichtheregularitiespostulatedbytheMendeliantheoryofhereditywereusedtoaddressotherquestionsaboutthestructureandfunctionoflivingsystems(Waters2004;Falkinpress).Thisexperimentalpracticeimposedstrongconstraintsontheconceptofthegene.IntheearliestdaysofMendeliangenetics,WilliamCastleshybridizationexperimentswithhoodedratschallengedthediscretenessandconstancyofMendelianfactors.Inthoseexperimentsallelesappearedtobecontaminatedbytheallelestheyhadsharedacellwithinpreviousgenerations.Theresultingdebateexposedacircularityofargu-mentation:unitfactors(individualMendeliangenes)canonlybeidentifiedbytheireffectonunitcharacters(thosethatdisplayasingle,consistentMendelianpatternofinheritance),buthowCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGene87canaunitcharacterthatissupposedtostandforaunitfactorbedelimited?Thiscircularitywasresolvedbydefinition:Mendelizingtraitsaredeterminedbyasinglegene,andnon-Mendelizingtraitsarecontrolledbymorethanonegene.TheinstrumentalgeneisbydefinitionaMendelizingunititistheretodoajobthatdependsonthisstipulation.Thevisible,heritablecharactersoforganismsmustbeinterpretedinsuchawayastopermitgeneticanalysisofthosetraits.Ifacharacterdoesnotcorrespondtoagenethenitmustbedecomposedintosimplercharactersthatdo(laterdescribedaspri-marycharacters).Inthesamespirit,quantitativetraits,whichvarycontinuouslybetweenindividualsandthuscannotoccurinMendelianratios,weretreatedastheeffectofmanyhypotheticalgenes,eachofwhichmakesanequalandinseparablecontributiontothecharacter,givingrisetothedisciplineofquantitativegenetics.thematerialgeneTheMorganschoolrapidlyestablishedthechromosomaltheoryofheredity,accordingtowhichgenesarearrangedinalinearfashionalongthechromosomesthatcytologistshadobservedinthecellnucleus.TheywereabletoexplainmanydeviationsfromthestandardMendelianpatternofinheritanceintermsoftheobserv-ablebehaviorofchromosomes.Mostimportantly,theywereabletocorrelatecloselythelinkagemapsgeneratedbygeneticanalysiswithobservablechangesinthestructureofchromosomes,anachievementfacilitatedbythediscoveryofhuge,polytenicchromo-somesinthesalivaryglandsofDrosophila.Linkagewasthusbotha(functional)measureoftheprobabilitythattwogeneswouldbeinheritedtogetheranda(structural)factabouttherelativepositionofvisiblebandsonthesalivaryglandchromosomes.Butdespitetheseachievements,mostmembersoftheMorganschooldidnotconcernthemselveswiththematerialnatureofgenes,bothbecausethiswasnotaquestionthatcouldbepursuedviageneticanalysisandbecausethepursuitofgeneticanalysisdidnotrequireittobeanswered.Molecularbiologywasbornwhengeneticists,nolongersatisfiedwithaquasi-abstractviewoftheroleofgenes,focusedontheproblemofthenatureofgenesandtheirmechanismofaction(Morange1998,2).ForemostamongthesewasHermanJ.Muller,astudentofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n88paule.griffithsandkarolastotzMorgansnotsatisfiedbythepurelyinstrumentalnotionofthegeneasanunknownphysicalentitylocalizedonchromosomes.ForMullertheseparticulate,atomicentitieswerethebasis,thesecretoflife,andtheessentialentitiesonwhichtheDarwinianprocessofevolu-tionrests.Inordertofulfillthesefunctionsgenesneededtohavethepropertiesofautocatalysis(self-replication)tomakethemunitsofheredity,heterocatalysistoallowthemtocontributetothepheno-type,andmutabilitytocreateheritablevariation.Mullersetuparesearchprogrammetostudythematerialnatureofthegeneandrevealthephysicalbasisoftheseproperties.In1927Mullerdis-coveredthemutageniceffectofx-raysandusedthistomakethefirstestimatesofthephysicalsizeofanindividualgene.Forourpurposes,Mullersemphasisonthematerialgeneisimportantbecauseofhiscommitmenttofindinganepistemicpathwaytothegenethatbypassedtheobservedeffectofthegeneofthephenotype.Whenthiscommitmentstartedtobearfruititbecamepossibletoadvanceaconceptofthegenethatabandonedsomeofthecommitmentsrequiredifgenesweretobeepistemicallyaccessibleviageneticanalysis.Featuresofthegenethatpreviouslycouldnotbemeaningfullycalledintoquestionandthatwerethustreatedasdefinitionalbecamefeaturesthatcouldbetestedandpotentiallyrejected.Thematerialnatureofthegenewasprogressivelyrevealedbythenewdisciplineofbiochemistry,whichcameintobeingintheinterwaryears.Oneaimofthisdisciplinewastounderstandthesynthesisoftheagentsoforganicspecificityorganicmoleculesthatinteractonlywithaverynarrowclassofothermoleculesandthusallowtheveryprecisechemistryrequiredbylivingsystems.Fromthemid-1930sitbecameincreasinglyclearthatthespecificityoforganicmoleculesisexplainedbyconformationandweakinter-actionsbetweenmolecules.Theconformationofamoleculeisitsthree-dimensionalshape,whichdetermineswhetherspecificsitesonmoleculescancometogether.Theinteractionsbetweenthosesitesaremuchweakerthanthecovalentbondsofstandardinorganicchemistry,sothatinteractionsbetweenmoleculesandthecon-formationofindividualmoleculescanbealteredbyrelativelylowenergies.Theseprinciplesturnedouttounderliethestructureandfunctioningofallformsoflife(Morange1998,15).TheconceptofspecificityrapidlybegantobeappliedtotherelationshipbetweenCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGene89genesandtheirproducts,aswellastotherelationshipbetweenenzymesandtheirsubstrates.Iftheactivityofthecellisexplainedintermsofmolecularspec-ificityitisnaturaltosupposethattheeffectsofgenesonphenotypesaremediatedbytheproductionofbiomoleculeswithappropriatespecificity.Thusin1941theonegeneoneenzymehypothesis,whichhelpedtoforgeanexperimentalassociationbetweenbio-chemistryandgenetics,wasborn.GeorgeBeadleandEdwardTatumchosetoattacktheproblemofgeneactionbygeneticanalysisofaknownbiochemicalprocess.TheyproducedandisolatedmutantstrainsofthefungusNeurosporaeachunabletosynthesizeoneofseveralchemicalsinvolvedinasinglebiosyntheticpathway.Geneticanalysisofthesemutantsshowedthateachdeficiencywastheresultofamutationinasinglegene.OnlythreeyearslaterOswaldT.AveryproducedexperimentalevidencethatgenesweremadeofDNA.Lookingback,hisevidenceseemscompelling,butitneededanothereightyearsandadifferentlineofexperimentforittochangethereceivedproteinmodelofthegene.Iftherelationshipbetweengenesandenzymeswasoneofspecificity,liketherela-tionshipbetweenenzymesandtheirsubstrates,thenitseemedunlikelythatDNAcouldberesponsibleforgeneticspecificity.ThelittlethatwasknownaboutDNAsuggesteditwasanunspecificandmonotonousmolecule,perhapswithastructuralroleinthechro-mosome.Historianshavestressedtheverysubstantialchangesinapproachproducedbytheinfluxofscientiststrainedinphysicsintobiologyduringthe1940s.Thesechangesmovedgeneticsandbiochemistryclosertogetherandpavedthewayforthemolecularconceptionofthegenethatprevailedfromthe1950stothe1970s.Oneoftheseformerphysicists,MaxDelbru¨ck,wasconvincedthatunderstandingthesecretoflifewouldrequireaphysicalapproachandanorganismassimpleandpureasabacterialvirusanorganismsosimplethatitcouldbeconceivedasanakedgene.Thebacteriophageappearedtohavehardlymorethantheonekeycharacteristicoflife,self-replication,andwasthusdeemedperfecttostudythispropertywithoutopeningthebiochemicalblackbox(Morange1998,45).ThephagegrouparoundDelbru¨ck,SalvadorLuria,andAlfredHersheyhelpedtoestablishbacterialgeneticsandtheprokaryoticageingeneticresearch.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n90paule.griffithsandkarolastotzdoingwithoutgenes?TheclashbetweentheleadinggeneticistRichardGoldschmidtandhiscontemporariesinthe1940sandearly1950sprovidesfurtherinsightintotheclassicalgeneconcept.ThesuccessesoftheMorganschoolindeterminingthelinearorderofgenesonchromosomesallowedthediscoveryofpositioneffectsinwhichachangeintherelativepositionofgenesonthechromosomeisassociatedwithachangeintheirphenotypiceffects.Thisinturnraisesquestionsconcerningthenatureofmutation.Todaywedefineamutationasanyheritablechangeinthenucleotidesequenceofachromosome,whichmayoccureitherbythesubstitutionofonenucleotideforanotherorbythetranslocationorinversionofachromosomesegment.Inclassicalgenetics,however,mutationwasnecessarilydefinedasachangeintheintrinsicnatureofanindividualgenemanifestinaheritabledifferenceinphenotype.Mutationswerethusdistinguishedfrompositioneffects,inwhichanintrinsicallyiden-ticalgenehasadifferenteffectbecauseithaschangeditslocation.Goldschmidtchallengedthisdistinction.Astherewasnodirectevidencethatchromosomeshavedistinctivestructuralpartscorre-spondingtoindividualgenes,hesuggestedthatmutationsandpositioneffectsweresimplysmallerandlargerchangesinthestructureofthechromosome.Becausechromosomalchangesonverydifferentscaleswereknowntohavephenotypiceffects,Goldschmidtarguedthatchromosomesprobablycontainedahierarchyofunitsoffunction.Famously,hedeniedthatgenesexist,bywhichhemeantthatnouniquestructuralunitcorrespondedtotheunitoffunctionofclassicalgenetics.AlthoughGoldschmidtseffortsfrom1940to1958standoutasoneofthefirstattemptstodevelopatheorywhichintegratedmodelsofgeneticstructure,geneticaction,develop-mentalprocessesandevolutionarydynamics(Dietrich2000a,738),hisviewswerecompletelyunacceptabletomostofhiscon-temporaries.Effectively,Goldschmidtwasinsistingthatbothaspectsofthedualidentityoftheclassicalgeneconvergeonasingleunitthematerialgenemustcorrespondtotheinstrumentalunitofgeneticanalysis.Evidencetothecontraryisthusevidencethattherearenogenesintheclassicalsense.Goldschmidtscontemporariesperhapsdifferedinthattheyweremorehopefulthatfuturedis-coverieswouldrevealauniqueunitofgeneticfunctionattheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGene91molecularlevel.TheycertainlydifferedintheircommitmenttocontinuingexistinglinesofresearchandunwillingnesstoundertaketheradicalreorientationthatGoldschmidtwassuggesting.‘neo-classical’geneticsandthemoleculargeneBythemid-1950sDNAwasestablishedasthegeneticmaterial,itsstructurehadbeenanalyzedbyJamesWatsonandFrancisCrick(1953),andCrickhadstatedtheCentralDogmaofmolecularbiologyanditsrelatedsequencehypothesis(1958):thelinearsequenceofnucleotidesinasegmentofaDNAmoleculedeterminesthelinearsequenceofnucleotidesinanRNAmolecule,whichinturndeterminesthesequenceofaminoacidsinaproteinbyinfor-mationalspecificity,thatis,viathegeneticcodewhosedetailsweretobeelucidatedintheearly1960s.Thesameperiodsawaseachangeinthegeneconceptitself,onethatPetterPortinhaslabeledthetransitionfromtheclassicaltotheneo-classicalgene(Portin1993).Itmayappearslightlyconfusingthatthelatterconceptionhasalsobeenlabeledtheclassicalmoleculargene(Neumann-Held1998),butasPortinsneo-classicismispreciselyamolecularizedclassicism,thetwonamesarecomplementary.Thenew,molecularconceptofthegenewastheresultoftech-nicaldevelopmentsthatallowedmuchmoredetailedmapsofthechromosome(finestructuremapping)andtheinterpretationoftheresultsofthisenhancedformofgeneticanalysisinthelightofthenewunderstandingofthematerialgene.Thenewconceptiondepartedfromtheclassicalinrecognizingthatthegeneisnotthefundamentalunitofmutationorofgeneticrecombination.Recombinationinclassicalgeneticswastheprocessinwhichallelesfromtwocopiesofachromosomewerecombinedonasinglecopyasaresultofcrossingoverbetweenhomologouschromosomepairsduringmeiosis.Recombinationwasthusrecombinationofanalleleofonegenewithanalleleofanothergene,sothatgenesthemselvesweretheminimalunitofrecombination.Workingwithbacteriophagefrom1954to1961SeymourBenzerwasabletoincreasetheresolutionofthecis-transorcomplementationtestsoastomapoutindetailthelocationofdifferentmutationswithinthesamegeneanddemonstratedconclusivelythatrecombinationCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n92paule.griffithsandkarolastotzcanoccurbetweendifferentpartsofasinglegene.Twomutationsaresaidtobeincis-positionwhentheyareonthesamecopyofachromosome.Theyareintrans-positionwhenoneisoneachoftwohomologouschromosomes.Thelogicofthecis-transtestdependsonthefactthatmostmutationsarerecessiveintheheterozygote.Hence,ifanoffspringderivesamutantalleleofonegenefromoneparentandamutantalleleofanothergenefromtheotherparent,itshouldalsoreceiveamutation-free,functionalcopyofeachgenefromtheotherparentandappearphenotypicallynormal.If,however,anoffspringreceivesadifferentmutationfromeachparent,buttheyareinthesamegene,thenitwillhavenomutation-freecopyofthatgeneandwillbeaphenotypicmutant.Thus,crossingtwomutantlinestoproduceoffspringwiththetwomutationsintrans-positiontestswhethertheyareinthesamegene.If,however,geneticrecombinationcanoccurwithinasinglegene,thenasmallproportionoftheoffspringofacrossbetweencarriersoftwodifferentmutantallelesofthesamegenewillreceiveacopyofthegenethatrecombinestheundamagedportionfromonemutantallelewiththeundamagedportionfromtheothermutantalleleandisthusrestoredtonormalfunction.Benzerusedananalogueofthecis-transtestinbacteriophagetodemonstratethatthegeneasafunctionalunitdefinedbythecis-transtest(thecistron)canberepresentedasalinearrecombinationmapofmutatedsites.Thisacknowledgmentledhimtodistinguishbetweenunitsofrecombination,recons,mutation,mutons,andgeneticfunction,thecistron.BenzersworkcouldhavebeenseenasavindicationofGoldschmidtandotherskepticismabouttheunified,particulategene(Holmes2000;Falk2005).Butthiswasnothowitwasviewedbyhiscontemporaries.Instead,thecistronwasmoreorlessimmediatelyidentifiedwiththegene.Fromthisfollowedthecon-ventionalgeneconceptofmolecularbiology.OnereasontheresultswereinterpretedinthiswaywasthatthephysicalstructureoftheDNAmoleculewasnowknownandofferedanaturalinterpretationforBenzersfindings.Theunitofrecombinationandmutationisthesinglenucleotide,whilsttheunitofgeneticfunction(hetero-catalysis)isthesequenceofnucleotidesfromwhichasingleRNAistranscribed,correspondingtoasingleprotein,andthusvindicatingtheexistingdoctrineofonegeneoneenzyme.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGene93challengestotheclassicalmoleculargeneconceptBythemid-1960smanyscientiststhoughtthatthemajorproblemsofmoleculargeneticshadbeensolvedandwereinclinedtoleaveotherinvestigatorstoironoutthedetails(Stent1968).ButtheclaimthatwhatistrueforE.coliistruefortheelephantturnedouttobepremature,anditseemsunlikelythatmoleculargeneticistswillfindthemselvesoutofworkanytimesoon.Accordingtotheclassicalmolecularconceptionageneisaseriesofcontiguousnucleotideswhosesequencecorrespondstothesequenceofaminoacidsinasinglepolypeptidechain(oneormoreofwhichmakesupaprotein).ItwassoonrealizedthatsomegenescodeforfunctionalRNAsthatarenottranslatedtoaprotein,butthisfactiseasilyaccommodatedbytheclassicalconception.AsC.KennethWatershasstressed,thefundamentalmoleculargeneconceptisthatofaDNAsequencethatdeterminesthestructureofsomegeneproductbylinearcorrespondence(Waters1994,2000).ThemoleculargeneistheimageintheDNAofthemoleculewhosebiologicalactivityisofinteresttotheexperimenter(RobD.Knight,pers.comm.).Theclassicalmoleculargeneseemedtounitethetwoidentitiesoftheclassicalgeneinasinglenaturalunit.Thefunctionaldefinitionofthegenethatunderlaygeneticanalysisandthestructuraldefinitionofthematerialgenehadturnedouttobetwowaystopickouttheverysamething.Lookedatmoreclosely,however,thefunctionaldefinitionhadbeensignificantlyrevisedsoastotakeaccountoffindingsaboutthematerialgene.InMullersoriginalvisiongenesreproducethemselves(autocatalysis),influencethephenotype(heterocatalysis),andmutate.Theclassicalmoleculargene,how-ever,isnottheunitofreplication,whichisthewholeDNAmole-culeofwhichitisapart.Norisittheunitofmutation.Theonlyfunctionwithrespecttowhichthemoleculargeneistheunitoffunctionisthatofcontributingtothephenotype(Mullershetero-catalysis).Sothefunctionalroleofthegenewasrevisedtofitthemolecularrealitythathadbeenuncovered.Furthermore,thecon-ceptofthegenewasrestrictedtosequencesthatfulfilledthisnewfunctionalrole:notallsegmentsofchromosomesthatbehaveasMendelianfactorscountasgenesunderthenewconception.UntranscribedregulatoryregionsnotimmediatelyadjacenttotheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n94paule.griffithsandkarolastotzcodingsequencestheyregulatecansegregateindependentlyofthosecodingsequences,andsocanfunctionasseparateMendelianfactors,buttheyarenotseparatemoleculargenes.Nevertheless,theclas-sicalmoleculargenewasahighlysuccessfulexampleoftheresearchstrategyofidentifyingafunctionalrole,searchingforthemecha-nismthatfulfillsthatroleatalowerlevelofanalysis,andusingknowledgeofthatmechanismtorefineunderstandingoffunctionattheoriginal(inthiscasephenotypic)levelofanalysis.Sincethe1970s,however,furtherinvestigationoftheunderlyingstructuralunithastendedtounderminetheideathattherevisedfunctionalroleofthegenedeterminingthestructureofageneproductisfilledbynaturalunitsofstructureattheleveloftheDNA.Thestructuresinthegenomethatplayagenelikeroleneednotbephysicallydistinct:theycanoverlaponeanotheroroccurinsideoneanother(inthesamedirectionontheDNAmoleculeorinreverse).Therelationshipbetweenstructuralgenesandgenelikefunctionsisnotonetoonebutmanytomany:somegeneproductsaremadefrommorethanonestructuralgeneandindividualstruc-turalgenesmakemultipleproducts.Finally,thesequenceofele-mentsinthegeneproductdependsonmuchmorethanthesequenceofnucleotidesinthestructuralgene:differentsequenceelementscanberepeated,scrambled,andreversedintheproduct,andtheprecisesequenceofageneproductcanreflectposttranscriptionalandtranslationalprocessingaswellastheoriginalDNAsequence.Toputfleshontheseboneswewillbrieflydescribesomeofthesemechanismsandgiveanexample(Figure5.1).(Formoreexamples,details,andreferences,seeStotzandGriffiths2004;Stotz,Bostanci,andGriffiths2006;Stotz2006.)Ineukaryotes(organismswhosecellshaveanucleusandorgan-elles,includingfungi,plants,andanimals)theDNAsequenceistranscribedintoapremessengerRNA(pre-mRNA)fromwhichthefinalRNAtranscriptisprocessedbycuttingoutlargenoncodingsequences,calledintrons,andsplicingtogethertheremainingcod-1ingsequences,theexons.Biologistsspeakofalternativecis-splicingwhenmorethanonematuremRNAtranscriptresultsfromtheseprocessesthroughthecuttingandjoiningofalternativeexons.Adjacentgenesaresometimescotranscribed,thatis,transcribedtogethertoproduceasinglepre-mRNAthatisthenspliced.SplicingmayalsooccurbetweenageneandanadjacentpseudogenethatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGene95pre-mRNA2bcmaturemRNAabcdepre-mRNA1apre-mRNA3dpre-mRNA4mat-reFigure5.1Acontemporarymoleculargene.Linesdenoteintrons;boxesdenoteexons.Subunit1oftherespiratorychainNADHdehydrogenaseisencodedbythegenenad1,whichinthemitochondrialgenomesoffloweringplantsisfragmentedintofivecodingsegmentsthatarescatteredoveratleast40kbofDNAsequenceandinterspersedwithotherunrelatedcodingsequences.Inwheat(illustrated)thefiveexonsthattogetherencodethepolypeptideof325aminoacidsrequireonecis-splicingevent(betweentheexonsb/c)andthreetrans-splicingevents(betweenexonsa/b,c/d,andd/e)forassemblyoftheopenreadingframe.Inaddition,RNAeditingisrequired,includingaCtoUsubstitutiontocreatetheinitiationcodonforthisORF.InsomemossesandinmammalstheORFforNAD1isanuninterruptedstretchofnucleargenomicDNA.Finally,inwheat,aseparateORFforamaturaseenzyme(mat-r)isencodedintheintronupstreamofexone(ChapdelaineandBonen1991).Formoreexamples,visithttp://representinggenes.org.wouldbeincapableofproducingaproductonitsown.Alternativegeneproductsmayalsobederivedfromso-calledoverlappinggenes.Inthesecases,thegenes,inthesenseoftheopenreadingframes(ORFs)thataretranscribedintoRNA,arenotlineduplikesomanypearlsonastring,butinsteadmayoverlaponeanotherorevenbecompletelycontainedonewithinanother.Whilesomecasesofalternativesplicingproducearangeofproteinsthatarestructurallyrelatedtooneanother,inothercasestheproductsarequitedifferentfromeachother(inwhichcasetheyareoftendescribedasproductsofoverlappinggenes,ratherthanalternativesplicingofthesamegene).Thedegreeofdifferencebetweentheproductsdependsontheextentofoverlapbetweentheirexons,andonwhetherthesesharedsequencesarereadinthesamereadingframe.ItistheprecisenucleotideatwhichreadingbeginsthatdetermineswhichcodonsaDNAsequencecontains.StartingatadifferentnucleotideiscalledCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n96paule.griffithsandkarolastotzframeshift,aphenomenonthatwouldlooklikethisinanEnglishsentence:AgeneisaflexibleentitybecomesAgeneisaflexibleentity.Butunlikeanyhumanlanguage,aDNAsequenceisalwaysmadeupofmeaningfulthree-letterwords(codonsthatspecifyanaminoacidduringtranslation)nomatterwherereadingbegins.Thismeansthatverydifferentproductscanbereadfromthesamesequencemerelybyframeshiftingbyonenucleotide.AswellasalternativetranscriptsfromaDNAsequence,multiplesimulta-neoustranscriptscanoccur,asisthecaseoftheparallelprocessingoffunctionalnoncodingRNAs(suchasmicroRNAs)fromtheintronicregionsoftheprematuretranscript,whichmaybeinvolvedintheregulationofcodingtranscriptofthesamegene.Intheprocessoftrans-splicingafinalmRNAtranscriptispro-cessedfromtwoormoreindependentlytranscribedpre-mRNAs.Whilsttheprefixtransmightsuggestthatthesepre-mRNAsarederivedfromDNAsequencesfarapartfromeachother,thisisbynomeansalwaysthecase.Infact,twocopiesoftheverysamesequencecanbesplicedtogetherthisway,ascanalternativeexonsinwhatwouldatfirstglancelooklikeanormalcaseofcis-splic-ing.Moreover,untilveryrecentlyitwasthoughtthatonlyonestrandofDNAistranscribed,butinfactDNAcanbereadbothforwardsandbackwardsbythecellularmachinery,producingeitherdifferentormatching(complementary)products.Thelattercase,inwhichexactlythesamesequenceisreadinreverse,willresultinanantisensetranscriptwithlikelyregulatoryfunction,possiblythroughsilencingitscomplementarytranscript.RNAeditingisanothermechanismofmodificationthatcansignificantlydiversifythetranscriptomeorproteome(thetotalcomplementoffinaltranscriptsorproteinsinthecellsofanorganism).WhereasmostotherformsofposttranscriptionalmodificationsofmRNA(cap-ping,polyadenylation,andcis-splicing)retainthecorrespondenceoftheprimarystructureofcodingsequenceandgeneproduct,RNAeditingdisturbsthiscorrespondencebychangingtheprimarysequenceofmRNAafteritstranscription.Thecreationofcrypto-genesviaRNAeditingcanpotentiallyhaveradicaleffectsonthefinalproduct,dependingonwhethereditingchangesthesenseofthecodoninwhichitoccurs.WhiletherearelikelyasmanyvarietiesofRNAeditingasthereareorganisms,allbelongtoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGene97oneofthreeknownmechanisms:thesite-specificinsertionordeletionofoneorseveralnucleotides,ornucleotidesubstitution(cytidine-to-uridineandadenosine-to-inosinedeamination,uridine-to-cytidinetransamination).Althoughwewillnotdescribethemhere,otherprocessesmayoccurbeforethefinalmRNAtranscriptistranslatedintoaproteinsequenceorprocessedintoafunctionalRNA.TherelationshipbetweenDNAandgeneproductisindirectandmediatedtoanextentthatwasneveranticipatedwhenthebasicmechanismsoftranscription,RNAprocessing,andtranslationwereclarifiedinthe1960s.themoderngeneThemoderngeneasPortin(1993)hastermeditrepresentsafurtherstageinthedialecticofstructureandfunctiondescribed.Theclas-sicalgene,primarilydefinedbythefunctionalroleitplayedinheredity,becameidentifiedwiththestructuralgenerevealedbyearlymolecularbiology,primarilythroughthestudyofprokaryotesandbacteriophage.Asaresult,thefunctionalroleofthegenewasredefinedasthedetermination,bylinearcorrespondence,ofthestructureofageneproduct.Furtherinvestigationofthemannerinwhichawiderrangeofgenomesgenerateawiderrangeofgeneproductshasrevealedthatthisfunctionalrolecanbefilledbydiverse,highlyflexiblemechanismsattheleveloftheDNAitself:Wearecurrentlyleftwitharatherabstract,openandgeneralizedconceptofthegene,eventhoughourcomprehensionofthestructureandorganizationofthegeneticmaterialhasgreatlyincreased(Portin1993,173).Goldschmidtscritiqueoftheparticulategenehasbeenexplicitlyrevivedinthelightofournewunderstandingofgenomestructureandfunction:Theparticulategenehasshapedthinkinginthebiologicalsciencesoverthepastcentury.Butattemptstotranslatesuchacomplexconceptintoadis-cretephysicalstructurewithclearlydefinedboundarieswerealwayslikelytobeproblematic,andnowseemdoomedtofailure.Instead,thegenehasbecomeaflexibleentitywithbordersthataredefinedbyacombinationofspatialorganizationandlocation,theabilitytorespondspecificallytoaparticularsetofcellularsignals,andtherelationshipbetweenexpressionpatternsandthefinalphenotypiceffect.(Dillon2003,457)CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n98paule.griffithsandkarolastotzInaprescientpapertwentyyearsago,RaphaelFalkreviewedwhatwerethennewlyemergingchallengestotheclassicalmoleculargeneandconcluded:Todaythegeneisnotjustthematerialunitortheinstrumentalunitofinheritance,butratheraunit,asegmentthatcorrespondstoaunit-function,asdefinedbytheindividualexperimentalistsneed.Itisneitherdiscretethereareoverlappinggenes,norcontinuousthereareintronswithingenes,nordoesithaveaconstantlocationtherearetransposons,noraclearcutfunctiontherearepseudogenes,notevenconstantsequencesthereareconsensussequences,nordefiniteborderlinestherearevariablesequencesbothupstreamanddownstream.(Falk1986,169)Thus,asearlyas1986wewerewellonthewayfromthewelldefinedmaterialentitybacktoanabstraction,ahypotheticalcon-struct,ifnotaninterveningvariable,devisedbyscientistfortheirneeds(Falk1986,160).Focusingonthecuttingedgeofcontemporarygenomicscaninduceanextremelydeflationaryviewofthegene.Somemolecularbiologists,realizingthattheconceptsofgenetranscriptionorgeneexpressionmaynotsufficetocapturethevariationinexpressedgenomicsequences,haveproposedthemoregeneraltermgenometranscriptiontoallowfortheincorporationofRNAtranscriptsthatcontainsequencesoutsidetheborderofcanonicalgenes.Thisviewdoesnotsiteasilywiththeclassicalmolecularconceptionofgenes,whichfromthenewperspectiveseemlikestatisticalpeakswithinawiderpatternofgenomeexpression(FintaandZaphiropoulos2001).Onepragmatic,technologicalreasonthattodaysbiologistsarepreparedtoconsidersuchradicaloptionsisthatthechallengeofautomatedgeneannotationhasturnedtheapparentlysemanticissueofthedefinitionofgeneintoapressingandpracticaloneasthelimitationsofapurelystructural,sequenced-baseddefinitionofthegenehavebecomeapparent.Oneinfluentialrecentreviewcon-cludesthatonesolutionforannotatinggenesinsequencedge-nomesmaybetoreturntotheoriginaldefinitionofageneasequenceencodingafunctionalproductandusefunctionalgeno-micstoidentifythem(SnyderandGerstein2003,260).Thegeneconcept,however,playsaroleinmanyothercontextsbesidesthecuttingedgeofgenomics(Stotz,Griffiths,andKnight2004).Wesuggest,therefore,thatthereareatleastthreeanswerstoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGene99thequestionWhatisagene?,noneofwhichcanbeneglectedifwehopetodepictthestateofcontemporarybiologyaccurately.Thesearethetraditional,instrumentalgene;thepostgenomicmoleculargene;andthenominalgene.thetraditionalgeneBiologistscananddostillusegeneticanalysistheanalysisofthephenomenonofhereditybytheanalysisoftheresultsofhybridi-zation,eitherbetweenorganismsordirectlybetweenDNAmole-cules(Waters2004;Falkinpress).Geneticanalysisremainsakeytoolinaddressingbroaderbiologicalquestions.Forthesepurposesthegeneremainsaninterveningvariable,definedbytheinheritancepatternsthatitenablesustofollow,andthedifficultiesofprovidingaunivocalaccountofitsidentityasamaterialunitcanbeputtooneside.Thetraditionalgeneconceptisretainedinmuchthesamewayinpopulationgenetics.Inanimportantrecentanalysis,LennyD.MossintroducesthetermGene-Pforsomethingverylikeourtraditionalgene(Moss2003).ThePstandsvariouslyforpheno-type,prediction,andpreformationsincethesegenesareidenti-fiedintermsoftheirphenotypiceffects,areusedtopredictthephenotypicresultsofhybridization,andreflectwhatMosstermsinstrumentalpreformationismastrategicneglectofthewaysinwhichthegene-phenerelationshipdependsuponotherfactors.MosscontrastshisGene-PtoamaterialisticconceptofthegenethathecallsGene-D(fordevelopment).Genes-Daredefinedbytheirintrinsicchemicalcapacitytotemplateforgeneproducts.Here,wewishtodistinguishtwoimportantlydifferentwaystoconceptualizegenesthatfallwithinthegeneralareaofconceptualspacethatMoss2labelsGene-D.thepostgenomicmoleculargeneWeusethephrasepostgenomicmoleculargenetorefertotheentitiesthatcontinuetoplaythefunctionalroleofthemoleculargenemakinggeneproductsincontemporarymolecularbiology.Thepostgenomicmoleculargeneconceptembodiesthecontinuingprojectofunderstandinghowgenomestructuresupportsgenomefunction,butwithadeflationarypictureofthegeneasastructuralCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n100paule.griffithsandkarolastotzunit.Thesegenesarethingsyoucandowithyourgenome(Stotz,Bostanci,andGriffiths2006):althoughthegeneisstillanimageintheDNAofthetargetmolecule(themoleculewhoseactivitywewishtounderstand)thisimagemaybefragmentedordistortedtosuchanextentthatitcannotbediscerneduntilfunctionalgenomicshasrevealedhowthesesequenceelementsareusedinthebroadergenomicandcellularcontext.Thisconceptionofthegeneremainsacriticalaspectoftheepistemologyofmolecularbiosciencesimplybecauselinearcorrespondencebetweenmoleculesisfunda-mentaltobiologistsabilitytoidentifyandmanipulatethem,viatechnologiesrangingfromcDNAlibrariestomicroarraystoRNAinterference.Butalthoughitisimportanttoknowthegeneforsomemoleculeinthissense,itdoesnotmatterverymuchwhetherthatcollectionofsequenceelementsisagene!Toputitlesspara-doxically,theutilityofknowingtheDNAelementsthatunderlietheproductionofthetargetmoleculeoritsprecursorsdoesnotatalldependonwhetheritispossibletogiveaunivocaldefinitionofthematerialgene.Findingthegeneforthemoleculeinthissenseremainsimportantevenonthemostdeflationary,postgenomicviewofthemoleculargene.nominalgeneTheuseofdatabasescontainingnucleotidesequencesiswellestablished.Codifiedaspartofthisprocessisaparticularuseofgeneconceptsonthebasisofwhichonecanidentifyvariousgenesandcountthenumberofgenesinagivengenome....Icallgenes,pickedoutinthisway,nominalgenes.Agoodwayofparsingmyargumentisthatnominalgenesareausefuldeviceforensuringthatourdiscourseisanchoredinnucleotidesequences,butthatnominalgenesdonot,andprobablycannot,pickoutall,only,orexactlythegenesthatareintendedinmanyotherpartsofgeneticwork.(Burian2004b,6465)Itishardtodisagreethatformanypracticalpurposesgenesaresimplysequencesthathavebeenannotatedasgenesandwhoseannotationassuchhasbeenacceptedbythescientificcommunity.But,asBurianhimselfmakesclear,thisdoesnotimplythatthescientificcommunityhasaclearunderstandingofwhatmakesasequenceagenethatneedsonlytobemadeexplicit.ThomasFoglehasarguedpowerfullythatthisisnotthecase(Fogle2001).TheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGene101workingconceptofthegene,accordingtoFogle,issomethinglikeastereotypeorprototype:asequenceisageneifithasenoughsimi-laritiestoothergenes:forexample,itcontainsanopenreadingframe,hasoneormorepromoters,hasoneormoretranscriptsthatarenottoofunctionallydiversefromoneanother,andsoon.Thisismoreorlessadescriptionofautomatedgenediscoverymethods,andFoglessuggestionisthattheconceptofthegeneisnomoreprincipledordefinitionlikethanthis.Thevariousgenelikefeaturesarenotweightedagainstoneanotherinanyprincipled,theory-drivenway,butratherareweighteddifferentlyondifferentoccasionsinordertosegmenttheDNAsequenceintofairlytraditional-lookinggenes,sometimesgivinguponstructuralcriteriatosavefunctionalones(asintheexampleinFigure5.1),atothertimesgivinguponfunctionalcriteriatosavestructuralones(asincotranscriptionofageneandapseudogene).Fogleisquitecriticalofthisstateofaffairs,arguingthatcom-biningstructuralandfunctionalfeaturesintoasinglestereotype,whathecallstheconsensusgene,hidesboththediversityofDNAsequencesthatcanperformthesamefunctionandthediversefunctionsofparticularDNAsequences.Buriantakesamorepositiveview,emphasisingthevalueofsimplyhavingasharedcollectionofnamedsequencesknownorsuspectedtobeinvolvedinthepro-ductionofgeneproducts.conclusionThegenebeganlifeasaninterveningvariable,definedfunctionallyintermsoftheMendelianpatternofheredityinobservablepheno-typiccharacters.Itrapidlyacquiredasecondidentityasahypothet-icalmaterialunit.Aproductivedialecticbetweeninvestigationsofthegenethatidentifieditineachofthesetwowaysconcludedwiththeneo-classicalorclassicalmolecularconceptionofthegene.Thefunctionalroleofthegenewasredefinedtoexcludemutationandrecombination,whichbecamepropertiesoftheDNAinitsownright,ratherthanofindividualgenes.Thefunctionofthegenebecamethedeterminationofthestructureofgeneproductsvialinearcorrespondencebetweenmolecules.Thisfunctionalrolewasplayedbyanaturalclassofunitsatthemolecularlevelthestructurallydefinedmoleculargene.FurtherinvestigationofawiderCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n102paule.griffithsandkarolastotzrangeofgenomesandawiderrangeofgeneproductshasthrownintodoubtwhetheranadequatestructuraldefinitionispossiblethestructuralbasisuponwhichgeneproductsaregeneratedmaybeaverybroadclassofthingsyoucandowithyourgenome.Atthispointitremainspossibletothinkofgenesinthetraditionalmannerthatdatesbacktotheearlytwentiethcenturyasinterveningvari-ableinthegeneticanalysisofphenotypes.Itisalsopossibletothinkofthemastheoftencomplexcollectionsofsequenceelementsthatfillthefunctionalroleofthemoleculargene(postgenomicmolec-ulargenes).Finally,itispossibletothinkofgenesassimplythosesequenceswhosesimilarityonvariousdimensionstostereotypicalgeneshasledthemtobeannotatedasgenesandwhoseannotationassuchhasbeenacceptedbythescientificcommunity(nominalgenes).ACKNOWLEDGMENTSThismaterialisbaseduponworksupportedbytheNationalScienceFoun-dationunderGrants0217567and0323496andsupplementalfundingfromtheUniversityofPittsburgh.GriffithssworkonthischapterwassupportedbyanAustralianResearchCouncilFederationFellowship.Anyopinions,findings,andconclusionsorrecommendationsexpressedinthismaterialarethoseoftheauthorsanddonotnecessarilyreflecttheviewsoftheNSF.notes1.Incontemporaryusage,cis-elementsarethosetranscribedtogetheraspartsofasinglepre-mRNAwhereastrans-elementsaretranscribedseparatelyandunitedatsomestageofposttranscriptionalprocessing(trans-splicing).Thustrans-elementsinthemodernsense(trans-onmRNA)maybecis-locatedontheDNA.2.Moss(pers.comm.)suggeststhatourpostgenomicmolecularandnominalmaterialgenesareperspectivesongenes-Dcorrespondingtowhatare,somewhatperversely,calledforwardandreversegenetics.Thepostgenomicmoleculargeneembodiesthetraditional,forward,strategyoflocatingthetemplateresourcescorrespondingtoaknownphenotype.Thenominalgeneisatemplateresourcewhoseusewesetouttounderstand.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\npetergodfrey-smith6InformationinBiology1.introductionTheconceptofinformationhasacquiredastrikinglyprominentroleincontemporarybiology.Thistrendisespeciallymarkedwithingenetics,butithasalsobecomeimportantinotherareas,suchasevolutionarytheoryanddevelopmentalbiology,especiallywherethesefieldsborderongenetics.Themostdistinctivebiologicalroleforinformationalconcepts,andtheonethathasgeneratedthemostdiscussion,isinthedescriptionoftherelationsbetweengenesandthevariousstructuresandprocessesthatgenesplayaroleincaus-ing.Formanybiologists,thecausalroleofgenesshouldbeunder-stoodintermsoftheircarryinginformationabouttheirvariousproducts.Thatinformationmightrequirethecooperationofvariousenvironmentalfactorsbeforeitcanbe‘‘expressed,’’butthesamecanbesaidofotherkindsofmessage.Aninitialresponsemightbetothinkthatthismodeofdescriptionisentirelyanchoredinasetofwell-establishedfactsabouttheroleofDNAandRNAwithinproteinsynthesis,summarizedinthefamiliarchartrepresentingthe‘‘geneticcode,’’mappingDNAbasetripletstoaminoacids.However,informationalenthusiasminbiologypredatesevenarudimentaryunderstandingofthesemechanisms(Schrodinger1944).Andmoreimportantly,currentapplicationsofinformationalconceptsextendfarbeyondanythingthatcanreceiveanobviousjustificationintermsofthefamiliarfactsaboutthespecificationofproteinmoleculesbyDNA.Thisincludes(i)Thedescriptionofwhole-organismphenotypictraits(includingcomplexbehavioraltraits)asspecifiedorcodedforbyinformationcontainedinthegenes;103CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n104petergodfrey-smith(ii)Thetreatmentofmanycausalprocesseswithincells,andperhapsofthewhole-organismdevelopmentalsequence,intermsoftheexecutionofaprogramstoredinthegenes;(iii)Theideathatgenesthemselves,forthepurposeofevolutionarytheorizing,shouldbeseenas,insomesense,‘‘made’’ofinformation.Fromthispointofview,informa-tionbecomesafundamentalingredientinthebiologicalworld.Thereisnoconsensusabouttheproperformandstatusofthesekindsofdescription,andtheresulthasbeenthedevelopmentofafoundationaldiscussionwithinbothbiologyandthephilosophyofbiology.Somehavehailedtheemploymentofinformationalcon-ceptshereasacrucialadvance(Williams1992).Othershaveseenalmosteverybiologicalapplicationofinformationalconceptsasaseriouserror,onethatdistortsourunderstandingandcontributestolingeringgeneticdeterminism(Francis2003).Mostofthepossibleoptionsbetweentheseextremeviewshavealsobeendefended.Theseincludevariousargumentsthatsome,thoughnotall,ofthepopularusesofinformationalconceptsinbiologyarelegitimate(Godfrey-Smith2000,Griffiths2001).Theyalsoincludeargumentsthateventhemoretendentioususesoftheseconceptsarelegitimatesolongastheconceptsareappliedconsistently(Sterelny,Smith,andDickison1996,Jablonka2002).Otherphilosophersandbiologistsregardthewholematterasatempestinateacup;theydonotthinkthatthedevelopmentofaninformationallanguagefordescribinggenesmakesmuchofadifferencetoanything,asitisobviouslyaloosemetaphoricalusagethatcarriesnorealtheoreticalweight(Kitcher2001).Thephilosophicaldiscussionhasdevelopedfortworeasons.Oneisthegeneralphilosophicalinterestinabstractconceptualproblemsinparticularareasofscience–aninterestindebatesthatseemresistanttoempiricaladjudication,butdonotseemmerelytermi-nological.Sosomephilosophicalinteresthereisakintomorefamilarphilosophicalattentiontosuchbiologicalconceptsasfit-ness,species,andnaturalselection.Buttheconceptofinformationisnotmerelyanordinarytheoreticalconceptwithinaparticularpartofscience.Itisalsopartofafamilyofconceptsthathasbeenthefocusofintensestudyinseveralpartsofphilosophy,stretchingbackCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nInformationinBiology105forcenturies.‘‘Information’’itselfdoesnothavealonghistoryinphilosophy,butitiscloselyrelatedtoconceptsthatdo,suchastheconceptofmeaning,whichiscentraltophilosophyoflanguageandmuchphilosophyofmind.Sophilosophersarefamiliarwiththekindsofpuzzlesthataregeneratedbythisfamilyof‘‘semantic’’concepts.Itisnotthatphilosophershavedevelopedaconsensustheorythatcanbeapplied,inanoff-the-shelfway,tonewcases.Butphilosophersareintimatelyacquaintedwithmanyofthepuzzles,twistsandturns,redherrings,andtrade-offsthatariseinthisarea.Soasinformation(andrelatedconcepts)havebecomemorepromi-nentinbiology,somephilosophershavethoughtthatthisisanareawheretheyarequalifiedtohelpinthedevelopmentofusefulandcoherentbiologicalconcepts.Thischapterhastwomainsections.Thenextsectiongivesanoutlineofsomeoftheargumentsandoptionsdevelopedtodate.Thethirdsectionthendevelopssomemorenovelideas,whicharepresentedinacautiousandexploratoryway.Beforemovingtothesurveysection,therearetwootherpre-liminarypointstomake.First,thetopicofthisessayisnottheroleoftheconceptsofinformationandrepresentationinthepartsofbiologywheretheyaremostobviouslyrelevant;theessayisnotconcernedwithneuroscience,perception,languageprocessing,andsoon.Thetopicistheroleofinformation(anditsrelatives)inpartsofbiologywhereitsroleislessobvious,suchasthedescriptionofgenes,hormones,and(tosomeextent)signalingsystemsatthecellularlevel.Second,intheearlypartofthisdiscussionIwillnotputmuchemphasisonsomeofthefinerdistinctionsbetweentheconceptsofinformation,representation,meaning,coding,andsoon.Asthediscussionproceeds,distinctionsbetweenconceptswithinthisfamilywillbecomemoreimportant,butsomeofthesubtledistinctionswillbebackgroundedinitially.2.outlineofthedebateOnecommonwaytostartorganizingtheproblemistomakeadistinctionbetweentwosensesof‘‘information,’’ortwokindsofapplicationofinformationalconcepts.Oneoftheseisaweakorminimalsense,andtheotherisstrongerandmorecontroversial.Intheweakersense,informationalconnectionsbetweeneventsorCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n106petergodfrey-smithvariablesinvolvenomorethanordinarycorrelations(orperhapscorrelationsthatare‘‘nonaccidental’’insomephysicalsenseinvolvingcausationornaturallaws).ThissenseofinformationisassociatedwithClaudeShannon(1948),whoshowedhowtheconceptofinformationcouldbeusedtoquantifyfactsaboutcon-tingencyandcorrelationinausefulway,initiallyforcommunica-tiontechnology.ForShannon,anythingisasourceofinformationifithasanumberofalternativestatesthatmightberealizedonaparticularoccasion.Andanyothervariablecarriesinformationaboutthesourceifitsstateiscorrelatedwiththatofthesource.Thisisamatterofdegree;asignalcarriesmoreinformationaboutasourceifitsstateisabetterpredictorofthesource,lessinformationifitisaworsepredictor.Thiswayofthinkingaboutcontingencyandcorrelationhasturnedouttobeusefulinmanyareasoutsidetheoriginaltechno-logicalapplicationsthatShannonhadinmind,andgeneticsisoneexample.Thereareinterestingquestionsthatcanbeaskedaboutthissenseofinformation(Dretske1981),buttheinitiallyimportantpointisthatwhenabiologistintroducesinformationinthissensetoadescriptionofgeneactionorotherprocesses,sheisnotintroducingsomenewandspecialkindofrelationorproperty.Sheisjustadoptingaparticularquantitativeframeworkfordescribingordinarycorrelationsorcausalconnections.Consequently,philosophicaldiscussionshavesometimessetuptheissuebysayingthatthereisonekindof‘‘information’’appealedtoinbiology,Shannon’skind,thatisunproblematicanddoesnotrequiremuchphilosophicalattention.Theterm‘‘causal’’informa-tionissometimesusedtorefertothiskind,thoughthistermisnotideal.Whateveritiscalled,thiskindofinformationexistswheneverthereareordinarycontingencyandcorrelation.Sowecansaythatgenescontaininformationabouttheproteinstheymake,andalsothatgenescontaininformationaboutthewhole-organismpheno-type.Butwhenwesaythat,wearesayingnomorethanwhatwearesayingwhenwesaythatthereisaninformationalconnectionbetweensmokeandfire,orbetweentreeringsandatree’sage.Themorecontentiousquestionthenbecomeswhetherornotbiologyneedsanother,richerconceptofinformationaswellasShannon’sconcept.Informationinthisrichersenseissometimescalled‘‘semantic’’or‘‘intentional’’information.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nInformationinBiology107Whatisthedifferencebetweenthem,andwhymightwethinkthatbiologyneedstoemployaricherconcept?Thereisarangeofdifferencesbetweenthetwo.Firstandmostimportantly,informa-tionalconnectionsintheShannonsenseconnectenvironmentalconditionswithbiologicaltraitsinthesamewaythattheyconnectgenesandthosetraits.WithrespecttoShannoninformation,thereiswhatGriffithsandGraycalla‘‘parity’’betweentherolesofenvir-onmentalandgeneticcauses(GriffithsandGray1994,Griffiths2001).Inaddition,informationintheShannonsense‘‘flows’’inbothdirections,asitinvolvesnomorethanlearningaboutthestateofonevariablebyattendingtoanother.Sowecanreadoffsomethingaboutthephenotypefromthestateofthegenes,butwecanalsolearnsomethingaboutthegenesbyattendingtothephenotype.SometalkaboutinformationinbiologyisconsistentwiththesefeaturesofShannoninformation,butsomeisnot.Itisusuallythoughtthatatleastsomeapplicationsofinformationallanguagetogenesissupposedtoascribetogenesaspecialkindofcausalpropertythatisnotascribedtoenvironmentalconditions,evenwhentheyarecausallyimportant,andthatisalsounidirectional.Inaddition,amessagethatcarries‘‘semanticinformation,’’itisoftenthought,hasthecapacitytomisrepresent,aswellasaccuratelyrepresent,whatitisabout.Thereisacapacityforerror.Shannoninformationdoesnothavethatfeature;wecannotsaythatsomevariablecarriesfalseinformationaboutanother,ifweareusingtheoriginalShannonsenseoftheterm.Butbiologistsdoapparentlywanttouselanguageofthatkindwhentalkingaboutgenes.Genescarryamessagethatissupposedtobeexpressed,whetherornotitactuallyisexpressed.Oncewetaketheallegedsemanticpropertiesofgenesasseriouslyasthis,somesubtlequestionsarise.Ifgenesarecarryingamessageinthissense,themessageapparentlyhasaprescriptiveorimperativecontent,asopposedtoadescriptiveorindicativeone.Genescontaininstructions,notdescriptions.Their‘‘directionoffit’’totheireffectsissuchthatifgenesandphenotypedonotmatch,whatwehaveisacaseofunfulfilledinstructionsratherthaninaccuratedescriptions.SeveralphilosophersandbiologistshavearguedthatmuchinformationaltalkaboutgenesusesaricherconceptthanShan-non’s,butthisconceptcanbegivenanaturalisticanalysis.Itisnotalapsebackintounscientificteleologicalthinking.OnewaytoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n108petergodfrey-smithproceedistomakeuseofarichconceptofbiologicalfunction,inwhichthefunctionofanentityderivesfromahistoryofnaturalselection(Sterelny,Smith,andDickison1996,MaynardSmith2000,Sheaforthcoming).Thissortofmoveisfamiliarfromthephilosophyofmind,wheresimilarproblemsaroseintheexplanationofthesemanticpropertiesofmentalstates.Whenanentityhasbeensubjecttoandshapedbyahistoryofnaturalselection,thiscanprovidethegroundingforakindofpurposiveornormativedescriptionofthecausalcapacitiesofthatentity.Tousethestan-dardexample(Wright1976),thefunctionofaheartistopumpblood,nottomakethumpingsounds,becauseitistheformereffectthathasledheartstobefavoredbynaturalselection.Thehopeisthatasimilar‘‘teleofunctional’’strategymighthelpmakesenseofthesemanticpropertiesofgenes,andperhapsotherbiologicalstructureswithsemanticproperties.Thereareseveralwaysinwhichthedetailsofsuchanaccountmightbedeveloped(Godfrey-Smith1999),somefocusingontheevolvedfunctionsofthegeneticmachineryasawhole,andothersonthenaturalselectionofparticulargeneticelements.Allversionsofthisideaofferedsofarhaveproblemsofdetail.Oneproblemisthatthereisnooverallconnectionbetweenbiologicalfunctionandsemanticproperties;havingafunctionintherichhistoricalsenseisnotgenerallysufficientforhavingsemanticproperties.Legsareforwalking,buttheydonotrepresentwalking.Enzymesareforcata-lyzingreactions,buttheydonotinstructthisactivity.Therearethingsthatlegsandenzymesaresupposedtodo,butthisdoesnotmakethemintoinformationcarriers,inarichbeyond-Shannonsense.Whyshoulditdosoforgenes?Sterelny,Smith,andDickisonseemtothinkthereisaquiteintimateconnectionbetweenevolutionaryfunctionandsemanticpropertiesinthecaseofbiologicalstructuresthathavebeenselectedtoplayacausalroleindevelopmentalprocesses.Theyarguethatgenes,invirtueofthesefunctionalproperties,representtheout-comestheyaresupposedtoproduce.Theyadd,however,thatanynongeneticfactorsthathaveasimilardevelopmentalrole,andhavebeenselectedtoplaythatrole,alsohavesemanticproperties.SoSterelny,Smith,andDickisonwanttoascribeveryrichsemanticpropertiestogenes,butnotonlytogenes.Somenongeneticfactorshavethesamestatus.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nInformationinBiology109Proposalsthatappealtoevolutionarydesignto‘‘enrich’’theinformationalpropertiesofgeneshaveproblemsofdetail,buttheyalsohaveattractivefeatures.ItisstrikingthatJohnMaynardSmith,whenhegrappledwiththestatusofhisenthusiasmforinforma-tionalconceptsinbiology,optedforsomethingalongtheselines(2000).Theresultingoverallpicturehasgoodstructuralfeatures.Wewouldhavealoose,uncontroversialShannonsenseofinformationthatappliestoallsortsofcorrelations,andan‘‘overlay’’ofrichersemanticpropertiesincaseswherewehavetherightkindofhistoryofnaturalselection.Genesandahandfulofnongeneticfactorswouldhavetheseproperties;mostenvironmentalfeaturesthathaveacausalroleindevelopmentwouldnot.Theneatnessoftheresultingpictureprovides,forsomepeople,goodreasontoperseverewithsomeaccountalongtheselines.SofarinthissectionIhavemostlydiscussedtheconceptofinformation;therehasnotbeenmuchtalkof‘‘coding.’’Andtheideasdiscussedsofardonotputanyemphasisonthespecialfeaturesofgeneticmechanismsthemselves,suchasthecombinatorialstructureofthe‘‘geneticcode.’’Butsurelythesefeaturesofgeneticmecha-nismsprovidemuchoftheunderlyingmotivationfortheintroduc-tionofsemanticconceptsintobiology?Itmightseemso,butalotofdiscussionshaveineffecttreatedthisasanopenquestion.Asnotedpreviously,theenthusiasmforsemanticcharacterizationofbiologi-calstructuresextendsbackbeforethegeneticcodewasdiscovered.(SeeKay2000foradetailedhistoricaltreatment.)Butanotherlineofthoughtintheliterature,overlappingwiththeprecedingideas,hasfocusedonthespecialfeaturesofgeneticmechanisms,andontheideaof‘‘geneticcoding’’asacontingentfeatureofthesemechanisms.BothGodfrey-Smith(2000)andGriffiths(2001)havearguedthatthereisonehighlyrestricteduseofafairlyrichsemanticlanguagewithingeneticsthatisjustified.Thisistheideathatgenes‘‘codefor’’theaminoacidsequenceofproteinmolecules,invirtueofthepeculiarandcontingentfeaturesofthe‘‘transcriptionandtransla-tion’’mechanismsfoundwithincells.Genesspecifyaminoacidsequenceviaatemplatingprocess,whichinvolvesaregularmappingrulebetweentwoquitedifferentkindsofmolecules(nucleicacidbasesandaminoacids).Thismappingruleiscombinatorial,andapparentlyarbitrary(inasensethatishardtomakeprecise–Stegmann2004).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n110petergodfrey-smithTheargumentisthatthesefeaturesmakegeneexpressionintoacausalprocessthathassignificantanalogiestovariousparadigmaticsymbolicphenomena,suchastheuseofnaturallanguage.Somehavearguedthatthisanalogybecomesquestionableoncewemovefromthegeneticsofsimpleprokaryoticorganisms(bacteria),tothoseineukaryoticcells.ThishasbeenathemeofSarkar’swork(1996).Mainstreambiologytendstoregardthecomplicationsthatariseinthecaseofeukaryotesasmeredetails,whichdonotcom-promisethebasicpicturewehaveofhowgeneexpressionworks.Anexampleistheeditingand‘‘splicing’’ofmessengerRNA(mRNA)transcriptsintoaprocessedmRNAthatisusedintranslation.Thisisabiologicallyimportantprocess,anditdoesmaketheDNAamuchlessstraightforwardpredictorofaminoacidsequence,butitcanbearguedthatthisdoesnotmuchaffectthecrucialfeaturesofgeneexpressionmechanismsthatmotivatetheintroductionofasymbolicorsemanticmodeofdescription.SotheargumentinGodfrey-Smith(2000)andGriffiths(2001)isthatthereisonekindofinformationalorsemanticpropertythatgenesandonlygeneshave:codingfortheaminoacidsequencesofproteinmolecules.Butthisrelation‘‘reaches’’onlyasfarastheaminoacidsequence.Itdoesnotvindicatetheideathatgenescodeforwhole-organismphenotypes,letaloneprovideabasisforthewholesaleuseofinformationalorsemanticlanguageinbiology.Genescanhaveareliablecausalroleintheproductionofawhole-organismphenotype,ofcourse.Butifthiscausalrelationistobedescribedininformationalterms,thenitisamatterofordinaryShannoninformation,whichappliestoenvironmentalfactorsaswell.InthissectionIhavedistinguishedonelineofthoughtthatlooksatShannoninformationandits‘‘enriched’’relatives,andanotherlineofthoughtthatlooksatthepeculiarfeaturesofthemechanismsofgeneexpression,andtheoriginalnarrowideaofa‘‘geneticcode.’’Butthetwolinesofthoughtcanbemarriedinvariousways.May-nardSmith,inresponsetoproblemswithhisteleofunctionalaccount,appealedatonepointtosomespecialfeaturesofgeneticmechanisms,includingtheapparent‘‘arbitrariness’’ofthegeneticcode.Thisideahasbeenpopular,butishardtomakeprecise.Thekeyproblemisthatanycausalrelationcanlook‘‘arbitrary’’ifitoperatesviamanyinterveninglinks.Thereisnothing‘‘arbitrary’’CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nInformationinBiology111abouttheproximalmechanismsbywhichamolecularbindingeventoccurs.Whatmakesthegeneticcodeseem‘‘arbitrary’’isthefactthatthemappingbetweenbasetripletsandaminoacidsismediatedbycontingentfeaturesofthesequencesoftransferRNA(tRNA)molecules,andoftheenzymesthatbindaminoacidstotRNAmolecules.Becauseweoftenfocusonthe‘‘long-distance’’connec-tionbetweenDNAandproteinandpaylessattentiontotheinter-veningmechanisms,thecausalrelationappearsarbitrary.Ifwepickedoutandfocusedonstepsinanyotherbiologicalcascadethatareseparatedbythreeorfourinterveninglinks,thecausalrelationwouldlookjustas‘‘arbitrary.’’Hereitisalsosignificantthatthestandardgeneticcodeisturningouttohavemoresystematicandnonaccidentalstructurethanpeoplehadoncesupposed(Knight,Freeland,andLandweber1999).Iwilldiscussthreemoretopics,inamoreself-containedway,tofinishthesurvey.Thefirstistheideathatgenescontainaprogram,inasenseanalogoustothatincomputerscience(Mayr1961,Moss1992,Marcus2004).Thisideahasnotbeendiscussedinsuchaconcertedwaybyphilosophers,thoughitisseenconstantlyinbio-logicaldiscussion.Herethefocusismoreonthecontrolofprocessesbygenes,asopposedtothespecificationofaparticularproduct.The‘‘program’’conceptseemstobeappliedinbiologyinanespeciallybroadandunconstrainedway,oftenguidedonlybyveryvagueanalogieswithcomputersandtheirworkings.First,wemightisolateaverybroadusage,inwhichtalkofprogrammingseemsmerelyaimedatreferringtotheintricatebutorderlyandwell-coordinatednatureofmanybasicprocessesinbiologicalsystems.Here,themostthattalkof‘‘programs’’couldbedoingisindicatingtheroleofevolutionarydesign.Anexamplemightbetalkof‘‘pro-grammedcelldeath’’inneuroscience,whichisaveryimportantprocesswithinneuraldevelopmentthatcouldjustasaccuratelybedescribedas‘‘orderlyandadaptivecell-deathinaccordancewithevolutionarydesign.’’Second,however,wemightisolateasenseinwhichtalkof‘‘programs’’inbiologyisdrivenbyacloseanalogybetweensomebiologicalprocessandthelow-leveloperationofmoderncomputers.Onecrucialkindofcausalprocesswithincellsiscascadesofup-anddown-regulationingeneticnetworks.Onegenewillmakeaproductthatbindstoandhencedown-regulatesanothergene,whichisthenCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n112petergodfrey-smithpreventedfrommakingaproductthatup-regulatesanother–andsoon.WhatwehavehereisacascadeofeventsthatcansometimesbedescribedintermsofBooleanrelationshipsbetweenvariables.Oneeventmightfollowonlyfromtheconjunctionofanothertwo,orfromadisjunctionofthem.Down-regulationisakindofnegation,andtherecanbedoubleandtriplenegationsinanetwork.Generegulationnetworkshavearichenoughstructureofthiskindforittomakesensetothinkofthemasengagedinakindofcomputation.Computerchip‘‘and-gates,’’neural‘‘and-gates,’’andgenetic‘‘and-gates’’havesomegenuinesimilarities.Mostotherbiologicalpro-cesses,thoughjustasmuchtheproductofevolutionarydesign,donothaveastructurethatmotivatesthissortofcomputationaldescription.Andonceagainwefind,asinthecaseof‘‘geneticcod-ing,’’thatthedomaininwhichthiscomputationallanguageiswellmotivated,whenappliedtogenes,isconfinedtothecellularlevel.Lesselaboratecascadesofthiskindcanalsobefoundintheendo-crinological(hormone-using)systemswithinthebody.Here,too,informationallanguagecanseemnaturallyapplicableandmaybejustifiedbyasimilarlineofargument.ThesecondofthethreetopicsIwilldiscusstofinishthissectionisthelinkbetweeninformationaldescriptionandgeneticdeter-minism.Anumberofcriticshavearguedthattheinformationalorsemanticperspectiveongeneactionfostersorencouragesnaiveideasaboutgeneticdeterminism(Oyama1985,Griffiths2001).Othersthinkthatgeneticdeterminism,whenitisfalse,isanordinaryerroraboutcausalrelationsthathasnoparticularlinktotheinformationaldescriptionofthoserelations.Isidewiththecriticsherewhosaythatthereissomethingdefiniteaboutinfor-mationaldescriptionofgenesthatencouragesfallaciesaboutgeneticcausation.ThekeypointhasbeensummarizedbyGriffiths.Henotesthatincomplexsystems,almostallcausalfactorsarecontextdependent,andusuallyitisnothardtorememberthis.Ifwethinkinordinarycausalterms,itisstraightforwardtonotethatageneticcausewillhaveitsnormaleffectsonlyifaccompaniedbysuitableenvironmentalconditions,andanenvironmentalcausewillhaveitsnormaleffectonlyifaccompaniedbysuitablegeneticconditions.(Ifthesensitivityoneithersideishigh,thentalkof‘‘normal’’effectsitselfmaybemisleading.)But,Griffithssuggests,theinformationalmodeofdescribinggenes(andotherfactors)fosterstheappearenceofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nInformationinBiology113contextindependence.‘‘Geneticcausationisinterpreteddetermi-nisticallybecausegenesarethoughttobeaspecialkindofcause.Genesareinstructions–theyprovideinformation–whilstothercausalfactorsaremerelymaterial....Agaygeneisaninstructiontobegayevenwhen[becauseofotherfactors]thepersonisstraight’’(2001,395–96).Sotheideaisthattheinferentialhabitsandasso-ciationsthattendtogoalongwiththeuseofinformationalorsemanticconceptsleadustothinkofgenesashavinganadditionalandsubtlekindofextracausalspecificity.Thesehabitscanhaveaneffectevenwhenpeoplearewillingovertlytoacceptcontextdependenceof(most)causesincomplexbiologicalsystems.Relat-edly,theideaofinternalgeneticmessagesmayalsofosteratendencytowardakindofessentialistthinking;themeaningoftheinternalmessagetellsuswhatthe‘‘truenature’’oftheorganismis,regard-lessofwhetherthisnatureisactuallymanifested.Myfinaltopicinthissurveyisthemoststrongandtendentiousemploymentofinformationallanguageforgenes,whicharisesinthecontextofevolutionarybiology.Ithasbeencommonforsometimetosaythat,intheevolutionarycontext,weshouldthinkofageneintermsofitssequence,whichispreservedovermanyreplicationevents,andnotintermsofparticularDNAmolecules,whichcomeandgo(Dawkins1976,1986).Theideathatsequencecanbepre-servedacrosschangesintheunderlyingmoleculesiscertainlyrea-sonableandimportant.Butthismessage,importantasitis,hasbeenexpressedinextremeandphilosophicallymysteriouswaysbysometheorists.G.C.Williams(1992),forexample,holdsthatbecauseofthesefactsaboutthepreservationofgenesequenceacrosschangesinmolecules,weshouldthinkofinformationasakindoffundamentalingredientoftheuniverse,alongwithmassandenergy,thatexistsinitsown‘‘domain.’’Thismakesthecausalconnectionsbetweentheinformationaldomainandtheordinaryphysicaldomainquitemysterious,andWilliamshimselffindsthisanimportantproblem.Buttheappearenceofaproblemarisesonlyfromanunnecessaryreificationofinformation.Wecaninsteadsaythatwhathasbeenlearnedfromworkontheevolutionaryfeaturesofgenesisthatvariousdifferentphysicalobjectscansharetheirinformationalproperties.Theseinformationalpropertiesareexplicableintermsofthelower-levelphysicalpropertiesoftheobjects,andthecontextsinwhichtheobjectsareembedded.SuchaviewdoesraisesomefurtherCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n114petergodfrey-smithquestions,butitdoesnotintroducetheideaofinformationasaseparate‘‘stuff’’whoserelationstoordinaryphysicalthingsaretenuousandproblematic.Theenthusiasmforareifiedtreatmentofinformationcanleadtoothertheoreticalproblemsinbiology.Someoftherecentadvocatesof‘‘intelligentdesign’’creationismhavetriedtousethespecialandmysteriouspropertiesofinformationtomountanti-Darwinianarguments(Dembski1997,criticizedinGodfrey-Smith2001).Theseargumentshavenorealforce.Indeed,theresultingviewstendtobelessplausiblethanearlierversionsoftheargumentfromdesign,becauseevenroutineandlow-levelformsofevolutionbynaturalselection,suchastheevolutionofdrugresistanceinbacteria,tendtoberuledoutasimpossibleinprinciple.Buttheinformationalter-minologyinwhichtheargumentsareexpressedlendsthemaspuriousappearenceofrigor.3.thenextsteps?InthissectionIwillcautiouslyintroducesomeideasthatapproachthewholeproblemsomewhatdifferently.Iwillmotivatethechangeintackbyaskingwhatappearstobeanoddquestion.Istheinfor-mationalorsemanticdescriptionofgenesmetaphorical,ornot?Thisshouldbeaneasyquestiontoanswer,butinfactseemstobesurroundedbyuncertainty.Ontheonehand,biologistssometimessaythattheintroductionofaninformationalframeworkwasacrucialtheoreticaladvance.Thissuggeststhatitisnotatallametaphor.Ifelectricalcharge,andentropy,werecrucialtheoreticaladvancesintheirday,itwasnotbybeingmetaphors.Butifonepresseshardonwhattheseinformationalpropertiesaresupposedtobe,especiallyoncewegetbeyondthesimpleideaofacombinatorialmappingfromnucleicacidstoaminoacids,itiscommontoencounteraretreattotheideaofgeneticinformationasametaphor.Itisnotliterallytruethatgenesareprogrammingdevelopmentorrepresentingthewhole-organismphenotype,butthisisametaphorthathasprovedinvaluabletobiology.Ofcourse,wehavetoexpectsomevaguenesshere.Andwecannotexpectbiologiststobeexpertsontheanalysisofliteralandnonliterallanguage.Butwhatmakesthesituationoddisthefactthatifsomeonetriedcarefullytoadjudicatethisquestion,heorsheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nInformationinBiology115wouldrunimmediatelyintothefactthatinthecaseofascriptionsofsemanticproperties,thereisnoclearandwell-understoodborderbetweenliteralandmetaphorical.Thereisnotaclearandwell-demarcatedsenseofwhattheliteraldomainis,towhichmetapho-ricalcasesarebeingcompared.Thesameproblemarises,tosomeextent,incognitivescience,whichisoftenbasedontheideathatthemind/braincanbeseenasacomputer.Doesthismean,thereissuchathingascomputation,andthebrainliterallydoesit?Oristheideaofneuralcomputationsomethingmorelikeametaphor?Theabstracttheoryofcomputa-tion,withinmathematics,isnotespeciallyhelpfulforansweringthisquestion(Smith2002).Inthecaseofcomputationincognitivescience,thequestioncanbedeflectedinitiallybysayingthatcomputationisbeingtreatedasa‘‘model’’forthemind.Buttheterm‘‘model’’issoambiguousthatthisdoesnothelpmuch.Sometimes‘‘model’’meansaprovisionalandcautiouslydefendedtheory.Thisdoesnothelpherebecausecautionisnottheissue.Wewanttoknowwhetherinformationprocessing,computation,representation,andsoon,arerealnaturalkindsthatbrainactivity–andgeneticactivity–mightbeliteralinstancesof,orwhethersomeotherstoryabouttheroleoftheseconceptshastobetold.IntheremainderofthissectionIwillsketchonealternativestoryofthiskind.Itisdesignedtocontrastwiththesimplerideathatinformationalpropertiesaredefinitebutelusivepropertiesthatgeneseitherdoordonothave.Instead,informationaldescriptionofgenesismotivatedbyafamilyoffactors,whichIwillgroupintothreecategories.First,itismotivatedbysomerealanduncon-troversialfeaturesofgenesandDNAthemselves,whichwouldnotalonebesufficienttomotivateanelaborateinformationaldescrip-tion.Second,theuseofinformationalandsemanticlanguageintroducesintobiologyaparticular‘‘causalschematism,’’derivedfromeverydaycontextsinwhichsymbolsareused.Theschematismfunctionsasamodel,inasensediscussedinsomerecentphilosophyofscience.Third,theinformationalframeworkreflectsandreinforcesacommitmenttoawayofdemarcatingthescientificallyimportantfeaturesofgenesandassociatedmechanisms.Theframeworkfore-groundsonesetofpropertiesandbackgroundsanother,andtheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n116petergodfrey-smithpropertiesofgenesandothermoleculesthatarebeingforegroundedaresequenceproperties,asopposedtoalltheirotherchemicalproperties.Theresultofthisanalysisisanaccountoftheroleofinformationallanguageinbiologythatismorefocusedontheentiredisciplinaryroleoftheinformationalframeworkandlessonspecificinformationalpropertiesthatmightormightnotbereal.Iwillsaymoreabouteachofthethreecategoriesinturn.First,theinformationalframeworkismotivated,ofcourse,bysomerealanduncontroversialfeaturesofgenesandDNAthemselves.Someofthesewerehighlightedintheprevioussection,andtheyincludethecom-binatorialstructureandregularityofthemappingrulefromnucleicacidstoaminoacids.ButsomemotivationmayalsoarisefromafeatureofDNAthatisnotsooftenremarkedoninthiscontext.Thisisthepassivity,orcomparativeinertness,ofDNA.HerewefocusonsomefactsaboutwhatDNAdoesnotdo,aswellaswhatitdoes.TheevolutionofDNAasarepositoryofsequenceinformationisoftensaidtobedueinparttoitschemicalstability.OriginoflifeworkemphasizesthefactthatRNAisagoodinitialreplicatormoleculebecauseithassomeenzymaticactivity,butDNAismorestableonceproteinshavebeendevelopedforenzymaticwork.Andinmoderncells,DNAdoesnotdoverymuchinchemicalterms;almostallofitseffectsgoviaaparticularindirectcausalpathwaybywhichDNAsequenceistranscribedandtranslated.(ThemainexceptiontothisisDNA’sdirectinteractionwithtranscriptionfactors,ingeneregulation.)Proteins,asisalwaysnoted,domostoftheactualchemicalworkinthecell.DNAspecifiesaminoacidsequenceanddoesnotdomuchelse.SotocallDNAan‘‘informa-tional’’molecule,inamoderncontext,isoftenagesturetowardwhatitdoesnotdo,aswellastowhatitdoesdo.Mysuggestionforasecondsetofmotivationsismoretenden-tious.Itinvolvesageneralanalysisofwhenandwhypeopleintro-ducesemanticconcepts(includinginformation)intoscientificandotherexplanatorycontexts.Thesuggestionisthattheuseoftheseconceptsisgenerallyguided–notalwaysconsciously–bythepos-tulationofananalogybetweenaparticulareverydayformofsymboluseandthedomainthatthetheorististryingtounderstand.Thisanalogycanbeverypartial,whilestillexertinginfluenceonhowthephenomenaaredescribedandunderstood.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nInformationinBiology117Howdoestheanalogywork?Acentralaspectofeverydaysymboluseisthatoneobjectisusedto‘‘standfor’’another.Moreprecisely,apersonguidesbehaviordirectedononeobjectordomainbyattendingtothestateofanother.Thisistheschematiccoreofeverydaysymboluse,anditshowsupabstractlyinmanyphilosophicalanalysesofsemanticphenomena(e.g.,Millikan1984),aswellasinmodelsofsignalinggamesandtheevolutionofmeaning(Skyrms1996).Thisbasicpatternisalsoinstalledinthebasicpic-turethatShannonusedinhistheoryofinformation:wehaveasource,andasignalwhosestatecanbeconsultedtolearnsomethingaboutthesource.Acentralfeatureofthis‘‘causalschematism’’isthedistinctionbetweensomemechanismthatreadsorconsumesthesignalandthesignalitself.Inthegeneticcase,theideathatsemanticdescriptionisguidedbythismodelisquitehelpful.First,weseethatthebasiccell-levelmachineryoftranscriptionandtranslationis,infact,afairlygoodinstanceoftheschematicstructureinquestion.Theribosomal/tRNAmachineryis,ineffect,areaderorconsumerofnucleicacidsequence,withthefunctionofcreatingproteinproductsthatwillhaveavarietyofuseselsewhereinthecell.Wealsoseethatthisrealizationofthecausalschematismappliesonlyatthecelllevel,atthelevelatwhichthetranscriptionandtranslationapparatusshowsupasadefinitepartofthemachinery.Soweseewhyitistrue–ifitistrue,asIthinkitis–thattheuseofinformationalorsemanticlanguageinexplaininghowproteinmoleculesaremadeislegitimateandwellmotivated,whiletheuseofthislanguagewhentalkingabouttheroleofgenesinproducingwhole-organismphenotypesisnot.Oncewethinkintermsoftheinfluenceofanalogyandacausalschematismhere,wecanalsonoteaconnectiontothediscussionofthecomparativechemical‘‘passivity’’ofDNAdiscussedearlierinthissection.Paradigmaticcasesofmessagesineverydaylifeareratherphysicallypassive,too,havingtheirsignificanteffectsonlyviatheirinter-pretationbyareaderorconsumer.Thissecondcategoryoffactorsmotivatinginformationaldescriptionofgenesinvolvesakindofmodel-basedtheorizing,inasensethatwasdevelopedfortheanalysisofverydifferentpartsofscience(Giere1988,Godfrey-Smithforthcoming,WeisbergCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n118petergodfrey-smithforthcoming).Theterm‘‘model’’givesussomedefinitepurchasehereafterall.Mythirdcategoryinvolvesarolefortheinformationalframeworkthatisnotpartofacausalhypothesis,apositedmechanism,oranythingofthatkind.Instead,itinvolvesacommitmenttoawayofdemarcatingandcategorizinganentiredomain.Viatheinforma-tionalframework,acommitmentismadetotheimportanceofonesetofpropertiesandtheunimportanceofanother.Onesetofprop-ertiesofbiologicalmoleculesisforegrounded,byintroductionofalanguagethatcannaturallyaccomodatethem,whileanothersetofpropertiesisbackgrounded.Whatarebeingforegroundedaresequenceproperties,asobjectsofstudy,asopposedtoalltheotherchemicalpropertiesofgenesandassociatedbiologicalmolecules.ThesuggestionisthatratherthanattributingsomeparticularcausalpowerstoDNAsequence,theinformationalframeworkoftenfunctionstomakesequencesingeneraltheprimaryfocusofstudy.Whatresultsisaformofabstractionakintothatseeninstatisticalmechanics;thereisafocusonadistinctivelevelofdescriptionandaparticularsetofstatisticalfeaturesofinteractionsbetweenparticles,abstractingawayfromlotsofotherproperties(Griesemer2005).Theinformationalframeworkalsobringswithitasetofconceptualtoolsthataresuitedfortheanalysisofsequenceproperties,asopposedtootherchemicalproperties.However,itshouldbeaddedherethatthereareconspicuoususesofinformationallanguageinbiologyincontextswheresequencepropertiesarenottreatedascentral,forexampleinthedescriptionofhormonalsignaling.InthesecasesIwouldemphasizethesecondofthethreefactorsdiscussedinthissection,theroleofacausalschematismderivedfrompublicsymboluse.Supposetheactualpatternsofuseofinformationallanguageingeneticsareinfactguidedbyfactorslikethese,inacontext-sensi-tivemixture.Theuseoftheinformationalframeworkisguidedbysomerealfeaturesofgeneticmechanisms,butalsobyapplicationofschematiccausalmodelthatguidesmanyormostusesofasemanticlanguage.Itreflectsandreinforcesageneraldisciplinaryfocusonsequencepropertiesasopposedtoothers.Thiswouldsteerusawayfromtheideathatthereissomedefinitebuthiddensetofpropertiesbeingpositedbysuchlanguage,whichmightormightnotbereal.IwillleavesomewhatopenhowthissetofideasrelatestothemoreCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nInformationinBiology119standardlinesofthoughtoutlinedintheprevioussection.Insomeways,thetwocancomplementeachother.Inotherways,thereisprobablysometension.ACKNOWLEDGMENTSIamindebtedtoArnonLevyandNicholasSheaforcommentsanddiscus-sionoftheseissues.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nalexanderrosenberg7Reductionism(andAntireductionism)inBiologyAcceleratingdevelopmentsinmolecularbiologysince1953havestronglyencouragedtheadvocacyofreductionismbyanumberofimportantbiologists,includingCrick,Monod,andE.O.Wilson,andstrongoppositionbyequallyprominentbiologists,especiallyLewontin,alongwithmostphilosophersofbiology.Reductionismisametaphysicalthesis,aclaimaboutexplanations,andaresearchprogram.Themetaphysicalthesisthatreductionistsadvance(andantireductionistsaccept)isphysicalism,thethesisthatallfacts,includingthebiologicalfacts,arefixedbythephysicalandchemicalfacts;therearenononphysicalevents,states,orprocesses,andsobiologicalevents,states,andprocessesare‘‘nothingbut’’physicalones.Thismetaphysicalthesisisonereductionistssharewithantireductionists.Thereductionistarguesthatthemetaphysicalthesishasconsequencesforbiologicalexplanations:theyneedtobecompleted,corrected,mademoreprecise,orotherwisedeepenedbymorefundamentalexplanationsinmolecularbiology.Theanti-reductionistdeniesthisinference,arguingthatnonmolecularbio-logicalexplanationsareadequateandneednomacromolecularcorrection,completion,orgrounding.Theresearchprogramthatreductionistsclaimfollowsfromtheconclusionaboutexplanationscanbeframedasthemethodologicalmoralthatbiologistsshouldseeksuchmacromolecularexplanations.Antireductionistsarguethatsuchexplanationsareeitherorbothunnecessaryorunattainable.Reductionistsarguethattheirviewmustbedistinguishedfromeliminativism,thethesisthatmolecularbiologynotonlyprovidestheexplanans(whatdoestheexplaining),butalsodescribesallthe120CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)121factstobeexplained(theexplanantia).Antireductionistsholdthatreductionismmustinevitablycollapseintoeliminativism.1.reductionism–thepostpositivistpictureTherelationshipoftherestofbiologytomolecularbiologywasoriginallyenvisionedbyphilosophersofsciencetovindicateamodelofreductionassociatedwithphilosopherswhohadbeenpositivistsortheirstudents(whencethelabel‘postpositivist’).Inthelocusclassi-cusofreduction,ErnestNagel’sStructureofScience(1961),reductionisaformofintertheoreticalexplanationcharacterizedbythededuc-tivederivationofthelawsofthereducedtheoryfromthelawsofthereducingtheory.Thedeductivederivationrequiresthatthecon-cepts,categories,andexplanatoryproperties,ornaturalkindsofthereducedtheory,becapturedinthereducingtheory.Todoso,termsofthenarrowertheorymustbelinkedtoconceptsinthebroadertheory.AsexponentsofreductionsuchasSchaffner(1976)andRuse(1976)noted,themostdifficultandcreativepartofareductionisestab-lishingtheseconnections,formulatingbridgeprinciplesthatlinktheconceptsofthetwotheories.Thus,inparticular,toreduceMendeliantomoleculargeneticsrequiredtheNobelPrize–winningdiscoveriesofWatsonandCrickthatidentifiedthegenewithDNA.ItwasD.Hullwhofirstnoted(1976b)thedifficultyofactuallycharacterizingMendelianpropertiesbyconceptsdrawnonlyfrommolecularbiology.Mostsignificantly,asthebasicunitofpheno-typicexpression,mutation,andrecombination,thegenecouldnotbeidentifiedaseitheroneDNAbase(thesmallestunitofmutation),oronestretchofDNAthatconstitutesastructuralorregulatorygene,ortheamountofDNAsequenceminimallyinvolvedinrecombina-tion.Moreover,becauseoftheredundancyofthegeneticcode,manydifferentDNAsequencescouldcodeforthesamegene.TherelationbetweenmolecularDNAsequencesandgenesidentifiedbytheirfunctionwastherefore‘‘many-one’’and‘‘one-many’’:thesameDNAsequenceisimplicatedinmanydifferentgenesandmanydifferentsequencescandischargethefunctionofasinglegene.OfcourseinthesubsequentthirtyorfortyyearsmattersmadetherelationsbetweenDNAandgenesevenmorecomplicated:regulatorysequencesandsites,intronsandexons,posttranscriptionalandtranslationalmodification,promoters,operons,openreadingframes,junkDNA,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n122alexanderrosenbergrepeats,transposons,RNAviruses,allthesemadecompletelyintractabletheproblemofdefiningorcharacterizingeitherthegeneingeneral,orevenparticulargenesforparticulartraits,orforthatmatterevenparticularimmediategeneproductsintermsofthestructureofDNAmoleculesthatcomposethem.Thereisofcoursenotroubleidentifying‘tokens’–particularbitsofmatterwecanpointto–ofgeneswithparticular‘tokens’oftheirmolecularconstituents.Buttokenidentitieswillnotsufficeforreduction,eveniftheyareenoughforphysicalismtobetrue.Thesecondproblemfacingtraditionalreductionisminbiologywastheabsenceoflaws(beyondthoselaws,ifany,embodiedinthetheoryofnaturalselection),eitheratthelevelofthereducingtheoryorthereducedtheoryorbetweenthem.Indeed,agooddealofthephilosophyofbiology’ssearchforrealnomologicalgeneralizationsinbiologyhasbeenmotivatedbythecontroversyaboutreduction;bothitsproponentsandopponentsneededreallawsthattheycouldshowwereeitherreducibleorirreducibletoothermorefundamentallaws,thoseofphysicalscience.Inparticular,forWatsonandCrick’sachievementtovindicatetraditionalreductionism,therehadtobelawsofclassical,Mendelian,orpopulationgenetics,whichtheidentificationofthegeneasasequenceofDNAwouldenableustoderivefromlawsofmolecularbiology.ButMendel’slawsarenotlaws:theybegantoberiddledwithexceptionsalmostfromthemomenttheywerefirstrediscoveredintheearly1900s:cross-over,linkage,mitoticdrive,autosomalgenes,andsoforth.Andwhenprotectedfromexceptionsbyceterisparibusclauses,theyparticularlydefyderivationfrommorefundamentalprinciples.(Othercandidatesforthestatusofnomologicalgeneralizationsinbiology,e.g.,theHardy-Weinberg‘‘law’’orFischer’sfundamentaltheorem,turnouttobeeithertautologies,orconsequencesofotherequallycontroversialcandidatesfornomologicalstatus,suchastheprincipleofnaturalselection.)Moreover,therewerenodistinctivenomologicalgeneral-izationsofmolecularbiologyeither,justthestocheometricequationsoforganicchemistry.Asweshallsee,theabsenceoflawsbedevilsbothpostpositivistreductionismanditsdenial.Thattherearenolawsofbiologytobereducedtolawsofmolec-ularbiology,andindeedthattherearenolawsofmolecularbiology,canbeshownbythesameconsiderationsthatexplainwhygenesandDNAcannotsatisfyreduction’scriterionofconnection.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)123Theindividuationoftypesinbiologyisalmostalwaysviafunction:tocallsomethingawing,orafin,orageneistoidentifyitintermsofitsfunction.Butbiologicalfunctionsarenaturallyselectedeffects.Andnaturalselectionforadaptations–thatis,environmentallyappropriateeffects–isblindtodifferencesinphysicalstructurethathavethesameorroughlysimilareffects.Naturalselection‘‘chooses’’variantsbysomeoftheireffects,thosethatfortuitouslyenhancesurvivalandreproduction.Whennaturalselectionencouragesvar-iantstobecomepackagedtogetherintolargerunits,theadaptationsbecomefunctions.Accordingly,thestructuraldiversityofthetokensofagivenMendelianorclassicalorpopulationbiologicalgeneorgenerallyanyfunctionallyidentifiedbiologicalsystemorstructureisinevitable.Andnobiologicalkindwillbeidenticaltoanysinglemolecularstructureormanageablyfinitenumberofsetsofstructures.Toseewhytherecanbenolawsinbiology,noranythingthatcouldsatisfythecriterionofconnectabilitybetweenanitemfunc-tionallyandstructurallycharacterized,considertheformofageneralizationaboutallFs,whereFisafunctionalterm,likegene,orwing,orcamouflage.Thegeneralizationwilltaketheform(x)[Fx!Gx],alawaboutFsandGs.Gxwillitselfbeeitherastructuralpredicateorafunctionalone:thatis,eitheritwillpickoutGsbysomephysicalattributecommontothem,orGxwillpickoutGsbydescriptionsofoneoftheeffectsthateverythingintheextensionofGxpossesses.ButifFxisafunctionalkind,itwillhavebeenshapedbynaturalselection.Accordingly,therewillbenosinglephysicalfeaturecommontoallitemsintheextensionofFx.Itwillbeaphysicallyheterogeneousclasssinceitsmembershavebeenselectedfortheireffects.SoGxcannotbeastructuralpredicate.OfcoursesomestructuralfeaturemaybesharedbyallofthemembersofFx.Butitwillnotbeabiologicallyinterestingone.Ratheritwillbeapropertysharedwithmanyotherthings–likemassorelectricalresistance.ThesepropertieswillhavelittleornoexplanatoryrolewithrespecttothebehaviorofmembersoftheextensionofFx.Forexample,thegeneralizationthat‘‘allendothermsarecomposedofconfinedquarks’’doesrelateastructuralproperty–quarkconfinement–toafunctionalone–endothermy–andisexcep-tionlesslytrue.Butisnotalawofbiologicalinterest.Gxcannotbeastructuralkind.CanGxbeanotherfunctionalkind,likeFx,sharedbyallmembersofFx?TheexistenceofanotherfunctionalpropertyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n124alexanderrosenbergdifferentfromFthatallitemsintheextensionofthefunctionalpredicateFxshareishighlyimprobable.IfFxisafunctionalkind,thenbecauseoftheblindnessofselectiontostructurethemembersoftheextensionofFxarephysicallydiverse.Assuch,anytwoFshavenonidentical(andusuallyquitedifferent)setsofeffects.WithoutafurthereffectcommontoallFs,selectionforeffectscannotproduceanotherselectedeffect;itcannotuniformlyselectallmembersofFforsomefurtheradaptation.Thus,thereisnofurtherfunctionkindallFsshareincommon.Whetherfunctionalorstructural,therewillbenopredicateGxthatislinkedinalawtoFx.Ofcourse,thesortofconnectionreductionismrequiresbetweentheMendeliangeneandthemoleculargenewillhavetobeanidentity,Fx¼Gx,whichimplies(x)[Fx!Gx].So,ourconclusionprecludessatisfactionofeitherofthetworequirementsofpostpositivistreduction.Theunavoidableconclusionisthatasfarasthepostpositivistmodelofintertheoreticalreductionisconcerned,neitherofitsconditionsforreductionaresatisfiedbyrelationsbetweennon-molecularandmolecularbiology.2.antireductionismvindicated?Ifantireductionismweremerelythedenialthatpostpositivistreductionobtainsamongtheoriesinbiology,itwouldbeobviouslytrue.Butantireductionismisnotjustanegativeclaim.Itisthethesisthata)therearegeneralizationsattheleveloffunctionalbiology,b)thesegeneralizationsareexplanatory,c)therearenofurthergeneralizationsoutsidefunctionalbiologythatexplainthegeneral-izationsoffunctionalbiologyorexplainmorefullywhatthegeneralizationsoffunctionalbiologyexplain.Allthreecomponentsofantireductionismaredauntedbyatleastsomeofthesameproblemsthatvexreductionism:thelackoflawsinfunctionalbiologyandtheproblemsfacinganaccountofexpla-nationintermsofderivationfromlaws.Iftherearenolawsand/orexplanationisnotamatterofsubsumption,thenantireductionismisfalsetoo.Butbesidesthefalsepresuppositionsantireductionismmaysharewithreductionism,ithasdistinctproblemsofitsown.Indeed,theseproblemsstemfromtheverycoreoftheantireduc-tionistargument,theappealtoultimateexplanationsunderwrittenbythetheoryofnaturalselection.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)125Toseethedistinctiveproblemsforantireductionism,consideranexampleofputativelyirreduciblefunctionalexplanationadvancedbyantireductionists(duetoKitcher1984,1999).Theexampleisthebiologist’sexplanationofindependentassortmentoffunctionalgenes:Theexplanandumis(G)Genesondifferentchromosomes,orsufficientlyfarapartonthesamechromosome,assortindependently.AccordingtoKitcher,thefunctionalbiologistproffersanexpla-nansfor(G),whichweshallcall(PS):(PS)Considerthefollowingkindofprocess,aPS-process(forpairingandseparation).Therearesomebasicentitiesthatcomeinpairs.Foreachpair,thereisacorrespondencerelationbetweenthepartsofonememberofthepairandthepartsoftheothermember.Atthefirststageoftheprocess,theentitiesareplacedinanarena.Whiletheyareinthearena,theycanexchangesegments,sothatthepartsofonememberofapairarereplacedbythecorrespondingpartsoftheothermembers,andconversely.Afterexactlyoneroundofexchanges,oneandonlyonememberofeachpairisdrawnfromthearenaandplacedinthewinnersbox.InanyPS-process,thechancesthatsmallsegmentsthatbelongtomembersofdifferentpairsorthataresuffi-cientlyfarapartonmembersofthesamepairwillbefoundinthewinnersboxareindependentofoneanother.(G)holdsbecausethedistributionofchromosomestogamesatmeiosisisaPS-process.Kitcherwrites,‘‘ThisIsubmitisafullexplanationof(G),anexplanationthatprescindsentirelyfromthestuffthatgenesaremadeof’’(Kitcher,1999,199–200).Considerwhy,accordingtotheantireductionist,nomolecularexplanationof(PS)ispossible.Itisforthesamereasonnofunctionalbiologicalkindcanbeidentifiedwiththoseofmolecularbiology.Becausethesamefunctionalroleisalmostalwaysrealizedbyavarietyofstructures,andbecausenaturalselectionisblindtothisvariety,thefullmacromolecularexplanationfor(PS)orfor(G)willhavetoadverttoarangeofphysicalsystemsthatrealizeindepen-dentassortmentinmanydifferentways.Thesedifferentwayswillbeanunmanageabledisjunctionofalternativessogreatthatwewillnotbeabletorecognizewhattheyhaveincommon,ifindeedtheydohavesomethingincommonbeyondthefactthateachofthemwillgenerate(G).Eventhoughweallagreethat(G)obtainsinvirtueCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n126alexanderrosenbergonlyofmacromolecularfacts,nevertheless,becauseoftheirnumberandheterogeneity,thesefactswillnotexplain(PS),stilllesssup-plant(PS)’sexplanationof(G).Butthisargumentisunsatisfactory.Beginwith(G).Iftheargu-mentoftheprevioussectionisright,(G)isnotalawatall.Atmostitreportsparticularfactsaboutaspatiotemporallyrestrictedkind,‘‘chromosomes,’’ofwhichthereareonlyafinitenumberextantoveralimitedperiodatonespatiotemporalregion(theEarth).Accord-ingly,(G)isnotalawandsonotacandidateforreduction,andtheimpossibilityofreducingittomolecularbiologyisnoobjectionagainstreductionism.Moreimportant,itisunclearwhatcertifies(PS)–theaccountofPS-processesgiven–asexplanatory,andwhatpreventsthevastdisjunctionofmacromolecularaccountsoftheunderlyingmechanismofmeiosisfromexplaining(PS),orforthatmatterfromexplaining(G)(cf.Sober1999).Thereductionistholdsthat,contrarytoKitcher’sclaim,thereissomethingthatthevastdisjunctionofmacromolecularrealizationsof(PS)haveincommonthatwouldenabletheconjunctionofthemtoexplain(PS)fullytosomeonewithagoodenoughmemoryforthedetails.EachwasselectedforbecauseeachimplementsaPSprocessandPSprocesseshavebeenadaptiveinthelocalenvironmentoftheEarthfromabouttheonsetofthesexuallyreproducingspeciesonward.SinceselectionforimplementingPSprocessesisblindtodifferencesinmacromolecularstructureswiththesameorsimilareffects,theremayturnouttobenothingelsecompletelycommonandpeculiartoallmacromolecularimplementationsofmeiosisbesidestheirbeingselectedforimplementingPSprocesses.Ofcourse,itmaybethatweneverknowenoughofthe‘‘gorydetails’’fora(disjunctive)macromolecularanswertothequestionofwhy(G)obtains.Similarly,wemayneverknowenoughforamacro-molecularexplanationof(PS)tobeacompleteanswertoourques-tion‘‘WhydoPSprocessesoccur?’’Butthiswouldbehollowvictoryforantireductionism,evenifwegrantthetendentiousclaimthatwewillneverknowenoughforsuchexplanationstosucceed,foritrelegatesantireductionismtothestatusofaclaimaboutbiologists,notaboutbiology.Suchphilosophicallimitationsonourepistemicpowershavebeenrepeatedlybreechedinthehistoryofscience.Antireductionistswillneedadifferentargumentfortheclaimthatneither(G)nor(PS)canbeexplainedbythedisjunctionofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)127macromolecularmechanismsthatrealizeit,andfortheclaimthat(PS)doesexplain(G)and(G)doesexplainindividualcasesofrecombination.Oneargumentforsuchaconclusionrestsonthemetaphysicalthesisthattherearenodisjunctivepropertiesorthatifthereare,suchpropertieshavenocausalpowers.Hereishowtheargumentmightproceed:thevastmotleyofalternativemacro-molecularmechanismsthatrealize(PS)havenothingincommon.Thereisnoproperty–andinparticularnopropertywiththecausalpowertobringaboutthetruthof(G)–thattheyhaveincommon.Physicalism(whichallantireductionistspartytothisdebateembrace)assuresusthatwheneverPSobtains,somephysicalpro-cessorotherobtains.Thuswecanconstructtheidentity(oratleastthebiconditional)that_____ðRÞPS¼P1;P2:::Pi;:::Pm;wheremisthenumber,averylargenumber,ofallthewaysmac-romolecularprocessescanrealizePSprocesses.Antireductionismcannotarguethat(R)hasnoexplanatorypower,becauseitismerelyalocalfactabouthowPSprocessesarerealizedonEarth.Forthisisalsotrueof(G)andantireductionistsinsistthat(G)hasexplanatorypower.Acausaltheoryofexplan-ationmightruleoutRasexplainingPSonthegroundthattheWWWWWdisjunction,P1;P2:::Pi;:::Pm;isnotthefullcauseofPSprocesses.Thismightbeeitherbecauseitwasincomplete–thereisalwaysthepossibilitythatstillanothermacromolecularrealizationofPSwillarise–orbecausedisjunctivepropertiesjustarenotcauses,havenocausalpowers,perhapsarenotreallypropertiesatall.Thus,theantireductionistalleges,(PS)and(G)arethebestandmostcompleteexplanationstowhichbiologycanaspire.Considertheclaimthat(R)isnotcomplete,eitherbecausesomedisjunctshavenotoccurredyetorperhapsthatthereareanindefi-nitenumberofpossiblemacromolecularimplementationsfor(SP).Thisinfactseemstobetrue,justinvirtueofthefactthatnaturalselectioniscontinuallysearchingthespaceofalternativeadapta-tionsandcounteradaptations,andthatbiologicalthreatstotheintegrityandeffectivenessofmeiosismightinthefutureresultinselectionfornewmacromolecularimplementationsof(PS).Butthisisnoconcessiontoantireductionism.ItismorelikeanargumentCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n128alexanderrosenbergthatneither(PS)nor(G)reportsanexplanatorygeneralization,thattheyareinfacttemporarilytrueclaimsaboutlocalconditionsontheEarth,andsonotcandidatesforpostpositivistreduction.Supposethat(R)canbecompletedinprinciple,perhapsbecausethereareonlyafinitenumberofwaysofrealizinga(PS)processbut,itisclaimed,thedisjunctionisnotacausalorarealpropertyatall.Thereforeitcannotfigureinanexplanationofeither(PS)or(G).Thereareseveralproblemswithsuchanargument.First,thedis-WWWWWjunctsinthedisjunctionofP1;P2:::Pi;:::Pmdoseemtohaveatleastoneorperhapseventworelevantpropertiesincom-mon:eachwasselectedforimplementing(PS)andcausallybringsaboutthetruthof(G).Second,weneedtodistinguishpredicatesinlanguagesfrompropertiesinobjects.Itmightwellbethatinthelanguageemployedtoexpressbiologicaltheory,theonlypredicateweemploythatistrueofeveryPiisadisjunctiveone,butitdoesnotfollowthatthepropertypickedoutbythedisjunctivepredicateisadisjunctiveproperty.Philosophylongagolearnedtodistinguishthingsfromthetermswehitupontodescribethem.Howmighttheantireductionistargueagainstthecausalefficacyofdisjunctiveproperties?Onemightholdthatdisjunctivepropertieswillbecausallyefficaciousonlywhentheirdisjunctssubsumesimilarsortsofpossiblecausalprocesses.Ifweadoptthisprinciple,thequestionatissuebecomesoneofwhetherthedisjunctionWWWWWofP1;P2:::Pi;:::Pmsubsumessimilarsortsofcausalprocesses.Theanswertothisquestionseemstobethatthedisjunctionsharesincommonthefeaturesofhavingbeenselectedforresultinginthesameoutcome–PSprocesses.Thus,thedisjunctivepredicatenamesacausalproperty,anaturalkind.Ifso,antireduc-tionistswillbehardpressedtodenythetruthandtheexplanatorypowerof(R).Besidesitsproblemsinunderminingputativemacromolecularexplanationsof(PS),(G),andwhat(G)explains,antireductionismfacessomeproblemsinsubstantiatingitsinitialclaimsthat(PS)explains(G)and(G)explainsindividualcasesofgeneticrecombi-nation.Theproblemsofcoursestemfromthefactthatneither(PS)nor(G)isalaw,andthereforeanaccountisowedofhownon-lawlikestatementssuchasthesecanexplain.Wecanconcludethatsofaraspostpositivistreductionismanditsantireductionistoppositionareconcerned,neitherviewisrelevantCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)129totherealissueabouttherelationbetweenfunctionalandmolec-ularbiology.Ifthereisarealdisputehere,itcannotbeaboutthederivabilityorunderivabilityoflawsinfunctionalbiologyfromlawsinmolecularbiology,astherearenolawsineithersubdiscipline.Norcantherealdisputeturnontherelationshipbetweengeneraltheoriesinmolecularandfunctionalbiology.Oncethisconclusionisclear,thequestionofwhatreductionismwasinthepostpositivistpastcanbereplacedbythequestionofwhatreductionismisnow;fortheobsolescenceofthepostpositivistmodelofreductionhardlymakesthequestionofreductionismoritsdenialobsolete.Indeed,developmentsinmolecularbiology–therateofsequencing,thesuccessofcomputationalin-silicogenomics,andespeciallytheelucidationofthemolecularmechanismofdevelopment–makethisquestionmorepressingthanever.Butthequestionhastobereformulatedifitistomakecontactwithrealissuesinbiology.3.reductionismafterpostpositivismThedebatebetweenreductionismandantireductionismcannotbeoneaboutlawsastheyhavetraditionallybeenunderstoodinphilosophyofscience.Thedisputeaboutwhethertherestofbiologyisreducibletomolecularbiologywillhavetobeoneaboutthebest,mostcomplete,mostcorrect,ormostadequateexplanationofpar-ticularfactsaboutlifeonEarth,someofthesefactsobtainingovergeologicalepochs,butallofthemultimatelythecontingentresultsofgenerallawsofnaturalselectionoperatingontheboundarycon-ditionsthatobtainedhere3.5billionyearsago–thatis,historicalfacts.Reductionismneedstoclaimthatthemostcomplete,correct,andadequateexplanationsofthehistoricalfactsuncoveredinfunctionalbiologyarebyappealtootherhistoricalfactsuncoveredinmolecularbiology,plusperhapssomelawsoforganicchemistry.Antireductionismmustclaimthatthereareatleastsomeexpla-nationsinfunctionalbiologythatcannotbecompleted,corrected,orotherwiseimprovedbyadducingconsiderationsfrommolecularbiology.Theonlywaytodothisistoarguethatexplanationsinfunctionalbiologyareallultimatelyadaptational,asaresultofthefunctionalvocabularyinwhichtheyareexpressedoroftheirimplicitappealtonaturalselectioninconnectingabiologicalexplananstoabiologicalexplanandum.IfthetheoryofnaturalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n130alexanderrosenbergselectionisnotitselfreducibletophysicalscienceand/orisitselfindispensabletoexplanationinmolecularbiology,thelatterwillbenomorereducibletophysicalsciencethantherestofbiologyisreducibletoit.ExamplesofsuchargumentsaretobefoundinKitcher(1984,1999),Sober(1993),andelsewhere.Torefutethissortofargument,reductionistsneedtodotwothings.Firsttheyneedtoshowthatevolutionaryexplanationsinfunctionalbiologyareunavoidablyinadequate,andinadequateinwaysthatcanbeimprovedonlybyevolutionaryexplanationsfrommolecularbiology.Andthentheyneedtoshowthatthetheoryofnaturalselectionisnotitselfabarriertothereductionofmolecularbiologytophysicalscience.Thesearebothdauntingtasks.Letusconsiderthefirstchallenge,thatofshowingwhatmakesevolutionaryexplanationsinfunctionalbiologyinadequateinwaysonlyevolutionarymolecularexplanationscancorrect.Supposeweaddressthequestion,Whydobutterflieshaveeye-spotsontheirwings?Theexplanationispresumablyanadaptationalistonethataccordsafunction,incamouflage,forinstance,totheeyespotonbutterflywings.Eyespotsonbutterflyandmothwingshavebeenselectedforoveralongcourseofevolutionaryhistory.Onsomebutterfliesthesespotsattracttheattentionandfocustheattacksofpredatorsontopartsofthebutterflylessvulnerabletoinjury.Suchspotsaremorelikelytobetornoffthanmorevulnerablepartsofthebody,andthislossdoesthemothorbutterflylittledamage,whileallowingittoescape.Onotherbutterflies,andespeciallymoths,wingsandeyespotshavealsobeenselectedfortakingtheappearanceofanowl’shead,brows,andeyes.Sincetheowlisapredatorofbirdsthatconsumebutterfliesandmoths,thisadaptationprovidesparticularlyeffectivecamouflage.Thereductionisthasnodifficultywiththisevolutionaryexpla-nation,asfarasitgoes.But,onthereductionist’sview,suchanexplanationdoeslittlemorethansettheresearchagendarequiredtocashintheexplanation’spromissorynotesofferedtheoriginalexplanation.Thefunctionalexplanationleavesunexplainedseveralbiologicallypressingissues,pressingenoughtodenyitcomplete-ness,correctness,orexplanatoryadequacy.Thesearethequestionofwhatalternativeadaptivestrategieswereavailabletovariouslineagesoforganismsandwhichwerenot,andthefurtherquestionofhowthefeedbackfromadaptednessoffunctionaltraits–liketheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)131eyespot–totheirgreatersubsequentrepresentationindescendantswasactuallyeffected.Themostdisturbinglacunaintheoriginalexplanationisitssilenceonthecausaldetailsofexactlywhichfeedbackloopsoperatefromfortuitousadaptednessoftraitsinoneormoredistantlypastgenerationstoimprovedadaptationinlatergenerationsandhowsuchfeedbackloopsapproachthebiologicalfacttobeexplainedasalocallyconstrainedoptimaldesign.Thesedemandsallstemfromwidelysharedscientificcommitments:thedemandforcompletecausalchains,thedenialofactionatadis-tance,andthedenialofbackwardcausation.Naturalselectionatlevelshigherthanthemacromolecularissilentonthecruciallinksinthecausalchainthatconverttheappearanceofgoaldirectednessintotherealityofefficientcausation.Onlyamacromolecularaccountoftheprocesscanprovidethem.Suchanaccountwoulditselfalsobeanadaptationalexplanation:itwouldidentifystrate-giesavailableforadaptationbyidentifyingthegenes(orothermacromolecularreplicators)thatdeterminethecharacteristicsoflepidopterans’evolutionaryancestorsandthatprovidetheonlystockofphenotypesonwhichselectioncanoperatetomovealongpathwaystoalternativepredation–avoidingoutcomes–leafcolorcamouflage,spotcamouflage,orotherformsofBatesianmimicry,repellanttastetopredators,Mullerianmimicryofbadtastingspe-cies,andsoon.Thereductionist’smorecompleteexplanationwouldshowhowtheextendedphenotypesofthesegenescompetedandhowthegenesthatgeneratedtheeyespoteventuallybecomepre-dominant,thatis,areselectedfor.Notethatthereductionist’sfullexplanationisstillahistoricalexplanationinwhichfurtherhistoricalfacts–aboutparticulargenesandbiosyntheticpathways–areaddedandareconnectedbythesameprinciplesofnaturalselectionthatareinvokedbytheoriginalevolutionaryexplanation.Butthelinksinthecausalchainofnaturalselectionarefilledintoshowhowpastadaptationswereavailableforandshapedintotoday’sfunctions(cf.Rosenberg2000).Antireductionistswilldifferfromreductionistsnotonthefactsbutonwhethertheinitialexplanationwasincomplete,incorrect,orinadequate.Theywillagreethatthemacromoleculargeneticandbiochemicalpathwaysarecausallynecessarytothetruthoftheoriginalevolutionaryexplanation.Buttheydonotcompleteanotherwiseincompleteexplanation.TheyaremerelyfurtherfacetsofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n132alexanderrosenbergthesituationthatmolecularresearchmightilluminate(Kitcher1999,199).Theoriginaladaptationalanswertothequestion,Whydobutterflieshaveeyespots?doesprovideacompleteexplanatoryanswertoaquestion.Whoisrighthere?Ifexplanationfollowscausation,thereduc-tionismhasmuchtorecommenditselfasamethodology.However,onaneroteticviewofexplanations,higher-levelandlower-levelexplanationsmaybeacceptedasreflectinganswerstodifferingquestionsadvancedindifferentcontextsofinquiry.Thereductionistmayadmitthattherearecontextsofinquiryinwhichthenon-molecularanswerstoquestionssatisfyexplanatoryneeds.Butthereductionistwillinsistthatinthecontextofadvancedbiologicalinquiry,asopposed,say,tosecondaryschoolbiologyinstruction,forexample,thenonmolecularquestioneitherdoesnotariseor,havingariseninanearlystageofinquiry,nolongerdoes.Moreover,theonlyassurancethatanythinglikethenonmolecularadaptationalexpla-nationisontherighttrackisprovidedbyamolecularexplanationthatcashesinitspromissorynotesbyestablishingtheadaptiveoriginsofthefunctionaltraitsinmoleculargenetics.Thelatter-dayreductionistholdsthatcompleteexplanationscanonlybeprovidedinbiologybyadvertingtothemacromolecularstates,processes,events,andpatternsonwhichnonmolecularhistoricaleventsandpatternssupervene.Thereductionistdoesnotclaimthatbiologicalresearchortheexplanationsiteventuatesincandispensewithfunctionallanguageoradaptationalism.Muchofthevocabularyofmolecularbiologyisthoroughlyfunctional.Norisreductionismtheclaimthatallresearchinbiologymustbe‘‘bottom-up’’insteadof‘‘top-down’’research.Sofarfromadvocatingtheabsurdnotionthatmolecularbiologycangiveusallofbiology,thereductionist’sthesisisthatweneedtoidentifythepatternsathigherlevelsbecausetheyaretheexplanandaforwhichmolecularbiologyprovidestheexplanantia.Whatthereductionistassertsisthatfunctionalbiology’sexplanantiaarealwaysmolecularbiology’sexplananda.Thereremainsaseriouslacunainthisargumentforreductionisminbiology,onepotentiallylargeenoughtodriveadecisiveanti-reductionistobjectionthrough.Althoughlatter-dayreductionismclaimstoshowthatnonmolecularexplanationsmustbecashedinformolecularones,theseexplanationsarestilladaptational,stillCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)133evolutionary–theystillinvokethemechanismofblindvariationandnaturalselectionoperatingonmacromolecules.Anduntilthisprinciplecanshowunimpeachablereductionistcredentials,itremainsopentosaythatevenatthelevelofthemacromolecules,biologyiscausallyautonomousfromphysicalprocesses.Incon-sequence,anirreducibleroleforlawsgoverningnaturalselectioninmolecularbiologywillsubstantiateanautonomyfromphysicalsciencethatnoreductionistcanaccept.Theprospectsforreductionismintheendturnontherelationshipbetweenthetheoryofnaturalselectionandphysicalscience.Threealternativessuggestthemselves.(a)Thelawsgoverningnaturalselectionareunderivednomicgeneralizationsaboutbiologicalsystems,includingmacromolecularones,andareemergentfrompurelyphysicalprocesses.Thisalter-nativewouldvindicatetheirreducibilityofallofbiology,butitwouldmakebiologydifficulttoreconcilewithphysicalism.Naturalselectionwouldturnouttobeaprocessnotfixedbythephysicalfacts.(b)Biologicalnaturalselectionisderivablefromsomelawsofphysicsand/orchemistry.Thisalternativewouldvindicatethereductionistvisionofahierarchyofscientificdisciplinesandtheories,withphysicsatthefoundations.Boththetemporalasym-metryofnaturalselectionandthemultiplerealizabilityoffitnessinavastsetofdifferentphysicalrelationsbetweeninteractorsandtheenvironmentmakethisalternativeunattainableforreductionism.(c)Naturalselectionasaprocessreflectsahithertounnoticedandunderivedlawaboutphysicalsystems(includingnonbiologicalones),andfromittheevolutionofbiologicalsystemscanbederived.Thisalternativehasbeenlittlecanvassedbutwouldreconcilereductionism’scommitmenttophysicalismandantireductionism’scommitmenttotheautonomyofbiologyfrom(therestof)physicalscience.4.genicreductionismandgenocentrismIndependentlyoftheissuesofphysicalism,thenatureofexplana-tion,andmacromolecularreductionismasaresearchprograminbiologythereisanother,relatedissueinthephilosophyofbiologyinCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n134alexanderrosenbergwhichthelabel‘reductionism’figures.Thetermisemployedtostigmatizeaviewitsopponentscall‘genicreductionism.’Inthissense,‘reductionism’isnotaclaimaboutintertheoreticalexplana-tionbutanamefortheerrorofignoringimportantcausalvariablesintheexplanationofsomeprocessorphenomenon.Thusforexampleitwouldbewronglyreductionistictotracethecauseofsicklecellanemiatoamutanthemoglobingenethatresultsinaserinemoleculeinthewrongplaceontheoutsideofthesufferer’shemoglobinmolecules,whichinitiallyresultsintheirclumpingtogether,thenarterialblockage,andeventuallyproducesanemia.Tobeginwiththerealsoneedtobearteries,redbloodcells,oxygen,andofcoursethewholestructuralandregulatorygeneticmachineryofhemoglobinsynthesisandalltheenvironmentalinputsrequiredfortheotherwisesuccessfuldevelopmentofthepersoncarryingthisgene.Thenotionthatthegenebyitselfsufficesforatraitorevenasingleproteinisapuerilemistakethatnobiologistcouldmake.Butsomephilosophersandbiologistsarguethatthetendency,wide-spreadamongbiologists,toaccordaspecialexplanatoryroleindevelopmentorsomaticregulationtothegenesconstitutesamoresophisticatedbutequallyegregiousversionofthisreductionisticmistake.‘‘Genicreductionism’’isthenamegiventothisreductiveerror.Thissortoferroneousreductionismofcourseencouragesandisinturnencouragedbytheprospectsforintertheoreticalexplana-toryreductionbetweennonmolecularandmolecularbiology.Ofcourse,intertheoreticalreductionismdoesnotimplyorrequiregenicreductionism,butitcertainlydoesencouragetheviewthatthegenehasaspecialexplanatoryroleinbiology,aviewthatmaycheerfullyacceptitsopponent’slabelof‘‘genocentrism.’’Accordingtothisview,thespecialcausalandthereforeexplanatoryroleofthegeneisduetoitsinformationalroleinstoringandtransmittinghereditaryinformation,itsroleinprogrammingthedevelopmentoftheembryo,anditsregulatoryfunctioninsomaticcells.Intertheoreticalreductionofnonmoleculartomolecularbiologyencouragestheseattributionsbothprogrammatically,astheDNAstructureofthegeneisourbestcurrentsourceofinformationaboutallbiologicallyactivemacromolecules,includingallenzymesandstructuralproteins,andsystematically,asmoleculardevelop-mentalgeneticsisbeginningtoprovidethedetailedexplanationofdevelopmentthatbiologyhassoughtbutfailedtosecuresinceCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)135Aristotle.Inthewordsofoneopponentofintertheoreticalreductionandgenicreduction,thetwentiethcenturywas‘‘thecenturyofthegene’’(Fox-Keller2000).Accordingthegeneaspecialcausalandexplanatoryroleinheredity,development,andregulationwouldbeaverystrongincentivetopursuingintertheoreticalexplanationsofthesortthereductionistadvocates.Inconsequence,thereisalivelydebateamongphilosophersofbiologyandbiologistsaboutwhethergeneshavethespecialcausalandexplanatoryrolesgenocentrismaccordsthem.Thedebateischanneledthroughthefollowingissuesparticularlyaboutdevelopmentmoleculargenetics:TherehavebeenseveralstrikingsuccessesintracingoutthedetailsofembryologicaldevelopmentamongorganismssuchasCaenorhabditiseligansandDrosophilamelanogaster.Thisresearchshowsthatprettywellalltheactionisingeneregulationandgeneexpression(evenforepigeneticprocessessuchasmethylation).ItmightbearguedagainstgenocentrismthatthesesuccessesareinsufficienttogroundtheexpectationthattherestofthedetailsofDrosophiladevelopmentandbehaviorareequallyintelligiblefromapurelymacromolecularperspective.WhysupposethatdevelopmentamongvertebratesshouldbejustlikeDrosophilaembryogenesis?Isitreasonabletoextrapolatefromonecasetotheclaimthatamacromolecularprogramwillexplaindevelopmenteverywhereandalways?Thegenocentristwillreplythathomologiesinthereg-ulatorydevelopmentalgenesofallmulticellularcreatures(e.g.,theHomeoboxgenes)aresogreatthattheburdenofproofhasbeenshiftedtothosewhowoulddoubttheextrapolationfromthefruitflytoHomosapiens.Amoreseriousobjectiontogenocentrism,onepersistentlyraisedagainstintertheoreticalexplanatoryreductionaswell(Kitcher1984),istheobservationthatmuchofthestoryofdevelopmentinthesecasesisexpressedinnonmacromolecularconcepts.Thedescriptionofdevelopmentadvancedinmoleculardevelopmentalgeneticsmakesrepeatedreferencetocells,theirpropertiesandbehavior.IftheroleofwholecellsisindispensabletothegeneticprogramforDrosophilaembryogenesis,wewillhavetobeconfidentthatthereisanadequate,purelymacromolecularexplanationoftheroleofthecellanditscyclebeforewecanconcludethatmolecularbiologyalone,andunaided,providestheexplanationfordevelop-ment.AndiftheexplanatoryroleofthewholecelloritsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n136alexanderrosenbergnonmolecularpartsisbothindispensableandirreducible,thenreductionism’sexplanatoryclaimmustbesurrendered.Vindicatinggenocentrismagainstthisargumentrequiresananalysisthatshowsthatmacromolecularexplanationsadvertingtoorganelles,cells,tis-sues,andsoon,citetheirpurelyphysical,topographic,geographic,andspatialpropertiesandsoareinnocentofseriouslypresupposingthebiologicalprocessestheypurporttoexplain,orelseitrequiresthattheroleoforganelles,cells,tissues,andsoforth,bethemselvesfullyexplainedinmacromolecularterms.Reductionistsarecon-fidentbothofthesetasksareattainable.Avariantontheobjectionthatnonmolecularfactorsarerequiredeveninmolecularexplanation,expressesthemoreradicalskepti-cismofacoalitionofbiologistsandothersdubiousaboutthegeneticexplanationofdevelopment.Whysupposethatthegeneshaveanyspecialroleindevelopment?Thereisavastrangeofotherconditions–physiologicalandenvironmental–causallynecessaryforfertiliza-tionandembryogenesisalongwiththeproductsofthegenes.Accordingtothe‘‘casualdemocracythesis’’advancedforexamplebyDevelopmentalSystemsTheorists(GriffithsandGrey1994,forexample),eachofthesecasuallynecessaryfactorsisonaparwiththeothers;noneisevenprimusinterpares;anddependingontheexplanatoryinterestsandpracticalfocusofabiologistanyoneofthemcantakecenterstageinoneoranotherexplanationofdevelopment.Accordingly,genocentrism’sattributionofaspecialroleindevelop-menttothegenesisunwarranted,andalongwithitseclipseweshouldrejectreductionismaswell(LewontinandLevins1985).Genocentristswillacknowledgethattherearemanyotherfactorscausallynecessaryalongwiththegenesfortheireffects.Buttheywillinsistthatwhatmakesthegenes‘‘special’’andvindicatesgenocentrismistheirstatusasinformationalmoleculesandtheirroleinprogrammingthesequenceofnecessarystepsandtheirmaterialsintobuildingtheembryo.Onthisviewthegeneticcode,whichmakessixty-fourdistinctmessageunitsoutofthecombinationoffournucleicacidbasesingroupsofthreebases(theso-calledcodons),hastheliteralpropertiesofaninformationsignalingsystemthat(redundantly)namesthetwentyaminoacidsandallowsfortheexpressionofinstructions(inmessengerRNA)totheribosomestoproducealltheproteinsinthebodyintheorderandamountsrequiredfordevelopmentandfunction.ThecodeisonthisviewanCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReductionism(andAntireductionism)137arbitraryconvention,inCrick’sterms,a‘‘frozenaccident’’(MaynardSmith2000).Opponentsdenythatthegenomeisliterallyinforma-tional.Somedosobyappealingtoalternatetheoriesoftheoriginsofthegeneticcodethatpurporttoshowitisnonarbitrary.Thisdebateisvexedbyimportantphilosophicalproblemssurroundingthenotionofinformationandthepropertyofintentionalityinits‘‘homebase’’–psychology–andisunlikelytobesettledindebateamongbiologistsandphilosophersofbiology.Butthegenocentristisonstrongergroundinclaimingthatthegenesareprivilegedbecauseoftheirroleinliterallyprogrammingtheembryo.Herethegenocen-tristargumentissimpleanddirect.DNAsequenceshavealreadybeenusedliterallytoprogramcomputers–albeitnotelectronicbutmacromolecularones–andtoenablethemtocompletecalculationsthatelectroniccomputersareincapableofundertaking,includingforexampleparticularlylargevaluesoftheNP-hard‘‘travelingsalesman’’problem.IfDNAmoleculescaninstantiateaprogramtosolvethesesortsofproblems,theninstantiatingaprogramthatsolvesmuchsimplerproblems,suchasbuildinganembryo,isifnotchild’splay,atleastwellwithintheirpowers.Thesequenceofpolynucleotidebasesinthegenomecanliterallyprogramtheembryo.ButtoinferfromthepoweroftheDNAsequenceliterallytoprogramdevelopmenttotheconclusionthatthegeneticprogramcandosoproceedsonthereductiveidentificationofgeneswithDNAsequences.AswesawinSection1,thisisahighlyprob-lematicalassumption.Infact,opponentsofgenocentrismseizeuponthedifficultyofprovidinganumericalidentityofthegeneasaunitoffunctionwiththeDNAsequenceasastructuredphysicalobject,toargueagainsttheverycoherenceofthegeneconceptandsoagainsttheprospectsforgenocentrism.Thenotionofthegeneisindeedquiteunproblematic.Genocentrismandintertheoreticalreductiveexplanationrequirethattherearegenes,thattheycanbedistinguished,individuated,counted,andotherwisetreatedastherelevantunitsofhereditarytransmissionanddevelopmentalcontrol.Butithasbeenallegedthehistoryofgeneticsinthelatetwentiethcenturyhasshownthatthenotionofageneasonethingindifferentlyidentifiedbyitseffectsinproteinsynthesisandbyitschemicalcompositioncannolongerbecountenancedbyinformedbiologists(seeGriffithsandNeuman-Held1999,forexample).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n138alexanderrosenbergThoughforalongtimeproductiveandfertileinitsencouragementofscientificprogress,theconceptofthegenehasbeenovertakenbyevents.Thecomplexitiesinheredityanddevelopmentthatmolec-ularbiologyhasuncovered,onthisview,makethegeneanobsoleteidea.Itisanideathatwillbeeclipsedinthenextcenturybyothernotions,perhapsonesthatwillvindicateantireductionistconclu-sionstothesameextentthatthetwentiethcentury’sfixationonthegenereflectedreductionisticones.Andwhenthegeneissuperceded,sowillgenocentrism,itsassociatedreductionisticresearchprogram,anditsphilosophyofbiology.Respondingtotheseargumentsandprovingthatthereportsofthedeathofthegenehavebeengreatlyexaggeratedremainamongthechiefpreoccupationsoflatter-dayreductionism.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nlindleydarden8MechanismsandModelsGenerallyspeaking,makingmodelsforunknownmechanismsisthecreativeprocessinscience.Harre´1970,401.introduction:mechanismsandmodelsBiologistsoftenseektodiscovermechanisms.Knowledgeofbiolo-gicalmechanismsisvaluablebecausedescriptionsofthemoftenplaytherolesattributedtogeneralscientifictheories.Theyprovideexplanationsofpuzzlingphenomena.Theyenablebiologiststomakepredictions.Theyaidthedesignofexperiments.Theymayexplaindomainsofwidescope.Theymaymakepossiblemedicalorbiotechnologicalinterventionsforpracticalpurposes.Especiallyinmolecularbiology,theoriesconsistofsetsofmechanismschemas,suchasthoseforDNAreplicationandproteinsynthesis.Biologistsusemanytypesofmodelstorepresentanddiscovermechanisms:diagrammaticmodels,physicalscalemodels,analoguemodels,modelorganisms,invitroexperimentalsystems,mathe-maticalmodels,computergraphicandsimulationmodels.Modelsrepresentandsubstituteforthethingmodeled,whilebeingeasiertounderstand,manipulate,orstudy.Choiceofanappropriatemodeldependsontheproblemtobesolvedusingit.Inmedicine,animalmodelsareoftenusedwhenthegoalistounderstanddiseasemechanismsinhumans.Molecularbiologistsusebacteriaandvirusesasmodelsformechanismswithdomainsofverywidescope,suchasDNAreplication.Thetopicsherearemechanismsinbiologyandmodelsthataidtheirdiscovery.Section2providesacharacterizationofmechanisms,based139CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n140lindleydardenoncasesfrommolecularandneurobiology.Section3introducesthedistinctionsamongmechanismschemas,theirinstantiations,andincompletesketches;thetermmodelinthesenseofatheoreticalmodelmayrefertoanyofthethree.Severalkindsofmodelsaidthediscoveryofmechanisms,especiallyanaloguemodels,modelorganisms,andinvitroexperimentalsystems;theyarethesubjectofSection4.Section5examinestheuseofsuchmodelsinreasoningtodiscovermechanisms,whichisanextendedprocessofgenerating,testing,andrevisingmechanismsketchesandschemas.Finally,theconclusionpointstogeneralphilosophicalissuesandtounansweredquestionsinthisnewresearchprogramonmechanismsinbiology.2.characterizationofbiologicalmechanismsAmechanismissoughttoexplainhowaphenomenonisproduced.Mechanismsmaybecharacterizedasentitiesandactivitiesorganizedsuchthattheyareproductiveofregularchangesfromstartorsetuptofinishorterminationconditions(Machamer,Craver,andDarden2000,3).Thenatureofthephenomenonforwhichamech-anismissoughtprovidesimportantguidanceindiscovery.Biologistsseekthelocationofthemechanismandfindplacesforitsbeginning,ending,toppingoff,bottomingout,andboundaries,guidedinpartbythenatureofthephenomenon.Manybiologicalmechanismsareregularinthattheyusuallyworkinthesamewayunderthesameconditions.Theregularityisexhibitedinthetypicalwaythatthemechanismrunsfromstarttofinish,therebyproducingandrepro-ducingagivenphenomenon.Forexample,thephenomenonofDNAreplicationisproducedbythemechanismofDNAreplication.Themechanismbeginswithonedoublehelixandendswithtwo.Onedoublehelixunwindsandeachhalfprovidesatemplatealongwhichcomplementarybasesarealigned,yieldingtwoidenticalhelicesattheend.ThedescriptionofthismechanismbottomsoutatthelevelofpartsoftheDNAmolecule,includingthebasesandtheircharges.Theprecisehydrogenbondingbetweenbases(usually)producesaccuratecopyingoftheorderofthebasesfromtheparentstrandstothedaughterones.Thetopping-offpointforthedescriptionofthismechanismistheentiredoublehelix,amacromoleculewithinthenucleiofcells.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels141Mechanismsarecomposedofbothentities(withtheirproperties)andactivities.Activitiesaretheproducersofchange.Entitiesarethethingsthatengageinactivities.Forexample,twoentities,aDNAbaseanditscomplement,engageintheactivityofforminghydrogenbondsbecauseoftheirpropertiesofgeometricshapeandtheirarrangementsofweakpolarcharges.Entitiesandactivitiesareinterdependent.Forexample,polarchargesarenecessaryforhydrogenbondformation.Appropriateshapesarenecessaryforlockandkeydockingofenzymesandsubstrates.Thisinterdependenceofentitiesandactivitiesallowsbiologiststoreasonaboutentitiesonthebasisofwhatisknownorconjecturedabouttheactivities,andviceversa.Suchreasoningbyforwardorbackwardchainingaidsdiscoveryofsubsequentorpriorstages,basedonwhatisknownorconjecturedaboutadjacentones.Forthepurposesofagivenbiologist,researchgroup,orfield,therearetypicallyentitiesandactivitiesthatareacceptedasrelativelyfundamental.Inotherwords,descriptionsofmechanismsinthatfieldtypicallybottomoutandtopoffinparticularplaces.Thoseplacesmaybemoreorlessarbitrarilychosen.Forexample,memorymechanismsareinvestigatedatmanymechanismlevels,fromamouselearningamazetotwoneurons(cells)exchangingneuro-transmitters(molecules).Alternatively,appropriatebottomingoutandtoppingoffmaybedictatedbythenatureofthephenomenonandthekindsofworkingentitiesthatareactiveinmechanisms.Inmolecularbiology,mechanismstypicallybottomoutindescriptionsoftheactivitiesofmolecules(macromolecules,smallermolecules,andions)andcellorganelles(e.g.,ribosomes).Theseentitiesaretheworkingentitiesofmolecularbiologicalmechanisms,suchasDNAreplicationandproteinsynthesis.Smaller(orlarger)entitiesdonothavetherequisitesizes,shapes,charges,orotheractivity-enablingpropertiestoplayrolesinthesemolecularmechanisms.Mechanismshaveproductivecontinuitybetweenstages:thatis,theentitiesandactivitiesofeachstagegiverisetothenextstage.Therearenogapsfromthesetuptotheterminationconditions.Mechanismshaveabeginningandanend,againmoreorlessarbi-trarilychosen.Forinstance,anaturalbeginningpointforthemechanismofDNAreplicationisonedoublehelix,andanaturalendingistwodoublehelices.However,inanongoingseriesofmechanisms,someofwhichmightbecyclic,thechoiceofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n142lindleydardenstartandstoppointsmaybebasedmerelyontheconvenienceofinvestigation.Mechanismsshouldbedistinguishedfrommachines.Amachineisacontrivance,withorganizedpartsdesignedtoworktogethersmoothly.Mechanismsareoftenassociatedwithmachinesbecausemechanismsaremostconspicuousinhumanartifacts,suchasthemechanicalclock.Astoppedclockisstillamachine,butitisnotamechanism.Mechanismsareactive.Humanartifactsmayexhibittheoptimalresultsofengineerseffectiveandefficientdesigns.However,livingthingsarearesultofevolutionarytinkeringandsatisficing.AsMichaelRusenoted,theideathattheworldisfullofdesignedmachineshasbeenreplacedbytheideathatitcontainsevolvedmachines,builtinaramshacklewayasevolutionfashionstheiradaptationsfromavailableparts.Althoughorganismsmaybeviewedasramshacklemachines,anorganism,asawhole,isnotamechanism.Manymechanismsoperatewithinalivingorganism.Moreover,lookingupinsteadofdown,organismsmaybesaidtoplayrolesinhigher-levelmechanisms,suchastheisolatingmechanismsleadingtospeciation.3.descriptionsofmechanisms:schemas,sketches,andtheoreticalmodelsBiologicaltheoriesrepresentedbysetsofmechanismschemasmaybecontrastedwithphilosophersanalysesoftheoriesassetsofsyntacticformalaxiomsorasabstractandidealizedformalsemanticstructures.(Thesenseofformalmodelinthissemanticconcep-tionoftheorieswillnotoccupyushere.)Analysisofmechanistictheoriesinbiologydoesnotimportaformalstructuretounderstandtheories,butinsteadstrivestocharacterizemechanismsandtheirrepresentationsinamannerfaithfultobiologistsownusages.Scientistsusetheoriestodescribe,explain,explore,organize,pre-dict,andcontroltheitemsinatheorysdomain.Descriptionsofmechanismsaidallofthesetasks(Craver2002a).Adequatedescriptionsofmechanismsincludeadescriptionofthephenomenonproducedbythemechanism,theentitiesandactivitiescomposingthemechanism,theirsetupconditions,alongwiththeirproductivelycontinuousspatialandtemporalorganization.Spatialorganizationincludeslocalization,structure,orientation,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels143connectivity,andcompartmentalization(ifany).Temporalorgani-zationincludestheorderinwhichactivitiesoccur;theirrate,duration,andfrequency;andtheoverallorderofthestagesofthemechanism(Darden2006,chap.12).Amechanismschemaisatruncatedabstractdescriptionofamechanismthatcanbefilledwithmorespecificdescriptionsofcomponententitiesandactivities.Anexampleisthisdiagramofthecentraldogmaofmolecularbiology:DNA!RNA!proteinThisisaschematicrepresentation(withahighdegreeofabstraction)ofthemechanismofproteinsynthesis.Alessschematicdescriptionofamechanismshowshowthemechanismoperatestoproducethephenomenoninaproductivelycontinuouswayandsatisfiesthecomponency,spatial,temporal,andcontextualconstraints.Thegoalinmechanismdiscoveryistofindadescriptionofamechanismthatproducesthephenomenon,andforwhichthereisempiricalevidenceforitsfeatures.Amechanismschemacanbeinstantiatedtoyieldsuchanadequatedescription.Incontrast,amechanismsketchcannot(yet)beinstantiated.Componentsare(asyet)unknown.Sketchesmayhaveblackboxesformissingcomponentswhosefunctionisnotyetknown.Theymayalsohavegrayboxes,whosefunctionalroleisknownorconjectured;however,whatspecificentitiesandactivitiescarryoutthatfunctioninthemechanismare(asyet)unknown.Thegoalinmechanismdiscoveryistotransformblackboxes(componentsandtheirfunc-tionsunknown)tograyboxes(componentfunctionsspecified)toglassboxes(componentssupportedbygoodevidence).Aschemaconsistsofglassboxesonecanlookinsideandseealltheparts.Incompletesketchesindicatewherefruitfulworkmaybedirectedtoproducenewdiscoveries.Thetransitionfromsketchtoschemamaybeacontinuousprocess,asvariousportionsofthemechanismarediscoveredinapiecemealway.Aninstantiatedschemashowsdetailsofhowthemechanismoperatesinaspecificinstancetoproducethephenomenon.Hence,mechanistictheoriesexplainthephenomenaintheirdomains.AsWilliamBechtelandAdeleAbrahamsen(2005)noted,explainingaphenomenoninvolvesdescribingthemechanismresponsibleforit,oftenbyconstructingamodelthatspecifieskeyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n144lindleydardenparts,operations,andorganizationandthatcansimulatehowtheirorchestratedfunctionstransformcertainparts.Thetermmodelherereferstoamodelofamechanismorwhatphilosophersmightcallamechanistictheoreticalmodel.Modelinthissensemayrefertoanyofthethreetermsdiscussed:amechanismschema,aninstantiationofamechanismschema,orasketch.Sometimesthetermstheoryandmodelinthissenseoftheo-reticalmechanisticmodelareusedsynonymously,inwhichcaseamechanismschemaorasetofmechanismschemasisappropriate.Sometimesamodelissaidtobeaninstanceofatheory,showinghowanabstracttheoryistobeappliedinaparticularcase,inwhichcaseaninstantiationofamechanismschemaisappropriate.Sometimesaproposaliscalledamodelbecausecomponentsareasyetunspeci-fied,inwhichcaseitdesignatesanincompletemechanismsketch.Thescopeofthedomainmodeledvaries.Amechanismschemamayrepresentasingleuniquecase(e.g.,amechanismproducingauniquehistoricalevent)orarecurringmechanisminonlyonespecies(e.g.,adisease-producingmechanisminoneformofhumancancer).Moreoften,mechanismschemashaveamiddlerange(Schaffner1993)ofapplicability;thatis,theyarefoundinsomesubsetofbiologicalcases,suchasmemorymechanismsinthehip-pocampusofvertebrates.Inafewcases,aschemamaybeclaimedtoapplytoallknowncases,suchasallinstancesofproteinsynthesisinlivingthingsonEarth.Again,considerthediagramforthemechanismofproteinsynthesis:DNA!RNA!proteinThisisasimplifiedandgeneralschemaoftheproteinsynthesismechanism.Itisveryschematicandabstract;atthisdegreeofabstractionitmaybeinstantiatedinadomainofverywidescope.ItappliestomostinstancesofproteinsynthesisinlivingorganismsfoundonEarth.Butcomparethefollowingschema:RNA!DNA!RNA!proteinThisdiagramisatthesamedegreeofabstractionasthepreviousone,butithasadomainofmuchnarrowerscope,namely,retroviruses.Hence,thedegreeofabstractionwithwhichthemechanismschemaisrepresentedandthescope(thatis,thegenerality)ofthedomainCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels145modeledaredistinct.Increasingthedegreeofabstractionmayproduceaschemawithahigherdegreeofgenerality,butnotnecessarilyso,astheseexamplesillustrate.Theamountofdetailspecifiedinanabstractmechanismschemaiscalledthedegreeofabstraction.Abstractionhierarchieshaveanincreasinglossofdetailasoneascendsthehierarchy.Conversely,asonedescendsthehierarchy,thereisincreasingspecificationofdetailuntiltheschemaisinstantiated,resultinginthedescrip-tionofaparticularmechanism.Thus,amechanismschemashouldnotbeviewedasamodelwithmerelythetwo-placerelationofavariableanditsvalue;mechanismschemahierarchiesmayhavearangeofdegreesofabstraction.Thetermdegreeisusedtorefertorungsinabstractionhierarchieswhilethetermlevelreferstorungsinpart-wholehierarchiesamongnestedmechanisms(usuallyrepresentedatroughlythesamedegreeofabstraction).Inbiologyandphilosophyofbiology,thetermmodelisusedinmanyways,suchastorefertomechanismschemas,theirinstan-tiations,sketches,andhierarchicalmechanistictheories.Adifferentbutalsocommonusageisemployedtorefertosomethingrelevantlysimilartothemechanismofinterestandusedinitsrepresentationordiscovery.Tomodelsofthislattertypewenowturn.4.modelsfordiscoveringmechanismsManykindsofmodelsaidthediscoveryofmechanisms.Modelshavebothasubjectandasource.Themechanismtobedis-coveredisthesubjectofthemodel.Thesourceofthemodelmaybethesubjectitself,asinthecaseofphysicalscalemodelsandcomputersimulationmodels.Incontrast,thesourceofamodelmaybedifferentfromthesubjectmechanismofinterest,asinthecaseofananaloguemodeloramodelorganismusedasasubstituteforstudyingmechanismsinhumans(Harre´1970).Diagramsareatypeofmodelusedtorepresentthemechanismofinterest;theyhavethesamesourceandsubject.Diagramsareespeciallypropitiousforrepresentingmanymechanisms.Theyshowoverallspatialorganizationofthepartsanddepictmoreorlessstructuraldetailoftheentities.Activitiesaremoredifficulttorepresentinstaticdrawings.Sometimesarrowsillustrateactivities,butarrowsarealsooftenusedtoshowmeremovementortimeCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n146lindleydardenslices.Cognitivepsychologistshavestudiedhowhumansmanip-ulatevisualrepresentationsinordertorunmentalsimulationsofmechanisms.Thismethodenablesthepersontoseehowsomemechanismsworkandtousetherepresentationtomakepredictions(discussedinBechtelandAbrahamsen2005).Butinmorecomplexcases,humansuseaids,suchascomputersimulations,torepresentthecomplexmechanismandtorunasimulationtomakeapredic-tionandexplorewhat-ifscenarios.JimGriesemer(2004,43839)intriguinglynoted:Althoughinteractivegraphicsextendedthetraditionofphysicalmodeling,theyalsoconstitutedanewmodeofinteractionwithnumericaldata,allowinguserstointervenekinestheticallyinthesimulationprocess.Thisisterraincognitoforconventionalphilosophiesofscientificknowledge.AphysicalscalemodelofDNAhasthesamesourceandsubject.Inafamouscase,thex-raycrystallographerRosalindFranklinproducedx-rayphotographsbybombardingcrystallizedDNA(boththesourceandthesubject).JamesWatsonandFrancisCrickusedherphotographsinchoosingtheshapeanddimensionsoftheirphysicalscalemodeloftheDNAdoublehelix.Althoughthex-raycrystal-lographicdata(aswellasotherdataaboutthechemicalcompositionofDNA)constrainedthespaceofpossiblemodels,thoseconstraintswereinsufficienttodetermineallthephysicalpropertiesofDNA(thesourceandsubject)touseinbuildingthescalemodel.InTheDoubleHelix,WatsonrecountsthemomentwhenhephysicallymanipulatedaccuratelyconstructedphysicalmodelsoftheDNAbases.Thattactilemanipulationintwodimensions(basedontheassumptionthatthebaseswereflatandinaplane)allowedhimtodiscoverthegeometricfitandhydrogenbondingbetweencom-plementarybases.Thisdiscoveryillustratestherolethatmaybeplayedbyphysicalmanipulationsofscalemodels.Itwasonestepamongmanyinthistwo-yearextendeddiscoveryepisode.Thedoublehelixmodelwithitstwostrandsofcomplementarilybondedbasessuggestedtoitsdiscoverershowitcouldcarryoutoneofthefunctionsofthegeneticmaterial.TheyimmediatelyproposedthemechanismofDNAreplicationviatheactivityofcomplementarycopyingtofilltheblackboxofgeneticreplicationintheseriesofhereditarymechanisms.Thus,themodelofDNAservedasamodelforinvestigatingthefunctionsofthegeneticmaterial.(Formoreonthisdistinctionofmodelofandmodelfor,seeGriesemer2004.)CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels147Two-dimensionaldiagrams,three-dimensionalphysicalscalemodels,andcomputergraphicmodelsmayallbegeneratedbyusingdataaboutthesubjectbeingmodeled.Toseehowthestructurescanpossiblyfunction,researcherslocatetheactivity-enablingpropertiesoftheentities.Examplesincludethehydrogenbondsbetweenbasesortheactivesitesofenzymes.Thoseactivity-enablingpropertiessuggesttohumansortocomputationaldiscoveryprogramswhattheactivitiesandstagesofthemechanismmaybe.Drugdiscoveryprogramsreadilyexploitsuchknowledgeofactivesitesonmole-culestodesignnewchemicalstoplaydesiredrolesindiseasepre-ventionmechanisms.Standingincontrasttosuchmodels,inwhichthesourceandsubjectarethesame,areanaloguemodels,inwhichthesourceforthemodeldiffersfromthesubjectitself.Distantandnearanalogies,modelorganisms,andmodelexperimentalsystemsareexamples.Ananaloguemodelismoreorlesssimilartothesubjectofinterest.InherclassicModelsandAnalogiesinScience,MaryHesse(1966)coinedusefultermsforthesimilaritiesanddifferencesbetweentheanalogueandthesubject.Thecomponentsthattheybothsharearethepositiveanalogy.Thecomponentsthataredissimilararethenegativeanalogy.Thecomponentswhoserelationhasyettobedeterminedatagivenstageintheuseoftheanalogyaretheneutralanalogy.Scientistshaveoftenusedanalogiesindiscoveringnewscientifictheories.ExamplesaboundinKeithHolyoakandPaulThagards(1995)MentalLeaps:AnalogyinCreativeThought.Thediscoveryofmechanismsisnoexception.Thoseworkingontheuseofanalogicalreasoningtoconstructscientifichypothesesbreakitdownintostages:problemfinding,analogueretrieval,extractionofanabstractcausalstructurefromtheanalogue,mappingfromanaloguestructuretothesubjectarea,adjustmentstofitthesubject,andtestingofthenewlyconstructedhypothesis.First,oneidentifiestheproblemtobesolved.Forexample,onewishestounderstandthemechanismofregulationofthegenesproducingtheenzymesforsynthesizingtheaminoacidtryptophan(trp).Thenonesearchestoretrieveanappropriateana-logue.Forexample,onemightbefamiliarwiththemodelforreg-ulatingthesetofgenesforproducingtheenzymesfordigestingthesugarlactose;thelacoperonmodelworksviaaderepressionmechanism.ThenextstageisextractinganabstractmechanismCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n148lindleydardenschemafromthedetailedanalogue.Forthelacoperonmodel,onemightdropthedetailsspecifictothelactosecasetoconstructanabstractderepressionmechanismschema.Insuchanabstractschema,ageneproducesarepressorproteinmoleculethatbindstoanoperatorgeneontheDNAjustupstreamforasetofcoordinatelycontrolledstructuralgenes.Whenanexternalinducerispresent(themilksugarlactose),itservestobindtotherepressorandchangeitsshape;asaresulttherepressorfallsofftheDNA.OncetherepressorfallsofftheDNA,theadjacentstructuralgenesbecomeactive.Thisabstractderepressionschemamaybemappedtothesubjectareaforthetryptophancase.Adjustmentsmustbemadebecauseofthedifferencesinthetrpcase.Thetrpgenesareexpressedwhencertainconcentrationsoftryptophanarenotpresent.Sotryptophan,whenboundtotherepressor,allowsittobindtotheDNAandrepressthegenes.Intheabsenceofsuitableconcentrationsoftryptophan,therepressordoesnotbindtotheDNA.Thus,onecangenerateamechanisticmodelfortheregulationofthetryptophangenesbyanalogywiththelacoperonandappropriatemodifications.Onceonehasgeneratedamechanistichypothesisbyanalogy,itmustbeevaluatedtoseehowwellitfitsthesubjectarea.Oneusesittopredicttheoutcomeofexperimentsonthetrpsystem.However,anomaliesaroseduringtestingofthetrpsystem.Infact,thetrpoperonwasfoundtobemorecomplexthanthelacoperon.Thedepressionmechanismwasacting,butsomethingmorewasalsohappening.Resolutionoftheanomaliesrequiredtheadditionofanotherregulatorymechanism,calledattenuation.Thissecondarymechanismoperatedtofinetunetheconcentrationoftheen-zymesproducingtryptophan,dependingontheconcentrationoftryptophaninthebacterialcell.Sometimeswhenanomaliesarise,theunexploitedneutralanalogyintheoriginalanaloguemodelmaybearesourceforideasabouthowtorevisethemechanistichypothesis.However,inthetrpcase,newcomponentsnotfoundatallinthelaccasehadtobeaddedtoresolveseveralanomalies(Karp1989).Afterthesuccessoftheoperonmodel,itwasusedasananaloguetoconstructplausiblehypothesesabouthowothergeneswereregulated.Someoperatebyaderepressionmechanism,butothersdonot.Therepertoireoftypesofgeneregulationmechanismscontinuestogrow(Beckwith1987).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels149Inadditiontoconceptualanaloguemodels,biologistsusephysicalanaloguemodels,namely,modelorganismsandinvitroexperi-mentalsystems.Suchmodelsystemsmaybeused,insomecases,tomapdirectlytothesubjectofinterest;inothercases,theymaybeusedinthediscoveryofageneraltheory.Whenthegoalistodiscovermechanismsinhumans,oftendiseasemechanisms,thenanimalmodelsaresoughtorconstructedandmappingsaredirectfromtheanimalmodeltohumans.However,inveterinaryanimalmedicine,humansmaybethemodelorganismsformechanismsinvolvingpossiblepainproducedbydrugsorprocedures.Humanscanreporttheirpainsensations,whileinferencesonthebasisofphysiologicalorbehavioralcuesinanimalsaremuchlessreliableforjudgingthepresenceandseverityofpain.Manysuchconsiderationsguidethechoiceofmodelorganisms,includingthenatureofthemechanismsoughtinthesubjectofinterest,thebeliefthatsuchamechanismorarelevantlysimilaroneoperatesinthemodel,theeaseofmanipulation,andtheamountofworkthathasalreadybeendoneonthemodelorganismthatcanserveasabasisforfurtherwork(Burian1993).Schaffnerextensivelyexaminedtheuseofmodelorganismsinmolecularbiologyand,morerecently,inbehavioralgenetics(Schaffner1993,1998,2001).Asaresult,heviewedsomebiologicaltheoriesashavingthestructureofoverlappingtemporalmodels.Thesearetheoriesofthemiddlerange:thatis,theirscopeisnotuniversalbut,withvariations,applicablebeyondasingleinstancetoadomainofmiddle-rangescope,suchasprokaryotesorvertebrates.Componentsofsuchtheoriesarepresentedascollectionsofenti-tiesundergoingaprocess.Inthemechanisticperspectiveproposedhere,Schaffnerwasreferringtotheoriescomposedofmechanismschemaswithvaryingscope.Amodelorganismoramodelexperi-mentalsystemprovidestheprototype;then,howwidelytheprototypicalmechanism(orslightvariantsofitoritsmodules)occurshastobedeterminedempirically.Inbiological,asopposedtoapplied,research,asSchaffnernoted,thegoalisoftentofindgeneralizations.Amanipulablemodelissoughtthatwillprovideresultsthatcanbegeneralized.Themodelorganismisthesourceforthegeneralmechanismschema(thesubject),aswellasaninstanceofit.Thehistoryofbiologyisrepletewithexamplesnotonlyofexcellentmodelorganisms,namely,thoseCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n150lindleydardenwithtypicalmechanismsthatweresuccessfullygeneralized,butalsooffailedones,namely,thosewithoddquirksthatledtheirusersastrayinthesearchforgeneraltheories.GregorMendelspeas(Pisumsativum)havewhatotherslaterdiscoveredtobegeneralhereditarymechanisms,whileMendelsattemptstoextendhisresultstohawkweed(Hieracium)failed.Hawkweed,itwaslaterfound,canreproduceasexually,andthuswasaverypoorchoiceasamodelorganismforgeneticcrosses.HugodeVriesstudiedthesud-denappearanceofnewtruebreedingformsofeveningprimrose(Oenothera).Hebelievedthattheeveningprimrosewasanexcellentmodelorganismforestablishingaresearchprogramofexperimentalevolutioninhisbotanicalgarden.However,theextremelyrarechromosomalmechanismsintheeveningprimrosearenotgeneralatall.Considerableempiricalworkensuedtounravelthequirkymechanismsofhawkweedandeveningprimrose,morethanwouldlikelyhavebeendonehadtheynotplayedtheroleofanomalousmodelorganismsinthesearchforgeneralhereditarymechanisms.Incontrastarethetriumphaltales.T.H.Morganschoiceofthefruitfly(nownamedDrosophilamelanogaster)forhisgeneticstudiesyieldedunderstandingofverygeneralhereditarymechanisms.Similarly,JacquesMonodschoiceofthebacteriaEscherichiacoliledtothediscoveryofregulatorygenes,auniversallyfoundcom-ponentofgeneregulationmechanisms,eventhoughallarenotthedepressiontype(DardenandTabery2005).However,manyanoma-liesaroseforMonodsfamousquip:WhatstrueforE.coliistruefortheelephant,onlymoreso.Othermolecularbiologists,desiringtostudymechanismsofcellulardifferentiationnotfoundinbacteria,SydneyBrenner,forexample,carefullychoseandperfectedstrainsofthenematodeworm,Caenorhabditiselegans(discussedinAnkeny2000),andFranc¸oisJacob(1998)chosethemouse,Musmusculus.Evenwhenitfails,itisagoodstrategytogeneralizefromamechanismdiscoveredinoneexperimentalsystemtootherspro-ducingsimilarphenomena.Evolutionoftendoesreusemechanismsortheircomponents.AsFrancisCollins,thenheadoftheU.S.HumanGenomeProject,said:Becauseallorganismsarerelatedthroughacommonevolutionarytree,thestudyofoneorganismcanprovidevaluableinformationaboutothers.MuchofthepowerofmoleculargeneticsarisesformtheabilitytoisolateandCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels151understandgenesfromonespeciesbasedonknowledgeaboutrelatedgenesinanotherspecies.Comparisonsbetweengenomesthataredistantlyrelatedprovideinsightintotheuniversalityofbiologicmechanismsandidentifyexperimentalmodelsforstudyingcomplexprocesses.(Collinsetal.1998,68687)Becauseofthecommonevolutionarydescentofbiologicalorgan-isms,biologyhasastrongerbasisforappealingtosimilaritythanotherfieldsemployingcross-fieldanalogies.Manysimilaritiesarearesultofevolutionaryhomology;thatis,thesimilaritiesresultfromthesubjectandsourcessharingacommonancestor.Thus,modelorganismsandmodelexperimentalsystemsmayserveashomologuesforstudyingthemechanismofinterest.Butevolutionworksbothbycopyingandbyediting,thatis,bothbyinheritanceandbyvaria-tion.So,itisanempiricaljourneytofindtheappropriatefamilyresemblances,intheliteralsenseofthatterm(Schaffner2001).AsMarcelWeber(2005)pointedout,phylogeneticinferencesbasedonhomologyprovideasounderbasisforgeneralizationthanmereinductionbysimpleenumeration.Whenamechanismisfoundinorganismsdistantontheevolutionarytree,theassumptionismadethatallthedescendantsoftheircommonancestorsharethesamemechanism.Thisis,henoted,anargumentfromparsimony,butonethatisplausiblebecauseoftheunlikelihoodthatthesamemechanismdidariseindependentlyinwidelyseparatedevolu-tionarypaths.Hisexamplesofwidelysharedhomologousmechan-ismsincludedthemechanismsofDNAreplicationandproteinsynthesisinalleukaryotes.Evenmoreoftenthantheevolutionaryconservationofentiremechanisms,modulesofmechanismsarereusedinothermecha-nisms.ModelorganismshavesuppliedwhatWebercalledmaterialsforpreparativeexperimentation.Forexample,DNAsequencesextractedfromDrosophilawereusedasprobestofishforhomo-logousDNAsequencesingenomiclibrariespreparedfromtheDNAofavarietyofotherorganisms.TheimportanthomeoboxgenesdiscoveredinDrosophilaareanexample.Homeoboxgenescontrolthedevelopmentofthefrontandbackpartsofthebody.DNAsequencesalmostidenticaltothosefromthefruitflywerequicklyfoundinmiceandhumans(Weber2005,16264).Nowthatmanywholegenomeshavebeensequenced,thegenomedatabasesandthegrowingproteindatabasesserveaswhatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n152lindleydardenmightbecalledcannedmodelorganisms.Theseinsilicodataallowsearchestofindorthologousgenesthatcanbetracedbacktoacommonancestor.Thenumberofsharedgenesandproteinswithsimilaractivitiesissurprising.Mostorganismsshareasubstantialnumberofmolecularmechanismsormodulesofmechanismsthatareveryancient.Evolutionappearstoworkbyfashioningnewarchitecturesfromoldpieces.Whensuchorthologousgenesarefoundbuttheirfunctionisunknowninhumans,modelorganismsprovideresearcherswithauniquemethodforfindingthemechanismsinwhichthegenesfunction.Thismethodiscalledthemodifierscreen(HariharanandHaber2003).Randommutationsareinducedintheorganismknowntohaveaspecificmutationinageneofinterest.Theaddedmutationsinothergenesmaymodifytheusualphenotype,therebyprovidingcluestothemolecularmechanisminwhichthegeneofinterestisimportant.ManydifferentmutantsinDrosophilacanbeinducedandscreenedtodetectaneffect.Genesthatundergoamutationthatcausesaworseningofthephenotypearecalledenhancers,whereasgenesthatcauseacorrectionofthemutantphenotypearecalledsuppressors.Additionalgeneticexperi-mentsallowtherolesoftheenhancersandsuppressorsinthemechanismofinteresttobedeterminedandorthologousgenessoughttoinvestigatethescopeofthenewlydiscoveredmechanismcomponents.Somephilosophersraisedconcernsaboutwhethertheuseofsimplemodelorganismsmightskewresults,butSchaffnerrepliedthatmodelorganismsarenotonlyintendedtoberepresentativeprototypes,butalsotobeidealizedinthesenseofsharpenedandmoreclearlydelineated.Thevalueofsharpened,simplifiedideali-zationsisalessonthatthephysicalsciencescanstillteachus....Oncesimpleprototypesarepreliminarilyidentified...thenvaria-tions(oftenintheformofaspectrumofmutants)aresought(orre-examined)toelucidatetheoperationofsimplemechanisms(Schaffner1998,280;italicsinoriginal).Thusfarinmolecularandneurobiology,simplemodelorganismshaveprovedveryuseful.Theextenttowhichsimplemodelsystemsandthesearchforsimplemechanisticaccountsmustbesupple-mentedinthefaceofbiologicalcomplexityremainstobeseen(forexampleoffailuresofthemechanisticresearchprograminthefaceCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels153ofcomplexity,seeBechtelandRichardson1993).Nonetheless,manytypesofmodelshaveprovedveryfruitfulinthediscoveryofbiologicalmechanisms.5.reasoningstrategiesfordiscovery:generating,testing,andrevisingAnalyzingreasoningindiscoveryisamuchmoretractabletaskwhenwhatistobediscoveredisamechanism(ratherthanavaguelycharacterizedexplanatorytheory).Further,discoveryisanextendedprocessofgeneration,evaluation,andrevisionofmechanisticschemasandsketches.Modelorganismsandmodelexperimentalsystemsmayplaymanydifferentkindsofrolesinallstagesofdis-covery.Theymaybeusedforexploratoryexperimentation,priorto(orinplaceof)usingaconceptualanalogy,todiscoverpossiblecomponentsofthemechanism.Inthediscoveryofthemechanismofproteinsynthesis,PaulZamecnikandhiscolleaguesworkedtoperfectaninvitroexperimentalsystemthatwouldincorporateradioactiveaminoacidsintopolypeptides(componentsofproteins).Theycentrifugedratliverstoextractcomponents,includingmicroscopicallyvisibleparticles(latercalledribosomes).Theyfoundthattheyhadtoputintotheinvitrosystemaparticularcentrifugefractionextractedfromtheratliversinordertoproduceincorporationofaminoacidsintopolypeptidechains.Thiscaseshowsthataninvitroexperimentalmodelsystemcanbecon-structedbyphysicallydecomposinganactualorganismandtheninvestigatingtheworkingparts(Rheinberger1997).Ideallyonecanisolateallthemechanismcomponentsanddeterminetheirroleswithinthemechanism.Butevenbeforeathoroughcharacterizationisavailable,arunningmechanismmaybeconstructedinvitrotoallowfurtherexploratoryexperimentationofitsparts.Suchexplorationisoneofmanywaysthatgenerationandtestingarecloselytiedduringmechanismdiscovery(Darden2006,chap.3).Moreoften,modelorganismsandmodelexperimentalsystemsareusedtotestplausiblemechanistichypothesesgeneratedviaanalogy,presumedhomology,orothermeans.Craver(2002b)detailedexperi-mentalstrategiesfortestingahypothesizedmechanism.Suchexperimentshavethreebasicelements:(i)anexperimentalsetupinwhichthemechanism(orapartofit)isrunning,(ii)aninterventionCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n154lindleydardentechnique,and(iii)adetectiontechnique.Themechanismsketchorschemabeinginvestigatedmayprovideanabstractframeworkforconstructinganexperimentalprotocol:intervenehere;detectthere.Biologistssetupmanykindsofexperimentalmodelstotestmechanistichypotheses.Intactlivingorganismshavemanymechanismsrunning;thechallengewithintactorganismsistofindwaysofindividuatingsinglemechanismsandrulingoutconfoundingfactors.Invitropreparationssolvesomeproblemsencounteredwithinvivoones.Thechallengeistofindtheappropriatecomponentsandmakethemworkinaninvitroexperimentalsystem.Craver(2002b)discussedseveraldifferentkindsofinterventionstrategiesthathavebeenusedhistoricallytotestamechanistichypothesisinanexperimentalmodel.Firstareactivationstrategies,inwhichthemechanismisactivatedandthensomedownstreameffectisdetected.Oneexampleoftheuseofamodelorganismistoputaratintoamazeanddetectactivityinitsbraincellswitharecordingdevice.Acommonbiochemicalinterventionistoputinatracer,suchasaradioactiveelement;activatethenormalmechanism;anddetectthetracerasitrunsthroughthemechanism.Goodrecordingdevicesandtracersdonotsignificantlyaltertherunningoftheactivatedmechanism;theymerelyallowobservationofitsworkings.Secondaremodificationstrategiesthatinvolvenotmerelyacti-vatingbutmodifyingthenormalworkingofthemechanismoper-atinginthemodelsystem.Awaytolearnaboutamechanismistobreakapartofitanddiagnosethefailure(Glennan2005).Afruitfulwaytolearnabouttheactionofageneistoknockitoutandnotetheeffectsintheorganism.Aswiththenotoriousablationexperimentsinphysiologyinthenineteenthcentury,theproblemwithgeneknock-outtechniquesinintactanimalsisthatsuchamissingpartmayhavemultipleeffectsthataredifficulttodisentangle,giventheoftencomplexreactionsbetweengenotypeandphenotype.Anotherkindofmodificationstrategyisanadditivestrategy.Somecomponentinthemechanismisaugmentedorover-stimulated,theneffectsaredetecteddownstream.Craversexamplewasofengineeredmicewithmoreofaspecifickindofneuralreceptor.Thosemicelearnedfasterandretainedwhattheylearnedlonger,therebyprovidingevidencefortheroleofsuchreceptorsinlearningandmemory.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels155Craver(2002b)suggestedusingallthreetypesofstrategiesactivation,ablation,andaddition.Consistentresultsstrengthentheevidenceforthehypothesizedmechanism.Eachhelpstocompen-satefortheweaknessesoftheotherstoyieldarobust(Wimsatt1981)conclusion,namely,aconclusionsupportedbyavarietyoftypesofevidence(Lloyd1987).Inadditiontomanipulatinganintact,operatingmechanism,onemayseekevidencefortheexistenceandnatureofhypothesizedentities,activities,and/ormodulesseparately.Forexample,anionchannelproteinmaybeisolatedanditsstructureinvestigatedtofinditsroleinaneuronalmechanism.Theproteincouldbegeneticallyalteredtohaveanabnormaladditionalparttoinvestigatehowthataffectsitsfunctioning.ApositiveresultofsuchinvestigationofahypothesizedpartofamechanismisanexampleofwhatElisabethLloyd(1987)calledindependentsupportforaspectsofthemodel,todistinguishitfromtheoutcomeofthemodel(thelatterisoftencalledtestingapredictionofthemodelasawhole).Strategiesforcredentialingexperimentalevidence,ingeneral,are,ofcourse,importantforassessingtheevidenceobtainedfromamodelorganismormodelexperimentalsystem.Theseincludeuseofadequatecontrols,reproducibilityofresults,appropriateuseofrandomization,anddemonstrationoftheadequacyofinstruments,tonameonlyafew.Finally,evidencefromtwoormorefieldsfurtherstrengthenstheclaimthattheconclusionisrobust.Sometimesasingleresearcherusestechniquesfromtwodifferentfields.Some-timesresearchersfromdifferentfieldsprovideevidencefordifferentmodulesofthemechanism,asdidthebiochemistsandmolecularbiologistsforthemechanismofproteinsynthesisandtheworkingofthegeneticcode.Thestudyofinterfieldrelationsbydifferentresearchgroupsandthecoordinationoftheirresultstoprovideevidenceforacoherentpictureofthemechanismisoneofthemanyimportantsocialaspectsofthecollectivescientificenterprise.Adescriptionofaparticularmechanismmaybelocatedinthelargermatrixofbiologicalknowledge(Morowitz1985),whichincludeshierarchicallyorganizeddescriptionsofmechanismsinwhichonemechanismservesasapartofalargerone.Thematrixalsoincludeslongertemporalseriesofmechanismsthatindicatewhichmechanismsoccurbeforeandafteragivenone.TheserequirementsforanadequatedescriptionofamechanismconstrainCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n156lindleydardenandguidemechanismdiscovery,asadescriptionofeachissought,anymissingcomponentsarefilled,andcoherencewithinawidercontextisexplored.Aswehaveseen,discoveryofmechanismsinvolvesgeneratingofpossibleandplausiblemechanistichypotheses(e.g.,byanalogyorhomology),testingthosehypothesesinmodelorganismsandmodelexperimentalsystems,anddecidingwhetheranewlyproposedmechanismfitscoherentlyintothematrixofotherknownmechanisms.Anotherpartofthediscoveryprocessiserrorcor-recting.Duringtesting,afailedpredictionyieldsanempiricalanomaly.Themechanistichypothesismaybeinneedofrevision(Wimsatt1987).Afruitfulandoftemployedstrategyistoover-generalizefromasuccessfulresult,useitanalogicallyinothercasestoconstructplausiblehypotheses,thenspecializewhenanomaliesarise.Asystematicsearchforanomaliesallowsthescopeofamechanismschematobedetermined.Further,anomaliesguidethegenerationofhypothesesaboutalternative,variantmechanismsthatdonotfitthehypothesizedschema.Asintheextendeddiscoveryprocessesofgenerationandtesting,theviewthatwhatistobediscoveredisamechanismprovidesguid-anceinreasoningtoresolveanomalies.Suchreasoninginanomalyresolutionis,first,adiagnosticreasoningtask,andthenaredesigntask.Thelocationofthefailureissought.Philosophershavebeenundulypessimisticabouttheabilitytolocalizethesiteoffailureinsomeholisticwebofbeliefs.Inpractice,scientistsoftenlocalizetheerroneouspartofamechanismschemaandcorrectit.Diagramsofthemechanismsstagesaidlocalizationoftheproblematiccomponent.Then,dependingonthesiteoflocalization,aredesignprocessmaybeneededtoimprovethehypothesizedmechanism.Thehypothesizedmechanismormechanismschemaaidsbothdiagnosistolocalizethefailureand,ifrequired,redesigntosupplyanimprovedmodule.Asafirststepintheanomalyresolutionprocess,theanomalousresultmustbecredentialedtoensurethatitisnottheresultofanobservationalorexperimentalerror.Experimentsrevealingananomalymaybereproduced,usingcarefulcontrols,orinvestigated,usingothercredentialingstrategiesforexperimentalresults(formoreoncharacterizinganomalies,seeElliott2004).Oncetheanomalousresultisconfirmed,thelocationofthefail-ureneedstobediagnosed.OnthebasisoftheextentofrevisionCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels157required,ananomalymaybecategorizedasamonster,specialcase,ormodelanomaly.Iftheanomalycanbelocalizedoutsidethedomainofthemechanismschema,thennorevisionisrequired.Anotherpossibilityisthattheanomalymightresultfromadiseaseorotherabnormality.Suchmonsteranomaliescanbebarredfromrequiringachangeinthenormalmechanismschema.Anexampleofmonsterbarringoccurredwhenlethalgenecombinationsproducedanomalousgeneticratios;normallythecombinationoftwogeneticallelesdoesnotleadtothedeathoftheembryo.Norevisioninclaimsaboutnormalgeneticmechanismswasrequiredwiththediscoveryoflethals;akindoffailurehadbeenfound.Sometimes,theanomalyrequiresasplittingofthedomaininwhichthemechanismisclaimedtooperate.Iftheanomalyonlyoccursinasmallpartofthedomain,theanomalyisaspecialcaseanomaly.Forthesmalldomainconsistingonlyofretroviruses,aRNA!DNAstepwasaddedtotheusualmechanismschemaforproteinsynthesis.Incontrasttomonsterandspecialcaseanomalies,modelanomaliesindicatewhatisnormalforadomainofwidescope.Thus,theanomalyisamodelinthesenseofanexemplar.Theremaybenosharpdividebetweenspecialcaseandmodelanomaliesasdomainsaresplittoaccommodatevariationsinthewaysmechanismsoper-ate.Theboundarybetweenspecialcaseanomaliesandmodelanomaliesisnotsharp.Oncetheanomalyisjudgedtorequirerevisionofamechanismschema,furtherguidanceresultsfromadiagrammaticrepresenta-tionofamechanismorothermeansoflocatingitsmodules.Inthemid-1950s,theribosomewashypothesizedtoplaythefunctionalroleofthetemplatefortransferringtheorderofthebasesintheDNAtotheorderoftheaminoacidsinaprotein.DNA!templateRNA!proteinDNA!ribosomaltemplate!proteinAnomaliesbegantoaccumulatefortheribosomaltemplatehypothesis.AsDouglasAllchinnotedinexaminingothercases,presenceofmultipleanomalieslocalizedinthesamesiteofahypothesisstrengthenstheconfidencethatrevisionisrequired.AttemptingtoresolvetheanomalyinwhichthebaseratiosofDNAandribosomalRNAdidnotcorrespond,Crick(1959)atCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n158lindleydardenfirstsystematicallygeneratedalternativehypothesestosavetheribosomeastemplatehypothesis.ThisanomalyindicatedaproblemabouttheDNA!RNAstepintheproposedmechanism.Heproposedalternativeslocalizedinthismoduleofthemechanism.Eachcomponentofthismoduleservedasalocationforgeneratinghowpossiblyredesignhypotheses.Thiscaseshowsasingleresearchersystematicallygeneratingasetofalternativeredesignhypothesisatthesiteoffailure.Conservatively,thesetofalternativesCrickdiscussedin1959didnotincludethepostulationofanasyetundiscoveredtypeofRNAhavingabasecompositionlikethatofDNA.ThiswastheideaofaseparatemessengerRNA(mRNA),differentfromtheknowntypesofRNA.ThediscoveryofsuchamessengerRNAwasthewaytheanomalywassoonresolved.TracerexperimentssupplieddirectevidencefortheexistenceofmRNAs.Thefunctionalrequirementofatemplate,atthatstageofthemechanism,withappropriaterela-tionstothestagesbeforeandafterit,acquiredanewrolefiller,namely,messengerRNA(discussedinDarden2006,chap.3).Thisribosomeanomalycaseshowsthatwhenwhatistoberevisedisamechanismschema,thatschemafurnishesmuchguidanceforanomalyresolution.Diagramsandotherrepresenta-tionsofthemodulesofmechanismsguidelocalizationandredesign.Whenananomalyislocalizedtoastage,thenredesignmayneedtobedonebyaddingsomethingbeforeorafterthestageorchanginghypothesizedentitiesand/oractivitieswithinthestageitself.Fur-thermore,theentitiesandactivitiesofastagemustgiverisetothenext,therebyimposingconstraintsonthecomponentsofasub-sequentstage,onthebasisofwhattheprioronecanproduce.Also,themodulesofthemechanismnotimplicatedbytheanomalymustbeshowntocontinuetofunction.Thedesideratumofhavingaproductivelycontinuousmechanismthusaidsredesignduringanomalyresolution.Insum:reasoninginthediscoveryofamechanismisguidedbythedescriptionofthephenomenonofinterest,aidedbythecharacter-izationofwhatamechanismis,andelaboratedbyspecifyingthefeaturesthatanadequatedescriptionofamechanismshouldsatisfy.Mechanismdiscoveryinvolvestightrelationsamonggeneration,evaluation,andrevisionofmechanismschemasofvaryingscope.PhilosophersshouldnotviewdiscoveryasaprocessofflounderinginCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMechanismsandModels159anunconstrainedspaceofvaguelycharacterizedtheories.Ifthegoalistodiscoveramechanism,muchcannowbesaidaboutreasoningstrategiesandexperimentalmodelstoaidthattask.6.conclusionPhilosophersofbiology,workingoncasesfrommolecularbiology,cellbiology,andneurobiology,havecharacterizedmechanismsasusedinthosefields.Analoguemodels,modelorganisms,andmodelexperimentalsystemsaidthediscoveryofmechanisms.Reasoningingeneration,evaluation,andrevisionconvertsincompletemechanismsketchestowell-supportedmechanismschemas.Thisnewperspectiveonmechanisms,arisinginthephilosophyofbiology,allowsthereexaminationoftraditionaltopicsinthephilosophyofscience.Thesetopicslookdifferentwhenonestartswithmechanisms(ratherthan,e.g.,perspectivesarisingfrommathematicalphysicsorformallogic).Thisresearchprogramisjustbeginning;thecitationsherepointtorecentwork.Philosophersarguethatappealtomechanismsprovidesanaccountofcausation(Glennan1996,Machamer2004,Tabery2004,Bogen2005),dis-covery(Thagard2003,BechtelandAbrahamsen2005,Glennan2005,Darden2006),explanation(Machamer,Darden,andCraver2000,Glennan2002,BechtelandAbrahamsen2005),functionalanalysis(Craver2001),interfieldintegrationandunity(Craver2005,Darden2006,chap.3),andreduction(Craver2005,Darden2006,chap.4).TheseauthorsstresstheimportanceofmechanismsinsuchfieldsasMendeliangenetics,molecularbiology,cellbiology,neuroscience,cognitivescience,andlinguistics.Asyetunsolvedaretheissuesofhowthisviewofmechanismsappliestoanalyzingthemechanismofnaturalselection(SkipperandMillstein2005)andtoanalyzingmathematicalmodelsinpopulationgeneticsandecology.Mathe-maticalandcomputersimulationmodelsofmechanismsusuallyhaveequationsorfunctionstoproducestatetransitions,whileomittingrepresentationsofstructuresandtheactivitiesthatproducethetransitions.Couldtheseimpoverishedmathematicalmodelsbeimprovedbyaddingthedetailsoftheworkingpartsofmechanisms?Thisnewmechanisticperspectiveisprovingfruitfulforreexamingissuesinphilosophyofsciencefromthepointofviewofphilosophyofbiology.Itislikelytocontinuetoprovidenewinsights.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nandreariew9TeleologyIdistinguishbetweenPlatonicandAristotelianteleologies.IdetailWilliamPaleysPlatonicteleologicalargumentfortheexistenceofGodfromnaturaldesignandofferDarwinstheoryofnaturalselec-tionasanantiteleologicalresponse.But,sinceAristotlesteleologyisdistinctfromPlatosteleologyIasktowhatextentisDarwinstheoryanti-Aristotelian.TeleologyinbiologyismakingheadlinenewsintheUnitedStates.ConservativeChristiansareutilizingateleologicalargumentfortheexistenceofasupremelyintelligentdesignertojustifyleg-islationcallingfortheteachingofintelligentdesign(ID)inpublicschools.Teleologicalargumentsofoneformoranotherhavebeenaroundsinceantiquity.Thecontemporaryargumentfromintelli-gentdesignvarieslittlefromWilliamPaleysargumentwrittenin1802.Botharguethatnatureexhibitstoomuchcomplexitytobeexplainedbymindlessnaturalforcesalone.Whatissoremarkableaboutcomplexdesigns?Compareawatchwithastone.AccordingtoPaley,watchesarecomplexandstonesarenotbecauseawatchsfunctioningdependsonitsprecisearrangement;astonesdoesnot.MichaelBehe,acontemporaryadvocateofintelligentdesign,labelsthesortofcomplexityirre-duciblecomplexity:ifyouremoveanypart,thewholestructureceasestofunction(Behe1996).Thebacterialflagellum,Beheargues,isirreduciblycomplex.Itactsasatinypropellerspinningatmorethan20,000revolutionsperminute.Thirtydifferentproteinsarepreciselyformedandadjustedtoproducethemotion.Ifanyoneofthemisremoved,theflagellumceasestopropelthebacterium.TheexampleforwhichPaleyismostfamousisdifferentthemammalianeyebuttheargumentformisthesame.160CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology161Howdoweexplaintheexistenceofcomplexdesigns?Tradi-tionally,therearetwomainpositions,thematerialistandthetele-ologist.Amaterialistbelieves,roughly,thatallnaturalphenomenaaretheproductofthecausalinteractionsofmatter.Thereisnoroominthematerialistsontologyforpurposeorgoals.Matterisnotimbuedwithit.Materialarrangementsarenotformedbypurposiveagents.Allseeminglygoal-directedactivityisultimatelythepro-ductofmatterandcause.Evenourownintentionsandgoals(andconsciousness)areexplainedmateriallywithoutreferencetogoal-directedprocesses.Onthematerialistviewcomplexitemsandarrangementsareexplainedbynonpurposivenaturalforces,matterandcause.Philosophersinthecontextoftheteleologydebatessometimescallthematerialistanaturalist.Iwillusethetermsinterchangeably.Ateleologistbelievesotherwise:theexplanationforgoal-directedarrangementscallsforgoal-directedcauses.Thereareroughlytwosortsofteleologists,onewewillcallaPlatonistandtheotheranAristotelian.APlatonistbelievesthatsimplematterandcausedonotexplainthegoodnessororderlinessofanarrangement.Toillustrate,considerananalogousquestion,WhydoesSocratessitinprison?WhilefactsaboutphysiologyofferacompleteexplanationofSocratescurrentpositioninprison,thefactsdonotprovidetherealreasonforSocratespredicament.HeremainsinprisonbecauseremainingratherthanescapingiswhatSocratesdeemsthebestcourseofaction.IntheTimaeus,Platoappliesthisreasoningtoexplaincosmologicalorder.Hefindsthematerialistexplanationwithitsreferencetothesimplemotionsatisfactoryfordescribingastronomicalpatterns,butinsufficientforexplainingthebeauty,orderliness,andgoodarrangementofthecosmos.Thetruecauseisagencyworkingforthebest,concludesPlato(2000,48a);referencetointentionalityisthematerialistsmissingingredient.Whenitcomestoexplainingbiologicalarrangementandfunc-tioninganAristotelianeschewsPlatonicdesignersforaninherentpurposiveorgoal-directedforcethatresidesinthematerialproper-tiesoflivingentities.Goodarrangementsarenotthehandiworkofacreator;rathertheyareduetosomeinnerprincipleofchangewithinlivingorganisms.Thisforceexplains,amongotherthings,naturalregularities.Forexample,manycarnivorespossesssharpteethinthefrontandbroadmolarsintheback.ThisparticularCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n162andreariewdentalarrangementallowscarnivorestoflourishinthewild.Carnivoreswhosedentalarrangementssignificantlyvarytendtoperish.Thematerialistexplanationintermsofmatterandcausewouldfailtoexplainwhythecarnivorousarrangementissosuccessfulwhileitsvariantsarenot.Thatisbecause,accordingtotheAris-totelian,thematerialisthasnoontologicalcategorylikeAristotlesthatforthesakeofwhich,ortelos.Sharpteethgrowinfrontandbroadmolarsgrowinthebackforthesakeofanorganismsflour-ishing.Thegoalisinherentinthenatureofdentalgrowth.Tosumup,thePlatonicandAristotelianagreethatmaterialistexplanationsinsufficientlyexplaingoodarrangements,especiallyonesthatallowthesystemtoflourishorevenonesthatperformtheiractioninthebestpossiblemanner(asPlatobelievedaboutthecosmos).Theydisagreeonwheretolocatethepurposiveforceinnature.AmoderndayadvocateforintelligentdesignaswellasitsVictoriancousin(Paleys)isaPlatonist.Shebelievesthatgoodnaturaldesignmustbetheproductofadesignerjustaswell-designedartifactsmustbetheproductofintelligentandtalenteddesigners.TheAristotelianpositionconstitutesanalternativetothePlatonicmetaphysics,eschewingdesignersforinherentgoal-directedtenden-ciesimbuedinthematteroflivingorganisms.Onthefaceofit,ourdiscussionofthedifferencebetweenPlato-nistsandAristotelianschangedtheessenceofwhatteleologicalargumentspurporttoexplain.ForPaleyandcontemporarypropo-nentsofintelligentdesignwhatistobeexplainediscomplexity.AristoteliansandPlatonistsaimtoexplaineitherorderlinessinthecosmosorbiologicalarrangementsthatleadtoflourishingoforganisms.But,oncloserinspectionratherthanachangeofsubjectournewdiscussiondeepenedourpreviousanalysis.Implicitlywedefinedcomplexityinfunctionalterms.Anitemiscomplexifachangeinapartleadstotheceasingofthesystemsfunctioning.Butwhatofsystemsthathavearedundancybuiltin?Althoughanairplaneiscomposedofmanyinterworkingparts,itisnottruethatthefailureofapartwillforcetheplanetoceaseflying.Iftheauto-maticpilotsystemfails,thepilotcanswitchtomanual;ifoneengineconksout,theotherscompensate;ifthesystemresponsibleformaintainingappropriateoxygenlevelsinthecabinfails,oxygenmasksdropdownfromabovetheseat.Theredundantsystemsensurefunctioningevenwhenpartsfail,butthefactthatthepartsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology163arenotessentialforfunctiondoesnotmeanthatairplanesystemsarenotcomplex.Quitethereverse,itmeansthattheyareextremelycomplex.Idonotmeantoengageinalongdiscussiononexactlywhatcomplexitymeans,butitisworthpointingoutthattheissueisnotexactlystraightforward(foropposingaccountsseeDembski1998andFitelson,Stephens,andSober1999).1.paley’sinferencetothebestexplanationPaleysargument(aswithBehes)startswithananalogybetweenlivingorganismsandhumanartifacts.Briefly,ifyoucameacrossawatchandinquiredastoitsexistenceyouwouldnottakeseriouslytheconclusionthatwatchesaretheproductofnaturalforces.Itishighlyimprobablethatnaturalforceswouldrandomlycoalescematterintoawatch.Thepossibleexistenceofadesignerwhocanmanipulatethepartsforhisownpurposemakestheexistenceofwatchesmuchmorelikely.APlatonicconclusionfollows:theexistenceofadesignerbestexplainswatchesandlivingorganisms.OnmyviewmanyofthemoreinfluentialteleologicalargumentsinhistoryareinstancesofinferencetothebestexplanationorIBE(seeAriew2002,Cooper1987,Johansen2004).OneinferstothetruthofahypothesisfromtheexistenceofphenomenathatthehypothesisthatbestexplainsPaleysinferencetotheexistenceofawatchmaker(andlaterhisargumentfortheexistenceofadivinecreator)worksexactlyinthisway.Theexistenceofwatchmakersissupportedbytheexistenceofwatchessinceexistenceofwatch-makersbestexplainshowsuchcomplexthingscouldcometoexist.AninterestingfeatureofIBEargumentsisthatwiththemwecaninfertheexistenceofunobservablephenomena.AsPaleyputit,alreadyhavinginferredtheexistenceofawatchmakerfrominspectionofawatchfoundonadirtpath:Norwouldit,Iappre-hend,weakentheconclusion,thatwehadneverseenawatchmade;thatwehadneverknownanartistcapableofmakingone;thatwewerealtogetherincapableofexecutingsuchapieceofworkmanshipourselves,orofunderstandinginwhatmanneritwasperformed(Paley1828,4).ThestrengthoftheinferencetoawatchmakerdependsnotonourwitnessingwatchmakersmakingwatchesbutintherelativelikelihoodthatwatcheswouldexistifskilledCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n164andreariewwatchmakersweretoexist.Comparethattothelikelihoodofthematerialisthypothesisthatwatcheswouldexistifonlytherandomactionofnaturalforcesweretoexist.Naturalprocessesaloneareunlikelytomakeawatch.Acreatorwithforethoughtmorelikelywill.WhileperhapsPaleywouldhaveunderstoodtheconceptoflikelihoodrathercolloquially,inthecontemporaryphilosophyofscienceliteratureitsmeaningismademorepreciseinthecontextofprobabilitytheory(seeSober2000).ThefeatureofinferringtheexistenceofunobservablecausesdistinguishesIBEfromgardenvarietyinductivearguments(Ruse2003,43).Letusseewhy.Thestrengthofaninductiveargumentdependsonthesizeandbiasnessofthesample.Yet,asPaleysug-gests(intheprecedingquote),wemayneverhaveseenwatchmakersmakewatches.Ifso,oursamplesizeiszero.LikewiseforthesampleoftimesinwhichanyofushasseenGodcreatinglivingthings(asHumepointedout).Ifteleologicalargumentswereinductivelybased,thentheywouldbenonstarters.GoodthingforPaleythathisinferenceisnotinductive.InanIBEtheissueisnotthefeaturesasamplehasincommon,butratherwhatexplainsanobservedphenomenon.LetustakeanotherexampletoillustratethedifferencebetweenIBEandinductivearguments(theexampleistakenfromSober2000).SupposeGregorMendelsargumentfortheexistenceofgeneswasinductivelybased.Inasenseitwas,sincehesupposedlyobtainedhisdefinite3:1phenotypicratiosforvariouscharactersofpeaplantsonthebasisofnumerousplantings.ButMendelstheoryofgeneswasnotinferredbytherepeatedplantings;histheoryof3:1phenotypicratioswas.Hethenaskedadistinctivequestion:Whatwouldexplainwhytheoffspringexhibittheseratiosfromparentalcross-ingsfromvarioustypes?Forexample,whydidheget3:1ratiosofwrinkled:smoothpeasintheoffspringcrosses?Here,heemployedanIBE:supposethehereditarymaterialactedasaparticlethatobeyscertainrules.Thatwouldbestexplaintheratios(comparetotheblendinghypothesisthatDarwinfavored).Notice,Mendelhadnodirectevidenceoftheexistenceoftheseparticles(thatwenowcallgenes).Rather,heinferredthemasthebestexplanationfortherepeated3:1phenotypicratios.SomephilosophersbelievethattheuseofIBEisoneofthehallmarksofgoodscience.Ifso,perhapsitshouldbesaidthattheuseofIBEinteleologicalarguments(bybothCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology165PaleyandIDers)shouldmarkthemasscientifichypotheses.Iwillnotpushthissuggestivelinetoofar,butnoticeitdoesgivethelietothetypicalargumentfromanti-IDersthatthecentralIDargumentisunscientific.IBEargumentsarenotunscientific.Paleynotesthatthestrengthoftheinferencetowatchmakersisnotinvalidatedifthewatchisfoundtobeinaccurate(Nei-ther...woulditinvalidateourconclusion,thatthewatchsometimeswentwrong,orthatitseldomwentexactlyright[1828,4]).Thisisanotherdistinctivefeatureofteleologicalexplanations,andonenotobviouslyavailabletomaterialisticexplanations.RecallAristotlesargumentagainstthematerialist:thematerialistcannotaccountforthereasonforwhycertaindentalarrangementsallowtheirpossessorstoflourishwhileothersleadtotheirdemise.Thesuccesstermsarenotpartofthematerialistvocabulary;thematerialisthasonlymatterandcausetoexplainorganicevents.Andyetthematerialetiologyoftheeventsdoesnotdistinguishonethatsucceedsfromonethatfails.Theteleologisthasanextraconcept,anevaluativetermthatexplainssuccessandfailure.ForAristotletheextraconceptisthatforthesakeofwhich.Certaindentalarrangementsexistforthesakeofacarnivoresflourishing.Carnivoreswithdefec-tiveoralternativearrangementsaremalfunctioningindividuals.TheargumentisthesameforPaley,butthenatureofthetelosisdistinct.Allwatcheshaveafunction,evenbrokenwatches.ThedistinctionbetweenbrokenandrunningwatchesiswhetherornottheyfulfilltheirGod-givenfunction.Sincetheteleologicalexplanationprovidesanaccountoftheessentialdifferencebetweenfunctioningandmalfunctioningitemswhilethemate-rialistexplanationdoesnot(thematerialistwouldhavetoascribethedifferencestochance),teleologicalexplanationshaveanadvantageovermaterialistexplanations.1.1Paley’sSelf-ReplicatingWatchNext,Paleyconsiderswhatwouldhappenifwefoundaself-replicatingwatch.Thepassageisalovelyearlyexampleofsciencefiction:Suppose,inthenextplace,thatthepersonwhofoundthewatchshouldaftersometimediscoverthat,inadditiontoallthepropertieswhichhehadCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n166andreariewhithertoobservedinit,itpossessedtheunexpectedpropertyofproducinginthecourseofitsmovementanotherwatchlikeitselfthethingisconceivable;thatitcontainedwithinitamechanism,asystemofpartsamold,forinstance,oracomplexadjustmentoflathes,baffles,andothertoolsevidentlyandseparatelycalculatedforthispurpose;letusinquirewhateffectoughtsuchadiscoverytohaveuponhisformerconclusion.(1828,9)Inadditiontoservingthefunctionoftellingtime,thiswatchhasafurtherextraordinaryfeature:itproduceswell-functioningoffspring.Thediscoveryofaself-replicatingwatchaffectstheformercon-clusiontheexistenceofwatchmakersinseveralimportantways.First,itfurtherillustratesthatthestrengthofPaleysinferencedoesnotdependoneverhavingseenanywatchmakersmakewatches.AsIarguedearlier,thisfeaturedistinguishesIBEargumentsfrominductiveonessinceitdoesnotdependonsamplingfromapopulationofevents.Second,andmoreimportant,thediscoveryoftheself-replicatingwatchstrengthenstheinferencetotheexistenceofadesigneratthesametimethatitweakenstheinferencetothehypothesisthattheitemistheproductofnaturalforcesalone.AsPaleyputsit,Ifthatconstructionwithoutthisproperty,or,whichisthesamething,beforethispropertyhadbeennoticed,provedintentionandarttohavebeenemployedaboutit,stillmorestrongwouldtheproofappearwhenhecametotheknowledgeofthisfurtherprop-erty,thecrownandperfectionofalltherest(1828,9).Theprob-abilityofnaturalforcesrandomlyproducingawatchisveryverysmall,buttheprobabilityofnaturalforcesrandomlyproducingsomethingasextraordinaryandexquisiteasaself-replicatingwatchisevensmaller.Thegenerallessonis,Themorecomplextheparts,thestrongertheevidenceofadesigner.ItandthenextfeatureplayalargeroleinPaleysultimateinference,theexistenceofaGod.Thethirdeffectthattheself-replicatingwatchexamplehasontheformerconclusionis,inPaleyswords,toincreasehisadmirationofthecontrivance,andhisconvictionoftheconsummateskillofthecontriver.Howevercomplexwatchesare,mostofuswithaverageintelligenceandskillscouldimaginelearning,afterextensivetrain-ing,howtocreatewatches.Buttohavetheskillofthemakerofaself-replicatingwatchwouldbeextraordinaryorevensupernatural.TheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology167generallessonhereis,Themorecomplexthedesign,themoreintelligent(orskillful)isthedesigner.1.2EasySteptoDesign:LivingOrganismsOncethesecondandthirdfeaturesofthenewIBEfromself-replicatingwatchesareinplace,Paleyhasonlytoconvinceusthatlivingtissues,organs,organisms,andecosystemsaremuchmorecomplexthanself-replicatingwatches,thattheirpartsaremuchmoreattunedtothefunctionsthattheyserve.InasensethatistheintentionofmostofPaleysbook,NaturalTheology,fromwhichhisfamousargumentfortheexistenceofanintelligentdesignerisarelativelysmallsection.Thelaterchaptersaremoreorlessazoologicaltextbook,detailingthewonderofnaturaladaptations.Themostfamouspassagesarefoundinthesectionthatdescribestheanatomyandfunctionoftheeye.Forinstance,heexpressesamazementathowtheanatomyofeyesfromanimalslivingindistinctenvironmentsdiffersaccordingtothelawsoftransmissionandrefractionofraysoflight.DoesPaleysargumentsucceedinprovingthatlivingorganismsarecreatedbyGod?Amonghiscontemporariesheprovidedapow-erfulargumentagainstamaterialistwhoseemstohavelittletoaccountfornaturaladaptation.Yet,thesuccessofIBEargumentsdependsontherelativesuccessofthegivenhypotheses,and,asDarwinwouldshow,thereareotherhypothesestoconsiderbesidestherandomactionofmatterandcauseandanintelligentdesigner.ThereinliesaformallimitationofIBEinferences.ThestrengthofanIBEisonlyasgoodastheprofferedhypotheses.Foranygivensetofhypothesestheinterlocutorhasalwaystheoptiontoremainagnosticastothecauseofthephenomenoninquestion.Tosuggestotherwise,forinstance,toarguethatsinceGodisabetterexplana-tionthanmatterandcause,Godmustexist,istocommittheonlygameintownfallacy(Sober2000).AsforprovingtheexistenceofanomnipotentandomniscientGod,PaleyscommentatorsandcriticsoftenpointedoutwhilePaleysinferencemighthavestronglysuggestedtheexistenceofasupremelyintelligentdesigner(asupernaturaldesigner),itstopsshortofprovingthatthedesignerisGod.IsuspectthatthisisthesortofthingthatmodernIDerslikeBehehaveinmindwhentheyexpressagnosticismaboutthenatureofthedesigner(Orr2005).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n168andreariew1.3IBEversusArgumentbyAnalogyInsteadofanIBE,somecommentatorsbelievethatPaleysargumentisbestreadasaninstanceofanargumentbyanalogy.True,anana-logybetweenfunctioningorgansandartifactsfeaturesinalotofwhatPaleysays.ButIthinkifPaleysargumentwereessentiallyanargu-mentbyanalogy,thenitwouldbealltheweakerforit.Asweshallseelateron,IthinkthesameofthosewhoreadAristotlesmanyrefer-encestoasinart,natureasconfirmationthathisteleologicalargumentsareessentiallyargumentsbyanalogy.ThereareformaldifferencesbetweenIBEargumentsandargumentsbyanalogy.Thestrengthofanargumentbyanalogydependsonthedegreetowhichtheitemorphenomenoninquestionresemblestheanalog.Winnie-the-Poohsomewhatresemblesreal-wordhoney-lovingbears,butnotenoughtowarranttheconclusionthatreal-worldbearsaremadeofcottonstuffing.Dowatchessufficientlyresemblelivingorganismstowarranttheconclusionthatthelatteraredesignedbecausetheformerare?Probablynotifweregardthevarietyofextantlivingorganisms.Watchesaremadeofmetalliccogsandwheels,haveglassfaces,andfitaroundoneswrist.Koalabearsareendothermic,climbtrees,anddrawcrowdsattheSanDiegoZoo.PaleysargumentfaresmuchbetterasanIBEthanasanargumentbyanalogy.Sure,theanalogybetweenwatchesandlivingorganismsplaysacentralroleintheIBE,butasameanstostrengthentheIBEinferencefromcomplexadaptationtodesigner,notasaanalogybetweenthefeaturesofwatchesandthoseoflivingorganisms.Inotherwords,therelevantfeaturesincommonbetweenwatchesandlivingorganismsaretheircomplexityandwell-suitednesstothefunctionthattheyserve.Ifdesignersarethebestexplanationforhighlyfunctioningartifactstheywillbeevenbetterexplanationsfortheevenmorecomplexandhighlyfunctioningformsfoundinnature.2.darwin’stwoalternativestopaley’screatorEarlierIsaidthatthestrengthofPaleysinferencetoadesignerdependsontherelativeweaknessofalternativehypotheses.Paleysalternativeisastraightforwardmaterialistwho,likeAristotles,isforcedtoexplainfunctionalarrangementsbycitingchance.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology169DarwinsgreatreplytoPaleyistoprovideanaturalisticalternativeinwhichchanceisnotthereasonwhylivingorganismsaremade.Infact,Darwinofferedtwoalternativehypotheses;bothserveasanalternativetosomeaspectofPaleysinference.Darwinsfirsthypothesisisthatalllivingorganismsarerelatedbycommonancestry.Differentspeciesevolvefromthecommonlineageasbranchesgrowonatree.Notallspeciesevolve;somegoextinctandotherscontinuetheirheritagerelativelyunchanged.TheideathatthereisatreeoflifefromwhichallformsevolveisaradicalalternativetotheChristianideathateachspeciesistheuniquecreationofanall-goodGodandthereafterimmutableandeternal(foranexcellentoverview,seeWaters2003).Darwinssecondcentralideaexplainstheconditionsforevolu-tionofspeciesandextinction,andultimately,thewayneworgan-ismalformsemergefrompreexistingvariation.Darwinarguedthatallorganismsundergoastruggleforexistenceinpartbecausemoreindividualsarebornthancouldpossiblysurvivegiventhelimitedresourcesoftheirenvironment(hegotthisfromreadingMalthus).Organismspossessavarietyofheritabletraits(theypassthemtotheiroffspring),andhenceorganismsarevariouslyequippedtohandlethestruggleforexistence.Thoseindividualswhohappentopossesstraitsthathappentomakethembetteradaptedtotheirlocalenvironmentalconditionswillpassthemtotheiroffspring.Thoselineagesthatfailtoreproducetheirtraitswilleventuallygoextinct.Reiteratethisnaturalselectionovermanygenerationsandovermanydifferentvariantsandeventuallylineagesevolvedistinctivefeaturesthatmakethemoverallbetteradaptedtotheirconditionsthantheywerebefore.Darwindidnotknowhowvarietiesareproduced;nordidheunderstandthedetailsandrulesofheritability.Buthistheorydidnotneedthedetails;itrequiredonlythatheritablevarietiesexistasthefuelfornaturalselection.DarwinstheoryofnaturalselectionisanalternativetoboththematerialistandtheChristiantheoriesfortheexistenceofcomplexadaptations.Aswesaw,thematerialistisforcedtoexplaincom-plexityandwell-adaptednessasachancesideeffectofnaturalpro-cesses.Darwinstheorydoesnotresorttochance.Hespecifiesthegeneralbiologicalconditionsthatdeterminetheexistenceofadap-tations.Darwinstheoryisradicallydistinctfromtheteleologicalexplanation:adaptationsarenotcreatedonDarwinstheory,butCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n170andreariewtheyemergeoutoftheconfluenceofmanydistinctbiologicalprocesses,theproductionofindividualdifferences,themechanismsofheritability,thebasicrulesofdemography,andtheconsequenceofdifferentialadaptationtolocalenvironmentalconditions.ItisonethingtoofferalternativestotheChristiantheoriesofspeciationandbiologicaladaptationandanothertoclaimthatthealternativesbetterexplainthenaturalphenomenawesee.First,letuscompareDarwinstreeoflifehypothesistotheChristiantheoryofimmutablespeciesintheformalcontextofanIBEargu-ment.WewillfollowwithacomparisonofDarwinstheoryofnaturalselectionwithPaleyshandofGod.HowcouldDarwindemonstratethesuperiorityofhistreeoflifehypothesis?ShouldDarwininferevolutionfromthesameperfec-tionsandintricaciesthatPaleyviewedasevidenceforGodshand-iwork?No.AsthecontemporarycommentatorS.J.Gouldputit,Idealdesignisalousyargumentforevolutionforitmimicsthepostulatedactionofanomnipotentcreator.OddarrangementsandfunnysolutionsareproofofevolutionpathsthatasensibleGodwouldnevertreadbutthatanaturalprocess,constrainedbyhistory,followsperforce(Gould1980,20).Darwinsargumentagainstacreatorandforanonintentionalforceofnatureisfoundintheawkwardnessofdevelopmentalpatterns,andtheseeminglypoordesignsofnature.Baleenwhalesdevelopteethinneotonyonlyforthemtobereabsorbedintothebaleenstructurethattheyusetofeedonkrill.Pandasgetatthetendershootsofbamboothroughtheinefficientprocessofrunningthestalksalonganinflexiblespurofbonethatjutsoutlikeathumb.Paleyarguedthatdesignisevidentinmishapsaswell,forthepurposeisclearevenifthesystemdoesnotachieveit(Paley1828,67).However,Paleyisreferringtoinstancesoffaileddevelopment,thatis,deformedindividuals.Darwinsmishapsareflawsoftypedesignflawsfromaengi-neerspointofview.Onemightsupposethatthesignificanceoftheapparentdesignflawsisarhetoricalpointagainsttheperfectionofthecreatorspowers.Wecouldask,Whywouldaperfectdesignercreatesuchawkwardandinefficientdesigns?ButthatleadsusawayfromamoresignificantpointandgetsusembroiledinthedebateabouttheperfectionofGodsoverallplan.Themoresignificantpointiswhataboutthesestrangebiologicalfeaturesrevealtousabouttheirprovenance.AsDarwinwrites,RudimentaryorgansCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology171maybecomparedwiththelettersinaword,stillretainedinthespelling,butbecomeuselessinthepronunciation,butwhichserveasaclueinseekingforitsderivation(quotedinGould1980,27).DarwinarguedthateachlineageisrelatedbycommonancestrywhiletheChristianviewisthateachspeciesrepresentsanindependentactofcreation.FlaweddesigninrudimentaryorgansprovidesevidenceofmodificationfromacommoncauseratherthanindependentcreationinthesamewaythatthesimilarityofwordsfornamesofnumbersamongSpanish,French,andItaliansuggeststhatthewordsshareacommonancestralrootratherthanarisingdenovoforeachlanguage(Sober2000b,42).Analogously,thereabsorbtionofwhalesteethinitsmotherswombisevidencethatwhaledevelopmentisnotaseparateactofcreationbutsurvivesasaremnantandmodification(bynaturalselection)ofanancestraldevelopmentalpattern.Withtheuseofastrangebonespurtosplicebamboostalks,thepandaseatinghabitsareawkward.Naturalselectionexplainsitsexistenceasamodifica-tionfromtheanatomyofpandaancestors.Onceevidencemountsthatorgansandorganismsevolvefrompreexistinglineages,thentheseeminglyimprobableideathatcomplexandwondrousadaptationsaretheresultofcertaincondi-tionspertainingtoblindbiologicalprocessesratherthanthehandofacreatorseemsmuchlessprobable.AsGoodrichputit:Darwinsgreatmeritistohaveshownthat[evolution]canbeseenatworkatthepresenttime,canbetestedbyobservationandexperiment,andleavenoroomforanymysteriousgoverningcausesinaddition;that,infact,acompletescientificaspectoftheprocessofevolutioncanbedescribedasanunbrokenseriesofnaturalevents,asequenceofcauseandeffect,aseriesofstepseachonestrictlydeterminedbythatwhichcamebeforeanddeterminingthatwhichfollowsafter.(Goodrich1912,28)2.1DarwinandAdamSmithToseehowthisworks,wewillcompareDarwinsanswertoPaleywithAdamSmithstheoryoflaissez-faire.TheanalogybringshomethepointechoedinGoodrichthatDarwinstheorypavedthewayfornaturalisticexplanationsforadaptationsandcom-plexity(thisisasimilarstrategyfoundin(Schweber[1977]andGould[2002]).ThequestionforAdamSmithandothereconomistswas,Whatmakesawell-orderedeconomy?OneanswerisakintoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n172andreariewtheteleologists:congregateeconomicexpertstocreateandimposeprinciplesandrulesontheeconomicactivitiesofthepeople.AdamSmithsanswerisverydifferent;infact,itistheoppositeoftheexperttheory:youallowindividualstotransactwithoutconstraint.Fromthecollectiveofallthoseunfetteredtransactionswillemergeawell-orderedeconomy.Accordinglytherearenorulesorprinciples;ratherorderemergesasasideconsequenceofindividualactivitywhoseintentisorthogonaltothecollectiveeffectindividualsareactingontheirowndesireforprofit,notforthegoodofthewhole(Gould2002).Smithinvokestheinvisiblehandtodescribethesedistinctivefeaturesofthistheory:thereisnohandofeconomycreation.Darwinstheorycanbeviewedinthesamelight.ToachievegoodadaptationsandharmonyinnatureyoudonotneedahandofGod;ratherthesephenomenaaretheresultofthecollectiveactivityofindividualsthataresimplystrivingtosurviveandreproduceontheirown,withoutamindtothecollectiveeffect.Theindividualsvaryintheirfeaturesandintheirabilitiestosurviveandreproduceinthelocalenvironmentalconditions.Naturalselectionistheinvisiblehand,or,tomixmetaphors,theblindwatchmaker.EmergentismistoPaley-stylecreationismasCopernicussSun-centeredcosmologyistoPtolemysEarth-centereduniverse:itstandstheconventionalbiologicalwisdomonitshead.AninterestingconsequenceofthisanalogybetweenDarwinstheoryandAdamSmithsisthat,contrarytowhatisreportedinthepopularpress,themainalternativetoIDisnotnecessarilythespecificsofDarwinstheoryofnaturalselectionbuthisbroaderconceptionthatcomplexadaptationsemergeasthecumulativeresultofblindbiologicalprocesses(asGoodrichindicates).ItmightturnoutthatthespecificsofDarwinsconditionsforevolutionareincompleteorfalse,butifso,itdoesnotfollowthatIDisthebestexplanationforbiologicalcomplexity.Thereareotherpossiblenaturalisticexplanationsofcomplexity,thatis,explanationsthatdonotrefertosupernaturalagency.HencethedebateisnotreallyDarwinversusGodbutnaturalismversusteleology.3.platonicversusaristotelianteleologyNaturalselectionandlaissez-faireareantiteleologicaltheoriesinthattheyprovideanexplanationfortheexistenceofapparentlygoodCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology173designswithoutreferencetoacreator.Yet,neitherisanti-teleologicaltoutcourt.Aswesawbefore,betweenPlatoandAristotletherearetworatherradicallydistincttheoriesofthenatureofthetelos.Torecapsomeofthesedifferences:forPlato,thecreatorgovernsallmotionexternally,fromonhigh(asitwere),wheretheactionisforthebestfromacosmologicalpointofview.ForAristotle,teleologyisnonpurposive,nonrational,nonintentional,andimmanent,residinginaninnerprincipleofchange.Thevalua-tionisnotaboutwhatisbestfromadesignerspointofview,butwhatisusefultoanindividual.Telosisabuilt-infeatureofmattertoaccountforthefinalcauseofnaturalarrangements.So,fromanalternativetoaPlatonictheoryofcreationitdoesnotfollowthatDarwinstheoryisantiteleological.PerhapsDarwinstheoryisAristotelian.InanexchangewithAsaGrayDarwinhintsatthepossibilitythatinsomesenseDarwinstheoryisteleological.AsaGraycomments,LetusrecogniseDarwinsgreatservicetoNaturalScienceinbringingbacktoitTeleology:sothatinsteadofMorphologyversusTeleology,weshallhaveMorphologyweddedtoTeleology(quotedinLennox1993,409),towhichDarwinreplies:WhatyousayaboutTeleologypleasesmeespeciallyandIdonotthinkanyoneelsehasevernoticedthepoint(Lennox1993,409).LetuslookmorecloselyatAristotlesdistinctiveteleologicalarguments.IntheendIwillarguethatsomeofDarwinstheorycanbeinterpretedasprovidinganalternativetoAristotlesteleology,butsomecannot.PerhapsthereasonwhyPlatosteleologydifferssofundamentallyfromAristotlesarisesfromtheirdistinctviewsabouttheoriginsofthecosmos.ForPlato,thecosmosbegantoexistataparticulartime.ForAristotlethecosmosiseternal.So,forAristotlethereisnoneedtoexplainhowthecosmoscametoexistbecauseitalwaysexisted(Zeyl2006).Further,AristotlesteleologicalargumentstendtoconcernbiologicalarrangementswhilePlatosaremoreglobalinthattheyconcerntheexistenceofthecosmos.Forhisbiologicalteleology,Aristotlerecognizestwocategories,formalandfunctional.Informalteleologythetelosisaninherentpropertyofbiologicaldevelopment.Itexplainsadevelop-mentaleventbycitingtheorganismsbiologicalneeds.Plantsrequirenourishment,sorootsextenddownwardsratherthanupwardsforthesakeofnourishment(Phys.199a29allcitationsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n174andreariewarefromAristotle1984).Infunctionalteleologythetelosisascribedtotherelationbetweentheplacementandfunctioningofpartsforthesakeofthewholeorganism.Sharpteetharelocatedinthefrontofthemouthforthesakeoftearing(199b24);thatiswhytheyarethere.Inotherwordssharpteethpersistinnatureamongcarnivoresbecausetheycontributetotheflourishingofcarnivores.3.1Aristotle’sTeleologicalArgumentfromFlourishingEarlierIdescribedAristotlesargumentfromflourishingwiththeexampleofdentalarrangements.Accordingly,thematerialistcannotexplainwhycertaindentalarrangements(sharpteethinfront,broadmolarsintheback)regularlyleadtotheflourishingofcarnivoresthatpossessthearrangementwhilealternativedentalarrangementsoftenleadtothecarnivoresdeath.Thematerialistexplanatorytoolkitofmatterandcauseistoolimitedtoexplainthedifference;hencethematerialistisforcedtoascribethedifferencetochance.Butthe(immanent)teleologistcanrefertotheinnerprin-cipleofchangethatischaracteristicofthespecies:thewinningdentalarrangementoccursforthesakeoftheflourishingofthespecies.Besidestheargumentfromflourishing,Aristotleoffersatleasttwomoreargumentsforteleology,theargumentfromhypotheticalnecessityandtheargumentfrompattern.Letuslookatthembothinturn.3.2Aristotle’sTeleologicalArgumentfromHypotheticalNecessityInPhysicsII.9Aristotlearguesthatinadditiontothenatureandmovementofsimplebodies(materialnecessity)andchance,biolog-icalexplanationsrequireathirdingredient:hypotheticalnecessity.Whatishypotheticalnecessity?Takeeyelids,forexample.Eyelidsareflapsofskinthatprotecteyesfromeasyexternalpenetration.AccordingtoAristotletheeyelidmaterialtheflapsofskinisnecessaryforthesakeofeyeprotection.Thenecessityshouldbereadasaconstraintonmaterialsgiventhespecificpurposeforwhichthepartwillbeused.Notanymaterialwilldoforthesakeofeyeprotection,onlyeyelidmaterialgiventhespecificformofeyeCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology175protectionthathumansandotheranimalsrequire.Thisismeanttobetakenstrongly:theactualmaterialsthatcomposeanorganarerequiredforthecompletionoftheprocesswherecompletionisthegoalofdevelopment(Cooper1987,255).Theconceptofhypotheticalnecessitymakescleartherelationshipbetweenfunctionalteleologyandformalteleology.Considertheexample:eyelidmaterialispres-entforthesakeofeyeprotection(thatisthefunctionofeyelidmaterial).So,eyelidmaterialhasafunctionalroletoplayinthegrowthofeyeprotection.Further,eyeprotectionisnecessaryforseeing,andseeingoccursforthesakeoftheorganismsgrowth.Thenecessityisgrantedtomatter,eyelids,andisconditionalinthatitcontributestothegoalofnaturalgrowth.Eyelidmaterialcon-tributestonaturalgrowthbyaffordingeyeprotection,whichitselfiscrucialforthefunctionofseeing(Cooper1987).AristotlesargumentfortheexistenceofhypotheticalnecessityisaninstanceofanIBE.Toillustrate,IfollowCooper(1987)andswitchexamplesfromeyelidstothedevelopmentofanewbornfromsperm,egg,andtheusualbackgrounddevelopmentalconditions(theexampleisCoopers,notAristotles).Aristotlewouldsaythatthematerialistcannotaccountforthewaythesematerialsconspiretoproducefetusesnearlyeverytime.Inotherwords,byappealtosimplemotionandmaterialcause,materialistscannotfundamen-tallydistinguishbetweenphysicalforcesthatareunconstrainedtoproducearangeofdifferentpossibleoutcomesandphysicalforcesthatnearlyalwaysresultinthesameproductanewborn.Thematerialistsonlyrecourseisanappealtocoincidence.Aristotlesreplyisthatcoincidenceisinsufficienttoaccountfortheregularityoftheconjugationseeninorganicdevelopmentbecausechanceoperatesonlyinunusualcircumstances(198b35199a3).Theprin-cipleofhypotheticalnecessitybetterexplainstheregularityofdevelopment:thematerialsarethereforthesakeofproducingtheconjugationthatleadstothedevelopmentofnewborns.3.3Aristotle’sTeleologicalArgumentfromPatternInhisargumentfrompattern,Aristotlerecognizesthatthesameteleologicalschemeappliestoexplainaparticularsortoforganiza-tionthatregularlyoccursbothwithinhumanactionandinthenonhumannaturalworld.TheorganizationhehasinmindisCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n176andreariewexemplifiedinallofthefollowing:housebuilding,leavesgrowingtoshadefruit,rootsdescendingfornourishment(ratherthanris-ing),nestbuildinginbirds,andwebmakinginspiders(theseareAristotlesexamples).Inallofthesecaseswerecognizeacertainpatternofarrangementandsequentialorder.Forexample,indevelopmentofanartifact(suchashousebuilding)orinnature(asinrootsdescendingdownwards)allthestepsofdevelopmentoccurinasequencethatleadsuptothefinalstate.Thesepatternsdonothappenbyaccident.Rathertheyoccurineveryinstanceinwhichtherelevantorganizationisfound,forexample,intheintentionalproductionofartifacts(housebuilding)orthenondeliberatefor-mationofnaturalobjects(webmaking,nestmaking,rootsde-scending,leavesshadingfruit).ItisinthisrespectthatAristotlefamouslyremarksthatasinart,soinnature(Phys.199a910)andasinnature,soinart(199a1516).Thesamepatternthatexplainscertainorganizationsfoundinnaturealsoexplainsthesameorganizationsfoundinartifacts(Charles1995,115).Thiscertainorganizationisjustgoal-directedactivity.Aristotleinfersteleologyfrompatternsoforderandarrangement.Wewillcallthistheargumentfrompattern.Tostrengthenthisargument,Aristotlepresentsthefirstinstanceinwhichteleologypreservesadistinctionbetweenfunctionandaccident;however,forAristotlethetermisamistake.Mistakesoccurwhenoneofthestagesrequiredtoachievethegoalhasfailedtocompleteitsroleintheproductionofthegoal,forexample,whenadoctorpoursthewrongdosageorwhenamanmiswritesorwhenmonstrositiessuchasman-headedox-progenyorolive-headedvine-progenydevelop.Thesameteleologicalpatternwherebyeachstageofdevelopmentoccursinorderforthesakeofthegoalallowsustoexplainthedifferencebetweenwhatoccursbyartornature,ontheonehand,orbymistake,ontheother.Whatoccursbyartornaturefollowsthepatternsuccessfullywhilemistakesorthecrea-tionofmonstersfeaturesafaileddevelopmentalstage.Contrastorderlinessamongthenormalbeingswithdisorderfoundinmon-strosities.Forthematerialistthereisnoessentialdifferencebetweenfunctionandmalfunction;bothareexplainedintermsofmatterandcause.Aristotlearguesthatteleologybetterexplainsthedifference:justaspurposefulmanufacturingproducesfunctionalartifacts,disorderedmanufacturingleadstomistakes.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology177PerhapswecanviewAristotlesteleologyaspointingoutadis-tinctivefeatureofbiologicalprocessesthat,unlikethecosmologicalanddemographic,seemtobegforateleologicalargument.Thecosmosfeaturedaregularityororderlinessthatrequiredexplana-tion.Butbiologicaleventsfeaturemeans-endsprocesses(IborrowherefromAmundson1999,16,wherebydevelopmentalprocessesandbiologicalarrangementsoccurforthesakeofthegoodoftheorganism[recallSocratesrealreasonforremaininginprisonitisforhisowngood]).4.darwin’salternativetoaristotelianteleologyRecalltheimportanceofDarwinsanswertoPaleysargumenttotheissueofteleology.Darwin,ingatheringevidenceforhistreeoflifehypothesis,debunksPaleysPlatonicteleologywherebyorganictraitsareintentionaldesignsofasupremecreator.Evidencefromvestigialandoddarrangementssuggeststhatorganictraitsarenotderivedfromapurposefulactofcreationbutratherorganictraitsarederivedandmodifiedfromthetraitsoftheirancestorsthroughnaturalselection.Thatis,Darwinreplacesthehandofcreationwithadescriptionoftheconditionsrequiredforvariousindividualsandtheirtraitstoevolveintoadaptations.However,bydebunkingPlatonicteleology,itdoesnotfollowthatDarwinhasdebunkednaturalteleologyaltogether.Platonicteleologyisonlyonesort;Aristotelianteleologyisanentirelydifferentsort.So,whatisthestatusofteleologyontheDarwinianscheme?Inwhatsenseisitanti-Aristotelian?ForAristotlethereareatleasttwoinstancesinwhichteleologicalexplanationsarecalledfor,toexplainfunctionalarrangements,whereanitemsexistenceisexplainedintermsofitsusefulness,andtoexplainregulargrowthpatternsthatmembersofaspeciesorgenusshare.IwouldliketoarguethatDarwinstheoryofnaturalselectionoffersanalternativetoteleologicalexplanationsforfunc-tionalarrangements.InwhatwayisDarwinstheoryofnaturalselectionanalternativetoAristotlesteleologicalexplanationoffunctionalarrangements?ForAristotle,theexplanationoffunctionalarrangementsdependsonthesametelosthatdeterminesgrowth.So,theexistenceofaCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n178andreariewusefularrangement(sharpteethinfrontandmolarsintheback,forinstance)isapropertyoftheformalfeaturesofgrowththeydevelopforthesakeoftheusefulnessofthearrangement.Recallthenotionofhypotheticalnecessity:thematerialexists(andhencethedevelop-mentoccurs)forthesakeofpreservingtheusefularrangement.Toputthepointdifferently:thepurposeexistsfirst;theprocessandmaterialsfollowthereafter.Darwinstheorydiffersinatleasttwoways.First,Darwinmakesadramaticdistinction,antitheticaltoAristotle,betweentheinternalprocessesthatgenerateanorgan-ismandexternalprocesses,theenvironmentthatdeterminesadaptability(Lewontin2000,42).ForDarwin,thedistinctionisabsolute:theinternalconditionsforgrowththatdeterminevaria-bilityandinheritancearenotresponsivetotheenvironmentalforcesthatultimatelyselectordeterminewhichorganismaltraitswilleventuallyprevailovertime.So,growthexplanationsarecompletelydistinctfromadaptationexplanations.Inexplaininghowspeciescometobesowelladaptedtotheirenvironmentalconditionsthereisnoneedtomentionthefactorsthatdeterminehowvariantsariseinthefirstplace;variantspreexisttheirselection.Second(andfollowingfromthefirstpoint),ratherthanfunctionsconstrainingthepresenceofmaterialsasitdoesforAristotleshypotheticalnecessity,forDarwin,itistheotherwayaround:materialsconstrainfunction.Individualvariantsaregeneratedbyadistinctinternalprocessandpreexisttheirselection.Theendresultistheexistenceofcontrivances,organswhoseprovenanceisancestralyetmodifiedbynaturalselectiontothenewlocalenvironmentalconditions.Thatiswhypandaspossesssuchanawkwardmechanismformanipulatingbambooshoots.Itisamodificationofthepawstructureofancestorsofpandas.Thepan-dasthumbisacontraption,modifiedfromtheanatomyofwhatwasavailabletobeoperateduponbyselection.Thislastpoint,Ithink,beginstoexplainAsaGraysremarkthatDarwinweddedMorphologytoteleology.ThereferencetoMorphologyistoaschoolofthoughtthatadvancedaunityofplantheoryoforganicdiversity.Accordingly,membersofataxonomicgroupareaccountedforintermsofresemblancesbetweenmembersofthesameandothertaxonomicgroups.TraitsthatresembleeachotheracrosstaxonomicgroupsarecalledhomologuesandindicateCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology179acommonplanthroughoutnature.Morphologiststhoughtthatpickingouthomologousstructuresconstitutedpickingoutessentialcategoriesinnature.Thatis,theexistenceofhomologousstructuresindicatesthefundamentallawsofbodyplans.However,Darwinwonderedhowtoexplaintheprevalenceofvariantstothecommonplans.Tothisheinvokesnaturalselection.Naturalselectionoper-atesoverpreexistingstructurescompetingforlimitedresourcesinacommonenvironment.So,whilestructurespreexisttheiradaptiveusesitisnaturalselection,adistinctivecausalmechanism,thatexplainsmorphologicalchange.ContemporaryphilosophersofsciencehavetakenforgrantedthatDarwinstheoryofnaturalselectionnaturalizesteleologicalexplanation.Yet,aswehaveseen,thereareavarietyofteleologicalexplanationsintheancientandearlymodernliterature,onlysomeofwhicharepreservedinDarwin.Inparticular,theAristotelianexplanationfortheexistenceoftraitsintermsoftheirusefulnessispreservedinDarwin,butwithoutreferencetothetelosthatdeter-minesindividualgrowth.PostDarwinwhenwesaythatbirdshavewingsbecausewingsareforflying,weunderstandthewhat-forexplanationintermsofanevolutionaryargumentwherebyposses-singwings(whichthemselvescametobebysomemechanismthataccountsforindividualvariation)providesafitnessadvantageandhencewingswereselectedandeventuallybecameprevalentinbirdpopulations(Enc¸andAdams1992,Mayr1988.ForcontrastingaccountsseeCummins2002,Lewens2004).So,theformoftheteleologicalexplanationremains,butthetermsoftheexplanationsarecompletelynaturalized:noreferencetothetelosofgrowthinsofarasgrowthisaconditionbutnotamechanismofselection.Forthisreason,ErnstMayrarguesthatabetterdescriptionofthenatureofDarwinsexplanationisteleonomyratherthantele-ology(Mayr1988).Darwinsnaturalselectionalsoprovidesusawayofnaturalizingthedistinctionbetweenfunctionandmalfunction.Wingsareforflyingexplainswhybirdshavewingsevenifanyparticularmemberofapopulationhasabrokenwing.ForAristotle,malfunctioningorgansarethosethatfailtoachievetheirfinalends,butforthisapplicationofDarwinstheory,malfunctioningorgansarethosethatdonotperformthefunctionforwhichtheywereselectedinthepast(Neander1991).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n180andreariewIhavejustarguedthatDarwinstheoryofnaturalselectionoffersusanalternativeexplanationtotheteleologicalarguments(bothPlatonicandAristotelian)fromfunctionalarrangements.Yet,asIhavediscussedbefore,Aristotleoffersusanotherreasontoinferatelos,intheexplanationofanindividualsgrowth.SincetherootofDarwinstheoryisanabsolutedistinctionbetweentheinternalconditionsofanindividualsgrowthandtheexternalconditionsthatdeterminewhetherornottraitswillbeselected,itshouldnotbesurprisingtousthatDarwinstheoryissilentontheconditionsthatdeterminegrowth.KarlvonBaercritiquesDarwinonthisverypoint(Lenoir1982,270).Accordingtothiscritique,ifblindnecessityistheonlyforceoperating,thenthefundamentalquestionsofbiologydevelopment,adaptation,andthelikewillremainunintelligible.Anexplanationthatstringstogethermechanicalprocesseslacksthefundamentalprinciplethatconnectstheprocessestoaparticularend(Lenoir1982,271).IinterpretvonBaerscriticismtobeclosetoAristotlesargumentfromregularity:thematerialistlackstheprinciplethatdistinguishesonematerialprocessfromanyother.Consequentially,whatdistinguishesdevelopmentalprocessesthatleadtolivingnewbornfromonethatfails?5.conclusionAttheoutsetofthisessayIclaimedthatthecontemporaryargumentfromintelligentdesignvarieslittlefromWilliamPaleysargumentwrittenin1802.Botharguethatnatureexhibitstoomuchcom-plexitytobeexplainedbymindlessnaturalforcesalone.Bothconcludetheneedtopostulatetheexistenceofanintelligentdesigner,acreatorwithforethoughtandpurpose.Buttherearedif-ferencesbetweenPaleysargumentandthemodernargumentfromID.PaleyconcludedthattheChristianGodexists,whilemodernIDsupportersclaimtobesilentaboutthefeaturesofthecreator.WhilePaleysinterlocutorswerematerialistsofthesortdescribedbyPlato,thetargetofIDsargumentisCharlesDarwin.TheclaimisthatDarwinstheoryofnaturalselectioncannotaccountforirreduciblycomplexnaturaldesignssuchastheflagellum,sinceitsbuildingcouldhaveonlybeendoneinonestep(givenitsincredibleintricacy),notagradualprocessofselection.SomecriticsofIDpointoutthatbiologyalreadyhasexamplesofnaturalselectionsputtingcomplexCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nTeleology181designsinplace(Ruse2003,320,discussestheKrebscycle).Otherspointoutthatelaboratesystemsmightevolveforsomereasonandthenbeco-optedforanentirelydifferentfunction:Whosaysthosethirtyflagellarproteinswerentpresentinbacterialongbeforebac-teriasportedflagelli?Theymayhavebeenperformingotherjobsinthecellandonlylatergotdraftedintoflagellum-building.Indeed,theresnowstrongevidencethatseveralflagellarproteinsonceplayedrolesinatypeofmolecularpumpfoundinthemembranesofbacterialcells(Orr2005).Stillothersargueagainstthecriteriaofhowtodetectnaturaldesigns(Fitelson,Stephens,andSober1998).Thecriticalpartofthepresentessayhasaconsistenttheme,pointingoutargumentinsufficiencies.JustasIarguedthatDarwinsandSmithsnoncreationistargumentsareinsufficienttonegateanAristotelianteleology(althoughDarwinstheoryeventuallydoes),IalsoarguedthatevenifDarwinsversionofnaturalselectiondoesnotexplainhowcomplexitemscometoexistinthefirstplace,IDdoesnotnecessarilyfollow.AllthatisrequiredtodefeatIDorcreationismistodemonstratethepossibilitythatcomplexdesignscanemergefromnaturalisticprocesses.ACKNOWLEDGMENTSInthisessayIhavereliedonanearlierwriting,Ariew2002.Iwouldespe-ciallyliketothankDanielGoniprowandPaulDiMartinoforcommentsonanearlierdraft.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nkimsterelny10Macroevolution,Minimalism,andtheRadiationoftheAnimals1.minimalistmodelsofmacroevolutionPalaeobiologyisourmainsourceofdirectevidenceaboutthehistoryoflife.Butwhilethathistoryisfascinatinginitself,palaeobiologysmostdistinctivecontributiontoevolutionarytheoryistheinsightitprovidesontheimportanceofscale.Palaeobiologistsseetheresultsofevolutionaryprocessessummedoverhugesweepsofspaceandtime.Asaconsequenceofthatwindowontheeffectsofdeeptimeandvastspace,wehaveachancetoseewhetherthepalaeobiologicalrecordenablesustoidentifyevolutionarymechanismsthatareinvisibletocontemporarymicroevolutionarystudieswiththeirlocalspatial,temporal,andtaxonomicscales.Palaeobiology,inotherwords,isthedisciplineofchoiceforprobingtherelationshipmacroevolutionarypatternsandmicroevolutionaryprocesses.Thischapterwillbeorganisedaroundanimportantframingidea:thatofaminimalistmodelofthisrelationship.Ishalldiscussminimalismindetailshortly,butasaroughfirstapproximation,accordingtominimalism,macroevolutionarypatternsaredirectreflectionsofmicroevolutionarychangeinlocalpopulations;theyarereflectionsofchangesofthekindwecanobserve,measure,andmanipulate.Forexample,MichaelBenton(forthcoming)discussesmodelsofglobalspeciesrichnessthatdependonscalingupinspaceandtimeequilibriummodelsoflocalecologicalcommunities.Totheextentthatsuchscaled-upmodelsareadequate,macroevolu-tionarypatternsarenothingbutlocalchangessummedovervastsweepsofspaceandtime.Thischapteraimstomakeexplicitthe182CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation183patternsforwhichthisminimalistmodelisappropriate;toconsiderthequitedifferentwaysinwhichthatmodelcanbeenriched;andtohighlightsomephenomenathatsuggestthatminimalismdoes1indeedsometimesneedtobeenriched.Thechallengetominimalismisthatscalematters:thereareaspectsofthehistoryoflifethatarenot(solely)theresultofpopula-tion-levelprocessesaccessibletoneobiologicalinvestigation.Becausescalematters,palaeobiologytellsussomethingwecannototherwiseknowaboutevolutionaryprocesses.Butwhenisamacroevolutionarypatternjustamereaggregateoftheresultsoflocalprocesses?AsIseeit,minimalmodelsaresimple,perhapssimplifiedmodelsoftherelationshipbetweenmicroevolutionaryprocessandmacroevolu-tionarypattern.Buttheyaresimpleinfourindependentaspects.i.First:theyareindividualist.Inthesemodelsofevolutionarychangethefitnessvaluesthatmatterarefitnessesofindividualorganisms.Onemovebeyondminimalismistodevelopevolu-tionarymodelsinwhichspeciesthemselvesareselectableindivi-dualsinapopulationofspecies.Intherecentpalaeobiologicalliterature,thisideahasbeencentrestage;agooddealhasbeenwrittenonhowtocharacteriseandempiricallytestspeciesselectionmodels(seeforexample,Vrba1989,GouldandLloyd1999,Sterelny2003).Butwhilespeciesselectionmightbeimportantinexplaining,forexample,patternsofsurvivalinmassextinctionevents,thecaseforitsimportanceremainstobemade.Incontrast,asweshallseeinconsideringtheoriginsandelaborationofmulticellularity,someformofgroupselectionisalmostcertainlyimportantindrivingmajortransitionsinevolution.ii.Second,extinctionandspeciationprobabilitiesareeffectsofindividual-organismfitness.Thespeciationandextinctionprob-abilitiesofaspeciesstandinasimple,directrelationshiptoselec-tiononindividualsinthepopulationsofwhichthespeciesiscomposed.Forexample,ifitsextinctionprobabilityishigh,itishighbecauseindividualorganismsarenotwelladaptedbycomparisontotheircompetitors,notbecause(say)thespecieslacksgeneticvar-iationorbecauseitlacksmetapopulationdynamicsinwhichmi-grationbuffersindividualpopulationsagainstlocalextinction.Minimalistmodelsidealiseawayfromtheevolutionarycon-2sequencesofspecies-levelproperties.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n184kimsterelnyiii.Third,novelties,keyinnovations,andthelike,ariseasordinaryvariationsinextraordinarycircumstances.Someevolutionaryinno-vations(forexample,theinventionofsex)expandthespaceofevolutionarypossibilityopentoalineage.Butthereisnothingun-usualaboutthegenetic,development,orselectiveoriginofthoseinnovations.Noveltiesariseandareestablishedasordinary,smallvariantsfromancestralconditions.Possibility-expandingchangesare,inDennettshelpfulphrase,retrospectivecoronations(Dennett1995).Dawkinshasarguedthattheoriginsofphylaareretrospectivecoronations:thegreatbranchesofanimallifebeganwithordinaryspeciationevents;nothingaboutthosespeciationsatthattimemarkedthosebranchingsasofespecialimportance,thoughitturnedoutthattheywereimportant.Likewise,lookingbackwards,wecanseethat(say)theevolutionofflowerswasapossibility-changingkeyinnovation.Itledtoanecologicalandevolutionaryrevolutioninplantcommunities.Buttheimportanceofthisinnovationinpolli-nationwouldnothavebeenidentifiableatthetime.Keyinnovationsaregenuinelyimportant.Buttheirimportancecanonlyberecognisedretrospectively:akeyinnovationisonethathappenstotakeplaceintherightplace,time,andtaxon.Itsoriginandestablishmentinthattaxonarenottheresultofanyunusualevolutionaryprocess.Wemoveawayfromminimalistmodelsifwethinkthattheoriginorestablishmentofnovelties(sometimes)requiresspecialexplanation.iv.Fourth,wecanidealiseawayfromthechangesinthedevel-opmentalandselectivebackgroundofgeneticvariationinpheno-types.Wecantreattheselectiveanddevelopmentalbackgroundofchangeasafixedbackgroundcondition.Inmicroevolutionarystudiesforexample,modelsoftheresponseofguppiestosexualandnaturalselectionwecanusuallytreatthedevelopmentalsystemasastablebackgroundconditionofwithin-populationmicroevolutionarychange.Wedothiswhenwetreatgenesasdif-ferencemakers:thesubstitutionofonealleleforanothermakesaselectablephenotypicdifference:it(say)makesamaleguppybrighter.Genesaredifferencemakers,butonlyifweholdfixedthecausalbackgroundinwhichtheyact.Insuchrestrictedcontexts,theconceptofagenotype-phenotypemapandtheseassociatedcausalclaimsmakesgoodsense.Obviously,whenourinterestturnstomacroevolutionarypattern,thisassumptionbecomesmuchmoreproblematic;modelsthatmakethesestabilityassumptionsidealiseCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation185radically,forselectiveenvironmentanddevelopmentalsystemarelabileondeepertimeframes.Wecannotextrapolatefromcon-straintsonshort-termevolutionaryresponsestosimilarconstraintsonlong-termevolutionaryresponses.Understoodthisway,minimalismisamodel,notadoctrine.Everyonewouldacceptthatthefourelementsofminimalismideal-iseawayfromsomeofthecomplexitiesofthebiologicalworld.Noonesupposesthatselectiveanddevelopmentalenvironmentsareinvariant.Almosteveryonewouldconcedethatselectioncanactoncollectivesofindividualsorthatlarge-effectmutationsmightoccasionallybeimportant.Theideaofminimalismisthatitistypicallyfruitfultoidealiseawayfromthesecomplications.Soinsteadofthinkingofminimalismasadoctrinetobedefendedorundermined,weshouldinsteadfocusonidentifyingtherangeofcasesforwhichminimalmodelsareappropriate,andthosecasesinwhichthesemodelsneedtobesupplemented.Aswerelaxthesimplepictureoftherelationshipbetweenpopulationleveleventsandspeciesdynamics,wethusgetaspaceofmodels.Ourproblembecomesoneofidentifyingtheevolutionaryphenomenaforwhichmodelsindifferinglocationsinthatspaceareappropriate.2.fourviewsoflifeToshowthatminimalismisfruitful,weneedtoshowthatwecandevelopinsightfulminimalmodelsofmajormacroevolutionaryphe-nomena;thatwecanexplainthetreeoflifesmoststrikingfeatures.Toshowthelimitsofminimalism,weneedtoidentifythosepalaeobio-logicalphenomenaforwhichminimalmodelsarenotadequate.Thatisbeyondthescopeofanysinglechapter,notleastbecausethereisvigorousdisagreementaboutthephenomenatobeexplained.Giventherichnessofpalaeobiologicalphenomena,andthevigourof3palaeobiologicaldebate,mystrategywillbeillustrativeratherthanexhaustive.Ishallbeginwithfourrepresentativeexamplesofoverallviewsoflifeshistory,andaninitialcharacterisationoftheirrela-tionshiptominimalistmodels.Twoofthemseemtofitminimalmodels;theothertwosuggestthatweneedextensionsofthosemod-els.Ithenexploreinmuchmoredetailonecrucialandcontroversialepisodeinthehistoryoflife,theearlyradiationoftheanimals.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n186kimsterelnyView1:TheSpreadofVariationInhis(1996),StephenJayGouldarguedthatwhiletherereallyhasbeenariseinboththemaximumandmeancomplexityoflivingagents,thisfactofhistoryisbestunderstoodasanexpansionofvar-iationincomplexity.Ifprocessesofdifferentiation,speciation,andextinctionactindependentlyofcomplexity,variationincomplexitywillnonethelessincreaseovertime.Ifwegraphchange,withcomplexityonthehorizontalaxisandtimeonthevertical,variationwillspreadbothtotheleftandtotherightfromthepointoflifesorigin.However,evenifdifferentiation,speciation,andextinctionareindependentofcomplexity,thespreadneednotbeequalinbothdirections,forlifescomplexityhasaminimumbound(setbythebiomechanicsofmetabolismandreplication)butnoupperbound.Moreoverlifeoriginatednearthisminimumbound.Sovariationwillspreadtotheright,inthedirectionofgreatercomplexity,butveryslightlyifatalltotheleft.Thusmaximumandmean(butnotnecessarilymodal)complexitywilldriftupwardsovertime.Givenaminimumboundandapointoforiginclosetothatbound,thenullexpectationisanincreaseinmaximumandmeancomplexity.Soevenabiasedspreadofvariationneednotchallengeminimalism.Thelocalevolutionaryprocessesofadaptationofpopulationstotheirspecificlocalcircumstances,perhapssignificantlymodifiedbydrift,andbydevelopmentalandgeneticconstraints,wouldgenerate4spreadingvariation.View2:EscalationandArmsRacesOnedividewithinbiologyisbetweenthosewhoemphasisetheimportanceofinteractionsbetweenbiologicalagentsandthosewhoemphasisetheimportanceofabioticfactorsforthelifeoforganisms.Manyevolutionarybiologistshaveemphasizedhostilecoevolu-tionaryinteractionsbetweenlineages:armsracesbetweendifferingcompetitorsforthesameresource;betweenpredatorsandprey,plantsandherbivores,hostsandparasites.Forexample,GeeratVermeijhasdefendedanescalationmodelofthehistoryoflife,takingashismodelinteractionsbetweenbivalvesandtheirpred-ators.Ashereadsthelonghistoryofbivalveevolution,itisoneofthegradualimprovementofdefence.BivalveshaveevolvedtheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation187capacitytodigdeeperandfasterintothesubstrate,andtheyhaveevolvedthicker,spikiershells.Predatorefficiencytoohasrampedupovertime.Predatorshavebecomebetteratdiggingup,drillinginto,crushing,orbreakingopenshells.Vermeijthinksofthesebivalvehistoriesasindicativeoflifeshistoryasawhole:organismsbecomebetteradaptedovertime(Vermeij1987,1999).ThinkingofthehistoryoflifeasanescalationofadaptationdrivenbyarmsracesistothinkofthathistoryinamoreselectionistwaythandoesGould.Escalationscenariospresupposeboththatselectiontendstodriveevolutionarydynamicsandthatthereisasystematicbiasinthedirectionofselection.Butthisideaiscom-patiblewithoveralltrendsinevolutionandecologyreflectinglocallydeterminedevents.Thecausalengineofescalationiseco-logicalinteractioninlocalcommunities.Escalationisaminimalmodelofthehistoryoflife,thoughonewhoseempiricalpre-suppositionsmaynotbemet.Externaleventscaninterrupttheassociationbetweenlineages,breakingcoevolutionaryconnections.Forexample,thechangingbiogeographyandclimateofAustraliamightchangethesuiteofinsectstowhichtheeucalyptlineageisexposed,thusabortingarmsracesbetweeneucalyptsandphytophagousinsectlineages.View3:TheIncreasingSpaceofEvolutionaryPossibilityWhenweconsiderlifeataparticulartime,weshouldseeitashavinganupperboundaswellasalowerbound(Sterelny1999d,KnollandBambach2000).Forexample,theevolutionoftheprokaryoticcellhadsignificantevolutionarypreconditions.Untilgeneswereorga-nisedintochromosomes,andthefidelityofgenereplicationimproved,prokaryoticcellswerenotinthespaceofevolutionarypossibility.Thereisanupperlimitonthecomplexityofquasi-bioticsystemsthatlackadivisionoflabourbetweenmetabolismandreplication.Thesameistrueofothergradesofcomplexity.Forexample,NicoleKinghaspointedoutthatonlytheevolutionofsignallingandcelladhesionmechanismsinprotistsmadepossibletheevolutionofmulticellularlife(King2004).Atatime(touseGouldsmetaphor)thereisawalltotherightaswellasontheleft.Overtime,though,thatwallshifts,forthepreconditionsforanewCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n188kimsterelnygradeofcomplexityandofdifferentiationareassembled.Thespaceofevolutionarypossibilityforlifeasawholeexpands,becausetherightwallmovesforsomelineages.Asitexpands,someofthosepossibilitiesarerealised.Thebest-knownmodelofthiskindisthatdevelopedbyJohnMaynardSmithandErsSzathmary(MaynardSmithandSzathmary1995,1999).Theyseeevolutionaryhistoryascharacterisedbyaseriesofmajortransitions.Theseincludetheshiftfromindepen-dentlyreplicatingstructurestotheaggregationofcodependentreplicatorsintochromosomes;theshiftfromRNAasthecentralreplicatortoDNAreplication;theevolutionoftheeukaryoticcell;theinventionofcellulardifferentiationandtheevolutionofplants,animals,andfungi;theevolutionofcolonialandsocialorganismsfromsolitaryones;andeventheevolutionofhumanlanguage.Manyofthesetransitionshavetwocrucialcharacteristics:(i)theyarerevolutionsofbiologicalinheritancesystems,involvingtheexpan-sionofthetransmissionofheritableinformationacrossthegen-erations,and(ii)theyarerevolutionsinselection,fortheyinvolvetheassemblyofanew,higher-levelagentoutofpreviouslyinde-pendentagents.Minimalistmodels,tradinginthefitnessonlyofindividualorganisms,donotseemwellsuitedtogiveanaccountofsuchrevolutionsinselection.Yet,jointly,theserevolutionsleadtoanexpansionofthespaceofbiologicalcomplexity.Sothisviewofevolutionaryhistorydoesraiseissuesaboutminimalism,foritsuggeststhatminimalistmodelscannotgiveanadequateaccountofthedistinctionbetweentwoverydifferentkindsofevolutionarychange:theexpansionofpossibilityversustheexplorationofpossibility.Standardmicroevolutionarytheoryseemstobeaboutpossibility-exploringchange,notpossibility-expandingchange.Thissamedistinctionbetweenpossibility-expandingandpossibility-exploringinnovationsseemstobeimplicitinournextpictureoftheoverallpatternoflifeshistory,too.View4:ExpandingEcospaceAndrewKnollandRichardBambach(2000)developaviewofevolutionarytransitionsthatfocusesontheecologicalchangesthatareconsequencesofmorphologicalinnovations;thusthisposessimilarchallengesinunderstandingtheoriginandestablishmentofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation189novelty.ForKnollandBambach,thecrucialpatterninthehistoryoflifeisanexpansionofthehabitatsorganismsexploit,anexpansionoftherangeofresourcesorganismsexploitwithinthosehabitats,andanexpansionofthecomplexityandvarietyoftheecosystemsthatareassembledasaresultoftransitionsinmorphologicalcom-plexity(KnollandBambach2000).Indevelopingtheirmodeloftheexpandingecospace,KnollandBambachidentitysixmegatrendsthatmaproughlyontothemajortransitionsofMaynardSmithandSzathmary:trendsthatyokemorphologicalinnovationtoecologicalrevolution.Theseare(i)thetransitionfromtheprebiostolifeasweknowit,(ii)theradiationoftheprokaryotes,(iii)theprotistradia-tion,(iv)theevolutionofaquaticmulticellularity,(v)themulti-cellularinvasionoftheland,(vi)theevolutionandglobaldispersalofintelligence.Intheirview,thesetrendsallinvolveecologicalrevolutions.Theradiationoftheprokaryotesexpandedtherangeofenergysourcesintowhichlifecouldtap.Theevolutionoftheprotists,accordingtoKnollandBambach,addeddepthtoecosystems,foritinvolvedtheevolutionofpredation.Asprokaryotesmostlyextractedenergyfromabioticsources,thestructureofbacterialecologicalcommunitieswasfairlysimple.Eukaryotescanengulfparticles,includinglivingones,andhencetheyhaveaddedlayerstoecologicalcommunities,increasingtheirverticalcomplexity:grazing,predation,decomposi-5tion,areeukaryotespecialties.Theevolutionofmarinemulti-cellularitywasobviouslyaverydramaticmorphologicaltransition.But,equally,itwasanecologicaltransformationaswell.Thistransitionresultedingreatlyincreasedverticalcomplexityofcom-munities,andgreatlyincreasedecologicalengineering.Forexample,withtheradiationofmarineanimalsintheCambrian,thenatureoftheseafloorchanged.Insteadofsedimentsonthefloorbeingstable,theyweremixedbyanimalsburrowingthroughthem(thisisknownasbioturbation).Coralsandothershelledformscreatedsubstrateandhabitatforotherorganisms,asdidkelpforests(andmuchmorerecently,seagrasses).Richcoevolutionaryinteractions(bothsym-bioticandparasitic)betweenmulticelledandsingle-celledorgan-ismsbecamepossible.Theinvasionoftheland,likewise,wasanecologicalrevolution:awholearrayofphysicalenvironmentsbecamehabitats.Asaresultofthatinvasion,therearenewkindsofcommunitiesandnewkindsofcoevolutionaryinteractions(mostCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n190kimsterelnystrikingly,thosebetweenfloweringplantsandinsects).Thedirec-tionofevolutionismarkedbytheestablishmentofsuccessivelymorecomplexecologieswithmoreandmoreverticallayers.Eukaryoteswereaddedtothetopofabacterialfoundation,andvariousmulticelledlayerswereeventuallyaddedtopersistingthoughdoubtlesstransformedprokaryote-protistsystems.IfGeeratVermeijisright,theseincreasinglycomplexcommu-nitiesarealsoincreasinglydominatedbyenergy-greedyorganisms:organismsthatharvesttheavailableenergyateverhigherrates,andasaconsequencehaveincreasingimpactontheirlocalenviron-ments(Vermeij1999).Inhisview,boththeradiationoftheflow-eringplantsattheexpenseofthegymnospermsandtheradiationofmammalsandbirdsattheexpenseofamphibiansandreptilesexemplifythistrend(Vermeij1999).Ecosystemshavebecomeincreasinglydominatedbythesehigh-activityorganisms.Attimes,theseactasdefectorsinatragedyofthecommons.Theyharvestmoreresourcesthanothersatthesametrophiclevel,thussuckingresourcesoutofthesystematincreasingrates,fortheirnumbersexpandatothersexpense.Butbecausetheyusesomuchenergy,theyexertmorepoweroverthelocalenvironment,increasingtherateatwhichenergyandnutrientscyclethroughlocalecosystems.Theiroveralleffectistorampupthepaceoflife,therebyfurtherselectingforagentswithsimilarlylargeenergyappetites.Inthenextsection,Ilinkminimalismtoacrucialcase:theradiationoftheanimals.InSection4IpresentIhopeanear-consensusviewofthelarge-scalehistoryofthatradiation.InSections5and6,Idiscussevolutionaryexplanationsofthatradiationandtherelationshipbetweenmacroandmicroevolution.Ithenverybrieflyconclude.3.majortransitions:achallengetominimalism?Intheremainderofthischapter,Ishallexploreminimalismthroughconsiderationofmajortransitionsinevolutionand,inparticular,theradiationofmulticelledanimallife.Ishallsuggestthatthesetransitionstakeusbeyondminimalistmodelsofevolutionforthreereasons.First,asmuchofthemajortransitionsliteratureempha-sises,amultilevelperspectiveisessentialtounderstandingtheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation191selectiveregimesresponsibleformovementsoftherightwall.Transitionsincomplexityhaveofteninvolvedindependentagentscomingtoshareacommonevolutionaryfate.Thesetransitionsinvolveatrajectoryfromapopulationofinteractingagentstoamorecomplexcollectiveagent.Perhapsinafewcases(conceivably,thefirsteukaryote)thistransitiontookjustasinglestep.Butinmostcasestherewas(andoftenstillis)atransitionalregimeinwhichthefitnessoftheincipientcollectiveandthefitnessofitsnowsemi-independentelementswerebothimportant.Inunderstandingsuchtransitionalregimes,onecrucialproblemistounderstandthemechanismsthatpreventthefunctionalorganizationofthenewcollectivefrombeingunderminedbyselectiononitscom-ponentsfordefectionandfree-riding.Soonethemeinvolvestheinteractionbetweenlevelsofselection,andtheprocessthroughwhichselectiononthecollectiveresultsinthecomponentsofthatcollective(moreorless)ceasingtobeDarwinianpopulations(Buss1987,Michod1999).Asecondchallengeinvolvestheevolutionofnoveltyandtheexpansionofevolutionarypossibility.Thepossibilityspaceacces-sibletoalineagedependsinpartonitscurrentlocationinmor-phospace.Alineageinwhich(say)thearthropodadaptivecomplexhasevolvedhasevolutionarypossibilitiesopentoitthatarenotopento(say)velvetworms:segmentedanimalsbutwithoutthestructuralcomplexityandskeletalsupportofthearthropods.Butaccesstopossibilityspacedependsaswellonthemechanismsofinheritanceanddevelopmentthatcharacterisealineage,onthevariationsfromcurrentlocationthatarepossible.Developmentalmechanismsmakesomeregionsofspacemoreaccessiblefromalineagescurrentlocation,andotherslessaccessible(seeforexam-ple,Arthur2004,SchlosserandWagner2004).AsIhavenoted,minimalistmodelstreattheevolutionofnoveltiesasaspeciesofretrospectivecoronations.Noveltiesariseandareestablishedinlocalpopulationsthroughmechanismsthatareindistinguishablefromthosedrivingordinarymicroevolutionarychange.Onepoten-tialproblemforminimalistmodelsofnoveltyistheroleofmacro-mutation(i.e.,mutationswithdramaticphenotypiceffects)inevolution,butinmyview,theimportanceofmacromutationforminimalismhasbeenoverstated.Noonethinksthatmacromuta-tionsareimpossible.EvensomeoneasscepticalasDawkinsaboutCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n192kimsterelnytheroleinevolutionofmacromutationallowsthattheyhaveprobablyplayedsomeroleinexpandingthespaceofevolutionarypossibility,perhapsintheevolutionofsegmentation(Dawkins1996).Butevenifraremacromutationaleventshaveplayedanimportantroleinexpandingthespaceofpossibility(asisquitelikely),macromutationsarecertainlynotcommonenoughtobethe6normalexplanationoftheoriginofnovelty.Henceinmodellingtheevolutionofnoveltyandtheexpansionofpossibility,itmaywellbereasonabletoneglectthiscomplication.Inassessingminimalism,itisimportanttoavoidafalsedichot-omy.Ontheminimalistmodel,apossibility-expandingnoveltyisanordinaryvariantcrownedretrospectively.Onewayofbeingextraordinaryistobetheresultofamacromutation.Butthereareotherandmoreimportantways,forevolutionarypossibilityismulti-factorial:itdependsonalineagescurrentpositioninphenotypespace,thearrayofpotentialvariationsintheheritabledevelop-mentalresourcesavailableinthatlineage,andinvariationsinthewaysgeneticandotherdevelopmentalresourcesareused.Andwhiletherearegoodreasonstothinkthatsuddenmacromutationalshiftsinmorphospaceareindeedveryrare,thosereasonsdonotgeneralisetotheotherfactorsonwhichevolutionarypossibilitydepends.Forexample,inSterelny(2004)Iarguedthattheformationofsymbioticalliancesofteninvolvesmajorshiftsinevolutionarypossibility.TheacquisitionofmicrobialsymbiontshasgivenmanyMetazoansaccesstolifewaysthatwouldotherwisebeclosedtothem.Andwhiletheevolutionaryoriginationofsymbioticalliancesisnotaneverydayfeatureofthebiologicalworld,itisnotvanishinglyrare,either.Possibility-expandinginnovationsmaybechangesinthecontrolofdevelopment,andtheseneednotbeminimalvariationsofpriorsystemsofdevelopmentalcontrol.Thecrucialpointhereisthatthemechanismsofdevelopmentalplasticitycancushiontheimmediatephenotypiceffectofsignificantchangesinthedevelopmentalsys-tem.Thatiswhylargechangesinthegenesthemselveschromo-someinversions,duplication,andthelikearenotalwaysfatal.Significantdevelopmentalchangesneednotresultinlarge(hencealmostcertainlycatastrophic)jumpsinmorphospace(West-Eberhard2003).ThepowerofthesemechanismsisillustratedbyWest-Eberhardsextraordinaryexampleofthegoatbornwithoutfrontlegs.Itadoptedatwo-leggedpostureandmovedinsomewhatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation193kangarooishfashion.Theadjustmentwasnotjustbehaviouralbutmorphological.Thegoatdevelopedwithchangedhindlegandpelvicstructures,acurvedspine,strongneck,andassociatedmuscularchanges.Possibility-expandingchangesinalineagemayoftenhavetheirorigininicebergmutations:variationsthatarenotradicallydiffer-entinphenotypefromtheirancestorsandcontemporaries,butthatgeneratethatphenotypefromanimportantlychangedsetofdevel-opmentalresources.Onerelativelyuncontroversialexampleofapossibility-increasingchangeindevelopmentisonethatincreasesmodularity.Ifsomeaspectsofphenotypeareundermodulardevel-opmentalcontrol,theywillberelativelymoreevolutionarilylabile(Lewontin1985,Kauffman1993).Butthereareotherexamples:forexample,MarkRidleyarguesthatmorphologicalcomplexityisconstrainedbythefidelityofinheritance.Theerrorratecharacter-isticofprokaryotereplicationwouldbefataltomulticellularani-malswiththeirlargergenomes;theirevolutiondependedontheevolutionofamoreaccuratesystemoferrorcorrection(Ridley2000).InSection6,wewillconsiderinsomedetailtheclaimthattheevolutionofcomplexanimalbodiesdependsontheevolutionofnewmechanismsofgenecontrol.Athirdchallengeistheintegrationofinternalandexternalfactorsinexplainingtransitions.Selection-drivenmicroevolutionisoftenconceivedasahill-climbingprocess:apopulationfindsitselfsub-optimallylocatedinanadaptivelandscapeandrespondstothatlocationbyoptimisation.AsRichardLewontinhaspointedout,evenwhendoingmicroevolutionarystudiesweoftenshouldnotthinkofenvironmentsasfixedandorganismsaslabile(Lewontin1985,Odling-Smeeetal.2003).Inevolutionarytransitions,selectiveenvironmentsanddevelopmentalpossibilitiesarelabile,sosuchanidealisationislesslikelytobeappropriateforpossibility-expandingtransitions,suchastheinventionandelaborationofmulticellularity.ItistothattransitionInowturn.Thereisnosingletransitiontomulticellularity;instead,thereisaclusteroftransitionsthattookplaceatdifferenttimesandtodifferentdegrees.Bonners1998reviewnotes,inadditiontotheobviousmulticellularcladesofplants,fungi,andanimals,thegreen,red,andbrownalgaeandavarietyofothermoreexoticcases(Bonner1998).Inthesemoreexoticcases,multicellularityhasbeenestablishedwithoutmuchCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n194kimsterelnydifferentiation(asinthecaseofthealgae).Indeed,insomeofthecasesitisnotclearwhetherweshouldthinkofthesesystemsasasinglemulticelledagentorasasocial,cooperativepopulationofsingle-celledagents(asincellularslimemoldsandmyxobacteria,whichaggregatetoformstalkedstructureswithsporesatthetop).Ishallfocusonjustoneofthesetransitions,theMetazoanradiation,foritisanespeciallyimportantcase.Somystalkinghorsewillbetheearlyevolutionofcomplexanimallife.4.themetazoanrevolutionTheCambrianExplosionnamestheradiationofanimallifeintheearlytomid-Cambrian,fromabout543millionyearsago(mya)toabout505mya.TheanimalsoftheCambrianwerenottheearliestmulticelledanimals.TheywereprecededinthefossilrecordbyanenigmaticEdiacaranfauna:anarrayofdiscoidandfrond-shapedformswhoserelationsbothtooneanotherandtolivinganimalsremaincontroversial(Narbonne2005).ButtheCambriansawthefirstappearanceofundoubtedancestorsofcontemporaryanimals.Bythenthefirstbivalves,arthropods,echinoderms,molluscs,andchordateshadcertainlyevolved.TherearehintsofsuchanimalsbeforetheCambrian,intracefossilsandfossilembryos.Butbythemid-Cambrian(about530mya),theywererichlypresent.Moreover,thisexplosionwasofenormousconsequence.Incontrastto(say)theevolutionofmulticelledredalgaeandthefewmulticelledlineagesofciliatesanddiatoms,theevolutionofthemetazoachangedtheevolutionaryandecologicallandscape.Arguably,thistransitionwasfastandvast.Itresultedinahighlydisparate,taxon-richclade.Itwastheinventionnotjustofmulticellularitybutofthecontrolofdif-ferentiationandafulldivisionoflabour.Themetazoavarywidelyamongstthemselves,yetarecharacterisedbycomplexthoughreli-abledevelopmentalpathways,involvingmanycelldivisionsanddifferentiationintomanycelltypes.Allbuttheearliest(orthemostsecondarilysimplified)Metazoansareverticallycomplex,withcellsorganisedintotissues,whichinturnareorganisedintoorgansandorgansystems.Manyhavecomplexlifehistoriesinvolvingradicalchangesinphenotypeovertime:manyMetazoa,inotherwords,areadaptednotjusttoasingleenvironmentbuttodifferentenviron-mentsatdifferentstagesoftheirlifehistory.InmanyanimalcladesCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation195thereisafundamentaldivisionoflabourbetweenreproductionandinteractionwiththeenvironment,forintheselineagesthereisanestablishedsoma/germlinedistinction,withtheearlysequestrationindevelopmentofthosecellsthatwillbethefuturegametes.SotheMetazoanradiationisnotjustatransitiontomulti-cellularity.Itisatransitiontostructurallyandbehaviourallycom-plexagents.Theevolutionofsuchagentsdependsontheevolutionofthereliabledevelopmentalcontroloflargenumbersofdiffer-entiatedcellsandthecomplexestheymake.Theirdiversitydependsontheproliferationofavastnumberofdistinctdevelopmentalcontrolsystems,andthereliabletransmissionofthosecontrolsys-temstothenextgeneration.Moreover,theinventionandradiationoftheanimalswastheinventionoffullytreelikeevolutionaryhis-tories,withlimitedhorizontalgeneflowand(comparatively)well-definedspeciestaxa.Notoriously,thesystematicvocabularythathasbeendevelopedtodescribeanimallineagesfitsotherradiationslesswell.Moreover,theMetazoahavebeenprofoundagentsofecologicalchange.TheradiationoftheMetazoaestablishedwhollydifferentkindsofecologicalcommunitiesbasedonwebsoforgan-ismsthatlivebyeatingotherorganismsandtheirproducts.Thesecommunitiesprofoundlychangedthephysicalenvironmentinwhichorganismslived.DavidBottjer,forexample,haswrittenoftheCambriansubstraterevolution:Cambriananimalsprofoundlychangedthephysicalsubstrateofshallowseafloors.Beforethisradiationthesewerestabilised(andhenceavailableashabitatfortheEdiacaranbiota)bymicrobialmats.Onceburrowinganimalsradi-ated,thesematswerebrokenup,andthemudsbecameunstable.Thesoftseabedswerenolongeravailabletosessileorganismsthatlackedspecialstabilisingadaptations(holdfastsofsomekind)forsuchseafloors(Bottjeretal.2000).Likewise,theinventionoftheturdrevolutionisedthechemistryofoceanwaters:itpackedwastechemicalsinaformthatwasheavyenoughandcompactenoughtosinktotheseafloorratherthandispersesuspendedinthewater(Loganetal.1995).TheMetazoanradiationisanespeciallycentralcaseforunder-standingevolutionarytransitionsandthelimitsofminimalistmodelsofevolution.Indeed,ifthefossilrecordisatruerecord,wegofromseasandseafloorswithafewsimpleanimals(sponges,jelly-fish,perhapsafewwormlikebilaterians)toseasandseafloorsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n196kimsterelnyteemingwithanimallife.TheMetazoaappearfairlysuddenlyandrichlyintherecord.Doesthisrapiddiversificationinthefossilrecordreflectevolutionaryhistory,andifso,doesitshowthatunusualevolutionarymechanismswereresponsibleforthisradia-tion?Famously,inWonderfulLife,StephenJayGouldarguedforaffirmativeanswerstoboththesequestions,arguingthattheextraordinaryfossilsoftheBurgessShaleshowedthatanunprece-dentedlydisparateanimalfaunaevolvedextraordinarilyrapidlyinCambrianseas,andthereafterthemechanismsthatgeneratedsuchgreatdisparityshutdown(or,atleast,slowedverydramatically)(Gould1989).However,ourviewofthelifeintheCambrianhassincebeentransformed.ThereisreasontosuspectthatGouldsestimateofitsdisparityrestsonataxonomicillusion.GouldthoughtthattheCambrianfaunawasmoredisparatemoremorphologicallydiffer-entiatedthananysubsequentfauna.ManyoftheBurgessShalefossilsdidnotmatchthebodyplansofanyoftheextantgreatclades.Forexample,manyBurgessarthropodsdidnothavethecharacter-isticsoftrilobites;nordidtheyhavethebodyplancharacteristicsofspidersandtheirallies;norofinsectsandtheirallies,norofcrus-taceans.Butweshouldnotexpectearlymembersofalineagetofitbodyplanspecificationsusedtoidentifythelivingmembersofalineage(foraparticularlyforcefulstatementofthisargument,seeBuddandJensen2000).Livingcrustaceans(forexample)havelimbandsegmentationpatternsonethat,forexample,includestwopairedantennaethat(inallprobability)wereassembledincre-mentallyandhavethenbeeninheritedbythelivingcrustaceans.Thosetaxawiththatinheritedpattern,livingandextinct,arethecrowngroupcrustaceans.Ontheassumptionthatthislimb/segmentationpatternwasassembledgradually,therewillbestemgroupcrustaceans.Thesearetaxaonthelineagethatleadstothelivingcrustaceans,taxathatlivedafterthatlineagedivergedfromtheotherarthropodsandbeforethedefinitivecrustaceanpackagewasassembled.Stemgrouptaxaareboundtolookstrange.AndtheremusthavebeenstemgroupMetazoansaplentyintheCam-brian,forthatwastheperiodinwhichthegreatMetazoancladesweredivergingandacquiringtheirdistinctivemorphologies.GouldsextremeCambriandisparitypulsemaybenothingbuthis7encounterwiththisarrayofstemgroupMetazoans.TotheextentCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation197thattheCambrianchallengetominimalistmodelswasbasedontheideathatCambriananimalevolutiongeneratedextraordinaryamountsofdisparity,andthenthesupplyofnewdisparitydriedup,thischallengenowlookslesspressing.Evenso,ifthefossilrecordistobetrusted,therewasarapiddiversificationoftheMetazoafromtheearlytothemid-Cambrian.Butisthatrecordtobetrusted?Overthelasttenyearsorso,molecularmethodshaveservedasanindependentcheckontherelationshipbetweenthefossilrecordofearlyMetazoanevolutionandthetruepatternofthatevolution.AvarietyofmolecularclockshavebeenusedtocalibratethedivergencetimesoftheMetazoanphyla.Theideais(a)tocomparehomologous,slowlyevolvinggenesindifferentphyla,andcalculatetheextenttowhichthosegenesequenceshavedivergedonefromanother;(b)tocalibratetherateofevolutionusingtaxawitharich,reliable,andwell-datedfossilrecord;(c)tousethatratetocalibratedivergencetimesforother8taxa.Theearlierresultsofusingthesemolecularmethodswereverystrikingindeed:theypushedtheapparentdivergenceofthebasalMetazoanslongbeforetheCambrian.SomeofthesestudiesestimatedthedivergencetimesbetweenthespongesandotherMetazoans(andeventhelatersplitbetweenthecnidariansandthelineageleadingtothebilaterallysymmetricalMetazoa)asoveronebillionya(fordiscussionoftheseearlierstudies,seeBromhametal.1998,Lee1999).Whilethedivisionofalineageneednotimplythesimultaneousevolutionoftheirdistinctbodyplans,thesedatesimplyaverylongperiodofcrypticevolution.Iftheyareright,thechallengetoevolutionarybiologybecomesthatofexplainingwhyMetazoananimalevolutionwasinvisibleforsolong,andhowandwhyitsuddenlybecamevisible.However,morerecentmolecularclockestimatesofdeepMetazoansplitsaremorecongruentwithfossildates.OnedeepdivideamongsttheMetazoaisbetweentheprotostomeanddeu-terstomedevelopmentalpathways,andKevinPetersonandhiscolleaguesestimatethisdivergenceintherange573656mya(Petersonetal.2004).Theyarguethattheverydeepdivergencetimesdependedonusingvertebrateevolutiontocalibratetheclock;thisgaveamisleadingresultbecausetherelevantgenesinverte-bratesseemtohaveevolvedmuchmoreslowlythaninother9lineages,makingtheclockrunfasterthanitshould.SodivergenceCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n198kimsterelnytimesaredeeperthanthefossilrecordsuggests,thoughnothun-dredsofmillionsofyearsdeeper.Forexample,PetersonandNicholasButterfieldestimateMetazoanoriginsat664mya(PetersonandButterfield2005,9549).Ifthosedatesareright,Metazoansexistedforcloseon100millionyearsbeforetheyleftanunmis-takablerecordoftheirpresence.Ifthefirstanimalsweretiny,soft-bodiedelementsoftheplankton,thatinvisibilityisnosurprise.JamesValentinehasprovidedaveryhelpfultableofthefirstappear-anceinthefossilrecordofthevariousanimalphyla(Valentine2004,186).TheEdiacaranisthefinaleraoftheProterozoiceon;itimmediatelyprecedestheCambrian.Onlytwophyla(CnidariaandPorifera)haveafirstappearanceinthisEdiacaranera,whereasthereisalargeclusteroffirstappearancesintheCambrian.However,twelvephylahavenofossilrecordatall.Theseareallsmall,soft-bodiedanimals.SothereisnothingextraordinaryinthesuggestionthatsmallanimalscouldhavebeenpresentintheEdiacaranworld10withoutleavingafossilrecord.Moreover,whilethereisnodirectfossilevidenceofsurvivingMetazoanlineagesuntilabout570mya,perhapsthereisindirectevidence.PetersonandButterfieldarguethatancientMetazoanshaveleftanindirectecologicalsignature.TheyclaimthatthereisasignalofprotistsbecomingvulnerabletoMetazoanpredationatthebaseoftheEdiacaranperiod(i.e.,ataround635mya).Sponge-gradeorganisms,becauseofthebasicdesignoftheirbodies,cancaptureonlybacteriaandsimilarsizeparticles.Anervoussystemandagutareneededtocaptureeukaryotes:theseinnovationswereestab-lishedbythelastcommonancestorofthecnidariaandthebilater-ians.PetersonandButterfieldthinkthereisapalaeobiologicalsignatureofthatnewvulnerabilityofprotiststopredation.Beforethatperiod,acritarchs(single-celledorganismsofunknownaffilia-tion)hadhyperstableevolutionarydynamicsandlittleapparentdiversity:aftertheperiod,theyshowbothdiversityandrapidturn-over.PetersonandButterfieldthinkthemorphologicaldiversityoftheEdiacaranacritarchsisasignatureofbiologicalinteraction,inparticular,aresponsetopredationofprotistsbyearlyanimals.Insummary:ourbestguessofthepatternoftheMetazoanradiationgoessomethinglikethis.Metazoansprobablyexistedfor80100millionyearsorsobeforethebaseoftheCambrian.Buttheirecologicalandmorphologicaldiversitywaslow.OnlyafewoftheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation199livingMetazoanlineageshadseparatedbeforetheCambrian,andmostoftheseearlyanimalsweresimple,withoutmuchinthewayofcomplextissues,organs,ormusculature.Theywere,perhaps,mostlyverysmallaswell.However,thesizeandorganisationalcomplexityoftheLastCommonAncestorofthebilateriancladestheUrbilateriaremainscontroversial.AftertheCambrian,therewasagenuinelyrapidecological,phylogenetic,andmorphologicalradiation.Araftofmorphologicalcomplexeshadevolvedbythemid-Cambrian;complexesthathadexistedatbestinveryrudi-mentaryformattheEdiacaran-Cambrianboundary.TheCambrianExplosionisnoillusion.However,itwasprobablytheperiodinwhichmanyofthemodernphylaacquiredtheircrowngroupchar-acteristicsratherthantheperiodinwhichthosecladesfirstdivergedfromtheirsistertaxa.Itwas,inallprobability,anextraordinaryperiodfortheevolutionofnovelty.Puzzlesremain.Whydidanimalsremain,collectively,smallandunobtrusiveforthebestpartof100millionyearswhentherewasroomatthetop?Afterall,thepost-Cretaceousradiationofthemammalshastakenonlytwo-thirdsthattime.Ifunobtrusivemicrofaunalivedfor80millionyearsormorewithoutmuchexpansioninsizeorcomplexity,thissuggeststheexistenceofarightwallblockingtheevolutionoflargerandmorecomplexMetazoa.WeshouldlookforanexternalenvironmentalbarrierorconstraintsinternaltotheMetazoanclade.Thereremainphenomenatobeexplained.5.minimalmodelsofthecambrianradiationTheMetazoanradiationresultedintheevolutionofadazzlingarrayofmorphologicalnovelties:itwasaperiodinwhichthespaceofevolutionarypossibilityexpanded.InotedinSection1thatminimalistmodelsofnoveltyemphasisetheroleoftheexternalenvironmentinexplainingpossibility-expandinginnovations.Keyinnovationsareordinarychangesinextraordinarytimes.TheoriesoftheCambrianexplosionthatemphasisetheroleofexternalenvir-onmentaltriggersfitminimalism.Accordingtothislineofthought,armour(say)wasindeedakeyinnovation.Butinthegenerationofrelevantvariationanditsestablishmentinlocalpopulations,thespreadofbiomineralisedstructuresin(forexample)stemgroupCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n200kimsterelnybrachiopodslookedjustlikeanyothersmalladjustmenttolocalcircumstances.ThedifferencebetweentheCambrianandotherper-iodsofanimalevolutionlaynotintheprocessesthroughwhichvariationsaroseandspreadinlocalpopulations,butintheglobalextentandimportanceofdrivingenvironmentalchange.Therevo-lutionwasarevolutionincircumstances,arevolutioninwhatwasnecessary.Therearequiteplausibletheoriesofthiskind.OneideaisthattheMetazoanradiationwasinitiatedbyaninjectionofnewresourcesintotheecosystemsofthelateProterozoicworld.Onesuchresourceisoxygen.Atmosphericoxygenisabiologicalproduct:itisnotanancientfeatureoftheworld.Athresholdlevelofoxygenisneededtopowertheaerobicmetabolismsof(most)animals.Whiletinyani-malscanliveinlow-oxygenenvironments,large,active,orwell-armouredanimalscannot.Soonetraditionalhypothesiswasthattheradiationwastriggeredbyoxygensreachingthatthreshold(Nursall1959).Knolldiscussesthisideasympathetically,arguingthatthereisindeedgeochemicalevidenceforanincreaseinatmo-sphericoxygeninthelateProterozoic,aftertheyoungestofthethreesevereglobaliceagesthatprecededtheEdiacaranbiota.Theevo-lutionofthatbiotamightwell(heargues)havebeentriggeredbythatoxygenpulse(Knoll2003,21720).Thereareotherversionsofthisresource-drivenviewoftheradiation.Vermeijarguesthatmajorpulsesofevolutionaryinno-vationarecausedexternally,bysuddenpulsesintheavailabilityofresources(Vermeij1995).Inparticular,hethinksthatthetimeofMetazoanevolutionischaracterizedbytwogreatinnovationrevo-lutions:onefromtheCambriantothemid-Ordovician,theotherinthemid-Mesozoic.Underseavolcanismgeneratedalargepulseof11extraresources,andthosepulsesledtobiologicalrevolutions.HearguesthatanexternalmechanismmustplayacrucialroleintheMetazoanradiation,formanyinnovationsoccurredindependentlyinseverallineages,forexample,theevolutionofaskeleton,rapidandcontrolledlocomotion,thecapacitytoburrow.Theevolutionarydynamicsofonelineageareoftensensitivetochangeinothers.PerhapsthentheMetazoanradiationisaneffectofrunawaycoevolutionaryinteractions.Akeyinnovationinonelineagetriggerscoevolutionaryresponsesinothers.Thosere-sponsesthemselvesmayincludefurtherkeyinnovation,inducingCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation201feedback-drivendiversification.Thereisaraftofmutuallycom-patiblesuggestionsalongtheselines.Themostwidelydiscussedoneisbasedontheinventionofmacropredation,abreakthroughthattriggeredahostofdefensivecounteradaptationsandhencefurtheradaptationsforpredatorylifeways(Vermeij1987,McMenaminandMcMenamin1990).BuildingshellsandskeletonscertainlytakesoffintheCambrian:thaterasawtheevolutionofhardstructuresinbivalves,molluscs,brachiopods,arthropods.Therearetwomorerecentideas:Bilaterianswithatruebodycavityinventedtheca-pacitytotunnelthroughthesubstratebothforfoodandforrefuge,destroyingonecommunitytype(basedonsedimentsstabilizedbymicrobialmats)andestablishingothers.Therewasabiologicalrevolutioninthenatureoftheseafloor(Bottjeretal.2000).Mostrecently,AndrewParkerhasarguedthattheCambrianradiationisacentralnervoussystemrevolution.Theinventionoftruevisionistheinventionofaspecialkindofactiveagency.Throughvision,agentsgetfast,accurate,andpositionalinformationabouttheirlocalenvironment,andthatsetsupselectionforrapid,guidedresponse.Theresultistheevolutionofanewkindofagent,onewhosebehaviourinbothforaginganddefenceisguidedbyspecificandup-to-dateinformationaboutitslocalenvironment.Noothersensemodalityhasthiscombinationofrange,specificity,andspeedofinformationtransmission(Parker2003).SuchexplanationsfitwithminimalistmodelsoftheCambrianradiation.Theselectiveenvironmentchangedinanimportant,sustained,andglobalway.Thosechangesaffecteddifferentlineagesinsimilarways:perhapstheyeasedresourceconstrainsonallofthem;perhapsalltheMetazoanlineagesforthefirsttimewereatriskfrompredation.ThusbroadlysimilarresponsesevolvedinparallelindifferentMetazoanlineages.Astheystand,though,theseexternalistmodelsareincomplete:theygivenoaccountoftheori-ginsandestablishmentofthestrikingmorphologicalinnovationsthekeyinnovationscharacteristicoftheCambrianradiation.Minimalistmodelspresupposeratherthandeliveraminimalistaccountoftheoriginsofnovelties.Yetthemorphologicalinnova-tionsoftheCambrianaretrulyextraordinary.Moreover,oncewerecalltherearealternativestominimalismotherthanmodelsinvokingmacromutationsdeliveringawholefunctioningnewsysteminonego,thereisnoreasontosupposesomeminimalistCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n202kimsterelnyaccountofnoveltymustbetrue,evenifwedonotknowwhatitis.ThusweshallconsiderinthenextsectionanextendedargumentfortheclaimthatthebilateriannoveltiesoftheCambriandependedonfundamentalchangesinMetazoandevelopmentalsystems.WecannotexcludethepossibilitythatkeyCambrianinnovationsreallydidoriginateandestablishinjustthesamewaythat(say)variationsinthefacetsoftrilobitecompoundeyesevolvedandestablished.Butinthelightofthesedevelopmentalconsiderations,itislikelythattheMetazoanradiationarosethroughsomeformofcomplexfeedbackamong(a)exogenouslycausedenvironmentalchanges,(b)bioticallytriggeredenvironmentalchanges,(c)changesinevolv-ability.InthenextsectionIdiscusstheideathattheCambrianexplosiondependedonadevelopmentalrevolution,andthenreturntothekeyproblemforthischapter:whatdoesthisradiationshowabouttherelationshipbetweenmicroevolutionaryprocessandmacroevolutionarypattern?6.thedevelopmentalrevolutionhypothesisSeanCarrollhasarguedthatthediversificationpulseoftheCambrianrepresentsachangeindevelopmentalprogramratherthanachangeinselectiveregime.He,likeothers,emphasisestheimportanceofmodularitytoevolutionarypossibility.Innovationsthatincreasethespaceofevolutionarypossibilityaremadepossible(orperhaps,muchmoreprobable)bymodularconstruction.Ifthedevelopmentofonestructureislargelyindependentofthedevelopmentofothers,thosestructurescanvaryindependentlyofoneanother;thestruc-turescanbeindependentlymodified.Suchdevelopmentalcom-partmentalizationdecouplesphenotypictraitsfromoneanother,enablingalineagetoescapefromdevelopmentalconstraintsthatwouldotherwiselimittherangeofvariation.Moreovernewstruc-turescanbemadebyrepetitionfollowedbydifferentiation,aswitharthropodlimbs.Carrollarguesthatthereisevidenceforanincreaseincomplexityinarthropodlimbdesign(andhencearthropodeco-logicalversatility)sincetheCambrian,andheinterpretsthisasaninstanceofmodulardevelopmentsallowingrepetitionfollowedbydifferentiation(Carroll2001;seealsoLewontin1985,Kauffman1993,WagnerandAltenberg1996).OnCarrollsview,theseCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation203evolutionarychangescannotbeunderstoodasthereplacementofoneallelebyanotherinthecontextofinvariantsystemsofgeneregula-tionandexpression.TheCambrianradiation(and,moregenerally,theevolutionofnovelty)requiresevolutionarychangesinhowgenesareusedaswellaschangesinthegenesthemselves.Theyareessentiallyevolutionarychangesingeneregulation(Carroll2005).CarrollhasdefendedoneversionoftheideathattheCambrianradiationisaradiationindevelopment.ButthemostarticulateddevelopmentalistexplanationoftheCambrianradiationisduetoEricDavidson.TheMetazoanradiationisreallyaradiationofonedeepbranchoftheMetazoans.Thespongesandjellyfishdidnotexperienceanexplosiveincreaseindiversityanddisparity;thatbursttookplaceinthebilateralanimals(KnollandCarroll1999).Majormorphologicalinnovationsseparatethesedevelopmentallysimplespongesandjellyfish(thediploblasts)fromtheearliestbilaterallysymmetricalanimals(theUrbilaterians).Theseincludeathrough-gut,athirdgermlayer,acentralizednervechord,abodyvascularsystem,primitiveorgans.EricDavidsonandhiscolleagueshavesuggestedthatthesemorphologicalinnovationsandthediversifica-tionthatfollowedfromthemdependedonadevelopmentalrevolu-tion.ThatdevelopmentalrevolutionisthecrucialevolutionarychangethatmadetheCambrianExplosionpossible(Petersonetal.2000,PetersonandDavidson2000,ErwinandDavidson2002).Manyadultbilateriansdevelopfromlarvaethatliveasverysmall,butfree-livingmembersoftheplankton.Thislifehistoryisknownasindirectdevelopment,asjuvenilesaremorphologicallyandeco-logicallyverydifferentfromtheadultsintowhichtheydevelop.Suchlarvaeconsistofonlyafewthousandcells.Theyhaveonlytentotwelvecelltypes.Moreover,theyareorganizationallysimple.Theydonothavemultilayered,organisationallycomplexinternalstructures.Finally,theselarvaedevelopinadistinctiveway.Theirgenesareturnedonearlyindevelopment,andthuscelllineagesdifferentiateearly,withabouttenroundsofcelldivision.Cellsfindtheirfinalpositionandroleintheorganismunderlocalsignallingcontrol.Incontrasttothestandarddevelopmentalpatternofadultbilaterians,theselarvaedonotfirstdifferentiateintoembryonicregionspriortocelldifferentiation.TheHoxgenescontrolthissystemofdevelopmentalregionalisationonthefront-to-backaxisinadultbilaterians,andthosegenesarenotactiveintheselarvae.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n204kimsterelnyDavidsonthinksthatthedevelopmentalmodeexemplifiedbytheseplanktoniclarvaeisarelicoftheearliestbilaterians,andarguesthatdevelopmentwithearlydifferentiationandlocalcontrolsufficesonlyforthisgradeofmorphologicalcomplexity.Morecomplexmorphologiesrequiredadevelopmentalrevolutionthathadtwokeyingredients.Oneisregionalregulation,andhencedelayedcellulardifferentiation.Crucialgenesdeterminingcelltypearenotswitchedonuntilafterdevelopmentalregionsareestablished.Theotheristheevolutionofapopulationofset-asidecells:cellsthatretainalltheirpotentialforcelldivisionandthatarenotcommittedtospecificcellfates.Theseset-asidecellpopulationsexistinthelarvalformofindirectlydevelopingbilateria,andadultmorphologicalstructuresarerecruitedfromthoseset-asidecells.Contemporaryadultbilateriansshareadevelopmentalrecipethatincludesadifferentiatedaxisofsymmetryfromfronttobackandasystemofrecursiveregionalization.Thedevelopingembryoisdividedintoasetofregions,eachofwhichisunderfairlyinde-pendentdevelopmentalregulation.Oftentheseregionsarefurthersubdivideduntilthespecificdetailsofadultmorphologyareconstructed.Sohowdidthedevelopmentalrevolutiontakeplace?Davidsonsevolutionarynarrativegoessomethinglikethis:Early-differentia-tionembryogenesisevolvedaftertheCnidariasplitofffromthestemMetazoa.Thethirdtissuelayeroftheembryo,theendomesoderm,thenevolved,andthiswascrucialtothelaterevolutionofstruc-turallycomplexorgans,forthesehavetheirdevelopmentalorigininthisthirdlayer.AftertheCtenophorasplitofffromthestem,bila-teriansymmetryandthePax-6genes(involvedinvision)evolved.Thefinalinnovations,justpriortothecrowngroupradiationofthebilaterianphyla,wastheevolutionofafullHoxclusterwithfront-to-backdifferentiationandset-asidecells.Hoxgenesevolvedearlier,whenthecnidarianssplitfromthebasalMetazoans,buttheydidnotoriginallyfunctiontocontrolfront-to-backdifferentiation.ThefullHoxcomplementwiththecontemporaryHoxfunctionsevolvedrelativelylate.Thisnarrativeleadstotheirportraitofthecommonancestoroftheextantbilaterianclades:thecommonancestor,thatis,tothevastmajorityofmulticelledanimaldesigns.Morphologically,thecreaturewassmall;perhapsverysmall,andpossiblypelagic.ButitCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation205wasmorestructurallycomplexthanitsCnidariansistergroup.Thecommonancestorhadamesodermallayer,acentralnervoussystem,andtwo-endedgutwithamouthandanus.Itthushadafront-to-backaxis.Thesearegenuinebilaterianhomologiesnotsharedbyjellyfish.Moreover,somecommondevelopmentalmechanismswereavailabletothecommonancestor,includingtheHoxsystem,notjustforbasicbody,front-back,andup-downdifferentiation,butforstructuresattachedtothebody.Thusthemorphologicalsimi-laritiesbetweenthebilaterianphylainpartdependoncrucialandverydeepmorphologicalhomologies.Buttheyarealsotheresultofparallelevolutionworkingwithahomologousdevelopmentaltoolkit.Manyorgansystemsacrossthebilateriansasawholeheads,hearts,sensorysystemsareanalogsnothomologs.Butthecelltypesofwhichtheseorgansarecomposedarehomologsandexplainsomeoftheirfunctionalsimilarities,asdothesehomol-ogousdevelopmentalmechanisms.Thecommonancestorhadthedevelopmentaltoolkitneededforacomplexmorphology.Butitwasnotitselfcomplex.Whilethecommonancestorwasitselflikelytobearelativelysmallandsimpleorganism,itwaspreadaptedtomorphologicaldifferentiationinresponsetotherightbioticandabiotictriggers.Asitwasinitiallyformulated,thedevelopmentalrevolutionhypothesisentailsapuzzleofitsown.Whatselectedforthesedevelopmentalchanges?AsGrahamBuddandSorenJensen(2000)argue,ifdevelopmentalinnovationprecededmorphologicalinno-vation,theselectiveadvantagesofthedevelopmentalinnovationsofset-asidecells,modulardevelopmentalregulation,andtheHoxsystemareobscure.Whatwasthefunctionoftheseinnovationsiftheyprecededratherthanpostdatedoraccompaniedgrowthinsizeandcomplexityofbodilyorganization?Davidsonandhiscolleaguesplacethesedevelopmentalchangesverydeepinthebilaterianstemlineage,longbeforeanytracefossilevidenceofbilaterianmorpho-logicalinnovations.Theyseemtocommitthemselvestotheideathattheseanimalswereminuteandnondescriptevenatthecom-pletionofthedevelopmentalrevolution(Petersonetal.2000,12).Thereis,however,anaturalmodificationoftheideaofthedevelopmentalrevolutionthatleadstoamoreintegratedconceptionoftheradiation.Davidsonthinksthatearly-differentiationem-bryogenesisisnotsufficientlypowerfultobuildadult,crowngroupCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n206kimsterelnybilaterianbodyplans.Thatcouldmeanthatitissimplyimpossibletobuildanadultcrowngroupbilaterianbythismodeofdevelop-ment.Butperhapstheconstraintislessabsolutethanthat.Perhapsearly-differentiationdevelopmentcanbuildonlyasomewhatsimpleversionofanadultbilaterian;forexample,onewithoutacomplexsensorysystemorcomplexlocomotor-manipulationsystem.Alter-natively,perhapstheprimitivemodeofbilateriandevelopmentcanbuildcomplexadultsonlyatthecostofahighrateofdevelopmentalerror.Thismoremodestviewofdevelopmentalconstraintsoncomplexityleadstoanaturalratchethypothesis:apositivefeedbackbetweendevelopmentalandmorphologicalchange.Earlybilaterianmorphologicalinnovations(forexample,thetwo-endedgut)wouldselectforimproveddevelopmentalcontrol,toreducetherateofdisastrousdevelopmentalerrors.Oncetheseevolve,theypermittheevolutionoffurtherearlybilateriannovelties,whichinturnselectforfurtherimprovementsindevelopmentalcontrol,andsoon.Earlyelementsinthebilaterianbodyplandidoccurfirst.Butthelimitsonearly-differentiatingdevelopmentinconstructingsuchbodiesselectedforthekeyelementsofthedevelopmentalrevolution,bothtomakedevelopmentmorereliableandtosupportspecificadaptivecomplexesgraftedontothebasicplan.7.finalthoughtsThekeypointthetakehomemessageisthatminimalmodelsareindeedminimal,andtheycanbeenrichedinavarietyofways.Oneisbyextendingtemporalscale:onmicroevolutionarytimescales,wecanoftentreatfeaturesofboththeenvironmentanddevelop-mentasfixed.Butthesearenotfixedonmacroevolutionarytimescales.Inparticular,itisunlikelythatwecaningeneralmodeltheevolutionofnoveltyinaclassicallyDawkinsianway,bythinkingofalternativeallelesasdifferencemakers:onereplacesanotheraseachmakesaconsistent,selectablephenotypicdifference,butonlyrelativetoafixeddevelopmentalandgeneticenvironment.Intheevolutionofpossibility-expandinginnovations,theseenvironmentsarenotfixed.Theindividualistperspectiveonselectionissome-timestoolimiting.Itisclearthattransitionsinindividualityrequiregroupselectionofsomekind.ButitmaywellbethecasethatthestructureandorganizationofspeciesthemselvesarebothimportantCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation207andnosimplereflectionofwithin-populationchangeinthecon-stituentdemesfromwhichspeciesareformed.Itisarguablethatspeciesstructureplaysaroleinexplainingfine-grainedmacro-evolutionarypatterns,forexample,thepunctuated-equilibriumpatternoftypicalspecieslifehistories(Eldredge1995,Sterelny1999b).LetmefinishbyconnectingthesegeneralmoralstotheMetazoanradiation.First:theoriginationoftheMetazoansinvolvedatransi-tioninindividuality,andhencerequiresamultilevelperspectiveonevolution.TherewillhavebeenaperiodintheearlyevolutionoftheMetazoanswhenthefitnessofindividualcellswithinaprotoanimalandthefitnessofthatprotoindividualinapopulationwerebothimportant.Second:theradiationoftheMetazoanswasnotanevo-lutionaryradiationtakingplacewithinafairlyconstantenviron-ment.Theradiation,whateveritscauses,profoundlychangedboththeselectiveandthephysicalenvironment.Likewise,theCambrianradiationwasevolutioninachangingdevelopmentalenvironment.Davidsonmaynothavecorrectlyidentifiedtheprimitivemodeofbilateriandevelopmentorthesequencethroughwhichthecon-temporarybilateriandevelopmentaltoolkitwasbuilt.Butitisclearthattheradiationinvolvedprofounddevelopmentalchange,forwhileprotistpreadaptationsforcomplexdevelopmentwereimpor-tant,theradiationrequiredtheevolutionofbothcellulardiffer-entiationandverticalcomplexity,ascellsareorganisedintotissues,organs,andorgansystems.Veryearly,simpleMetazoanshavefewcelltypes:Placazoahavefour;spongeshavefivecelltypes;cnidar-ianshaveten.Valentineestimatesthatstemgroupbilaterianshadbetweentwelveandfortycelltypes,dependingonthephylum.Crowngroupbilaterianshavemanymore(Valentine2004,7475).Thereisnothinglikethisintheprotistworld.HencetheevolutionofMetazoans,especiallythebilaterians,requiredamajorrevolutionindevelopmentalcontrol.Onecannotmodelthisevolutionarytransitionastheresultofthesubstitutionofvariantallelesfortheirpredecessorsinarelativelyfixeddevelopmentalenvironment.Therewerecrucialchangesnotjustinorganismsgeneticcomplementbutinthewaysgenesareused.Finally,thereistheopenissueofnovelty.WerethecrucialnoveltiesoftheCambrianradiationbuiltunob-trusively,withtheirsignificanceonlybecomingapparentlater?TheextenttowhichwecanfittheevolutionofnoveltyintotheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n208kimsterelnygradualistframeworkofminimalmodelsremainsopen.ButifCarrollorDavidsonisright,somenoveltiesphenotypicicebergsreallyaredifferent,fortheirevolutionisaccompaniedbychangesinthedevelopmentalarchitecturethatmakefurtherchangesmuchmorelikely.ACKNOWLEDGMENTSThankstoBrettCalcott,PatrickForber,PeterGodfrey-Smith,DavidHull,AndyKnoll,TimLewens,DanMcShea,SamirOkasha,andMichaelRusefortheirhelpfulcommentsonanearlierdraftofthisessay.notes1.Gouldhaswrittenhereofextrapolationism(Gould2002).IhaveavoidedthistermbecauseGouldsworksuggeststhatthereisasinglecontrastbetweenminimalistandextendedmodelsofthemicro/macrorelationship,onethatturnsontheacceptanceorrejectionofhigh-levelselection.Incontrast,Ithinkthereareanumberofwaysofgoingbeyondminimalism,andsomeofthesehavenothingtodowithlevelsofselection.2.Suchpropertiescanbeimportantwithoutspeciesselectionsbeingimportant,inpartbecausespecies-levelpropertiesmaynotbeheritable.3.Thereisconsiderabledebateaboutthehistoryoflife,evenatacoarsegrainofanalysis.Itisnotsurprisingthattheissueofprogresshasalwaysbeencontentious(seeRuse1996).Butevenmoretechnicalclaimsaboutcomplexityanddiversityhavegeneratedrichdebate:see,forexample,McShea1998a,KnollandBambach2000,Benton2006.4.InfittingtheSpreadofVariationviewwithinaminimalistframe-work,IpartcompanywithDanMcSheasimportantandinfluentialworkontheseissues.Weboththinkthecrucialelementofminimalismislocaldeterminism:forminimalmodelstobeadequate,thefateofalineagedependsonthefateofitsconstituenttaxa,andtheirfateinturndependsonlocalcircumstances.Butwehavedifferentviewsonhowtoidentifylocaldeterminism.Forexample,unlikeme,McSheacountssensitivitytoaleftwallofminimalcomplexityasaviolationoflocaldeterminism:hethinksofitasafeatureoftheglobalenvironmentaffectingalltaxa.SeeMcShea1996,Alroy2000,McShea2000.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMacroevolution,Minimalism,Radiation2095.Thismayexaggeratetheuniquenessoftheprotistthreattobacteria.Thereispredationinthebacterialworld,andhenceprobablyinancientbacterialenvironments.Bacteriapredatebylysingandenvagination.Inthefirst,amobofbacteriareleaseanenzymethatcausesthemembraneofthepreyspeciestobreach,spillingthecontentsofthecellandtherebymakingtheaminoacids,nucleicacids,andotherbuildingblocksofmetabolismavailableforacquisition.Inthesecondmethod,abacteriumbumpsinto,surrounds,takesin,andlysesasmallermicrobe(Lyons,personalcommunication).6.Insayingthis,itisimportanttodistinguishbetweenlarge-effectmutationsandmacromutations.Large-effectmutationsinvolvesignificantquantitativechangesinexistingtraits;macromutationsinvolvethesingle-stepcreationofnewstructures.Recentpopulationgeneticshasembracedtheideathatlarge-effectmutationsareimportantcausesofordinaryevolutionarychange.Minimalistmodelsarecertainlynotcommittedtotheviewthatallstructuresarebuiltbytinyincrementsovercountlessgenerations.Foragooddiscussionoftheresourcesavailabletominimalism,seeLeroi2000.7.For,first,stem-grouporganismswillnotfittaxonomicstereotypesderivedfromcrowngrouporganisms.Second,theywilllookstrangebecauseoursenseofanormal-lookingorganismforinstance,anormalcrustaceanisderivedfromourexposuretoahostofcrowngroupcrustaceans.Ourpatternrecognitionheuristicsaretrainedoncrowngrouptaxa.Thereremains,asDanMcSheapointsouttome,thepossibilitythattheseearlyanimalsareweird,disparity-expandingorganismsnotjustbecausetheyfailtofittaxonomicstereotypesbutbecausetheyhaveanextraordinaryloadofuniquetraits.Hemayberight;unfortunately,welackwaysofmakingsuchintuitionsofweirdnessrigorous.8.Simpleversionsofthisapproachobviouslymakeriskysimplifyingassumptionsabouttheconstancyofratesofchangeacrosstimesandbetweenlineages.Butthoseconcernscanbeaddressedbyusingavarietyofdifferentgenesanddifferentcalibrationpoints.ForasystematicdiscussionofthereliabilityofvariousclocksandtheirapplicationtotheradiationoftheMetazoans,seeBromham2003.9.Thedefendersofthedeepdatesdisputethis,arguingthatthe(relatively)recentdivergencetimesestimatedbyPetersonandcompanydependnotonthechoiceofcalibrationtaxa,butonthetreatmentofthefossildatesasthemaximumageofdivergence,asifthefirstappearanceofafossilintherecordwasthefirstappearanceofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n210kimsterelnythatorganismonEarth(BlairandHedges2005).PetersonrespondsinPetersonandButterfield2005.10.Nothingextraordinary,solongasthepreservationpotentialofsoft-bodiedorganismshasnotchangedradicallyovertheEdiacaran-Cambrianboundary.Thisassumptionmaynotbesafe:thereareEdiacaranfossilsofearly-stageembryos,andtheseareoftinyorganisms.Moreover,ithasbeenarguedthattheEdiacaranfossilswereformedonlybecauseEdiacaranpreservationconditionswereverydifferentfromthoseoftheCambrianandsubsequenteras(Narbonne2005).11.Thereseemstobeaseriousproblemwiththisidea,fortheinnovationmechanismrestsontheideathatuntiltheresourcebudgetincreases,innovationsaretooexpensive.Apulseofresourcesintotheenvironmenteasesresource-basedconstraintsonpotentialinnovations.Butthisassumesthatanincreaseinoverallproductivityleadstoanincreaseinpercapitaaccesstoresources.Butifpopulationgrowthkeepspacewiththegrowingresourceenvelope,thenthepercapitaavailabilityofresourcesmaynotchange.Vermeijnotesthisproblem(1995,134),butthenendsuprespondingtoadifferentproblem,theideathatasuddenresourcespurtmaybedestabilizing,apossibilityhearguesisconfinedtorelativelyundiverseecosystems.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nmaureenkearney11PhilosophyandPhylogeneticsHistoricalandCurrentConnectionsPhilosophicalargumentshaveplayedaninfluentialroleinthedevelopmentofphylogeneticsystematics–thefieldofbiologythatseekstoreconstructthegenealogicalrelationshipsamongspecies,discoverthepatternofeventsthathasledtothedistributionanddiversityoflife,andusethisknowledgetoconstructnaturalclassi-ficationsofspecies.Threesetsofdiscussionsclearlydemonstratethisconnectionbetweenphilosophyandphylogenetics:inferencemodesandtheirrelevancetocompetingphylogeneticmethods,thenatureandtreatmentofspeciesandhighertaxa,andthenatureandtreatmentofphylogeneticevidence(characterdata).Withineachoftheseareas,systematistshaveusedphilosophicalargumentstodefendparticularconceptsandmethodologicalapproaches,ortoproposenewones.And,withineachoftheseareas,philosophershavescrutinizedtheargumentsofsystematistsandcontributedtheirown.Vigorousdebateamongstsystematistsregardingthesetopicsispervasive.Acommonunderlyingtensionthathelpsdrivesuchdebatesrevolvesaroundtheproperrolesofprocesstheories,assumptions,andtrainedjudgmentinphylogeneticsresearch.Forexample,concernsaboutobjectivityandtestabilityhavesometimesledsystematiststorejectmethodsthatdependonevolutionaryprocesstheories,butsuchrejectionstypicallydonot‘stick’forverylong.Thus,acyclicalpatternisevident–attemptstoinfusetheo-reticaldependenceintophylogeneticsresearchhaverepeatedlybeencounteredbychargesofnon-objectivityanddecreasedtestability,yetattemptstoavoidthemhaverepeatedlybeencounteredbychargesofoperationalism.Twomainquestionsemergefromthis:Whatmustbeknownaboutevolutioninordertoanalyzephylogeny?Whatdoesitmeantobeobjectiveasaphylogeneticist?211CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n212maureenkearney1.phylogeneticsystematicsSystematicsmaybetheoldestbranchofbiology,oftentracedbacktoAristotleandtheancientGreeks.Aristotle(384–322BC)heldanessentialisticviewofspeciesaseternalandimmutable,andchar-acterizedfeaturesoforganismssimilarly.Thistypologicalviewofnaturepersistedforcenturies,andbiologicalclassificationvialogicaldivision(i.e.,legs/nolegs,blood/noblood)wasthedominantapproach.Linnaeus’(1707–78)systemofclassificationwasfunda-mentallybasedontheAristoteliantraditionoflogicaldichot-omizationandbecameformalizedunderthebinomialsystemoftaxonomicnomenclaturethatissofamiliartoallbiologists.Addi-tionally,upuntiltheearlynineteenthcentury,apervasiveideaofthenaturalorderoftheworldwastheGreatChainofBeing,orScalaNatura(LadderofNature),anunbrokensequencefromthemostprimitiveorganismstothemostadvanced(humankind)(Lovejoy1936).Thislinearsequenceoflifewasrootedinearlyideasabouttheprogressivestructureoftheworld,evermovingtowardsper-fection.However,theobservedstructureofvariationinthebio-logicalworldeventuallyrejectedhypothesesofprogressiveordering.Darwin’s(1859)evolutionarytheorylaidthegroundworkforrejectinganessentialisticnotionofspecies,emphasizingthevari-abilitythatmustexistinorderfornaturalselectionandtransfor-mationtooccur.Theemphasisonvariabilitystandsinobviousoppositiontonotionsoftypes.Likewise,iteventuallybroughtanendtotheScalaNaturaandrelatedideasabout‘naturalprogression’.Ultimately,thesewerereplacedby‘tree-thinking’,withentitiesrelatedthroughhierarchiesofcommonancestry.Darwinalsorevolutionizedthedisciplineofsystematicswiththenotionthatclassificationshouldbebasedongenealogicalrelationships(Darwin1859),althoughnotallofhiscontemporariesagreedwiththisidea.Duringthelatenineteenthandearlytwentiethcenturies,biologywasstronglyinfluencedbyextensivestudiesofpopulationsandtheirvariability,leadingtotheModernSynthesis–aunificationofvariousfieldsofbiologysuchaspalaeontology,systematics,andgenetics(e.g.,Dobzhansky1937,Fisher1930,Huxley1942,Mayr1942,Simpson1953).BuildingonDarwinianprinciplesandnewCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics213evolutionarystudies,the‘populationthinking’oftheModernSynthesisbiologistsfurtherstressedtheuniquenessandvariabilityoforganismsandpopulations.Withvariationseenasfundamentaltobiologyandthenotionoftypesrejected,averydifferentworldviewemerged–onethatisarguedtohaveprofoundlyaffectedthedisciplineofsystematics(Mayr1959).Thelasthalfofthetwentiethcenturywitnessedseveralmethod-ologicalrevolutionsinsystematics,whicharedescribedbelow.Theprominenceandreputationofsystematicswithinthebroaderfieldofevolutionarybiologygrewsteadilythroughoutthoseyears.Today,systematicshasanintimateconnectiontomanyotherareasofbiologybecausetheresultsofphylogeneticanalysis(phylogenetictrees)allowbiologiststotestprecisehypothesesaboutevolutionarypatternsandprocesses.Aresomegroupsmorediversethanothersand,ifso,why?Dofeaturesoforganismsco-evolve?Howmanytimesdidanecologicalassociationorastructureevolve?Istheevolutionofabehaviorcorrelatedwiththeevolutionofamorpho-logicalfeature?Howdogeneticanddevelopmentalregulationvaryacrossgroups?Dogeneticchangesoccurmorerapidlyinsomegroupsthaninothers?Today,werecognizethatanswerstoallofthesequestionsdependatleastpartiallyuponphylogenetictrees.Modernbiologytellsusthatthereisasingleevolutionarytreeoflifeforallspecies–atleast1.7millionspecies,astaggeringnumberthatstilldoesnotreflecttotalhistoricaldiversitybecauseoffossilandextantspeciesnotyetdiscoveredordescribed.Initssimplestconception,phylogeneticsystematicsistheorganizationofthistreeoflife,ortheorderingofbiodiversity.Theorderingsystemisaphylogenetictree,ahierarchicalsystemthatgroupstaxaaccordingtorelativerecencyofcommonancestry,basedonhomologousfea-turesderivedfromcomparativestudiesofphenotypicandgeneticdata.Thus,thetaskofthesystematistcanbeseenastheknittingtogetherofspeciesviaevidenceofcommonancestryintoaphylo-genetictree.Virtuallyallcontemporarybiologistsagreethatevolu-tionoccurs,thattheresultofitisthevastbiodiversitywitnessedaroundus,andthatknowledgeofhistoricalphylogeneticrelation-shipsisnecessaryfortestingevolutionaryandecologicalhypoth-eses.However,theystillargueaboutwhatthatmeansforthepracticeandmethodsofsystematics.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n214maureenkearney2.methodsformakinginferencesaboutphylogenyEvolutionarytaxonomy(e.g.,Mayr1969,Simpson1961)grewoutoftheModernSynthesis,andwasheavilyrootedinDarwinianevolu-tionarytheory.Themethodsofevolutionarytaxonomybeginwithevolutionaryfirstprinciplessuchasnaturalselection,adaptation,andhomology.Theseprinciples,inconjunctionwithextensivecomparativestudiesoforganisms,areusedtoassesstherelativeimportanceand/orreliabilityoforganismalfeatures(characters)forinferringgenealogicalrelationshipsand,ultimately,toreconstructevolutionaryrelationshipsamongspeciesbasedonthosecharacters.Anemphasisonheterogeneousratesofevolutionacrossgroupsandoncausallyimportantevolutionaryinnovationsleadstothecon-structionoftaxonomicgroupsbasedonacombinationofrecencyofcommonancestryandpurportedadaptivelyimportantsimilarities.Thus,anevolutionarytaxonomicclassificationmayreflectbothevolutionarybranchingpatternsandevolutionarydisparitybetweengroups.Asanexample,thereiscurrentlyconsiderablesupportfortheideathatbirdsandcrocodiliansshareamorerecentcommonancestrythaneitherdoeswithotherextantgroups(suchasturtles,snakes,or‘lizards’).However,evolutionarytaxonomistsprefertogroupcrocodilianswithturtles,snakes,and‘lizards’inthegroup‘Reptilia’(totheexclusionofbirds).Becausebirdshavemanyuniquecharactersandareconsideredtohavedivergedsignificantlycom-paredtorelatedgroups,theyarerecognizedasaseparatetaxondespiteevidenceofasharedevolutionaryhistorywithcrocodilians.Thesamekindofargumenthasbeenappliedtohumansinrelationtotheirclosestrelatives.Evolutionarytaxonomywascriticizedforalackofexplicitmethod-ology,subjectivejudgmentsaboutthephylogeneticutilityofdata,andaneclecticapproachthatoftenproducedcompetingclassificationsforthesamegroup.Evolutionarytaxonomistswereportrayedastoospeculativeandintuitive,transcendingempiricaldatatoproduceauthoritariananduntestableviewsofphylogeny.Mostimportantly,criticsnotedthepotentialforcreatingartificial(non-monophyletic)groupswiththesemethodssincefactorsotherthancommonancestryweresometimesusedtogrouptaxa.Twoverydifferentschoolsdevel-opedinoppositiontoevolutionarytaxonomy–numericaltaxonomyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics215andcladistics.However,somefundamentaltenetsofevolutionarytaxonomyremaininsystematicstoday.Thearchitectsofevolu-tionarytaxonomypublishedthefirsttextbooksdedicatedtosystematicmethods,whicharewidelycitedtodayaslandmarksthatofferedalexiconandmorepreciselyhonedconceptsforsystematics.Neartheendofthe1950s,somescientistsbeganadvocatinganapproachtosystematicsthatusedcomputer-assisted,quantitativemethods.Thesescientistsproposedanexplicitandmore‘objective’methodologyforsystematics,leadingtotheriseofnumericaltax-onomyor‘phenetics’(SneathandSokal1973,SokalandSneath1963).Toalargeextent,pheneticsmaybeviewedasabacklashagainstwhatwereperceivedasthesubjectiveandunrepeatablemethodsofevolutionarytaxonomy,combinedwiththeburgeoningapplicationofcomputersciencetovariousbiologicaldisciplines.Pheneticistsarguedthatevolutionarytheoryshouldnotenterintoclassificationstudies;objectivityinsystematicswastobefoundinpurportedly‘theory-free’,quantitativemethods.Indeed,thetwoprincipalaimsofnumericaltaxonomywere‘repeatability’and‘objectivity’(SneathandSokal1973,11).Inordertoaccomplishthesegoals,pheneticistsadvocated1)theuseofaveraged‘overallsimilarity’measuresforgroupingorganisms,2)equalweightingofallcharacters,3)theuseoflargenumbersofcharacters,4)quanti-tativecharactercoding,and5)a‘theory-free’approachtocharacteridentificationusing‘rawsimilarity’asaguide.Apheneticclassifi-cationtypicallydepictsgroupsthatareclusteredquantitativelyonthebasisofaveragedsimilarity(ordistance)values.Distinctionsarenotmadebetweenhomologousversusnon-homologoussimilarity,norbetweenprimitiveversusderivedsimilarity.Pheneticswasintendedprimarilyforclassification,notgenealogy(whichwasconsideredunknowable).Theapproachwasmeanttoproducethemostefficient‘informationstorageandretrievalsys-tem’,oranall-purposeclassificationoforganisms.Itwascriticizedformanyreasons,includingthefactthat‘overallsimilarity’isnotabiologicallymeaningfulbasisforsystematics(e.g.,Farris1979,1983,Mayr1965).Further,itsnaivete´vis-a`-vis‘theory-free’characteridentificationwasdescribedasthe‘‘look,see,code,cluster’’approach(Hull1994).Despitetheidealisticnotionof‘overallsimilarity’,numericaltaxonomyalsoleftimportantlegaciestosystematics–theCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n216maureenkearneynumericalcodingofcharactersandtheuseofcomputeralgorithmstoanalyzedataprovedtobelastingchangesinsystematicmeth-odology.Somewouldalsoarguethattheantitheorystanceofpheneticspersistsinvariousformsinthefieldtoday.AsdidDarwinandothers,WilliHennig(1950)arguedthattax-onomyshouldreflectphylogeny,thatgenealogicalrelationshipsamongspeciesshouldbebasedon‘specialsimilarity’orsharedderivedcharacters,andthattheserelationshipsshouldbearrangedinahierarchicalmannertoreflectthetheoryofdescentwithmod-ification.Hennig’sphylogeneticsystematicsemphasized:1)theuseofonlyshared,derivedcharacters(synapomorphies)asevidenceforidentifyingnatural(monophyletic)groups;2)comprehensivestudiesofhomologydeterminationbasedoncharacteranalysis;and3)anexplicitlygenealogicalinterpretationofrelationshipsamongspe-cies.Incontrasttoevolutionarytaxonomy,phylogeneticsystem-aticsacceptsonlymonophyletictaxonomicgroups–forexample,thosegroupscomposedofthemostrecentcommonancestoroftheincludedspeciesandallofitsdescendants.Incontrasttophenetics,phylogeneticsystematicsisrootedinthetheoreticalprincipleofdescentwithmodification,incorporatesbiologicalevaluationofcharacters,andusesdiscretesynapomorphiesratherthanoverallsimilarityvaluestodiagnosegroups.Theresultisacladogramdepicting‘sister-group’relationships,orrelativerecencyofcommonancestryamonggroups.Theimportantdistinctionsbetweenmonophyleticgroups,para-phyleticgroups,andpolyphyleticgroupsisoneofHennig’smostimportantlegacies.Amonophyleticgroupisdiagnosedbysynapo-morphyandcomprisesacommonancestorandallofitsdescen-dants;aparaphyleticgroupisdiagnosedbysymplesiomorphyandcomprisesacommonancestorandsome,butnotall,ofitsdescen-dants;apolyphyleticgroupexcludesthemostrecentcommonancestorofitsmembersbecauseitsdiagnosticcharacteraroseseparatelyintwoormorephylogeneticallydisparatelineages.Onlymonophyleticgroupscanbeconsidered‘natural’or‘real’entitiesaccordingtoHennigbecauseonlyinthosegroupsisgenealogicalhistorycaptured.InHennig’ssystem,theimportantdistinctionbetweenhomologousandnon-homologousderivedsimilaritymustalsobeanalyzed.Twoormoretaxamayshareaderivedsimilarity(synapomorphy)foreitheroftworeasons:eitheritwasacquiredCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics217throughdescentfromacommonancestor(homology),oritwasacquiredconvergently(homoplasy).Thedistinctionisrevealedthroughphylogeneticanalysis–theanalysisofobservedfeaturesoforganismsrelativetoahierarchy.ThelegacyofHennig’sworkinsystematicsisprofound.Indeed,shortlyafterthetranslationofHennig’sbookintoEnglish(Hennig1966),systematicsunderwentanotherrevolutionwiththedevel-opmentofcladistics(e.g.,EldredgeandCracraft1980,KlugeandFarris1969,NelsonandPlatnick1981).ExpandingonHennig’sviews,cladistsarguedagainstbothevolutionarytaxonomyandphenetics.Theyadvocatedthatphylogeneticsoughttobeanempiricalandtestablescience(incontrasttotheintuitiveand/orauthoritarianapproachofevolutionarytaxonomy)andthatsharedderivedfeaturesprovidetheonlybasisfortaxonomy(incontrasttotheuseof‘rawsimilarity’inphenetics).Fromthebeginning,cladistshavealsobeencloselyassociatedwiththeideathatthephiloso-phicalprincipleofparsimonyshouldbeanintegralpartofphylo-geneticmethods–inpractice,thisprincipleisusedtominimizeadhochypothesesofhomoplasyinphylogeneticanalysis(e.g.,Farris1983).Theuseofparsimonyisusuallyjustifiedwithanappealtoexplanatorypower–mostparsimoniousphylogenetichypothesesaresaidtoexplainasmuchoftheavailableevidenceaspossibleashomology,therebyavoidingadhochypothesesofhomoplasy(Farris1983).The‘cladisticrevolution’intaxonomyisconsideredahighlysignificantparadigmchangeinthefield(Hull1988),initiatedbyHennig’sstrongfocusongenealogicalrelationshipsbetweenspecies,andrevolutionaryinthesenseofreplacingintensionalwithexten-sionalthinkinginsystematics(Dupuis1984).Thismaybeso,butitisalsothecasethatsincethebeginningofcladistics,therehasexistedatensionbetweenthosewhoemphasizegenealogicalrela-tionshipsandmoreorlessembraceevolutionarytheoryandthosewhoemphasizeclassificationandresisttheincorporationofevolu-tionarytheoryintosystematics.Thelattergroup–the‘patterncladists’–arguedthatcladisticsitselfisnotaboutevolution,butonlyaboutthepatternofrelativerelationshipsamongsttaxaasindicatedbycharacterdistributions(NelsonandPlatnick1981,Patterson1982,Platnick1979).Somesystematistscontinuetoarguethatcladisticsisanevolutionary-theory-freeclassificationmethod.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n218maureenkearneyPatterncladistics,or‘transformedcladistics’,grewoutofskepticismregardingtheabilityofsystematiststoreconstructphylogeny,aswellasconcernaboutmethodologicalcircularity–inotherwords,ifsystematistswishtousephylogenetictreestotesthypothesesaboutevolution,thentheyshouldnotuseevolutionarytheorytoconstructtrees.Thedistinctionbetweenobservedpatternandexplanatoryprocesstheoryisparamountinthesediscussions:theexplanandum(inthiscase,thehierarchyofgroupswithingroups)andtheexpla-nans(inthiscase,phylogeny)shouldnotbeconflated(Brady1985).Thepurportedindependenceofobservationandinterpretationandtheappealtoobservationaslogicallypriortophylogenyseemtobeargumentswithrootsinempiricismandidealisticmorphology.Inanycase,accordingtopatterncladists,classificatorycladograms–withtaxaorganizedinsetswithinsetsbasedontheparsimoniousdistributionofcharacterdata–areallthatcladisticscanclaimtoachieve.Theuseofparsimonymethodsinthiscontextissometimesjustifiedbasedonhighinformationcontentfoundinparsimoniousclassifications.Itis,however,difficulttoarguefortheprimacyof‘classification’over‘phylogenyreconstruction’whenoneexaminestheutilizationofcladogramsbybiologists.Thecontemporaryliteratureindicatesthatsystematistsarenotinterestedininformationstorageandretrievalsystems,Venndiagrams,orefficientsummariesofchar-acterdistributions.Instead,mostsystematiststodayseemtobeconcernedwithphylogenyreconstruction(i.e.,inferringhistoricalpatternsofcommonancestry),andwiththeuseofphylogenetictreestotestbroaderhypothesesinevolutionarybiology–oratleastthisishowphylogenetictreesaretreatedonceproduced,regardlessofwhatisclaimedbytheirauthorsabouttheirinitialontologicalstatus.Argumentsaboutinferencemodeshavealsoplayedanimportantroleinthehistoryofmethodologicaldebatesinsystematics.Farris(1983)proposedahypothetico-deductiveapproachtophylogenetics,alsosuggestingthatweshouldchoosethosephylogenetichypoth-eseswiththehighestexplanatorypower.Thesehypothesesaresaidtobethemostparsimoniousones,whicharethosethatrequirethefewesthypothesesofhomoplasy(convergenceorparallelism).TherootsofthisideacanbefoundinHennig’sprinciplethat‘‘thepresenceofapomorphouscharactersindifferentspeciesisalwaysCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics219reasonforsuspectingkinship...,andthattheiroriginbycon-vergenceshouldnotbeassumedapriori’’(Hennig1966,121).Thisstatementisinterpretedbymostcladiststomeanthathomologyshouldbepresumedintheabsenceofevidencetothecontrary,or,inotherwords,thathomoplasyshouldbeminimizedinphylogeneticanalysis.EarlycladistsalsoinvokedthefalsificationistphilosophyofKarlPopper(1959,1962)asameanstoincreasethetestabilityofphylo-genetichypotheses,andtosupporttheclaimthattheleastfalsified(mostcorroborated)phylogenetichypothesiscorrespondstothemostparsimoniouscladogram.Later,cladisticswastiedtoaPopperianphilosophyofscienceviathe‘testofcongruence’–thematchingversusnon-matchingofcharacterstatements,whichplaytheroleofpotentialfalsifiersinthissystem(e.g.,Kluge1997).Accordingtothis,themaximallycongruentsetofcharactersgivesthemostparsimonioustree,whichisthehypothesisthatisleastfalsified(andmostcorroborated)bythedata.Somesystematistsandphilosophersdisagreedwiththeideathatcladisticscanbeconstruedasafalsificationistendeavor.Manyviewedparsimonymethodsaseitherinductiveinference(relyingonthemaximalcongruenceofcharacterstatementstoobtainthebest-supportedtree)orabductiveinference(inferencetothebestexplanation).Thecruxofthematteristhatallphylogeneticmethodspermitsomelevelofhomoplasy;inotherwords,phylogenetichypotheses(particulartreetopologies)donotlogicallyforbidanyparticularcharacterdistribution(Sober1988),makingitdifficulttoconcludethatphylogenetichypothesescanbefalsifiedinaPopperiansensebyphylogeneticcharacterdata.Nevertheless,theputativehypothetico-deductivenatureofcladisticsremainsanissueofvigorousdebateamongstsystematists(e.g.,deQueirozandPoe2001,Hull1999,Kluge1997,2001,Rieppel2003).Inadditiontotheargumentsdescribedabove,apotentiallyser-ious‘flyintheointment’forfalsificationisminsystematicsisthetreatmentofphylogeneticcharacterdata.Thestancetakenbymanycontemporarysystematiststhatcharacterdatamustnotbebiolog-icallyevaluatedcancauseaseriousunderdeterminationofphylo-geneticcharacters(whicharesupposedtobepotentialfalsifiersinthissystem).Thesesystematistseschewinvestigationsofpotentialcharacterinterdependence,developmentalorfunctionalcorrelationofcharacters,ordifferentialweightingofcharactersbecauseofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n220maureenkearneyconcernsaboutsubjectivity,andprefertouseanyandallobserva-tionsascharacterdatausingaglobalcongruencetest.However,intheabsenceofanycausalgroundingforcharacters,characterredef-initionandrecodingcaneasilyimmunizephylogenetichypothesesagainstrejection(seeSection4).Thus,the‘characterproblem’playsanimportantandneglectedroleinthedebateabouttheframeworkofphylogeneticinference.Felsenstein(1978)identifiedconditionsunderwhichparsimonymethodscouldbestatisticallyinconsistent,layingthegroundworkfortheriseofmaximum-likelihoodmethods(e.g.,Edwards1972,Fisher1925)inphylogeneticanalysis.Proponentsofmaximum-likelihoodapproachesarguethatrobusthypothesesofphylogeneticrelationshipsareobtainableonlyonthebasisoffairlyspecificassumptionsabouttheunderlyingevolutionaryprocess,andwiththeuseofrigorousstatisticalmethodsofanalysis(Hillis,Huelsenbeck,andSwofford1994).Unsurprisingly,theriseofmaximum-likelihoodmethodsinphylogeneticscoincidedwiththeincreasinguseofnucleotideposi-tionsinalignedDNAsequencesascharacterevidenceinsystematics,andaconcomitantinterestindevelopingmodelsofnucleotideevo-lution.Suchmodelsformamajorcomponentofmaximum-likelihoodalgorithmsforphylogeneticanalysis,andarealsoamajorpointofcriticismbydetractorsofthesemethods.Opponentsofmaximum-likelihoodphylogeneticmethodsarguethatlikelihoodanalysescanbeperformedonlyinthecontextofmodelsthatmakeoverlyrestrictive,simplifyingassumptionsaboutevolutionaryprocesses,andthatlikelihoodmethodsmaythem-selvesfailtobestatisticallyconsistentundercertainconditions(Kluge2001).Someauthorsargueagainstlikelihoodmethodsasinductiveand‘verificationist’incontrasttothepurportedlydeduc-tive/falsificationistnatureofcladisticparsimony,andhaveat-temptedtoexplicatearelationshipbetweenfalsificationismandcladisticparsimonyusingPopper’scorroborationformalism(Kluge1997,1999),aneffortthathasstimulatedthe‘Popperdebate’onceagain(deQueirozandPoe2001,FaithandTrueman2001,Farris,Kluge,andCarpenter2001,Kluge2001,Rieppel2003).SomesystematistsarguethatonlycladisticparsimonyconformstoPopper’sfalsificationistphilosophy;somearguethatlikelihoodmethodsofphylogeneticinferencearejustasconsistentwithPopper’sconceptofcorroborationasareparsimonymethods;someproposeaCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics221frameworkforphylogeneticsthatispurportedlybasedonPopperiancorroboration,yetnotonfalsificationism;andstillothersargueonceagainthatPopperianismhasnothingtodowithphylogenetics.Meanwhile,thefieldmarcheson.Mostrecently,Bayesianinferencemethodshavebeenappliedtophylogenetics(Huelsenbecketal.2001).Unlikecladisticmethods(whichidentifythephylogenetichypothesisthatismostparsimo-niousgivencertainassumptions),andunlikemaximum-likelihoodmethods(whichidentifythephylogenetichypothesisforwhichtheobserveddatahavethehighestprobabilitygivenacertainmodelofevolution),Bayesianmethodsidentifythephylogenetichypothesiswiththehighestposteriorprobability.Thelatterentityisdependentuponthepriorprobabilityofthehypothesisandontheprobabilityoftheobserveddatagiventhehypothesis.Asappliedtophylogeneticinference,aBayesiananalysisdeliverstheposteriorprobabilitydistributionoftreesbyassigningprobabilitiestotreesconditionalonthedata.OneofthemainargumentsagainstBayesianinferencemethodsinphylogeneticshasbeentheselectionofthepriorprob-abilities,whicharesubjective.Computationally,Bayesianphylo-geneticmethodsaremuchfasterthanmaximum-likelihoodanalysesintermsofanalyzinglargedatasetsandassessingsupportforalternativetrees,andmanysystematistspreferthemforthisreason.However,evaluationandcomparisonofsupportvaluesderivedfromBayesianversusmaximum-likelihoodanalysesarecurrenttopicsofdebate.MuchofthedebateoverthemeritsofBayesianmethodsmirrorsthatbetweencladistsandlikelihoodists,butthereisalsoanemergingdisagreementbetweenlikelihoodistsandBayesians,whichwillbeofinterestinthecomingyears.Ofcourse,Bayesianandlikelihoodmethodsweredebatedinstatisticalfieldslongbeforetheywereappliedtophylogenetics,andthosedebatesmaybeexpectedtobereplayedtosomeextentinthecontextofphylogeneticanalysis.Thediscussionsdescribedaboveillustrateindecisionamongsystematistsoverthepropermethodologicalframeworkforphylo-geneticinference,aswellassomeresistancetotheuseofexplicitlystatisticalapproaches.Manysystematistsstriveforahypothetico-deductivemodeofinferenceinphylogeneticanalysis.Somestriveforafalsificationistsystematics.AsattractiveasPopper’sphiloso-phyofsciencehasbeentosystematists,theabsenceofadeductiveCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n222maureenkearneylinkbetweenanyparticulartreetopologyandanyparticularchar-acterdistributionmakesitdifficulttojustifyphylogeneticmethodsonhypothetico-deductivegrounds(Sober1988).Otherfactors,suchasthenatureofphylogeneticcharacterstatements(seeSection4),mayalsofavorthisconclusion.3.thenatureofspeciesandhighertaxaLifeiswildlydiverse,butitisalsoperceptiblydiscontinuous;biodiversitycomprisesmoreorlessdiscreteentities,whichbiolo-gistscallspecies.Aconceptofspeciesisoneofthecoreconceptsofsystematicsandevolutionarybiology–thatofafundamentalunitofcomparisonandperhapsafundamentalinteractorintheevolu-tionaryprocess.Butwhatexactlyisthenatureoftheseentitiesthatsystematistsaretryingtoidentify,compare,andclassify?Thistopichasengenderedagreatdealofconceptualdiscussionanddebate.Biologicalspeciesconceptsarerootedintheprocessesthoughttocreatespecies(suchasreproductiveand/orgeographicisolation)andtomaintainspecies(suchasinterbreedingand/orcohesiveness).Thebiologicalspeciesconceptrejectstheuseofmorphologicaldis-tinctnessinrecognizingspeciesandinsteaddefinesspeciesasgroupsofpopulationsseparatedbyreproductivegaps:‘‘Speciesaregroupsofactuallyorpotentiallyinterbreedingnaturalpopulations,whicharereproductivelyisolatedfromothersuchgroups’’(Mayr1942).Thisconceptwaslaterrestatedas‘‘Aspeciesisareproductivecommunityofpopulations(reproductivelyisolatedfromothers)thatoccupiesaspecificnicheinnature’’(Mayr1982).Practicalproblemsinapplyingthebiologicalspeciesconcepttoalloflifeexist:asexual,polytypic,andhybridizingentitiesalloccurinnature–aretheyspecies?Thishasledsometosuggestthatapluralisticapproachtospeciesmaybenecessary(e.g.,MishlerandDonoghue1994).Inaddition,documentationofreproductiveprocessesinrealpopula-tionsisdifficultatbest.Simpson’s(1961)evolutionaryspeciesconceptallowsforasexualspecies:‘‘Anevolutionaryspeciesisalineage(anancestral-descendantsequenceofpopulations)evolvingseparatelyfromothersandwithitsownunitaryevolutionaryroleandtendencies.’’TheemphasisonprocessintheprecedingspeciesconceptscausedsomesystematiststonotethatoperationalizingtheseconceptstoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics223recognizespeciesisproblematic.Pheneticspeciesconceptsdefinespeciesonthebasisofoverallpheneticsimilarity–inotherwords,speciesaregroupsofsimilarorganisms.Pheneticistsbelievedthatbiologicalspecies,justlikeevolutionaryrelationshipsbetweenspecies,areunknowableintheabsenceof‘directproof’andreplacedthenotionofspeciesasthefundamentalunitofclassificationwith‘operationaltaxonomicunits’,orOTUs.Pheneticspeciesconceptsattempttoavoidtheoreticalinputandtomakespeciesidentifica-tionsstable:‘‘Wemayregardasaspecies(a)thesmallest(mosthomogeneous)clusterthatcanberecognizeduponsomegivencri-terionasbeingdistinctfromotherclusters,or(b)apheneticgroupofagivendiversitysomewhatbelowthesubgenuscategory’’(SneathandSokal1973,365).Butpheneticsimilaritymeasuresarearbitrary,anddifferentwaysofmeasuringsimilaritywillgivedifferent‘species’.Moreover,biologiststendtorejecttypologyandrecognizethatorganismswithinaspeciesarenotalwaysverysimilartoeachother;therearebothcrypticandpolytypicspecies.(Sincetheadventofmolecularbiology,manycrypticspecieshavebeendiscovered,makingspeciescriteriaandconceptsevenmorechallenging.)Phylogeneticspeciesconceptsidentifyspeciesassegmentsofaphylogenetictree:‘‘Aspeciesisthesmallestdiagnosableclusterofindividualorganismswithinwhichthereisaparentalpatternofancestryanddescent’’(Cracraft1983).Theemphasishereisoncladogenesis,andonthesystematist’sabilitytodiagnosespeciesthroughphylogeneticanalysis.Variouspermutationsofthephylo-geneticspeciesconceptexist.‘‘Wedefinespeciesasthesmallestaggregationsofpopulations(sexual)orlineages(asexual)diagnosablebyauniquecombinationofcharacterstatesincomparableindivid-uals(semaphoronts)’’(NixonandWheeler1990).Ingeneral,phylo-geneticspeciesconceptstendtofocusondiagnosability(NixonandWheeler1990)ormonophyly(Donoghue1985).Operationally,aspeciesisadiagnosablelineage(i.e.,whereafixedqualitativedif-ferencecanbeidentified).However,ifallthatisrequiredforspeciesstatusisasingledifferentiatingfeature,thenmalesandfemalescanbeseparatespecies,larvaandadultcanbeseparatespecies,andasinglemutationcancreateanewspecies.Despitethenumerouspublicationsdebatingthe‘speciesproblem’,theremaybemoreunityofopinionthanappearsonthesurface(deQueiroz2005).ThemajordifferencebetweenthemyriadCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n224maureenkearneyspeciesconceptsisbetweenthosethatemphasizetheprimacyofspeciationprocesses(e.g.,interbreeding,reproductiveorgeographicisolation)versusthosethatemphasizecriteriaforidentifyingordelimitingspecies(e.g.,monophyly).AccordingtodeQueiroz(2005),ifthedistinctionbetweenspeciesconceptsandspeciescri-teriaismadeclear,thenthereismoreunderlyingcommonalityamongvaryingspeciesconceptsthanonemightimagine.Thatcommonalityis,‘‘Speciesaresegmentsofpopulation-levelevolu-tionarylineages.’’Asidefromspeciesconcepts,howtothinkaboutspeciesisanothertopicofmuchcurrentdiscussioninthefield,andthatdis-cussionisoftenphilosophicallybased.Hull(1965,1976)andGhiselin(1974)arguedthatevolutionarytheoryprecludesviewing1speciesasclassesornaturalkindsbecauseclassesandkindsaretiedtoanessentialismthatisinconsistentwithanevolutionaryworld-view.Ratherthanspeciesrepresentingcollectionsoforganismsmeasuredbysomedegreeofsimilarity,bysomedefiningfeature,orbynecessaryandsufficientconditions,theseauthorsarguethatspeciesarediagnosedbytheirhistory.Forexample,despitetheabsenceoflimbs,asnakeisatetrapodbyvirtueofitsphylogenetichistory(asnakedoesnothave‘nolegs,’but‘modifiedlegs’).TherelateddistinctionbetweenclassificationandsystematizationmadebyGriffiths(1974)hasalsobeenhighlyinfluential.Classesorsetsimpartamembershiprelation,whichmakesitdifficulttorevisethemempirically.Incontrast,individualsareparticularswithspatiotemporalextension;theyarenotsubjecttoamembershiprelationbuttoapart-wholerelation.Thus,speciesaresaidtobe2conceptualizedasindividuals.Whetherspeciesandhighertaxacanalternativelybeviewedashomeostaticpropertycluster(HPC)nat-uralkindswithinarealistperspective(rejectingstrictlydichot-omousthinkingaboutclassesversusindividuals)isanotherissuestillbeingdebated(Boyd1991,1999,deQueiroz1992,Ghiselin1997,Keller,Boyd,andWheeler2003,Mayr1987,Ruse1987).Suchaviewpointrequiresdivorcingthetraditionalconceptofnaturalkindsfromdefinitionsbasedonnecessaryandsufficientconditionsinordertoaccommodatethecomplexityofthebiologicalworld.Indeed,somebiologistsandphilosophersviewthestrictclass/indi-vidualdistinctionasinadequate(e.g.,Grene1990,2002,Griffiths1999,Kelleretal.2003,Mayr1987,Rieppel2006).TheseauthorsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics225suggestthat,whilevariationisprominent,itisnotthesolefeatureofthenaturalworld;thefactthatweareabletorecognizedifferentspeciesandmakescientificallyinterestinggeneralizationsillus-tratessomethingmorethanstrictindividualism.Nevertheless,itisevidentthattheindividualitythesisforspecieshashadahugeimpactonthefieldofphylogenetics,includingthecurrenteffortbysometooverturnthetraditionalandlong-standingrank-basedsystemforgoverningtaxonomicnames.Forthepast250years,theLinnaeanhierarchyhasformedthebasisoftaxonomy,withrankedtaxonomiccategories(Kingdom,Class,Order,etc.)towhichtaxaareassignedduringclassification.Acomplexsetofrulesandconventionsgoverningthenamingoftaxaisalsoanintegralpartoftraditionaltaxonomy.AproposedchallengetotheLinnaeansystemoftaxonomyisaphylogeneticsystemoftaxonomybasedontheevolutionaryprincipleofdescentwithmodification.Proponentsofphylogenetictaxonomy(e.g.,deQueiroz1992,deQueirozandGauthier1990,1994)arguethatspeciesandhighertaxashouldbeorderedintoanaturalsystembasedontheirgenealogicalrelation-shipsratherthanthepossessionofdefiningcharacteristics.Oneofthecentralissuesinphylogenetictaxonomyisthemannerinwhichtaxonnamesaredefined.UndertheLinnaeansystem,thenameofafamilyoforganismsmightbedefinedasthefamilythatcontainscertainlower-leveltaxa;underthephylogeneticsystem,thatfamilynamewouldbedefinedasthemostrecentcommonancestorofthelower-leveltaxa,plusallofitsdescendants.Thus,theconceptualdrivingforcebehindthedevelopmentofthe‘PhyloCode’(CantinoanddeQueiroz2003)isrejectionoftheessentialismbelievedtounderlietheLinnaeansystemofclassification.Detractorsofthis‘nomenclaturalrevolution’argueagainstphylogeneticnomen-clatureonvariousgrounds–empirical,philosophical,andpractical(e.g.,Kelleretal.2003,NixonandCarpenter2000,Rieppel2006).Apparently,whetherornot‘PhyloCode’successfullyescapesessentialismviatheostensivedefinitionoftaxonnamesremainsamatterofdebate.Inaddition,pragmaticissuesofnomenclaturalstabilityareofgreatconcerntobothsidesofthedebate.Theulti-mateacceptanceorrejectionofphylogeneticnomenclatureversusthelong-standingrank-basedsystemwillbeoneofthemoreinter-estingareastofollowinthecomingyearsforbothsystematistsandphilosophersofscience.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n226maureenkearney4.thenatureofphylogeneticevidenceSystematistsareinthebusinessoftryingtoevaluatealternativephylogenetichypothesesforvariousgroups.Theyhaveonlytheendproductsofthebranchingprocess–organismsandtheircharacter-istics–thatcanbeobservedtodayandusedasevidenceformakinginferencesaboutphylogeneticrelationshipsamongsttaxa.Featuresthatdiagnosegroupsareproposedtobehomologues.Becausetherelationofhomologyisanunobservable(i.e.,becausehomologyisidentifiedbycomplexinferencesratherthansimpleobservation),characterstatementsthatarebasedonobservedsimilaritiesanddifferencesinphenotypicorgeneticdataareusedasevidenceinphylogeneticanalysis.Today,thosecomparativeobservationsaretypicallytransformedintonumericalcodesandenteredintoadatamatrix(characters·taxa).Someoptimalitycriterion(e.g.,parsi-mony,maximum-likelihood)isthenusedtoanalyzethatdatamatrix,usuallywiththeaidofacomputerprogram,andtoobtainaphylogenetichypothesis.Fromtheverybeginningofthehistoryofsystematics,therehasbeengreatdifficultyindeterminingwhattheusefulphylogeneticcharactersoforganismsmightbe.Thenatureofphylogeneticchar-acterevidenceandtheidentificationofcharacterscontinuetogen-eratecontroversyinthefield.Evolutionarytheoryandcomparativestudiestellusthatorganismsaremadeofpartsthatare,tosomeextent,dissociable,recombinable,andchangeableovertime.Thesepartsaretheevidence,ordata,ofbiologicalsystematics.Butwhatexactlyconstitutesapart?Itisclearlyinappropriatesimplytoreduceorganismstoaggregatesoffeatures,characters,orrawobservationsbecauseorganismsaredevelopmentallyandfunction-allyintegratedwholes.However,phylogeneticanalysisrequiresthedecompositionoftheorganismalwholeinordertogeneratechar-acterdataforphylogeneticanalysis.Asaresult,toproposephylo-geneticcharactersisfarfromtrivial–amongotherthings,thesystematistmustdecidewhetheranobservedfeatureisone,two,ormanycharacters,andwhetheraspecificcharacterisareliableindicatorofhomologyorpossiblyamisleadingconvergence.Mostsystematistsagreethatthecharacterscapableofindicatingphylo-geneticaffinityarenotjustanyfeatures,butevolutionaryhomo-logues.And,atleastsinceDarwin,thedefinitionofhomologyforCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics227mostbiologistsisacorrespondenceofpartsduetocommondescent.Fromthisviewpoint,itwouldseemthatinsightintounderlyingcausalityincharacterevolutionwouldbehelpfultosystematistsintheirworkofidentifyingandcodingcharacters.However,historyshowsthatthisisnotalwaysthecase,andforfamiliarreasons.Theevolutionarytaxonomists’approachtohomologyandchar-acterswasrootedinextensiveorganismalstudies,andcharacterweightingwasbasedonpresumedphylogeneticreliability.Issuessuchaspotentialnon-independenceofcharactersduetoevolu-tionaryprocessesofconstraint,selection,adaptation,andcorrela-tionwereconsideredveryimportant.Suchevaluationisadmittedlyimprecise,requiringjudgmentsabouttherelativephylogeneticutilityoforganismalfeatures,acomprehensiveunderstandingofthecharactersandorganismsunderstudy(the‘expertproblem’),andconsiderationofevolutionaryprocessesactinguponcharacterevolution(aconsiderationthatmanysystematistsseeastooassumption-laden).Hennigianphylogeneticsystematicsalsoemphasizedinitialcharacteranalysisasanecessaryguidetohomology.Hennig(1966)usedavarietyofcriteria–detailedcomparativemorphologicalstud-ies,topology,connectivity,ontogeny,functionalanatomy,geologi-calprecedenceinthefossilrecord,andecology–toidentify,analyze,andpolarizecharacters.Evaluationofcharacterqualityandutilitywasbasedonboththeoreticaljustificationsandempiricalinvesti-gations.Althoughonemaydisagreewiththeuseofanyoralloftheseguidelinesforcharacterdelineation,itisinstructivetonotethathomologywassomethingtobecomprehensivelyinvestigatedpriortotreeconstructionforHennig,notsolelytheresultofphy-logeneticanalysis.Characterqualityandutilitywereevaluatedusingtheoreticaljustifications,empiricalinvestigations,andesti-mationsaboutthelikelihoodofconvergenceversushomology(seealsoHennigandSchlee1978).Pheneticistsconsideredsuchjudgmentsaboutcharactersarbi-traryandsubjective.SokalandSneath(1963,87)emphasizedthatapproachestocharacterdataneednotbebasedonbiologicaleval-uation,butshouldbeobjective,explicit,quantitative,andrepeat-able:‘‘Onewaytodealwithproblemsofhomologyistoignoredetailsofstructure.’’(Itisimportanttonotethatinthisconceptionof‘objectivity’boththeorydependenceandqualitativedescriptionsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n228maureenkearneyofcharacterstatesdiminish‘objectivity’.)Fundamentally,thephe-neticapproachtocharacterdatareducescharacterstorawobserva-tions,andthisuncriticalempiricismisonefactorthatultimatelyledtothemethod’sdemise.However,theoverallphilosophydoesnotseemtohavebeencompletelyovercomeinmodernsystematics,atleastformorphologicalcharacters.Somecontemporarysystematistsparadoxicallyacknowledgethatnotheory-freeobservationispossible,yettheyrejecttheoreticalandempiricalevaluationsofcharactersinfavorofaputativelyrigorousmethodoftesting–congruenceofcharactersrelativetoahierarchy.Arelatedargumentemphasizesourignorancewithrespecttoallofthecausalcorrelatesofphylogeneticallyinformativecharactersandseeksasunbiasedanapproachtocharacterdelineationasishumanlypossible.Bothapproachesmaintainthatbiologicalevaluationofcharactersisirrelevantandimpossible,andthatanyobservationcanbeacharacter,andbothultimatelydefertocongruenceunderpar-simonyasthesolemethodoftestinghomology.Citingtheprincipleof‘totalevidence’,theyadvocatethatphylogeneticstudiesshouldincludeallpreviouslypublishedcharacterdatainaglobalcon-gruencetest,thisbeingthemostobjectiveandrigorouswaytotestcharactersandhomology.Thisstancehasgeneratedanewdebateaboutthe‘characterproblem’amongstsystematists(e.g.,KearneyandRieppel2006,Kluge2003,RieppelandKearney2002).Theheartofthedebateseemstobethatsomesystematistsgivethephylogenetictreelogicalpriorityovercriticalcomparativestud-iesofcharacterdata–fromsuchaviewpoint,itisonlythetree,notempiricalcharacterevaluation,thatcaninformusabouthomologyandwhatalegitimatecharactermightbe(Ha¨rlin1999).Othersystematistsacknowledgethelimitsanddifficultyofcharacterevaluationbutareuncomfortablewiththecontentionthatknowledgeofhomologyandphylogenycanbederivedfromthesimplecoher-enceoftheory-freeobservationreports.AsRuse(1988,60)notes:‘‘Assoonasonestartsbreakingorganismsintoparts,onemustbringintheory...Taketwobears,onewhiteandonebrown.Dotheydifferinonefeature,ordoesonetakeeachhairseparately...Thepointiswhethersomeonewhoexplicitlyeschewsthetheoryhastherighttocombineallthehairsintoonefeature.’’Itisinstructivetonoteintoday’scontextthatnumericaltax-onomistspreviouslystressedthe‘empiricalapproach’intaxonomy,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics229withanemphasison‘firmobservation’ratherthanphylogeneticorevolutionaryassumptions.Today,mostsystematistswouldagreethatnosuchtheory-free‘observationlanguage’exists,yetmanystilladmit(atleastpotentially)anyobservationreportintothetotalevidenceunderevaluationanddisallowempiricalrejectionofthesame.Oneconcernaboutthisapproachisthethreatofinstru-mentalism–thatcharacterstatementsmaybecomemereinstru-mentsusedtoachieveahypothesisofphylogeny,ratherthanbeinggroundedempiricallyandcausallyintheorganismsunderstudy.Arelatedconcernisthatthestanceagainstevaluationofcharacters,oragainstanycriteriaforhomologyhypotheses,cancauseaseriousunderdeterminationofphylogenetichypotheses(Richards2002,2003).Throughdefinitionandredefinition,virtuallyanycharacterstatement(certainlyofmorphologicalcharacters)canbemadetocoherewithanysetofothersuchstatements,andthroughsplittingorlumpingofthenumberofcharacterstatements,thesamecanbeachieved.Thisisparticularlytrueif‘anything’canbeacharacteronthesoleconditionofitscoherencewithothercharactersrelativetoahierarchy.Thus,whilecoherenceofcharacterstatementsrelativetoahierarchymaybeanecessaryconditionofphylogenyreconstruc-tion,itseemsunlikelytobeasufficientcondition.Theclaimthatseverityoftestincreasesexclusivelywithanincreasingnumberofcharactersusedinphylogeneticanalysis,nomatterthenatureofthosecharacters,alsoseemsquestionable.Thismightbetrueifeachcharactercorrespondedtosomebitofinfor-mationthatcouldbeempiricallygraspedbyeveryworkingsystematistandwerefullyindependentfromallotherbitsofinfor-mation.This,however,isnotthecase,forbiologicalaswellasepistemologicalreasons.Incontrast,tobringtheinsightsofdevel-opmentalbiology,functionalanatomy,andotherevolutionaryconsiderationstobearoncharacterdelineationandinterdependenceappliestheorytotheproblemofcharacterdelineation.Criteriasuchastopologicalcorrespondenceandconnectivityhavemoreorlesssuccessfullybeenusedtohelpmakethecommonhistoricaloriginofhomologuesempiricallyaccessible,eveninfaceofthefactthattopologicalrelationshipscanthemselvesevolve.Itisassumedthatthisisso,notbecauseofanyarbitrarynotionofsimilarity,norbecauseofamerelyconventionaluseoftopologyandconnectivityinthesearchforhomology,butbecausetheseguidesareatleastCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n230maureenkearneyapproximatelyalignedwithcausalevolutionaryanddevelopmentalprocesses.Suchcriteriaarearguablywhatallowtranscendenceof‘primitive’similarity(i.e.,theoutermostearossicleofamammalandthelowerjawofasharkarenotphenotypicallysimilarbuttheysharesimilartopologicalrelations;suchguidelineshavearguablyledtothesuccessfuldiscoveryofhomologywhereas‘primitive’similaritycouldnot).Butsystematistsalsorecognizethatsuchcriteriaarenotfoolproof,andthuscharactercongruenceisanimportantpartofevaluatinghomologyhypotheses.5.discussionandconclusionsAlthoughconceptualandmethodologicaldialoguesinsystematicsseemtoreplayaneternaldebateindifferentforms,thefieldhasalsotranscendedthesedebatestoagreatextent–realprogresshasbeenmadeinunderstandingthetreeoflifeformanygroups,andsys-tematicscontinuestobecomemoreandmoreintegratedwithotherareasofevolutionarybiology.Itisnowrecognizedasthefoundationforresearchinevolutionarybiology,ecology,behavior,andbiogeog-raphy.Inaddition,thefieldcontinuestobeinfluencedbynumerousdevelopments,fromnewdiscoveriesaboutevolutionarymechan-ismsofinheritanceanddevelopment,tothewidespreaduseofcomputersthatcananalyzelargeamountsofdata,tonovelmethodsforextractingandsequencingDNA,andothers.Yet,containedwithinthedebatesdescribedaboveisevidenceofapersistentstrugglewithnotionsofobjectivity,theorydependence,andtestability.Thiswasexpressedinthemethodologicaldebatebetweenpheneticistsandevolutionarytaxonomists,andinthedif-ferentmethodologicalviewpointsofphylogeneticcladistsversuspatterncladists.Today,asimilartensionexistsbetweenlike-lihoodistswhoseektoincorporateinformationabouttheevolu-tionaryprocessintosystematicsthroughmodel-basedanalyses,andothersystematistswhorejecttheuseofthesemodelsastootheo-reticallyassumptive.Withindebatesaboutspecies,somesuggestthatspeciesarethesmallestphylogeneticallydiagnosableunits,whereasotherssuggestthatsomethingmoremaybenecessary.Differentapproachestocharacterdataalsoreflectthistheme.Pheneticistsadvocatedanalyzingasmanytraitsaspossible‘objec-tively’intoquantitativeunitcharacters,incontrasttothebiologicallyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nPhilosophyandPhylogenetics231steepedapproachofevolutionarysystematics.Earlycladistsrejectedthetenetsofnumericaltaxonomy,yetphenetictendenciesincharacterdelineationpersist.Concernsabout‘objectivity’anditsconnectionto‘testability’haveledsystematiststocritiqueandsometimesrejectmethodsthataredependentupontheoriesorjudgment.However,attemptstoavoidtheoryandtrainedjudgmentinphylogeneticsoftenreachdeadends,whichmayillustratethatsuchavoidancedoesnotwork.Thecharacterdebateisanexcellentexampleofthis–relianceonatheoreticalobservationsascharactersyieldsthepredicamentofmyriad,user-definedwaystodelineatecharacters,andanapproachthatfailstotranscendsubjectivity.Indeed,intheabsenceofcausalgrounding,observationssimplybecomemoredefinitionalandphy-logenetichypotheseslesstestable.Incontrast,itmaybearguedmoresuccessfullythatlinkingobservationstocausalmechanismsmayincreaseobjectivity.Manysystematistsandphilosophersofbiologyhavenotedthattheinfluenceofevolutionarytheoryhasnotyetbeenfullyintegratedinsystematics.Oneexplanationofferedforitsincompleteintegra-tionisthatsystematistsstillfailtograspthedistinctionbetweenclassificationandsystematization–thatis,thedistinctionbetweenorderingthingsintoclassesonthebasisofpropertiesandorderingthingsintosystemsonthebasisofanaturalprocessthroughwhichtheirpartsarerelated(e.g.,deQueiroz1988).Perhapsthereisanotherreason,onethatmayberesolvedbyfurtherdiscussionsbetweenphilosophersandsystematists:incorporatingtheoreticalandcausalconsiderationsintophylogeneticsresearchwithoutsacrificingobjectivityortestabilityhasprovedtobedifficult.Fertilegroundforfuturediscussionbetweensystematistsandphilosophersliesinthecriticalexaminationofwhatitmeanstobeobjectiveandscientificwithinanevolutionaryworldview.ACKNOWLEDGMENTSThankstoDavidHullandMichaelRuseforinvitingmetocontributetothisvolume,andtoR.Boyd,K.deQueiroz,D.Hull,A.Larson,B.Patterson,R.Richards,andO.Rieppelforreadingthechapterandofferinghelpfulcriticism.Thisworkwassupported,inpart,bygrantsfromtheNationalScienceFoundation(DEB-0235628andEF-0334961).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n232maureenkearneynotes1.‘Class’hasaspecialuseinthisdebate,meaningsomethingcloseto‘setdefinedbynecessaryandsufficientahistoricalmembershipconditions.’Onalternativeconceptions(e.g.,Boyd,1991,1999),speciesandhighertaxacouldbehistoricallydefinedkindsthatlacknecessaryandsufficientdefiningconditions,ratherthanindividuals.2.Itisnotclearthatthehomeostaticclusteringofcharactershonorstheprevailingconceptionofmonophyly,makingtheHPCconceptionforhighertaxapotentiallymorecomplexthanthatforthespecieslevel.Atthespecieslevel,boththeHPCconceptionandthespecies-as-individualsapproachmaybeabletoexplainthehistoricityofspecies.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nfranciscoj.ayala12HumanEvolutionTheThreeGrandChallengesofHumanBiologyManisbutareed,theweakestinnature,butheisathinkingreed.BlaisePascal,Pense´es,number347asummaryoftheargumentHumanbiologyfacesthreegreatresearchfrontiers:ontogeneticdecoding,thebrain-mindpuzzle,andtheape-to-humantransfor-mation.Byontogeneticdecoding,ortheegg-to-adulttransformation,Irefertotheproblemofhowtheunidimensionalgeneticinforma-tionencodedintheDNAofasinglecellbecomestransformedintoafour-dimensionalbeing,theindividualthatgrows,matures,anddies.Cancer,disease,andagingareepiphenomenaofontogeneticdecoding.Bythebrain-mindpuzzleIrefertotheinterdependentquestionsof(1)howthephysicochemicalsignalsthatreachoursenseorgansbecometransformedintoperceptions,feelings,ideas,criticalarguments,aestheticemotions,andethicalvalues;and(2)how,outofthisdiversityofexperiences,thereemergesaunitaryreality,themindorself.Freewillandlanguage,socialandpoliticalinstitutions,technologyandart,areallepiphenomenaofthehumanmind.Bytheape-to-humantransformationIrefertothemysteryofhowaparticularapelineagebecameahominidlineage,fromwhichemerged,overonlyafewmillionyears,humansabletothinkandlove,todevelopcomplexsocietiesandsubjecttoethical,aestheticandothervalues.Thehumangenomedifferslittlefromthechimpgenome.233CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n234franciscoj.ayalaTheegg-to-adulttransformationisessentiallysimilar,andsimilarlymysterious,inhumansandothermammals.Thebrain-to-mindtrans-formationandtheape-to-humantransformationaredistinctivelyhuman;theydefinethehumanum,thatwhichmakesusspecificallyhuman.Nootherissuesinhumanevolutionareofgreatercon-sequenceforunderstandingourselvesandourplaceinnature.Erectpostureandlargebrainaretwoofthemostsignificantana-tomicaltraitsthatdistinguishusfromnonhumanprimates.Buthumansarealsodifferentfromchimpanzeesandotheranimals,andnolessimportantly,intheirbehavior,bothasindividualsandsocially.Distinctivehumanbehavioralattributesincludetoolmak-ingandtechnology;abstractthinking,categorizing,andreasoning;symbolic(creative)language;self-awarenessanddeathawareness;science,literature,andart;legalcodes,ethics,andreligion;complexsocialorganizationandpoliticalinstitutions.Thesetraitsmayallbesaidtobecomponentsofhumanculture,adistinctivelyhumanmodeofadaptationtotheenvironmentthatisfarmoreversatileandsuc-cessfulthanthebiologicalmode.Culturaladaptationismoreeffectivethanbiologicaladaptationbecause(1)itsinnovationsaredirected,ratherthanrandommutations;(2)itcanbetransmitted‘‘horizontally,’’ratherthanonly‘‘vertically,’’todescendants;and(3)becauseculturalheredityisLamarckian,ratherthanMendelian,acquiredcharacteristicscanbeinherited.lifetohumanTheoldestknownfossilremainsoflivingorganismsaredatedsomewhatearlierthan3,500millionyearsago,justafewhundredmillionyearsaftertheEarthhadcooled.Theorganismsweremicroscopic,individualcells,buthavingalreadyconsiderablecomplexityoforganizationandelaboratebiochemicalmachinerytocarryonthefunctionsoflife.Wedonotknowwhenlifestarted,butitlikelywasatleastonehundredmillionyearsearlier.Thereareseveralhypothesesabouthowlifefirststarted,butnoneofthesehypothesesissufficientlywellsupportedbyevidenceand,thus,noneofthemisacceptedbyallscientists.Butthefactthatittook‘‘only’’oneorafewhundredmillionyearsfromtheformationoftheEarthtotheappearanceofthefirstsingle-cellorganisms,suggeststhatlifeinsomeformislikelytoappearinanyplanetthatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution235haswaterandafewotherelements(notably,inourplanet,carbon,nitrogen,phosphorus,andsulfur).Thetemperaturemustalsobe‘‘right’’,withinacertainrange,asitisthecaseforplanetEarth,becauseofthe150millionkilometersthatseparateitfromtheSun,sothatwatercanexistinliquidphase(ratherthanonlyaseithericeorvapor,ifthetemperatureistoolowortoohigh).TherearethreelargegroupsoforganismsonEarth:eucaryotes,bacteria,andarchaea.Theeucaryotesincludeanimals,plants,andfungi.Eucaryotesareorganismsthathavetheirgeneticmaterialenclosedinaspecialcapsule,ororganelle,calledthenucleus.Humansareeucaryotes.Animals,plants,andfungiaretheonlyorganismsthatwecandirectlyexperiencewithoursenses,andthustheyweretheonlyorganismswhoseexistencewasknowntohumansuptothreecenturiesago.Yettheyaccountforonlyafractionofthetotaldiversityoftheeucaryotes.Theothereucaryotesareallmicroscopic.Somecausewell-knowndiseases,suchasPlasmodium,whichcausesmalaria,orEntamoeba,whichcausessevereintestinalmaladies.Asecondgroupoforganismsarethebacteria.Humanshaveknownoftheexistenceofbacteriaformorethanacentury.Weassociatethemwithdiseases,butbacteriaperformmanyusefulfunctions,includingtheincorporationofnitrogenfromtheatmosphere,nitrogenthatani-malsandplantsneedbutarenotabletogetdirectlyfromtheatmo-sphere(whereitisveryabundant,about75percentofthetotal;therestismostlyoxygen).Also,bacteriaareresponsibleforthedecompositionofdeadmatter,aprocessthatisessentialinthemaintenanceofthecycleoflifeanddeath,becauseitmakesagainavailable,forneworganisms,valuablecomponentsthathadbeenincorporatedintothenowdeadorganisms.Thegeneticdiversityandnumberofspeciesofbacteriaareatleastaslargeasintheeucaryotes.Therearemanymorekindsofbacteriathantherearekindsofanimals,plants,andfungicombined.Andtheyaresoabundantthattheirtotalweight(their‘‘biomass’’)isatleastasgreatas(andprobablymuchgreaterthan)thatofallplants,fungi,andanimalscombined,eventhoughindividuallytheyaresomuchsmaller.Thisisahumblingthought.Weseeour-selves,thehumanspecies,asthesummitoflifeandwearethemostnumerousofalllargeanimals;andweseeanimalsandplantsasthedominantformsoflifeonEarth.However,modernbiologyteachesasthat,numericallyaswellasinbiomass,thenearlytwomillionknownspeciesofanimals(includinghumans)amountonlytoaverysmallCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n236franciscoj.ayalafractionoflifeonEarth.Fromtheperspectiveofnumbersandbiomass,thebacteriaalonecountmuchmorethanwedo.Thereisanothergroup,thearchaea,likelytobeaboutaslargeastheeucaryotesorthebacteria.Theexistenceofthearchaeaisaveryrecentdiscoveryofmolecular(modern)biology.Becausetheseorganismsdonotdirectlyinteractmuchwithus,biologistswerenotawareoftheirexistence.Threedecadesagoscientistsonlyknewafewspecies,suchasthosethatexistinthehotspringsofYellowstoneNationalParkintheUnitedStatesandinothervolcanichotsprings,wheretheythriveattemperaturesapproachingtheboilingpointofwater.Biologiststhoughtthattheseweresomeunusualformsofbacteria.Nowweknowthemtobelongtoaverydiverseandnumerousgroupoforgan-isms,abundantinthetopwaterlayersoftheseasandoceans.Abucketofseawaterstudiedwiththemoderntechniquesofmolecularbiologymayyieldtensorhundredsofnewarchaeaspecies.ThenumberoflivingspeciesonEarthisestimatedtobebetween10and30million,butsomebiologiststhinkthattheremaybeasmanyas100millionspecies,ifbacteriaandarcheaeaareincluded.Animalsrepresentasmallfractionofallspeciesnowliving.Morethan99percentofallanimalspeciesthatlivedinthepasthavebecomeextinctwithoutissue.Thisismostlikelytrueforallotherkindsoforganismsaswell.Thus,thetotalnumberofspeciesthathaveexistedsincethebeginningoftheEarthismorethanonebil-lion.Wehumansarebutoneofthem.Humansareanimals,butaverydistinctanduniquekindofani-mal.Ouranatomicaldifferencesincludebipedalgaitandanenor-mousbrain.Butwearenotablydifferentalso,andmoreimportantly,inourindividualandsocialbehaviors,andintheproductsofthosebehaviors.Withtheadventofhumankind,biologicalevolutiontranscendeditselfandusheredinculturalevolution,amorerapidandeffectivemodeofevolutionthanthebiologicalmode.Productsofculturalevolutionincludescienceandtechnology;complexsocialandpoliticalinstitutions;religiousandethicaltraditions;language,literature,andart;radioandelectroniccommunication.humanoriginsOurclosestbiologicalrelativesarethechimpanzees,whoaremorecloselyrelatedtousthantheyaretothegorillas,andmuchmoreCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution237thantotheorangutans.(Thechimpanzeesincludetwospeciescloselyrelatedtooneanother,butbothequallyrelatedtohumans,Pantroglodytes,orcommonchimpanzee,andPanpaniscus,orbonobo.)Thehominidlineagedivergedfromthechimpanzeelineage7–8millionyearsago(mya)anditevolvedexclusivelyintheAfricancontinentuntiltheemergenceofHomoerectus,somewhatbefore1.8mya(Cela-CondeandAyala2001).ThefirstknownhominidsaretherecentlydiscoveredSahelanthropustchadensis(dated6–7mya;Brunetetal.2002;Vignaudetal.2002),Orrorintugenensis(dated5.8–6.1mya;Senutetal.2001),andArdipithecusramidus(dated5.2–5.8mya;Haile-Selassie2001).Theywerebipedalwhenontheground,butretainedtree-climbingabilities.Itisnotcertainthattheyallareinthedirectlineofdescenttomodernhumans,Homosapiens;rather,somemayrepresentsidebranchesofthehominidlineage,afteritsdivergencefromthechimpanzeelineage.Aus-tralopithecusanamensis,dated3.9–4.2mya,washabituallybipedalandhasbeenplacedinthelineofdescenttoAustralopithecusafarensis,Homohabilis,H.erectus,andH.sapiens.Otherhomi-nids,notinthedirectlineofdescenttomodernhumans,areAus-tralopithecusafricanus,Paranthropusaethiopicus,P.boisei,andP.robustus,wholivedinAfricaatvarioustimesbetween3and1mya,aperiodwhenthreeorfourhominidspecieslivedcon-temporaneouslyintheAfricancontinent(seeCela-CondeandAyala2001foranextensivereviewofhominidevolution).ThefirstintercontinentalwandereramongourancestorswasH.erectus.ShortlyafteritsemergenceintropicalorsubtropicaleasternAfrica,H.erectusdispersedtoothercontinentsoftheOldWorld.FossilremainsofH.erectusareknownfromAfrica,Indo-nesia(Java),China,theMiddleEast,andEurope.H.erectusfossilsfromJavahavebeendated1.81±0.04and1.66±0.04mya,andfromGeorgiabetween1.6and1.8mya.AnatomicallydistinctiveH.erectusfossilshavebeenfoundinSpainandinItaly,depositedabout800,000yearsago,theoldestknowninWesternEurope.FossilremainsofNeanderthalhominids(Homoneanderthalensis),withbrainsaslargeasthoseofH.sapiens,appearedinEuropearound200,000yearsago(200kya)andpersisteduntil40kya.TheNeanderthalswerethoughttobeancestraltoanatomicallymodernhumans,butnowweknowthatmodernhumansappearedatleast100kya,muchbeforethedisappearanceoftheNeanderthals.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n238franciscoj.ayalaMoreover,incavesintheMiddleEast,fossilsofmodernhumanshavebeenfounddatednearly100kya,aswellasNeanderthalsdatedat60and70kya,followedagainbymodernhumansdatedat40kya.Itisunclearwhetherthetwoformsrepeatedlyreplacedoneanotherbymigrationfromotherregions,orwhethertheycoexistedinthesameareas.RecentgeneticevidenceindicatesthatinterbreedingbetweenH.sapiensandH.neanderthalensisneveroccurred.Theoriginofanatomicallymodernhumansiscontroversial.SomeanthropologistsarguethatthetransitionfromH.erectustoarchaicH.sapiensandlatertoanatomicallymodernhumansoccurredconsonantlyinvariouspartsoftheOldWorld.Proponentsofthis‘‘multiregionalmodel’’callattentiontofossilregionalcon-tinuityinthetransitionfromH.erectustoarchaicandthenmodernH.sapiens.Theypostulatethatgeneticexchangeoccurredfromtimetotimebetweengeographicallyseparatepopulations,sothatthespeciesevolvedasasinglegenepool,eventhoughgeographicdifferentiationoccurredandpersisted,justasgeographicallydiffer-entiatedpopulationsexistinotheranimalspeciesandinmodernhumans.ThisclaimofinterbreedingbetweenH.erectuspopula-tionsdependsonthepostulateofpersistentmigrationsandinter-breedingbetweendistantpopulations,evenfromdifferentcontinents,ofwhichnodirectevidenceexists,althoughitisnottheoreticallyunlikelytohaveoccurred.However,itisdifficulttoconciliatethemultiregionalmodelwithfossilevidenceofthecontemporarycoex-istenceofdifferentspecies(H.erectusandH.sapiens)orforms(archaicandmodernH.sapiens)inChina,Indonesia,andotherregions.OtherscientistsargueinsteadthatmodernhumansfirstaroseinAfricabetween150kyaand100kya,andfromtherespreadthroughouttheworld,replacingelsewherethepreexistingpopula-tionsofH.erectusorarchaicH.sapiens.Thisiscalledthe‘‘OutofAfrica’’hypothesis,whichisnowfavoredbymostevolutionists.GeneticandmolecularevidenceshowsgreaterdifferencebetweenAfricanandnon-Africanpopulationsthanbetweenallnon-Africanhumanpopulations.ThispatternofdifferentiationendorsesthehypothesisthattheoriginofanatomicallymodernhumanswasinAfrica,whencemodernhumansexpandedtotherestoftheworld,startingabout100kya.Itisnotpossible,however,toexcludecompletelyapartialparticipationofarchaicH.sapiensfromtheOldCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution239Worldintheoriginofmodernhumans.SomeobservationsevincethepersistenceofolderanatomicaltraitsinmodernhumanpopulationsofCentralEuropeandtracesofancientmitochondrialDNAhavebeenfoundinAustralianpopulations(Wolpoffetal.2001,Adcocketal.2001).Inanycase,geneticanalysissupportstheoccurrenceofatleasttwo,notjustone,majormigrationsoutofAfrica,wellaftertheoriginalrangeexpansionofH.erectus(Templeton2002).IwroteearlierthatHomosapiens,ourspecies,isonlyoneofmorethanonethousandmillionspeciesthathavelivedonEarthsincethebeginning.Fromthatperspective,humansarebutaspeckonourplanet.Thisisalsothecasefromtheperspectiveoftime.Thehominidsdivergedfromtheapesabout7–8mya,andmodernhumanscomeintoexistenceabout100kya.Yet,lifehasexistedonEarthformorethan3,500my.Itisdifficulttothinkinmillionsofyears.Soletmetransformthetimelineofevolutionintoaone-yearscale,sothatlifearisesinourplanetonJanuary1,atzerohours,andsothatitisnowmidnightonDecember31.Inthisone-yearscale,forthefirsteightmonthsthereisonlymicroscopiclife;thefirstanimalsappeararoundSeptember1;theyaremarineanimals.ThelandiscolonizedaroundDecember1;theprimatesoriginateonDecember26;thehominidsseparatefromthechimpanzeesonDecember31,atnoon;andmodernhumansariseonthatlastdayoftheyearattwenty-threehoursforty-fiveminutes.Wehavebeenaroundforatotaloffifteenminutes.Thatalsoisahumblingthought.ButIhastentoaddthateventhoughweare‘‘butareed,’’asPascalfamouslyputit,weareathinkingreed,andtothisIshallpresentlyreturn.thehumangenomesequenceBiologicalheredityisbasedonthetransmissionofgeneticinfor-mationfromparentstooffspring,inhumansverymuchthesameasinotheranimals.ThegeneticinformationisencodedinthelinearsequenceoftheDNA’sfournucleotidecomponents(the‘‘letters’’ofthegeneticalphabet,representedbyA,C,G,T)inasimilarfashiontoencodingofsemanticinformationinthesequenceoflettersofawrittentext.TheDNAiscompactlypackagedinthechromo-somesinsidethenucleusofeachcell.Humanshavetwosetsoftwenty-threechromosomes,havingreceivedonesetfromeachCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n240franciscoj.ayalaparent.ThetotalnumberofDNAlettersineachsetofchromosomesisaboutthreethousandmillion.TheHumanGenomeProject,whichwasundertakenin1989,hasdecipheredthesequence(exceptforanumberofsmallsegments)ofthethreethousandmillionlettersinthehumangenome(thatis,inonesetofchromosomes;thehumangenomesequencevariesamongindividuals).IestimatethattheKingJamesBiblecontainsfewerthanthreemillionletters,punctuationmarks,andspaces.WritingdowntheDNAsequenceofonehumangenomedemandsonethousandvolumesofthesizeoftheBible.Thehumangenomesequenceis,ofcourse,notprintedinbooks,butstoredinelectronicform,incom-puterswherefragmentsofinformationcanberetrievedbyinvesti-gators.Butifaprintoutiswanted,onethousandvolumeswillbeneededjustforonehumangenome.Thetwogenomes(chromosomesets)ofeachindividualaredif-ferentfromoneanother,andfromthegenomesofanyotherhumanbeing(withthetrivialexceptionofidenticaltwins,whosharethesametwosets,sinceidenticaltwinsdevelopfromonesingleferti-lizedhumanegg).Therefore,printingthecompletegenomeinfor-mationforjustoneindividualwoulddemandtwothousandvolumes,onethousandforeachofthetwochromosomesets.Surely,again,therearemoreeconomicwaysofpresentingtheinformationinthesecondsetthanlistingthecompletelettersequence;forexample,byindicatingthepositionofeachvariantletterinthesecondsetrelativetothefirstset.Thenumberofvariantlettersbetweenoneindividual’stwosetsisabouttenmillion,aboutoneinthreehundred.TheHumanGenomeProjectoftheUnitedStateswasinitiatedin1989,fundedthroughtwoagencies,theNationalInstitutesofHealth(NIH)andtheDepartmentofEnergy(DOE).(Aprivateenterprise,CeleraGenomics,startedintheUnitedStatessomewhatlaterbutjoinedthegovernment-sponsoredprojectinachieving,largelyindependently,similarresults.)Thegoalsetwastoobtainthecompletesequenceofonehumangenomeinfifteenyearsatanapproximatecostofthreethousandmilliondollars,coincidentallyaboutonedollarperDNAletter.Adraftofthegenomesequencewascompletedaheadofschedulein2001.In2003theHumanGenomeProjectwasfinished.Thesequencehasbecomeknownwithasmuchprecisionaswanted.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution241Proponentsoftheprojecthadusedinflatedrhetorictoextolitsanticipatedachievements.Theprojectwascalledthe‘‘HolyGrail’’ofbiology,whichwouldmeetthebiblical‘‘Knowthyself’’injunc-tion.TheNobelistWalterGilbertsaidaboutacomputerdisk1encodinganindividual’sDNAsequenceinformation,‘‘thisisyou’’.(TheNobelistandfirstdirectoroftheproject,JamesWatson,2assertedthat‘‘ourfateisinourgenes’’.)DanielKoshland,editoratthetimeofScience,proclaimedthatwithknowledgeofthegenomesequence,‘‘wemaybeabletopreventthedamage’’causedbyviolent3behavior.HastheHumanGenomeProjectaccomplishedanyoftheseloftyobjectives?Hasknowledgeofthehumangenomesequenceaccomplishedtheanticipatedpromiseofcuringhumandiseases?threefrontiersofhumanbiology:beyondthehumangenomeHumanbiologyfacesthreegreatresearchfrontiers:ontogeneticdecoding,thebrain-mindpuzzle,andtheape-to-humantransfor-mation.Thistransformationinvolvedtheemergenceofculturalheredityandculturalevolution,anewandmuchmoreeffectivemodeofadaptationtotheenvironmentthanthebiologicalmode.Theconundrumishowthiswasaccomplishedthroughthechangeoflessthan2percentofthegenome.Onecanrefertothesethreeissuesastheegg-to-adulttransfor-mation,thebrain-to-mindtransformation,andtheape-to-humantransformation.KnowingtheDNAsequenceofhumanbeingsisofgreatuseasadatabasetobiologistsandhealthscientists.ButsuchknowledgeaboutthehumangenomedoesnotbyitselfcontributemuchtothesolutionofanyofthethreeconundrumsIhaveidentifiedhere,orto4thesolutionofanyotherfundamentalbiologicalproblem.ontogeneticdecodingTheinstructionsthatguidetheontogeneticprocess,ortheegg-to-adulttransformation,arecarriedinthehereditarymaterial.ThetheoryofbiologicalhereditywasformulatedbytheAugustinianmonkGregorMendelin1866,butitbecamegenerallyknownbyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n242franciscoj.ayalabiologistsonlyin1900:geneticinformationiscontainedindiscretefactors,orgenes,thatexistinpairs,onereceivedfromeachparent.Thenextsteptowardunderstandingthenatureofgeneswascom-pletedduringthefirstquarterofthetwentiethcentury.Itwasestablishedthatgenesarepartsofthechromosomes,filamentousbodiespresentinthenucleusofthecell,andthattheyarelinearlyarrangedalongthechromosomes.Ittookanotherquarter-centurytodeterminethechemicalcompositionofgenes–deoxyribonucleicacid(DNA).DNAconsistsoffourkindsofchemicalcomponents(nucleotides)organizedinlong,double-helicalstructures.Aspointedoutearlier,thegeneticinformationiscontainedinthelinearsequenceofthenucleotides,verymuchinthesamewayasthesemanticinformationofanEnglishsentenceisconveyedbytheparticularsequenceofthetwenty-sixlettersofthealphabet.Thefirstimportantsteptowardunderstandinghowthegeneticinformationisdecodedoccurredin1941whenGeorgeW.BeadleandEdwardL.Tatumdemonstratedthatgenesdeterminethesynthesisofenzymes;enzymesarethecatalyststhatcontrolallchemicalreactionsinlivingbeings.Itbecameknownlaterthataseriesofthreeconsecutivenucleotidesinagenecodesforoneaminoacid(aminoacidsarethecomponentsthatmakeupenzymesandotherproteins).Thisrelationshipaccountsforthepreciselinearcorre-spondencebetweenaparticularsequenceofcodingnucleotidesandthesequenceoftheaminoacidsthatmakeuptheencodedenzyme.Butchemicalreactionsinorganismsmustoccurinanorderlymanner;organismsmusthavewaysofswitchinganygeneonandoff.Thefirstcontrolsystemwasdiscoveredin1961byFranc¸oisJacobandJacquesMonodforagenethatdeterminesthesynthesisofanenzymethatdigestssugarinthebacteriumEscherichiacoli.ThegeneisturnedonandoffbyasystemofseveralswitchesconsistingofshortDNAsequencesadjacenttothecodingpartofthegene.(Thecodingsequenceofageneisthepartthatdeterminesthesequenceofaminoacidsintheencodedenzyme.)Theswitchesactingonagivengeneareactivatedordeactivatedbyfeedbackloopsthatinvolvemoleculessynthesizedbyothergenes.Avarietyofgenecontrolmechanismsweresoondiscovered,inbacteriaandothermicro-organisms.Twoelementsaretypicallypresent:feedbackloopsandshortDNAsequencesactingasswitches.ThefeedbackloopsensurethatthepresenceofasubstanceinthecellinducesthesynthesisofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution243theenzymerequiredtodigestit,andthatanexcessoftheenzymeinthecellrepressesitsownsynthesis.(Forexample,thegeneencodingasugar-digestingenzymeinE.coliisturnedonoroffbythepresenceorabsenceofthesugartobedigested.)Theinvestigationofgenecontrolmechanismsinmammals(andothercomplexorganisms)becamepossibleinthemid-1970swiththedevelopmentofrecombinantDNAtechniques.Thistechnologymadeitfeasibletoisolatesinglegenes(andotherDNAsequences)andtomultiplythem,or‘‘clone’’them,inordertoobtainthequantitiesnecessaryforascertainingtheirnucleotidesequence.Oneunanticipateddiscoverywasthatmostgenesoccurinpieces:thecodingsequenceofageneisdividedintoseveralfragmentsseparatedonefromthenextbynoncodingDNAsegments.Inadditiontothealternatingsuccessionofcodingandnoncodingsegments,mam-maliangenescontainshortcontrolsequences,likethoseinbacteriabuttypicallymorenumerousandcomplex,thatactascontrolswitchesandsignalwherethecodingsequencebegins.Muchremainstobediscoveredaboutthecontrolmechanismsofmammaliangenes.Thedauntingspeedatwhichmolecularbiologyisadvancingmakesitreasonabletoanticipatethatthemainpro-totypesofmammaliangenecontrolsystemswillbeunraveledwithinadecadeortwo.Butunderstandingthecontrolmechanismsofindividualgenesisbutthefirstmajorsteptowardsolvingthemysteryofontogeneticdecoding.Thesecondmajorstepwillbesolvingthepuzzleofdifferentiation.Ahumanbeingconsistsof1trillioncellsofsometwohundreddifferentkinds,allderivedbysequentialdivisionfromthefertilizedegg,asinglecell0.1millimeterindiameter.Thefirstfewcelldivisionsyieldasphericalmassofamorphouscells.Successivedivisionsareaccompaniedbytheappearanceoffoldsandridgesinthemassofcellsand,lateron,ofthevarietyoftissues,organs,andlimbscharacteristicofahumanindividual.Thefullcomplementofgenesduplicateswitheachcelldivision,sothattwocompletege-nomesarepresentineverycell.Moreover,experimentswithotheranimals(andsomewithhumans)indicatethatallthegenesinany5cellhavethepotentialofbecomingactivated.Yetdifferentsetsofgenesareactiveindifferentcells.Thismustbesoinorderforcellstodifferentiate:anervecell,amusclecell,andaskincellarevastlydifferentinsize,configuration,andfunction.ThedifferentialCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n244franciscoj.ayalaactivityofgenesmustcontinueafterdifferentiation,becausedifferentcellsfulfilldifferentfunctions,whicharecontrolledbydifferentgenes.Theinformationthatcontrolscellandorgandifferentiationis,ofcourse,ultimatelycontainedintheDNAsequence,butprobablyonlyinveryshortsegmentsofit.Whatsortofsequencesarethesecontrollingelements,wherearetheylocated,andhowaretheydecoded?Inmammals,insects,andothercomplexorganisms,therearecontrolcircuitsthatoperateathigherlevelsthanthecontrolmechanismsthatactivateanddeactivateindividualgenes.Thesehigher-levelcircuits(suchastheso-calledhomeoboxgenes)actonsetsratherthanindividualgenes.Thedetailsofhowthesesetsarecontrolled,howmanycontrolsystemsthereare,andhowtheyinteract,aswellasmanyotherrelatedquestions,arewhatneedtoberesolvedtoelucidatetheegg-to-adulttransformation.TheDNAsequenceofsomecontrollingelementshasbeenascertained,butthisisaminoreffortthatisonlyhelpedalittlebyplowingthewaythroughtheentirethreethousandmillionnucleotidepairsthatcon-stitutethehumangenome.Experimentswithstemcellsarelikelytoprovideimportantknowledgeasscientistsascertainhowtheybecomebraincellsinonecase,musclecellsinanother,andsoon.Thebenefitsthattheelucidationofontogeneticdecodingwillgivetohumankindareenormous.Thisknowledgewillmakepos-sibletheunderstandingofthemodesofactionofcomplexgeneticdiseases,includingcancer,andthereforetheircure.Itwillalsoconferanunderstandingoftheprocessofaging,theunforgivingdiseasethatkillsallthosewhohavewonthebattleagainstotherinfirmities.Cancerisananomalyofontogeneticdecoding:cellsproliferatealthoughthewelfareoftheorganismdemandsotherwise.Individualgenes(oncogenes)havebeenidentifiedthatareinvolvedinthecausationofparticularformsofcancer.Butwhetherornotacellwillturnoutcancerousdependsontheinteractionoftheoncogeneswithothergenesandwiththeinternalandexternalenvironmentofthecell.Agingisalsoafailureoftheprocessofontogeneticdecoding:cellsfailtocarryoutthefunctionsimprintedintheirgeneticcodescriptorarenolongerabletoproliferateandreplacedeadcells.In1985,healthcareexpendituresintheUnitedStatestotaled$425billion;in2004theysurpassed$1trillion.MostoftheseCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution245expendituresgoforsupportivetherapyandtechnologicalfixesthatseektocompensateforthedebilitatingeffectsofdiseasesthatwedonotknowhowtopreventortrulycure.Bycontrast,thosediseaseswhosecausationisunderstood–tuberculosis,syphilis,smallpox,andviralchildhooddiseases,forexample–cannowbe6treatedwithrelativelylittlecostandthebestofresults.Amere3percentofthenation’stotalhealthcareexpendituresisdevotedtobasicresearch.Doublingortriplingthispercentagewouldresultinonlyamodestriseintotalexpenditures,butwouldyieldlargesavingsinthenearfuture,ascancer,degenerativediseases,andotherdebilitatinginfirmitiesbecomepreventableorcurable,andthusnolongerrequiretheexpensiveandultimatelyineffectualtherapynowinpractice.thebrain-mindpuzzleThebrainisthemostcomplexandmostdistinctivehumanorgan.Itconsistsof30billionnervecells,orneurons,eachconnectedtomanyothersthroughtwokindsofcellextensions,knownastheaxonandthedendrites.Fromtheevolutionarypointofview,theanimalbrainisapowerfulbiologicaladaptation;itallowstheorganismtoobtainandprocessinformationaboutenvironmentalconditionsandthentoadapttothem.Thisabilityhasbeencarriedtothelimitinhumans,inwhichtheextravaganthypertrophyofthebrainmakespossibleabstractthinking,language,andtechnology.Bythesemeans,humankindhasusheredinanewmodeofadaptationfarmorepowerfulthanthebiologicalmode:adaptationbyculture.Themostrudimentaryabilitytogatherandprocessinformationabouttheenvironmentisfoundincertainsingle-celledmicro-organisms.TheprotozoanParameciumswimsapparentlyatrandom,ingestingthebacteriaitencounters,butwhenitmeetsunsuitableacidityorsalinity,itchecksitsadvanceandstartsinanewdirection.Thesingle-celledalgaEuglenanotonlyavoidsunsuitableenviron-mentsbutseekssuitableonesbyorientingitselfaccordingtothedirectionoflight,whichitperceivesthroughalight-sensitivespotinthecell.Plantshavenotprogressedmuchfurther.Exceptforthosewithtendrilsthattwistaroundanysolidobjectandthefewcarniv-orousplantsthatreacttotouch,theymostlyreactonlytogradientsoflight,gravity,andmoisture.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n246franciscoj.ayalaInanimalstheabilitytosecureandprocessenvironmentalinformationismediatedbythenervoussystem.Thesimplestner-voussystemsarefoundincoralsandjellyfishes;theylackcoordi-nationbetweendifferentpartsoftheirbodies,soanyonepartisabletoreactonlywhenitisdirectlystimulated.Seaurchinsandstarfishpossessanerveringandradialnervecordsthatcoordinatestimulifromdifferentparts;hence,theyrespondwithdirectandunifiedactionsofthewholebody.Theyhavenobrain,however,andseemunabletolearnfromexperience.Planarianflatwormshaveaboutthemostrudimentarybrainknown;theircentralnervoussystemandbrainprocessandcoordinateinformationgatheredbythesensorycells.Theseanimalsarecapableofsimplelearningandhenceofvariableresponsestorepeatedlyencounteredstimuli.Insectsandtheirrelativeshavemuchmoreadvancedbrains;theyobtainprecisechemical,acoustic,visual,andtactilesignalsfromtheenvironmentandprocessthem,makingpossiblecomplexbehaviors,particularlyintheirsearchforfoodandtheirselectionofmates.Vertebrates–animalswithbackbones–areabletoobtainandprocessmuchmorecomplicatedsignalsandtorespondtotheenvironmentmorevariablythaninsectsoranyotherinvertebrates.Thevertebratebraincontainsanenormousnumberofassociativeneuronsarrangedincomplexpatterns.Invertebratestheabilitytoreacttoenvironmentalinformationiscorrelatedwithanincreaseintherelativesizeofthecerebralhemispheresandoftheneopallium,anorganinvolvedinassociatingandcoordinatingsignalsfromallreceptorsandbraincenters.Inmammals,theneopalliumhasexpandedandbecomethecerebralcortex.Humanshaveaverylargebrainrelativetotheirbodysize,andacerebralcortexthatisdis-proportionatelylargeandcomplexevenfortheirbrainsize.Abstractthinking,symboliclanguage,complexsocialorganization,values,andethicsaremanifestationsofthewondrouscapacityofthehumanbraintogatherinformationabouttheexternalworldandtointegratethatinformationandreactflexiblytowhatisperceived.Withtheadvanceddevelopmentofthehumanbrain,biologicalevolutionhastranscendeditself,openingupanewmodeofevolu-tion:adaptationbytechnologicalmanipulationoftheenvironment.Organismsadapttotheenvironmentbymeansofnaturalselection,bychangingtheirgeneticconstitutionoverthegenerationstosuitthedemandsoftheenvironment.Humans,andhumansalone,haveCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution247developedthecapacitytoadapttohostileenvironmentsbymod-ifyingtheenvironmentsaccordingtotheneedsoftheirgenes.ThediscoveryoffireandthefabricationofclothingandshelterhaveallowedhumanstospreadfromthewarmtropicalandsubtropicalregionsoftheOldWorld,towhichwearebiologicallyadapted,toalmostthewholeEarth;itwasnotnecessaryforthewanderinghumansthattheywaituntilgeneswouldevolveprovidinganato-micalprotectionbymeansoffurorhair.Norarehumansbidingtheirtimeinexpectationofwingsorgills;wehaveconqueredtheairandseaswithartfullydesignedcontrivances,airplanesandships.Itisthehumanbrain(thehumanmind)thathasmadehumankindthemostsuccessfullivingspecies,bymostmeaningfulstandards.TherearenotenoughbitsofinformationinthecompleteDNAsequenceofahumangenometospecifythetrillionsofconnectionsamongthe30billionneuronsofthehumanbrain.Accordingly,thegeneticinstructionsmustbeorganizedincontrolcircuitsoperatingatdifferenthierarchicallevels,asdescribedearlier,sothataninstructionatoneleveliscarriedthroughmanychannelsatalowerlevelinthehierarchyofcontrolcircuits.Thedevelopmentofthehumanbrainisindeedoneparticularlyintriguingcomponentoftheegg-to-adulttransformation.Butwemustfocusnowontheissueathand,namely,howthisawesomeorgan,thehumanbrain,works.Withinthelasttwodecades,neurobiologyhasdevelopedintooneofthemostexcitingbiologicaldisciplines.Anincreasedcommit-mentoffinancialandhumanresourceshasyieldedanunprece-dentedrateofdiscovery.Muchhasbeenlearnedabouthowlight,sound,temperature,resistance,andchemicalimpressionsreceivedinoursenseorganstriggerthereleaseofchemicaltransmittersandelectricpotentialdifferencesthatcarrythesignalsthroughthenervestothebrainandelsewhereinthebody.Muchhasalsobeenlearnedabouthowneuralchannelsforinformationtransmissionbecomereinforcedbyuseormaybereplacedafterdamage,aboutwhichneuronsorgroupsofneuronsarecommittedtoprocessinginformationderivedfromaparticularorganorenvironmentalloca-tion,andaboutmanyothermatters.But,forallthisprogress,neu-robiologyremainsaninfantdiscipline,atastageoftheoreticaldevelopmentcomparableperhapstothatofgeneticsatthebeginningofthetwentiethcentury.Thosethingsthatcountmostremainshroudedinmystery:howphysicalphenomenabecomementalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n248franciscoj.ayalaexperiences(thefeelingsandsensations,called‘‘qualia’’byphilos-ophers,thatcontributetheelementsofconsciousness),andhowoutofthediversityoftheseexperiencesemergesthemind,arealitywithunitaryproperties,suchasfreewillandtheawarenessofself,thatpersistthroughanindividual’slife.Idonotbelievethatthesemysteriesareunfathomable;rather,theyarepuzzlesthatthehumanmindcansolvewiththemethodsofscienceandilluminatewithphilosophicalanalysisandreflection.AndIwillplacemybetsthat,overthenexthalf-centuryorso,manyofthesepuzzleswillbesolved.Weshallthenbewellonourwaytowardansweringtheinjunction‘‘Knowthyself.’’theape-to-humantransformationKnowingthehumanDNAsequenceisafirststep,butnomorethanonestep,towardsunderstandingthegeneticmakeupofahumanbeing.ThinkoftheonethousandBible-sizedvolumes.Wenowknowtheorderlysequenceofthethreethousandmillionletters,butthissequencedoesnotprovideanunderstandingofhumanbeingsanymorethanwewouldunderstandthecontentsofonethousandBible-sizedvolumeswritteninanextraterrestriallanguage,ofwhichweonlyknowthealphabet,justbecausewewouldhavedecipheredtheirlettersequence.Humanbeingsarenotgenemachines.Theexpressionofgenesinmammalstakesplaceininteractionwiththeenvironment,inpatternsthatarecomplexandallbutimpossibletopredictinthedetails–anditisinthedetailsthattheselfresides.Inhumans,the‘‘environment’’takesanewdimension,whichbecomesthedomi-nantone.Humansmanipulatethenaturalenvironmentsothatitfitstheneedsoftheirbiologicalmakeup,forexample,usingclothingandhousingtoliveincoldclimates.Moreover,theproductsofhumantechnology,art,science,politicalinstitutions,andthelike,becomeadominantfeatureofthehumanenvironment.AsIhavementionedearlier,adistinctivecharacteristicofhumanevolutionisadaptationbymeansof‘‘culture,’’whichmaybeunderstoodasthesetofnon–strictlybiologicalhumanactivitiesandcreations.Twoconspicuousfeaturesofhumananatomyareerectpostureandlargebrain.Wearetheonlyvertebratespecieswithabipedalgaitanderectposture.Birdsarebipedal,buttheirbackbonestandsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution249horizontalratherthanvertical(penguinsareaminorexception);kangaroosaremostlybipedal,butwithoutpropererectpostureorbipedalgait.Brainsizeisgenerallyproportionaltobodysize;relativetobodymass,humanshavethelargest(andmostcomplex)brain.Thechimpanzee’sbrainweighslessthanapound;agorilla’sslightlymore.Thehumanmaleadultbrainhasavolumeof1,400cubiccentimeters(cc),aboutthreepoundsinweight.Inearlierdecades,evolutionistsraisedthequestionwhetherbipedalgaitorlargebrainoccuredfirst,orwhethertheyevolvedconsonantly.Theissueisnowresolved.Ourhominidancestorshad,sinceatleastfourmillionyearsago,abipedalgait,buttheirbrainwasstillsmall,nomorethan450cc,apoundinweight,untilabouttwomillionyearsago.BrainsizestartedtoincreasenotablywithourHomohabilisancestors,whohadabrainabout650ccandalsobecametool-makers(hencethenamehabilis),andwholivedforafewhundredthousandyears,startingabouttwoandahalfmillionyearsago.TheirimmediatedescendantswereHomoerectus,withadultbrainsreachingupto1,200ccinsize.(IusethenameHomoerectus,asitisoftenused,inabroadsensethatencompassesafairlydiversegroupofancestorsandtheirrelatives,whichcurrentpaleoanthropologistsclassifyinseveralspecies,includingHomoergaster,Homoantecessor,andHomoheidelbergensis.)Ourspecies,Homosapiens,hasabrainof1,300–1,400cc,aboutthreetimesaslargeasthatoftheearlyhominids.Ourbrainisnotonlymuchlargerthanthatofchimpanzeesorgorillas,butalsomuchmorecomplex.Thecerebralcortex,wherethehighercognitivefunctionsarepro-cessed,isinhumansdisproportionallymuchgreaterthantherestofthebrainwhencomparedtothatofapes.biologicalevolutionversusculturalevolutionCulture,asIdefineithere,hasanindividualandasocialcomponent.Itincludesideas,habits,dispositions,preferences,values,andbeliefsofeachindividual.Italsoincludesthepublicresultsofhumanintellectualactivity;technology;humanisticandscientificknowledge;literature,music,andart;codesoflawandsocialandpoliticalinstitutions;ethicalcodesandreligioussystems.TheindividualandsocialcomponentsofculturecorrespondtotheWorldCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n250franciscoj.ayala2andWorld3oftheeminentphilosopherKarlPopper.Thediffer-encebetweenthetwobecomesapparentwhenweconsiderthattheextinctionofhumankindonEarthwouldeliminateWorld2,whileWorld3couldsurviveinpartoronthewholeandcouldbeassimi-latedbyhumansorhumanoidsfromadifferentplanet.Theadventofculturebroughtwithitculturalevolution,asuperorganicmodeofevolutionsuperimposedontheorganicmode,whichhas,inthelastfewmillennia,becomethedominantmodeofhumanevolution.Thereareinhumankindtwokindsofheredity–thebiologicalandthecultural–whichmayalsobecalledorganicandsuperorganic,orendosomaticandexosomaticsystemsofheredity.Biologicalinheritanceinhumansisverymuchlikethatinanyothersexuallyreproducingorganism;itisbasedonthetransmissionofgeneticinformationencodedinDNAfromonegenerationtothenextbymeansofthesexcells.Culturalinheritance,incontrast,isbasedontransmissionofinformationbyateaching-learningprocess,whichisinprincipleindependentofbiologicalparentage.Cultureistransmittedbyinstructionandlearning,byexampleandimitation,throughbooks,newspapersandradio,televisionandmotionpictures,throughworksofart,andbyanyothermeansofcommunication.Cultureisacquiredbyeverypersonfromparents,relatives,andneighborsandfromthewholehumanenvironment(Dobzhansky1962,Ehrlich2000,Cavalli-SforzaandFeldman1981,BoydandRicherson1985,RichersonandBoyd2005).Culturalinheritancemakespossibleforhumanswhatnootherorganismcanaccomplish–thecumulativetransmissionofexperi-encefromgenerationtogeneration.Animalscanlearnfromexperi-ence,buttheydonottransmittheirexperiences,their‘‘discoveries’’(atleastnottoanylargeextent)tothefollowinggenerations.Animalshaveindividualmemory,buttheydonothavea‘‘socialmemory.’’Humans,ontheotherhand,havedevelopedaculturebecausetheycantransmitcumulativelytheirexperiencesfromgenerationtogeneration.Someculturaltransmissionhasbeenidentifiedinchimpanzeesandorangutanpopulations,butthe‘‘cultures’’devel-opedbytheseapesamounttotrivialrudimentswhencomparedtohumancultures(Whitenetal.1999,Whiten2005).Culturalinheritancemakespossibleculturalevolution,anewmodeofadaptationtotheenvironmentthatisnotavailabletoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution251nonhumanorganisms–adaptationbymeansofculture.Organismsingeneraladapttotheenvironmentbymeansofnaturalselection,bychangingovergenerationstheirgeneticconstitutiontosuitthedemandsoftheenvironment.Buthumans,andhumansalone,canalsoadaptbychangingtheenvironmenttosuittheneedsoftheirgenes.(Someanimalsbuildnestsandmodifytheirenvironmentalsoinotherways,butthemanipulationoftheenvironmentbyanynonhumanspeciesistrivialcomparedtohumankind’s,eveninthecaseoftheapes.)Forthelastfewmillennia,humanshavebeenadaptingtheenvironmentstotheirgenesmoreoftenthantheirgenestotheenvironments.Inordertoextenditsgeographicalhabitat,ortosurviveinachangingenvironment,apopulationoforganismsmustbecomeadapted,throughslowaccumulationofgeneticvariantssortedoutbynaturalselection,tothenewclimaticconditions,differentsourcesoffood,differentcompetitors,andsoon.Thedis-coveryoffireandtheuseofshelterandclothingallowedhumanstospreadfromthewarmtropicalandsubtropicalregionsoftheOldWorldtothewholeEarth,exceptforthefrozenwastesofAntarctica,withouttheanatomicaldevelopmentoffurorhair.Humansdidnotwaitforgeneticmutantspromotingwingdevelopment;theyhaveconqueredtheairinasomewhatmoreefficientandversatilewaybybuildingflyingmachines.Peopletraveltheriversandtheseaswithoutgillsorfins.Theexplorationofouterspacehasstartedwithoutwaitingformutationsprovidinghumanswiththeabilitytobreatheunderlowoxygenpressuresortofunctionintheabsenceofgravity;astronautscarrytheirownoxygenandspeciallyequippedpressuresuits.FromtheirobscurebeginningsinAfrica,humanshavebecomethemostwidespreadandabundantspeciesofmammalonEarth.Itwastheappearanceofcultureasasuperorganicformofadaptationthatmadehumankindthemostsuccessfulanimalspecies.Wheneveraneedarises,humanscandirectlypursuetheappro-priatecultural‘‘mutations,’’thatis,designchangestomeetthechallenge.Thesechangesarethediscoveriesandinventionsthatpervadehumanlife.Theinventionanduseoffire,theconstructionofbridgesandskyscrapers,thetelephoneandtheInternet,areexamplesoftechnologicalculturalmutations;science,art,politicalinstitutions,codesofethicsandreligioussystemsalsoareculturalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n252franciscoj.ayalamutations.Onthecontrary,biologicaladaptationdependsontheaccidentalavailabilityofafavorablemutation,orofacombinationofseveralmutations,atthetimeandplacewheretheneedarises.Culturalheredityandbiologicalhereditydrasticallydifferintheirmodeoftransmission,withimportantconsequencesinthespeedwithwhichafavorableadaptationspreads.Biologicalheredityistransmittedonlyvertically,fromparentstotheiroffspring,whileculturalheredityspreads‘‘horizontally’’aswellasvertically,asnotedearlier.Afavorablegeneticmutationnewlyariseninanindi-vidualcanbetransmittedtoasizablepartofthehumanspeciesonlythroughinnumerablegenerations.However,anewscientificdiscov-eryortechnicalinnovationcanbetransmittedtothewholeofhumankind,potentiallyatleast,inlessthanonegeneration.Wit-nesstheworldwidespreadofcellularphonesortheInternetinlessthanadecadeorofthepersonalcomputerinlessthanaquarter-century.BiologicalheredityisMendelianbecauseonlythegenesreceivedfromone’sownparentsaretransmittedtotheprogeny.(Thepresenceinanindividualofnewlyacquiredgenevariationsbyspontaneousmutationdoesnotmateriallychallengethisstatement.)Butacquiredcharacteristics,thatis,theinventions,technologicaldevelopments,andanykindoflearningorexperienceacquiredthroughoutanindi-vidual’slife,canallbetransmittedtootherhumans,whetherornottheyaredirectdescendantsoftheindividual.CulturalheredityisLamarckianinthissense,because‘‘acquiredcharacteristics,’’andnotonlyinheritedones,canbetransmittedtoothers.ThedraftDNAsequenceofthechimpanzeegenomewaspub-7lishedon1September2005.Inthegenomeregionssharedbyhumansandchimpanzees,thetwospeciesare99percentidentical.Thedifferencesappeartobeverysmallorquitelarge,dependingonhowonechoosestolookatthem:1percentofthetotalseemsverylittle,butitamountstoadifferenceof30millionDNAlettersoutofthethreebillionineachgenome.Oftheenzymesandotherproteinsencodedbythegenes,29percentareidenticalinbothspecies.Outoftheonehundredtoseveralhundredaminoacidsthatmakeupeachprotein,the71percentofnonidenticalproteinsdifferbyonlytwoaminoacids,ontheaverage.Thetwogenomesareabout96percentidenticalifonetakesintoaccountDNAstretchesfoundinonespeciesbutnottheother.Thatis,alargeamountofgeneticmaterial,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nHumanEvolution253about3percentorsome90millionDNAletters,hasbeeninsertedordeletedsincehumansandchimpsinitiatedtheirseparateevolu-tionaryways,7or8millionyearsago.MostofthisDNAdoesnotseemtocontaingenescodingforproteins.Comparisonofthetwogenomesprovidesinsightsintotherateofevolutionofparticulargenesinthetwospecies.Onesignificantfindingisthatgenesactiveinthebrainhavechangedmoreinthehumanlineagethaninthechimplineage.Alsosignificantisthatthefastestevolvinghumangenesarethosecodingfor‘‘transcriptionfactors.’’Theseare‘‘switch’’proteins,whichcontroltheexpressionofothergenes,thatis,whentheyareturnedonandoff.Onthewhole,585geneshavebeenidentifiedasevolvingfasterinhumans,includinggenesinvolvedinresistancetomalariaandtuberculosis.(Itmightbementionedthatmalariaisamuchmoreseverediseaseforhumansthanforchimps.)GeneslocatedintheYchromosome(thechromosomethatdeterminesmaleness;femaleshavetwoXchromosomes;maleshaveoneXandYchromosome,theYbeingmuchsmallerthantheX)havebeenmuchbetterprotectedbynat-uralselectioninthehumanthaninthechimpanzeelineage,whereseveralgeneshaveincorporateddisablingmutationsthatmakethegenesnonfunctional.Thereareseveralregionsofthehumange-nomethatseemtocontainbeneficialgenesthathaverapidlyevolvedwithinthepast250,000years.OneregioncontainstheFOXP2gene,whichhadearlierbeendiscoveredtobeinvolvedintheevolutionofspeech.Extendedcomparisonsofthehumanandchimpgenomesandexperimentalexplorationofthefunctionsassociatedwithsig-nificantgeneswillsurelyadvanceconsiderablyourunderstanding,overthenextdecadeortwo,ofwhatitisthataccountsforthehumanum,whatmakesusdistinctivelyhuman.Surelyalso,fullunderstandingwillonlyresultfromthejointsolutionofthethreeconundrumsthatIhaveidentified.Thedistinctivefeaturesthatmakeushumanbeginearlyindevelopment,wellbeforebirth,asthelinearinformationencodedinthegenomegraduallybecomesexpressedintoafour-dimensionalindividual.Inanimportantsense,themostdistinctivehumanfeaturesarethoseexpressedinthebrain,thosethataccountforthehumanmindandforhumanidentity.Itishumanintelligencethatmakespossiblehumanculture.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n254franciscoj.ayalanotes1.CitedbyD.NelkinandM.S.Lindee,TheDNAMystique.TheGeneasaCulturalIcon,W.H.Freeman,NewYork,1995,p.7.2.QuotedinLeonJaroff,‘‘TheGeneHunt,’’Time,20March1989,pp.62–67.3.D.Roshland,‘‘Elephants,MonstrositiesandtheLaw,’’Science25(4February1992),p.777.4.IamnotchallengingherethattheHumanGenomeProjecthasmanypublichealthapplicationsorthatthedecipheringofthegenomesofotherspeciesisofgreatconsequenceinhealthcare,agriculture,animalhusbandry,andindustry.ThequestionishowmuchitcancontributetosolvethethreefundamentalproblemsfacedbyhumanbiologythatIamexpounding.5.Thesheep‘‘Dolly’’wasconceivedusinggenesextractedfromacellinanadultsheep.6.Thisstatementisoverlyoptimistic,anditmaybeoutrighterroneousifthephrase‘‘understoodcausation’’isnotpreciselyconstrued.MalariaandAIDSaretwodiseaseswhosecausationisunderstoodatanumberoflevels,yetwefailtotreatthem‘‘withrelativelylittlecostandthebestresults.’’Inanycase,onecananticipatethatincreasedknowledgeoftheetiologyofthesediseasesmayleadtosuccessfuldevelopmentofeffectivevaccinesordrugs.7.Nature437(1September2005);seealsoScience309(2September2005).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\ndavidj.buller13VarietiesofEvolutionaryPsychologyintroductionWhatisevolutionarypsychology?Theanswertothisquestioniscomplicatedbythefactthatthetermevolutionarypsychologyiscommonlyusedintwodistinctsenses.Inonesense,evolutionarypsychologyissimplythestudyofhumanbehaviorandpsychologyfromanevolutionaryperspective.Inthissense,evolutionarypsychologyisafieldofinquiry,alooseconfederationofresearchprogramsthatdiffersignificantlyintheoreticalandmethodologicalcommitments.Thesediverseresearchprogramsattempttoexplainawidevarietyofphenomena,rangingfromforagingandbirthspacingintraditionalhunter-gatherersocietiestoencephalization(thepro-gressiveincreaseinbrainsizerelativetobodysizeinthehumanlineage)andtheevolutionofaltruismandlanguage.Whatunitestheseresearchprogramsisnotasharedcommitmenttospecifictheoriesregardingtheevolutionofhumanbehaviorandpsychology,butonlyacommitmenttoarticulatingquestionsabouthumanbehaviorandpsychology,andarticulatinganswerstothosequestions,withconceptualandtheoreticaltoolsdrawnfromevolutionarytheory.Inthisbroadsense,evolutionarypsychologydatesbacktoDarwinsTheDescentofMan(publishedin1871)andTheExpres-sionoftheEmotionsinManandAnimals(publishedin1872).But,despiteDarwinsearlyefforts,therewasrelativelylittleconcertedstudyofhumanbehaviorandpsychologyfromanevolutionaryperspectiveuntilthelatterhalfofthetwentiethcentury,whenseveralresearchprogramsemergedandattractedsignificantnum-bersofresearchers(LalandandBrown2002).Theearliestofthese255CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n256davidj.bullerresearchprogramswashumanethology,exemplifiedbyKonradLorenzs1963bookOnAggression.Thefieldreallytookoff,how-ever,withtheemergenceofhumansociobiologyinthe1970s,andintheensuingdecadeadditionalresearchprogramsknownasevolu-tionaryanthropologyandhumanbehavioralecologyemerged.Theseresearchprogramsdifferinthemethodsbywhichtheyapplyevolutionarytheorytothestudyofhumanbehaviorandpsychology,andtheydifferintheirtheoreticalaccountsofhowevolutionhasaffectedthehumanmind.Nonetheless,inthebroadsenseoftheterm,evolutionarypsychologyencompassesalloftheseresearchprograms.Inanarrowersense,thetermevolutionarypsychologyoftendesignatesjustaspecificresearchprogramwithinthefieldofevo-lutionarypsychology,theforemosttheoreticiansofwhicharetheanthropologistsJohnToobyandDonaldSymonsandthepsycholo-gistsLedaCosmidesandDavidBuss.Thisgroupofresearchersisunitedinthebeliefthatadoptionofanevolutionaryperspectiveonhumanpsychologyimmediatelyentailsanumberofveryspecifictheoreticalandmethodologicaldoctrines,andoftenthetermevo-lutionarypsychologyspecificallyreferstothissetofdoctrines.Soastoclearlydistinguishthefieldofinquiryfromthespecificresearchprogram,Iwillrefertothefieldofinquiryasevolutionarypsychology(inlowercase)andtheresearchprogramasEvolu-tionaryPsychology(capitalized).Sinceitsemergenceinthelate1980s,EvolutionaryPsychologyhasbecomethesinglemostdominantresearchprograminthefieldofevolutionarypsychology,havinggarneredthelionsshareofattentionbothwithinacademiaandthroughoutthepopularmedia.ButthereismoretoevolutionarypsychologythanEvolutionaryPsychology.Inparticular,whileEvolutionaryPsychologyhasoccupiedthelimelight,humanbehavioralecologyhasquietlybecomeavibrantresearchprogramwithimpressivecredentials.Indeed,itisthestrongestrivaltoEvolutionaryPsychologywithinthefieldofevolutionarypsychology.Inthischapter,IstrivetogivesomesenseofthediversityofresearchinevolutionarypsychologybycomparingandcontrastingthetheoreticalandmethodologicalprinciplesofEvolutionaryPsychologyandhumanbehavioralecology.Bothoftheseresearchprograms,however,grewoutofhumansociobiology,sothatiswherewewillbegin.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology257humansociobiologyAlthoughmanyresearchershavecontributedtotheprogramofhumansociobiology,withoutdoubtitsleadingtheoreticianhasbeenEdwardO.Wilson.Inthemid-1970s,Wilsonpublishedseveralworksthatshowcasednumerousapplicationsofevolutionarytheorytotheexplanationofanimalbehaviorandthatarticulatedatheoreticalframeworkwithinwhichtoviewthem.Wilsoncalledthisframe-worksociobiology,whichhedefinedsimplyastheextensionofpopulationbiologyandevolutionarytheorytosocialorganization(1978,x).Wilsonfurtherarguedthattheveryprinciplesthatsuc-cessfullyexplainthesocialorganizationofbeehivesanddominancehierarchiesinspidermonkeyscouldbeextendedtohumansocialbehavioraswell.Thisextensionofsociobiologytohumanbehaviorbecameknownashumansociobiology,andWilsonconceiveditasthestudyofthebiologicalbasisofhumansocialbehavior.ThecoreideaofWilsonssociobiologywasthatbehaviorhasevolvedundernaturalandsexualselectionjustasaspectsoforganicformhave.Evolutionbynaturalorsexualselectionoccurswhenorganismsinapopulationexhibitphenotypicvariation,thatvaria-tionisheritable,andorganismswithoneofthephenotypicvariantsare,onaverage,betteradaptedtotheirenvironmentthanorganismswiththealternativephenotypes.Whentheseconditionsaremet,selectioncausesthebetter-adaptedphenotypetoincreaseinfre-quencyinthepopulation,andthepopulationasawholebecomesbetteradaptedtoitsenvironment.Oververylongstretchesoftime,selectionhasthiseffectonmanydifferentphenotypes,andpopula-tionstherebybecomewelladaptedtotheenvironmentstheyin-habit.ThesimpleideaatthefoundationofWilsonsprogramwasthattheseexplanatoryprinciplesareapplicabletobehavioral,notjustmorphologicalandphysiological,phenotypes.Forexample,femalesofmanyspecieschooseamateonthebasisofthequalityofmalecourtshipdisplays.Ifmalescourtshipdisplaysvaryinquality,andthatvariationisheritable,thensexualselectionwillcausethesuperiordisplaytoincreaseinfrequency,andmaleswillbecomebehaviorallyadaptedtofemalepreference.Inthisway,selectioncanshapethewaythatorganismsbehavejustasitshapestheirbodies.Thissimpleideahastwoimportantcorollaries.First,itentailsthatbehaviorsthathavebeenshapedbyselectionareadaptations.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n258davidj.bullerThus,justasorganismsinapopulationpossessanatomicaladapta-tions,theypossessbehavioraladaptationsaswell.Accordingly,partofWilsonsprogramwasanefforttoprovideadaptationistexplana-tionsofhowcertainformsofbehaviorevolved.Second,sinceselectionhasshapedbehaviortotheenvironmentinwhichitoccurs,andsinceanorganismstotalenvironmentincludesitssocialenvironment,Wilsonssimpleideaentailsthatsomebehav-iorsareadaptationstosociallife.Accordingly,Wilsonsprogramwasprincipallyconcernedwithexplaininghowindividualsinapopulationarebehaviorallyadaptedtosociallifewithoneanotherexplainingbehavioraladaptationsfordominancehierarchies,formanifestinganddealingwithaggression,andformating.Indeed,Wilsontookthecentraltheoreticalproblemofhisprogramtobeexplainingtheevolutionofaltruismexplainingwhysomanyorganismshaveevolvedtoperformactsthatbenefitotherorganismsatacosttothemselves.ToillustratetheseaspectsofWilsonsprogramofhumansocio-biology,considersexdifferencesinhumanmatingbehavior.Bothsexesneedtoreproduceinordertobesuccessfulintheevolutionarylonghaul,butreproductionentailsverydifferentcostsforthetwosexes.Inordertoproduceasinglechild,awomanmustinvestoneofherverylimitednumberofeggs,physiologicalresourcesforanine-monthgestation,andthemetaboliccostsoflactation(oftenlastingtwoorthreeyears).Moreover,duringpregnancyandlactation,awomanisunabletoreproducewithmalesotherthanandpossiblybetterthanthefatherofherchild.Incontrast,inordertoproduceasinglechild,amaleneedonlyinvesttheenergyexpendedincopulationandthecontentsofasingleejaculate.Afterafruitfulcopulation,amancanreproducewithotherwomen,whereasawomaniscommittedtothecostlyactofchildbearing.Thisisaradicalasymmetryintheminimumobligatoryparentalinvestmentrequiredofthesexesinordertoproduceasingleoffspring:Womenareobligatedtoafarhigherinvestmentinoffspringthanaremen.Giventhisasymmetry,selectionshouldhavemadewomenverychoosywhenselectingamate,sincetheyhavetoinvestagreatdealinasingleoffspringand,hence,haveagreatdealtolosebychoosingapoorsire.Incontrast,sincemenincursuchaminimalobligationinordertoproduceanoffspring,andsincetheycan(theoretically)impregnateinnumerablewomenduringthetimeittakesawomanCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology259tobearonemanschild,selectionshouldhavemademenindis-criminatelypromiscuous.AsWilsonsays,selectionshouldhavecreatedmalestobeaggressive,hasty,fickle,andundiscriminating,andfemalestobecoy(1978,125).Inotherwords,inhumans,malepromiscuityandfemalecoynessarebehavioraladaptations.evolutionarypsychologyThestartingpointofEvolutionaryPsychologyisacorrectivetothecoreideaofWilsonssociobiology.EvolutionaryPsychologistsarguethattreatingbehavioralphenotypesasjustlikemorphologicalandphysiologicalphenotypesobscuresafundamentaldifferencebetweenthem,forbehaviorsareevents,whicharetheoutputofaninformation-processingbrainreactingtoinformationalinputaboutthecurrentconditionsinboththeenvironmentandthebrainitself.Theonlywaythatselectioncanaffectbehavior,then,isbyalteringtheinformation-processingstructureofthebrain(ToobyandCosmides1992).So,whenabehaviorhasevolvedunderselection,thereisanimportantsenseinwhichitisnotthebehavioritselfthathasbeenselectedfor,butratherthepsychologicalmechanism(cognitiveormotivational)thatiscausallyresponsibleforproducingthatbehaviorunderappropriateconditions.Sincebehavioralevo-lutioninvolvesselectionforthepsychologicalmechanismsthatcausebehavior,theadaptationsthatemergeintheprocessofbe-havioralevolutionarethepsychologicalmechanismsthatcausebehavior.Consequently,EvolutionaryPsychologistsconclude,sociobiologywasmistakeninseekingadaptationatthelevelofbehavior;adaptationmustbesoughtatthelevelofthepsychologicalmechanism(ToobyandCosmides1992).ThegoalofEvolutionaryPsychologyisthustodiscoveranddescribetheinformation-processingstructureofourpsychologicaladaptations(Buss1995).Fromthisstartingpoint,EvolutionaryPsychologistsderiveanumberoftheoreticalandmethodologicaldoctrines.First,theyargue,ourpsychologicaladaptationsareundoubtedlycomplex,andtheconstructionofcomplexadaptationstypicallyrequireshundredsofthousandsofyearsofcumulativeselection.OurancestorsspentthePleistocenetheepochspanning1.8millionto10,000yearsagolivinginsmallhunter-gatherergroups,butonlythepast10,000yearslivingasagriculturistsandthepastfewhundredyearslivinginCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n260davidj.bullerindustrialsocieties.Consequently,itishighlyimprobablethathumanshaveevolvedadaptationstopost-Pleistoceneenviron-ments.Rather,EvolutionaryPsychologistsargue,ourpsychologicaladaptationsmusthavebeendesignedduringthePleistocenetosolvetheadaptiveproblemsfacedbyourhunter-gathererancestors(Symons1992).AsCosmidesandToobycolorfullyputit,OurmodernskullshouseaStoneAgemind(1997,85).Adaptiveproblemsarecommonlycharacterizedasproblemswhosesolutionsenhancetheabilitytosurviveorreproduce.AndtheadaptiveproblemsfacedbyourPleistoceneancestorsrangedfromacquiringmatesandformingsocialalliancestoavoidingpredatorsandinedibleflora.Theseproblemsareverydiverseincharacter,andeachrequiresauniquebehavioralsolution;asuccessfulbehavioralsolutiontooneproblemwouldnothavetransferredtoanother.Thus,EvolutionaryPsychologistsargue,eachadaptiveproblemwouldhaveselectedforitsowndedicatedproblem-solvingpsychologicalmechanism(Symons1992).More-over,sinceourPleistoceneancestorsfacedsuchanenormousvarietyofadaptiveproblems,CosmidesandToobyconcludethatthebrainmustbecomposedofalargecollectionofcircuits,withdifferentcircuitsspecializedforsolvingdifferentproblems.Onecanthinkofeachspecializedcircuitasaminicomputerthatisdedicatedtosolvingoneproblem.Suchdedicatedminicomputersaresometimescalledmodules(1997,81).Indeed,CosmidesandToobyestimatethatthehumanmindcontainshundredsorthou-sandsofsuchmodules,andthisviewhasaccordinglybeendubbedthemassivemodularitythesis.AccordingtoEvolutionaryPsychologists,evolvedmoduleshavethefollowingproperties(CosmidesandTooby1997,ToobyandCosmides1992).First,theyaredomainspecific,specializedtodealonlywitharestrictedtaskdomain.Assuch,theirinformation-processingproceduresareactivatedby,andsensitiveto,onlyinfor-mationaboutaparticularaspectoftheworld,inmuchthewaytheearisresponsiveonlytospecificvibratoryfrequencies.Second,theyareequippedwithsubstantialinnateknowledgeabouttheirproprietaryproblemdomainsandwithasetofinnateproceduresspecializedinemployingthatknowledgetosolveproblemsintheirdomains.And,third,theydevelopreliably,andwithoutformalinstructionintheirproblemdomains,ineverynormalmemberofourspecies.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology261Sinceevolvedmodulesarecomplexadaptations,andsinceselectionusuallytendstomakecomplexadaptationsuniversalornearlyuniversalinaspecies,EvolutionaryPsychologistsarguethathumansmustshareacomplex,species-typicalandspecies-specificarchitectureofadaptations(ToobyandCosmides1992,38).Indeed,EvolutionaryPsychologistsbelievethatevolvedpsychologicalmodulesconstituteauniversalanduniformhumannature(ToobyandCosmides1992,79).Accordingly,EvolutionaryPsychologistsinterpretdifferencesbetweenindividualswithinthesameculture,anddifferencesbetweenindividualsindifferentcultures,astheproductofacommon,underlyingevolvedpsychologyoperatingunderdifferentenvironmentalcircumstances(ToobyandCosmides1992,45).However,becauseournetworkofmodulesouruniversalhumannatureevolvedtosolvetheadaptiveproblemsfacedbyourPleistoceneancestors,andbecausetheenvironmentswenowinhabitdifferenormouslyfromthoseinhabitedbyourPleistoceneancestors,EvolutionaryPsychologistsarguethatourevolvedmod-ulesoftenfailtoproduceadaptivebehavioramongmodernhumans.Forexample,fearevolvedasanemotionalalarmthatsignalsathreattosurvival.But,sincehumanfearsevolvedduringthePleistocene,humanstendtofearsnakesbutnotcarsandguns,despitethefactthatmorepeoplearekilledbycarsandgunsthanbysnakes.Inaddition,peopleinmodernindustrializedsocietiescouldmaximizetheirreproductivesuccessbydonatingtheirspermoreggstocryo-banks,butveryfewpeoplepursuethisreproductiveoption.ThereasonisthatthisoptionwasnotavailableinthePleistocene,andwehavemindsdesignedtomaximizereproductivesuccessonlyunderPleistocene-likeconditions,inwhichsuchsuccesswasachievedonlythroughthepursuitofcopulation.Becauseofthismismatchbetweenhumannatureandcontemporaryhumanenvir-onments,Symonsarguesthatthestudyofwhethercontemporaryhumanbehaviorisadaptivewillrarelyshedlightonhumannatureortheselectiveforcesthatshapedthatnature(1992,146).Thus,EvolutionaryPsychologistsclaim,inordertodiscovertheevolveddesignofthemind,wemustreverseengineerthemindfromthevantageofourevolutionarypast.ThemethodbywhichEvolutionaryPsychologistsproposetoreverseengineertheevolvedstructureofthemindisevolutionaryCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n262davidj.bullerfunctionalanalysis(ToobyandCosmides1992,Buss1995).Evolu-tionaryfunctionalanalysisbeginswiththespecificationofanadaptiveproblemthatPleistocenehumanspresumablyfaced.Thatadaptiveproblemisthenanalyzedintoanumberofsubproblemswhosesolutionscollectivelyconstituteasolutiontotheadaptiveproblem.(Forexample,Pleistoceneeramalesfacedtheproblemofintrasexualcompetitionforreproductiveaccesstofemales,andsolvingthisproblempresumablyrequiredsolvingthesubproblemsofacquiringtheresourcesdesiredbyfemales,successfullycourtingfemales,andretainingmates,amongotherthings.)Thenextstepistodeterminewhatformsofbehaviorwouldhaveconstitutedadap-tivesolutionsunderPleistoceneconditionstothesesubproblems.Amoduleisthenpostulated,whichisassumedtohaveevolvedtogeneratesolutionstoallofthesesubproblems.Thefinalstepistodeterminetheinformation-processingproceduresbywhichthemodulegeneratesitsbehavioralsolution(s)fromitsinputs.Evolu-tionaryPsychologiststhenconductstandardpsychologicalexperi-mentsinordertodeterminewhetherpeoplebehaveinwayspredictedbythemodularhypothesisgeneratedintheselasttwosteps.EvolutionaryPsychologistsclaimtohavemademanydiscoveriesregardingtheevolvednatureofthemindbyemployingthismethod.Considerjustoneexamplebywayofillustration.ThroughoutourevolutionasPleistocenehunter-gatherers,meninvestedresources(food,protection,andpaternalcare)intheoffspringoftheirmates.Butbecauseovulationisconcealedandfertilizationoccursinter-nallyinourspecies,aPleistocenehumanmalecouldneverbe100percentcertainwhenhewaslikelytoimpregnatehismateor,ifhismatewaspregnant,whetheritwashewhohadimpregnatedher.Thisposedthefollowingproblemforanancestralmale:Ifhismatewassurreptitiouslyunfaithful,amancouldwastehisresourcesonachildthatwasnothisown.Pleistocenehumanwomen,incon-trast,werealways100percentcertainthatoffspringborntothemweretheirown.Anancestralwomansproblemwasthatinfidelitybyhermatecouldleadtohisfallinginlovewithanotherwoman,abandoningher,andwithdrawingtheresourcesonwhichshedependedtorearherchildrensuccessfully.EvolutionaryPsychologistsarguethatjealousyevolvedasanemotionalalarmtoprotectagainsttheserespectivepotentiallossesduetoamatesinfidelity.However,sincethethreatsposedbyinfidelityweredifferentforthesexes,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology263EvolutionaryPsychologistsargue,malesandfemalesmusthaveevolveddifferentpsychologicalmechanisms:Theinputsthatacti-vatejealousyformenwillfocusheavilyonthesexactperse,whereasforwomentheywillfocusoncuestothelossofthemenscommit-mentandinvestment(Buss1995,14).EvolutionaryPsychologistshaveconductednumerousstudiestotestthisprediction,andtheyclaimthatitisconfirmedbytheevidence(Buss1995,1415).Thus,theyconclude,menandwomenhaveevolveddistinctpsychologicaladaptationsformonitoring,andemotionallyrespondingto,cuesofpotentialinfidelity.humanbehavioralecologyWhereasEvolutionaryPsychologyisanattempttoblendevolu-tionarytheorywithcognitivepsychology,humanbehavioralecologyderivesfromthebranchofbiologyknownasbehavioralecology.Behavioralecologyisthestudyofhowanimalbehaviorisadaptivelyresponsivetoconditionsinanimalsphysicalandsocialenvironments.Thefundamentalpremiseofbehavioralecologyisthatanimalsaremaximizersofonesortoranotherefficientpredatorsorforagers,orelusiveprey.Theusualgroundforbelievingthisisthepresumptionthatnaturalselectionhasmadethemso(Grafen1991,5).Behavioralecologistsviewanimalsasbehavingsoastomaximizetheirsharesofavarietyofcurrenciesthatarecorrelatedwithsurvivalandreproductivesuccess.Theseevolu-tionarilysignificantcurrenciesincludecaloricintake,offspringsurvivability,clutchsize,territory,numberofcopulations,qualityofmate,numberofsperminaninseminate,numberofinseminatesharvestedperfertileperiod,andnumberofmatesperfertileper-iod.Behavioralecologistspresupposethatanimalstendtoadoptbehavioralstrategiesthatenablethemtomaximizethesecurren-ciesintheparticularenvironmentalconditionsinwhichtheyfindthemselves.Andthispresupposition,inturn,entailsthatanimalsarecapableofbehavingadaptivelyacrossaverywiderangeofeco-logicalconditions,flexiblyalteringtheirbehaviorinresponsetocurrentconditionsinordertomaximizetheirchancesofsurvivalandreproductivesuccess.Behavioralecologistsstudyanimalbehaviorwithoptimalitymodelsandevolutionarygametheoreticmodels.SuchmodelsbeginCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n264davidj.bullerwiththespecificationofsomecurrencywhosemaximizationistobestudied.Forexample,amodelmaystudyclutchsize(thatis,numberofoffspringbornandcaredforatthesametime)insomespeciesofbird.Themodelsthenidentifyanumberofalternativestrategiesthatanimalsmaypursuebywayofattemptingtomax-imizethatcurrency.Ifthecurrencyisclutchsize,thealternativestrategieswouldbevariousclutchsizes:onestrategywouldbetohavetwochicks,anotherwouldbetohavethree,andsoon.Themodelsthenidentifythecostsandbenefitsassociatedwitheachoftheavailablestrategies.Inthecaseofclutchsize,thebenefitsofthealternativestrategiesareeasilymeasuredintermsofnumberofoffspringrearedtoreproductiveviability.Accordingly,benefitsappeartoincreasewithincreasingclutchsizes.However,offspringneedtobefedandcaredfor,andthoseactivitiesexactahighcostinparentalenergy;indeed,thegreaterthenumberoffledglings,themorefoodthatneedstobecapturedandreturnedtothenest.Moreover,ifclutchsizebecomestoolarge,parentscannotade-quatelyprovideforallthechicksinthebrood,andfledglingmor-talityincreases.Sotherearealsocostsassociatedwitheachstrategy(eachclutchsize).Behavioralecologistscalculatethecostsandbenefitsofeachstrategyinordertodeterminewhichoftheavailablestrategiesmaximizestheaverageratioofbenefitstocoststhatis,inordertodeterminewhichistheoptimalstrategy.Forexample,behavioralecologistsmaypredictthat,foraparticularspeciesofbird,fivefledglingsistheoptimalclutchsize.Theythenpredictthatthestudiedanimalswillpursuethatoptimalstrategy,andtheytesttheirpredictionagainsttheactualbehaviorinapopulationofthestudiedspecies.Therearetwopointstonoteaboutmodelinginbehavioralecol-ogy.First,theparticularcostsandbenefitsassociatedwithapartic-ularbehavioralstrategydependheavilyonthespecificfeaturesoftheenvironmentinwhichthatstrategyispursued.Inanenviron-mentinwhichfoodisscarceanddifficulttoobtaintheoptimalclutchsizewillbesmallerthaninanenvironmentinwhichfoodisabundantandeasilyobtainable.Thus,predictionsregardingtheoptimalstrategyinapopulationarealwaysrelativetotheparticularenvironmentinhabitedbythepopulation.Second,althoughsome-timesananimalsoptimalstrategyisindependentofthestrategiesofotherpopulationmembers,atothertimesitisnot.Forexample,forCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology265manyanimalsthereisanoptimalamountoftimespentforagingforfood,whichmaximizestheenergyintakeperunitofforagingtime,andthisoptimumisindependentoftheamountoftimeotherpopulationmembersspendforaging.However,whenpopulationmembersdirectlycompetewithoneanotherforresources(includingmembersoftheoppositesex,whoarereproductiveresources),theoptimalstrategyforanyparticularpopulationmemberwilldependonthestrategiesofotherpopulationmembers.Ifmostmalescom-petingforterritoryonlyengageinthreateningdisplaysandretreatwhenattacked,atacticofextremeaggressionmaybegreatlybene-ficial.But,ifmostmalesareextremelyaggressive,thenaggressioncouldentailthecostsofinjuryordeath.Sothecostsandbenefitsofabehavioralstrategyinacompetitiondependonthestrategiesadoptedbyotherpopulationmembers.Insuchcases,apopulationmaybecharacterizedbyanevolutionarilystableratioofalternativebehavioralstrategies.Whenbehavioralecologistsfindthatanimalsare,infact,pursu-ingthestrategypredictedbyanoptimalityorevolutionarygametheoreticmodel,theyareconfidentthattheirmodelhascorrectlyidentifiedtheselectiveforcesintheenvironmenttowhichanimalbehaviorisresponsiveandthecost-benefitstructuresoftheavail-ablealternativestrategiesinthatenvironment.However,ifanimalbehaviorfailstoconformtothepredictedoptimalstrategy,behav-ioralecologistsassumethatthemodelneedstoberevised.Modelscanberevisedbyalteringthesetofstrategiesassumedtobeavail-abletothepopulationorbychangingthecostsandbenefitsasso-ciatedwiththestrategiesintheset.Butwhenamodeldoesnotaccuratelypredictbehavioralstrategies,behavioralecologiststypi-callyassumethatthemodelhasfailedtoincludesomevariablestowhichanimalsarerespondinginchoosingabehavioralstrategy.Inparticular,behavioralecologiststypicallyassumethatthestudiedanimalsarenotpursuingthepredictedstrategybecauseofatrade-offamongcompetinglifedemands.Theassumptionisthattheneedtomaximizeanothercurrencyplacesconstraintsonthewaysinwhichpopulationmemberscanpursuemaximizationofthecurrencyinthemodel.Infact,forthetypicalanimal,lifeislittlemorethanaseriesoftrade-offs(LalandandBrown2002,11718).Inverygeneralterms,animalsfaceatrade-offbetweensomaticeffort(effortexpendedCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n266davidj.bullertowardbodilygrowthandmaintenance)andreproductiveeffort.Withinthecategoryofreproductiveeffort,thereisatrade-offbetweenmatingeffort(effortexpendedtoincreasethenumberofoffspring)andparentingeffort(effortexpendedtocareforalreadyproducedoffspring).And,withinthecategoryofparentingeffort,parentsoftwoormoreoffspringfaceatrade-offbetweencaringforoneoffspringandcaringforanother.Accordingly,whenanimalbehaviorfailstoconformtothepredictionsofamodel,behavioralecologiststypicallyassumethattheanimalsaretryingtosimulta-neouslymaximizeseveralcurrenciesandthateffortstomaximizeonecurrencyplaceconstraintsoneffortstomaximizeanother.Unlikearobotdesignedtoexcelatsweepingorstamping,behav-ioralecologistsbelieve,naturalselectionisunlikelytodesignorganismstomaximizeoutputsofanyparticulartask;rather,selectionshouldfavororganismsthatoptimizetheseabilities(tradeoffamountsandefficienciesineach),thusmaximizingtheirchancesofsurvivingandreproducing(Smith,BorgerhoffMulder,andHill2001,130).Thus,thepresuppositionunderlyingmodelinginbehav-ioralecologyisthatselectionhasdesignedanimalstoachieveanoptimalallocationofeffortamongcompetinglifedemands.Intheideallimit,then,behavioralecologyaimstoprovideasetofinter-connectedmodelsshowinghowanimalbehaviorstrikestheoptimalcompromiseinpursuingallevolutionarilysignificantcurrencies.Humanbehavioralecologyissimplytheapplicationoftheseideastohumans,andittherebyinvolvesseveraltheoreticalcommitmentsregardinghumanbehavior.First,humanbehavioralecologyassumesthathumandecisionmakingisflexiblyresponsivetocurrentenvironmentalconditions,resultinginthechoiceofbehavioralstrategiesthatwilloptimizetheallocationofeffortamongcom-petinglifedemandsandmaximizelifetimereproductiveoutputrelativetotheconstraintsimposedbytheenvironment(BorgerhoffMulder1991,70).Asaresult,second,humanbehavioralecologyseesbehavioraldifferencesbetweenindividualsasadaptivere-sponsestodifferingenvironmentalconditions.Humanbehavioralecologythusseekstodeterminehowecologicalandsocialfactorsaffectbehaviouralvariabilitywithinandbetweenpopulations(BorgerhoffMulder1991,69).Accordingly,humanbehavioralecol-ogistsofteninterprethumanbehaviorastheresultofconditionalstrategies,behavioralstrategiesoftheformInenvironmentalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology267conditionsA,dox;inconditionsB,doy;inconditionsC,doz(Smithetal.2001,128).Third,humanbehavioralecologistsassumethathumanbehaviorisadaptiveacrossaverywiderangeofenvir-onmentalconditions,includingmanyenvironmentalconditionstowhichourspecieswasneverexposedduringitsevolutionaryhistory.Thus,whereasEvolutionaryPsychologyexpectshumanbehaviortobefrequentlymaladaptiveincontemporaryenvironments(becauseevolutioninourpsychologicaladaptationsislaggingbehindtherapidchangesinpost-Pleistocenehumanenvironments),humanbehavioralecologyexpectshumanbehaviortobewell-adaptedtomostfeaturesofcontemporaryenvironments,andtoexhibitrelativelylittleadaptivelag(Smith2000,30).Inaddition,humanbehavioralecologistsbelievethatadaptivebehavioralresponsescanbeproducedandreproducedbyavarietyofdifferentmechanisms.Thesameadaptivebehaviorcouldbeachievedbyoneindividualthroughtheoutputofaninnatemodule,butbyanotherindividualastheresultofdomain-generallearning.Moreover,thesameadaptivebehaviorcouldbegeneticallytrans-mittedacrossgenerationsthroughgenesformodulesorlearningbiases,throughdirectteachingbyothers,orthroughindirectcul-turaltransferoflearnableinformation.Sinceadaptivebehaviorcanbeachievedthroughavarietyofdifferentmechanisms,humanbehavioralecologistsadoptamethodologicalstrategyknownasthephenotypicgambit:Theyignoredetailsaboutunderlyingmechanisms(whicharetypicallynotknownanyway)inthebeliefthatthesedetailswillnotmatterwithrespecttounderstandinghumanbehavior.Thatis,humanbehavioralecologistsbelievethatafocusonevolutionarilysignificantecologicalconditions,andtheadaptivedemandstheseplaceonhumans,willenablethemtounderstandwhyhumansbehaveastheydoevenintheabsenceofknowledgeofthemechanismsresponsibleforproducingthatbehavior(which,inanycase,mayvaryfromoneindividualtoanother).Thus,humanbehavioralecologistsaregenerallyagnosticaboutmechanisms(includingthequestionofcognitivemodularity)(Smith2000,30).Toillustratetheseprinciplesofhumanbehavioralecology,considerthephenomenonofpolyandry,amaritalsysteminwhichonewomanhasmorethanonehusband.NearlyallsystemsofmarriageinethnographicallyrecordedhumansocietiesareeitherCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n268davidj.bullermonogamousorpolygynous(inwhichonemanhasmorethanonewife).But,of849recordedsocieties,polyandryispracticedinfour,allofwhicharelocatedintheHimalayanhighlands(BorgerhoffMulder1991,82).Atfirstglance,polyandryappearstodefyevolu-tionarylogic,forawomanslifetimereproductiveoutputislimitedbythenumberofpregnanciesshecancarrytoterm,whereasamanslifetimereproductiveoutputislimitedonlybythenumberofwomenhecanimpregnate.Atthetheoreticallimit,awomancanachievehermaximalreproductiveoutputwithasinglemate,whereasamancanachievehismaximalreproductiveoutputonlywithmultiplemates.Thus,polyandryappearstoentailnorepro-ductivebenefitsforwomen,whileinvolvingavastlysuboptimalreproductivearrangementformen.Fromanevolutionarycost-benefitstandpoint,polygynywouldappeartoprovidethegreatestbenefitsformen,whilenonethelessallowingwomentoachievetheirmaximallifetimereproductiveoutput.Sowhywouldmeneveragreetoenterapolyandrousmarriage?Humanbehavioralecologistsstudypolyandrouspopulationswithaneyetounderstandingtheecologicalfactorsthatmaymakepolyandryanadaptivechoice,andtheyhaveidentifiedseveraleco-logicalfactorsthatmayaffectthecost-benefitcalculationsinthedecisionmakingofthosewhoenterpolyandrousmarriages.Humanbehavioralecologistshavediscoveredthatpolyandrousmarriagesaretypicallyfraternalthatis,marriagesinwhichonewomanismarriedtotwoormorebrothers.Thishelps,inpart,tooffsetthecostsofpolyandrytothecohusbands,sincetheirresourcesarepooledtorearonlyoffspringtowhichtheyareallgeneticallyrelated.Moreover,humanbehavioralecologistshavealsodiscoveredthatpolyandrytypicallyoccursamongbrotherswhohaveinheritedfarmlandthatistoosmalltobedividedintoparcelsthatcouldeachsustainafamily.Inaddition,farmingtheinheritedlandishighlylaborintensive,sothatnooneofthebrotherscouldsuccessfullycultivateitinordertosupportafamily.Finally,wherepolyandrousmarriagesoccur,therearenotalternativesourcesofincomeavail-abletothebrothers;cultivatingthefamilyfarmistheonlyavailablemeansofsubsistence.Thus,humanbehavioralecologistshaveconcluded,polyandrypaysbrothersundersuchcircumstances,sincetheydobetterbymaintainingjointpossessionofthefarm,workingittogether,marryingonewoman,andrearingtheirjointoffspringCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology269thantheywoulddobytryingtogotheirownways(BorgerhoffMulder1991,84).Asfurtherconfirmationofthishypothesis,humanbehavioralecologistshavediscoveredthatwhenalternativesourcesofincomesufficienttoraiseafamilybecameavailable,youngerbrotherstypicallyleavetheirpolyandrousmarriagesinordertostartafamilyoftheirown(LalandandBrown2002,12324).Humanbehavioralecologiststhereforebelievethatpolyandryisanadaptivemarriagesystemintheecologicalconditionsofthosewhochooseit.complementaryresearchprogramsorcompetingparadigms?ThereareseveralapparentdifferencesbetweenEvolutionaryPsy-chologyandhumanbehavioralecology(summarizedinTable13.1).First,whereasEvolutionaryPsychologystrivestodiscoverpsycho-logicaladaptationstoPleistoceneenvironments,humanbehavioralecologystudieshumanbehaviorandhowitisadaptivelyresponsivetoecologicalconditions.Second,EvolutionaryPsychologyexpectshumanbehaviortobefrequentlymaladaptiveincontemporaryenvironments,becauseofadaptivelaginpsychologicalevolution,Table13.1.ComparisonofEvolutionaryPsychologyandHumanBehavioralEcologyEvolutionaryHumanbehavioralpsychologyecologyWhatisevolutionaryPsychologicalAdaptivebehavioraltheoryemployedtoadaptationsstrategiesexplain?IscontemporaryhumanNoYesbehaviorgenerallyadaptive?FromwhatvantageOurPleistocenepastThepresentpointareevolutionaryprinciplesapplied?CommittedtomassiveYesNomodularity?CommittedtoauniversalYesNohumannature?CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n270davidj.bullerwhilehumanbehavioralecologyexpectshumanbehaviortobefairlywelladaptedtocontemporaryenvironments.Accordingly,third,EvolutionaryPsychologybelievesthattheevolvednatureofthehumanmindmustbereverseengineeredfromthevantageofourspeciessPleistocenepast,whereashumanbehavioralecologybelievesthatevolutionaryprinciplescanbeappliedinstudyinghumanbehaviorincontemporaryenvironments.Fourth,whereasEvolutionaryPsychologypostulatesthathumanbehavioriscausedbyhundredsorthousandsofmodules,whicharespecial-purposeminicomputersadaptedtospecificadaptiveproblemsfacedbyourPleistoceneancestors,humanbehavioralecologyisagnosticaboutthenatureandnumberofpsychologicalmechanismsthatarecaus-allyresponsibleforadaptivehumanbehavior.Finally,EvolutionaryPsychologystrivestodiscoverauniversalhumannatureunderlyingbehavioraldifferencesbetweenculturesandbetweenindividualsinthesameculture,whereashumanbehavioralecologystudieshowenvironmentaldifferencesbetweenindividualsaffectbehavioraldifferencesbetweenthem.SomehavearguedthatthesedifferencesbetweenEvolutionaryPsychologyandhumanbehavioralecologyaremoreamatterofexplanatoryemphasisthansubstantivescientificdisagreement(Smith2000,3336;LalandandBrown2002,chap.8).Accordingtothisecumenicalview,EvolutionaryPsychologyissimplytheinvestigationofthepsychologicalmechanismsaboutwhichhumanbehavioralecologyremainsagnosticinitsfocusonbehavior.Whereashumanbehavioralecologystudiesthebehavioraloutsideofthehumanorganism,EvolutionaryPsychologystudiesthebehav-iorallygenerativepsychologicalinside.So,whilehumanbehav-ioralecologyaspirestoexplainhowourbehaviorisadaptivelyresponsivetoourecologicalconditions,EvolutionaryPsychologyaspirestoexplainhowourpsychologicaladaptationscausethatbehavior.Similarly,theargumentgoes,humanbehavioralecologyseekstoexplainhowvariationinecologicalconditionsaffectsbehavioralvariationbothwithinandbetweenpopulations,whereasEvolutionaryPsychologyseekstoexplainhowthisbehavioralvar-iationiscausedbyauniversalhumannaturerespondingdifferen-tiallytodifferingenvironmentalconditions.Thus,itispossibletoseeEvolutionaryPsychologyandhumanbehavioralecologyasofferingcomplementary,ratherthancompeting,explanations.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology271Buttheecumenicalpositiongreatlyexaggeratestheextenttowhichthetworesearchprogramsarecompatible.For,whilehumanbehavioralecologyiscompatiblewithsomeevolutionaryaccountofthemechanismsunderlyingthebehavioritstudies,itisnotcompat-iblewithEvolutionaryPsychologysaccountofthosemechanisms.Toseewhy,reconsiderEvolutionaryPsychologysmassivemod-ularityhypothesis.Accordingtothishypothesis,eachadaptiveproblemourlineagefacedinitsPleistocenepastwassolvedbyadedicatedmodule;adaptivebehaviorwasachievedineachproblemdomainbyanexpertsystem,whichwasdesignedtoachieveadaptiveperformanceinitsproblemdomain,butwasineffectiveoutsideitsareaofexpertise.Indeed,EvolutionaryPsychologistsclaim,humansoftenfailtobehaveadaptivelyincontemporaryenvironmentsbecausemodules,withtheirtunnelcognition,areincapableoffunctioningeffectivelywhennotencounteringpreciselytheconditionsforwhichtheyweredesigned.Thiscontrastssharplywithhumanbehavioralecologyspre-suppositionthathumanscanflexiblyaltertheirbehavioralstrategiessoastostrikeoptimaltrade-offsamongnumerousadaptiveprob-lems.Accordingtohumanbehavioralecologists,effectiveadaptivedesignrequiresintegrativemechanismsformeasuringtradeoffs(whichthemselvesvaryincomplexandcontingentways),andadjustingbehavioraccordingtotheweightedeffectofdifferentactivitiesonreproductivesuccess(Smithetal.2001,13031).But,sincetheproblemofstrikingtheoptimaltrade-offacrossadaptiveproblemdomainsisnotaprobleminanyofthosedomains,nodomain-specificmodulecouldweighthecostsandbenefitsofalternativetrade-offstrategiesandadjustbehavioraccordingly.Suchstrategictrade-offscouldbestruckonlybysomedomain-generalpsychologicalmechanism.Moreover,thatdomain-generalmechanismcouldnotsimplybeamechanismthatturnedonandoffthevariousmodulesthatarerelevanttoonescircumstances,leavingthemodulestosolvetheirownproblemsintheirownways.Itwouldhavetobeadomain-generalmechanismthatcouldadjustbehaviorwithineachadaptiveproblemdomaininawaythatallowedfortheoptimalallocationofeffortandefficienciesamongcompetingdemands.Butanypsychologicalmechanismcapableofadjustingbehaviorwithinproblemdomainssoastostrikeoptimaltrade-offswouldnotneedtobesupplementedwithmechanismsthatareCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n272davidj.bullerspecializedforfunctionswithineachproblemdomain.Thus,althoughhumanbehavioralecologyspresuppositionthathumanscanflexiblyalterbehaviorsoastooptimizetrade-offsiscompatiblewithsomeaccountofthepsychologicalmechanismsthatmakesuchadaptivetrade-offspossible,itisnotcompatiblewithEvolutionaryPsychologysmassivemodularityhypothesis.Ifhumanbehavioralecologyisrightabouttheflexibleadaptivenessofhumanbehavior,EvolutionaryPsychologyiswrongaboutthepsychologicalmechan-ismsunderlyingthatbehavior.Thissubstantivedifferencebetweenthetworesearchprogramsisrelatedtoanotherdifference,whichEvolutionaryPsychologistshavetakentobesubstantive,butisonlypartlyso.Becausehumanbehavioralecologyseekstoexplainadaptivehumanbehavior,whileremainingagnosticaboutthepsychologicaladaptationsunderlyingthatbehavior,EvolutionaryPsychologistshaveoftenclaimedthatitisnotagenuinelyevolutionarytheoryofhumanbehavior.Accord-ingtoEvolutionaryPsychologists,thetheoryofevolutionbynaturalselectionisatheoryofadaptation,sonothinginthetheoryofevolutionbynaturalselectionjustifiesanadaptation-agnosticsci-enceofadaptiveness(Symons1992,150).SinceEvolutionaryPsy-chologysgoalistodiscoverhumanpsychologicaladaptations,itclaimstobetheonlytrulyevolutionarytheoryofhumanbehaviorandpsychology.Buthumanbehavioralecologyisnotagnosticastowhetheradaptationsunderlietheadaptivebehavioritstudies.Indeed,humanbehavioralecologistsassumethathumandecisionsareguidedbycomplexprocessesofobservation,evaluation,recalledexperience,experimentationandstrategizingwhich...havethemselvesbeenshapedbypastselectionpressures(BorgerhoffMulder1991,70).Inthisrespect,humanbehavioralecologistsarenomoreagnosticaboutadaptationsthanEvolutionaryPsychologists.Humanbe-havioralecologymerelyrefusestocommititselftohypothesesregardingtheprecisecausalmechanismscomprisingtheadaptationsthatunderlieadaptivehumanbehavior.Therearetworeasonsforthisrestraint,andthesereasonssubstantivelydifferentiatehumanbehavioralecologyfromEvolutionaryPsychology.First,humanbehavioralecologistsbelievethatthebestwaytodiscoverhumanpsychologicaladaptationsistostudythewaysinCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nVarietiesofEvolutionaryPsychology273whichhumansrespondadaptivelytotheirecologicalconditionsratherthantoattempttoreverseengineerthemfromthevantageofourPleistocenepast(Smithetal.2001,13132).Thus,wewillachieveknowledgeoftheadaptationsunderlyinghumanbehavioronlyafterweunderstandtheactualdecisionshumansmakeregardingsurvivalandreproduction.Second,thebeliefthathumanscanflexiblyaltertheirbehavior,soastobehaveadaptivelyeveninevolutionarilynovelenvironments,presupposesthathumanpsy-chologicaladaptationsaremechanismsofadaptiveplasticity(inwhichasinglegenotypeproducesmorethanonephenotypebyrespondingappropriatelytoenvironmentalconditions).Sincemechanismsofadaptiveplasticityarenotyetwellunderstood,humanbehavioralecologistscurrentlytreatthemasblackboxesintheirstudiesofhumanbehavior.Butwhateverthedetailsaboutthecausalworkingsinsidesuchblackboxes,mechanismsofadap-tiveplasticitycontrastsharplywithEvolutionaryPsychologysmodules,whicharefunctionallyspecializedtoproduceparticularformsofbehaviorandwhichdevelopreliablyacrossabroadrangeofenvironmentalconditions.Consequently,whilehumanbehavioralecologydoesnotpresupposehypothesesaboutspecificpsychologicaladaptations,itdiffersfromEvolutionaryPsychologyregardingthekindsofadaptationthatunderliehumanbehavior.Thus,althoughhumanbehavioralecologycanbefruitfullysup-plementedwithexplanationsofthenatureandevolutionofthepsychologicalmechanismsunderlyingadaptivehumanbehavior,itstheoreticalcommitmentsprecludebeingconjoinedwiththeparticularexplanationsthatEvolutionaryPsychologyoffers.Despitesomesuperficialcomplementarities,humanbehavioralecologyandEvolutionaryPsychologyareactuallycompetingparadigmsratherthancomplementaryresearchprograms.Asthetheoreticalandempiricalfortunesofoneprogramwane,thoseoftheotherwillwax.conclusionThewidelypopularizedresearchprogramofEvolutionaryPsychologyisnottheonlygameonthefieldofevolutionarypsychology.Indeed,humanbehavioralecologyisavibrantalternativeparadigmforCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n274davidj.bullerunderstandinghumanbehaviorfromanevolutionaryperspective.And,sincemuchrecentresearchhasdetailednumerousproblemswiththetheoryandmethodologyofEvolutionaryPsychology(see,forexample,Buller2005),humanbehavioralecologyistheparadigmthatholdsthegreatestpromiseforthefutureofevolutionarypsychology.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nvaleriegrayhardcastle14NeurobiologyMostoftheissuesfoundintraditionalphilosophyofsciencearerecapitulatedinthephilosophyofneurobiology.Inparticular,philosophersofneurobiologyworryaboutwhatcountsasappro-priateempiricaljustificationforatheoreticalclaim,howtodeter-minewhichleveloforganizationisthecorrectoneforascientificexplanation,whatexplanationsshouldlooklike,whetherallexplanationswillorshouldreducetosomeprimitives,andhowwhatwelearnaboutthemind/brainshouldaffectlargersocial,economic,andpoliticaldecisions.Inaddition,philosophersofneurobiologyconcernthemselveswithsometraditionalaspectsofphilosophyofmind,includingworryinghowitisabraincanrepresent,ifitdoes,andhowandwhetherthisrepresentationtiestoothernotionsofrepresentationincognitivescienceandbeyond.Itisdifficulttofocusononlyoneoftheseconcernstotheexclusionoftherest.Mostlikely,aswecometounderstandsomeparticularaspectofthepracticeofneurobiology,wewillalsounderstandothersaswell.Inwhatfollows,Idiscusstheseareasofconcernastheydifferfromtraditionalarguments.Thisdiscussionthereforeshouldbelaidontopofandbeseentocomplementtheveryrichliteratureintraditionalphilosophyofscienceandphilosophyofmind.1.theoriesinneurobiologyBrainsarecomplicatedandmessyaffairs;theoriesaboutbrainssharethesesametraits.Thedifficultyisthatinordertomakeasimplegeneralizationabouthowsomeaspectofthebrainfunctions,scientistshavetoretreattosuchabroadlevelofabstractionthattheirassertionsbecomealmostempiricallymeaningless.Inorderto275CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n276valeriegrayhardcastlemaketheirclaimstestableinalaboratory,neurobiologistshavetoconfinetheirideastoparticularanimals,toparticularexperimentaltasks,ortoboth.Asaresult,scientistsendupwithneurobiological‘‘theories’’thatcontaintwodistinctparts:abroadstatementoftheoreticalprincipleandasetofdetaileddescriptionsofhowthatprincipleplaysoutacrossdifferentanimalmodelsandexperimentaltasks.Thoughthedetaileddescriptionsfallunderthegeneralprin-ciple,theyarenotimmediatelyderivablefromit.Moreover,thedetaileddescriptionscanbeincompatiblewithoneanother,thougheachwillmaintainafamilyresemblancewiththeothers.(SeeHardcastle1995,Schaffner1993,Suppe1989forsimilarapproachestounderstandingtheoriesinthebiologicalsciences.)Atagrosslevel,mammalianbrainsareremarkablysimilartooneanother.Indeed,thecentralnervoussystem(CNS)ininvertebratesisnotallthatdifferentfromthemammalianCNSeither.Thereareinnumerablehomologousareas,celltypes,neurotransmitters,peptides,chemicalinteractions,andsoforth.However,oncewescratchthesurfaceofdifferentanimalbrains,wedofindimportantdifferences.Forexample,considerthesemicircularcanal.Allmammalshaveroughlythesamefiveendorgansintheirearstosupporttheirauditoryandvestibularsystems,andtheyallworktokeeptheirlateralsemicircularcanalsintheirearsparalleltothehorizontalplanerelativetotheEarth,forkeepingitinthatpositionallowsthemtogetthebestpossibleinformationaboutheadpositioninspace.(Thelateralcanalismaximallyexcitatorytoayaw(left-to-right)headmotion;keepingthecanalinlinewiththehorizontalplaneallowstheorgantodetectthismotionwiththegreatestaccuracy.)Butrodentsambulatewiththeirnecksextended,whichkeepstheirheadsinanextremedorsalposition,whilehumansinclinetheirheadsabouttwentydegreeswhenwalkingnaturally.Ingeneral,scientistscancorrelatethedifferencesintheshapeofthesemicircularcanalsintheearwithskullshapeandthepositionthatananimal’sheadisnormallyin.(Itisanunansweredbutintriguingquestionwhetherscientistsfindthecanalstructurestheydobecausedifferentheadsevolvedtobeorientedindifferentdirectionsorwhetheranimalsnaturallyholdtheirheadsindifferentpositionsbecausetheirsemicircularcanalsevolveddifferently.)CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNeurobiology277Foranotherexample,considertheretina.Therearestrikingdifferencesbetweenherbivoresandpredatorsinbrainstructure,forcreatureswhomunchongrassesandtreesrequiremuchlesspreciseenvironmentalinformationthanthosewhohuntmovingtargetsinordertosurvive.Asaresult,rodentshavenofoveae.Tomaintainvisualfixationonapoint,theymovetheirnecks,usingwhatisknownasthevestibular-colicresponse.Thevestibularsystemintheirearstellsthemhowtheirheadisorientedandtheyusethatinformationtoreorienttheirheadsinordertokeepwhateverobjectcurrentlyfascinatesthemintheirlineofsight.Incontrast,primateshavefoveaeandtheymovetheireyeballstokeeptheirtargetwithinthefovealarea,usingthevestibular-ocularresponse.Thisisamuchmorepreciseorientingmechanism,whichallowsthemtomovetheireyestocompensateforchangesinheadpositionsuchthattheycankeepobjectsfoveatedforaslongastheywish.Forsomeindicationofhowimportantcomputinghorizontaleyemotionistoourbrains,considerthattheabducens(orsixth)nerveinhumans,whichcontrolshorizontaleyeabduction,feedsintooneofthebiggestmotornucleiinthebrainstem.Thisocularnucleus,whichcontrolsonlyoneverytinymuscle,isonlyslightlysmallerthanthenucleusthatcontrolsallofourtwentyorsofacialmuscles.Inmorestrikingcontraststill,batsdonotmaintainocularposi-tioninthesamefashionastherestofthemammals.Becausetheyflyandsohavegreaterfreedomtomoveinthree-dimensionalspace,maintainingbodypositionrelativetohorizontalisnotaneasyoption.Asaresult,theyuseothersenseorgans,primarilyhearing(theotherhalfoftheeighthnerve),todeterminehowtheireyesshouldbeoriented.Consequently,theyneednotrelyonvestibular-ocularresponsesaswedo,eventhoughtheirbodiesareequippedwithsuchreflexmachinery.Alloftheseanatomicalandphysiologicaldifferencesareimpor-tantwhenneurobiologistswanttoinvestigatesomethinglikethewaythebrainlearnstocompensatefordamagetothevestibularpathways.Whatmayseemassmallandinsignificantdifferencesfromabroadmammalianperspectivebecomehugelyimportantasscientistsseektounderstandtheparticularmechanismsofbrainplasticity.Cantheyuseanimalswithnofoveaeandavestibular-colicresponsetolearnabouthowfoveatedmammalsrecovertheirCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n278valeriegrayhardcastlevestibular-ocularresponse?Moregenerally,theyneedtoknowhowwellparticularanimalmodelstranslateacrosstheanimalkingdom.Shouldtheybeallowedtogeneralizefromexperimentsonasinglespecies(orsetofspecies)tothewaynaturefunctions?Inallvertebrates,aunilaterallabyrinthectomy(UL),oralesionofthelabyrinthinestructureinoneear,givesrisetotwotypesofocu-larmotordisorders.Therearestaticdeficits,suchasabiasto-wardlookingtowardthelesionedsidewhentheheadisnotmoving,anddynamicdeficits,suchasabnormalvestibular-ocularreflexes(VORs),whichoccurinresponsetoheadmovements.InonlytwoorthreedaysaftertheULprocedure,thebrainstartstocompensateforitslossandthestaticdeficitsdisappear.Sincelabyrinthinestruc-turesdonotregenerate,andperipheralneuronscontinuetofireabnormally,whateverthebrainisdoingtorecoverhastobeacentraleffect(ShaeferandMeyer1974).Singleneuronrecordingsfromavarietyofanimalsindicatethatthevestibularnuclei(VN)onthesamesideofthebrainasthelesionstarttoshownormalrestingrateactivityasthebrainlearnstocompensateforitsinjury.Scientistsdobelievethatwhateverthemechanismis,itisalsolikelytobeageneralprocedurethebrainusesforrecovery,fortheyfindsimilarrestingraterecoveriesofthesorttheyseewiththeipsilateralvestibularnucleiafterdenervationinthelateralcuneatenucleus,thetrigeminalnucleus,andthedorsalhorn,amongotherareas.Exactlyhowanargumenttodefendtheseconvictionsissupposedtorun,though,isunclear,sinceitisfairlyeasytofindsignificantdifferencesinthewaysorganismsrecoverandcompensateforves-tibulardamageacrosstheanimalkingdom.Frogs,forexample,appeartorelyoninputfromtheintactlabyrinthtoregulatetherestingactivityofthevestibularnuclei.Mammals,however,donot.Therecoveryoftheirvestibularnucleioccursindependentlyoftranscommissuralinputs(Flohretal.1981).Inaddition,staticsymptomsfollowdifferenttimecoursesindifferentanimals.Inrats,spontaneousnystagmusdisappearswithinhoursafterUL,whileintherabbitandguineapig,itpersistsforseveralweeks(BaarsmaandCollewjin1975,Sirkin,Precht,andCourjon1984).Inhumans,itmaycontinueinoneformoranotherforseveralyears(Fisch1973).Thereisafundamentaltensioninneurobiologybetweenthebigpicturestoryandwhatisfoundinparticularinstances.AllsciencesCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNeurobiology279stripawayfeaturesoftherealworldwhentheydevisetheirgeneralizations.Physicistsneglectfriction;economistsneglectaltruism;chemistsneglectimpurities,andsoon.However,whatneurobiologistsaredoingisnotanalogoustowhatphysicists,economists,andchemistsaredoing.Ineachoftheothercases,thescientistsaresimplifyingthenumberofparameterstheymustconsiderinordertomakeusefulandusablegeneralizations.Incontrast,ifneurobiologistsweretoignorethedifferencestheyfindacrossspecies,thentheywouldhavenodatalefttobuildatheorywith.Thereisnotanythingleftover,asitwere,onceneurobiologistsneglecttheanatomicalandphysiologicaldifferencesfoundinthebrainacrosstheanimalkingdom.Thereismuchleftoverwhenphysicistsneglectfriction;mostofclassicalmechanicsisleft,infact.Indistinctiontotheothersciences,inneurobiologywefindatensionbetweenthegeneralrulesonehopestofindthatdescribeallbrainsandtheparticularcasesneurobiologistshappentostudy.Whatshouldthescopeanddegreeofgeneralizationforneuro-biologicaltheoriesbe?Itappearsscientistsareconfrontedwithanunpleasantchoice.Eithertheysettleforlarge-scaleabstractgen-eralizations,whichglossoverwhatmaybeimportantdifferences,ortheyfocusonthedifferencesthemselves,attheexpenseofwhatmaybeusefulgeneralizations.However,despiteappearances,theydonothaveaneither-orpropositionthattheyhavetoresolvebeforetheycanmoveahead,foraproperneurobiologicaltheorycontainsbothgeneral(andfairlyvague)abstractionsanddetailedcommentsonspecificanatomiesandphysiologies.Theparadigmtheoriesforphysicsaresimpleelegantequationswithuniversalscope.Theoriesinneurobiologyreadmorelikealistofgeneralprinciplesplusdetailedcommentaries.Onefeelsthetugofthedilemmaposedaboveonlyifoneisoperatingwitharestrictednotionofwhatascientifictheoryis.Sometheoriesarepithyandsuccinct;somearenot.Neurobiologicaltheoriesareofthelattersort.Inneurobiology,scientistsstartwithatheoreticaldescriptionatthemostgenerallevel;itiswhatwemightcallthe‘‘theoreticalframework’’–themostgeneralcomponentinaneurobiologicaltheory.Oncetheyadopttheframework,theycanmakemoreprecisehypothesesasawayoffillingouttheirtheoreticalproposal.Theseclaimscanbelocaltoparticularphylaorspecies;hence,theyarenotintendedtobeamoredetailedspecificationofthegeneralCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n280valeriegrayhardcastleframework.Instead,theycanbethoughtofasinstancesorexamplesofhowtheframeworkmightbecashedoutinparticularcases.However,itisnotthecasethatall‘‘fillingsout’’failtogeneralize.Forexample,thedynamicsymptomsofULrecoverbyusingadif-ferentmechanism(probably).Onehypothesisisthatbrainsuseaformofsensorysubstitutiontocompensateforthevestibular-ocularreflex(Berthoz1988,MilesandLisberger1981).Inthiscase,thebrainusesinternallygeneratedsignalsfromthevisualorsomato-sensorysystemstocompensateforthevestibularloss.Itmaysub-stitutecomputationsfromthesaccadicoravisualpursuitsystem,bothofwhich(probably)reconstructheadvelocityinternally,forvestibularthroughputs.Datadrawnfromexperimentsonfrogs,cats,andhumansindicatethattheyallapparentlyusethesamemechanism,thoughitremainstobeseenwhetherthisproposalwillbeapplicabletoallcreaturesandwhetheritcanbegeneralizedmuchbeyondvestibularreflexes.Therearedifferentdegreesofabstractiononemightuseoncesometheoreticalframeworkisadopted.Somediscussionsaregoingtoberestrictedtoasinglespecies,ormaybeevenonedevelopmentalstagewithinaspecies;otherswillincludeseveralunrelatedspeciesorphyla.Botharelegitimatewaysofcashingouttheframeworkinparticularinstances,andneitheristobepreferredtotheother.Thedatawilldictatethescopeofsubhypotheses,andscopecanvarydramatically.Andthisishowtheoriesinneurobiologyarebuiltandstructured.Detailedconclusionsregardingasingleanimalmodelgiverisetogeneraltheoreticalprinciples.Theseprinciplesinspirenewexperi-mentsdonewithotheranimalmodels,whichinturngiveusnew(andprobablyincompatible)detailsbutalsonewgeneralprinciples.Thesenewprinciplesthenconnecttootherdetailedstudiesusingdifferentprotocolsonstillotheranimals,andsoitgoes.Attheendoftheday,wehaveasetofrelatedtheoreticalprinci-plesthatjointlycomposeageneraltheoreticalframework.Andtheseprinciplesareheldtogetherbythedetaileddatafromawidevarietyofanimalstudies.Neurobiologycontinuallymovesbetweentwodifferentwaysofunderstandingthenervoussystem,firstinbroadandsweepingstrokesandsecondbysubmergenceintheminutiae.GeneraltheoreticalprinciplesariseoutofandthenfeedbackintoparticularanimalexperimentsdoneondifferentanimalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNeurobiology281models.Becausephysiologydiffersacrossspecies,specificexperi-mentalprotocolsareappropriateonlyforspecificmodels.Some-timesthedataarisingoutofthedifferentanimalmodelsanddifferentexperimentalproceduresoverlap,butlargelytheydonot.Hence,sometimesthedetailedconclusionsareconsistent,butsometimes–alotofthetime–theyarenot.Neurobiologistsweaveastorythroughtheiranimalmodelsandexperimentalprotocolsunitedbyacommonguidingtheoreticalthread.Theybothfindcom-monaltiesanddefinedifferences.Andthisentireexercise,takentogether,fashionsthetheoreticalstructureofneurobiology.2.theory-ladenobservationsandsingle-cellrecordingsItisalmostatruisminphilosophyofsciencethatthereisnounproblematicdistinctionbetweenobservationandtheory.Thatis,anyscientificobservationswemakearefilteredthroughandbyapriortheoreticalframework.Rawdatabecomeobservationsasweinterpretthewaystheyeitherfitorbelieourhypotheses(Woodward1989).Inshort:whatcountsasanobservationandhowthatobser-vationfunctionsinthebusinessofscienceareheavilymediatedbytheory.Inneurobiologyinparticular,itiseasytochangethefun-damentalnatureofourobservationsusingacceptedmethodologicaltechniquesformanipulatingrawdata.Gooddataallowscientiststodiscriminateamongcompetingclaimsaboutphenomena(Suppe1989).Theparticularpracticesofthescientificsubfieldtellushowtojudgewhetherdataaregood.Sometimesthesepracticesinvolveexplicitcalculationsandformalderivations;sometimestheyinvolvemattersofpersonaljudgmentandskill.Thecasesinneurobiologyinvolveboth.Inparticular,itisamatterofpersonaljudgmentintheworldofsingle-cellrecordingwhentoemploycertaincomputationalprocedures.Differentsortingtechniquesgiverisetodifferentdata,sowhichtechniquestoemployisanimportantquestion.Butthatisalsoaquestionforwhichnoeasyoracceptedanswerexists.Ithasonlybeenduringthelastdecadeorsothatneurobiologistshavebeenabletorecordfromtheextracellularspaceofalargenumberofneuronsfromawakeandbehavinganimals.Whentheyrecordwithanelectrodenearasinglecell,theydopickupthecells’CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n282valeriegrayhardcastleactionpotentials,whicharecommonlybelievedtobethemeansthroughwhichneuronscommunicate.Buttheyalsorecordthingsthatlooklikeactionpotentials,butareinsteadvoltagesgeneratedbyaxonalbundlesorthefieldpotentialsfromparallelsetsofdendrites.Moreover–andespeciallyifthemicroelectrodehasarelativelylowimpedence–extracellularelectrodespickupsignalsfromseveralneuronsatthesametime,recordingfromallthecellsinanearbyarea.Theproblemishowtodifferentiatethecontributionsofthedif-ferentcellsandcellpartsfromasinglelumprecording.Inmanycasesscientistsonlycareaboutoneparticularactionpotential;therest,fromtheirperspective,isbackgroundnoise.Thechallengeishowtoseparatewhattheywantfromalltheelectricalsignalstheydonotwant.Thechallengeishowtomovefromtherecordingsoftheelectrode’soutputtogenuine,reliable,andinformativedata.Thischallengeiscompoundedbythenoisynatureoftherecordingsthemselves.Someofthenoiseismechanicalandarisesfromtheamplifiersthemselves,butsomeisbiologicalandcomesfromtheneurons.Braincellsjitteraroundconstantly(cf.ConnorsandGutnick1990).Neuronsarenotquietuntiltheyfireoffaspike,assomemightthink.Instead,theyarealwaysproducingsomeactivityorother.Allinall,scientistshavetoculltheirdatafromquiteadin.Finally,becausethecomponentsinarecordingarenotconstant,itisdifficulttogetatheoreticalhookintothewaveform.Spikeshapescanchangeovertime;electrodescandriftduringrecordingsession,changingpositionrelativetothecells,whichwouldalsoalterthespikeamplitudes;andtheelectricalpropertiesofelectrodesvarywithchangesintipconditionorbackgroundimpedance.Gatheringdatafromsingleunitactivitypresentsneurobiologistswithaserioustechnicalchallenge.Inordertogetusabledata–togetgenuineobservations–outofwhattheelectrodetransmits,scientistsmustisolateeachneuron’scontributionstotherecordedwaveform.Theyfirstneedtoascertainexactlyhowmanyneuronstherecordedwaveformreflects.Howcantheydothisiftheyhaveamessofoverlappingactionpotentialsandfieldpotentialsfromavarietyofcellsatdifferentandunknowndistancesfromtheelectrode?Thisquestionbecomesparticularlyvexingifotherneuronsinthesameareahavespikesofthesameorasimilarshapeandamplitude.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNeurobiology283Thereareseveraldecompositionalgorithms;however,eachisimperfect(seeLewicki1998).Eachrepresentsadifferentwaytomovefromrawoutputtointerpretedandinterpretabledata,givingscientistsdifferentwaysofrefiningthewaveformstheyhaverecordedsothattheycanlaterinterpretthem.Eachiswhatphilosophersarethinkingaboutwhentheytalkaboutthetheory-ladennessofdata.Scientistshavetochoosewhattodowiththeirmeasurementsinordertogetsomethingthatcanbescientificallyuseful.Andthewaytheychooseisdeterminedbypreviouslyacceptedtheories.Butevenwithalltheseadvancedsortingtechniques,itisstillhardtopredictthenumberofneuronselicitingthedata.Ideally,scientistswouldliketoclaimthatoneneurongenerateseachclusterofspikeswehaveidentified,butifthecellsarefiringincomplexbursts,orifthereisnonstationarynoise,orifthespiketrainsoverlaponeanother,theycannotgetaccurateclassificationsatall.Itissimplyanunsolvedproblemhowtodecomposecoincidentactionpotentialswithvariablespikeshapes.Thebestscientistscandoatthispointisguess.Theirguessesareinformedbytheiryearsofexperience,buttheyareguessesnonetheless.Guessingisnotquitewhatphilosophersofsciencehaveinmindwhentheytalkaboutthetheory-ladennessofobservation.Theirvisionofcreatingdataisoneofmore‘‘scientificmethod.’’Thatis,topulldataoutofthedialmovementsorchangesincolororsquigglesonthepage,philosophersgenerallyholdthatthereissomeexplicitbackgroundtheory,devisedinsomeotherscientificinquiry,thatscientistslearnandthenusetointerpretwhattheyareseeingormeasuringassomethingusefulfortheirstudies.Butthereisathe-oreticalgap,asitwere,inthemovefromrawrecordingstogenuinedata,agapscientistscannotfillwithanysortofdecision-makingalgorithm.Thebestscientistscandoatthispointissimplyleapacrossthegap,onblindfaith,withaneyetowheretheywanttogo.Neurophysiologytravelsinacognitivecircle;scientistsusewhattheyknowtoculldatathatsupportwhattheybelievetobethecase.Nevertheless,progressisnotstymied.Knowledgeaccruesinsmallincrements,witheachsetofsingle-cellrecordingsalteringthefaceofwhatisknownaweebitatatime.Becauseneurobiologicalsortingtechniquesrelysoheavilyonpreviouslyacceptedneurobiologicalhypotheses,therewilllikelyneverbeanabruptordramaticconceptualrevolution.ButwhatisknowncanevolveslowlybutCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n284valeriegrayhardcastlesurelyuntilthefinalrestingpositionisquitefarremovedfromtheplacewheretheinvestigationbegan.3.localizationandreductionWhenscientistsdosingle-unitrecordingsfromasetofneuronstheyassumethattheyareexaminingadiscretesystem.Theyhavebeenwildlysuccessfulusingthisstrategy,identifyingatleastthirty-sixdifferenttopographicalvisualprocessingareasincortex(DeGelder2000),differentiatingthe‘‘what’’fromthe‘‘where’’objectproces-singstreams(DeYoeandVanEssen1988;Mishkin,Ungerleider,andMacko1983),anddistinguishingmotiondetectionfromcontourcalculations(Barinaga1995),tonamebutafewexamples.Mapsofbrainfunctionaregettingmoreandmorecomplicatedasmoreandmoreislearnedabouttheprocessingcapacitiesofindividualcells.Andalltheseprojectsarefoundedonthebeliefthatbrainshavediscreteprocessingstreamsthatfeedintooneanother.Yetthemostneuronsscientistshaveeverbeenabletorecordfromsimultaneouslyareafewhundred;themostcellstheycaneverseesummedlocalfieldpotentialactivityoverareafewthousand.Butbrainareashavehundredsofthousandsofneurons,severalordersofmagnitudemorethancanbeaccessedatanygiventime.Andtheseneuronsareofdifferenttypes,withdifferentresponsepropertiesanddifferentinterconnectionswithothercells,includingothersimilarneurons,neuronswithsignificantlydifferentresponseproperties,andcellsofcompletelydifferenttypes.Anyconclusionsscientistsdrawaboutthebehaviorofwhatevercellstheyarerecordingfromaregoingtobelimitedtoverybasicstimulus-responseandcorrelationanalysesofwhateverneuronalsubtypetheyarecurrentlyexamining.Hence,thefunctionalitytheyascribeonthebasisoftheserelativelymeagersortsofexperimentsmightbemuchmorerestrictedthanwhatthecellsareactuallydoing.Theyinsertanelectrodeinornearacellandthenrecordwhatitdoesastheystimulatetheanimalinsomefashion.Theyrecordfromacellinavestibularnucleusandthenmovetheanimal’sheadabouttoseewhetherdoingsochangestheactivityoftheneuron.Ifitdoes,thentheymoveitmoreortheymoveitdifferentlyandseehowthatchangestheneuronaloutput.Ifitdoesnot,thentheyeithertryanothernearbycellortrysomeotherstimulus.ButwhattheyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNeurobiology285cannotdoisrecordfromalltheneuronsinsomeisolatedarea,eveniftheareaisverysmall.Andwhattheycannotdoistestanygivencellforalltheknownfunctionalcontributionsofbraincellsingeneral.So,whattheyconcludeaboutanycellwillonlyreflectthecellstheyhaveactuallyrecordedfromusingstimulitheyhaveactuallyused.Thisresearchstrategysystematicallyunderestimateswhenneuronsactuallyrespondandunderwhatconditions.Unitstudiesattempttocombinescores,hundreds,oreventhousandsofsingle-unitrecordingstotrytoanalyzethepopulation.Theoretically,scientistscouldperhaps,inprinciple,delineateanervoussystemregionstereotaxicallyifithadreproduciblecorre-lationsbetweenafferentandefferentconnectionssuchthattheycouldultimatelyarticulatetheneurobiologicalfunctionofthedefinedregion.However,thelikelihoodofsuccessforthistypeofstudydecreasesasthecomplexityoftheorganismincreases.Sci-entistscandrawfunctionalconclusionsregardingtheactivitiesofneuronsintheabdominalgangliaofAplysia,orthesegmentalgangliaoftheleech.Butthearchitectureoftheseorganisms’centralnervoussystemissodifferentfrommammals’thattheprobabilityofsuccessfullyusingsimilartechniquesforunderstandinghumansisverylowtozero.Inaddition,theactualprocessingofinformationthatgoesoninthosecellsinvolveslotsofdifferentkindsofexcitatoryandinhibi-toryinputsfromotherareasinthebrainstem,cerebellum,andcerebralcortex.Thedorsalhornissupposedtointegrateafferentnociceptiveinformationfromtheperipheryandpassitontothemotorsystem(amongotherthings),butitdoesnotdothatsegre-gatedfromtherestofthebrainandwhatthebrainistryingtodo.Itisintegratingandpassingastheorganismistryingtopursuepreyorfleefromanenemy.Moreover,thebrainregionsthatperformthesetasksareoftenconnectedtotheveryareascientistsarerecordingfrom.Themotorsystemfeedsbackdownintothedorsalhorn,asdothethalamusandsignificantpartsofcortex.Theimpactoncognitiveprocessingofsuchrampantfeedbackconnectionsinthebrainisonlyjustnowstartingtobeexploredinneurobiologicalresearch,thoughexactlyhowtodothisisadifficultquestiontoanswer.Ofcourse,neurobiologistsdesigntheirexperi-mentskeepinginmindtheknownanatomicconnectionsbetweenandamongtherelevantstructures.Atthesametime,anyactualCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n286valeriegrayhardcastleexperimentalobservationsofalltheremoteinfluencesonthedorsalhorn,forexample,areimpossible,despitehowevermanyindividualneuronsscientistsrecordfrom.Theysimplydonothaveanywayofconductingsuchextensive,invasivetestsonliveanimals.Atbest,theparticularinfluencesassumedinanyparticularrecordingseriesareamatterofpreviouslyacceptedgospel,dogma,andfaith.Ideally,neurobiologiststrytoconjoinsingle-cellstudieswithsomesortoflesionexperiment.Oncescientistsconstructageneralflowchartoftherelevantstructuresbasedonanatomyexperiments,andtheyhaveestimatednormalunitbehaviorfromaseriesofsingle-cellstudies,theythentrytoknockoutthehypothesizedfunctionsbyplacinglesionsinotherwisenormalanimals.Theyruntheirexperimentsonthebasisoftheassumptionthattheselesions,placedinregionsknowntobeimportant,willchangetheunitbehaviorofcellstheyarestudyinginaconsistentfashion.Iftheywitnesssuchachange,theyusethatinformationtoexplaintherelativefunctionalcontributionsofthelesionedregiontothecellsunderscrutiny.Inotherwords,theyareusinglesionstudiestotrytoderiveafunctionalboxologyforthebrain,justascognitivepsy-chologistsusereactiontimedistributionsanderrormeasurementstofindoneforthemind.Butthereisalargertheoreticalconcern.Whatneurobiologistsknow,butgenerallyignore,isthatanyfunctionalchangeinthecentralnervoussystemwillleadtocompensatorychangeselse-where(e.g.,Merzenichetal.1983).Becausethebrainishighlyplastic,lesioningitinoneplacewillprovokeittoreactinsomefashioninsomeotherplace.Usuallytheseotherplacesarenotcomponentsinthesystemorregionbeingstudied.Buteveniftheyare,neurobiologistsignoreplasticityofthebraininfavorofassumingaconsistentfunctionalalterationascausedbythelesionandnothingmore.Howareinvestigatorssupposedtoevaluatesomeobservedfunctionalchangewhenthedifferencetheyseemighthavebeenevokedbythebrain’sattempttocompensateforitslossandnotbyanyspecificdeficitinducedbythelesion?Theshortansweristhattheycannotiftheyarerestrictedtosingle-cellrecordingsandlesionstudies.Toanswerthisquestionweneedtobeabletoseetheactivityoftheentirebrainatonceandovertime.Theexcitementoverfunctionalmagneticresonanceimaging(fMRI)andotherimagingtechniquesconcernsexactlythisCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNeurobiology287point:thereisawayoflookingattheactivityofthewholebrainatonetimeastiedtosomecognitiveactivityorother.Butmagneticresonanceimaging,thebestnoninvasiverecordingdevicewecur-rentlyhave,onlyhasaspatialresolutionofabout0.1millimeterandeachscansamplesafewsecondsofactivity.Thisimprecisionforeclosesthepossibilityofdirectlyconnectingsingle-cellactivity–whichoperatesthreetofourordersofmagnitudesmallerandfaster–withlargerbrainactivationpatterns.Methodologicaldifficultieswithcurrentimagingtechniquesarenowwellknown(Bechtel2000,CabezaandNyberg1997,2000).MostcenteraroundthefactthatMRIisablood-oxygen-level-dependent(BOLD)measure,whichcanonlybeimperfectlycorrelatedtobrainactivity.Thatis,MRImeasureschangesintheoxygenationlevelofblood;itdoesnotdirectlymeasureanythingaboutactualneuronalactivity.Othersaretiedtothefactthatthemeasurecannotdiffer-entiatebetweeninhibitoryandexcitatoryactivation,andthatcanconfoundthewaytheimagesareinterpreted.Anareamightbe‘‘read’’asbeingpartoftheprocessingstreamforsomeinput,eventhoughwhatisshowingupintheMRanalysisisthatareaactivelydampingdownactivity.Athirdsetoflimitationsistiedtothesparsedis-tributionofsomeprocessingsystems.Ifasystem–nociceptioninsomatosensorycortexmightbeoneexample–iswidelybutsparselydistributed,thenitsactivitylevelmightneverreachwhatisrequiredforaBOLDmeasuretonotice,giventhatcellssurroundingthesystemarenotactivatedbytheparticularstimulusinquestion.Thefinalsetofconcernsrevolvesaroundthesubtractionmethodusedinimagingstudiestoculldata.Inbrief,hereishowthattechniqueworks.Theexperimenterpickstwoexperimentalcondi-tionsthatshebelievesdifferalongonlyonedimension:theydifferonlywithrespecttothecognitiveorperceptualprocessshewantstoinvestigate.Shethencomparesbrainactivityrecordedunderoneconditionwithwhathappensinthesecondcondition,lookingforregionswhoseactivitylevelsdiffersignificantlyacrossthetwo.Theseareas,shebelieves,constitutetheneuralsubstratesofthetaskunderscrutiny.Bysubtractingonesetofscansfromtheother,thehopeisthatonehasremovedactivitynotspecificallyrelevanttothetaskathand.Letussetasidethefactthatthismethodhasnowayofdeter-miningwhetherthedifferencesfoundareactuallytiedtotheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n288valeriegrayhardcastlecognitiveprocessandnottosomethingelseoccurringconcurrentlybutcoincidentally.Letusalsosetasidethefactthatsomeactivitymightbebothrelevanttothetaskathandandrelevanttothebaselinetask.Noticethathowwellthesubtractionmethodwillworkdependsuponthesensitivityofthemeasuringdevicessuchthattheworsetheinstrumentis,thebetterthemethodseemstobeforlocalizationstudies.Lowsignal-to-noiseratio(SNR)meansthatscientistswillfindonlyafewstatisticallysignificantdifferencesacrossconditions.Andthesearethesortsofresultsneurobiologistsneedinordertobolsteranyclaimsidentifyingparticularcognitiveprocesseswithdiscretebrainregions.ButastheimagingtechnologyimprovesandtheSNRincreases,scientistsseemoreandmoresitesthatdifferacrosstrials.Themoresitestheyget,themoreitappearsthatessentiallytheentirebrainisinvolvedineachcognitivecomputation.Andthemoreitappearsthattheentirebrainisinvolvedineachthought,thelessitistheycanjustifyanyassumptionoffunctionalspecificityinthebrain.Ifweextrapolatefromwhatscientistsmightlearnwithmoresensitivemeasures,wecaneasilyseethattherewillbeatimewhenthiswholeapproachjustwillnotworkanymore.Putintheharshestterms,brainimagingseemstosupportreductionismbecausethescienceisnotverygoodyet.Forexample,Brodmanarea6appearssignificantlyactiveaftersubtractioninstudiesofphoneticspeechprocessing,voluntaryhandandarmmovements,sight-readingofmusic,spatialworkingmemory,recognizingfacialemotions,binoculardisparity,sequencelearning,idiopathicdystonia,pain,itch,delayedresponsealterna-tion,andcategory-specificknowledge,tolistonlyasubsetofactivitiesinwhichitissignificantlyanddifferentiallyactive.Itcouldbethecasethatifscientistskeepondoingthesortofsub-tractionstudiesthattheycurrentlyaredoing,theneventuallytheywillfindaunifyingandpithywaytodescribewhatpremotorcortexisdoinginhumans.Inthisinstance,neurobiologywouldbeontherighttracktodeterminingbrainfunction,buttheystillhavealongwaytogo.Butitcouldalsobetruethathowaregionfunctionsdependsheavilyonthe‘‘neuralcontext.’’Itsfunctionalroleinacognitiveeconomydependsonhowitisconnectedtootherareasandhowthoseotherareasareresponding.(ThefunctionoftheseareaswouldalsobedependentontheirparticularconnectivityandCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nNeurobiology289thecurrentpatternsofactivation.Andsoitwouldgo.)Ifthisiscorrect,thensearchingfor‘‘the’’functionofparticularareasismisguided,fordifferentbrainregionsplaydifferentrolesdependinguponthecognitivetasksathand.4.neuroethicsAsprogressismadeintounderstandinghowthebrainworksandhowtoinfluencebrainfunctioning,seriousethicalquestionsariseconcerninghowthemedical,insurance,andgovernmentalleadersshouldreacttonewinformationandpossibilities(Marcus2004).Neuroethicsisanewlyburgeoningareaofresearch,withnationalattentiononlynowbeingfocusedontheissues.Particularquestionsthatphilosophersofneurobiologywillhavetoanswerconcernhowandwhetherweshouldalternormalfunctioningbrains,howandwhetherweshouldusebraintechnologytotrackindividuals’socialbehavior,andhowandwhetherwhatwelearnaboutthebrainchangesthewaywethinkofourselvesashuman.Weknowalotabouthowmemoryworks,and,moreimportantly,howitfailsus.Sevenbasicwaysinwhichmemorycanfailaredecreasingaccessibilitytomemoriesovertime,lapsesinattention,temporaryinabilitytoaccessstoredinformation,falserecognitionofsomething,falsememoryofsomething,contaminationofstoredinformationbycurrentbeliefs,andrememberingofitemsatinap-propriatetimes.Alloftheseprocessesareperfectlynormalandoccurinallofusatsometimeoranother.Supposewehavesomewayofcorrectingsomeorallofthesedeficits.Shouldwe?Orshouldweacceptless-than-perfectmemoriesasthewayweare?Neurobiologistsarealreadytrackingwhereandhowmoraldeci-sionsaremadeinthebrain;theyarealsolookingatbraindifferencesbetweennormalandsociopathic,psychopathic,andviolentlyimpulsiveindividuals.Weknowthatsuchindividualsrespondtoviolentorotherwisedisturbingsituationswithincreasedactivityintheamygdalaanddecreasedactivityinthefrontallobesrelativetonormalindividuals.Wecannowidentifysuchtrendsinindividualsbeforetheyactuallycommitanycrime.Shouldwe?Andwhatshouldwedowithsuchinformationoncewehaveit?Ifwecometobelievethatviolenceisbiologicallybased,asareallotherbehavioraldecisions,thenwhatdoesthissayaboutnotionsofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n290valeriegrayhardcastleselforfreewill?Howmightthisalterourcourtsystems,sincetheyoperateundertheassumptionthatoneisguiltyifonecouldhavedoneotherwiseinasituationbutchosenotto?Similarquestionsarisewithgenderdifferencesinthebrains.Weknowthatfemalebrainsdifferfrommales’.Whateffect,ifany,shouldthisfacthaveonoureducationalsystems,oursocialexpectationsofgenderedbehavior,ormen’sandwomen’sprofessionallives?Weareonlybeginningtoconfrontthesesortsofquestions,asourtechnologyisonlybeginningtoallowustounderstandandchangethebraintoanysignificantdegree.Asourknowledgeofthemind/braincontinuestoincreaseexponentially,theseandothersimilarquestionswillonlybecomemorepressing.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nchristopherhorvath15BiologicalExplanationsofHumanSexualityTheGeneticBasisofSexualOrientationDogenesmakepeoplegay?Thisisaquestionscientistswhostudyhumansexualityhearallthetimefrompoliticians,students,andpeoplejustcuriousabouttheirownsexualorientation.Manypeoplebelievethatthecausationofsexualorientationhasimportantimplicationsforthemoralstatusofhomosexuality.Bothcommonsenseandmoraltheorytendtoevaluatetheacceptabilityofapersonsbehavioronthebasisofthecontrollabilityofthatbehavior.Onecannotbeheldmorallyblameworthyforthingsoverwhichonehasnocontrol.Becauseindividualshavenocontroloverthegenestheyinherit,accordingtothistheorytheyshouldnotbeheldresponsibleforbeinggayifitisgeneticallydetermined.ThisviewisapparentinsurveysofAmericansthathavefoundanassociationbetweenthebeliefthathomosexualityisgeneticallydeterminedandlessnegativeattitudestowardsgaypeople(Schmalz1993,Tygart2000).Thepurposesofthisessayaretoreviewwhatthebestsciencecurrentlyavailablesaysaboutthegeneticcausesofsexualorienta-tionandtodiscusssomeofthenormativeimplicationsand,moreimportantly,nonimplicationsofthatscience.Aswithanyissueinwhichthereissubstantialpoliticalandscientificcontroversy,thereisongoingdebateabouttheverydistinctionsthatlieatitsfounda-tion.Nevertheless,termslikeheterosexualandgayarecom-monplaceinboththepopularandthescientificliterature.Mypurposehereisnottodebatethedistinctions,buttodiscussthebestsciencewehaveavailabletodayasitisappliedtothedistinctionsascommonlyunderstoodbybothexpertsinthefieldandthepublicatlarge.Ihavestructuredthisarticlearoundthesortsofquestionsstudentsorpeopleunfamiliarwithrecentworkinthefieldmightask.291CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n292christopherhorvathhowdoscientistsdefinesexualorientation?TheKinseyScale(Kinsey,Pomeroy,andMartin1948)hasbecomethestandardwaytooperationalizesexualorientationinalmostallscientificstudiesofthephenomenon.Studyparticipantsareaskedaseriesofquestionsabouttheirsexualdesires,behaviors,fantasies,andself-identity.Onthebasisoftheanswersgiventotheseques-tions,eachparticipantssexualorientationisclassifiedsomewherealongaunidimensionallinearscalewhere0¼exclusivelyhetero-sexualand6¼exclusivelyhomosexual.Someresearchersweighttheanswerstocertainquestionsmoreheavilythanothers,dependingonthenatureofthestudy.Thosewhoareinterestedininvestigatingsexualorientationasasetoferoticdesiresdirectingsexualattractiontowardsparticularkindsofpeopleusuallyweightheanswersaboutdesiresandfantasieshigherthantheanswersaboutbehavior.Thiswouldseemaperfectlyrea-sonablebias.Tosaythatvirginshavenosexualorientationsimplybecausetheyarevirginswouldbetoflyinthefaceofhundredsofyearsofromanticliteratureaswellasmostpeoplespersonalexperience.Itisalsoobviouslytruethatpeoplewhofindthemselveslivingorworkingwithinveryhomophobicenvironmentsmightneveractontheirtruesexualdesiresoutoffearofreprisalorrejection.CriticsofbiologicalexplanationsofsexualorientationarguethattheconceptionofsexualattractionimplicitintheKinseytestisinadequatebecauseittreatssexualorientationasadiscretetraitwithclearlydefinedboundaries(e.g.,Stein1999).IntermsoftheKinseytest,everybodyfitsneatlyintooneofsevencategories.Sexualorientationisclearlyacontinuouscharacter.Peoplearegenerallyattractedtootherpeoplewithvaryingdegreesofintensity.Onepersonmightbehighlyattractedtoothermalesandonlyslightlyattractedtofemales.Asecondpersonmightbehighlyattractedtobothmalesandfemales.TheKinseyscalecollapsesattractionstobothmalesandfemalesintoasingleunidimensionalscaleanddividesthatspectrumintosevendiscretecategories.However,thepointsofdivisionandthenumberofcategoriesarearbitrary.Inpractice,sincemostresearchersareonlyinterestedinpeoplewhosesexualorientationliesononeortheotheroftheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nExplanationsofHumanSexuality293extremes,theycollapsethesevenKinseycategoriesintoamoreeasilymanagedtwoorthree.ItispossibletousethesameKinsey-styleinstrumentfocusingonattractionandbehavior,butmeasuresexualorientationasacon-tinuousvariable(Ellis,Burke,andAmes1987).ThisprocedureresultsinaparticularpatternofdistributionconsistentwiththeresultsofthemorestandardKinseytest.Mostpeopleclusteraroundtheheterosexualendofthedistribution,whilethereisasmallerclusterattheother,homosexualend.AlsoasinthemorestandardKinseytest,therearemoremalesthanfemalesatthehomosexualendofthespectrum.Femalesarealsomorewidelydistributedalongthemiddleofthespectrumwhilemalestendtoclusterattheextremes.Ifyouwanttoturnacontinuouscharacterintoadiscreteoneforsomepracticalorheuristicreason,youcanalwayscreatethenecessarycategoriesbydividingthespectrumintodiscretehunkswithboundaries.Itistobehopedthattheseboundarieswouldnotbeplacedatcompletelyarbitrarypointsinthespectrum.Instead,onewouldhopeforsomeprincipledreasonsbehindthecreationofthevariouscategories.Forexample,withsexualorientation,mostresearcherstakecaretomaketheseboundariesconsistentwithpeoplesself-reportedsexualidentity.issexualorientationgeneticallydetermined?ThisquestionmostcommonlytakestheformofIsthereagaygene?Forseveralreasonsitisimportanttorealizethereisnosuchthingasagaygene.First,thetermgaygeneisamisnomerinthesamewaythatstraightgenewouldbe.Thesamegenevariantthatpredisposessomeonetohomosexualitywhenpresentwouldpre-disposetowardsheterosexualitywhenabsent.Thus,itismoreappropriatetoask,Issexualorientationgenetic?Second,becausesexualorientationisacomplexphenomenon,itisunlikelythatanysinglegenedeterminessexualorientationwithtotalcertainty.Rather,itismorelikelythatseveralgenesinteractwitheachotherandwithenvironmentalfactorstoinfluencethedevelopmentofsexualorientation.Iftheyexist,suchgeneswillpredisposesomeonetowardsaparticularsexualorientationratherthandeterminehisorherorientationwithcertainty.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n294christopherhorvathThreeresearchmethodshavebeenusedtoexploretheexistenceofgeneticinfluencesonsexualorientation:familypedigreestudies,twinstudies,andmoleculargeneticstudies.Eachcontributesdis-tinctandusefulinformationaswellashavingspecificlimitations.Familystudiesareusedtoanswertwomainquestions.First,doesthetraitofinterestruninfamilies?Iftheanswerisno,itispointlesstocontinuetolookforgenes.Notalltraitsthatruninfamiliesaregeneticallydetermined,buttraitsthatdonotruninfamiliescannotbegeneticallydetermined.Second,bylookingforpatternsinthewayatraitispassedfromgenerationtogenerationitispossibletogetanideaaboutwhereinthegenomethepredisposinggenesmightbelocated(i.e.,onautosomesorsexchromosomes).Inordertodemonstratethatgenesinfluenceatraitsuchassexualorientation,onemustshowboththatitrunsinfamiliesandthatitoccursinthesefamiliesatlevelshigherthanthebaserateinthegeneralpopulation.Studieshavefoundthat,onaverage,therateofhomosexualityinmenwhohavegaybrothersisapproximately9percent(seeBaileyandPillard1995,Mustanski,Bailey,andChivers2002,forreviews).Femalehomosexualityalsoappearstoruninfamilies,althoughestimatesofhomosexualityamongsistersoflesbiansvarywidely,rangingfrom6to25percent(BaileyandPillard1995).Inordertocomparetheserateswiththepopulationbaserate,datafromalargesexualitysurveyareneeded.Themostrecentlargesurveyofsexualbehavior,conductedinthesamecountrywherethefamilydataweregenerated(UnitedStates)andusingthesamecri-teria(self-identificationasgay),producedratesofhomosexualityof2.4percentinmenand1.4percentinwomen(Laumann,Michael,andGagnon1994).Thefactthattherateofhomosexualityinrela-tiveswithagayfamilymemberishigherthanthepopulationbaseratesuggestsevidenceforthefamilialityofbothmaleandfemalesexualorientation.Familystudiesarealsousefulfordeterminingwhetherthepat-ternoftransmissionofatraitisautosomalorsexlinked.Humanshavetwenty-threepairsofchromosomes,twenty-twoofthemclassifiedasautosomesandonepairclassifiedassexchromosomes.Sex-linkedtraitsarethosethathavepredisposinggenestransmittedontheXorYchromosomes,andtheyhavedistinctpatternsofinheritance.SincetheYchromosomeissmallandisbelievednottocontainmanygenes,mostsex-linkedtraitsareinfluencedbygenesCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nExplanationsofHumanSexuality295ontheXchromosome.MalesinherittheirYfromtheirfatherandtheirXfromtheirmother;thusmaternaltransmissionofatraitisthestrongestevidencethatageneislocatedontheXchromosome.Thefirstfamilystudytoexaminepatternsoftransmissionfoundgaymenhadmoregayunclesandcousinsonthematernalsidethanthepaternalsideoftheirfamily(i.e.,maternaltransmission;Hameretal.1993).Aspreviouslymentioned,suchapatternofresultssuggeststhatageneassociatedwithmalehomosexualitymaybelocatedontheXchromosome.Onesubsequentstudyreplicatedthefindingsofmaternaltransmission(Riceetal.1999),whiletwoothershavenot(Baileyetal.1999,McKnightandMalcolm2000).Moreresearchwillbeneededtoclarifythisincongruity.Ifthematernaltransmissionhypothesiswereconfirmed,itwouldaddtoagrowingbodyofevidencesuggestingthegeneticsofsexualorientationaredifferentinmenandwomen.Familystudiescandemonstratebothwhetheratraitispasseddowninfamiliesandpossiblythepatterninwhichitisinherited,butsimplythatatraitispasseddowninafamilydoesnotnecessarilyindicatethatitisgenetic.Religiousaffiliation(e.g.,MethodistversusCatholic),forexample,tendstobepasseddownfromparentstochildren,butisnotgeneticinnature.Cultureandfamilytraditionareconfoundedwithgenetictransmissioninfamilystudies.Twinstudiesovercomethisissuebycomparingthesimilarityofmono-zygotic(identical)twinswhoshare100percentoftheirgenestodizygotic(fraternal)twins,whoshare50percentoftheirgenesincommononaverage.Becausebothtypesoftwinsareraisedinthesamefamilyandsharethesameenvironmentalexperiences(includingsharingthesameuterineenvironment)differencesintheirdegreeofsimilarityshouldbedueprimarilytodifferencesintheirdegreeofgeneticsimilarity.Usingstatisticalprocedurestocomparetwincorrelations,theheritabilityoftraitscanbecalcu-lated.Heritabilityreferstotheproportionofvariabilityinatraitwithinthepopulationthatisduetogeneticeffects.ItistypicallyrepresentednumericallyonascalefromH¼0(nogeneticinfluences)toH¼1(totallygeneticallycontrolled).Severalrecenttwinstudieshavebeenconductedonlarge,systematicallyrecruitedsamples(reviewedinMustanskietal.2002).Onerecentstudyof756pairsoftwinsintheUnitedStatesestimatedaheritabilityof.62forself-reportedsexualorientationinCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n296christopherhorvathmenandwomen(Kendleretal.2000).Anotherrecentstudyof1,405twinpairsidentifiedthroughanAustraliantwinregistryproducedheritabilityestimatesof.58formenand.30forwomen(Kirketal.2000).Toputthisheritabilityestimateinperspective,theherita-bilityofbodymassindexhasbeenestimatedatapproximately.80(Pietilainenetal.1999)andalcoholismatbetween.50and.60(McGue1999).Althoughtwinstudiesarehelpfulindeterminingwhethergenesareimportant,theyarenotabletotellyouwhichparticulargenesareatwork.Moleculargeneticresearchisneededforthispurpose.Onetypeofmoleculargeneticstudy,calledlinkageanalysis,hasbeenusedtostudymalesexualorientation.Linkagestudieslookforspecificchromosomalregionsthatarepasseddowninfamiliesalongwithatrait(e.g.,sexualorientation)atprobabilitylevelsgreaterthanchance(50percentforsiblings).Becausesomefamilypedigreestud-ieshavesuggestedthatsexualorientationistransmittedthroughthematernalline,thestudiespublishedtodatefocusonlyontheXchromosome.Hamerandassociates(1993)reportedthefirstlinkagestudyinfortyfamiliesofgaymenandfoundanassociationbetweenmalesexualorientationandaregionoftheXchromosomenamedXq28.Thisfindingwasreplicatedintwoindependentsam-ples(Huetal.1995,Sanders1998),andnotreplicatedinanother(Riceetal.1999).Furthermore,theregionwasshownnottobeanimportantdeterminantoffemalesexualorientation(Huetal.1995).GiventhemixedevidenceforanassociationbetweenXq28andmalesexualorientation,futureresearchwillbeneededbeforeanydefin-itiveclaimscanbemadeforitsimportanceinhelpingtodeterminesexualorientationinmales.ItisimportanttonotethatXq28isachromosomalregion,notaspecificgene.Findinglinkagetoachromosomalregiontellsyouthatagenelocatedwithinthisregionisrelatedtoaparticulartrait.IfXq28provestobeassociatedwithmalesexualorientation,thenextstepwillbetodeterminewhichgenewithinthisregionisimportant.Futureresearchwilllikelyexplorethis,aswellasgenesonotherchromosomes.Researchersmaylookforgenesonotherchromo-somesbecausetheevidenceformaternaltransmissionismixed,andbecauseevenifgenesontheXchromosomeareimportant,thatimportancedoesnotprecludegenesonotherchromosomesfromalsohavinginfluence.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nExplanationsofHumanSexuality297Althoughnospecificgenesformalesexualorientationhavecurrentlybeenidentified,andnochromosomalregionshavebeenimplicatedinwomen,itispossibletodrawseveralconclusionsabouttherolethatgenesplayindeterminingsexualorientation.Twinandfamilystudieshavedemonstratedthatgeneticinfluencesdoexplainaportionofthevariabilityinsexualorientationwithinthepopulationsthathavebeenstudied.Genesalsoseemtobemoreinfluentialinmalesexualorientation.Finally,thereissomeevidencethatthechromosomalregionXq28mayplayaroleinmalesexualorientation,butfurtherresearchisneededbeforeanydefinitiveconclusionscanbedrawn.howcouldhomosexualityhaveevolved?Peoplewhoaskhowhomosexualitycouldhaveevolvedareusuallywhatphilosophersofbiologyhavecometocallempiricaladapta-tionists(Godfrey-Smith2001).Theyconceiveofnaturalselectionastheonlyevolutionaryprocessofanyrealconsequenceandwonderhowthegenesthatcontributetothedevelopmentofatraitsuchashomosexualitycouldbemaintainedinapopulationwhentheyarelikelytodecreaseratherthanincreasethereproductivesuccessoftheorganismsthatbearthem.First,thepresenceofgenesthatcontributetohomosexualityincontemporarypopulationsisexplainedbytheirrelativefitnessinpriorgenerations.Whileitmayseemintuitivelyobviousthathomosexualhumanshavealowerfitnessthanheterosexualhumansasmeasuredbyactualreproductivesuccess,therearenogoodempiricalstudiesthatprovideactualevidencethathomosexualshavehadsubstantiallylowerfitnessthroughouttherelevantevolutionarypast.Ourintuitionsaboutthereproductivesuccessofhomosexualsarebasedonourmodernconceptionofgayandlesbianpeople.Whenwethinkofhomosexuals,wethinkofthesingleurbangaymanorwoman,orwethinkofthenicegaycouplelivinginthehousedowntheblock.Theideaoforganizingonesidentityandoneslifearoundhomosexualdesiresisverycontemporary(Halperin1990)andlargelyconfinedtoaveryfewcultures(Herdt1997).Forthevastmajorityofhumanhistoryandeventodayinthemajorityoftheworldscultures,peoplewhohavehomosexualdesiresarefarmorelikelytostructureCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n298christopherhorvaththeirlivesinheterosexual-typicalways.Ifculturalmoresrequireadultstomarryandtoproduceoffspring,thenhomosexualmembersofthecultureareunderasmuchpressuretoaccedetotheseexpec-tationsasheterosexualmembers.Whatlittleweknowaboutthewayhomosexualmembersofthepopulationhavelivedforthemajorityofhumanevolutionaryhistoryunderminestheintuitionthattheiractualreproductivesuccessissubstantiallydifferentfromthatoftheheterosexualmembersofthepopulation.Ofcourse,itwouldbenicetohavesomeactualempiricaldataonwhichtobaseanevolutionaryhypothesis.Unfortunately,suchdataabouttheactualreproductivesuccessofhomosexualsthroughouthumanhistorydonotexist.Forthelastthirtyyears,thekinselectionhypothesiswasthemostwidelyheldevolutionaryhypothesisregardingthemain-tenanceofhomosexuality.In1975,E.O.Wilsonarguedthatinsteadofproducingtheirownoffspring,homosexualmenreproduceindirectlybycontributingtheirresourcestotheoffspringoftheirrelatives.Wilsonhypothesizedthathomosexualmalemembersofafamilymighthaveactedashelpersinhuntingorinvariousotherdomesticchores.Otherstheorizedthatthehomosexualfamilymemberscontributionmighttaketheformofsharingincome,bequeathingwealth,orassistinginchildrearing(Ruse1981,Weinrich1987).Averyimportantpartofthekinselectionhypothesisisthathomosexualmalespossessasetofcharacteristicsthatmakethembettersuitedtolifeasdomestichelpersthanasactualfathers.Ruse(1981)andWeinrich(1987)arguedthathomosexualmalespossessasetofcharacteristicsthatsubstantiallyreducedtheirchancesofsuccessinthecompetitionformates.Thereisagreatdealofevidencetosuggestthathomosexualmentendtoexhibitarangeofgender-atypicalbehavioralcharacteristicsthatwouldhavecon-tributedtodecreasedmatingsuccessintheevolutionarypast(Baileyetal.1994).Forexample,studiesofcontemporaryhomosexualmalesshowthattheytendtobelessphysicallyaggressivethanheterosexualmales(BaileyandZucker1995).Homosexualmalesalsotendtobelaterbornsonswithadis-proportionatenumberofolderbrothers(BlanchardandBogart1996).Bythetimethehomosexualsoncamealong,thefamilymayhavealreadyinvestedthemajorityofitsresourcesintheearlierbornsons.Thus,thelaterbornhomosexualsonmayhavehadagreaterchanceCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nExplanationsofHumanSexuality299ofgettinghisgenesinthenextgenerationbycontributingtothereproductivesuccessofhisoldersiblingswhowerebetterequippedtocompetesuccessfullyformates.Inadditiontothepresenceofcharacteristicsmakingthemlesssuitedforthecompetitionformates,thekinselectionhypothesisincludesthesuppositionthathomosexualmaleshadsex-atypicaltraitsthatmadethemmorewillingtocontributeresourcestotheirniecesandnephewsthantheirheterosexualsiblings.Femalestypi-callyexhibitagreaterdegreeofempathythanmales.In1995,SalaisandFischerreportedevidencethathomosexualmenwereonaveragemoreempatheticthanheterosexualmen.Proponentsofthekinselectiontheoryhypothesizedthattheincreasedlevelofempathymighttranslateintoanincreaseinaltruisticbehaviortowardsrelatives.Whilethekinselectionhypothesishasbeenthedominantevo-lutionaryhypothesiswithrespecttohomosexualitysinceitscon-ception,thereisnoshortageofscientistswhorejectitasaviablealternative.Somerejectitontheoreticalorconceptualgrounds(e.g.,Dickemann1995).Othersrejectitbecausethefewempiricalstudiesdirectedattestingthehypothesishavefailedtoyieldanyconvincingsupportingevidence.Forexample,BobrowandBailey(2001)inves-tigatedwhethercontemporaryhomosexualmenweremoreorlesslikelythanheterosexualmentocontributefinancial,emotional,andsocialresourcestosiblings,nieces,andnephews.Theyfoundthat,contrarytothepredictionsofthehypothesis,heterosexualmentendtocontributemorefinancialresourcestokinthanhomosexualmen.Thiswouldseemtobestrongevidenceagainstthekinselectionhypothesis.Despiteitsconsiderableconceptualandtheoreticalweaknessesandtheempiricalevidenceagainstit,thekinselectionhypothesishasnotyetbeenrejected.Iwouldarguethatthisisdueinlargeparttotheadaptationistbiasprominentinsociobiologyandevolutionarypsychology.Another,muchlesswellrecognizedreasonfortheper-sistenceofthekinselectionhypothesisisthefactthatwhilemostresearchonsexualorientationhasbeenconductedwithinahighlyhomophobiccontext,thishypothesistreatshomosexualityasanadaptation.Onthekinselectionhypothesis,homosexualityisnotseenasadevelopmentalflawordisease,butasanormalvariationinhumansexualitythatconveysapositiveadvantagetothefamiliesCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n300christopherhorvathinwhichitisfound.Apreferenceforviewinghomosexualityinthispositivelightisnotsomuchareasonforbelievingthehypothesisasitisamotivationtoresistthealternativenonadaptationisthypotheses.Onesuchnonadaptationisthypothesisdevelopedoutofthebirth-ordereffectdescribedearlier.Thehigherincidenceofhomosexualityinlaterbornsonsmightbearesultofamaternalimmuneresponsetooneormoreofthemale-specific,Y-linked,minorhistocompat-abilityantigens(referredtocollectivelyasH-Yantigens)carriedbythemalefetus(BlanchardandKlassen1997).ThemothersimmuneresponsewouldbetriggeredbythepresenceoftheH-Yantigensofthefirstmalefetusshecarriedandwouldbecomestrongerwitheachsubsequentpregnancy.Thereisgoodevidence,independentofthishypothesis,thatH-Yantigensplaysomeroleinthesexualdiffer-entiationofvertebrates(Wachtel1983).Theyareusuallypresentintheheterogameticsex(malemammals)andabsentinthehomo-gameticsex(femalemammals)andtheyhavebeenhighlyconservedthroughoutvertebrateevolution(Wachtel1983).Forthesereasons,BlanchardandKlassen(1997)hypothesizethattheH-Yantigensareinvolvedinthedevelopmentofsex-typicalpsychologicalandbe-havioralcharacteristics.TheygoontohypothesizethatexposuretoH-Yantibodiesproducedbythemotherwouldaffectthemalefetussdevelopmentandpossiblyproducemaleswithsex-atypicalsexualbehavior(BlanchardandKlassen1997,374).Thus,theinteractionbetweentheH-Yantigensofthemalefetusandtheimmunesystemofthemotherwouldproducemaleoffspringwitharangeofgender-nonconformingbehaviorthatmightwellincludemorefemale-typicalsexualdesiresthatis,sexualdesiresdirectedatmales.ifsexualorientationisheritable,doesthatmeanitiscompelled,immutable,andinnate?doesthatmakeitmorallyacceptabletoactonhomosexualdesires?Despitecommonassertionstothecontrary,evidenceforbiologicalcausationdoesnothaveclearmoral,legal,orpolicyconsequences.Theargumentthatbiologicalcausationdoesrenderhomosexualitymorallyacceptablemightproceedasfollows:ApersoncannotbeheldmorallyblameworthyforaparticularbehaviorifthatbehaviorCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nExplanationsofHumanSexuality301isaresultofneurophysiologyandifheorshepossessesgenesthatpredisposehimorhertohavesuchneurophysiology.Putthisway,itiseasytoseethatsuchreasoningisirrational.Inorderforapersontobemorallyblameworthyforsomeaction,thatactionwouldhavetobenotaresultofgeneticallyinfluencedneurophysiology.Buttherearenobehaviorswhoseproximatecauseisnotneurophysiologicalandbrainsareallaproduct,atleastinpart,ofgenes.(ForamoredetailedversionofthisargumentseeGreenbergandBailey1993).Itisalsofrequentlyassumedthattraitsthathavegeneticunder-pinningsarecompelled,unchangeable,andinnate.QuestionsalongtheselinesusualtaketheformofDopeoplechoosetheirsexualorientation?Canpeoplechangesexualorientation?Arepeopleborngay?Whiletheseareinterestingquestionsintheirownright,itisimportanttounderstandthatknowingthatatraitisheritabledoesnotprovideaclearanswertoanyofthem.Atraitissaidtobecompelledifitiscompletelydeterminedbyfactorsoverwhichonehasnocontrol.Traitsthathavebothhighandlowheritabilitycanbothbecompelled.Eyecoloriscompletelygeneticallydeterminedandtotallyinvoluntary,whilethelanguageyouaretaughtinchildhoodistotallysociallydeterminedandyetequallycompelled.Environmentalorgeneticdeterminationisuninformativeabouttheextenttowhichapersonchoosessexualorientation.Forexample,onehypothesisaboutthedevelopmentofsexualorientationbasedonlearningtheorypositsthatwomenbecomelesbiansbecauseofnegativeexperienceswithmen.Yet,ifthiswerethecase,homosexualitywouldbenomorevoluntarythanifitweregeneticallydeterminedbecauseitwascausedbyeventsbeyondthewomanscontrol.Thefactthatatraitisheritabledoesnotmeanthatitisimmu-table,orresistanttochange.Thisisbecauseheritabilityestimatescan,anddo,changeovertimeandwhenthedistributionsofgenesand/orenvironmentschange.Forexample,recentresearchhasdemonstratedthatthepercentageofyoungadultsandtheamountofmigrationindifferentcommunitiesstronglymoderatetheherita-bilityofadolescentalcoholism(Dicketal.2001).Thissuggeststhatcommunitieswithmoreyoungadultsandgreatersocialmobilityallowforincreasedexpressionofgeneticdispositions,whereascommunitieswithfewerolderpeersandgreatersocialstructurefacilitategreatersocialcontroloverdrinking.Suchinteractions,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n302christopherhorvathwhichscientistsbelieveareimportantformostcomplextraits,demonstratethatheritabilityestimatesshouldnotbeusedtodeterminewhetheratraitisimmutable.Instead,itispropertoreviewtheevidenceconcerningattemptsatchangingit.Theresultsofsuchreviewssuggestnoevidencethatpsychotherapeuticorreli-giousconversiontherapiesareeffectiveinchangingsexualorienta-tion(Haldeman1994).Perhapsthemostcommonlymisunderstoodassociationisbetweenheritabilityandinnateness.Thegeneralpublicthinksofatraitasinnateifitispresentatbirth.Theassumptionseemstobethatifatraitispresentatbirthitmustbetheproductofgeneticratherthanenvironmentalfactors.Ofcourse,biologistsknowthatalltraitsareaproductoftheinteractionbetweengeneticandenvironmentalfactorsandthatagreatdealofthisinteractionoccursbeforebirth.Nevertheless,thedistinctionbetweengeneticallydeterminedtraitsandlearnedtraitscontinuestobestronglyheldbythegeneralpublic.Toillustratethedisconnectbetweentheconceptsofheritabilityandinnatenessconsiderthefollowingexample:Thetendencytohavetwothumbsoneoneachhandisinnateandmostpeoplehavetwothumbs.Whatvariationdoesexistinthumbnumberislargelyenvironmental(i.e.,losingathumb).Becauseheritabilityisdefinedastheproportionofvariabilitywithinapopulationduetogeneticfactors,andvariabilityinthumbnumberislargelyenvironmental,theheritabilityofthumbnumberislow.Butfewpeoplewouldarguethatthumbnumberisnotaninnatechar-acteristic.Mostresearchersinthefielddobelievethatsexualorien-tationisinfluencedbyprenatalfactors,butnotbecauseitisheritable.Instead,theyareconvincedbyevidenceofassociationsbetweenhomosexualityandtraitsthateithershowupearlyindevelopment(suchaschildhoodgenderedbehaviors;BaileyandZucker1995)orarebelievedtobedeterminedbeforebirth(suchashandednessandfingerlength;seeMustanskietal.2002forreview).Peoplelookingforanswerstonormativequestionsabouthomo-sexualitybyexaminingitsbiologicaloriginsarefocusingonthewrongissue.Thevalueofacharacteristicsuchashomosexualitydependsonitseffectsratherthanitscauses.Homosexualrelation-shipsbetweenconsentingadultsdonotinthemselvescauseharmoratleastdosonomoreoftenthanheterosexualrelationshipsdo.Furthermore,attemptsbyhomosexuallyorientedindividualstoliveCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nExplanationsofHumanSexuality303heterosexuallyarefraughtwithharm,bothtotheselfandtoothers,especiallythosewhoformheterosexualrelationshipswithsuchindividuals,suchastheirwivesandhusbands.conclusionScientistswhostudyhumansexualityareinterestedinavarietyofquestionsabouthumanbehaviorandtheneurophysiologythatunderliesit.Nevertheless,thequestionsthataremostinterestingtothepublicatlargearequestionsabouttherelationshipbetweengenesandsexualorientation.Althoughtherearemanyquestionsthatcannotbeansweredatthistime,therearesufficientscientificdatatoallowsomedefinitivestatementstobemade.Homosexualitydoesappeartoruninfamiliesandstudieshaveestablishedthatgenesplayasubstantialroleinexplainingindividualdifferencesinsexualorientation.ThereissomeevidencethataregionontheXchromosomemayinfluencemalesexualorientation,butfurtherresearchisneededandisunderway.Thesumoftheevidencefromthesestudiesalsostronglysuggeststhatthegeneticsofhomo-sexualitymustbedifferentformenandwomen.Thematernalimmuneresponsehypothesisisreplacingthesociobiologicalkinselectionhypothesisasthedominanttheoryregardingtheultimatecauseofmalehomosexuality.Thematernalimmuneresponsehypothesisprovidesanonadaptationistexplana-tionthatistheoreticallymorecoherentandmoreconsistentwithalloftheempiricaldatawehaveatthispoint.However,itcannotexplaintheexistenceoffemalehomosexuality.Giventhatthedatacollectedtodatestronglysuggestthatmaleandfemalehomo-sexualityhavedifferentpatternsofinheritance,onemightexpecttheretobedifferentcausalexplanationsforthemaintenanceofthebiologicalsystemsthatproducethem.Interestinthegeneticcausationofhumansexualorientationisdriveninlargepartbythesocialandpoliticalcontroversyoverthestatusofhomosexualsandhomosexualrelationshipswithinmodernculture.Nevertheless,itisimportanttoappreciatethefactthattheheritabilityofatraitprovideslittleinformationabouttheextenttowhichitiscompelled,immutable,innate,or,mostimportantly,morallyacceptable.Allpeopleshouldbetreatedwithdignityandrespectregardlessoftheirsexualorientation.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nzacharyernst16GameTheoryinEvolutionaryBiology1.introductionGametheoryisnowastandardtoolforexplainingpuzzlingandcounterintuitivebehavior.Butinspiteofthefactthatgametheorywasdevelopedtoinvestigaterationalandeconomicbehaviorofmodernhumans,ithasfoundequallyvaluableapplicationinbiol-ogy.Forexample,ithasoftenbeenobservedthatwhenfightsbreakoutbetweenmembersofthesamespecies,theantagonistsoftendisplayrestraintbynotinflictingseriousinjuryoneachotherduringthefight.Totakeanotherexample,individualguppieswillsome-timesgooutoftheirwaytoswimbesidealargerfishthatmayturnouttobeapredator(DugatkinandAlfieri1991a,b).Andchimpan-zeeswilloftenraiseanalarmtotherestoftheirgroupwhentheyspotadangerouspredator,inspiteofthefactthatanindividualwhodoessowillattracttheattentionofthepredator.Thesebehaviorsshouldbepuzzlingtoanyonewhothinksofevolutionas‘natureredintoothandclaw’.However,theycanallbeexplainedinaverysatisfyingwaybyapplyingsimplegame-theoreticanalyses.Thepurposeofthischapteristoillustrateenoughgametheorytoshowhowitmaybeappliedtothetaskofexplainingsuchpuzzlingbehavior.Althoughtherangeofsuchbehaviorsisextremelylarge,andtherangeofavailablegame-theoretictechniquesisequallylarge,Ishallfocusongame-theoreticexplanationsofoneparticularkindofbehavior–namely,altruism.Inwhatfollows,Ishallrefertoanybehaviorthatlowersthefitnessofitsbearer,butraisesthefitnessofanotherindividual,asaltruistic.Whenbiologistsrefertoabehavioras‘altruistic’,they304CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology305meanonlythatithasthiseffectonfitness–noclaimismadeaboutthepsychologicalstatesoftheindividualsinvolved.Indeed,thereisnoassumptionthattheindividualsevenhavepsychologicalstatesatall.Forexample,itturnsoutthatsomevirusesreproducemuchmoreslowlythantheycould.Suchlowvirulencetherebybenefitsothervirusesinthehost(bykeepingthehostalivelonger)whileloweringthefitnessofthevirus.Wemaythereforeproperlycallthevirusan‘altruist’,eventhoughitobviouslyhasnocapacityforhavingaltruisticorself-sacrificialfeelingsandsentiments.Furthermore,thebehaviorofthisvirusiseverybitaspuzzlingasthepsychologicallyladenbehaviorofachimpanzeeormodernhuman.Explainingtheevolutionofaltruisticbehavioriscommonlycalledtheproblemofaltruism.Theproblemofaltruismisanexcellentillustrationofthepowerofgametheory.Thisisbecausetheveryexistenceofaltruismseems,atfirstglance,toberuledoutbyevolution.Afterall,anygeneticpropensitytobehavealtruisticallywould,bydefinition,resultinaloweringofthatindividual’sfitness.Therefore,weshouldexpectthatevolutionwouldrelentlesslyeliminatealtruismfromanypopulation.Itturnsout,however,thatnaturalselection’seffectonaltruismismuchmoresubtleandinteresting.Gametheorythrowsthesesubtlefeaturesofnaturalselectionintosharprelief.2.gametheoryandtheproblemofaltruismGametheoryisusedtoanalyzesituationsinwhichtwoormore1individualshaveconflictinginterests.Typically,gametheoristsdonotcareaboutmanyofthefeaturesofthesituationthatthelay-personwouldnormallythinkareimportant.Forexample,supposethatBettyisdecidingwhethertopaybackaloanthatAlgaveherlastweek.IfalaypersonwereaskedtopredictBetty’sbehavior,thelaypersonwouldprobablytakeintoconsiderationsuchfactsaswhetherBettyhaspromisedtopayitback,andwhethersheappearstobeanhonestandtrustworthyperson.Incontrast,agametheoristdoesnotnormallycareaboutsuchthings.Whenwesubjectasituationtoagame-theoreticanalysis,westripawaymostofthecontextualfeaturesofthesituation,focusingCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n306zacharyernstPlayerIIASPlayerIAB–C,B–C–C,BSB,–C0,0Figure16.1Theproblemofaltruismasagameonlyonafewkeyfacts.First,wewanttoknowwhatbehaviorsareavailabletotheindividuals.Inanalogywithordinarygamessuchascheckersorchess,werefertotheavailablebehaviorsasstrategiesandwerefertotheindividualsinvolvedasplayers.SoBettyhastwooptions–shecanpaybacktheloanornot.Accordingly,wesaythatherstrategysetconsistsofthetwostrategies‘payback’and‘donotpayback’.Thestrategiesselectedbytheplayerswillleadtoanoutcome;someoftheoutcomesmaybemorebeneficialtosomeplayersthantoothers.Wemeasurethedesirabilityofanoutcomebyitspayofftoeachplayer.Forinstance,ifBettyrepaystheloan,thenoneresultofherbehavioristhatshegivesAlsomemoney.Thus,ifweareinterestedonlyintheamountofmoneythatBettyandAlhaveattheendoftheday,thenBetty’spayoffwillbenegativeifsheplaysthestrategyofrepayingtheloan.Earlier,wedefinedaltruisticbehaviorasbehaviorthatraisesthefitnessofanotherindividualatsomecosttooneself.Ingame-theoreticterms,wesaythatthealtruistraisesthepayoffofanotherindividualwhileloweringherownpayoff.Letussaythatanindi-vidualhasachoiceastowhethertobestowabenefitBontoanotherindividualatsomecostCtoherself,andthatthepotentialbenefitisgreaterthanthecost–inotherwordsthatB>C.Iftwoindividuals–callthemPlayerIandPlayerII–aretointeractwitheachother,thenwemayrepresenttheirsituationinthesimplematrixshowninFigure16.1.There,theplayershaveachoicebetweenbehavingAltruistically(A)orSelfishly(S).IfaplayerchoosesstrategyA,thensheincursacosttoherselfof
C;thestrategySincursnocost.PlayingstrategyAbestowsabenefitofBtoone’spartner;playingSdoesnot.Thisgameisprobablythemostfamousoneingametheory.Itisusuallycalledthe‘Prisoner’sDilemma’,becausethefollowingstoryusuallygoesalongwithit.TwobankrobbershavebeenarrestedbyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology307PlayerIIDRPlayerID1,110,0R0,104,4Figure16.2ThePrisoner’sDilemma.Thenumbersrepresentthejailsentencesforeachplayer,solargernumbersareworsethepolice,butthepolicedonothaveenoughevidencetoconvictthemoftherobbery.Sothepoliceputthetwosuspectsindifferentcellswheretheycannottalktoeachother,andthepolicemakethefollowingspeechtoeachsuspect:Weknowyourobbedthebank,butwecannotconvictyouwiththeinformationthatwehave.Soifyouagreetoratoutyourpartner,andyourpartnerdoesnotratyouout,thenwewillletyougofreewhilewegiveyourpartnertenyearsinprison.Ontheotherhand,ifyoudonotratoutyourpartner,butyourpartnerratsyouout,thenyouwillgotoprisonfortenyearswhileyourpartnergoesfree.Ifyoubothrateachotherout,thenyouwillbothgetfouryearsinprison.Ifneitheroneofyouratstheotherout,thenwewillmakesurethatyoubothgotoprisonforoneyearontrumped-upcharges.WerepresentthedilemmafacedbytherobbersasinFigure16.2.Iftheprisonersarefairlysophisticated,thentheycouldreasonasfollows:Supposethatmyaccompliceratsmeout.Ifso,thenIambetteroffrattinghimout,sinceIwouldrathergetfouryearsinprisoninsteadoften.Ontheotherhand,supposethatmyaccomplicedoesnotratmeout.Inthatcase,Iamstillbetteroffrattinghimout,becauseIwillgetnotimeinprisonratherthanoneyear.Sonomatterwhatmyaccomplicedoes,Iambetteroffrattinghimout.Thetroubleisthatbothprisonersgothroughexactlythesamelineofreasoning.Soweexpecteachtowindupwithfouryearsinprisonasaresult.Butthisseemsparadoxical,foriftheycananticipategettingfouryearsinprison,whynotbothremainsilentandgetoneyearinstead?AquickcomparisonofthePrisoner’sDilemmaandtheproblemofaltruismrevealsthattheyarereallythesamesituation.TheonlydifferencebetweenthetwomatricesisthatwehaverepresentedCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n308zacharyernstpositivenumbersaspayoffsinthefirstfigure,andaspenaltiesinthesecond.Butclearly,thismakesnosubstantivedifference.Aswehaveseen,theprisonershaveaconvincinglineofrea-soningthatleadsthemtowardbehavingselfishly,ultimatelyto2theirowndetriment.ButifthePrisoner’sDilemmaisreallyjustarestatementoftheproblemofaltruism,thenwehaveapuzzle:howisitpossiblefornaturalselectiontoleadtotheevolutionofanyaltruistictendenciesatall?Forifwehaveasoundargumentinfavorofrattingoutone’saccompliceinthePrisoner’sDilemma,thenthesameargumentshouldfavorselfishnessintheproblemofaltruism.3.twogame-theoreticapproachesAsiswellappreciatedbyphilosophersofscience,thebestexplana-tionsfrequentlydisplayagreatdealofgenerality,applyingtoawide3varietyofdifferentcasesacrossmanydifferentcircumstances.Indeed,itisoneofthestrengthsofgametheorythatitshowsuswhatrelevantsimilaritiesexistacrossdifferentphenomena.Thus,gametheoryholdsouthopeofconstructinggeneralexplanationsthatapplytomanydiversephenomena.Game-theoreticexplanationsmaybedividedintotwocate-gories.Ontheonehand,wemighthopethatthereissomechar-acteristicofthestrategiesthatmakesthemmuchmorelikely.Ifso,thenwemaylargelyignorethequestionofhowtheplayssettleonthosestrategies.Ishallcallthisthestaticapproach.Ontheotherhand,wemightfocusonmodelingtheprocessbywhichagentsdeterminetheirstrategiesandlargelyignorethecharacteristicsofthestrategiesuponwhichtheysettle.Ishallcallthisthedynamicapproach.Theadvantagetothefirstmethodisthatvarioussituationswillfollowdifferentprocessesanddynamics.Thus,ifwecanconstructasatisfactoryexplanationforanobservedsetofbehaviorsthatignoresthoseprocesses,thentheexplanationmayapplytoawiderangeofothercases.Ofcourse,thissimultaneouslysuggeststhebiggestdrawbacktotheequilibriumapproach–itmayturnoutthatthereisnogeneralexplanation,andthatwehavenochoicebuttobedrawnintoananalysisofthoseevolutionaryprocesses.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology3093.1TheStaticApproachHistorically,thestaticapproachistheonethathasdrawnthemostattentionfrombiologists,philosophers,andeconomists.Indeed,ithasfoundapplicationnotonlyinbiologyandeconomics,butalsoinphilosophicaldiscussionsconcerningtheoriginsofcommunicationandcoordinatedconventions(forwhichtheclassicsourceisDavidLewis1969).Onewayintuitivelytomotivatethestaticapproachistoconsiderastrategicsituationfacedbytwoormoreplayersasasysteminwhichthegameisrepeatedanindefinitenumberoftimes.Some-times,theplayerswillsettleuponanoutcomeinwhichtheyhaveanincentivetochangestrategies;sometimestheoutcomewillbestable,leavingtheplayerswithnoincentivetochangetheirstrate-gies.Wesaythatsuchastableoutcomeisanequilibrium.ThissimpleandelegantideawaspioneeredineconomicsbyJohnNash(1950,1951).Nash’sinsightgaveanintuitivelycompellingandmathematicallyprecisecharacterizationofthesetsofstrategiesthatmakeupaso-calledNashequilibrium.AccordingtoNash,weshouldconsiderasetofstrategiestocomposeanequilibriumjustincasenosingleplayercoulddobetterthanhercurrentpayoff,giventhecurrentstrategiesplayedbytheotherindividuals.Inotherwords,aNashequilibriumexistswhennoplayerhasanincentiveunilaterallytoswitchstrategies.Inthisway,theNashequilibriumconceptisapredictivetoolthattellsusthat,allotherfactorsbeingequal,ifagameweretoberepeatedanindefinitenumberoftimes,andwerandomlyselectatimeatwhichweobservetheplayers’choices,thereisahighprobabilitythatwewillobserveasetofstrategiesinNashequilibrium.Nash’soriginaltreatmentwasconcernedwithbargainingsitua-tionsbetweenrationalagents;infact,inNash’soriginalconception,thepreequilibrium‘experimentation’phasewastobeconductedonlyhypothetically,inthemindsofrationalagentswhoweretobargainonlyonce.Thus,theveryterm‘equilibrium’wasonlyametaphorinNash’sintendedcontext,becausethereisnodynamicprocessthatcouldbeinequilibriuminthefirstplace.However,eveninamoreliteralcontext,inwhichthe‘experimentation’isguidedbytrialanderrororbyevolutionaryprocesses,weshouldstillexpecttoobservesetsofstrategiesinNashequilibrium.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n310zacharyernstItwasJohnMaynardSmithwhogaveabiologicalmotivationforacloselyrelatedapproach(1973,1974,1982).Inhisformulation,wemayexplaintheevolutionofabehaviorbyshowingthatitisstableinaslightlydifferentsense.Hissenseof‘stable’isthatthebehavior,onceithasbeenadoptedbyapopulation,cannotbeinvadedbyasmallnumberof‘mutants’whobehavedifferently.Moreprecisely,weconsiderapopulationwhosemembersexhibitaparticularkindofbehavior.Thatbehaviorwillinthepopulationentailthattheindividualshaveaparticularfitness.Thebehavioris‘uninvadable’,accordingtoMaynardSmith,ifasmallgroupofmutantswhobehavedifferentlycannothaveahigherfitnessiftheyweretointeractwiththatpopulation.Anystrategyhavingthisstabilitypropertyiscalled4an‘evolutionarilystablestrategy’,oran‘ESS’forshort.WemayillustratetheESSconceptusingthePrisoner’sDilemma.LetussupposethatapopulationofplayersfacesthePrisoner’sDilemma,andthateveryoneinthepopulationplaysDon’tRat.ThisisnotanESS,becauseamutantwhoplaysRatwillenjoyamuchhigherpayoff.Thus,thestrategyofplayingDon’tRatisinvadablebymutantswhoRat.Thus,MaynardSmith’sESSconceptpredictsthatpopulationswillplayRatinthePrisoner’sDilemma,andthattherefore(becauseitisreallythesamegame),theywillbehaveselfishlyratherthanaltruistically.TheobservationthatthestrategyDon’tRatisnotanESSispuzzlingbecausewestillhavetheobservationthataltruismdoesexistinnature.Soourmodelmustbeflawedorincompleteifitpredictsonlyselfishness;inotherwords,theremustbeimportantbiologicalfactsthatweareoverlooking.Fortunately,gametheoryprovidesaflexibleenoughframeworkforincludingotherbiologicalfacts.Butthisrequiresustoadopt(whatIhavecalled)thedynamicapproach.3.2TheDynamicApproachInordertointroducethedynamicapproach,considerthefollowingsimplegame,whichissometimescalled‘Hi-Lo’.InHi-Lo,thetwoplayershaveachoicebetweenplayingstrategyAandstrategyB.Theyreceiveapayoffonlyiftheyplaythesamestrategy,buttheybothreceiveamuchhigherpayoffiftheybothplayA.Forinstance,wecouldsupposethatthepayoffiftheybothplayAis10,whiletheirpayoffiftheybothplayBismerely1.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology311Inthisgame,itisobviousthatweshouldexpecttheplayerstoconvergeontheAoutcome.However,closerinspectionrevealsthatneithertheNashequilibriumconceptnortheESSconceptmakesthisprediction.ThisisbecausetheoutcomesinwhichtheybothplayAandtheoutcomeinwhichtheybothplayBareNashequilibriaandevolutionarilystablestrategies.Afterall,apopulationofBplayerscannotbeinvadedbyamutantAplayer,becausethemutantwillreceiveapayoffof0,whichislowerthanthepopulation’spayoffof1.Sowewouldliketohaveaprincipledtestthatwouldtellus(inthiscase)thattheplayersaremorelikelytoconvergeontheBstrategy.Criteriathateliminatesomeequilibriafromconsiderationarecommonlycalled‘equilibriumrefinements’,andthereisalargeliteratureonthesethatmaybefoundineconomics.However,asmostoftheserefinementsmakerecoursetothecharacteristicsofrationaldecisionmakers,theyarelargelyoutsidethescopeofour5considerationhere.Ifweareworkingwithinabiologicalcontext,weneedtorefocusourattentionontheevolutionaryprocessesthatyielddeterminatebehaviorsamongpopulationsofnonrationalagents.Toreturntoourearliercoordinationexample,itisimplausibletosupposethatmostevolutionaryprocesseswillequallyfavorcoordinationonAandcoordinationonB,giventheradicallyunequalpayoffsinthosetwocoordinatedstates.Rather,itseemsclearthatmostevolutionaryprocesseswilltendtofavorcoordinationonthemoreprofitablebehavior.Thusweareledtoconsidertheevolutionaryprocessitself.Soletussupposethatwehaveaverylargepopulationofnonrationalagentswhopairupperiodicallyandatrandomtoplaythiscoordi-nationgame.Tosimplifymatters,imaginethatinitially,thepopu-lationisdividedroughlyevenlybetweenA-playersandB-players.Sinceweareworkingwithinabiologicalcontext,weinterpretpayoffasfitness,orexpectednumberofoffspring.Asistheusualcustomwiththissortofexample,weassumethateachagent’sbehavior(thatis,thechoiceofstrategy)isdeterminedgenetically,sothattheoffspringofA-playerswilltendtobeA-players,andsimilarlyfortheoffspringofB-players.ItisclearthatthissimpleevolutionaryprocesswilldramaticallyfavortheevolutionofB-players,inagreementwithourexpectationsforthisexample,forsupposethatyouareanA-player.AbouthalftheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n312zacharyernsttime,youwillmeetanotherA-playerandreceiveasmallpayoff;theotherhalfofthetime,youwillmeetaB-playerandreceivenothing.ButifyouareaB-player,youwillmeetanotherB-playerabouthalfthetimeandreceiveamuchlargerpayoff(sinceapairofB-playerswillhaveahigherpayoffthanapairofA-players).Thus,onaverage,B-playerswillenjoyamuchhigherfitnessthanA-playersunderthisdynamic,asthefollowingsimpleequationsshow:expectedpayoffforA=.5(1)+.5(0)=.5expectedpayoffforB=.5(0)+.5(10)=5Sincepayoffistobeinterpretedastheexpectednumberofoffspring,therewillbemoreB-playersinthepopulationinthenextgeneration.ThisiswherethedynamicbeginstomagnifythepositiveeffectofbeingaB-playerandleadstothespreadofBthroughouttheentirepopulation,foriftherearemoreB-playersthanA-playersinthepopulationinthenextgeneration,thenA-playerswillreceiveevenlowerpayoffsthantheydidduringthefirstgeneration.Iftheyareintheminority,thentheyarelesslikelytomeeteachotherandthereforearelesslikelytoreceiveanypayoffatall.TheoppositeisclearlytrueofthosewhoarefortunateenoughtobebornwithatendencytoplayB.Theywillbemorelikelytomeeteachotherinthenextgenerationandwillthereforebemorelikelytoenjoyhighpayoffs.Inthisway,thespreadoftheBstrategythroughoutthepopulationwillacceleratefromonegenerationtothenext.Ifthereisnootherprocesstostopit,weshouldexpectthatwithinrelativelyfewgenerations,thepopulationwouldbecomposedalmostentirelyofindividualswhoplaystrategyB.However,itisimportanttonotethattheseconsiderationsdonotimplythattheBstrategywillalwaysbeobserved,forsupposethatchanceeventscausethepopulationtostartinastateinwhich95percentoftheindividualsplayA.ThenanindividualwilltendtoencounteranA-player95percentofthetime,andwecanthereforecalculatetheaveragepayoffforeachtypeofplayerasfollows:expectedpayoffforA=.95(1)+.05(0)=.95expectedpayoffforB=.95(0)+.05(10)=.5Thus,ifthepopulationhappenstobegininastatethatisheavilyskewedtowardtheAstrategy,thenthepopulationwilltendtomovetoastateinwhicheveryoneplaysthatstrategy.SowhattheseCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology313considerationsreallysayisthatmostofthetimeweshouldobservethepopulationplayingstrategyB,butthatasmallpercentageofthetime,populationswillplayA.Thissimpleexampleisaninformalillustrationofaformalmodelcalledthereplicatordynamics,whichisdueoriginallytoTaylorandJonker(1978),andwhichhasplayedadominantroleinso-called6evolutionarygametheory.IncontrasttotheequilibriumapproachinventedbyNash,thedynamicapproachofevolutionarygametheoryisconcernedwiththeprocessbywhichapopulationofagentssettlesuponabehavior.Inthereplicatordynamics,whichisthesimplestandmostcommonofsuchmodels,thatprocessistakentobeevolutionandnaturalselection,wherestrategiescompetewitheachotherasreplicators.Incontrasttotraditionalgame-theoreticmodelsinwhichpayoffsareinterpretedasmoneyorwelfare,thereplicatordynamicsinterpretspayoffsasfitness.Thus,individualswhosestrategiestendtoyieldhigherpayoffsinthepopulationwillenjoyhigherfitnessandwilltherebytendtohaveagreaternumberofoffspringinsubsequentgenerations.Soifweweretolookatapopulationofagentsovermanygenerations,weshouldexpectthemostsuccessfulstrategieseventuallytopredominateinthepopu-lation.Butwhatoftheproblemofaltruism?Unfortunately,itturnsoutthattheselfishstrategyisfavoredbythereplicatordynamics.Toseethis,wesimplycalculatetheexpectedpayoffsofaltruismandselfishnessasfollows.Letussupposethatthepopulationisrandomlydividedbetweenaltruisticandselfishtypes.CalltheproportionofthepopulationthatisaltruisticPA,andtheproportionthatisselfishPS.BecausePAþPS¼1,wecanreplacePSwith(1
PA).UsingthepayoffmatrixfromFigure16.1,wemaycalculatetheexpectedpayoffofaltruistsandselfishtypes:payofftoaltruists¼PAðB
CÞþð1
PAÞð
CÞpayofftoselfishtype¼PAðBÞþð1
PAÞð0Þwhichsimplifiestopayofftoaltruists¼PAðBÞ
Cð1þ2PAÞpayofftoselfishtypes¼PAðBÞClearly,itisbettertobeselfish–afterall,selfishtypesgetallthebenefitsofinteractingwithaltruists,butneverincuranycosts.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n314zacharyernstThus,thereplicatordynamicspredictsthatpopulationswillmoveinexorablytowardastateinwhicheveryoneisselfish.4.additionstogame-theoreticexplanationsSoitbeginstoappearthattheproblemofaltruismisaverythornyone,andthatthegame-theoretictoolssuggestonlywhatweknewalready–namely,thatweshouldexpectindividualstobeselfish.Specifically,wefindthatselfishnessisthepredictedbehavioronboththestaticanddynamicapproaches,forapopulationinwhicheveryoneisselfishismorestablethanonewhichhasaltruists,andthereplicatordynamicsshowsusthatthesimplestpopulationdynamicfavorsthespreadofselfishnessinanymixedpopulation.However,wemustreturntotheempiricalfactthatmanyspeciesdoexhibitaltruisticbehavior,sotheremustbemoretothestory.Intheremainderofthisessay,Ishalldiscusstwoadditionalfeaturesthathavebeenprominentinexplainingtheevolutionofaltruism.ThenIshalloffersomespeculationsaboutpromisingareasforfutureresearch.4.1IterationOneofthemostinfluentialproposalsforexplainingtheevolutionofaltruismreliesuponthecommonsenseobservationthatmanyindi-vidualsinteractwitheachotherrepeatedly.Asweallknow,itiseasytoacquireareputationforinteractinginaparticularway,andonecancarrythatreputationtofutureencounters.Thus,thesuggestionisthatweshouldexplaintheevolutionofaltruismbyconsideringiteratedinteractionsinsteadofthesimpler‘one-shot’interactions.SoweneedtodifferentiatebetweentwodifferentversionsofthePrisoner’sDilemma.Ontheonehand,wehavethe‘one-shot’Prisoner’sDilemmainwhichindividualspairup,playthePrisoner’sDilemmaonce,andneverinteractagain.Ontheotherhand,wehavetheso-calledIteratedPrisoner’sDilemma,inwhichindividualspairupandplaythePrisoner’sDilemmarepeatedlywiththesamepartners.Itisimportanttonotethattheone-shotPrisoner’sDilemmaandtheIteratedPrisoner’sDilemmaareverydifferentgames.InCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology315PlayerIIASPlayerIAR,RS,TST,SP,PFigure16.3ThePrisoner’sDilemmaingeneralformparticular,theIteratedPrisoner’sDilemmahasamuchrichersetofavailablestrategies.Forexample,onecanplaytherelativelysimplestrategies‘AlwaysBeAltruistic’(henceforth,AA)or‘AlwaysBeSelfish’(AS).Butonecanalsoplayastrategythattakesone’spart-ner’spastbehaviorintoaccount.Forinstance,onecanplaytheappropriatelynamed‘Tit-for-Tat’strategy(TFT).WhenplayingTFT,onebeginsbybehavingaltruisticallytowardone’sopponent.How-ever,theTFTstrategyretaliatesagainstaselfishopponentbybehavingselfishlyontheroundaftertheopponent’sselfishbehav-ior.Conversely,theTFTstrategyrewardsanaltruisticopponentbybehavingaltruisticallyintheroundafterthatbehavior.Letusana-lyzetheproblemofaltruismusingthethreepreviousstrategies.Forsimplicity,IwillfollowthestandardpracticeofrewritingthematrixinFigure16.1usingtheseshortcuts:B¼TðTemptationtodefactÞB
C¼RðRewardformutualcooperationÞ0¼PðPunishmentformutualdefectionÞC¼SðSucker0spayoffÞAsFigures16.1and16.2makeclear,weshouldrequirethatT>R>P>Ssothatthebestpossibleresultforaplayeristobeselfishagainstanaltruist;thenextbestistobealtruisticwith7anotheraltruist;andsoon.Foreasyreference,IplaceanequivalentversionofFigure16.1usingthenewabbreviationsasFigure16.3.ConsidertwoplayerswhopairuptoplaytheIteratedPrisoner’sDilemmafortenroundsusingthepayoffmatrixfromFigure16.3.SupposethattheplayersbothplaytheselfishstrategyAS.Thenineveryround,bothplayersdefect,eachreceivingapayoffofP.Thus,theiroverallpayofffortenroundswillbe10P.Ontheotherhand,supposethataTFTplayerinteractstentimeswithaselfishASplayer.Inthefirstround,theTFTplayerwillbealtruisticandtheselfishplayerwillbeselfish,sothepayoffstothetwoplayerswillbeCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n316zacharyernstTFTAAASTFT10R,10R10R,10RS+9P,T+9PAA10R,10R10R,10R10S,10TAST+9P,S+9P10T,10S10P,10PFigure16.4PayoffsfortheIteratedPrisoner’sDilemmaSandT,respectively.Butafterthatpoint,theTFTplayerwillretaliateagainsttheselfishASplayerbyplayingtheselfishstrategy.Thus,fortheremainingninerounds,bothplayerswillreceivethepayoffofP.Sooverthecourseoftenrounds,theTFTplayer’spayoffwillbeSþ9P,andthepayofftotheASplayerwillbeTþ9P.Soagain,themoreselfishplayerwilldobetter(sinceT>S).Obviously,whenahighlyaltruisticAAplayergoesupagainstaselfishASplayer,thealtruistwillbeataseveredisadvantage.TheAAstrategygivesapayoffof10S,whiletheASstrategyreceivesthevastlyhigherpayoffof10T.Ontheotherendofthespectrum,whentwoAAplayers,twoTFTplayers,oranAA/TFTpairinteractwitheachother,noplayeriseverselfish,fortheTFTstrategyonly8retaliatesagainstselfishbehavior.Soineachofthesecases,thepayofftobothplayersis10R.KeepinginmindthatT>R>P>S,itisclearthatinanyiteratedinteraction,themoreselfishstrategydoesbetterthanitsopponent.WemaysummarizethepayoffsfromthelasttwoparagraphsasinFigure16.4.Butsurprisingly,thefactthatASdoesbetterineveryinteractiondoesnotimplythatitdoesbestoverthelongrun.Toseethis,wehavetoadoptadynamicperspectiveontheevolutionaryprocess.Imaginethatwebeginwithapopulationthatisdividedroughlyevenlyamongthosethreestrategies,andthattheindivid-ualspairuprandomlytoplaytheiteratedgamefortenrounds.Asinthereplicatordynamics,weassumethattheplayers’fitnessdependsontheirtotalpayoffforthosetenrounds.Intuitively,theproblemfacedbyanASstrategyoverthelongrunisthatitssuccessdependsuponvictimizingaltruisticAAplayersinthepopulation.However,theASplayersdramaticallylowerthefitnessofthoseAAplayers–forrecallthatinanAS/AAinteraction,theAAplayergetsthelowestpossiblepayoffof10S.Sooverthelongrun,theASplayerswilltendtodrivetheAAplayerstoextinction.OncethosealtruisticAAplayershavebeendrivenoutofthepopu-lation,theselfishAStypesnolongerhavethosehighlyprofitableCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology317interactions.Sointhisway,themostselfishstrategyisself-defeatingoverthelongrun.Ontheotherhand,thisisnotthecaseforTFTorAAplayers.Theybenefitfrominteractionswithother‘nice’strategies.Butmostimportantly,theydonotharmthoseplayersuponwhomtheydependforhighpayoffs.SoincontrasttotheselfishASplayers,thenicerTFTandAAstrategieshavethepotentialtodowelloverthelongrun,byhelpingtheirpartnerstoremaininthepopulation.Infact,theseinformalconsiderationshavebeenverifiedinafamouscomputertournamentorganizedbyRobertAxelrodattheUniversityofMichigan(1984,1997).HeinvitedpeopletosubmitstrategiesfortheIteratedPrisoner’sDilemmaandprogrammedacomputertoplayeachofthestrategiesagainsteveryotherstrategyinaround-robincompetition.Althoughavarietyofhighlysophis-ticatedstrategiesweresubmitted,thestrategythatyieldedthehighestpayoffinthetournamentwasnoneotherthanTFT.Axel-rod’sanalysisoftheresultsparallelstheprecedingconsiderations.Inorderforastrategytodowelloverthelongrun,itmustthriveonceithasspreadthroughthepopulation.Andinorderforthattohappen,itmustdowellwhenitplaysagainstitself.Therefore,strategiessuchasAAandTFThaveadistinctadvantage–theyyieldtherelativelyhighpayoffofRineachroundwhentheyplayagainstthemselves.ButselfishstrategiesaremorelikelytoyieldthelowerpayoffofPinanygivenround,becausetheytendtofallintoapatternofmutualdefection.However,strategiessuchasTFTaresuperiortomore‘naive’strategieslikeAAinthefollowingrespect.TheAAstrategycanbevictimizedbyselfishstrategiestoamuchgreaterextentthanTFT,forwhenaTFTplayerinteractswithaselfishtype,itbeginstobehaveselfishlyitself.Thus,itguaranteesthatitwillreceivethelowestpayoffSlessoften;ontheotherhand,a‘naive’strategylikeAAmayreceiveapayoffofSrepeatedlyinanygiveninteraction.Sowelearnafewimportantlessonswhenweconsiderrepeatedinteractions.Inordertobesuccessful,astrategymustexhibittwokeyfeatures.First,thestrategymustdowellwhenitplaysagainstitself–forifitdoesnot,thenitwillbeunlikelytopredominateinthepopulation.Second,thestrategymustprotectitselfagainsthighlyselfishoropportunisticstrategiesbyretaliating.Further-more,theadvantageofastrategylikeTFTisincreasedifthenumberCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n318zacharyernstofrepeatedinteractionsishigher.Forexample,iftherewere1,000repeatedinteractionsineachround,thentherearemoreopportu-nitiestoreceivethe‘reward’payoffofR.Similarly,thehighlyaltruisticAAplayerswillbedriventoextinctionfaster,followedbythedemiseoftheselfishAStypeswhodependonthepresenceofnaivealtruistsfortheirsurvival.Finally,itappearsasthoughwehavemadesomeprogressontheproblemofaltruism.Byconsideringtheeffectofrepeatedinterac-tionsontheevolutionofaltruisticbehavior,wecanmakesomedefinitepredictionsthatcanbeconfirmedordisconfirmedinabiologicalcontext.Letusgothroughsomeoftheselessonsindividually.First,theprecedingconsiderationssuggestthatwearemorelikelytoobservealtruisticbehaviorwhenindividualshavetheopportunitytointeractwitheachotherrepeatedly.Inabiologicalcontext,thisimpliesthataltruismwillbeobservedmoreofteninspeciesinwhichindividualslivetogetherinsmallgroupsoveralongperiod.Andindeed,thisisconfirmedbyobservation.Consider,forexample,ourevolutionarycousinsamongthegreatapes.Chimpanzees,bonobos,andgorillaseachliveinrelativelysmallgroupsovermuchoftheirlives.Moreover,ineachspecies,onegenderleavesthenatalgroupuponreachingsexualmaturityinamigrationpatternknownasphilopatry.Hence,thegenderthatremainsinthenatalgroupwillhaveincreasedopportunitiesforrepeatedinteraction.Thus,weexpecttofindagreaterdegreeofaltruismamongthegenderthatremainsinthenatalgroup.Indeed,thisisthecase.Ifthefemalesremaininthenatalgroupafterreachingsexualmaturity,thenanthropologistsrefertothatspeciesas‘femalebonded’.Similarly,speciesare‘malebonded’ifitisthemaleswhoremaininthenatalgroup.Itisacommonplaceobservationthatinfemalebondedspecies,cooperationandaltruismaremuchmorelikelytobeobservedamongthefemales,andcon-verselyifthespeciesismalebonded.Second,thegame-theoreticanalysessuggestthatweareunlikelytoobserve‘pure’altruism–instead,weshouldobserveamore‘restricted’formofaltruismsuchasTit-for-Tat.Successfulindivi-dualswillretaliateagainstselfishindividuals,soweshouldexpectan‘ethic’amongvariousspeciesaccordingtowhichselfishindi-vidualsarepunishedorisolatedfromtherestofthegroup.AmongCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology319nonhumanprimates,thisbehavioriswelldocumented.Inthemostfamousofsuchbehaviors,individualhunterswhosuccessfullyobtainmeatwillshareitwiththerestoftheirgroupinanovertlyaltruisticgesture.Butsuccessfulhunterswhofailtoshareareostracizedfromthegroupandgenerallydonotbenefitfromthealtruismofothers.Third,weexpectthatthereshouldbestrongeffectsofreputationwithinasocialgroup,forifsuccessfulstrategieswillbealtruistictowardaltruists,thenitbecomesimportanttoestablishareputationforbeingaltruistic.Incognitivelysophisticatedspecies,wewouldexpecttoseestrongeffectsofreputationwithinasocialgroup.Amonghumans,thiseffecthasbeendocumentedbyexperimentaleconomists,whohaveperformedexperimentsinwhichindividualsareobservedafterbeinggiventheopportunitytoestablisharepu-tation(CamererandThaler1995,FreyandBohnet1980,Thaler1988).Perhapsmoresurprisingly,itappearsthatthissameeffectoccursevenamongguppies.Experimentshaveshownthatguppiesprefertoassociatewithotherguppiesthatexhibitthealtruisticbehaviorofinspectingpotentialpredators(DugatkinandAlfieri1991a,b,surveyedinSoberandWilson1998).Thus,wemayhaveaformof‘reputation’evenwithinspeciesthathaveseverelylimitedcognitivepowers.4.2CorrelationAnotherhighlyinfluentialconsiderationforgame-theoreticanal-ysesistheeffectofcorrelationuponapopulation.By‘correlation’,wemeanatendencyforindividualswithsimilarstrategiestointeractwitheachother.Appliedtotheproblemofaltruism,anyassortativemechanismthatbringsaboutcorrelationwillincreasetheprobabilitythatanaltruistwillinteractwithanotheraltruist,andthatselfishtypeswilltendtomeeteachother.Correlationiscloselyrelatedtosomeoftheconsiderationsthatwehavemadeearlier.Foraswehaveseen,altruistswillbenefitfrominteractingwitheachother,whileselfishindividualsareharmedwhentheyinteractwitheachother.Theeasiestwaytoseetheeffectofcorrelationupontheevolutionaryaltruismistoconsideranextremecaseofperfectcorrelation.SupposethatindividualsinapopulationaresortedperfectlyaccordingtowhethertheyareCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n320zacharyernstaltruisticorselfish.Insuchacase,thealtruistswillalwaysreceiveapayoffofR,whiletheselfishtypeswillalwaysreceivethelowerpayoffofP.Thus,iftherewereperfectcorrelation,thenanindividualwouldbebetteroffasanaltruist.Ofcourse,correlationwillrarelybeperfect.Therewillalwaysbesomepositiveprobabilitythataselfishtypewillinteractwithanaltruist.Butcorrelationneednotbeperfectinorderforaltruismtohavetheadvantage.GivenanyvaluesofT,R,P,andS,onecaneasilycomputethedegreeofcorrelationthatisrequiredifaltruismistowinoutoverselfishness.Wewillomitthatcomputation9here.Butitturnsoutthatthereareavarietyofwell-knownmechanismsthatincreasetheprobabilitythatlikeinteractswithlike.Perhapsthemostimportantoftheseissimplythetendencyofindividualstointeractwithgeneticrelatives,forifgeneticrelativestendtobeborninthesamegeographicarea,andgenesplayaroleindeterminingbehavior,thenindividualswilltendtomeetsimilarlybehavingindividualstothedegreetowhichtheirinteractionsare‘local’.Modelsalongtheselineshavebeenexploredinabiologicalcon-textbySewallWright(1943,1945,1969).InWright’smodel,apopulation’stendencytoremainwithinaparticulargeographicareaisits‘viscosity’.Highpopulationviscosityentailsthatthemembersofapopulationareverylikelytoremaininasmallarea,whilepopulationswithlowerviscositywilltendtowanderacrossalargerregion.Inphilosophicalwork,theeffectsoflocalinteractionhavebeenexploredbyBrianSkyrms(1999,2004)andJ.McKenzieAlexander(2000),whohaveshownthattheeffectsoflocalinteractionmaybesurprisinglypowerful.Thus,weareledtoanotherspecificpredictionthatemergesfromthegame-theoreticanalysis.Populationswithhighviscosityshouldbeexpectedtoexhibithigherdegreesofcorrelation,andhence,morealtruisticbehavior.Indeed,preciselythisphenom-enonisobservedamongnonhumanprimates.Amongthegreatapes,thespecieswiththelowestpopulationviscosityistheoran-gutan,anditistheorangutanthatexhibitstheleastaltruisticbehavior.Ontheotherendofthespectrum,chimpanzeesandbonoboshaveveryhighpopulationviscosity,andtheyexhibitagreatdealofaltruisticbehavior.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology3215.futuredirectionsSofar,Ihopetohavemotivatedtheconclusionthatgametheoryisavaluabletoolforgaininginsightintotheevolutionaryoriginsofaltruisticbehavior.Inspiteofitsabstractnessandgreatgenerality,ityieldsaseriesofspecificpredictionsthatcanbeconfirmedordisconfirmedbyempiricalobservation.Thus,game-theoreticmod-elsarenot‘just-sostories’–theyaretestablemodelscapableofyieldinginsightintotheevolutionaryoriginsofpuzzlingbehaviors.Butasprovocativeastheprecedingresultsare,however,Ibelievethatthemostfascinatingworkinthisareaisyettobedone.Asisillustratedbytheprecedingdiscussionofiterationandcorrelation,oneofthemostvaluablefeaturesofgametheoryisthatadditionalrelevantfeaturesoftheinteractionmaybeintegratedintothegame-theoreticmodels.Forexample,althoughgametheorydoesnotnecessarilytakeiterationandcorrelationintoaccount,itisaverysimplemattertobuildthosefeaturesintothemodels.Weshouldbelookingforotherfeaturesthatarelikelytohaveplayedanimportantroleintheevolutionofaltruism,aswellasotherimportantsocialbehaviors.Toconcludethissurvey,Iwillbrieflyindicatetwoareaswhereimportantworkremainstobedone.First,althoughcurrentworkonlocalinteractionishighlypro-vocative,thereisawealthofrelevantpopulationstructuresthatshouldbestudiedfurther.Forexample,anthropologistsareveryconfidentthatourevolutionaryancestorslivedinaspecifictypeofmetapopulationstructureinwhichlargepopulationsweredividedintosmallbandsofgeneticrelatives(PuseyandPacker1986).Membersofthosebandswouldtendtomigratetoothernearbybands,andtheirmigrationwasheavilybiasedaccordingtotheirageandgender(Cheney1983,1986).Onthefaceofit,thispopulationstructurecontainsallofthefeaturesthatwouldtendtofavortheevolutionofaltruism(Ernst2001).Butmuchmoreworkneedstobedoneontheevolutionoflargemetapopulationssuchasthese.Second,interactionsamongthenonhumanprimatesarehighlystructuredinamannerthatismuchricherthanthesimplecorrela-tionmodelsthathavebeendiscussed.Inmostnonhumanprimatespecies,thereisapowerfuldominancehierarchythatdictatesthetypeofinteractionsthatareobserved(Colvin1983,Essock-VitaleandSeyfarth1986).IndividualsthatarehighlyrankedinthehierarchyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n322zacharyernsttendtointeractprimarilywithotherhigh-rankingindividuals,andtherolesthateachrankplaysinthegroupareimportantlydifferent.Anthropologistsarewellawareofthefactthatrankplaysacrucialroleindeterminingthestructureofprimatesocieties.However,there10hasbeenlittleworkonthisfromagame-theoreticperspective.Althoughitwouldbeasignificantcomplicationtothegame-theoreticmodels,wecouldassumethateachplayerinthepopula-tionhasanassociatedrankthataffectsthepayoffstructureofaninteraction.Itisgenerallybelievedthattheexistenceofdominancehierarchiesexplainswhyprimatesformcoalitionsandcooperatewithinthosecoalitionstodefendthemselvesagainsthigh-rankingindividuals.Ithasbeensuggestedthatthiscoalitionformationbehaviorisqualitativelysimilartothesophisticatedpoliticsthatareobservedinmodernhumansocieties.Asaworkinghypothesis,itmaybereasonabletosupposethatcoalitionformation(aswellasitsassociatedcooperativeandagonisticbehaviors)emergedasadirectresultofdifferencesinrankbetweenindividuals.Agame-theoreticmodelthatshowedthisanalytically–orrefutedthisconjecture–wouldconstituteanimportantadvance.notes1.Thisisso-callednoncooperativegametheory.Thereisalso‘coopera-tive’gametheory,inwhichtheplayersareabletomakebindingagreementsbeforetheyplaytheirstrategies.However,Ishallignorecooperativegametheoryinthisessay.2.Itisnotunusualtocomeacrossargumentsthattheirlineofreasoningisunsound.ButseeBinmore(1998)foracompellingsurveyandrefutationofsucharguments.3.Kitcher(1988)isthecanonicalexpressionofthis‘generalist’viewofscientificexplanation.4.ForathoroughreviewoftheESSconcept,seeW.G.S.Hines(1987).5.See,forinstance,BergstromandStark(1993);Bernheim(1984);CamererandThaler(1995);Gu¨th,Schmittberger,andSchwarze(1982);Krepps(1990);McKelveyandPalfrey(1992);Ortona(1991);Rabin(1993);Selten(1975);andSopher(1993).6.AnexcellentformalintroductiontoevolutionarygametheoryisWeibull(1995),andabriefinformalintroductionmaybefoundinSoberandD.S.Wilson(2000).Forthosewhoareinterestedinasurveyfromaneconomicperspectivethatcombinestechnicalsophisticationwiththoroughinformalmotivation,IsuggestLarrySamuelson(1997).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nGameTheoryinEvolutionaryBiology3237.Technically,weshouldalsorequirethat(T
S)<2R,forifthatconditionisviolated,thenindividualswhoplayanIteratedPrisoner’sDilemmadobestbytakingturns‘exploiting’eachother.However,Iwillnotbeconcernedwiththisconditioninwhatfollows.8.Axelrod(1984)referstoastrategywiththispropertyas‘nice’.9.SeeSkyrms(1994)foranexcellentdiscussionofcorrelationinsimplegamesfromtheperspectiveofdecisiontheory,andSober(2000)forastraightforwarddemonstrationofhowthecostofaltruismisrelatedtothelevelofcoordinationrequiredforaltruismtoevolve.10.Kitcher(1999)alsomakesasuggestionalongtheselines,althoughhisproposedmodelbearslittleresemblancetotheobservedbehaviorofprimates.YashaRohwer(forthcoming)hasalsoarguedconvincinglythatrankmayplayanimportantroleintheevolutionofso-calledaltruisticpunishment,inwhichoneindividualpunishesanotheratacosttothepunisher,andtothebenefitoftherestofthegroup.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\njanemaienschein17WhatIsanEmbryoandHowDoWeKnow?Becauseofrecentpublicexcitementaboutcloningandembryonicstemcellresearch,morepeoplethanjustdevelopmentalbiologistsarebusilytalkingaboutembryos.Humanembryosarecentralplayersinproposedlegislationatstate,federal,andinternationallevels.Butwhatismeantbyanembryo?Rarelyisthetermdefinedordefinedclearly.Yetthetermisusedinquitedifferentwaysandhasevolvedovertime.Howhavemeaningschanged,andforwhatreasons?Whatistherelationshipbetweenpublicandscientificunderstandingsofembryos?Here,thefocusismostdirectlyonevolvingunder-standingsofthebiologicalembryo,includingrecentshiftingpublicmeanings.Ineachcase,bothmetaphysicalandepistemologicalconsiderationsareimportant.Yetonlyaftertheemergenceofinvitrofertilizationdidtheembryobecomeanobjectofsignificantethicalconcern,andonlywithcloningandhumanembryonicstemstemcellresearchwasitwidelyseenasanobjectofsocialconcern.Thisessayconsidersthechangingunderstandingsofembryos.Sinceatleast1771,withtheappearanceofthefirsteditionoftheEncyclopediaBritannica,theembryohasbeenseenastheearliestandundifferentiatedstageofanindividualorganismsdevelop-ment.Theembryonicstagewasclearlyseparatedfromthefetalstage,withthefirstgivingwaytothesecondasformgraduallyemergedfromunformedmatter.Specifically,theembriowasinphysiology,thefirstrudimentsofananimalinthewomb,beforetheseveralmembersaredistinctlyformed;afterwhichperioditisdenominatedafoetus.Thefoetusdenotesthechildwhileitiscontainedinthemotherswomb,butparticularlyafteritisformed,tilwhichtimeitismoreproperlycalledembrio(Encyclopedia324CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?325Britannica1771).Intheeighteenthcentury,thechangewastakenasoccurringatquickeningorafter,whiletodaytheshiftfromembryotofetusisdefinedforhumansasoccurringatroughlyeightweeks.Clearly,theembryohaslongbeenseenasunformed,asundif-ferentiated,and(followingAristotle)ashavingthepotentialtobecomeanindividualoftheappropriatetypebutasnotyethavingbeenactualizedassuch.TheOxfordEnglishDictionaryoffersasimilarpicture.EarlyusagesofthetermintheseventeenthcenturyemphasizedtheEmbryoasAthinginitsrudimentarystageorfirstbeginning;agerm;thatwhichisstillinideaasopposedtowhathasbecomeactualinfact.Itremainsinanundevelopedstage,thatistobe,presumablyfromtheLatinEmbryonwithasuggestionofswellingwithin.Thepastcenturyofembryologytextbookshascontinuedtopro-videsimilarinterpretations,withemphasisonthecomingintobeingandgradualemergenceofformthroughtheprocesscalledepigenesis.Evenwhenembryologyhadbeguntobecalleddevelop-mentalbiology,andevenasthepresumedefficacyofgeneticinheritancebegantooverwhelmthepreviouspresumedcausalforceofepigeneticemergence,the1961Britannicacapturedatypicalunderstandingoftheembryo.TheEmbryoentrywaswrittenbyAuteRichards,anemeritusUniversityofOklahomazoologistwhohadbeendirectoroftheSchoolofAppliedBiologyandhadwrittenthewidelyused1931OutlineofComparativeEmbryology.Heportrayedtheembryoasbeginningwiththebiologicalactionoffertilizationandexistingthroughtheprocessofcelldivision,throughthecleavageofonefertilizedeggcellintomany.Theearlydevelopmentalprocessesoccurwithlargelyundifferentiatedcells,andonlygraduallydotheybecomeseparatedhistologicaltypes.Richardssummarizedthatitisnotuntilthesehistologicalchangesareaccomplishedthattheyoungembryoisreadytofunctionfullyandtotakeanindependentplaceintheworld(Britannica1961,Embryo)Ofcourse,somespeciesmovefromtheembryonicformthroughafetalorlarvalstage,buttheearliestembryonicformofallspeciesconsistsofundifferentiatedcellsandisunformed.Thiswastheepigeneticviewthattheembryoisthestagewhenformemergesgraduallyfromtheunformedmatter,whereunformedmeanslackinginorganicdifferentiationandwithoutthebodypartsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n326janemaienscheinandsystemsthatwillariselater.Epigenesiswasthedominantinterpretationoftheembryointothetwentiethcentury.Thealter-nativewaspreformation,withformpresumedtobealreadypresentfromthebeginning(seeMaienschein2003).Todocumenttheshiftsinepigeneticandpreformationistthinking,andtounderstandthechangingpatternsofdebate,itisusefultofocusonasequenceofsixselectedhistoricalepisodes.Underlyingmetaphysicalquestionsaboutthenatureoflife,organisms,andpartsprovidedonefocus,whileepistemologicalquestionsabouthowbesttounderstandtheemergingorganismprovidedanother.Whatwasthoughttobeatissueshiftedovertimeandindifferentcontextsinwaysthatareinstructiveforourunderstandingofcurrentdebatesaboutembryos.historicalinterpretationsOfcourse,therearemanywaystodivideupthemillenniastartingwithAristotle,butthefollowingselectedslicescapturetherangeofshiftingcentralissues.Eachepisoderaisesnewquestionsandintro-ducesnewrelevantfactors,buteachcasealsorevealsinstructivedecisionsaboutwhatisbeingstudied,howtodothestudying,andwhatrelevantfactorsshouldbebroughttobearininterpretingtheresults.keyperiodsinunderstandingembryosTheHypotheticalEmbryoremainedlargelyinvisibleandamatterfortheoreticalinterpretationfromAristotletotheEnlightenment.Eighteenth-centurydebateslaidoutthetraditionsofpreformation-ismandepigenesisthathavecontinued.ThePhysicalEmbryoofthemidnineteenthtoearlytwentiethcenturyintroducedcomparativestudyofembryosinmanyspeciestodescribethedetailsoforganismalchange.Whentheworkwasdoneinthecontextofevolutionarytheoryitinfluencedtheinter-pretationsofdevelopmentalstagesinimportantways.TheBiologicalEmbryogainedattentioninthe1920s30s,theembryologicalgoldenage,withemphasisontheorganizerandprocessesandcausesofdifferentiation.TheInheritedEmbryoofthe1950s60s,withfrogcloningandnucleartransfer,appearedwithanenthusiasmforgeneticismandeventuallygenomicism,andforreductionistmethodologies.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?327TheVisibleHumanEmbryostartedinthe1970swithinvitrofertilization(IVF)thattookembryosoutofthewomb,withNilssonswidelypublishedphotographsoffetusesthatcame(inac-curately)torepresentembryos,andwithuseofotherimagingtechniques,allinthecontextofabortionpolitics.TheConstructedEmbryoarrivedwithgeneticrecombination,cloning,andstemcellresearchthathaveallowedresearcherstoconstruct,deconstruct,andreconstructembryos.Becauseofthefearsandpromises,theembryobecomesapublicaswellasabio-logicalobject.Wemightalsopointtoaseventhperiod,oftheComputedEmbryo,withanemphasisoncollecteddatamanagementandinformatics,butthatraisesdifferentissuesandthereforewillnotbeconsideredinfurtherdetailhere.TheHypotheticalEmbryoTheHypotheticalperiod,drawingmoreontheoryandinferencethanonobservation,providesimportantbackgroundabouttheinterpretationsthatdominatedthinkingformorethantwomillennia.Aristotleoutlinedanepigenetichypothesisforembryonicdevelop-mentthatremainedtheonlyseriousinterpretationuntiltheeigh-teenthcentury.AccordingtoAristotle,embryodevelopmentwaspartofthenaturalprocessesofgenerationandcorruption.Genera-tionofanimals,throughsexualreproduction,involvescombiningfluids(orsemen)frombothparents.Thisminglingofmaleandfemalefluidsprovidesthematerialandthemotiveforcefordevelopment.Morespecifically,thefemalecontributesthematerialcausethatresidesinthemenstrualblood,andafterthedischargeisoverandmostofithaspassedoff,thenwhatremainsbeginstotakeshapeasafetus.Thismenstrualbloodisnotpure,however,andissimplythatoutofwhichitgenerates.Thematerialmustbeacteduponbythemalefluid,whichprovidestheformalcauseandinitiatestheefficientcauseforthedevelopmentthatfollows.Theformalandefficientcausesthereforebothactthroughthejoiningofthemaleandfemalefluids.Onlythencanthefinalcauseserveasthetelosforthelivingorganism.Aristotlesfourcausestogetherbringaboutgenerationofeachindividualorganism.ThusthingsarealiveinCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n328janemaienscheinvirtueofhavinginthemashareofthemaleandofthefemale.Themaleandfemaleserveastheprinciplesofgeneration(Aristotle1979,99,111,133,129).Aristotleurgedthattheformmustbeguidedbyinternalandnotoutsidecauses.Fromtheoutset,anindividualliferequiresasoulthatguidesthegradualunfoldingofform(orepigeneticpro-cess)fromtheunformedmatter.Thissoulconsistingofvegetativesoulforalllivingbeings,pluslocomotorysoulforallambulatorybeingscapableofpickingthemselvesupandmovingaround,plusrationalsoulforhumansalonesinceonlyhumanshavethepowertoreasongivesthepotentialtobecomeactualizedasanindividualorganism.Aristotledidnotpictureanembryoinoursenseofamaterialcellthatisfertilizedbyanothercelltoformanewunion.Rather,hisembryowasmoreprocessthanobject,anditwastheoreticalratherthanobservedformostspecies.Hewouldhavebeenabletoseeeggsonlyinsuchnonplacentalanimalsaschickens,frogs,orinsects.YetwhatAristotledidsee,especiallyinchicks,lookedinitiallyunformedandonlygraduallyactualizingthepotentialthroughtheformationeffectedbyepigeneticemergence.Asusual,Aristotlesinterpretationwasreasoned,accordedwithavailableobservableevidence,andprovidedanexplanationforthemanifestdevelopmentalprocessesofgrowthanddifferentiation.HisepigeneticinterpretationdominatedintotheeighteenthcenturyandfoundresonancewithleadingCatholicthinkers.SaintAugustineheldthattheprocessofgivingrisetoahumanlifewasgradualandthatthehumanonlybecomeshumanafterthefetusisformedandgrowing,afterquickening.Abortionwasconsideredasin,butnothomicideuntilafterfullhominizationhadoccurred.Anembryowasmaterialandwasalive,butitwasnotyetahuman.ForSaintThomasAquinas,thefetusfirstacquiresavegetativesoulandbeginstogrowlarger.Thenlateritacquiresananimalsoulandbeginstobecomedifferentiatedwithallitsanimalparts.Onlythendoesitacquireitsintellectualorrationalsoul,andonlythenisitfullyensouledandhuman.Thisinterpretationofdelayedhomini-zationdominatedearlyCatholichistory.EpigenesisprevaileddespiteinfluencessuchasPopeSixtusV,whoissuedthefirstpapaldeclarationonthesubjectin1588appar-entlybecauseofworryaboutrisingprostitutioninRome.SixtusCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?329decreedthatcontraceptionandabortionatanystageconstitutedhomicide.Yetthreeyearslaterwhenhedied,PopeGregoryXIVpointedtothestandardtheologicalunderstandingthathominiza-tionoccursonlygraduallyandreturnedtothelong-standinginter-pretationthatdevelopmentisepigeneticandthatthehumanemergesonlylaterwithfullensoulment.Withtime,additionalobservations,andadditionalphilosophicalreflection,however,otherinterpretationsemergedfromwithinnaturalphilosophytochallengeAristotlesepigenesis.Someresearcherswerepushedtowardanalternativebecauseoftheirmetaphysicalmaterialism.Ifallthatexistsarematteralongwiththemotionitexperiences,thentheyaskedhowepigeneticdevelopmentcouldyieldformfromnonform.Surely,suchemer-gencerequiressomeunacceptablevitalforceordirective,likeAristotleshypotheticalcauses.Thedemandformaterialisticmeta-physicthereforeledtopreformationism.Theformsimplymustbealreadyintheveryearliestmomentofanindividualslife.Otherwise,howcouldthenecessarysortsofchangeoccur(Roe1981)?Thisledtoheatedeighteenth-centuryargumentsaboutwhetheranorganismbeginsmoreorlessliterallypreformedandjustunfolds(orevolves),orwhetheritarisesgraduallyandepigeneticallythroughaprocessofembryonicdevelopment.Whilematerialistsemphasizedthemetaphysicalunacceptabilityofhypotheticalandapparentlynonmaterialorvitalisticcausesofemergenceofformfromnonform,however,epigenesistsinsistedontheprimacyofanepistemologybasedonobservation.Andobservationdidnotrevealtinylittlealreadyformedbeingsfromthebeginning.Inimportantways,thisdebateabouttherelativeprimacyofmetaphysicsorepistemology,aboutunfoldingoremergence,aboutpreformationorepigenesis,hasinformedalldiscussionssinceandevenliesattherootoftodaysheateddebates(seePinto-Correia1997).ThePhysicalEmbryoDebatebeganseriouslyinthelateeighteenthandearlynineteenthcenturies,whenembryosbecamePhysicalandvisibleinnonhumanspecies.Microscopesandanethosofnaturalphilosophyencouragedobservation.Thequestionswere,Whatcouldobservationsreveal,andwhatdidtheobservationsmean?CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n330janemaienscheinThisdebateplayedout,forexample,intheworkofCasparFriedrichWolffandCharlesBonnet,bothlookingatchickdevelop-ment.Theylookedatthesamethingandevenagreedaboutwhatitwasthattheysaw,buttheirconclusionsdiffered.Wolffwasanepigenesist,forwhomformemergesonlygradually;Bonnetwasapreformationist,whoinsistedthatformmustexistfromthebeginningofeachindividualorganism.Thisisastoryaboutcom-petingmetaphysicalandepistemologicalconvictions(Roe1981,Maienschein2003)In1759,Wolffstudiedthechickeggfromfertilizationthroughthetwenty-eight-hourstage,whichisshortlybeforetheheartbecomesclearlyvisibleandbeginstobeat.AsWolfflookedatchickafterchick,hourafterhour,hesawchange,yes,butnochick.Hedidnotseethechickform,abeatingheart,oranysmallpreformedchick.Instead,hewitnessedmovementandgradualchange.Hedidcon-siderthatperhapstheformwasjustnotvisibleyetbecauseitwassotiny,butthenstrongermicroscopesshouldrevealmoredetailandtheydidnot.Wolffconcludedthatweshouldtrustourobservations.Ifwecannotseesomething,thenwecanlegitimatelyassumethatitisnotthere.Thisisastrongepistemologicalassumptionaboutthenatureofknowledgeandjustification,andithelpedthatWolffsinterpretationalsoconformedtothestandardArisotelianepigeneticinterpretations.Tenyearslater,Bonnetalsounderstoodthepowerofempiricalobservation.Healsolookedcloselyatmanychicks,andheagreedthathedidnotseetheformedchickbeforethetwenty-eight-hourstage.HeagreedwithWolffthattheyarenotvisibleyet.ButBonnetconcludedthatthepreformedformofthechickmustbethere,justsomehowhiddenintheegg.Sinceweknowthatformexistslater,anditmustarisethroughtheactionsofmatterandmotion,thereforeitmustbepresentatthebeginning.Sincevitalforceswereunac-ceptabletothismaterialistandsincetherewasnootherexplanationforthegradualemergenceofform,Bonnetconcludedthattheformhadtoexistalready.Ifnaturalknowledgeistorelyonobservation,onlogic,andonapropermaterialisticmetaphysics,thentherecouldbenofurtherquestion.Anorganismmustbepreformedinsomeway.Inretrospect,weseeotheralternatives.Bonnetmighthavesaid,asNewtondidaboutgravity,Idonotknowhowformarises.Hypothesesnonfingo.ButBonnetwantedanexplanationoftheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?331originofindividualformandconcludedthatitmustresideinpreformation.Theresultsweredebatesonseverallevelsandthecoexistenceofcompetinginterpretationsofindividualdevelopmentwithepigenesisandpreformationism.Furtherobservationsofembryosintroducednewgroundsfordebate.Sofar,observershadseenchick,frog,andavarietyofinsecteggs,butitwasnotyetclearwhethermammalsalsohaveeggs.Someassumedthatallanimalssharethebeginningasanegg.In1827,KarlErnstvonBaerannouncedhisdiscovery(inafriendsdog,killedforthepurposeofexperimentalstudy)thatevenmammalshaveeggs,thoughitisdifficulttoobservetheirnormaldevelopmentsincetheeggsremaininsidemothers.Thisledtoenthusiasticstudyofthedevelopmentalstagesofembryosinasmanyspeciesasitwaspos-sibletostudy.Improvedmicroscopesandmicroscopictechniquesplayacentralrolehereastheymadeembryosinagrowingnumberofspecies,andatincreasinglyearlierstages,morevisible.Repre-sentationsinillustratedplatesandinwaxmodelswereimportantinpresentingtheembryotootherresearchersandtothepublic(Hopwood2002).Thelasthalfofthenineteenthcenturywasalsodominatedbytheimportanceoftheembryoforevolutionarytheory.Darwinpointedtoembryologyasfundamentalforinterpretingrelationships.Inchapter13oftheOrigin,heasked:How,then,canweexplaintheseseveralfactsinembryology,namelytheverygeneral,butnotuniversaldifferenceinstructurebetweentheembryoandtheadult;ofpartsinthesameindividualembryo,whichultimatelybecomeveryunlikeandservefordiversepurposes,beingatthisearlyperiodofgrowthalike;ofembryosofdifferentspecieswithinthesameclass,generally,butnotuniversally,resemblingeachother;ofthestructureoftheembryonotbeingcloselyrelatedtoitsconditionsofexistence,exceptwhentheembryobecomesatanyperiodoflifeactiveandhastoprovideforitself;oftheembryoapparentlyhavingsometimesahigherorganizationthanthematureanimal,intowhichitisdeveloped.Weknowthiswasarhetoricalquestion,andsureenoughhecon-cluded,Ibelievethatallthesefactscanbeexplained,asfollows,ontheviewofdescentwithmodification.Andthatfurthermore,theleadingfactsinembryology,whicharesecondinimportancetononeinnaturalhistory,areexplainedontheprincipleofslightmodificationsnotCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n332janemaienscheinappearing,inthemanydescendantsfromsomeoneancientprogenitor,ataveryperiodofthelifeofeach,thoughperhapscausedattheearliest,andbeinginheritedatacorrespondingnotearlyperiod.Embryologyrisesgreatlyininterest,whenwethuslookattheembryoasapicture,moreorlessobscured,ofthecommonparent-formofeachgreatclassofanimals.InhisGeneralleMorphologiederOrganismen,ErnstHaeckelwentfurther.Hesawontogenyasthebriefandrapidrecapitulationofphylogenyandsaweachindividualsdevelopmentasfollowingthesequenceof,andindeedbeingcausedby,theevolutionaryhis-toryofthatindividualorganismsspecies.Inhishighlypopularandwidelytranslatedbooks,Haeckelofferedpicturesofcomparativeembryology.See,heseemedtosuggest,thehumanformemergesfollowingtheevolutionarydevelopmentandadaptationsofitsancestors.Formarisesfromformoftheancestorsandunfoldsfollowingprescriptedstages.Darwinwasnotanembryologist;nordidhecontributedirectlytoourunderstandingoftheembryo.NordidHaeckel.ButwhileDar-winsuseoftheembryoinsupportingevolutionarytheoryandinhelpingtointerpretevolutionaryrelationshipswasconsistentwithvariousversionsofeitherepigeneticorpreformationistdevelop-ment,hisviewwasdecidedlypreformationist.Hiswasanotherpreformationistinterpretationbasednotonobservationsbutonthemetaphysicaldemandsofhisformofmonisticmaterialismandmotivatedbyhisdesiretoprovideevidenceforevolution.Thisprovidedthecontextinwhichthosestudyingcellsandembryosworkedattheendofthenineteenthcentury.Onlyin1869,thatis,shortlyafterDarwin,didtheCatholicChurchalteritslong-standingepigeneticinterpretation,whenPopePiusIXdecreedthathominizationisimmediateandbeginsimplic-itlyatconception.Unfortunately,hisApostolicaeSedisgivesfewcluesaboutwhatledtohisinterpretation,whichoverthrewcenturiesofAristotelianthinking,norwhetherhewasdrawingontherecentbiologicaldiscoveriesoffertilizationandofthemam-malianegg.Meanwhile,inGermanytheanatomistWilhelmHisturnedtohumanembryos.Hecollectedeveryhumanembryohecouldfindandsetupnetworksofphysicianstocontribute,seekingtoestablishthepatternsofhumandevelopment.TheAmericananatomistCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?333FranklinPaineMallstudiedwithHisandcarriedhumanembryologytotheUnitedStates.WhenHisdied,hiscollectionwenttoMall,thenattheJohnsHopkinsUniversity.In1914,MallpersuadedtheCarnegieInstitutionofWashingtontosupporthisgrowingembryocollection,thatledtothenationalHumanEmbryoCollection,whichisstillthemostimportantsourceofhumanembryomaterialandhistory(Maienschein,Glitz,andAllen2004).HisandMallsembryoswereallnecessarilydead,sincetherewasnowaytostudylivinghumanembryosinsidethemothers.Theembryoswereseenasmaterialobjects,withoutquestionsabouttheappropriatenessofcollectingandstudyingthem.EvidencethatthepubliclargelyagreedincludesthefactthatthestateofMarylandsDepartmentofPublicHealthurgedphysicianstocontribute.Therewerenoknowncomplaintsabouttheproject,perhapsinthebeliefthatimprovedunderstandingofhumanembryonicdevelopmentwouldhavemedicaltherapeuticvalue,overtime.TheBiologicalEmbryoThe1920s30sbrought,astheYaleembryologistRossHarrisonputit,agoldrushofstudiesofcausesandprocessesofdifferentiation.Whatcausestheunformedtobecomeformed:materialmechanicalchangeswithintheembryoitself,asHishadargued,orsomesetofspecialdirectiveforceswithinoroutsidetheembryo?Wastheresomethinguniqueaboutthelivingorganism?Didweneedwhatbiologistsbythetwentiethcenturyregardedasametaphysicallyquestionablevitalforcetoexplainemergenceofform(andHansDrieschdidgiveupembryologyandtakeupmetaphysicspreciselyonthisassumption),orisitsomethingaboutthenatureoftheorganicmatteranditsorganizationthatallowsdevelopmentanddifferentiationofcomplexforms?Howshouldembryosbestudied?HansSpemanntheorizedthatthetissuefromthedorsallipoftheblastoporeinamphibianshasspecialpowerstoinducetherestoftheembryoandtoserveasamaterialorganizertoproducedif-ferentiationandmorphogenesis(Hamburger1988,Maienschein1991).Dozensofresearcherstookupthechallengetofindtheprecisenatureoftheorganizer.Thereislittleevidencethatthistheoryreachedthegeneralpublicorevenbiologistsinotherspecialitiesoftenorinmuchdetail,butaneducatedlayaudiencedidaskCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n334janemaienscheinwhethersuchorganizationresultedfrommechanisticorvitalisticforces.ThephysicistErwinSchro¨dingerreachedthiswideaudiencewhenheasked,WhatisLife?andthatdiscussionwasclearlyinformedbyembryoresearch.Theepisodehasaninternallogicbutalsoraisesquestionsaboutourselectivehistoricalmemory.Thisepisodestartedwithtrans-plantationexperimentsofthe1890sandintotheearlytwentiethcentury.TheseincludedthefirststemcellexperimentsbyHarrisonwithneuroblastsintissueculture,andotherexperimentswithnucleartransplantation.Todaysstemcellresearchersknowlittleofthishistoryoftheirownresearch,whichleavesthemandthewiderpublicwiththeimpressionthatsomethingnewandamazing(orhorrifying)hasbeeninventedjustinthelastfewyears.Instead,stemcellandcloningresearchtodayisrootedfirmlyintraditionscarryingbacktoAristotle,throughtheworkofthetransplantationresearchersofthelatenineteenthandearlytwentiethcenturies.Harrison,Spemann,andtheirleadingcontemporariesassumedametaphysicalmaterialism.Whatexistsarematter,itspatternsoforganization,andchangeovertime.Thereisnoroomforvitalforcesorfluids,andtheyassumedthatdevelopmentisepigenetic.Thereisnoformfromthebeginning,butitemergesgradually,overtime,andguidedbyinternalforcesandfactors.Buthow?Thatwastheresearchprogram,focusedondiscoveringthematerialprocessesthatshapetheembryointoanindividualorganizedorganismoftherightsort.Epistemologically,theyassumedthatexperimentationwastheappropriateapproach.Sinceitwasnotpossiblesimplytoobservenaturalprocessesandtoseeinsidetheeggandembryo,itwasnecessarytocontriveexperimentalconditions.Manipulationofconditions,carefullycontrollingtheenvironmentasmuchaspos-sible,couldproducenewknowledge.Itwasthatnewknowledge,takencollectively,thatwouldrevealthepatternsandprocessesofwhatcametobecalledmorphogenesis,ortheappearanceofform.ThemostimportantoftheexperimentswereHarrisonsonnervefibersandSpemannsontheorganizer.Bothbeganwiththeideathattransplantingpiecesofadevelopingembryofromoneorganismtoanothercouldrevealtherelativecontributionsfromthedonorandthehost.Workingwithfrogs,whichareabundant,havelargeeggs,andareeasytomanipulateexperimentally,theybothsawthepowerofheteroplasticgrafting,ortakingandrecombiningpiecesfromCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?335animalsthatlookdifferentwithdifferentcolorsorsizes,forexample.Thismadeiteasytotellwhichtissuewasfromwhichorganisms.In190710,Harrisonrefinedthefirsttissueculturetechniquewithneuroblastcells(essentiallytodaysneuralstemcells,knowntogiverisetonervefibers).Heexplantedthesecells,transplantingthemoutofadevelopingfrogintoaculturemediuminadish.Theygrewoutasnervefibers,apparentlyjustastheywouldhavedoneundernormalconditions.Thissuggestedthatthecellscontainedanintrinsiccapacityfordifferentiation,yetundernormalconditionsthatdifferentiationwouldalsobeconstrainedanddirectedbytheenvironmentalconditionsforeachparticularcell.Theconclusionwasclear:anembryohasinternalcapacitiesfordevelopment,anditalsodependsoncuesandinputfromfactorsexternaltotheeggandembryoitself.Spemannfocusedontheearlierstagesofdevelopment,onthestagewhentheblastocystundergoesgastrulation.Thatisthefirststagewhentheembryobeginstobecomevisiblydifferentiated.Theundifferentiatedclumpofcellsundergoesrapidcellmovement,withaflowingofcellsintowhatiscalledtheblastoporeandtheforma-tionofgermlayers.Theclumpbecomesanorganizedballofthreelayersthatwillbecomedifferentpartsoftheorganism.Thisisthefirsttimewhenthereisclearorganization.Spemannasked,Howdoesthishappen?Whatcausestheapparentlyundifferentiatedmassofcellstobecomeorganized?Perhaps,Spemannhypothesized,thereisanorganizer.Thismustbematerialanditshouldbeaccessiblethroughexperimenta-tion.Indeed,hefoundthatthedorsallipoftheblastopore(thatis,aparticularsetofafewcellsatthestillunorganizedandundiffer-entiatedblastocyststageofdevelopment)inducesorganizationofthecells.Thisparticularmaterialseemedtosetupthelayeringofcellsintothethreelayers.Otherresearchsuggestedthattheprocessismuchmorecomplicated,butduringthe1920sand1930stherewastremendousexcitementaboutwhatseemedtobediscoverablematerialcausesoftheproductionofformfromtheunformedcells.Continuingresearchhasreinforcedthisearlyconvictionthatitistheblastocyststageofdevelopment,arounddaysfivetofourteeninhumans,thatisthebeginningoforganizationanddifferentiation.Thisisthestageatwhich,inhumansandothermammals,theCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n336janemaienscheinpreimplantationembryomustbeimplantedinthemotherandmustbegintoexchangenutrientswiththemother,oritwillnotsurvive.Itisalsothestageatwhichtheembryobeginstogrow,asitabsorbsnutrientsfromthemother.Anditisthestagewhentheclumpoflargelyidenticalcells,nowcalledembryonicstemcells,begintoundergodifferentiationanddevelopment.Inadditiontothebiologicalscientificresearch,thesocialandculturalcontextbegantohavesomeinfluence.HarrisonchairedtheU.S.NationalResearchCouncil(NRC),whichpromotedbloodtransfusionsduringWorldWarI,andhistissueculturediscoveryinspiredRockefellerUniversitysAlexisCarreltoexpandtissueculturestudyfortherapeuticapplications.Itisthisresearchandtheassumptionsthatunderlieitthathaveprovidedthetraditionofcelllinedevelopmentandapplicationthathaveledtotodayshopesforeffectivestemcelltherapies.HarrisonandSpemannwerestudyingfrogs,whileCarrelandhismedicalcolleaguesworkedwithhumans(Landecker2004).Itseemslikelythatevenifbiologistshadbeenabletostudyhumanembryos,thepublicresponsewouldhavebeenpositive.Thereisnoevidenceofearlytwentieth-centurysocialconcernaboutembryos.Humanembryoswereinvisible,insidewomen.Becausemammalianembryosremainedhidden,earlydevelopmentremainedamysterytomostpeople.Theassumptionwasthathumanembryodevelopmentwassimilartothatofotheranimals,buteventheexperimentalstudyofprimatereproductionthatbeganintheearlytwentiethcenturyinplacesliketheCarnegieInstitutionofWashingtonwasslowtorevealinsightsabouttheearliestdevelop-mentalstages(Maienschein,Glitz,andAllen2004).Embryosmustdevelopgradually,epigenetically,withformemergingthroughaprocessofstages.Buthow,andwhatdirectsthedevelopment?Istherereallyanorganizer,andifsowhatisitandhowdoesitwork?Thesequestionsremained.TheInheritedEmbryoThe1950syieldedonesortofanswerwiththeInheritedembryo.ThediscoveryofthestructureofDNA,thestuffofheredity,hasbeenwelldocumented(Olby1974,Judson1979).Geneticshademergedasafieldofstudy,andresearchershadbeenexploringboththeeffectsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?337ofthetheoreticalunitscalledgenesandthechromosomalstruc-turepresumedtocontainthegenes.Yetthestudywaslargelyabstract,andbiologistssuchasHarrisondidnotseegeneticsascontributinganythingtothestudyofdevelopment.AsdepartmentchairatYale,Harrisonsawnopointinhiringageneticist,forexample,sinceitwasthestudyoforganismsandtheirprocessesthathesawastheimportantworkforbiology.Onlyinthe1950sdidgeneticsbegintolinkeffectivelythroughmolecularbiologysuchthatitwaspossibletoimaginetheDNAanditspresumedgenesastheconcretematerialbasisforheredity.This,inturn,suggestedthatheredityprovidestheunderlyingcausalshapingofdevelopmentalprocesses.SinceeverycellcontainsthesameDNAandgenes,andyetthecellsbegintodifferentiateandtoproducemorphogenesisovertime,itmightbethatthegenesandDNAcarrythenecessaryinformationtoguidedevelopment.GenescouldbetheorganizerthatSpemannhadsought.In1938Spemannhadsuggestedafantasticalexperimentthatwouldgetattherelativecontributionsofthenucleuswithitschromosomesandgenesandcytoplasm.Itwasaconceptuallysimpletransplantationexperiment,removingthenucleusofoneeggandreplacingwithanothernucleus.Hedidnotcarryoutthisexperiment,ornotsuccessfully,butin1951RobertBriggsandThomasKingdid.Theytransferredfrognucleitoproduceanewkindofhybrid.ClonedfrogsgracedthecoverofpopularmagazinessuchasTimeandNewsweek.JohnGurdonwentfurtherinthe1960s,transplantingdonornucleifromfrogsinlaterdevelopmentalstagesintotheeggs.Hetransplanteddonornucleifromanalbinofrogintotheeggofanormallypigmentedmother,andtheoffspringturningouttobelikethedonornucleus,allalbino.Thissuggestedtheverystrongpredominantinfluenceofnuclearinheritanceoverdevelop-ment.Yes,epigeneticdevelopmentmightoccurgradually,throughtime,asformemergesfromtheunformedmaterial.Buttheguidingdirectionseemedincreasinglytooriginateinthenucleusandthegradualexpressionofgeneticinformationcodedthere.Accumulat-ingevidenceofthissortreinforcedtheideathatdevelopmentisnotonlylooselydirectedbutactuallycausedbythegenes.Andifso,thentheinformationanddeterminantsfordevelopmentanddifferentiationarealreadypresentatfertilization.Itseemedthatdevelopmentoccursbypreformationafterall,oratleastbysomeCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n338janemaienscheinversionofgeneticpredetermination.Towhatextentisdevelopmentactuallydeterminedorfixed,then,andtowhatextentcanitrespondtochangingenvironmentalconditions?Opinionshiftedtowardincreasingdeterminism,reinforcedbytheenthusiasmfortheHumanGenomeProjectofthelate1980sand1990s.Publicresponsetodiscussionsofdevelopmentduringthistimewaslimited,butwhatreactiontherewasremainedlargelycuriousratherthancriticalorconcerned.Theresearchmighthaveraisednewphilosophicalandbiologicalquestionsaboutindividualityandidentity?Afterall,ifaneggdidnotneeditsownnucleus,butanotherwoulddo,whatisthebiologicalbasisforconceptsofselfandother,ofidentityandautonomy?Butthisquestioningseemsnottohaveoccurredtoanysignificantextent.Thisresearchwaslargelytakenasjustthat,biologicalresearch,tobecarriedoutwithoutmuchthoughtaboutitsinterpretationsorimplications.TheVisibleHumanEmbryoIthasbeenonlysince1978,whenthehumanembryobecameliterallyvisible,thatembryoshavebecomeanobjectofwidepublicinterest.Thebirthofthefirsttesttubebaby,thelivelyandnormallittleLouiseBrown,firsttookhumanembryosoutofthemotherandintothepubliceye.Sociallyandculturally,wearestillsortingouttheimplicationsofthisdiscoverythattheearliesthumandevelopmentalstagescantakeplaceseparatelyfromthemother,andthequestionsraisedaboutwhatsinthedish.Anembryoisstillidentifiedasbeginningwiththeprocessoffertilizationuntil,inhumansandothermammals,thedevelopinganddifferentiatingorganismgainsallitsorgansystemsandbecomesafetus.Whathappenedafter1978wasinitiallynotaboutchangingscientificdefinitionsoftheembryo,thoughtheemphasisshiftedfromanimalstudiestohumanembryoresearchforthefirsttime.Theprimarychangesrightawaywerethattheembryogainedindividualmeaningforprospectiveparentsaswellaspublicmeaning,especiallyinthecontextofabortionpolitics.Clearly,technologyandimageshavebeenveryimportantinintroducingthisscientificresearchtothepublic.LennartNilssonsphotographshadprovidedabackgroundofassumptionsandinterest.Hisstunningpicturesweretakenwithascanningelectronmicroscopeandendoscopesandshowedthefetusinthewomb.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?339TheLifemagazinepresentationin1969gavemostpeopletheirfirstimagesofthedevelopinghuman.Thefactthatthesewerefetuses,andoftenlater-stagefetuses,andthefactthatthesewerehighlycoloredandcontrivedpictureswerenotpartofthepublicimpres-sion.Insteadtheseimagesoflittlepersons,suckingthumbsandlookinginnocent,becamethepublicimageofembryos.Manypeople,andperhapsasignificantmajority,stillimagineembryosasthesetinyclearlyformedbeingsfloatinginthewomb.Thisisnotthebiologicalembryo,butsinceithasbecomethepublicembryoformanypeople,anyattempttounderstandshiftingmeaningsmusttakethismisconceptionintoaccount.Anotherimportantsocialshiftinmeaningoftheembryocomeswiththerecognizedclinicalimportanceofthatclumpofundiffer-entiatedcellsinadishforpatients.AtsignificantpersonalcostintheUnitedStatesandwithsignificantpersonalinvestmentinanycase,thoseindividualsengaginginIVFhavecometoseetheembryoasthebeginningoftheirbaby.Thebiologicalfactthatbeforeimplantation,thisclumpofcellsisreallyjustthat,aclumpofundifferentiatedcells,doesnottakeawaythesocialandmedicalmeaninginvestedinthosecells.Makingthehumanembryovisible,withitspotentialformedicaladvantage,hasproducedincreasedfunding,aswellasincreasedpublicnervousness.WasIVFsafe?Forwhomandunderwhatcondi-tionswasitdesirable,andwhoshouldpayforthisinvitroprocessifitwasconsideredamedicaltreatment?Countrieshavehaddifferentresponsestoembryoresearch.IntheUnitedStates,IVFandembryoresearchgenerallyhasremainedunregulatedandlargelyprivatized.Whatregulationexistsisatthestatelevelthroughlawsandcourtcaserulings,withoverlappingandoftencontradictoryresults.TheonlyfederalrestrictionsareonfundingthroughtheappropriationsbillsfortheNationalInstitutesofHealth.Canada,Australia,andtheUnitedKingdomhaddifferentresponsesin1978andsince,acceptingIVFasapublicgoodwithpublicfundingandpublicregulation.Asaresult,thehumanembryoisdifferentlyvisibleineachcountry.TheConstructedEmbryoIn1997,IanWilmutannouncedthefirstclonedmammal,Dollythesheep(Wilmut2000).AswithBriggsandKingsfrogs,DollywasCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n340janemaienscheinproducedbytransplantingadonornucleusintoahosteggfromwhichthenucleushadbeenremoved.Inthiscase,however,thedonornucleuscamefromanadult,andtheresultwasseenasanunnaturalhybridthatchallengedassumptionsaboutwhatispossible.WilmutandhisteambroughtexperimentalembryologytothepublicverydramaticallyandtherebyfirstmadetheConstructedEmbryopublicthoughsuchresearchhadalreadybeenlongunderwayinsideresearchlaboratories.ThatDollyexisted,suddenlyandsurprisingly,asLouiseBrownhadexistedsuddenlyandsurprisinglyoutofIVF,challengedtreasuredassumptionswithinsocietyaswellaswithinscience.Per-hapscloneswoulddevelopdifferentlythannormalorganisms,orperhapstheywouldhaveadiminished(orenhanced)life.Whatisnormalandwhatisacceptableforembryos?thepublicbegantoask.Embryonicstemcellresearch,whichreachedpublicattentionjustayearlater,raisednewquestionsbothscientificandpublicandgeneratedprospectsforregenerativemedicaltherapies.Butthisresearchcan(atleastfornow)bestbecarriedoutbyharvestingundifferentiatedembryonicstemcellsfromembryos.Asweknowfromtheheateddebates,somemembersofthepublicfindthisunacceptablebecausetheymaketheassumptionthatanembryoisalreadyaperson(oratleastapotentialperson)andthereforewedonothavetherighttoharmit.Thisdiscussiongetsrightattheheartofwhatwemeanbyanembryo,andhowweknow.Scientistsurgethatwedrawonscien-tificunderstandingoftheembryo.Ahumanembryo,especiallybeforeitisimplantedinthemother,isreallyjustabunchofundifferentiatedcells.Tothebestofavailableknowledge,nosig-nificantgeneexpressionhasbegun;thereisnodifferentiation;thereisnosignificantgrowth.Thisisabunchofcellsdividinganddividing,atleastupthroughtheblastocyststage.Onlyatthatpoint,asHarrisonandSpemannrecognized,dodifferentiationandmor-phogenesisbegintooccur.Onlywithimplantationandgastrulationdoestheembryobeginactingandbegintheepigeneticprocessesofdevelopment,informedbythehereditycarriedinthegenes.conclusionsWhat,then,isanembryo?Somebiologistsprefertocallthepreimplantationstagesapreembryo.OthersurgethatitwouldbeCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nWhatIsanEmbryoandHowDoWeKnow?341easierandpoliticallysafertodropthetermembryoaltogether,thoughthatissurelypoliticallynaı¨ve.Whogetstodecidewhatanembryois?Onthefaceofit,thisisabiologicalquestionsinceembryosarebiologicalobjects.Therefore,biologistsshouldatleasthaveasay.Theyarequiteclearonthematter.Anembryoisnotyetformedinthesenseofstructuredwithfunctioningdifferentiatedparts,andapreimplantationembryoisreallylittlemorethanabunchofundifferentiatedcells.Atwo-,four-,oreveneight-celledpreimplantationembryocanbecometwins,quadruplets,orevenoctuplets.Upthroughtheeight-cellstage,thecellscanevenbepulledapartinthelabandtheseparatedcellscandevelopindivid-ually.Oroneortwoormorecellscanberemoved(perhaps,asiscommonlydonenowinfertilityclinics,totestoneofthecellsgenetically),andtherestcandevelopnormally.Thisisclear.Bio-logically,anembryoisnotyetformed,notyetdifferentiated,notyetrecognizablyhuman,andindeednotevenunalterablyanindivisiblesingleindividual.Yetbiologistsdonotaloneownembryos.Aspublicobjects,embryosaremuchmorecomplicated.Alargepoliticalgroupwouldliketodefineembryosasbeginningwithfertilization,andashavingstatusasembryonichumanpersonsatthattime.Whatwemustrealizeisthatindoingso,theyareinvokingmetaphysicalassump-tionsthatlieoutsidescienceandthatoftendependonreligiousassumptionsthatarenotsharedbythelargercommunityandcannotbejustifiedonscientificoranyotherclear-cutgrounds.Theirepistemicwarrantcomesfromsuchclaimsofintuitionordivineknowledgeorpureconjecture.Suchmeaningsare,ofcourse,highlyproblematic.Wehaveyettoworkthroughwaystodealwithcaseswherethebiologicalandsocialbecomesointertwinedasinthiscase.Weneedtoaskwhatauthorityandwhatprocesseswehaveforcarryingpoliticalandsocialdecisionstorestrictresearchbackintothelaboratory.Doestheloboratoryhaveanysortofprotection,assomeargue;istherearighttocarryoutscientificresearch?Ifso,whencantherebelimitsandhowaretheytobeimposed?ThequestionsWhatisanembryo?andHowdoweknow?remainworksinprogressbiologically,philosophically,andpublicly.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nmanfredd.laubichler18EvolutionaryDevelopmentalBiologyEvolutionarydevelopmentalbiologyisanemergingnewresearchareathatexploresthelinksbetweentwofundamentalprocessesoflife:developmentofindividualorganisms(ontogeny)andevo-lutionarytransformationinthecourseofthehistoryoflife(phy-logeny).Forsomeofitsmoreardentproponentsevolutionarydevelopmentalbiology,orevo-devoforshort,representsanewparadigmthatcompletestheModernSynthesisofthe1930sand1940s,whileothers,oftenthosewithamoreastutesenseofthehistoryofbiology,haveemphasizedthelong-standingconnectionsbetweenthesetwoareasofstudy.Butallagreethatevo-devoofferssomeofthemostpromisingtheoreticalperspectivesinevolutionarybiologyatthebeginningofthetwenty-firstcentury(seeforexampleAmundson2005,Carroll2005,Carrolletal.2005,Hall1998,KirschnerandGerhart2005,Laubichler2005,Mu¨ller2005,Wagneretal.2000).InthisessayIwillfirstsketchtheemergenceofpresent-dayevolutionarydevelopmentalbiologyduringthelastdecadesofthetwentiethcenturyfollowedbyabriefoverviewofthecentralquestionsandresearchprogramsofevo-devo.Iwillconcludewithadiscussionoftheoneproblemtheissueofhowtoexplainevolu-tionaryinnovationsandnoveltiesthathasthemostprofoundimplicationsforthephilosophyofbiology.Asthisexampleillus-trates,thetheoreticalpromiseofevo-devoliesintheintegrationofdifferentexplanatoryparadigms,thoseofevolutionarybiologyandpopulationgenetics,whicharebasedontheanalysisofultimatecausesinthesenseofErnstMayr,andofdevelopmentalgeneticsandphysiology,whichattempttogiveamechanisticaccountoftheoriginsoforganismalstructuresinformofproximatecauses,thatis,ofmolecularandcellularmechanisms(Mayr1961).Aswewillsee,342CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology343suchintegrationisnotpossiblewithoutfirstovercomingseriousconceptualandtheoreticaldifficulties.Thereforeitshouldbecomeclearinthecourseofthisessaythatanyfuturesynthesisofevo-devowillbeconceptualratherthansimplydatadriven.Theconceptofregulationhas,somewhatsurprisinglyperhaps,emergedasacentraltopicinevo-devo(see,e.g.,Carroll2005andCarrolletal.2005).Theideaofdevelopmentalregulationhas,ofcourse,alonghistorydatingbacktoearlyconceptionsoftheorganismasaself-regulatingindividual,ofregenerationandthemaintenanceofamilieuinterieur,andofcoursetoHansDrieschsfamousexperimentandinfamousideathatorganismsareself-reg-ulatingequipotentialsystemsgovernedbyentelechy,whichbecamethefoundationofneovitalismatthebeginningofthetwentiethcentury.Whilethemysterybehindthephenomenonofindividualdevelopmenthasalwaysbeenhowitispossiblethatsuchacomplexandfragileprocessgenerallyleadstoapredictableoutcomehumaneggsdevelopintorecognizablehumansandseasurchineggsintoseaurchinsevolutionhasbeenperceivedasanopen-endedprocessofconstanttransformation,thusinmanywaysastheexactoppositeofregulation(seeCanguilhem1979).However,developmentandevolutionaremorecloselylinkedthanonewouldexpect.AstheGermanevolutionarybiologistGu¨nterOscheoncedrylyremarked,evolutioncannotputoutasignClosedBecauseofReconstruc-tion.Indeed,allevolutionarytransformations,eventhoughtheymanifestthemselvesonapopulationlevel,needtomaterializewithintheconstraintsandpossibilitiesofafunctioningdevelopmentalsys-temofindividualorganisms.Andthemorewelearnaboutthedetailsofthesedevelopmentalsystemsabouttheintricateregulatorynet-worksandlinkedpathwaysthatcontrolgeneexpressionandthusdifferentiationduringontogenythemorewerealizethatevolu-tionarytransformationsareaconsequenceofchangesinthesereg-ulatorysystemsandotherdevelopmentalprocesses.Regulationhasthusbecomeaconceptthatallowsussimultaneouslytounderstandthestabilityofindividualdevelopmentandthepossibilitiesofevolutionarytransformations.Butbeforewecontinuethesediscussionsoftheconceptualstructureofevo-devoletusfirstbrieflyexploreitshistory(seeLaubichlerandMaienschein2007,Amundson2005,Laubichler2005,LoveandRaff2003).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n344manfredd.laubichler1.evolutionarydevelopmentalbiology:anewsynthesisor‘‘oldwineinnewbottles’’?ThestudyofindividualdevelopmentoforganismshasalonghistorythatdatesallthewaybacktoAristotle.Theoriesofgeneration,preformism,andepigenesishadaprofoundimpactonthewaylifewasunderstoodaseithercreativeandvitalormechanicalaswellasontheinterpretationoforganicformsandtheirhistory.Andindeeditwasinthecontextofdiscussionsaboutthedevelopmentoforganicformsthatideasrelatedtotheevolutionaryhistoryoflifefirstemerged.Darwin,ofcourse,putthesequestionsinfocuswhenhelaidoutthefollowingproblemsandobservations:(1)Bothorganismsandspeciesvary.(2)Thevariationofbothorganismsandspeciesisnotarbitrary,butclustered.Inotherwords,parentsandoffspringcloselyresembleeachother,butnotcompletely,anddif-ferentspeciescanbegroupedtogetherintohighersystematicgroupsthatareunitedbycommonfeatures.(3)Thespecificcausesforthesimilarityanddifferencesbetweenparentsandoffspringwereunknown,butitwasclearthatthesecauseshadtoactinthecourseofindividualdevelopment(ontogeny).(4)Theprinciplesforgroupingindividualspeciestogetherintohighersystematicgroupswerebasedoncomparisonandtheestablishmentofhomologies.Homologiesthesameorgansindifferentindividualsirrespectiveofformandfunctionwereconsideredthebasisforthehierarchicalsystemofclassification;itwas,however,notalwaysclearhowhomologiescouldbeestablished;(5)Embryologicalobservations(Entwicklungs-geschichte)hadrevealedthatearlier(lesscomplex)developmentalstagesofdifferentspeciesmorecloselyresembledeachotherthandidadultstages;(6)Thegenealogicalperspectiveandthegeologicalrecordsuggestedthatlesscomplexformsoflifeemergedearlierinthephy-logenetichistory(Stammesgeschichte)thanmorecomplexforms(anargumentthathadalreadybeenmadeinthecontextofculturalstagesinthehistoryofhumankind);(7)Itwasclearthatallexplanationsaswellasthespecificdetailsofthishistoryneededtobeinferred,often,aswiththegeologicalrecord,fromratherincompletedatasets,asdirectobservationwasnotanoption(Darwin1859).Darwinhimself,honedbydetailedobservationsofthedevelop-mentalstagesofbarnacles,carefullysuggestedaconnectionbetweenCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology345ontogenyandphylogeny:Thuscommunityintheembryonicstructurerevealscommunityofdescent(Darwin1859,449).Otherssoonfollowedsuit,exploringhowtheobservablepatternsofontogenycouldhelptorevealthehiddenpatternsofphylogeny,mostfamouslyErnstHaeckel,whoformulatedthebiogeneticlaw,firstcalledthatwayin1872.Theideathatontogenyrecapitulatesphylogenyandthedevelopmentalperspectiveimplicitinthebiogeneticlawalsoseemedtoofferinsightsintooneofthemajorproblemsofcomparativebiology,theestablishmentofhomologyrelations.Ashareddevelopmentalhistory,soHaeckel,shouldbeasolidbasisfortheassessmentofhomologies.Thislaterassumptionbecameoneofthefoundationsofevolutionarymorphology,apro-graminitiatedbyHaeckelandhisclosefriendCarlGegenbaur,whichfocusedontheestablishmentofaphylogeneticsystemasthegoalofcomparativeanatomyandembryology(seeNyhart1995,2002,Laubichler2003,LaubichlerandMaienschein,2003).Thisneworientationbroughtaboutbytheemphasisontherelationsbetweenontogenyandphylogenyledtosomespectacularinsights.Inthelate1860stheRussianembryologistAlexanderKowalevskydiscoveredsimilaritiesbetweentheontogeneticse-quencesofAmphioxusandvertebratesaswellastheexistenceofachordadorsalisinthelarvaeofascidians(Kowalvesky1867).Thesediscoveriessuggestedthatvertebrateswerederivedfromthelarvaeofascidians,atheorythatwassoonchallengedby,amongothers,AntonDohrn,thefounderofthezoologicalstationinNaples,whoproposedanannelidancestryforvertebrates(Dohrn1875).Thisandmanysimilarexamplesillustratehow,inthedecadesafterthepublicationoftheOriginofSpecies,thescientificproblemofontogenyandphylogenywasprimarilydeterminedbythecontextofevolutionarymorphology(phylogeny).Matterschangedaroundtheturnofthetwentiethcentury,whenevolutionarymorphologygraduallydisappeared.Moreexperimen-tallyorientedresearchprograms,suchasgeneticsanddevelop-mentalmechanicsandphysiology(Entwicklungsmechanik),carriedtheday.Entwichlungsmechanikinitiallyemphasizedtheroleofmechanicalandphysico-chemicalcausesinexplanationsofdevel-opment;onlylater,inthecontextofphysiologicalandgeneticapproachestodevelopment,didevolutionaryquestionsreturn.Theseconcernshad,inthemeantime,beentransformedbytheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n346manfredd.laubichlerresultsofexperimentalgenetics.BothRichardGoldschmidtandAlfredKu¨hnattemptedtointegrategenetics,development,andevolution;Ku¨hnwithhisconceptoftheWirkgetriebederErbanlagen,amechanismforrealizinghereditarydispositions,andGoldschmidtwithhisconceptionofphysiologicaldevelopmentalgeneticsandhisdiscussionofchromosomalrearrangements,macromutations,andhopefulmonsters(Goldschmidt1940,Ku¨hn1955).ThemostinclusiveresearchprogramdevotedtoquestionsofontogenyandphylogenywasdirectedbyHansPrzibramattheViennaVivarium,aprivatelyownedandfundedresearchstationdevotedtoexperimentalbiology.Przibramandagroupoflike-mindedandwell-to-doscientistsestablishedwhatwasatthetimethemostsophisticatedinstitutionforexperimentalresearchthatfocusedonthestudyofdevelopment,regeneration,endocrinology,experimentalevolution,andthelifehistoryoforganisms(bothplantsandanimals).Inaddition,thisgroupwasinterestedintheintegrationofexperimentalandtheoreticalapproaches.Theirworkwas,inmanyways,themostdirectforerunnerofcurrentevo-devo.Unfortunatelythepoliticalturmoilofthe1930sand1940sputanendtothisuniqueresearchenvironment(see,e.g.,Przibram1907).Thetwounifyingproposalsofmid-tolatetwentieth-centurybiologyweretheModernSynthesisandtheemergingmolecularbiology(e.g.,MayrandProvine1980,Morange1998).Theformerwasbasedontheconceptualintegrationallowedbythemathe-maticaltheoryofpopulationgenetics,whilethelatterwasbasedmoreonaunifyinglevelofanalysisandasharedrepertoireofexperimentalapproachesthananintegrativeconceptualframework.Whilethesetwoapproachesandtheirsometimesacrimoniousrelationsdominatedmidtwentieth-centurybiology,experimentalembryology,rechristenedasdevelopmentalbiology,continuedtoflourishinitsownscientificniche.Buttherewerealwayssome,suchasConradWaddington,whoemphasizedtheunityoftheevolutionaryanddevelopmentalresearchprograms.Hisideaofanepigeneticlandscapeandtheconceptofgeneticassimilationbothrepresentedattemptstoincorporatedevelopmentalmechanismsintoevolutionaryexplanations(Waddington1940).Bytheearly1970sthemainfocusofdevelopmentalbiologyhadturnedtowardquestionsofdevelopmentalgenetics.ThiswasalsothetimewhenCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology347arenewedinterestintheproblemofontogenyandphylogenybegantoemerge.Earlycontributorstowhatwouldlaterbecomeevo-devofocusedontheproblemofdevelopmentandevolutionbecausetheyweredissatisfiedwiththeexplanatoryframeworkoftheModernSynthe-sis.Severalrelatedissues,inparticular,wereattheheartoftheircritique.Forone,theassumptionoftheModernSynthesisthatmacroevolutionarypatternscanbeexplainedbyasimpleextensionofmicroevolutionaryprocesseswasquestioned.Palaeontologists,wholedthislineofattack,focusedontwophenomenathattheythoughtcouldnotbereconciledwithinthisframework:(1)Theobservationthatthefossilrecordofseveralcladesshowsperiodsofrapidevolutionarychangefollowedbyextendedperiodsofstasisledtothehypothesisofpunctuatedequilibrium,and(2)theobservationthatmostconceivablemorphologicalpatternsarenotrealized,thatis,thatthemorphospaceislargelyemptyandthatrealizedmorphologiesareclusteredincertaindomainsofthemorphospace,seemedtosupportideasthatmorphologicalchangeisprimarilyaconsequenceofdevelopmentalprocesses,suchasheterochrony,ratherthatstrictlydeterminedby(incremental)geneticfactors(e.g.,EldredgeandGould1972,Gould1977).AnothersourceofdissatisfactionwiththeModernSynthesiswastheprivilegedroleofadaptationindominanttheoriesofevolution(seeGouldandLewontin1979,asthelocusclassicusforthiscritique).Theadaptationistparadigmwaschallengedbothbymolecularbiologists(e.g.,withtheneutraltheoryofevolution)aswellasbyevolutionarydevelopmentalbiologists,whoemphasizedtheroleofinternalfactorsinevolution.Themostprominentearlyconceptsinthiscontextwerethenotionofdevelopmentalcon-straintasalimitationonpossiblephenotypicvariationandthereforeonadaptationandRupertRiedlsideaofburden,whichpostulatesaninternal,inadditiontotheexternalandenvironmentallyinduced,selectionpressure(Riedl1975,Maynard-Smithetal.1985).Theinternalconditions,whichactasasecond(orratherfirst)selectiveenvironment,arethoseofthedevelopingsystem.InthecontextofRiedlstheorytheseinternalfactors,orsystemcondi-tions,ashecallsthem,canexplainboththeevolvabilityaswellasthehierarchicalorganizationoforganismalforms(Riedl1975,WagnerandLaubichler2004).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n348manfredd.laubichlerBythemid-1980saconsensushadbeenreachedthattheprevailingversionofevolutionarytheoryneededtobereformed.However,eventhoughtherewasgeneralagreementthatdevelop-mentneededtobepartofevolutionarytheory,therewasfarlessagreementabouttheactualresearchagendasoraboutthebeststrategiestoaccomplishthisdesirednewsynthesis.Thesecen-trifugaltendenciesonlygrewstrongerafterthediscoveryoftheconservedsequencesofHomeoboxgenesprovidedamajorboosttothedevelopmentalgeneticversionofevo-devo.Thispluralityofapproachestotheproblemofontogenyandphylogeny,whichcontinuestothisday,despitegrowingattemptstoestablishagenuinenewsynthesisofevo-devo,isalsoreflectedinitscurrentfragmentedstructure.Anintegrativeframeworkthatwoulduniteallthesedifferentevo-devoapplicationswithinagenuinenewsynthesishas,asofyet,notemerged.Whatwecanobserve,however,isaclusteringofresearchquestionsaroundtwoorthreemainemphases,suchasevo-devo,devo-evo,anddevel-opmentalevolution(seeHall2000,Wagneretal.2000,Mu¨ller2005).2.researchprogramswithinevo-devoAswehaveseeninthisbriefhistoricaloverview,present-dayevo-devohasmanydifferentroots.Correspondingly,therearealsoseveraldistinctresearchprogramscurrentlypursuedunderthebannerofevo-devo.Someofthoseclearlyoverlap,andcollectivelythesedifferentresearchprogramsillustratethemethodologicalpluralismthatischaracteristicofevo-devo.Thismultitudeofapproachesalsohelpstounderstandthecurrentprominenceofevo-devorelatedresearchwithinorganismalbiology.However,thesamediversitythatmakesevo-devoresearchsoproductiveinmanywaysalsoprovidesaformidablechallengetoanyattemptstoarriveatasynthesisofdifferentperspectives(moreonthatlater).Themainresearchprogramswithinevo-devoarethefollowing(seealsoLaubichler2005,Mu¨ller2005):1.Thecomparativeprogram.Comparativeresearchisoneoftheoldestparadigmswithinbiology.Itcontinuestoraiseimportantquestionsandprovidesthefoundationofallfurtherinvestigations.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology349Todaycomparativeresearchcoversmanydifferentlevelsofbiologi-calorganization,frommorphologyandanatomytoembryologyandgenomics.Inthecontextofthelattertheemphasishasnowshiftedfromsimplesequencecomparisontocomparisonofgeneexpressionpatternsandgeneproductsandtheirinteractions,thusallowingacomparativestudyofdevelopmentalsequencesathighresolution.Thecomparativeprogramisalsoimportantforevo-devointhatitprovidesuswithdetailedphylogenies.Theseareimportantinassessinganynumberofgenuineevo-devoquestions,suchashypothesesaboutevolutionarytransformationsorhomology.2.Theexperimentalprogram.Theexperimentalprogramisacontinuationofthevenerabletraditionofexperimentalembryologyanddevelopmentalphysiologythatdominatedorganismalbiologyduringthefirstdecadesofthetwentiethcentury.Thefocusoftheseexperimentalapproachesliesintheelucidationofthemechanismsofdevelopment.Eventhoughmostexperimentalresearchindevelop-mentalbiologytodayfocusesontheroleofgenesindevelopment,thereisalsoanincreasingawarenessoftheimportanceofepigeneticandenvironmentalfactorsinmorphogenesisandevolution.Wehavealreadyseenthatregulationhasbecomeoneofthecentralconceptsindescribingbothdevelopmentalsystemsandtheirevolution.Epigeneticfactors,suchasdifferentialmethylationpatterns,geno-micimprinting,posttranscriptionalcontrol,andRNAediting,aswellasbiophysicalpropertiesofcellsandtissues,geometricalpatternsofself-organizingsystems,andenvironmentalfactorssuchastemperature,allcontributetotheregulatorymachineryofdevelopmentalsystemsandtheirevolution.Inmanipulatingtheparametersofdevelopingsystemstheexperimentalprogramwithinevo-devohasproducedsomeinterestingresults.Amongthosearetherecreationofancestralmorphologicalpatternsinlimbsortheelucidationoftherulesofdigitreduction(AlberchandGale1983,Mu¨ller1989).Inbothcasesmorphologicaltransformationsthatcloselyresembleactualevolutionarytransformationswereaccom-plishedbymeansofexperimentalmanipulationofnongeneticpartsofthedevelopmentalsystems,suchasconstrictingthesizeormanipulatingthegeometryoflimbbuds.Theseapproachescon-tributegreatlytoourunderstandingofthepatternsandprocessesofevolutionarytransformations.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n350manfredd.laubichler3.Theprogramofevolutionarydevelopmentalgenetics.Thefocusofthisresearchprogramisonthegeneticcomponentsofdevelop-mentalsystemsandtheirinteractions.Inmanywaysevolutionarydevelopmentalgeneticsisthemostvisiblepartofcurrentevo-devo.ThediscoveriesfirstofHoxgenesandofothertranscriptionfactorsthattogethermakeuptheregulatorygenenetworkscontrollingtheexpressionofgenesandthedifferentiationofembryonicanlagenandthenofthehighdegreesofconservationoftheseverygeneshavereceivedthemostattention.Thesediscoveriesalsocontributed(falsely,aswehaveseen)totheimpressionthatevo-devobeganwiththeHoxstory.Amongtheideasthatemergedinthecontextoftheseinvestigationsarethenotionofagenetictoolkitfordevelopmentasetofregulatoryelementsthatareinvolvedinthedevelopmentofthemainfeaturesofanimalbodies,suchassegmentalpatterningandaxisformationandtheproposaltoreconstructaso-calledUrbilateriaastheancestralconditionofallhigheranimals.Thelattercombinestheinsightsofthecomparativeprogramwiththediscoveryofthegenetictoolkitfordevelopment(seeCarrolletal.2005foranoverview).4.Thetheoreticalandcomputationalprogramwithinevo-devo.Evo-devohasalsotriggeredalotoftheoreticalresearch,especiallyduringthelastdecades.Thispartofevo-devoisonlybecomingmoreprominentasnewlydevelopedformalandmathematicalapproachesbegintoaddmorerigortolong-standingconceptualideas.Thetheoreticalprogramisespeciallyimportantasithasthepotentialtoprovideconceptualunificationtootherwisediverseexperimentalapproaches(seelaterdiscussion).Italsorepresentsadisciplinarycounterweighttotheprograminevolutionarydevel-opmentalgenetics.Whilethelatterislargelydominatedbyresearcherstrainedinmolecularordevelopmentalbiology,theformeristhedomainofevolutionarybiologists,whooftenrefertotheirapproachasdevelopmentalevolutionordevo-evo.InthiscontextGu¨nterWagner,atrainedevolutionarybiologistandeditor-in-chiefoftheJournalofExperimentalZoology,PartB:MolecularandDevelopmentalEvolution,oneoftheleadingevo-devojournals,alwaysspeaksofdevelopmentalevolution.Thesedifferencesaremorethanjustsemantics,astheyalsocorrespondtodifferentepistemologicalconvictionsandexplanatoryframeworks.Simplyput,developmentalbiologiststendtobemoreinterestedinCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology351structuralandtypologicalexplanationsbasedonmolecularandcellularmechanisms,whileevolutionarybiologistsfocusmoreondynamicprocessesonapopulationandspecieslevel(seealsoHall2000foradiscussionaboutthedifferencesofevo-devoanddevo-evoandAmundson2005foradetailedaccountofthedifferentepistemologicalandexplanatoryframeworks).Theintegrationofthesetwoapproachesisanythingbutstraightforward,althoughrecenttheoreticaldevelopmentsthatincludeworkongenotype-phenotypemapsincludingquestionsofhowbesttocharacterizephenotypespaceandmorphospace,theoreticalandempiricalanalysesofmodularityandrobustness,phenotypicplasticity,lifehistory,aswellasevolvability,allcontributetoaframeworkthatmightjustproveflexibleenoughtointegratethedifferentempiricalandtheoreticaltraditions(seeCallebautandRaskin2005,SchlosserandWagner2004,WestEberhard2003,Mu¨llerandNewman2003,HallandOlson2003).Theoreticalworkisalsogreatlyaidedbynewdevelopmentsincomputationalmethodsandrepresentations.Thedatabasesofthevariousgenomeprojectshavebeenindispensableforidentifyingdevelopmentallyactivegenesandestablishingtheirevolutionaryhistory.Asfunctionalannotationsofgenesinthesedatabasesincreaseandgeneontologiesbecomemoresophisticated,itwillsoonbepossibletoextractthekindofinformationaboutdevelopmentalgenesthatisnecessaryforamoredetailedunder-standingoftheevolutionofdevelopmentalsystems.WhatwehavelearnedforHoxandrelatedgenes,howtheirexpressiondomainsshiftindifferentspecies,andhowthiscorrelatedwithmorpholog-icalchanges,forinstance,orwhattheconsequencesofcertainduplicationeventshavebeen,willsoonbeavailableforalargenumberoftranscriptionfactors,signalinggenes,andreceptorproteins.Inaddition,computationalreconstructionsofgeneexpressionpatternsindevelopingembryoswillorganizedatainawaythatwillgreatlyaidmathematicalmodelingofdevelopmentalsystems.OneexampleofthisapproachisAdamWilkinssanalysisoftheevolutionofgeneticpathwaysindevelopmentalsystems,whichsuggeststhatcertainchanges,suchasupstreamadditionofcontrolelements,aremorelikelythanothers(Wilkins2002).Tosumup,thetheoreticalandcomputationalprogramwithinevo-devoisabouttogetagreatboostfromtheincreasingsuccessofexperimentalapproaches.Interestingtimesthuslieahead.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n352manfredd.laubichlerAnalyzingthedifferentresearchprogramswithinevo-devoisjustonewaytocapturethediversity,excitement,andpotentialofthefield.Anotherwaytohelpusunderstandwhyevo-devoissuchacentralpartofcurrentorganismalbiologyistoinvestigatesomeoftheconcreteresearchproblemsthatmakeupthecoreofpresentdayevo-devo.3.researchquestionsinevo-devoTherearemanyspecificquestionsthatarecurrentlyinvestigatedwithintheframeworkofevo-devosowewillhavetobeselectivehere.Butgenerallymostoftheindividualproblemswillbecon-nectedtooneofsixareasofresearch(Laubichler2005,Mu¨ller2005).1.Theoriginandevolutionofdevelopmentalsystems.Thisquestionfocusesontheevolutionarytransformationofthedevelop-mentalsystemsthemselves.Developmentalsystems,asanyothercharacteroforganisms,undergoevolutionarytransformations.Researchinthisarearevealsthemodulararchitectureofdevelop-mentalsystemsandinvestigatestheirrobustnessandhowdifferentdevelopmentalmodulesarecombinedandregulated.2.Theproblemofhomology.Theproblemofhomologyisoneofthecentralquestionsofallofbiology.Itisoftenseparatedinahistoricalhomologyconcept,usedmainlyinphylogeneticanalyses,whichdescribesthedistributionofhomologues,andabiologicalhomologyconcept,whichattemptstoexplaintheexistenceofhomologiesinthefirstplace.Anevo-devoframeworkemploysbothnotionsofhomologyandtriestointegratethem.3.Thegenotype-phenotyperelation.Mappingthegenotypeontothephenotypehasemergedasamainproblemwithinevolutionaryandquantitativegeneticsduringthelastdecades.Itisalsothequestionthatmostdirectlyinvolvesdevelopmentalmechanisms.Foralongtimepopulationgeneticmodelsassumedthatdevelop-mentdoesnotaffectthemappingofgenotypicontophenotypicvariationinanyimportantwaydevelopmentwasthustreatedasaconstant.Asthispositioncannolongerbeupheld,investigationsintotheformalpropertiesofthegenotype-phenotypemaphavebecomeamajorfocuswithinevo-devo.4.Thepatternsofphenotypicvariation.IthaslongbeenknownthatpatternsofphenotypicvariationarehighlyclusteredandCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology353constrained.Explanationsofthisphenomenonhavealwaysincludedreferencestothedevelopmentalsystem,mostlyintheformofdevelopmentalconstraintsthatlimitthepossiblephenotypicvariants.Recentlyithasalsobecomeclearthattheevolvabilityofcertainlineagesalsocruciallydependsontheexistenceofdevelop-mentalconstraints.Thisquestionisthusalsoamainconcernforevo-devoresearchers.5.Theroleoftheenvironmentindevelopmentandevolution.Morerecentlyenvironmentalfactorshavealsobeenincorporatedintotheevo-devoresearchprogram.Themorewelearnaboutthemolecularmechanisms,suchasDNAmethylationorendocrinedisruption,throughwhichtheenvironmentcaninfluencethephenotypicexpression,themoreitbecomesobviousthattheenvironmenthastobecomealargerpartofexplanationswithinevo-devo.6.Theoriginofevolutionarynovelties.Explainingtheoriginofnovelphenotypictraitshasbeenoneofthemajorchallengesofevolutionarybiology.Asevo-devooffersthefirstintegratedperspectivethathasthepotentialtocomprehensivelyaddressthisproblem,wewilldiscussthisquestioninmoredetail.4.theproblemofevolutionarynoveltiesThemajorchallengeforevolutionarybiologyistoexplaintheoriginofcomplexnovelstructuresandfunctions.DarwinalreadystruggledwiththisproblemintheOriginofSpecies.Whatgoodis5%ofaneye?isanoftenrepeatedquestionandeventodaytheideaofirreduciblecomplexityistakenasevidenceforintelligentdesignandfortheobviousshortcomingsoftheneo-Darwinianparadigmbasedonanunguided,unplannedprocessofrandomvariationandnaturalselection,whichsimplycannotbetrue.Partoftheproblemisthatproponentsofintelligentdesignarecon-sciouslymisrepresentingevolutionarybiologyasneo-Darwinismandaresystematicallyignoringalltheevidencethathasaccumu-latedoverthelasttwenty-someyearsinthecontextofevo-devo,evidencethathelpsusunderstandhowsomethingnewcanactuallyemergeinthecourseofevolution.But,asthecaseofintelligentdesignalsoshows,evo-devoandespeciallywhatitcanteachusabouttheoriginofevolutionaryinnovationsareofmorethanjustacademicinterest(seealsoKirschnerandGerhart2005).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n354manfredd.laubichlerSincethefirstdraftsequenceofthehumangenomewascom-pletedin2001,someofthesimplisticassumptionsabouttherela-tionshipbetweengenotypeandphenotypehavebeenchallengedbytherealizationthattheactualnumberofhumangenesisrelativelylow,onlyaround30,000,andthatmostofthesegenes(closeto19,000)arealreadypresentintinynematodeworms,suchasC.elegans.Manyoftheimportantregulatorygenesandtranscriptionfactors,suchastheHoxgenes,arealsohighlyconservedbetweenlineagesthathavebeenseparatedbyhundredsofmillionsofyears.Novelfeaturesarethusnotjustaconsequenceofnewgenesorevennewversionsofoldgenes.Whatthenaccountsfortheobviousphenotypicdifferencesbetweengroupsoforganismsandfortheemergenceofnovelstruc-turesinthecourseofevolution?Theshortanswertothisquestionisthatchangesinthedevelopmentalsystemsoftheseorganismsandmorespecificallychangesintheregulatorynetworksofgenesareresponsibleforthesedifferences.Inotherwords,thesamemechanismsthatleadtodifferentiationofcellsinthecourseofindividualdevelopment(ontogeny)alsoaccountforemergingdifferencesinthecourseofevolution(phylogeny).Intuitivelythismakessense.Allphenotypicdifferences,whethertheyarejustvariationsofacommonthemeorsomethingradicallydifferent,emergeduringthedevelopmentofindividualorganisms.Develop-mentalprocesseswillthusalwaysbetheimmediateorproximatecausesofphenotypicvariation.Stillseveralquestionsremain:Exactlyhowdodevelopmentalmechanismscontributetopheno-typicchanges,andhowcanweintegratesuchdevelopmentalexplanationsintothetheoreticalframeworkofevolutionarybiology?Thekeyideathathelpsusunderstandtheseissuesistheconceptofregulation.Developmentitselfisahighlyregulatedprocess.Howwoulditotherwisebepossiblethatdespiteconstantenvironmentaldisturbancestheoutcomesofdevelopmentaregenerallypredictablesuchthathumansgivebirthtohumansandseaurchineggsdevelopintoseaurchins?Whiledevelopmentalbiologyfocusesonindivid-ualdevelopmentalsequencesdescribingthetransformationfromsimple(afertilizedegg)tocomplex(anadultorganism)throughthedifferentiationofcellsandtheemergenceofanatomicalandhistologicalstructures,evo-devotriestounderstandhowintheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology355courseofevolutionsomethingnewcanemergewithintheseregu-lateddevelopmentalsequences.Butbeforewecanaddresspotentialexplanationsofsuchevolutionarynoveltieswefirsthavetodefinetheproblemmoreprecisely.Addressingtheproblemofevolutionaryinnovationsrequiresusfirsttodefineclearlywhatwemeanbynoveltiesand,second,todevelopasetofcausalhypothesesthatwillallowustoidentifythedevelopmentalchangesinvolvedintheemergenceofanevolu-tionarynovelty.IntheirseminalanalysisoftheproblemMu¨llerandWagnerdefinedamorphologicalnoveltyasastructurethatisneitherhomologoustoanystructureintheancestralspecies,norhomonomoustoanyotherstructureofthesameorganism(Mu¨llerandWagner1991,243).Whilethisrathergeneraldefinitionstillleavesopenmanydetails,itdoeshaveonepracticalimplication.Theproblemofidentifyingnoveltiesissquarelyplacedwithinthecomparativeprogramofevo-devoastheirrecognitiondependsonbothagoodphylogenyandadetailedassessmentofhomology.Settingaside,forthemoment,manyofthepracticalproblemsconnectedwithrecognizingnovelties(whicharesimilartotheproblemsofassessinghomology)wecanidentifysomeofthestepsrequiredforestablishingacausalhypothesisabouttheoriginofevolutionarynoveltieswithinthecontextofevo-devo.Thefirstquestionthatneedstobeaddressedis,Whatspecificdevelopmentalmechanismsareresponsibleforanewderivedcharacterstatethathasbeenidentifiedasanevolutionarynovelty?Answeringthisquestionrequiresthedetailedanalysisofthedevelopmentalmechanismsthatgenerateaspecificphenotypiccharacter.Itisthuspartoftheexperimentalprogramofevo-devo.Thesecondquestionbuildsonthisanalysis:Didthedevelopmentalmechanismsthatareresponsibleforthederivedcharacterstateoriginateatthesametimeasthischaracterstate?Thisisalreadyadifficultquestionthatrequiresustocomparethedevelopmentalmechanismsofancestralandderivedcharacterstates.Inmanycasesthiswillnotbepossible,asitisimpossibletoreconstructtheexactancestralconditionofdevelopmentalprocesses,especiallyifthetransformationinques-tionhappenedhundredsofmillionsofyearsago.Thesameproblemsalsoapplytothethirdandfourthquestions:Whatweretheexactdevelopmentalmechanismsresponsiblefortheinitialchangesinthecharacterstate?AretheobservedgeneticdifferencesbetweenCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n356manfredd.laubichlerthesetwodevelopmentalsystemssufficienttoaccountfortheobservedphenotypicdifferences?Theselasttwoquestionsfocusonthemechanisticdetailsofthechangesinthedevelopmentalsystemandtheextenttowhichobservablegeneticchangesprovideacom-pleteexplanationofevolutionarytransformations(Wagneretal.2000).Westillfacemanypracticaldifficultieswithmostcasesofevo-lutionarynoveltiescurrentlyunderinvestigation.Partoftheprob-lemcanbeattributedtotheselectionofexamplesandmodelorganisms,whichtendtofocusonmajormorphologicaltransfor-mations,suchasthefin-limbtransitioninearlyvertebrates.However,therearesomemodelsystemsthatdoallowustoaddressquestionsabouttheoriginofevolutionarynoveltiesexperi-mentallyaswellastheoretically.Forexample,socialinsectsdisplayaremarkablediversityinbehaviorrangingfromsolitarytoeusocial.Formanyofthesespeciesweknowtheirphylogeny,genetics,developmentalmechanism,aswellastheirphysiologicalandbehavioralrepertoire.Andwecanmanipulatethemexperi-mentallyinthelab,inseveralcasesactuallyinducingnoveltypesofsocialbehavioramongsolitaryspecies.Socialinsectsarethereforeanidealmodelsystemforthestudyofevolutionarynovelties.Thisworkisonlyjustbeginning,butwecanexpectthatitwillleadtomanyimportantinsightsintotheproblemofevolutionarynovelties.Anotherissuethatneedstobementionedhereiswhethergeneticdifferences(includingdifferencesinthegeneticpartsofregulatorynetworks)aloneprovideasufficientexplanationfortheoriginofevolutionarynovelties.Ourfourthquestionspecificallyaddressesthisissue.Bynowwehaveampleevidencethatepigeneticandenvironmentalfactors,whicharepartofthedevelopmentalsysteminmanyorganisms,playanimportantpartintheoriginofevolutionarynovelties,especiallyduringtheinitialphasesofchar-actertransformation.OnthebasisofsuchobservationsMu¨llerandNewmanhavesuggestedathree-stepmodelfortheoriginofevolutionarynoveltiesthatassumesthattheinitialemergenceofnewcharactersisoftencausedbyepigeneticandenvironmentalmechanismsthatarelaterstabilizedbyassociatedgeneticchanges(Mu¨llerandNewman1999,2003).Whilethisisstillarathercon-troversialidea,itcanbetested,especiallywithsocialinsects,amongwhichmanyemergentcolony-leveltraits,suchasdivisionoflaborCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology357andcastedistribution,areoftenaconsequenceofwithin-colonyinteractions,ratherthansimplemutations.5.evo-devoandtheproblemofinterdisciplinaryintegrationConceptualintegrationofratherdiverseresearchparadigmsisthemainchallengethatevo-devocurrentlyfaces.Therefore,itcaneasilybearguedthatthefutureofevo-devoasatrueinter-disciplinarysynthesiswilldependmoreontheoreticaladvancesthanonadditionalexperimentaldata.Asweallknow,appearancescanbedeceiving,andthedelugeofexcitingexperimentalresultsoverthelasttwodecadeshasinmanywayshiddenanunderlyingconceptualtensionthatwillhavetoberesolvedforevo-devotosucceed.InhisrecentbookRonAmundsonpointedoutingreatdetailhowdevelopmentalandevolutionaryexplanationsarebasedonratherdifferentepistemologicalfoundations(Amundson2005).TheformerarerootedinwhatAmundsonreferstoasastructuralistparadigmthatisbasedonacausalunderstandingofgeneralmolec-ularandcellularmechanisms,suchasmoleculargradients,cell-surfaceinteractions,andcell-cellsignaling,whilethelatterarepredicatedonamathematicalformulationofunderlyingpopulation-leveldynamics,suchasthereplicatorequation,thegeneralizedselectionequations,orstochasticprocesses,suchasrandomgeneticdrift.TheearliersuccessoftheModernSynthesisofthe1930sand1940s,whichisoftenheldupasamodelforanemergingevo-devosynthesis,wasbasedonmatchingpatternsofthetransmissionofgeneticinformation,ofspeciationandadaptation,andofthefossilrecordwithdynamicalprocesseswithinpopulations(seeMayrandProvine1980foranoverviewofissuesrelatedtotheModernSynthesis).Somethingsimilarwillhavetobeaccomplishedforevo-devo.Observedpatternsofmorphologicalandbehavioralevolutionwillhavetobematchedwiththepossibilitiesandconstraintsofdevelopmentalsystemsandthedynamicalprocesseswithinpopu-lations.Suchintegrationwilldependonaconceptualframeworkandassociateddynamicalmodelsthatadequatelyrepresentboththeactualphenomenatobeexplained(patternsofmorphologicalandbehavioralevolution)aswellastheunderlyingdynamicsthatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n358manfredd.laubichlergeneratedthem(seealsoWagneretal.2000,WagnersseriesofeditorialsinMolecularandDevelopmentalEvolution[2000,2001],WagnerandLarsson2003,Laubichler2005).Whatformcouldsuchintegrationtake?Aswehaveseeninourbriefhistoricaloverview,developmentalmechanismshavealwaysfeaturedprominentlyinattemptstoexplainpatternsofphenotypicdiversityandtransformation.Whetherthesewereconceptualizedaslawsofvariation(Darwin),recapitulationandterminaladdi-tion(Haeckel),homeoticmutationsandhopefulmonsters(Goldschmidt),orgeneticassimilation(Waddington),thegeneralargumenthasalwaysbeenthesame:phenotypicchangeshavetoariseduringontogeny;therefore,changesinontogenywillhavetoberesponsibleforobservedpatternsofphenotypicevolution.Thedetailsandspecificconcernsoftheseexplanationsdiffered,ofcourse,butthisvariationdoesnotdetractformtheirunderlyingsimilarity.Evo-devofallswithinthesameexplanatoryparadigm,butitalsodiffersinseveralimportantways.Forone,ourcurrentunderstandingofthemolecular,cellular,genetic,andepigeneticmechanismsofdevelopmentismuchmoredetailedandourabil-itiestomanipulatedevelopmentalsystemsexperimentallyhaveadvancedratherdramatically.Thisincreasedunderstandingofdevelopmentalmechanismshasledtoamorerefinedconceptualrepresentationofdevelopmentthatisnolongerbasedonsimplemechanicalforces,suchastheactionsofasomewhatmysteriousorganizer,orgradientsofmolecules.Ourcurrentunderstandingofdevelopmentaldifferentiationincludescomplexcausalpath-waysandinteractionsbetweengeneticandepigeneticregulatorynetworks.Wehavealreadyindicatedseveraltimesthatregulationhasbecomeoneofthecentralconceptsinbothdevelopmentalbiologyandevo-devo.Indeed,manyscientistsarguethatmodificationsintheregulatorynetworksso-calledregulatoryevolutioncanaccountfortheobservedphenotypictransformations(see,forexample,Carrolletal.2005,Davidson2001,2006,Wilkins2002).Whilechangesintheregulatorynetworksarecertainlyanimportantpartoftheexplanationofpatternsofphenotypicdiversity,ourexplanatoryschemawillhavetobeexpanded.Firstwewillhavetoestablishamoreadequateconceptualframeworkforwhatwewanttoexplain.EvolutionarynoveltiesareaprominentpartoftheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEvolutionaryDevelopmentalBiology359evo-devoexplanandum,butrelatedconceptssuchasfacilitatedvariation,developmentalconstraints,modularity,robustness,andevolvability,allconceptsthathaverecentlybeenanalyzedwithinthecontextofevo-devo,willhavetobefullyintegratedintoitsconceptualframework(see,forinstance,CallebautandGutman2005,KirschnerandGerhart2005,MaynardSmithetal.1985,SchlosserandWagner2004).Theseconceptsarealsotiedintothedifferentresearchprogramswithinevo-devo,Adetailedepistemo-logicalanalysis,ofthesortprovidedbyRonAmundson,oftheseprogramsandtheirunderlyingassumptionsisthereforecrucialforthefutureofevo-devoasasyntheticenterprise.Butbeforeweattemptsuchasynthesisweshouldalsocon-siderwhatwemeanbysynthesis,whatexactlyshouldbesynthe-sized,andwhatweexpectfromsuchasynthesis.TheModernSynthesiswassuccessfulbecause,afteraninitialphaseofintensediscussions,itemergedwithaclearexplanatoryagendahowpopulationgeneticmodelscanbeusedtoexplainadaptationandspeciation.Andeventhoughthesesimpleprinciplesortheirunderlyingassumptionswerealmostnevertrueforanyrealcase,theyneverthelessanchoredaveryproductiveexperimentalandtheoreticalresearchenterprise,whichultimatelyledtoourcurrentperiodofmethodologicalandconceptualpluralism.IfwetakethehistoricallessonoftheModernSynthesistobethattheroleofasynthesisistoprovideaconceptualfoundationfordifferentresearchprogramsthatwillultimatelyexplorethefuzzyedgesandareasbeyondthecoreofthesynthesis,thenweadvocateanopencon-ceptionofsynthesisratherthanaclosedviewbasedonintegrationofexistingparadigms.Seenthatwaytheroleofanevo-devosynthesiswouldbetopro-videasetofcoreconceptsandassumptionsthatallowfurtherresearch,realizingthatthisresearchwilleventuallytranscendtheexplanatoryframeworkofthecurrentsynthesis.Howwouldsuchanopen-endedsynthesislook?Itisprobablytooearlytotell,butitwillhavetobebasedonaconceptualstructurethatallowstheintegra-tionofdevelopmentalmechanismsintoevolutionaryexplanationsatahigherlevelofresolutionthanthecurrentideasaboutregulatoryevolutionandtheevolutionofthegenetictoolkitsuggest.Itwillalsohavetodevelopamorecomprehensiveconceptionofmechan-isticcausesforbothdevelopmentandevolutionthatincludesCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n360manfredd.laubichlergenetic,epigenetic,andenvironmentalfactors.Anditwillhavetodevelopasetofparadigmaticmodelsystemsthatwillallowustostudythesequestions.Aswehaveseen,therearesomeveryprom-isingnewmodelsystemscurrentlydevelopedthatmightjustbetherighttoolforthejob.Inanycase,theseareexcitingtimesforbothscientistsandphilosopherstocollaborateonsomeofthemostfundamentalproblemsofbiology.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\njasonscottrobert19MolecularandSystemsBiologyandBioethicsMolecularbiologyhassetitselfthetaskoflookingforthefundamentalpieceswithwhichthebiologicaljigsawistobeputtogether.Notsurprisingly(butwithsurpris-ingefficacy),ithasfoundmanyofthem,andtherearecertainlymoretocome.Oncefound,thesepiecescanbearrangedonapagenexttooneanotherinareasonablesequence,and...Behold!Anorganism!Well,notquite.CohenandRice1996,239Thephilosophyofmolecularbiologywas,foratime,entirelypre-occupiedwithreductionandreductionism:primarilythereductionofclassicalgeneticstomoleculargenetics(Kitcher1984,Waters1994,Sarkar1998),butalsoandmorerecentlythereductionofcomplexorganismalphenotypestogenes(Rosenberg1997,Sarkar1998).Whiletheseremainofsubstantialinterest,somenewareasofinteresthavealsoemerged,includingphilosophicalattentiontomolecularmechanisms(Machamer,Craver,andDarden2000,DardenandTabery2005)andmathematicalmodels(Keller2002,Sarkar2005).In-depthfocusontheintricatedetailsofthescienceisincreasinglycommonplace(e.g.,Schaffner2000,Burian2004,Sarkar2005).Molecularbiologyhasalsoprovedtobeofphilosophicalinterestnotonlyforitsownsake,butalsointheserviceofmolecularexplanationsofevolution(e.g.,Burian2004),disease(e.g.,Kitcher1996),andbehavior(e.g.,Schaffner2000),interalia.Justasthephilosophyofmolecularbiologyhaschangedinthepastfewdecades,sotoohasthescience,especiallywiththeintro-ductionofnewtoolsandtechniquesandthenovelopportunitiesaffordedbyadvancesingenomesequencing,computermodeling,361CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n362jasonscottrobertandbioinformatics.Indeed,somehaveproclaimedthatthe’omicseraheralds‘‘theendofmolecularbiologyasweknowit’’(Laubichler2000,287).Whilereductionismremainsofcentralconcern,itisevidentthatwitheverysuccessfulexperimentandnewdiscoveryatthemolecularlevelappearsanewpuzzleattheleveloforganisms,forhowtorelatefundamentaldiscoveriestophysiologicaloutcomesremainsalmostasopaqueasever.Thesentimentevidencedintheepigraphhasbeencapturedaswellbymanyotherscientistsandcommentators,includingEvaNeumann-Held:‘‘Sofar,biologycandescribeorganismsdowntothemolecularlevelofgenes.However,theinteractionsofgenes[withothergenesand]withother,non-geneticcomponentstoformanorganismisfarfrombeingunderstood....Inthedescriptionoforganisms(moregen-erally:ofsystems),biologystillhastoperformtheintegrativepart’’(Neumann-Held1999,107;seealsoRobert2004).Whattodowithgenomesequencedataisacaseinpoint.WhentheHumanGenomeProjectwasfirstproposed,itsproponentspromiseddramaticnewinsightsintohumanbeings,insicknessandinhealth,aswellasasteadysupplyofmedicaltreatmentsandmetaphysicalrevelations(see,e.g.,Cook-Deegan1994,NelkinandLindee1995).Skepticspointedout(correctly)thatthesepromiseswouldnotbeeasilyfulfilled,eveniftheseeminglyimpossibledreamofcompletelysequencingthehumangenomecouldbeachievedinthefirstplace(see,e.g.,Lippman1992).Remarkably,thelatterdreamwasachieved,withthepublicationin2000ofdraftsequencesandtheannouncementin2003ofthefinalizeddraft(Collinsetal.2003).Butunderstandingthehumangenomesequence(functionalgenomics),incomparisonwiththoseofotherorganisms(comparativegenomics),remainsanoutstandingtask.Asseveralearlycritics(e.g.,Lewontin1992,TauberandSarkar1992)understoodfullwell,humangenomevariabilitywouldprovecrucial,fornosinglesequencedgenome(orcompositeofgenomes)couldbemeaningfullysaidto‘represent’thediversityofgeneticvariationthatcharacterizesanyspecies.EpidemiologistsarethusactivelysamplingDNAfromvolunteersinthequestforpoly-morphisms(variants)bothbetweenandwithingroups,andbioin-formaticiansandothersarecontinuallydevelopingnewtoolsforanalyzingthevastquantitiesofgenomicsdatageneratedthroughsuchresearch.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandBioethics363Whilemakingsenseofthismorassofdataremainsadauntingtask,actuallygeneratingtreatmentsfordiseasesandimprovinghumanhealthoutcomesonthebasisoftheHumanGenomeProjectaremoredauntingstill.Ononehand,thisisbecauseofthecom-plexityofhealth(RobertandSmith2004):individualhealthanddiseasearesignificantlyaffectedbydevelopmental,ecological,andsocialcomponentsthatarenotreducibletogenesorparticularlyamenabletogeneticintervention;atthepopulationlevel,social,economic,anddemographicfactorsexplainhealthoutcomesfarbetterthangeneticvariations.Accordingly,identifyingindividualgeneticrisksmayhaveonlyminimaleffectsonhealthoutcomes,whetheratindividualorpopulationlevels.Ontheotherhand,organismaldevelopmentisfarmorecomplicatedthanglibpro-nouncementsaboutgeneticinstructionsandblueprintswouldsuggest(Robert2004):the‘pathway’fromgenetophenotypecom-prisesgene-gene,cell-cell,cell-tissue,andenvironmentalinterac-tions;epigeneticeffects;anddevelopmentalstochasticity;geneeffectsmaybepleiotropic(singlegeneshavemultiple,divergenteffects);manytraitsareepistatic(correlatedwiththeactivityofmanygenes);inshort,developmentitselfasadynamic,temporalrepertoireofprocessesdeterminesmanyaspectsofthemany-manygenotype-phenotyperelation.Accordingly,inmostcasesitisunclearhowtoproceedfromagenomesequencetotheidentifica-tionofgenestoafullunderstandingoftheaetiologyofaphenotype–andfromanyofthesetodeterminehowtointervenetopreventorpromoteitsmanifestation.Thatdevelopmentiscomplexshouldcomeasnosurprisetobiologists–thoughsomewouldprefertoignoredevelopmentalcomplexityinfavoroftherelativesimplicityofdevelopmentalgeneticsandmolecularbiology(whichthemselvesarefar,ofcourse,fromsimple).Indeed,uponannouncingthecompletionoftheHumanGenomeProject,theNationalHumanGenomeResearchInstitutealsoannounceditselaborate‘‘VisionfortheFutureofGenomeResearch’’(Collinsetal.2003).Thevisionincludesaseriesof‘‘grandchallenges’’forgenomicsresearch,intendedas‘‘bold,ambitiousresearchtargetsforthescientificcommunity’’(Collinsetal.2003,2)inbasicscience,health-relatedappliedscience,andpolicyscienceandethics.Thesegrandchallenges–liketheHumanGenomeProjectonwhichtheybuild–haveastheiraimtheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n364jasonscottroberttranslationofgenomicsdataintobiologicalunderstandingandhumanhealthoutcomesinthegenomic(or‘postgenomic’)era.Whethertheseaimswillbeachieveddependsonthedevelopmentofamoreintegrativebiology,andonbiologiststakingseriouslythechargetounderstandorganismallifeinallitscomplexity.frommolecularbiologytosystemsbiology?Systemsbiologyhasemergedasone‘field’withinwhichbiologistsandotherscientistsareattemptingtomakemolecularbiologywork‘‘intherealworldoftheorganism’’(CohenandRice1996,251).Giventhearrayofmoleculardataawaitinginterpretation,andinthecontextofagrowingawarenessofthelimitsofcertainkindsofsimplifyingstrategiesindealingwiththecomplexityoflivingorganisms,molecularbiologistshaveintensifiedcollaborationswithcomputerscientists,engineers,organismalandintegrativebiologists,andmanyothersunderthebannerof‘systemsbiology’(Fujimura2005).Whilethattermmayappeartobeafuzzy-soundingbuzzwordtosome,nonethelesstherearenowsystemsbiologyresearchcenters(suchastheInstituteforSystemsBiologyinSeat-tle),Ph.D.programs(asintheDepartmentofSystemsBiologyatHarvardMedicalSchool),andjournals(includingNature’srecentlaunchofMolecularSystemsBiology).Indeed,systemsbiologywasrecognizedbytheeditorsofScience(vol.310,23December2005)asarunner-upforthe2005BreakthroughoftheYear.Butwhatsystemsbiologyisisnotalwaysclear.ANatureedi-torialisthasprovidedasomewhatmacabrethoughaptresponse:‘‘Whatisthedifferencebetweenalivecatandadeadone?Onescientificansweris‘systemsbiology’.Adeadcatisacollectionofitscomponentparts.Alivecatistheemergentbehaviourofthesystemincorporatingthoseparts’’(Anonymous2005,1).Generically,then,asystemsapproachtobiologyreferstotheinterdisciplinarystudyofmanifoldandcomplexinteractionsamongDNA,RNA,proteins,cells,andbiomodulesofvarioussortsthatconstitutelivingentities.Asurveyofrecentreviewarticlesandcommentariesrevealsthatthename‘systemsbiology’isappliedtoanumberofdistinctresearchprograms,eachworthyofphilosophicalscrutinyinitsownright(see,e.g.,Hartwelletal.1999,Ideker,Galitski,andHood2001,KitanoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandBioethics3652002a,b,Auffrayetal.2003,Grant2003,Hoodetal.2004,Aderem2005,Kirschner2005,Moore,Boczko,andSummar2005,Sorger2005,Strange2005;cf.Newman2003andFujimura2005).Asystemsapproachhasbeenadvocatedandpursuedinmanydomains–forinstance,indevelopmentalbiology(e.g.,GilbertandSarkar2000,Robert2004),drugdiscovery(Butcher,Berg,andKunkel2004),neuroscience(Grant2003,WulffandWisden2005),physiology(Strange2005),psychology(Oyama1985),stemcellbiology(Robert,Maienschein,andLaubichler2006),andtoxicology(WatersandFostel2004).Theunderlyingappealofasystemsapproach,howeverrealized,isclear,asevidencedinthefollowingdepictionsofsystemsbiology:Systemsbiologyisanemergingdisciplinefocusedontacklingtheenormousintellectualandtechnicalchallengesassociatedwithtranslatinggenomesequence[data]intoacomprehensiveunderstandingofhoworganismsarebuiltandrun....Systemsbiologyisintegrativeandseekstounderstandandpredictthebehavioror‘‘emergent’’propertiesofcomplex,multicomponentbiologicalprocesses.(Strange2005,C968)Newtechnologieshaveinundatedresearcherswithadelugeofinformationongenes,proteins,cellulardynamics,andorganisms’responsestomuta-tionsandtheenvironment.Buttheyhaven’texplainedwhatmakeswholeorganismstick.Systemsbiologistsaretakingonthatchallenge,relyingheavilyonmathematicsandstatisticstointegratedataintoamorecompletepictureofhowbiologicalnetworksfromcellstowholeorganismsfunction.Theyarebuildingmodelsandmakingpredictionsabouthowbiologicalsystemswillbehave;theultimategoalistounderstanddeepmysteries–suchashowcellsdivide,animalsdevelop,plantsflower,andhumansbreathe.(Pennisi2003,1646)Thegoalofsystemsbiologyistoofferacomprehensiveandconsistentbodyofknowledgeofbiologicalsystemstightlygroundedonthemolecularlevel,thusenablingustofullyintegratebiologicalsystemsintomorefundamentalprinciples....System-levelknowledgeshouldbegroundedinsuchawaythatthesystemiscomposedofmolecules;andmoleculesfollowthelawsofphysics.However,howasystemoperatescanonlybedescribedbyasetoftheoriesthatfocusesonsystem-levelbehaviors.Thepointisthatsuchtheoriesmustreflecttherealitiesofbiologicalsystemsandmolecules,withoutabstractingtheessentialaspectsofbiology.(Kitano2002a,2).SystemsbiologyisacomprehensivequantitativeanalysisofthemannerinwhichallthecomponentsofabiologicalsysteminteractfunctionallyoverCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n366jasonscottroberttime.Suchananalysisisexecutedbyaninterdisciplinaryteamofinvesti-gatorsthatisalsocapableofdevelopingrequiredtechnologiesandcompu-tationaltools.(Aderem2005,511)Thefeaturesthatdistinguishthenewfieldofsystemsbiologyare:first,thegroundinginmolecules,primarilyatthelevelofthegene;second,thegoalofsystematic,comprehensiveandquantitativeanalysisofallofthecompo-nentsthatconstitutethesystem;third,theverticalintegrationofanalysistodevelopastructuretothesystem;fourth,thesimulationandcomputationalapproachestomodellingthesystem,andthebioinformaticsapproachesofdatahandling.(Grant2003,577)Thecarefulreaderwillhavepickedoutsomecommonthemesacrossthesepassages.Forinstance,onegeneralaimofsystemsbiologyistomovebeyondanalysisofindividualandisolatedcom-ponentsofasystemtowardquantification,modeling,andcompre-hensionoftheintegratedorganizationandinteractionofthecomponentsaspartofasystem.Understandingcomponentsofsystemswillremaincriticallyimportantinsystemsbiology(asanywhereinbiology),butfocusingonthestructuralandinteractivepropertiesofsystemswillhelptocompleteourunderstandingand1improveourpredictiveabilities.Tothisend,Strange(2005,C968)stressesthat‘‘asystemslevelcharacterizationofabiologicalprocessaddressesthreemainquestions.First,whatarethepartsofthesystem(i.e.,thegenesandtheproteinstheyencode)?Second,howdothepartswork?Third,howdothepartsworktogethertoaccomplishatask?’’Butthecarefulreaderwillalsohavenoticedsomeimportantdiscontinuitiesinthesevariouscharacterizationsofsystemsbiol-ogy,ofwhatitentails,howitproceeds.Forinstance,thesepassagessuggestboththatthefieldhasdevelopedprimarilyasapositiveresponsetothechallengeofdealingwithallthenewdatafrom‘‘’omics’’research(genomics,proteomics,metabonomics,etc.),andthatsystemsbiologyhasemergedprimarilyinresponsetothelimitsofreductionisminunderstandinglivingorganisms(FribouletandThomas2005;cf.theessayscollectedinHullandVanRegenmortel2002).Onthelatterview,howeverproductivereductionismhasprovedtobe,livingorganismsarenotentirelyreducible,andsonovelapproachesarerequirednotonlytomakesenseofthemo-leculardata,butalsotointegratedatafromotherlevelsofanalysis.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandBioethics367Thisisnottosaythatsystemsbiologistsareantireductionistsoremergentists–thoughsomearesoinclined(e.g.,Aderem2005,FribouletandThomas2005),othersmaintainamethodologicalcommitmenttoreductionismwithinasystemsbiologyframework(e.g.,Idekeretal.2001,Sorger2005).Thesetensionssignaltheneedforcriticalanalysisbyhistoriansandphilosophersofbiology,aswellassciencestudiesscholars,ideallybasedinrichanddetailedcasestudiesofparticularexemplarsofsystemsbiology.Ofspecialinterestwillbehistoricalanalysisof‘holistic’understandingsoforganismsinbiologyandmedicine(Laubichler2000)andconceptualarticulationofwhatisnew(andwhatisnot)insystemsbiology.fromsystemsbiologytosystemsbiomedicine?Ingeneral,systemsbiologyinvolvessynthesizingandintegratingbiologicaldataintomathematicalmodelsthatsimulatethebehaviorofbiologicalsystems(organisms,diseases,etc.)andallowforpredic-tionsoffuturestates.Thisideathatsystemsbiologywillcontributetogreaterpredictabilityunderwritesasystemsapproachtobiomedi-cine,whereindrugdiscoveryanddesign(NicholsonandWilson2003,Aderem2005,Strange2005)promisetobe‘rationalized’andtreat-mentspersonalizedaccordingtodynamicmodelsofdiseasemani-festationandprogression(Hoodetal.2004,WestonandHood2004).Thesearenotnewexpectations.Theyarethesortsofclaimscommonlymadeaboutbiotechnologiesingeneralandgenomicsresearchinparticular.Theideaofmolecularmedicineinparticularhasgeneratedandcontinuestogeneratemuchexcitement.AstrongmotivationforundertakingtheHumanGenomeProjectwasthepromisethatgenomicswill‘‘profoundlyalter’’medicine,byshiftingfromtreatmentofacutediseaseto‘‘preventionbasedontheiden-tificationofindividualrisk’’,forinstance,andbytailoringphar-maceuticalstoindividualgenotype(see,e.g.,GuyerandCollins1993andvanOmmen,Bakker,andDunnen1999).Issystemsbiol-ogyjustthelatestinanever-endingstringofoverhypedbiomedicalfads?Oristherefinallysomesubstancebehindtherhetoric?Mycautiousresponseis,alittleofboth.Whilethereisnodoubtthatsystemsbiologyisamongthecurrentpantheonofbio-logicalfetishes(systemsbiology‘‘promisestorevolutionizeourCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n368jasonscottrobertunderstandingofcomplexbiologicalregulatorysystemsandtoprovidemajornewopportunitiesforpracticalapplicationofsuchknowledge’’[Kitano2002b,1664]),thereisalsoreasonforoptimismthatthisinterdisciplinary,integrativeendeavorwillprovebiologicallyandbiomedicallyfruitful.Thesourceofmycautiousoptimismismolecularbiologists’apparentrealization(finally!)thatdevelopmentalsystems,suchaslivingorganisms,aredynamic,temporal,andmutableratherthanstatic(cf.GilbertandSarkar2000).Andwhiletheyareinternallycomplexandheterogeneous,organismsarealsoembeddedinexternallycomplexandheterogeneousenvironmentsthatcanjointlyandseverallyhavedramaticimpactsondevelopment,phy-siology,andbehavior(cf.vanderWeele1999).Tobegintotaketheseconsiderationsseriouslyinexperimentaldesign,modelbuilding,andtheoryelaborationistobegin,perhaps,torealizetheambitionsofmolecularandsystemsbiomedicine.Whethertheseambitionswillberealizeddependsonalargenumberofconsiderations,includingthewilloffundingagenciesanduniversities,andtheingenuityandcollaborativeabilitiesofbiolo-gistsandotherscientists.Italsodepends,inpart,onconstructivecollaborationswithanewbreedofbioethicisttoassessthesocietaldesirabilityofnoveldevelopments,therisksandbenefitsofpartic-ulareventualities,andtheappropriateapplicationofnewknowledgeintheserviceofhuman(andhumane)ends.Manyphilosophersofbiologyhavenointerestinethics;manybioethicistshavenointerestinthedetailsandcomplexitiesofbiologicalscience;and,todate,muchofthebioethicalliteratureongenetics,genomics,andmolecularbiologyhasbeenunfortunatelysuperficialinitstreat-mentoftherelevantscience.Butanemergingcohortofhybridscholarsengaginginbothconceptual(biophilosophical)andnor-mative(bioethical,political)analysismaybeabletorestructureandreinvigoratescientificandpublicdebatesaboutdesirableoutcomesofscienceandtechnology(e.g.,Kitcher2001,Maienschein2003,RobertandBaylis2003,Robertetal.2006).fromelsitosystemsbioethics?WhentheHumanGenomeProject(HGP)wasinitiatedintheUnitedStates,itsproponentsdedicatedaremarkable3percentoftheHGP’sCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandBioethics369totalannualresearchbudgettofundanEthical,Legal,andSocialImplicationsprogram(ELSI).ProgramssimilartoELSI,thoughsomewhatlesswellfunded,wereestablishedinCanada(theMedical,Ethical,Legal,andSocialImplicationscomponentoftheCanadianGenomeAnalysisandTechnologyprogram),theUnitedKingdom(throughtheWellcomeTrust,forinstance),andelsewhere.IntheUnitedStates,atleast,ELSIhadbothresearchandeducationalmandates.Intermsofresearch,theELSIprogramidentifiedseveralprincipalethical,legal,andsocialissuesforinitialstudy:theprivacyofgeneticinformation,theprovisionofsafeguardsagainstgeneticdiscrimination,theprotectionofparticipantsingeneticsresearch,andthesecureintroductionofgenetictestsintoclinicalmedicalpractice(Juengst1991).AgreatdealofresearchwasgeneratedthroughtheELSIprogram,someofitquitegood.Ofcourse,ELSIdidnotfundallofthebio-ethicalworkundertakeninrelationtotheHGP,butthroughtheELSIprogram,NorthAmericanphilosophers,ethicists,historians,sociologists,andclinicianswereabletoobtainunprecedentedfundingtoinvestigatethelikelyethical,legal,andsocialimpactoftheHGP.YetafterfifteenyearsoftheprolificELSIprogram,thereisstillnoendinsightinresolvingtheethicalproblemsraisedbytheHGP.Inpart,thisisbecauseELSIscholarshavetendedtofocusonissuesrelatedtoMendeliangenetics,andlesssoongenomics(orevendevelopmentalgenetics).Moreover,thefocushasbeenonraresingle-genedisordersratherthanmuchmorecommonconditionswithmultifactorialaetiologies.Finally,manyELSIanalyseswereatomistic(inthesenseofnotbeingintegratedwithlargerissuesinthephilosophyofbiology,philosophyofmedicine,healthtechnol-ogyassessment,orhealthpolicy)andreductionistic(inthesenseofbeingportrayedinsimplisticterms,suchasfalsedilemmasanddichotomies).LiketheHGPitself,theseELSIendeavorswerenecessarilypartialatbest–agoodstartingpoint,perhaps.Justasbiologistsarenowmovingtowardamoreintegrativesystemsbiology,perhapssotooshouldbioethicistsmovetowardamoreintegrative‘systemsbioethics’(Robertetal.2006).ResearchinsystemsbioethicswouldbeemergentfromELSI-styleresearch,asintegratedwithstudiesinthehistoryandphilosophyofbiologyandsocialandpoliticalstudiesofscienceandtechnology,andemployingavarietyofmethodsfromCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n370jasonscottrobertthehumanitiesandsocialsciences(especiallyGustonandSarewitz’s2002‘real-timetechnologyassessment’).Itwouldgeneratenewwaystoframe,explore,understand,andaltermoraldimensionsofscientificresearchbydefiningrelevantconcepts,interests,andvalues;probingtheirnature;andestablishinganunderstandingofthedynamic,interactiverelationsamongthedisparatecompo-nentsoftheparticularbioethical‘system’understudy.Thiswouldthenserveasabasisforproactivedeliberationaboutscientific,ethical,andpoliticalissuestogetherandinteractively,aimedatmakingbioethicsworkintherealworldofcomplex,pluralisticcivilsocieties.Thus,onecriticalaimofsystemsbioethicswouldbetomakenormativeandconceptualanalysisactuallymatterinpracticeandinpolicycontexts.Thismightbeachievedthroughthecollaborativeinteractionofmethodsfrommultiplefieldsofinquirytointegratekeyconceptsrelevanttoajointpublicandscientificdiscourseaboutbiologicalresearch.Buttranslatingbioethicsresearchintopolicyoptionsis,asbioethicistscanattest,noeasytask.Socialsystems,especiallyinpluralisticcivilsocietiessuchasdemocracieschar-acterizedbycompetinginterestsandvalueclaims,aretremendouslycomplex;togenuinelyunderstand(andalter)socialsystemsrequiresacoordinatedeffortacrosstraditionaldisciplinaryboundariesandacrossmultiplesectorsandspheresofinfluence,andnotjustthecoordinationofindependentefforts,muchlessthecoordinationofindependentdisciplinaryefforts.Alas,likesystemsbiology,thismaysoundniceintheabstract.Whetherandhowitmightactuallybeinstantiatedremaintobeseen.conclusionMolecularbiologyhasundergonetremendouschangesinthepastsevendecades.Whilesomeofthephilosophicalissuesremainconstant(e.g.,reductionism),newmethodologicalandconceptualconsiderationshaveemergedovertime,requiringphilosophicalclarificationandscrutiny.Inthisessay,Ihavenotsoughttoanalyzetheseissuessomuchastosketchinroughoutlinetherichterrainthatcomprisesthephilosophyofmolecularbiologyand,now,ofsystemsbiology.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandBioethics371Thisisfertilegroundforhistoriansandphilosophers–What,precisely,issystemsbiology,whetherinparticularcasesormoregenerally?What,ifanything,isnewinsystemsbiology,andwhatareitshistoricalprecursors?Whatareitsvariousconcepts,con-ceptualassumptions,andmotivations?Whatmethodologicalandanalyticalproblemsdoesitraise(and/orsolve)?Whatistherela-tionshipbetweenmolecularbiologyandsystemsbiology,andwhatisthenatureofmolecularsystemsbiology?–aswellasforsocialscientists,includingsciencestudiesscholars–Howisthisnewfieldorganized?Doessystemsbiologyreflectadramaticallynewstyleofinterdisciplinaryscholarshipinbiology,orisitpartofalongertradition(and,ifso,whichone[s])?Whatisthecultureofsystemsbiology,andwhatareitsemergingtraditions?–andethi-cists,too–Whatarethedeterminantsoftranslationalsystemsbiologyresearch,andwhatarethesocietalimplicationsofad-vancesinsystemsbiology?Whatarethebenefitsandopportunitycostsofstrategictranslationalresearch?Theseandrelatedques-tionsshouldpreoccupytheethics,history,philosophy,andsocialstudiesofmolecularandsystemsbiologyintheyearstocome,ideallyincollaborativeengagementwithscientistsandotherinterestedparties.Onlythenmaywedeterminewhethermolecularbiologycanbemadetoworkintherealworld–andmadetoworktowardbeneficialends.notes1.Onthisview,noticethatsystemsbiologyisnotonlythescienceofsystems,butalsothescienceofthe‘spacesbetween’moleculesandlargersystems–thestudyofwhathappensinthesespaces,how,andtowhatend.Cf.HansWesterhoff,ascitedinHenry(2003):‘‘‘Systemsbiologyisnotthebiologyofsystems,’heemphasizes.Instead,hesays,itistheregionbetweentheindividualcomponentsandthesystem,whichiswhyit’snew.‘It’sthosenewpropertiesthatarisewhenyougofromthemoleculetothesystem,’hesays.‘It’sdifferentfromphysiologyorholism,whichstudytheentiresystem.It’sdifferentfromreductionistthingslikemolecularbiology,whichonlystudiesthemolecules.It’sthein-between.’‘‘CambridgeCollectionsOnline©CambridgeUniversityPress,2008\ngregorym.mikkelson20Ecologyintroduction:undeaddogmasofempiricismIsuspectthedemandforevidenceaboutindividualsisabastardizedversionofanoldpositivistclaim:theclaimthattheoreticaltermsmustbedefinedinobservationalones,inparticularindividualsensoryexperiences.Kincaid1996,182Thephilosophyofbiologyhasmaturedquiteabitoverthelasttwodecades.Backin1988,Rusenotedaconspicuousdearthofworkonecology.Butby1999,SterelnyandGriffithsdevotedanentirechaptertoitintheirintroductiontothephilosophyofbiology.Thereisstillplentyofroom,andreason,formorephilosophicalattentiontoasciencesovitalforunderstandingandaddressingenvironmentalconcerns.Butatleastseveralpeoplenowmakephilosophyofecologytheiracademicspecialty.InthefollowingIshallverybrieflysurveyrecentdevelopmentsinbothecologyandthephilosophythereof.Oneimportantaspectofthedevelopmentswithinecologyisanexpansiontolargerspatio-temporalscalesofinvestigation.Thiswiderfocushasoften,thoughnotalways,resultedinashiftinperspective,fromviewingecolog-icalentitiesasclosedsystemstotreatingthemasopensystems.Iwilltakeacloserlookatthreeexamplesofscaleexpansionandteaseoutsomeoftheirimplicationsforenvironmentalpolicy,ononehand,andstill-commonreductionistphilosophiesofscience,ontheother.Finally,Iwillconsiderthephilosophicalimplicationsofthesearchformechanisms,whentheopennatureofthesystemsunderstudyisacknowledged.372CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEcology373Butfirst,somewidercontext.Intheearlytwentiethcentury,thelogicalpositivistsespousedacertainkindofethicalsubjectivism,acertainkindofreductionism,andacertainkindofinstrumentalism.Respectively,thesedoctrinesheldthatethicalstatementsaremeaninglessemotionaloutbursts,thattheoreticalpredicatesarereducibletoobservationalpredicates,andthatquestionsaboutthetruthorfalsityofstatementsregarding‘‘unobservable’’entitiesarealsomeaningless.Mostphilosophershavesincerejectedallthreeofthesepositions.Butotherversionsofsubjectivism,reductionism,andinstrumentalismpersistwithinbothphilosophyandscience.Bytheendofthetwentiethcentury,scientificrealismhadarguablyeclipsedinstrumentalismwithinphilosophyofscience.Never-theless,newversionsofinstrumentalismstillcropup.Onemightthinkthatafieldlikeecologyshouldbelargelyimmunetothewholedebate.Theentitiesthatitdealswitharemostly‘‘observable’’,andhencenotsubjecttotheskepticismtraditionallyleveledbyinstru-mentalistsagainst‘‘unobservable’’objectslikeatoms.YetSoberhasarguedforinstrumentalismregardinghypothesesaboutthedegreetowhichthecornplantsintwofieldsdifferinaverageheight(Sober1999).Ihaveshownthateveninthenew‘‘Akaikean’’statisticalframeworkinvokedbySober,theoriesachievepredictiveaccuracyinthewaythatrealistssaytheydo.Predictivesuccessresultsfromgettingattheunderlyingtruth–notfromthekindof‘‘cosmicaccident’’requiredbyinstrumentalistaccounts(Mikkelsoninpress).Itmaybethatmostphilosophersofsciencenowconsiderthem-selvestobeantireductionists.Andyetmostargumentsagainstreductionismattackonlyanextremeversionofit:theideathatlower-levelprocessescompletelyexplainhigher-levelprocesses(butnotviceversa).Isubmitthatitistimealsotoquestionaslightlymilderversionofreductionismthatseemstoguidescientificfundingpolicy,aswellasmanyscientists’viewsaboutpropermethodology.Accordingtothistypeofreductionism,lowerlevels‘‘merely’’playafarmoreimportantrolethanhigherlevels,inexplanationsofmostphenomena.Forexample,scientificfundingpatternsimplythatgeneticcausesofhumandiseasearefarmoreimportantthanenvi-ronmentalcauses.Inthefollowing,Ishalloffersomereasontodoubtthiskindofposition,atleastwithinecology.Thedebateoverethicalsubjectivismversusethicalrealismhasnotattractedmuchcommentinphilosophyofbiology(Sober2000isCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n374gregorym.mikkelsonanexception).Andyetthisdebatehasprofoundimplicationsforthequestionofwhethersciencecanorshouldbe‘‘value-free’’(Putnam2002).Inthefollowing,Ishalllimitmyremarkstonotingsomeofthepolicyimplicationsofrecentresearchonthecausesandcon-sequencesofbiodiversity.1.averybriefsketchofrecentecologyandphilosophythereofBeattycitedfoursetsof‘‘interestingfoundationalandmethodolog-icalproblems’’inecology(Beatty1998):1.‘‘[P]roblemsofclarifyingthedifferencesandcausalconnec-tionsbetweenthevariouslevelsoftheecologicalhierarchy(organism,population,community,ecosystem)’’2.The‘‘issueofhowcentralevolutionarybiologyistoecology’’3.‘‘[L]ong-standingissuesconcerningtheextenttowhichthedomainofecologyismorelaw-governedormoreamatterofhistoricalcontingency’’and4.The‘‘relatedquestionofwhetherecologistsshouldrelymoreonlaboratory/manipulativeversusfield/comparativemethodsofinvestigation.’’SinceBeattypublishedhisoverview,ecologistshavediscoverednumerousinterlevelcausallinksandotherlawlikegeneralizations(cf.1and3;see,e.g.,Kinsey2002,Marquetetal.2005).Someprogresshasalsooccurredwithregardtoatleastoneaspectoftherelationshipbetweenecologyandevolution,namely,theextenttowhichentireecologicalcommunitiesorecosystemsaretargetsofnaturalselec-tion(see2;Swenson,Wilson,andElias2000).And‘‘laboratory/manipulative’’and‘‘field/comparative’’researchprogramshavebothcontinuedtoproceedwithvigor(4).NordoIknowofanymajorshiftsinemphasisbetweenthetwo.Since1998philosophers,andscientistsofferingphilosophicalcommentary,havepaidthemostattentiontothethirdtopic;namely,lawsinecology(Cooper1998,ColyvanandGinzburg2003,Mikkelson2003).Severeambiguityaboutwhatlawsarestillplaguesthisbodyofwritings,justasitdogsmoregeneraldiscussionsoflawsinphilosophyofscience.Partlyforthisreason,IshallhereinCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEcology375avoidthequestionoflawhoodingeneralandfocusinsteadonotheraspectsofparticularecological‘‘laws’’.Someworkhasbeendoneonthefirsttwotopics,though.InSections2and3,Ioffernewexamplesandargumentsregardingtheissueoflinksbetweenlevelsofeco-logicalorganization.Asfortherelationshipbetweenecologyandevolution,philosophersofbiologyhavetendedunfortunatelytodepicttheformerasahandmaidenofthelatter(SterelnyandGriffiths1999).Iamnotawareofanysustainedphilosophicaldiscussionssince1998ofBeatty’sfourthtopic,differentempiricalapproachesinecology.Hisdichotomyof‘‘laboratory/manipulative’’versus‘‘field/observational’’obscuresimportantaspectsofthissubject,though.Foronething,manyofecology’smostimportantadvanceshaveresultedfrommanipulativefieldexperiments(e.g.,thoseofPaine1966,SimberloffandWilson1969,Likensetal.1970;seeDiamond1986forageneraldiscussion,andSection2cforanotherexample).AndyetsuchexperimentsstraddleBeatty’stwocategories,ratherthanfittingneatlyintoeitherofthem.Foranotherthing,therelativeprominenceofdifferentempiricaltechniquesseemstobealesscrucialmethodologicalissuethanthedegreeofintegrationamongtheory,experiment,andobservation.Asinanyfield,thesethreemodesofecologicalresearcharesometimestightlycoupled,andatothertimesfairlyinsulatedfromeachother.Intuitionaswellasacursoryhistoricalsurveyofscienceingeneralandecologyinparticular,suggestthatbetterintegrationyieldsmoresubstantialscientificprogress.TheexamplediscussedinSection2anicelyillustratesthebenefitsresultingfromimprovedintegrationoftheory,experiment,andobservation.Beforeproceedingfurther,Ishouldmentionatleastafewothercasesofrecentphilosophicalworkonecology.Odenbaugh(2001)dealtwithphilosophicalissuesstemmingfromaperiodofstridentreductionism,particularism,andpessimisminecology(fromthelate1970sthroughthe1980s).DeLaplante(2004)hasconsideredrelationshipsbetweenecologyandthesocialsciences.Andenviron-mentalethicistshavecontinuedtowriteaboutecologicalscience,astheyhaveforseveraldecades.Insomecases,theseeffortshaveyieldedastuteanalysesofecologyresultingfromlonganddeepreflectiononthenatureofthisscienceanditsmoralimplications(cf.Skipperetal.inpress).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n376gregorym.mikkelson2.goingmacroThecommonthemeunitingthethreecasestudiesconsideredinthissectionisashifttowardlargerscales.Historically,ecologistscon-centratedontheinternaldynamicsoflocalpopulations,commu-nities,andecosystems.Nowtheymoreoftenalsotakeintoaccount,orturntheirfocustoward,thelargerwholesthatcontainlocalecologicalsystems.Theseincludemetapopulations,metacommu-nities,landscapes,regions,bioticprovinces,and,atthelargestscaletodate,theentirebiosphereorecosphere.Inmanycases,suchachangeinfocushasresultedinpromisingnewinsights.Inthissection,Ishallassessthephilosophicalsignificanceofexpandedspatiotemporalperspectivesonrelationshipsbetween(a)plantsandherbivores,(b)areaandnumberofspecies,and(c)numberofspeciesandthetotaldensityofbiomasssummedacrossallspecies.a.ResponsesofDifferentTrophicLevelstoNutrientEnrichmentLeiboldandcolleaguesnotedastrikingmismatchbetweencertainmanipulativeexperiments,includingfieldexperiments,alongwiththeoreticalmodels,ononehand,andobservationalsurveysofunmanipulatedecologicalsystems,ontheother(Leiboldetal.1997).Thetheory,experiment,andobservationinquestionallfocusedon1changesinplantandherbivorebiomassdensityduetoincreasesin2thenutrientsthatplantsneedtogrow.Mostofthe‘‘nutrientenrichment’’experimentssurveyedbyLeiboldandcolleaguesfitoneoftwopatterns:asnutrientlevelsincreased,eitherplantsincreasedproportionallymuchfasterthanherbivores,orviceversa.Theseexperimentalresultsfitnicelywiththepredictionsofsomesimpletheoreticalmodels.Accordingtosuchmodels,iftheplantspeciesinquestionareallrelativelyedible,thenincreasingthenutrientsupplydoesnotincreasetotalplantdensityverymuch.Instead,herbivores‘‘chowdown’’onmostofthe‘‘extra’’plantpro-duction,leadingtoaproportionallymuchgreaterincreaseintheirowndensity.IlooselyfollowLeiboldincallingthisscenarioaclosed3‘‘foodchain’’(Leibold1996).SeethesteepestlineinFigure20.1.Similarmodelspredictedthatifsomeoftheplantsarerelativelyinedible,nutrientenrichmentshouldincreasetheirdensityCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEcology377IsolatedfoodchainsOpenfoodwebsLog(Totalherbivoredensity)IsolatedfoodwebsLog(Totalplantdensity)Figure20.1Schematicdepictionofcontrastingrelationshipsbetweentheresponsesofplantsandherbivorestonutrientenrichmentsubstantially.Thisisbecausetheinedibleplantsdonotsufferasmuchchowingdown.Thisrelativeimmunitygivesthemacom-petitiveadvantageagainstmoreedibleplants.Andthis,inturn,leavestheherbivoreswithrelativelylesstoeat.Sotheherbivoresincreaselittle,ifatall,indensitywithincreasednutrientlevels.4Thisisthe‘‘closedfoodweb’’scenario(cf.theshallowestlineinFigure20.1).Thenutrientadditionexperimentsandtheoreticalmodelsdescribedcontrastmarkedlywithobservationsofunmanipulatednaturalecosystems.Amongthesesystems,plantandherbivoredensitiesbothincreasesignificantlywithnutrientlevels,asdepictedbythemiddlelineinFigure20.1.AshiftinspatialandtemporalperspectiveenabledLeiboldandassociatestoreconciletheseobservationswiththeoryandexperiment.Theynotedthatthethe-oreticalandexperimentalmodelsathandbothtreatedecosystemsasiftheywereisolatedfromtheirsurroundinglandscapesorregions.Consequently,thesemodelsdidnotallowforthecolonizationofnewspeciesfromoutsideagivenecosystem.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n378gregorym.mikkelsonTheoreticalmodelsthattakecolonizationintoaccountpredictsignificantincreasesofbothplantandherbivoredensitywith5nutrientlevels(an‘‘openfoodweb’’scenario).Tounderstandthis,considerthatrelativelyinedibleplantsgenerallyrequiregreateramountsofnutrientsthandomoreediblespecies.Thisisbecauseresistancetoherbivory,forexample,throughtheproductionofnoxiouschemicals,canbephysiologicallytaxing.Nutrientenrich-mentthereforeallowsprogressivelymoreinediblespeciestocolo-nize.Colonizationbymoreinediblespecies,inturn,increasesoverallplantdensityandreducesherbivoredensity,relativetothefood-chainscenario.InFigure20.1,atiltdownwardfromthesteepestlinewouldrepresentthisoutcome.Thisraisesthequestionofwhethercolonizationbyinediblespeciesshouldcauseherbivoredensitytodeclinesodramaticallythatitresultsinthesamepatternfoundintheisolatedfoodwebscenario.ThereasonsgivenbyLeiboldandassociatesthatthisdoesnothappenaresomewhatobscure.Itwouldsuffice,however,fornutrientenrichmenttopermitthecolonizationofherbivorespecieswiththeabilitytoovercomethedefensesofthe‘‘inedible’’plants.Overcomingplantdefensescanalsobephysiologicallytaxing.Therefore,herbivoresabletodoitstandabetterchanceinthepres-enceofhighernutrientlevels,andthusgreateroverallplantpro-duction.ColonizationbythiskindofherbivorewouldresultinatiltupwardfromtheshallowestlineinFigure20.1.Justasinthesetheoreticalmodelsofopenecosystems,experi-mentalmodelsthatallownewspeciestocolonizefromoutsidetheecosystemresultinroughlyproportionalincreasesinplantandherbivoredensitywithnutrientlevels.Onemethodologicaltake-homemessageisthatmodelstreatinglocalecosystemsasopensystemsaccordwithobservedpatternsinnature.Theoriesandexperimentsthattreatsuchecosystemsasthoughtheywerecloseddonotfittheseobservations.Thistake-homemessagepotentiallyunderminesthetypeofreductionismdiscussedinSection3.Accordingtothatdoctrine,thepartsofanentityshouldplayamuchmoreimportantrolethanitsenvironmentinexplanationsofthatentity’sbehavior.Sinceaclosedsystem,byassumption,hasnoenvironment–nolargersystemthatcouldexertanymaterialinfluence–treatinganentityasaclosedsystemtakesthisreductionistprescriptiontoanextremedegree.ButCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEcology379anopen-systemapproachleavesopenthepossibilitythattheenvironmentplaysanequally,orevenmore,criticalexplanatoryrolethantheparts.Somereductionistshaverationalizedtheirpositionbyclaimingthatreductionis‘‘theonlymethodofattainingunitarysciencethatappearstobeseriouslyavailable’’(OppenheimandPutnam1958).Someantireductionistsconcedetheideathatreductionistheonlyviableroutetotheunityofscience(e.g.,Dupre´1993).Yettracingcausalinfluencesboth‘‘upward’’fromthepartsand‘‘downward’’fromtheenvironmentoffersmorechancestounifyourunder-standingofdifferentlevelsoforganizationthanwouldlimitingattentiontointernaldynamicsalone.b.Species-AreaRelationsAsintheplant-herbivorestudiesoutlinedpreviously,ecologistshavealsorecentlyexpandedthespatialscaleatwhichtheyinvestigaterelationshipsbetweenareaandnumberofspecies.The‘‘species-arearelation’’,or‘‘law’’,isoneofthemostvenerablepatternsinecology.Atleastinpartbecauseofpracticalconstraints,ecologistshistori-callybeganexploringspecies-arearelationsamongrelativelysmallpatchesofhabitat.Thesepatchesgenerallyfellintooneoftwocategories–eithercontiguouspartsoflargerhabitatsorislands.However,Rosenzweighasissuedaboldnewclaimaboutspecies-areapatternsamongentire‘‘bioticprovinces’’(Rosenzweig1995).Strictlyspeaking,Rosenzweigdidnotdistinguishcontiguouspatchesfromislandsfromprovincesonthebasisofsize(area).Instead,hedifferentiatedthemalonganaxisofimmigration.Con-tiguouspatchesexperiencethemostimmigration.Acontiguouspatchissowellconnectedtootherpatchesofthesamekindthatmostofthetime,whenpopulationsinsidethepatchdecline,theyare‘‘rescued’’byimmigrationfromoutside.Anisland,asRosenzweighasdefinedit,istooisolatedforthisrescueeffecttodominatepopulationdynamics.Ifagivenislandpopulationdeclines,itwillgoextinctunlessbirthsontheisland–ratherthanimmigrationfromoutsideit–turnthetide.Butanislandisstillconnectedenoughtoamainlandforimmigration,alongwithextinction,todominatediversitydynamics.Mostofthespeciesonanislandowetheirpresencetheretoimmigration,ratherthantospeciationontheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n380gregorym.mikkelsonIslandsBioticprovincesNumberofspecies95%habitatlossAreaFigure20.2Schematicdepictionofcontrastingrelationshipsbetweenareaandnumberofspeciesislanditself.Incontrast,aprovinceisanareasoisolatedfromother,similarhabitatsthatmostofitsspecies‘‘arrive’’throughinsituspeciation.ForRosenzweig,Hawaiiisisolatedenoughtocountasitsownprovince.Butmostprovincesaremuchlarger.Rosenzweig’sboldclaimisthatthespecies-arearelationshipamongprovincesislinear,ornearlyso.Contiguouspatchesandislands,incontrast,showadiminishing-returnspatternofspeciesrichnesswitharea;Figure20.2illustratesthecontrastbetweenislandsandprovinces.IfRosenzweig’sclaimprovesrobust,wewillhavetoincreasedramaticallytheestimatesofhowmanyspeciesourroadbuilding,agriculture,urbanization,andotheravenuesofhabitatdestructionaredrivingextinct.Importantly,species-areacurvesprovidenoinformationabouthowfastextinctionsduetohabitatlosswilloccur.Thelongerthoseextinctionstake,themoretimewewillhavetorestorelosthabitatandtherebypreventmassdie-offs.Alinearrelationshipimpliesthatifhumansdestroy95percentofthenativehabitatinagivenprovince–aswehavealreadydonetoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEcology381theAtlanticrainforestofSouthAmerica,thetallgrassprairieofNorthAmerica,andotherprovinces–eventually95percentofthespeciestherewillgoextinct.Incontrast,previousestimatesofanthropogenicextinctionwerebasedonextrapolationfromislandcurves.Thoseestimatespredictedthatlosing95percentofaprov-incewouldonlydrivearound50percentofitsspeciesextinct.Toseethecontrast,imaginestartingwiththelargestarearepresentedinFigure20.2–allthewaytotheright.Nowimaginedestroying95percentofthehabitat–thatis,moving95percentofthewayleftwardacrossthediagram.Andnotethedifferencebetweenthefractionsoftheoriginalspeciescountthattheisland-basedcurveversustheprovinciallinepredicttoremain.Astheexampleofplant-herbivoredensityrelationsdoes,thespecies-areaexampleillustrateshownewinsightsaregainedbyincreasingthescaleofecologicalinvestigation.Inthiscase,how-ever,thechangeinscaleresultedinashiftfromseeingprovincesasopensystemsanalogoustonear-shoreislandstoseeingthemasrelativelyclosed.Thelargeproportionofspeciesthatareendemicto–thatis,foundnowhereelsethanin–particularbioticprovinceswarrantstreatingprovincesasrelativelyclosedwithrespecttoimmigration(thoughofcoursenotisolatedfromsolarenergyinput,etc.).Thiscasethereforeremindsusthatinsomecases,andinsomerespects,theremaybegoodreasontotreatagivenecologicalentityasaclosedsystem.c.BiodiversityandEcosystemFunctionBesidesatrendtowardlargerscales,anotherchangeinthe1990swasashiftfromviewingthenumberofspeciesinacommunityoreco-systemstrictlyasadependentvariable–thatis,astheeffectofotherecologicallyimportantproperties,suchasarea,asdiscussedearlier–toinvestigatingitspotentialroleasacauseofimportant‘‘ecosystemfunctions’’aswell(Naeem2002).Oneprominentstrandofthisresearchdealswiththeinfluenceofplantspeciesrichnessontotalplantdensity.Plantdensity,inturn,affectstheabilityofecosystemstoprovide‘‘services’’suchascarbonstorage,floodanddroughtcontrol,andwildlifehabitat.Darwinassertedapositiveeffectofspeciesrichnessontotaldensity,sothisisnotexactlyanewtopic(Mikkelson2004).Whatisnewisthelargeamountofattention–andCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n382gregorym.mikkelson1.5081-m2patches1.25)221-kmpatches1.000.75Totalplantdensity(kg/m0.500.250204060NumberofplantspeciesFigure20.3Schematicdepictionofcontrastingrelationshipsbetweennumberofplantspeciesandtotalplantdensitycontroversy–thatdiversity-densityandotherdiversity-ecosystem-functionrelationshipshaverecentlyattracted.2Inoneexperiment,9-meterby9-meter(81-m)grasslandplotssownwithmorespeciestendedtohavesubstantiallygreatertotaldensitythanthoseplantedwithfewerspecies(Tilmanetal.2001).Theexperimentaltreatmentsrangedfrom1to16plantspeciesand2yieldedarangeofaround0.4to1.3kilograms/moftotalplant6density.SeethedottedcurveinFigure20.3.Forpracticalreasons,thesegrasslandfieldexperimentsandotherslikeithaveoccurredatscalesthataresmallrelativetothesizeofmostfarms,letalonelargermanagementunitssuchasnationalforestsorecoregions.Thespecies-arealawcanbeusedtopredictwhatshouldhappentothediversity-densityrelationshipatsuchlargerscales(Tilman1999).Atprogressivelylargerscales,itshouldtakemoreandmorespeciestoachieveanygivenbiomassdensity.Forexample,wecanusearecentestimateofthespecies-arearelationforcontiguousCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEcology383patchestoextrapolateTilmanandcolleagues’results.Accordingto0.15Rosenzweig(2003),roughlyS¼cA,whereSisnumberofspecies,cisanadjustableparameter,andAisarea.Thisequationentailsthat2foreach81-mplotinasquarekilometerofgrasslandtocontain162species,the1-kmpatchwouldhavetocontain66species.Thus,toattainthemaximumdensityrealizedinTilmanandcoworkers’2experimentthroughouta1-kmpatch,itwouldtake66species,ratherthanonly16(cf.thesolidcurveinFigure20.3).Onewaytounderstandthiscontrastbetweendiversity–total-densityrelationsatdifferentscalesistoconsiderthatthelargerthepatch,themoreheterogeneousthetemperature,moisture,andsoilconditions,andsoon,areacrossit.Thismeansthattheparticularsetofspeciesabletoattainacertaintotaldensityinonepartofthelargerhabitatwilldifferfromthespeciesthatarecollectivelyabletoattainthesamedensityinotherparts.This,inturn,entailsthatachievinganygivendensityacrossalargehabitatrequiresmorespeciesthanareneededtoachievethesamedensitywithinanysmallerpartofit.3.mechanisticresearchinopensystemsInallthreeareasofresearchexemplifiedintheprevioussectionofthispaper,ecologistshavesoughttogobeyondthediscoveryofso-calledphenomenologicalpatterns,byuncoveringthe‘‘mechan-isms’’responsibleforthem.Inthissection,Iaimtocorrectthemisconceptionthatmechanisticresearchisnecessarilyreduc-tionistic.Thismisconceptionseemscommonamongbothphiloso-phersandecologists,andamongbothadvocatesandopponentsofreductionism.Asoneantireductionistphilosopherofsocialscienceputit,‘‘Themicroexplanationtellsusthemechanismbywhichthemacroexplanationoperated’’(Garfinkel1981,58).Dardencrit-icizedanotherphilosopher,Thagard,forhis‘‘reductive’’viewofcertaingeneticmechanisms(Darden2000).MyargumentinthefollowingissimilartoDarden’s,butmorefullydeveloped,andframedwithinthecontextofecology.Ihaveelsewherespelledoutadifferent,complementaryreasonwhymechanisticresearchoftencontradicts,ratherthanfulfilling,reductionistprescriptions(Mikkelson2004).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n384gregorym.mikkelsona.DefinitionsBeforegoingfurther,letusconsiderwhatismeantbythewords‘‘mechanism’’and‘‘reductionism’’.Isubmitthatingeneral,toprovideamechanismforanecologicalrelationshipisto‘‘fillintheblank’’.GiventhatAcausesB,todescribeamechanismforthatrelationshipistoshowthatAcausesM,whichthencausesB.Inotherwords,themechanismA!M!BpartlyfleshesoutthecausalpathwayA>B.Thisisfairlyminimalistcomparedtotheexplica-tionsdiscussedbyTabery(2004).IamnotconvincedthatanymoreisreallyneededthanIprovidehere,butifmoreisrequired,thenperhapsmyinterpretationstillworksasapartialdefinitionoftheterm‘‘mechanism’’,providingnecessary,thoughnotsufficient,conditions.Toillustratethisconstrualofthe‘‘mechanism’’concept,supposethatanincreaseinonepopulationcausesasecond,competingpopulationtodecrease.TheLotka-Volterracompetitionequations–arelatively‘‘phenomenological’’approach–candescribethiskindofeffectwithreasonableaccuracy.More‘‘mechanistic’’modelsgobeyondtheLotka-Volterraequationsbyspecifyinghowpopulationsexertcompetitiveeffectsoneachother.Forexample,resource-competitionequationscanrepresentthefollowingcausalchain:Anincreaseinoneplantpopulationleadstoincreaseduptakeofacer-tainnutrientbythatpopulation,therebyreducingtheamountofthatresourceleftintheenvironment.Thisdecreaseinthenutrientsupplythencausesasecondpopulation,dependentonthesamenutrient,todecline.Whattype(s)ofreductionismis(are)atstakeinecology?Schoenerseemstohavecapturedatleastoneimportantkind.Healsoassumedwithoutdiscussionthat‘‘mechanistic’’means‘‘reductionistic’’:[A]dvocacyofareductionistapproachcoincideswithemphasizinginternal,ratherthanexternal,factorswhensimplificationisnecessary.Thusmechanisticpeoplewillstressbehavioralandphysiologicaldetailattheexpenseof,say,food-webdetail.(Schoener1986)Schoener’sarticlefocusedonexplainingpopulationdynamics.Hiscommentsthereforeimplythatmechanisticexplanationsofsuchdynamicsmustinvolvemoredetailabouttheorganismscom-posingthepopulationsthanaboutthecommunitiesorecosystemsCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEcology385containingthepopulations.Inotherwords,hearguedthatthemechanismsofpopulationdynamicsarepredominantlymicro-orlower-level.Attheextreme,thisformofreductionismentailstreatingpopu-lationsascompletelyclosedsystemsthatdonotinteractatallwithmacro-orhigherlevels.Thedensity,forexample,ofaclosedpopulationwouldbeexplainedstrictlyintermsofthepropertiesofthatpopulation,includingtheindividualorganismswithinit.Suchhigher-levelphenomenaasthedensitiesofotherpopulations,orthetotalnumberofspecies,withinthesamecommunity,wouldplaynoroleinreductionisticexplanationsofthiskind.b.AHigher-LevelMechanismLetusnowconsidertheimplicationsofarecenttheoreticalstudythatinsteadtreatspopulationsasopensystemsembeddedwithineco-logicalcommunities(KilpatrickandIves2003).Theauthorsofthisstudyofferedamechanisticexplanationfor‘‘Taylor’slaw’’.Thislaw2zrelatesthevariabilityofapopulationtoitsaveragedensity:r¼cl.2risthevarianceofpopulationdensity,cisacoefficientgreaterthanzero,lismeanpopulationdensity,andzisanexponentlessthan2.Ecologistshavefoundthispattern–observedamong‘‘morethan400speciesintaxarangingfromprotiststovertebrates’’–interestingbecauseitdiffersfromthe‘‘null’’expectation.Ifpopulationsexperiencedconstantpercapitavariability,zintheequationwouldequal2.Thefactthatzislessthan2formostspeciesmeansthatanincreaseinpopulationdensityleadstoadecreaseinper-capitavariability.KilpatrickandIvesexplainedthispatternintermsofpopulations’‘‘diffuseinteractions’’withtheotherspeciesintheircommunities.Roughlyspeaking,theirmechanismworksasfollows:Anygiven‘‘focal’’populationundergoessomevariationduedirectlytofluc-tuationsinitsphysicalenvironment.Fluctuationsinthepopula-tionsofitscompetitorsaddasecondsourceofvariationinthefocalpopulation.Supposethatthefocalpopulationexperiencesaperma-nentincreaseinitsmeandensity.Otherfactorsbeingequal,thiswouldforceadecreaseinthemeandensitiesofatleastsomeofits7competitors.This,inturn,wouldreducethepressureexertedbythosecompetitorsonthefocalpopulation.EachmemberofthefocalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n386gregorym.mikkelsonpopulationwouldthenbelessaffectedbyfluctuationsintheden-sitiesofthesecompetitors.Ifthedirectcontributionofenviron-mental‘‘noise’’toper-capitavariabilityremainsthesame,thenoverallper-capitavariationinthefocalpopulationshoulddecline.Wethushaveanexampleofahigher-levelmechanismforasame-levelrelationship.Thisisadifferentsenseof‘‘higher-levelmechanism’’thantheoneGlennanattributestoWimsatt(Glennan2002).Thelatterreferstoanymechanismthatdoesnot‘‘plunge’’allthewaydowntosome‘‘fundamental’’,suchasmicrophysical,level.Incontrast,hereImeanamechanismthatinvolvespropertiesatahigherlevelthantherelationshipbeingexplained.Inthiscase,acommunity-levelmechanismexplainsapopulation-levelrelation-ship,Taylor’slaw.Thiscasethereforecontradictsthecommonassumption,impliedbyGarfinkel(1981)andexpressedbySchoener(1986),thatmechanismsentailreductionism.Inotherwords,thiscasedemonstratesthatmechanisticresearchneednotempha-sizethepartsofafocalentityattheexpenseofitscontextorenvironment.Incidentally,thetwoexamplescitedinthissectionalsoillustratetherelativityofthephenomenological/mechanisticdistinction.TheLotka-Volterracompetitionequationsarephenomenologicalrela-tivetoresource-competitionequations.ButrelativetoTaylor’slaw,theLotka-VolterramodelthatKilpatrickandIvesusedtodescribehowmeanpopulationdensityaffectspopulationvariabilityismechanistic.concludingsummaryInthisessay,Ihavedescribedarecentshiftinperspectivefromsmallertolargerscalesinecology.Thisshifthasrevealedthatspe-ciesdiversityismoreimportant,andundergreaterthreat,thanwaspreviouslyknown.Italsomotivatesgreaterappreciationoftheroleplayedbyhigher-levelcausesinnature.Ihavealsoillustratedhowrecognizingtheopencharacterofmostecologicalsystemsleadsnaturallytothediscoveryofhigher-levelmechanisms.Earlierinthechapter,Isituatedthesepointswithinthecontextofrecentworkinthephilosophyofecology.AndIframedthediscussionintermsoflogicalpositivism’stenaciouslegacy.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nEcology387notes1.Livingmassperunitareaorvolume.2.Manyofthesesurveysactuallyinvolvedalgae,which,strictlyspeaking,arenotplantsbutplaythesameecologicalroleasplants,namely,‘‘photosynthesizer’’or‘‘primaryproducer’’.3.Thosefamiliarwithfoodchainsknowthatthisscenariopresupposesanevennumberoftrophiclevels(e.g.,justplantsandherbivores,orplantsplusherbivoresplus‘‘primary’’carnivoresplus‘‘secondary’’carnivores’’).4.RepresentedinLeibold’spaperby‘‘Model1’’.5.Leibold’sModels2and3.6.EstimatedbyvisualinspectionoftheYear2000datashowninfigure1BofTilmanetal.(2000).7.Eveniftheincreaseinthefocalpopulationdidnotreduceotherpopulations,itwouldstillreducetheirrelativedensities–relative,thatis,tothedensityofthefocalpopulation.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nsahotrasarkar21FromEcologicalDiversitytoBiodiversity1.introductionDuringthelastthreedecades,biodiversityconservationhasemergedasacentralfocusofenvironmentalconcerninmanyregionsoftheworld(Sarkar2005).Asaresult,large-scaleeffortsarebeingdevotedtodevisingsystematicprotocolsforconservation,sometimesinvolvingcomputationaleffortsunprecedentedinecology(MargulesandPressey2000).Theseeffortspresumethatasufficientlypreciseconceptofbiodiversityisathand.Somephilosophicalattentionhasalsobeenfocusedonelaboratinganadequatenormativebasisfor1conservation.Theseattemptednormativejustificationsforbiodi-versityconservationalsodependonwhatismeantbybiodiversity.Yet,‘‘biodiversity’’remainsacontestedterm.Theterm‘‘biodiversity’’wasfirstusedin1986asacontractionfor‘‘biologicaldiversity’’byWalterG.Rosenwhileplanningfora(U.S.)NationalForumonBiodiversity(Takacs1996).Subsequently,tem-porarilymutatedas‘‘BioDiversity,’’itwasusedasthetitlefortheproceedingsfromthatmeeting(Wilson1988).Noattemptwasmadetodefinethetermprecisely,evenasitsusespread–thechronology2inTable21.1includesthemostsalientepisodes.Conservationbiologyalsoemergedasadistinctinterdisciplinaryresearchareaduringthe1980swithitscentralaimtheprotectionofbiodiversity(Takacs1996,Sarkar2002,2005).Fromtheverybeginningtheterm‘‘biodiversity’’wasnotableforitspolyvalence,meaningdifferent3thingstodifferentusers.Suchpolyvalenceshouldneitheroccasionexcessivesurprisenor,byitself,betakentoindicatesomespecialdifficultywithaterm.Biologicaltermssuchas‘‘species’’or‘‘gene’’areatleastaspolyvalentwhileremainingsufficientlyprecisefor388CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity389Table21.1.EcologicalDiversity,Biodiversity,andConservationBiology:ABriefChronologyYearDevelopmentReferences1943Firstindexofecologicaldiversity.Fisheretal.19431948ShannonpublisheshisindexforthequantityShannon1948;ofinformationinamessage.PrestonPreston1948attemptstoimproveuponFisher.1949Simpson’sindexintroduced.Simpson19491955MacArthurusesShannonindexasameasureMacArthur1955ofstability,makesadiversity-stabilityclaimprecise.1958MargalefusesShannonindexasameasureMargalef1958;ofdiversity.EltonproducesempiricalElton1958evidenceofadiversity-stabilityrelationship.1959HutchinsonpayshomagetotropicalspeciesSarkar2005richness,raisingthequestionsofthedeter-minantsofthelatitudinaldiversitygradient.1960Whittakerdistinguishesamonga-,b-,andWhittaker1960c-diversity.1967MacArthurandWilsonpublishTheTheoryofMacArthurandIslandBiogeography.Wilson19671968BiologicalConservationstartspublication.Sarkar20041969PielouprovidesaunifiedtheoreticalanalysisPielou1969ofdiversityindicesinatextbooktreatment.1970HutchesonproducesfirstdissertationdevotedHutcheson1970toempiricalstudiesofdiversitymeasures.1971Hurlbertquestionstheuseofdiversityindices.Hurlbert19711973U.S.EndangeredSpeciesActispassed.MayMay1973publishesStabilityandComplexityinModelEcosystems.1975DiamondsuggestsislandbiogeographyastheDiamond1975;modelforthedesignofnaturereserves.Pielou1975Pieloupublishesbook-lengthtreatmentofecologicaldiversitymeasures.1976U.S.NationalForestManagementActisSarkar2005passed.1978HollingintroducesconceptofadaptiveSarkar2005management.(continued)CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n390sahotrasarkarTable21.1(continued)YearDevelopmentReferences1982PatilandTailliepublishstatisticalreviewPatilandTaillieofdiversitymeasures.Raoinventsthe1982;Rao1982;quadraticentropyindexcapturingSarkar2005distinctiveness.Marguleandotherseffectivelycriticizetheuseofislandbiogeographytodesignreserves.1983Kirkpatrickusesthefirstcomplementarity-Kirkpatrickandbasedareaselectionalgorithm.Harwood19831985FormationoftheSocietyforConservationSarkar2005;Soule´Biology(UnitedStates).Soule´publishesthe1985manifesto‘‘WhatIsConservationBiology?’’1986Publicitymaterialfor‘‘NationalForumonTakacs1996;BioDiversity’’includesfirstuseofthetermJanzen1986‘‘biodiversity.’’Janzenpublishesthemani-festo‘‘TheFutureofTropicalEcology.’’RecognitionofstalemateintheSLOSSdebate.1987ConservationBiologybeginspublication.Sarkar20041988PublicationofWilson,ed.,BioDiversity,Wilson1988;firstbookwith‘‘biodiversity’’initstitle.Margulesetal.Firstcomputeralgorithmbasedon1988;Magurrancomplementarity.Lastbook-length1988treatmentofdiversitymeasuresuntil2004.1989U.S.GAPAnalysisProgramlaunchedtoScottetal.1993identifyfeaturesofbiodiversitynotprotectedintheconservationareas.1991FirstissueofCanadianBiodiversity,thefirstSarkar2004journalwith‘‘biodiversity’’initstitle;namechangedtoGlobalBiodiversityin1993.1992ConventiononBiologicalDiversity(RiodeTakacs1996;Janeiro).WeitzmanattemptstorenewWeitzman1992interestinthegeneralstatisticsofdiversitymeasures.1999Tilmanmakesthecaseforarichness-stabilityTilman1999relation.2000MargulesandPresseypublishtheconsensusMargulesandframeworkforsystematicconservationPressey2000planninginNature.2004MagurranpublishesMeasuringBiologicalMagurran2004Diversity.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity391everydayuseinbiology.Theinterestingquestioniswhetherthereisanythingspecialabout‘‘biodiversity.’’Itwillbearguedlater,appearancestothecontrary,thatoperationalconceptsofbiodiversityusedwithincontemporaryconservationbiologysufferfromverylittleambiguity.However,thepolyvalenceof‘‘biodiversity’’andthefactthattherewasalmostnoattempttoquantifyitsmeasuresduringthefirstdecadeofitsusedosuggestthattheconceptofbiodiversitywasintroduceddenovo,thatis,indiscontinuitywiththedecades-oldtraditionofattemptstodefinequantitativemeasuresofecologicaldiversitywhichprecededit.ThenewconservationbiologistsintheUnitedStates,convincedthattheywereinvolvedintheestablishmentofadisciplinewithnoantecedent,atleastinadvertentlypromotedsucha4view.Theyalsoencouragedtheideathatanentirelynewconceptualapparatuswasbeingcreated.Thischapterwillexplorethisissueindetail,emphasizingtherootsofconceptsandmeasuresofbiodiversityinearlierquantitativemeasuresofecologicaldiversity.Thesecon-nectionsseemtohavebeenlargelyignoredinboththebiologicalandthephilosophicalliteraturebutareimportantforaphilosophicalunderstandingoftheconceptualstructureofconservationbiologyanditsrelationtoecology.Thegistoftheargumentisasfollows:Explicitscientificdiscussionsofbiodiversity,afterthecontractionwasfirstintroduced,implicitlypresumedwhathadbeenlearnedfromtheearlierdiscussionsofdiversitywithinecology.Theseearlierdiscussionsofecologicaldiversityhadreachedanimpassetoalargeextentbecauseofastalemateinthediversity-stabilitydebatebytheearly1980s.Consequently,itwasunclearwhetherquantitativemeasuresofecologicaldiversitycapturedanythingimportant.Conservationbiol-ogistsseemedtohavereactedtothissituationbyavoidingthesemeasuresintheirpractice.Anydiscussionofbiodiversityhadtokeeppartofitsfocusonthenormativejustificationforitsconservation,incontrasttotheearlierdiscussionsofecologicaldiversity.Thisencouragedashiftofdiscussionawayfromthequantitativemeasurestoconceptsthataremoreeasilyusedincontextsofdiscussionsofnormativity.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n392sahotrasarkarMoreover,biodiversityconservationalwaysoccurredinsociopoliticalcontextsinwhichconservationistshadtocontendwithotherclaimantsonland,forinstance,thosewhowouldconvertnaturalhabitatsintoagriculturalorindustrialproductionsystems.Discussionsofbiodiversitywereconstrainedbythiscontext,whichrequiredgeneralaccessibilitytomultiplestakeholdersandhelpedmovethediscussionsawayfromearliertechnicaltreatmentsofecologicaldiversity.Nevertheless,whenattemptsweremadetoquantifybiodiversitywithinconservationbiology,thoseearlierdiscussionsprovidedmanyoftheconceptualtoolsthatbecamerelevantthoughthehistoricalantecedentsofthenewmeasureswereoftenignored.Inparticular,measuresofwhatisknownasb-diversityarecriticallyrelevanttotheoperationallyquantifiedmeasuresofbiodiversitythatmustbeusedinsystematicconservationplanning.Thecrucialinnovationreponsibleforthisshiftwasthatplanningforconservationemergedasanalgorithmicprocessthatnecessitatedoperationalizationofbiodiversityasaquanti-tativeconcept.Philosophersandhistorianshaveexploredtherecenthistoryofecologyandhistoryofconservationbiologysosporadicallythatthehistoricalclaimsmadeduringthisargumentmustberegardedonlyasafirstattempttoreconstructthesedevelopments,subjecttowelcomerevisioninthefuture.2.background:measuresofecologicaldiversityThefirstquantitativeindexofecologicaldiversitywasproposedbyFisher,Corbet,andWilliams(1943)torelatethenumberofindi-vidualstothenumberofspeciesinasampledrawnfromanaturalcommunity.Startinginthelate1940s,Preston(1948,1962)attemptedtoextendFisher’swork.Allthesestatisticalmodelswerephenomenological,basedonthefitwithdata,ratherthanonbio-logicalprinciples,thoughFisher’sindexhasrecentlybeengivenatheoreticalfoundationbyHubbell(2001).However,duringthesameCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity393period,MacArthur(1957)approachedthesameproblemusingexplicitmechanisticmodelsbasedonpresumedbiologicalinterac-tions.Forinstance,MacArthur’s(1957)broken-stickmodel,whichwassupposedtopredictdiversity,assumedthatthespeciesina5communityapportionedresourcesatrandom.MacArthur’smanyinnovationsinthe1950spavedthewayforaresurgenttheoreticalecologyinthe1960sthatwentbeyondpopulationecology,leadingtothetheoryofislandbiogeographyand,eventually,thediversity-stabilityquestion(seeSection3).However,themostcommondiversitymeasuresusedinecologyemergedduringthesameperiodfromadifferentconceptualback-ground.Simpson(1949)introducedanindexof‘‘concentration’’Pn2Pnði¼1piwherepiisthefrequencyoftheithtypeandi¼1pi¼1Þ,thecomplementorinverseofwhichprovidesanaturalmeasureofdiversity.Margalef(1958)proposedtheuseofShannon’sinformationPnindexði¼1pilogpiÞasadiversityindex.Theintroductionofthesetwomeasuresledtoavarietyofothersandalargesetofempiricalstudiesinthe1960sthatmeasureddiversityinthefield(reviewedbyHutcheson1970;seeTable21.2).Nevertheless,thefundamentalquestionremainedopen:whatisthejustificationofthesemeasures?Hurlbert(1971)posedthequestionforcefullyandnofullysatisfactoryanswerhasyetbeenproduced.Conceptuallytherearetwoobviousoptionsforansweringthisquestion:(i)itcanbeshownthataproposedmeasurecapturessomeunproblematicintuitionaboutdiversitybylayingdownexplicitadequacyconditionsand,ideally,provingthatsomeproposedmea-sureistheonlyonesatisfyingtheseconditions;or(ii)theuseofameasurecanbejustifiedbyshowingitplaysanimportanttheoret-icalrolewithinecologybybeingusedinrelevantmodels.Thereareagaintwowaysinwhichthesecondoptioncanbecarriedout:(a)theproposeddiversitymeasuremaycapturesomefeaturethatisaresultofecologicalprocesses;or(b)itmaybesomethingthatplaysadeterminingroleinecologicalprocesses.Startinginthe1960s,allthreeofthesestrategiesofjustificationhavebeentriedbutwithlimitedsuccess.Proposedadequacyconditionsfordiversitymeasureshaverequiredthathigherdiversitymustresultinanincreaseof(1)therichnessornumberofspecies,(2)theevenness(orequitability)oftheirpresence,(3)theiraverageabundancerarityorlevelofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n394sahotrasarkarTable21.2.AdequacyConditionsforEcologicalaDiversityAdequacyconditionExplanationMeasuresincorporatingconditionRichnessNumberofRichness(PatilandTaillie1982);speciesatanareaSimpson(1949)measure;Shannonmeasure(Margalef1958)Evenness/EqualproportionalSimpson(1949)measure;Shannonequitabilityoccurrenceofallmeasure(Margalef1958);Hillspeciesatanarea(1973)index;Alatalo(1981)index;Pielou’s(1975)indices;redundancy(Patten1962);standarddeviation(Fager1972);numberofmoves(Fager1972);allmeasuresofintrasamplevariabil-itycanbeusedforthispurposeAverageLowproportionalSimpson(1949)measure;Shannonabundanceoccurrenceofameasure(Margalef1958);rarityspeciesatanareaHurlbert’s(1971)PIEindex;PatilandTaillie’s(1994)indexGeographicalLowtotaloccurrenceofRarity(Sarkaretal.2004)rarityaspeciesinaregionDistinctive-Phenotypic/taxonomicQuadraticentropy(Rao1982);allnessuniquenessofameasuresofphylogeneticspeciesatanareadifferencecanbeusedforthispurposeAbundanceTransferofRichness(PatilandTaillie1982);transferproportionfromaSimpson(1949)measure;Shannonlessabundance-raremeasure(Margalef1958);PatilandspeciestoamoreTaillie’s(1994)indexabundance-rarespecies(i.e.,therarespeciesincreasesinfrequencywhilethecommonspeciesdecreases)occurrence,(4)thehighestgeographicalrarityofspeciespresent,(5)theirdistinctiveness,and(6)theeffectofabundancetransferbetweentherarestandthemostcommonspecies.Table21.2explainstheseconditionsinmoredetailandlinkssomecommonmeasuresofecologicaldiversitytothem;theclassificationgiventhereisanextensionofthoseprovidedbyPielou(1975)andMagurran(1988)butCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity395probablystillnotcomplete;acompleteclassificationwouldnotbeentirelyunnoticedbyecologists.Withtheexceptionofcondition(4)allofthesemeasuresare‘‘local’’inthesensethattheirassessmentdoesnotrequireaccesstoinformationoutsideasystem;geographicalrarityhasglobalscope,buttherarityofspecieswithinasystemisstillafeatureofthatsystem–thispointwillbeimportantlater.Thetroublewiththisjustificatorystrategyisthatnosinglemeasurecansatisfyalltheseadequacyconditionsthoughthisresultapparentlyhasnotbeenprovedinfullgeneralityinaunifiedmathematicalframework.Pielou(1969)showedformallyhowtheShannonandSimpsonmeasuresincorporatedconditions(1)and(2).Ricotta(2005)showsthatcondition(5),formalizedasSchur-concavity,andcondition(6),formalizedasquadraticentropy(Rao1982),togetherleaveonlyrichnessasapossiblemeasureofdiversity(thatis,noothermeasurecansatisfythem).Thelastresultisunsatisfactorybecausediversityintuitivelymeansmorethanjustthenumberofspecies.Attheveryleastitmustrefertothedis-tributionofabundancesofthespecies(as,forinstance,capturedbytheevennesscondition).Thus,nosinglemeasureofdiversitycansatisfyallfiveconditionsofTable21.2.6Thesenegativeresultsmaybeinterpretedasshowingthatthereisnoconceptofdiversityatalltobeexplicated.Alternatively,theymaybeinterpretedasshowingthatnosinglemeasureofdiversitywillsimultaneouslyoptimizeallprevailingintuitions.Thesecondinterpretationismoreinteresting:(i)itsuggeststhattheseintuitionsbeexaminedtodeterminewhicharelessdispensablethanothers,and(ii)itleavesopentheoptiontopreferameasureofdiversityonthegroundthatitisconnectedtoecologicalprocesses.Table21.2showsthattheSimpsonandShannonmeasuresperformbetterthanallothermeasureswithrespecttothenumberofconditionstheymeet,apointthathasoftenbeenignoredwhennewmeasuresareproposed.Effortstofindconnectionsbetweendiversitymeasuresandeco-logicalprocesseshavealsofoundered.Withrespecttotheecologicaldeterminantsofbiodiversity,PatilandTaillie(1976,1979,1982)initiatedanambitiousprogramofderivingdiversity,interpretedasaverageabundancerarity,fromprobabilisticmodelsofinterspecificandintraspecificencountersbetweenorganisms.TheShannonandSimpsonmeasuresthenemergeasdifferentwaystocomputetheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n396sahotrasarkar7average.Mostcommentatorssincethe1980shaveignoredtheseefforts.Thetroubleisthatthesemodelshavenoobviousbearingonanyotherquestioninecology.Consequently,anydiversityindexemergingfromthesemodelsmaynotberelevantinotherecologicalcontexts;intheabsenceofsuchconnections,diversitywillplaynosignificanttheoreticalroleinecology.However,ifdiversityisinterpretedasrichness,anyattempttoderivespecies-areacurvesfromfundamentalprinciplesisamodelofdiversity.Similarly,thetheoryofislandbiogeography(MacArthurandWilson1967)isalsosuchatheoryofdiversityquarichness,as,moreexplicitly,isHubbell’s(2001)neutraltheoryofbiodiversity.Hutchinson’sjustlyfamousqueryaboutthesourceofvariationofspeciesrichnesswithlatitudealsoconstitutesacallforsuchatheoryofdiversityquarichness.But,asnotedearlier,thereismoretodiver-sitythanrichness.Diversity,byanymeasureotherthanrichness,isnoteasilyinterpretedasaresultofknownecologicalprocesses.3.stalemate:diversityandstabilityTurningtotheputativeecologicaleffectsofdiversity,amuchricher–andstillliving–traditionattemptstoconnectdiversitytoecologicalprocesses,toproductivity,andmoreimportantly,tostability(Til-man1999).Theideathatdiversityandstabilityaresomehowcon-nectedhasalongandcheckeredhistory(Pimm1991).MacArthurisprobablythefirsttohavemadetheclaimprecise,withElton(1958)andPimentel(1961)providinginitiallypromisingempiricalsupport.ConnellandOrias(1964)analyzedpatternsofdiversityandMacArthur(1975)producedtheoreticalargumentsinsupportofarelationship;however,Whittaker(1975)gavegroundsforcaution.Meanwhile,alsointhe1960s,theappropriatedefinitionofecolog-icalstabilitybecameascontroversialasthedefinitionofdiversity.Lewontin(1969a)introducedavarietyofinfluentialexactdefini-tions,noneofwhichcapturedallecologicalintuitions.Table21.3providesataxonomyofcommondefinitionsofstability.Aresultofthesemultipledefinitionsisthatanypresumeddiversity-stabilityrelationshipcanbeinterpretedinalargevarietyofways.Thereareatleast16differentmeasuresofa-diversityinTable21.2,and7differentmeasuresofstabilityinTable21.3,givingatleast112possiblepairedrelations.Thesearelowestimates:PimmCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity397(1991)identified5alternativeusesof‘‘stability’’butthesewereembeddedinahierarchyof3levelsofcomplexityand3levelsoforganization,whichresultin45measuresofstability.Onlyatinyfractionofthepossiblerelationshipshaveeverbeenexploredexperimentallyortheoretically.Thelastpointdoesnotseemtobegenerallyrecognizedintheecologicalliteraturethoughithasasignificantconsequence:anynegativeassessmentaboutthepossi-bilityofadiversity-stabilityrelationcannotbemorethantentative,subjecttofuturerevisionasmorepotentialrelationsareexplored.Thetheoreticalmodelsofthediversity-stabilityrelationthathavebeenexploredoftenproducedresultsthatseemedtodependonmodelingstrategies,whenmodelscouldbemathematicallyana-lyzedatall.Inparticular,May(1973)analyzedalargeclassofmodelsinwhichincreaseddiversity(intheformofwhathecalledcom-plexity)didnotleadtoincreasedstabilityandoftendelimitedthesetofconditionsinwhichstabilitycouldbemaintained.Manytheo-reticalmodelsremainpracticallyimpossibletotestinthefield.Inthe1980s,asconservationbiologywasemergingasanewrecog-nizablediscipline,thediversity-stabilitydebatehadreachedastale-mate.Thesituationisnotverydifferentsometwentyyearslater.Inthelastfewdecades,muchempiricalworkhasincreasinglyquestionedsupportforadiversity-stabilityrelation(Pimm1991,MuellerandJoshi2000).Recently,however,interestingpositiveresultswerereportedbyTilmanandcollaborators(LehmanandTilman2000),withdiversityinterpretedasrichnessandstabilityinterpretedasconstancy.Yet,thesereportshavebeenfollowedbyequallycompellingnegativeoneswithrichnessfoundtobeinverse-lycorrelatedwithstability,nowinterpretedasresilienceandresis-8tance(PfistererandSchmidt2002).Justasinthecaseofthestalemateoftheearly1980s,thedebateoverthediversity-stabilityrelationappearsoncemoretobeatanimpassethoughexperimentalworkcontinuesonseveralfronts.Yet,inthecontextofbiodiversityconservation,aputativediversity-stabilityrelationhasimmensenormativeforceandcon-9tinuestobeusedinthisfashion.Ifdiversityisatleastpartlydefinedbyspeciesrichness,sucharelationshipmayprovidearationalefortherivetargument(Sarkar2005):thedisappearanceofeachspeciesindividuallymayappeartobeinsignificant.But,likethefinalrivetthatdecidesaplane’sfate,onefinalextinctionmayleadtotheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nthestatevariableseffectsofaperturbation(e.g.,ofspecies’abundancesordensities)ortodecayrelativetotheinitialsizedensitiesorabundancesrelativetotheoriginalstate;(2)changeofspeciescompositionrelativetotheoriginalcompositionunits(perturbationmaybebioticorabiotic)MeasuredbythesameparametersasTheinverseofthetimetakenforthe(1)Inverseofthechangeofspecies’Perturbationsizemeasuredinnaturalordynamicafterachangeinthevalueofasystem’sstatevariablesareferencestateordynamicafteraperturbationinasystem,relativetoareferencestateordynamicafteraperturbationsustainandreturntoareferencestateordynamic(irrespectiveoftimetaken)Sizeoftheperturbationasystemcantolerance/domainofattractionCategoryLocalstabilityResilienceProbabilityofreturntoareferencestateDefinitionResistanceRateatwhichasystemreturnstoInverseofthemagnitudeofthechangePerturbationMeasureDefinitionsofStabilityModifiedfromSarkar(2005)basedcategoriesTable21.3.Perturbation-398CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nsustainsspecifiedminimumpopula-tionlevels,e.g.,nonextinctionofaproportionofitsspecies;(2)Thetimeasystemwillsustainspecifiedspeciescompositionsofsomeparameterofthesystemsuchasspeciesrichness,size,orbiomassabundanceandefficientlyasystemprocessesenergyandmaterialsandengagesinotherbiogeochemicalactivities(1)ThetimeforwhichasystemTheinverseofthesizeoffluctuationsMeasuredintermsofhowfaithfullyreferencestateordynamic(communityorpopulation)orpopulation)willcontinue‘‘func-tioning’’Probabilitythatasystem(community(LehmanandTilman2000)PersistenceAbilityofthesystemtocontinueinaConstancyReliabilityInverseofthevariabilityofasystemindependentcategoriesPerturbation-CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n400sahotrasarkarcollapseofthebiosphere.Someversionofthisargumentliesbehindmanyconsequentialistnormativerationalesforbiodiversitycon-servation.AsGoodman(1975,261)putsit:Thediversity-stabilityhypothesishasbeentrottedouttimeandtimeagainasanargumentforvariouspreservationistandenvironmentalistpolicies.Ithasseemedtoofferaneasywaytorefutethechargethatthesepoliciesrepresentnothingmorethanthesubjectivepreferencesofsomeminoritycon-stituencies....Fromapracticalstandpoint,thediversity-stabilityhypothesisisnotreallynecessary;evenifthehypothesisiscompletelyfalseitremainslogicallypossible–and,onthebestavailableevidence,verylikely–thatdisruptionofthepatternsofevolvedinteractioninnaturalcommunitieswillhaveuntoward,andoccasionallycatastrophic,consequences.Inotherwords,thoughthehypothesismaybefalse,thepoliciesitpromotesareprudent.Nevertheless,cogentnormativejustificationsofbiodiversityconservationmustrelyonotherarguments.Thesenegativedevelopmentsseemtosuggestthattheremaybenoveridicalconceptofecologicaldiversityand,consequently,nosuchconceptofbiodiversity.Butthiswouldbeanunwiseinterpretation:aconclusionaboutbiodiversitydoesnotfollowfromthepremiseaboutecologicalstabilityunlessthetwoconceptsarerelevantlyrelated.Inthecontextofbiodiversityanditsconservation,theconceptsofdiversitythatwereexploredearliermaywellhavebeeninappropriateinventory-basedconcepts,referringonlytowhatoccurswithinsys-temslocalizedatareas,ratherthandifference-basedconcepts,com-paringthedifferencesintheinventoriesattwoormoreareas.Whittaker(1960)elaboratedtherelevantdistinctionslongago,in1960.Hedistinguishedamonga-diversity,thediversitywithinanarea,b-diversity,thatbetweenareas;andc-diversity,orthetotaldiversityofaregion,includingbotha-andb-diversity.Thelattertwoarewhatarerelevanttobiodiversity,asthenextsectionwillemphasize.Themeasuresandadequacyconditionswehavebeenconsideringareallrelatedtoa-diversity;evengeographicalrarityofspeciesatanarea,aswasnotedearlier,refersdirectlyonlytowhatoccursatthatarea.4.biodiversityOnthebasisofasociologicalanalysis,Takacs(1996)andothershavearguedthattheterm‘‘biodiversity’’wasintroducedtomovepublicCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity401interestawayfromtheconservationofsinglecharismaticspecies(megafaunasuchaslargecolorfulbirdsorlargemammals)toallspecies.Thisissupposedtohavebeenparticularlyimportantforconservationintheneotropics,whicharelackingincharismaticspeciesbutwereunderspecialthreatduetoacceleratinghabitatdestructionsincetheearly1960s(Janzen1986).ThisviewispartlysupportedbyTakacs’scatalog(1996,46–50)anddefinitionssuchasthatusedintheConventiononBiodi-versity(ArticleII):‘‘‘Biologicaldiversity’meansthevariabilityamonglivingorganismsfromallsourcesincluding,interalia,terrestrial,marineandotheraquaticecosystemsandtheecologicalcomplexesofwhichtheyarepart;thisincludesdiversitywithinspecies,betweenspeciesandofecosystems.’’Morerecently,Wil-son(1997)argues:‘‘Biodiversityisdefinedasallhereditarilybasedvariationatalllevelsoforganization,fromgeneswithinasinglepopulationorspecies,tothespeciescomposingallorpartofalocalcommunity,andfinallytothecommunitiesthemselvesthatcomposethelivingpartsofthemultifariousecosystemsoftheworld.’’Thegratuitousreferencetoheredityaside,bothdefinitionsareessentiallythesame.Definitionssuchasthesearestillinventorybased,relyingonthecontentsofindividualsystemslikethedefinitionsofecologicaldiversitydiscussedinanearliersection;however,manyofTakacs’srespondentsexplicitlyappealinsteadtodifference-baseddefinitions,whichlookatdifferencesbetweenentitieswithinsystemsand,especially,betweensystems.Frombothpointsofview,earlierdiscussionsofdiversityintheecologicalliteratureseemtobeatbestmarginallyrelevanttothequestionofformulatingadefinitionofbiodiversityadequateforconservationbiology.However,atleastthreesetsofargumentsraisequestionsabouttheaccuracyofthelastthesis,thatis,thedenialoftherelevanceoftheearlierwork.First,noneoftheconceptsinvokedintheadequacyconditionsforecologicaldiversitymeasures(seeTable21.2)dependsoncharismaticspecies.Further,fromthebeginning,itwasrecog-nizedthatdefiningdiversitymeasureswasaformalstatisticalprob-lemwiththemeasuresapplicabletoanysetofentities,notjustspecies.Thisisclearinthediscussionsof(amongothers)Fisherandassociates(1943),Preston(1948),PatilandTaillie(1982),Rao(1982),andWeitzman(1992).Consequently,thegeneralizationofconcernCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n402sahotrasarkarfromcharismaticspeciestoallfeaturesofbiologicaldiversitydidnotprecludetheuseoftheearliermeasuresandanalyses.Second,evenifsomeoftheearlydiscussionsofbiodiversityignoredtheworkonecologicaldiversity,startinginthe1990s,quantitativemeasuresofbiodiversitybegantobeproposedwithinconservationbiologythough,initially,thesewereorientedtoward10taxonomicdiversity.Laterworkshowedthatmanyofthenon-taxonomicmeasurescanbeformallyrelatedtoecologicaldiversity11measures.Recentquantitativeworkonbiodiversityalsodrawsheavilyontheseearlierdiscussions(see,e.g.,Hubbell2001).Third,andmostimportant,onceb-andc-diversityareincluded(besidesa-diversity),theavailablemeasuresofdiversitycaptureeventhenewdifference-baseddefinitionsofbiodiversity(leavingoutonlythediversityofprocesses;seeSection6forelaborationofthispoint).Forinstance,someoftherecentworkondefiningbio-diversityforusewithinconservationbiologyhasfocusedonquan-titativeoperationalconcepts,especiallycomplementarity(Vane-Wright,Humphries,andWilliams1991,JustusandSarkar2002),whichcanbeexplicitlyrelatedtob-diversity(discussedlater).Theseoperationalquantifiedmeasuresofbiodiversityareimportantbecausesystematicconservationplanningemergedasanexplicitlyalgorithmicprocessthatmusthaveprecisegoalsbywhichtherelativesuccessofaplancanbeassessed.Themajorgoalofthealgorithmsistoselectthebestareaforconservationusingvariouscriteria,givenalistofareasandatleastlistsofbiodiversity‘‘sur-rogates’’presentinthem(withthesesurrogatesbeingfeaturesthataresupposedtomeasuretherelevantaspectsofbiodiversityforconservationplanning).However,thisraisestheinterestingunre-solvedphilosophicalquestionabouttherelationsbetweenthesemeasuresandtheintuitiveconceptsusedbythosewhoprovideddefinitionsforTakacs’s(1996)catalog(seethenextparagraph).Thequantitativecomplementarityvalueofanarea,relativetoasetofareasalreadyselectedforconservation,isthenumberofnew,orasyetonlyinadequatelyprotected,biodiversitysurrogates(suchasspecies,speciesassemblages,habitattypes)thatthisareacon-12tains.Areaswithhighercomplementarityvaluearegivenpref-erenceforprotectionoverareaswithlowervalues.Noticethatthisdefinitionassumesthattheproblemofwhatsurrogatesshouldbeusedhasalreadybeensolved.ButthedefinitionscatalogedbyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity403Takacs(1996)areintendedtospecifywhatshouldcountassurro-gates.Operationaldefinitionsofbiodiversityassumethattheappropriatesurrogatesaregivenandfocusontheirquantificationforconservationplanning.Theambiguity–andfreedom–inthechoiceofsurrogatesisunavoidable(Sarkar2002,2005),butmeasuresofecologicaldiversityareequallysubjecttothisproblem.Inecology,a-,b-,andc-diversityaretraditionallymeasuredusingspecies(as,also,wascomplementarity),butnobiologicalreasonprecludestheirbeingmeasuredbyusingothersurrogates–recallthatthesemea-suresareallultimatelystatisticalconcepts.Complementarityisacriticalconceptinthedesignofconserva-tionareanetworksaspartofbiodiversityconservationplanningbecauseitenablestheconstructionofsuchnetworkstomaximizebiodiversityrepresentationwhileminimizingtheareaincluded.Thisisimportantwhenbudgetaryconstraintslimittheamountof13landthatcanbededicatedtoconservation.Thealgorithmic(usuallyiterative)selectionofareasusingmeasuressuchascom-plementarityisknownbyvariousnames,including‘‘conservationareanetworkselection,’’‘‘placeprioritization,’’‘‘siteselection,’’and‘‘reserveselection’’inconservationbiology.Now,startingwithWhittaker(1960),quantitativemeasuresforb-diversityhavebeenproposedanddebatedthoughnottothesameextentasa-diversity.TraditionalstatisticalmeasuressuchastheJaccardindexandtheHammingdistancehavebeenusedtomeasure14b-diversity.Manyothershavebeensuggested.Itisclearthatthemeasuresofcomplementarityandb-diversityareformallyrelated.Perhapsbecausecomplementaritywasnotexplicitlyintroducedasaconceptofb-diversity,theconnectionhasonlyrecentlyreceivedan15overdueexplicitrecognitionintheecologicalliterature.AsMagurran(2003,172)putsit:‘‘Complementarityis...b-diversitybyanothername–themorecomplementarytwoareasare,thehighertheirb-diversity.’’Thisassessmentisbasicallycorrectbutrequiresanimportantmodificationinthecontextofconservationplanning.Becausecomplementarityisdefinedwithrespecttoanexistingselectedsetofareas,unliketraditionalmeasuresofb-diversity,itisnotasymmetricconceptexceptinthesimplest(ormathematicallydegenerate)caseinwhichtheexistingselectedsetconsistsofexactlyonearea.Ifareasareselectediteratively,asinmanyalgo-rithmsusedinconservationplanning,thecomplementarityvalueofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n404sahotrasarkaranareachangeswitheachiteration.Ingeneral,distancesbasedoncomplementaritymeasuresarenonmetric:theydonotsatisfythe16triangleinequality.Inanycase,theconceptualrelationbetweencomplementarityandb-diversityreiteratesthepointthattherearemoreconnectionsbetweentheolderworkonecologicaldiversityandconceptsofbiodiversitythanhasbeenrecognized.Moreover,complementarityisnottheonlyconceptbywhichbiodiversitycanbeoperationallyquantifiedforuseinplanning.Severalstudieshaveadvocatedtheuseofgeographicalrarityalongwithcomplementarity.Sarkarandcolleagues(2004)evenusedamixtureofcomplementarityandShannon’sandSimpson’sindicesfora-diversitythoughthesegenerallydidnotperformaswellinpracticeasamixtureofrarityandcomplementarityinrepresentingbiodiversityaseconomicallyaspossibleinaselectedsetofareas.Thus,eventheolderdiscussionofa-diversityreentersbiodiversityinthepracticeofcontemporaryconservationbiology.Whenmulti-pleconceptsareusedtoprioritizeareas,Sarkar(2002)hasarguedthatthecorrespondingalgorithmusedimplicitlydefinesbiodiversityinthatcontext.Therearemanyotherwaystoquantifybiodiversityoperationally,andc-diversitydoesnotenterthisdiscussiononlybecauseitisyettobesystematicallyused.Therearethustwosourcesofthepolyvalenceofbiodiversitywithwhichthischapterstarted:(i)thechoiceofsurrogatesand(ii)thechoiceofanoperationaldefinitionforbiodiversity.Onthisissue,biodiversityissimilartotheearlierecologicalconceptsofdiversity.Oncebiodiversitysurrogatesandoperationaldefinitionsarese-lected,nofurtherambiguityofthetermremains.Inthisrespect,biodiversityismuchbettersituatedthandefinitionsofspeciesorgenes.Forthelasttwoconcepts,unlikethecaseofbiodiversity,disambiguationproblemsremainevenafteraparticulardefinitionisselected.5.historicalspeculationsItremainstoexplainwhy,inthecontextofthe1980s,ascon-servationbiologyemergedasarecognizablescience,distinctfromecologybutdrawingheavilyfromit,biodiversityasaconceptwasnotexplicitlyconnectedtothewidelyavailableoldermeasuresofecologicaldiversity.IntheabsenceofsufficientlydetailedhistoricalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity405reconstructionsoftheemergenceofconservationbiology,anyexplanationmustbepartlyspeculative.Onelikelypossibilityisthatthestalemateinthediversity-stabilitydebategeneratedskepticismaboutwhethertraditionalconceptsofecologicaldiversitycanbesalvagedforanypurposeatall.Hurlbert(1971)arguedthatdiversityisa‘‘non-concept’’asearlyas1970andsuchconcernswereonlyexacerbatedbycontinuedfailurestoresolvethedebate.AsTable21.1indicates,bythemid-1980s,interestindevisingandstudyingmeasuresofecologicaldiversitylargelyevaporated.Evenwhentherewasarevivalofinterestinthediversity-stabilityquestioninthe1990s,richnesswastheonlymeasureofdiversitytobetrackedand,asnotedearlier,richnessisnotagoodmeasureofdiversity.Thiscontrastssharplywithstability;inthatcase,allthemeasuresinTable21.3havebeenusedinrecentyears(seeMcCann2000).Asnotedinthelastsection,thequestionwhetherrichnesscorrelateswithanymeasureofsta-bilityremainsopen.Howothermeasuresofdiversitywouldperformremainsinsufficientlyexplored.Givenallthisuncertaintyandlackofscientificsuccess,inthe1980sitwouldprobablyhavebeenatleastpracticallyimprudenttobasepoliticalargumentsforconservationontheexistingecologicaldiversitymeasures.Itisalsolikelythatnotmuchattentionwaspaidtotheearlierworkbecauseitwasnotenvisionedthatconservationbiologywouldrequireaquantitativeconceptofbiodiversityoreventhatitwouldbeuseful.Thesociopoliticalcontextofconservationrequiredcon-servationproposalstobeunderstoodbynonspecialiststakeholders.Popularscientificbooksratherthantechnicalmonographswereperhapsinevitablytheprimarymeansofcommunicatingwiththesestakeholders,andquantitativemeasuresofdiversity(whichwerebelievedtobeofdoubtfulutility,inanycase)hadlittleroletoplayin17suchacontext.Fromthebeginning,acentraltheoreticalandpracticalproblemhasbeenthedesignofnetworksofreservesandotherconservationareas.However,earlyeffortsconcentratedontheuseofislandbiogeographytheory(Diamond1975);eventhoughthattheorywasquantitative,itsusetosuggestheuristicrulesforreservedesignremainednonquantitative.Itspawnedthe‘‘single-large-or-several-small’’(SLOSS)debate,which,afteradecadeofacrimony,alsoendedinastalemate.MeanwhiletheuseofislandbiogeographytheoryforthispurposewasseverelycriticizedbyCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n406sahotrasarkarseveralgroupsonthegroundthattheanalogybetweenoceanicislandsandconservationareaswasfaultyandempiricallyungrounded(seeSarkar2005forahistory).Allthesediscussionsremainednonquantitative.Quantitativeplanning-orientedeffortsatdefiningbiodiversityoccurredinachangedcontextaftersystematicconservationplan-ning(MargulesandPressey2000)hademergedasanalgorithmicenterprisewithinconservationbiology.WhilethesealgorithmicapproachesdatebacktoKirkpatrick’sworkinTasmaniain1983(KirkpatrickandHarwood1983),withMargules,Nicholls,andPressey(1988)alreadyproducingsoftwareimplementingcom-plementarity-basedalgorithmsin1988,theyremainedcontroversialeveninthelate1990s(Sarkar2005).EventheGAPAnalysisPro-gram,launchedintheUnitedStatestoidentifygapsinwhatisprotectedinconservationareasnationally,wasslowtoadoptquantitativemethodsfortheselectionofnewconservationareas(Scottetal.1993).Itis,therefore,notsurprisingthattherewaslittleconcernforquantitativemeasuresofbiodiversityduringthefirstdecadeofconservationbiology,1986–96.Asseenearlierinthissection,thoughattemptstoconnectbiodiversitytoearliermeasuresofecologicaldiversitybeganintheearly1990s,almostallsuchworkdatesfromthelate1990sinthechangedcontextinwhichquanti-tativemeasuresofbiodiversitywerebeingdevisedandexploredbecausetheywerenecessaryforalgorithmicselectionofreservesandotherconservationareas.6.finalremarksThediscussionsoftheearliersectionsshouldhavemadeclearthatthereisadeepconnectionbetweenbiodiversity,whenoperationallyquantifiedthroughmeasuressuchascomplementarity,andthetraditionalconceptofb-diversity.Onceoperationalquantificationbeginstouseothermeasures,includinggeographicalrarity,evena-diversitybecomesimportant.Itisonlybecausebiodiversityoftenfunctionedasanormativeconceptinthe1980s,andbecausecon-servationbiologywasinitiallynotconceivedasrequiringquantita-tiveassessmentsofbiodiversity,thattheseconnectionshavenotpreviouslybeenexplored.Thepolyvalenceof‘‘biodiversity’’isalsorelativelyeasilyresolved.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity407However,animportantcaveatisnecessary:theconceptsofdiversity(ecologicaldiversityorbiodiversity)consideredreferonlytothestructuralandtaxonomichierarchiesofbiologicalorganiza-tionandnottoquestionsoffunctionalorbehavioralorganization.Yet,itisclearthatsomebehaviors,suchasthemigrationofNorthAmericanmonarchbutterflies(Danausplexippus),constituteendangeredbiologicalphenomena(BrowerandMalcolm1991)thatshouldformpartoftheconceptofbiodiversity.Sarkar(2002,2005)haspointedoutthatnotonlydoproposeddefinitionsofbiodiversityfailtocapturesuchbehavioraldiversitybutattemptstooper-ationalizethesedefinitionsareyettoshowsignsofplausibility,letalonesuccess.Twofinalpoints,oneaboutthenormativecontextofbiodiversityconservation,andtheotheraboutculturaldifferencesthatmustbefacedinpracticalcontexts,willendthischapter.Fromitsveryinception,conservationbiologywasconceivedofasnotonlybeingadescriptivescience,butalsoashavinganormativegoalthatmakesitmoreanalogoustomedicinethanbiology(Soule´1985).Therehavebeenmanyattemptstoviewbiodiversityasanormativeconcept.Meanwhile,therehasbeenmucheffortwithinconservationbiologytoconnectbiodiversitywithecosystemfunction,andtheseattemptsareclearlyanalogoustotheattemptstoconnectdiversitywithsta-bility.However,thereisacriticaldifference:inthecaseofbiodi-versity,theappealtoecosystemfunctiontypicallyformspartofanormative(consequentialist)argumenttodefendconservation(Sarkar2005).Inthecaseofecologicaldiversity,theputativecon-nectiontostability(orproductivity,etc.)doesnotnecessarilyhavethatnormativerolethoughthemotivationforthesearchforadiversity-stabilityrelationshipmaywelllieinthehopeforfindingsuchanormativeconclusion(Goodman1975).Moreover,biodi-versityconservationdoesnotoccurinasociopoliticalvacuum.Rather,conservationmustcompetewithotherpotentialusesofland,includingbiologicalandmineralresourceextraction,recrea-tion,conversionforagriculturalandindustrialdevelopment,andwildernesspreservation.Ifbiodiversityconservationmusttrumptheseotherclaims,theremustbeaconvincingnormativerationaleforitsgreatervalue.Aquestionthatbearsfurtherinvestigationiswhether–and,ifso,towhatextent–suchanormativeroleforbiodiversityhasinfluencedthewayitissupposedtobedefined.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n408sahotrasarkarVaryingsocialandculturalcontextsalsoimposedifferentcon-straintsonthegoalsandvalueofconservationplanning.Localandglobalstakeholdersmaydifferinwhatconstitutesthebiodiversitythatdeservesprotection.CriticsfromtheSouthhavelongarguedthatNorthernattitudestothenaturalworldareirrelevanttotheexigenciesofthosemateriallystrugglingforsurvival(Sarkar2005).However,thisdebatehasbeenmoreaboutthevalueofwildernessthanbiodiversitythoughthereisalsodisagreementaboutwhethertraditionalculturesvaluebiodiversityingeneraloronlythatpartofitthatshouldbeconservedasresources.Thediscussionsinthischaptersimplyignorethiskindofquestionaboutthemeaningofbiodiversity.Itassumesthatculturalconcernsaboutwhatconstitutesbiodiversitycanbeincorporatedintothechoiceofsurrogates.Ultimatelyitmaywellbethatthenormativityofconservationbiologyandthenecessityofincorporatingsociopoliticalconsidera-tionsindifferentculturalcontextsarewhatdistinguishbiodiversityfromearlierconceptsofecologicaldiversity.ACKNOWLEDGMENTSThanksareduetoJustinGarsonandJamesJustusforcommentsonanearlierdraftofthispaper.notes1.See,forexample,Norton(1987)andSarkar(2005).2.Thehistoryrelevantforthischapterisrelegatedtoatableforlackofspace.3.Gaston(1996)andTakacs(1996)providecatalogsofvariantdefini-tionsdefendedbyconservationbiologists.4.Sarkar(2005)givesadetailedhistoryandanalyzestherhetoricoftheearlyyearsofconservationbiology.5.MacArthuralsohadtwoothermodels,but,inthesediscussions,thebroken-stickattainediconicstatusasanullmodelwhiletheotherswerelargelyignored–theseotherswillalsobeignoredhere.6.Therearealargenumberofothersimilarresults–see,e.g.,Hill(1973),Alatalo(1981),Rousseauetal.(1999).TheyhavealsobeenreviewedbyPeet(1974),Poole(1974),Pielou(1975),Magurran(1988),SmithandWilson(1996),and,mostrecently,Ricotta(2005).CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nFromEcologicalDiversitytoBiodiversity4097.Evenearlier,Hurlbert(1971)hadmaderelatedsuggestions.8.Forthestateofplay,seethecommentarybyNaeem(2002).9.SeeMcCann(2000)forarecentexample.10.See,forexample,Vane-Wrightetal.(1991).11.Ricotta(2005)providesacriticaldiscussion.12.JustusandSarkar(2002)givethehistoryoftheuseofcomplemen-tarityinconservationplanning.13.Sarkaretal.(2004)reviewwhatcanandcannotbeachievedbycomplementarity-basedalgorithms,especiallyincomparisontooptimalalgorithmstosolvethesameformalizedproblems.Earlierliteratureiscondensedinthispiece.14.Koleffetal.(2003)provideafairlycomprehensivesurvey,withthelatterexploringtwenty-fourmeasuresapplicabletobinaryspeciespresence-absencedata.15.See,however,Hooper(1998)forsomescatteredremarksthatindicatesomerecognition.16.Intheliteratureonalgorithmdesign,theuseofcomplementarityisknownasthe‘‘greedyalgorithm.’’17.SomeoftheseissuesaredocumentedinTakacs(1996),butmuchmorehistoricalanalysisremainstobedone.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nrobertt.pennock22BiologyandReligionintroductionThehistorical,conceptual,andculturalinterplaybetweenbiologyandreligioninvolvesacomplexandphilosophicallyfascinatingsetofrelationships.Certainlythesimpleviewthatreligionhasuniformlybeenahindrancetobiologicalresearchisanunfaircar-icature.Religionhassometimeshadapositiveeffect,oftenindi-rectlystimulatingorevendirectlyencouragingscientificresearchofthebiologicalworld.Similarly,biologyhashadaprofoundeffectonreligion,sometimesofferingchallengesthatrequirebelieverstoreassessbasictheologicalassumptions.Scholarshaveexaminedbothdirectionsofinfluence,findingbothexpectedandunexpectedcon-nections(e.g.,Cantor2005,Rolston1999,Russell,Stoeger,andAyala1999).Thischapterwillgiveanecessarilyselectiveaccountofsomeofthemutualinfluencesbetweenbiologyandreligion.Itsmajorpurposewillbetohighlightadeeppatternunderlyingtheirinterplay,namely,thepervasiveeffectofthereligiousideaofthedivinelycreatednormativityofnature.Howthispatternisexemplifiedinmultiplewaysintheongoingcreationistcontroversywillbearecurringexample,butwewillalsoexaminehowitmaybeseeninawiderangeofissuesfromquestionsaboutwhatitmeanstobehuman;toreligiouswarningsagainst‘‘playingGod’’;toviewsaboutgenderrolesandsexualmorality,environmentalism,personhood,andthestatusofthefetus;andultimatelytometaphysicsandthequestionofultimatepriorityincreation.Webeginbylookingatbiologicalevolution,acriticaltopicthatintertwineswithmanyoftheseothercases.410CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion411religionandevolutionCertainlyoneofthebestcasesbywhichtoexaminetheinterplaybetweenscienceandreligionistheongoingreligiouscontroversyoverevolutionandcreation.Althoughthiscontroversyisoftenstereotypedaslittlemorethanasimpleattackuponbiologybynaı¨vebiblicalliteralists,thissuperficialviewmissesthewaythebattledistillstheessenceofthedeepconceptualdivisionsinthewaysthatpeopleconceptualizeanddealwithscientificandreligiousworld-viewsandtheirimplications.Aswewillsee,theissuetouchesnotonlyuponthegrossfeaturesoftheso-calledculturewarandthebroadstrugglebetweenscienceandreligion,butisconnectedtoawiderangeofthephilosophicallyinterestingtopicsintherelation-shipofbiologyandreligion,includingevidential,ethical,meta-physical,andevenexistentialquestions.BeingHumanTheeffectthatthediscoveryofevolutionhadandcontinuestohaveuponthereligious,especiallytheChristian,worldviewmaybeevenmoreprofoundthantheearlierscientificrevolutionsthatdisplacedhumanbeingsfromthephysicalcenterofcreation.Ifwewerenotspeciallycreated,whatdoesitmeantobehuman?DoeshumannessbeginandendwiththebiologicalnotionofthespeciesHomosapiens?ThetraditionalChristianunderstandingofhumandignitytookittobebasedintheGenesisnotionthathumansarecreatedinGod’simage.Whatcanberetainedofhumanuniquenessanddignityifweevolvedfromapesandare‘‘just’’onemorebranchontheevolutionarytree?Andwhathappensnowthatgenetictechnologyputsourfutureevolutioninourownhands?Throughitsdiscoveriesandthequestionsthattheseleadto,biologyhaschallengedreligiousthinkerstoreassessoldnotionsofImagoDei.Thoughaliteralreadingof‘‘God’simage’’wouldsuggestthathumanbeingsresembleGodinvisualshape,surelyitisanoddnotioninmorewaysthanonethathumanbeingsgettheirdignitybecausetheirbodylookslikeGod’s.Evolutiongivesfurtherimpetustoseparatethatnotionfromthebodyitselfandtofocusonmoreabstractcandidatesforthehumanessence.ItiscommonnowforreligiousthinkerstoidentifyinsteadtraitssuchashumanfreedomCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n412robertt.pennockandethics.SomelikethephilosopherofreligionandbiologyPatriciaA.Williamsplacethisexplicitlyinanevolutionaryframeworkandseehumanfreedomasemergingashominidsevolvedtobeabletorecognizeandmakechoices.WilliamshasnoproblemacceptingevolutionandarguesthatGod’slovewasfirstmanifestedintheuniversewhenitcouldbeexpressedthroughtheevolutionofcrea-tureswithsymboliclanguage.Thescientificdiscoveryofourevolu-tionaryconnectionstoalllivingthings,shesays,helpsusunderstandthatthecommandmenttoloveone’sneighborshouldembraceallofthelivingworld(Williams2001).WhileWilliams’sparticularaccountmaynotyetbewidelyshared,itdemonstrateshowreligiousviewsareabsorbingandpro-cessingdiscoveriesfrombiology.Similarprocessingandaccom-modationhavehappenednumeroustimesinthepast.Togivejustonefurtherexample,considerhistoricalchangesinviewsregardingthepossibilityoflivingbeingsonotherworlds,anideathatevolu-tionarybiologymakesoneconsidermoreseriously.ExtraterrestrialLifeDiscussionsofthepossibilityoflifeonotherplanetsgobackwellbeforeDarwin.IndeFontenelle’s1686classicConversationsonthePluralityofWorlds,aphilosopherandhishostessdiscusstheimplicationsofthenewCopernicancosmologyandfindtheideaofpeopleontheMoonandplanetstobealight-heartedandentrancingpossibility.ButWilliamWhewell’s1853treatmentoftheissuewasmoresober.InOfthePluralityofWorlds,Whewellarguedthatnolifeexistedanywhereelseintheuniverse(WhewellandRuse2001).Hisargumentborethestampofatheologicalworry:God’srela-tionshiptohumansissupposedtobepersonalandunique,withtheideaofHisappearingonEarthasaman,thesaviorJesus,beingsofundamentaltotheChristianviewthatitwouldbeimpossibleandequallyrepugnanttoimagineeitherotherworldswithanalogoussaviorsorbereftofsalvation.AdmittingthepossibilityoflivingbeingsonotherworldswouldthreatenourspecialrelationshipwithGod,nottomentionopeningthedoortosupportersofevolution.Indeed,ThomasPaineusedjustthiskindofanargumenttocriticizeChristianbeliefs,citingtheabsurdityoftheideaofGod’stravelingfromworldtoworldinanendlesssuccessionofdeathtoredeemtheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion413progenyofalienAdamsandEves.ButfewcontemporaryChristiansseembotheredbythepossibilityofextraterrestriallifeandtheCatholicphilosopherofscienceErnanMcMullin(2000)hasexplainedhowChristiantheologycanbeformulatedtoaccommodatesuchnotions.RethinkingReligioninLightofEvolutionItisimportanttorecognizetheextenttowhichmostmainstreamreligionshavealreadyaccommodatedevolutionarybiology.EvenChristiantheology,forwhichevolutionmightbethoughttoposethegreatestchallenges,hasforthemostpartmadepeacewiththefindingsofDarwinandtheevolutionarybiologistswhohavefol-lowed.MainstreamChristianityhasdonethisinmuchthesamewaythatiteventuallycametotermswiththeearlierchallengesfromphysicsandastronomy.RatherthanfollowingasimplisticreadingofScripture,religiousthinkershavefollowedGalileo’sadvicethattruthshouldnotcon-tradicttruthandthatbelieversshouldallowGodtospeakthroughthebookoftheworldandnotjustthebookoftheword.RatherthaninsistingthatGodmusthaveusedadirectformofmiraculouscreationtobringaboutbiologicalcomplexity,theyhavetakenamorebroad-mindedandgenerousviewofGod’spowersofcreation.Theyallow,oreveninsist,thatGodcreatedtheworldindirectly,byendowingitwithacompletesetoflawsthatdidnotrequireperiodicinterventionandadjustment.RatherthantyingChristiantheologytometaphorsofGod’s‘‘design’’or‘‘plan’’,theologianslikeJohnHaught(2000)suggestthatthesemaybemisleading,andthatnotionsofGod’s‘‘vision’’or‘‘dream’’fortheuniversemightbebetter.Ratherthangettinghungupovertheevolutionofourmaterialbody,theysay,rememberthatwhatisessentialtoChristiandoctrineistheimmaterialsoul.Fortheseandotherreasons,theCatholicChurchandotherChristiandenominationssaythatevolutionshouldnotbeseenasinconflictwithChristianfaith.Noraretheseonlyrecentaccom-modations.EveninDarwin’sdaymanyevangelicalChristianswereamongtheearliestdefendersofthetheory,asdocumentedinDavidLivingstone’s(1987)revealinghistoryoftheencounterbetweenevangelicaltheologyandevolutionarythought.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n414robertt.pennockCreationismHowever,evolutioncontinuestobeviewedastheologicallyanath-ematomostChristianfundamentalists,especiallyintheUnitedStates.WithaboutathirdofallAmericansidentifyingthemselvesasfundamentalistaccordingtopolls,andmanyevangelicalsholdingsimilarviews,thereisahugereceptiveaudienceforvariousformsofcreationism.Thegeneralconceptofcreationismistherejectionofthescien-tificaccountofevolutioninfavorofsupernaturalspecialcreation,buttherearemanyvariationsofthisideathatreflectdifferenttheologicalassumptions.ByfarthedominantformofcreationismisthevarietythatholdsthattheEarthis6,000to10,000yearsold.These‘‘youngEarth’’creationistscalculatefromthedaysandthegenerationslistedintheBibletogetthisfigure.However,othercreationistsreadthedaysofcreationinGenesisasbeinglongagesoftime,sinceadayfromGod’spointofviewneednotcorrespondtooursolarday,andsotheyacceptthestandardscientificchronology.Other‘‘oldEarth’’creationistsinterpretScriptureinotherwaysthatallowthis.Somecreationistsholdthatthemajorglobalgeographicalfeatureswerecausedbyacatastrophicworldwideflood,whileothersbelievethatNoah’sfloodmayhavebeenlocalor‘‘tranquil’’.Andsoon.Moreover,therearealsonon-Christiancreationistswhorejectevolutioninfavorofthecreationstoriesoftheirownreligions(Pennock1999).ThepubliccontroversyaboutcreationismasweexperienceittodaycanbetracedtotheEppersonv.ArkansasdecisionoftheUnitedStatesSupremeCourtin1968,whichruledthatlawsthatbannedtheteachingofevolutionwereunconstitutional.Whenthecourtsalsoruledthatteachingbiblicalcreationismwasuncon-stitutional,youngEarthcreationiststriedanewapproach.DroppingtheovertreferencestoScripture,theyclaimedtopromotemerelyascientificview,whichtheycalled‘‘creationscience’’.Theyhadsomesuccessatgettinglegislaturestopass‘‘balancedtreatment’’acts,butagainthecourtsfoundsuchlawstobeunconstitutional.ThetestimonyofthephilosopherofbiologyMichaelRuseonthenatureofscienceprovidedacriticalelementofthedecision.In1987,theSupremeCourtruledinEdwardsv.Aguillardthatcreationsciencewasnotsciencebutwasdisguisedreligion,andthusthatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion415teachingitinthepublicschoolsviolatedtheEstablishmentClauseoftheConstitution.Quicklyadaptingtothisloss,creationistschangedtheirterminology.Forinstance,manuscriptsofamajorcreationistbiologytextbook–OfPandasandPeople–thatwasinpreparationdroppedtheterm‘‘creation-science’’in1987immedi-atelyaftertheEdwardsdecisionandreplaceditwiththeterm‘‘intelligentdesign’’.Forobviousreasons,intelligentdesign(ID)advocatesnowdenythattheyarecreationists,buttheirhistoryandtheirsubstantiveviewsshowotherwise.Theintelligentdesignmovementismostcharacterizedbywhatitcalled‘‘theWedge’’,astrategydevisedbyPhillipJohnson,alawprofessorwhoiscreditedwithnegotiatingatrucebetweenyoungEarthandoldEarthcreationiststoimprovetheirchanceofsuccessbyunitingaroundabannerof‘‘merecreation’’.Solongastheystayedfocusedontheircommonviewthatevolutionarythinkingisprofoundlyanti-Christian,andthatGod,notnaturalprocesses,createdtheworldandhumanbeings,theycouldagreetemporarilytodroptheirbattlesaboutthefloodandtheageoftheEarth.Suchissuescouldbeintroducedlaterafterthesharpedgeofthewedgehadpenetratedtheconstitutionalbarrier.IDisoftenmisidentifiedasaformofoldEarthcreationism;infact,theyoungEarthersinthegrouphavesimplyagreedtoholdoffpressingtheircaseuntilIDreachestheclassroom.TheWedgewasalsoametaphortheyusedtospeakofhowtheywouldsplitapartthematerialist,naturalistworldviewofsciencesotheycouldreplaceitwiththeirtheisticscience.Whenlobbyingfortheirviewtobetaughtinthepublicschools,theycontinuedtheoldcreationscienceclaimthatIDwasbasedentirelyinscienceandwasnotreligious.Whenspeakingwithsupporters,however,theyrevealedthehiddenagenda.HereisarepresentativeexamplefromPhillipJohnson:MycolleaguesandIspeakoftheisticrealismorsometimes,merecreationasthedefiningconceptofourmovement.ThismeansthatweaffirmthatGodisobjectivelyrealasCreator,andthattherealityofGodistangiblyrecordedinevidenceaccessibletoscience,particularlyinbiology.WeavoidthetangledargumentsabouthoworwhethertoreconciletheBiblicalaccountwiththepresentstateofscientificknowledge,becausewethinktheseissuescanbemuchmoreconstructivelyengagedwhenwehaveascientificpicturethatisnotdistortedbynaturalisticprejudice.Iflifeisnotsimplymatterevolvingbynaturalselection,butissomethingthathadtobeCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n416robertt.pennockdesignedbyacreatorwhoisreal,thenthenatureofthatcreator,andthepossibilityofrevelation,willbecomeamatterofwidespreadinterestamongthoughtfulpeoplewhoarecurrentlybeingtaughtthatevolutionarysciencehasshownGodtobeaproductofthehumanimagination.(Johnson1996,emphasisadded)TheultimategoalofIDcreationistsistoreintroducesupernaturalexplanationsintobiologyandintosciencegenerally.Theyclaimthatbiologicalcomplexityinparticular,butalsootherfunctionalcom-plexitytheybelievetheycanidentifyintheuniverse,canbeexplainedonlybythepurposefulactionofatranscendentintelligence.religiousexplanationinbiologyOfcourse,creationistsarenotaloneinofferingmysticalexplana-tions.Norisitonlyintryingtoexplainthecreationoflifeanditsmyriadcomplexitiesthatreligiousbelievershaveappealedtothesupernatural.Wecanhereonlybrieflytouchuponafewrepre-sentativecasesofthemanyexamplesofpurportedreligiousexpla-nationsinbiology.Animportantsetofexamplesinvolveshowtounderstandillnessanddisease.EarlyinklingsofthetransitionfromareligioustoascientificexplanatoryframeworkareoftenillustratedintheHip-pocraticwritingsonepilepsy.Ratherthanthinkingofepilepticseizuresasa‘‘sacreddisease’’involvingsomesortofdivineposses-sion,Hippocratesrecommendedthatmedicaldoctorsunderstanditinnaturalterms.Theideathatsicknessistheresultofpossession,perhapsbyevilspirits,iscommonacrossmanyreligioustraditions,notjustanimisticones.RenaissanceChristiansofferedsimilarexplanationsforthedancingmania,whatbecameknownasSt.Vitusdance.Onthiskindofreligiousviewofdisease,curesneces-sarilywillinvolveanappeasementoforstrugglewithimmaterialspirits.Thematerialisticexplanationsofmedicalsciencemaybeviewedasirrelevantorevenassuspect.Arelatedreligiousconceptofdiseaseisthatitistheresultofsin.Insomecasesthediseaseistakentobeapunishmentforsin,sometimesevendirectlyinflictedbyGod.AclassicexampleofthisviewofdiseaseasdivinepunishmentviewwasseenintheplaguesoftheMiddleAges,whichledtoself-flagellationandotherdisplaysofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion417publicpenanceasbelieversattemptedtoatoneforsomeunknownoffensetoGod.PaintingsfromtheperioddisplayChrist’sthrowingboltsofplaguefromheaventopunishthesinners.Noraresuchreligious‘‘explanations’’intermsofimmaterialspiritsanddivinepunishmentconfinedtothedarkagesofhistory.AmongthemanyrecentexamplesoneparticularlysalientonewastheviewexpressedbymanyfundamentalistChristiansthatthediseasethateventuallywasidentifiedasAIDSbutthathadinitiallybeencalledthe‘‘gaydisease’’wasGod’spunishmentforthesinofhomosexuality.Evenafterthehumanimmunodeficiencyvirus(HIV)wasidentifiedasthecause,manycontinuedtoviewthedis-easeasamodernplaguesenttopunishsinnersandquestioned1whetherHIVwasindeedthecause.Suchexamplesofreligiousexplanationsofdiseaseareillustrativeofthewayinwhichthisinfluentialreligiousworldviewmixesempiricalandmoralissues.Whatweareseeinginsuchcasesistheplayingoutofthereligiousviewofthenormativityofnature.normativityofnatureTheideathatnaturehasabuilt-inmoralstructureiscommontomanyreligions,butherewewillfocusonthewaythisisseenfromaChristianperspective.Thebasicideaissimple.Goddesignedtheworldwithaplaninmind.Hesetitupandsawtoitthattheworldwasgood.LeavingasidethecomplexitiesassociatedwiththeFall,thegeneralideaisthatGodcreatedtheworldwithaninnatemoralstructure.PersonswhooverlookthisdivinelycreatedinherentnormativestructurecannotpossiblyhaveethicalrelationshipswiththeworldandwithGod.CreationandMoralityTheintelligentdesignleaderPhillipJohnsonhasexplainedwhatheandwhatallcreationiststaketobethesignificanceoftheevolution/creationdebate.Ifyouhaveabiblicalcreationstory,thengettingtherightrelationshipwithGodandgettingtoheavenarethemostimportantthings.Ifyouthrowthatoverboardandyouhaveanaturalisticcreationstory,thosethingsbecomeunimportantandwhatbecomesimportantishowweapplyscientificCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n418robertt.pennockknowledgetomakeaheavenhereonearth.That’sadreamofvariouskindsofreformprograms–socialism,forexample.(QuotedinGoode1999)ThisviewthattheChristiancreationstoryinformsusofhowtohavetherightrelationshipwithGodiscommonamongmanyChristianbelievers.Fundamentalistsinparticularlooktothebib-licalstoriesofcreationforguidanceabouttheplanthatGodhasforhumanbeings.Forinstance,thestoryofhowGodcreatedoneman,Adam,andthenlateronewomen,Eve,isseentobeinformativeoftheproperbiologicalorderofcreation.Itdefinesthepropercorecomponentsofthefamily–onemanandonewoman.Itdefineswhataresupposedtobetheappropriategenderroles:thewomenistobethehelpmateofherhusband.Itexplainswhywomenshouldsufferthepainsofchildbirth:theyareapunishmentforEve’ssin.Andsoon.Again,theworldisseentohaveabuilt-innormativity,muchofitrelatedtosexualmorality.Thisviewisbynomeanspeculiartocreationism,butitisper-vasiveincreationistwritings.AFreudiancouldeasilyanalyzecreationismasasublimationanddisplacementofarepressedsexualobsession.idandsexCreationsciencewritingsarerifewithwarningsabouthowevolu-tionistthinkingistoblameforsexualpromiscuity,divorce,pornog-raphy,abortion,andevenbestiality.Intelligentdesigncreationistsmakereferencetoexactlythesamelist.PhilipJohnsonandothersregularlyillustratewhattheytaketobeatstakeinthebattlebetweenthenaturalisticworldviewofevolutionandthetheisticworldviewofintelligentdesignusingexamplesinvolvingsex.Pre-maritalsex,adultery,divorce,andflexiblegenderroles,allpur-portedlythefruitsoftheformer,areputupagainstchastity,faithful,stablemarriages,and‘‘proper’’genderrolesthataresupposedlythefruitsofthelatter.InasinglebookJohnsontwicementionssex-educationclassesinwhichgirlspracticeunrollingcondomsovercucumbersasanexampleofthesorryeffectsoftheevolutionist,naturalistworldview.Butcreationiststypicallyreservetheirgreatestireforwhatthey,asdootherChristianfundamentalists,seeastheworstofthesexualCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion419sinsagainstnature.Inarguingagainstevolution,creationistsreg-ularlycitethelinesfromRomans1thatsaysthateversincethecreationoftheworldGod’sinvisibleattributesofeternalpoweranddivinityhavebeenabletobeunderstoodandperceivedinwhathehasmade.Thosewhorefusetoseethisarefoolswithoutexcuse.Johnsonsaysthattheself-deceptivethinkingofevolutionistsfur-theraffirmsthecorrectnessofthatchapterofthebiblicalworldview.AndwhatelsedoesRomans1say?Thatwhentheyexchangethetruthofcreationforthelieofevolution,Godhandsthesesinnersovertotheunnaturalpassionsofhomosexuality.homosexualityChristianfundamentalistsandevangelicalstakehomosexualitytobeoneofthemajorproblemsoftheday.Theculturewaragainstwhattheyseeastheimmoralityofhomosexualitygoeshandinglovewiththewaragainstevolution.Bothariseinpartfromthereligiousideaofnormativitybuiltintonature.Thisisanotherexampleinwhichbiologyasasciencebothaffectsandisaffectedbyreligion.Scienceingeneralandbiologyinparticularpotentiallyhavealottosayabouthomosexuality.ThepioneeringKinseyreportmadepeoplequestionlong-heldassumptionsabouttheprevalenceofhomosexualbehavior.Coulditbetruethat10percentofpeoplearehomosexuals?Studiesbyethologistsrevealedcasesofhomosexualbehavioramonganimalsinthewild.Aretherereallylesbiansea-gulls?Studiesbygeneticistssuggestedthepossibilityofageneticbasistohomosexuality.Istherereallyagaygene?SuchinformationisextremelysalientforreligiousbelieverswhothinkthatGodbuiltnormativityintocreation.ReligiousconservativestakesexualorientationtobeamatterofchoiceandbelievethathomosexualityisasinfulchoicethatgoesagainstnatureasGodintendedit.Butifhomosexualityhasabio-logicalbasis,thenitisnotsoeasytodismissitas‘‘unnatural’’.Ifpeopleare‘‘borngay’’,thencantheybeblamedforwhattheyare?Ifsexualorientationisfixedbiologically,thendoesitmakesensetosaythatcounselingcancureit?Studieshaveshownthatpeoplearelesslikelytoviewhomosexualityasinherentlyimmoraliftheybelieveitisbiologicallydetermined.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n420robertt.pennockThisreactiontoinformationfrombiologyrevealsatensioninthebasisofthereligiousobjectiontohomosexuality.MostChristianobjectionstohomosexualitystemfromtakingittobemorallycondemnedintheBible.ButifCreationwasdesignedwithanintrinsicnormativestructure,thenbiologicalinformationmaynecessitateareassessment.Indeed,thiskindofargumentfrombiologyissufficientlycompellingthatsomeChristiansmodifytheirviewtosaythathomosexualorientationisnotitselfimmoralbutthathomosexualbehavioris.Otherschallengethescience,arguingthatstudiessuggestingabiologicalbasisofsexualorienta-tionareflawedandthatgayscanindeedbecuredofwhatreallyisapathology.Insuchanatmosphere,itisnowonderthatbiologicalresearchonhomosexualitysuchasthatofDeanHamerandSimonLeVay,whoclaimedtofindevidenceofageneticinfluenceinmalehomo-sexuality,becomeshighlycontentiousandpoliticized.Thedebateismadeevenmorecomplicatedbysomephilosopherswhoaredis-missiveofbiologyandarguethathomosexualityissimplyasocialconstruction.Andthereisthemoregeneralcriticismthatmuchofthisargumentonbothsidesisbasedonafundamentalmistakeinethicalreasoninginvolvingthenaturalisticfallacy.Moreover,therearealsoreasonable,independentobjectionstogeneticreductionismandtheideathattherecouldbeagaygene.PhilosopherslikeMichaelRuse(1990)havesteppedintohelpsortouttheseandotherissues.environmentalismItisworthgivingonemoreimportantexampletoillustratethewayinwhichbiologyandreligionbecomeentangledbecauseofassumptionsaboutthenormativityofnature.Religiousassump-tionsmayhaveaprofoundeffectonthewaythatpeopleviewthevalueoftheenvironment.Inaninfluentialarticle‘‘TheHistoricalRootsofOurEcologicCrisis’’,LynnWhiteJr.arguedthatJudeo-ChristianreligiousassumptionsthatpermeateWesternculturearelargelythesourceoftheattitudesthatheblamedforenvironmentaldegradation.Amongtheseattitudes,heclaimed,isafaithinperpetualprogressthatisindefensibleapartfromaparticularkindofteleology.ItarisesfromCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion421astoryofcreationinwhichGodmakestheworldandalllivingthingsfortheexpressbenefitofman.(Womanisanafterthoughtandalsocreatedforman’sbenefit,topreventhimfrombeinglonely.)Accordingtothisreligiousviewnoiteminphysicalcreation,saysWhite,hasanypurposebuttoserveman’spurposes.Thisisareli-giousphilosophythatseesmanasthemasterofnatureandashavingdominionoverall,secondonlytoGodHimself.Psalm8,forexample,speaksofGodasgivingmandominionovercreation,puttingalllivingthingsunderhisfeet.AndofcoursethejustificationreachesbacktoGenesis1,whereGodsaystoAdamandEve,‘‘Befruitful,multiply,filltheearthandconquerit.Bemastersofthefishofthesea,thebirdsofheavenandalllivinganimalsontheearth.’’WhitesaysthatinthissenseChristianityisthemostanthropocentricreligiontheworldhaseverseen.Hewarnedthat‘‘weshallcontinuetohaveaworseningecologiccrisisuntilwerejecttheChristianaxiomthatnaturehasnoreasonforexistencesavetoserveman’’(White1967,54).WhiterecognizedthatChristianityiscomplexandlookedtoseewhetherithadtheologicalresourcesthatmightmitigatethisbasicproblem.HesuggestedthemodelofSaintFrancisofAssisi,whoemphasizedhumilityofmanindividuallyandasaspecies,andattemptedtopromotewhatWhitethoughtwasamoredemocraticvisionofGod’screation,exemplifiedinhisnotionofBrotherAntandSisterFire.Thiswasprobablytoosimplisticasolution.AlthoughSaintFrancisdidtalkwiththeanimalsasbrothers,whentheyspokebacktohimtheyrepeatedthesameproblematicbiblicalteleology,sayingthattheyexisted‘‘foryoursake,oman’’.However,ChristiantheologianswhohadfeltrebukedbyWhite’schargessubsequentlytriedtofindanalternativescripturalbasisforanenvironmentalethic.Shiftingemphasisfrompassagesinwhichthedominionmodelisrooted,theydrewuponpassageslikethatofGod’scovenantwithNoahandallofcreationandupontheparableofthegoodstewardintheGospelofLuke.TherainbowwasasignofGod’scovenantnotjustwithmanbutwiththeEarthandapromisenevertodestroyitagain.AndJesus’sparableofthegoodstewardreinterpretstheideaofdominiontoincludearesponsibilityofstewardship–‘‘Whenamanhashadagreatdealgiventohimontrust,evenmorewillbeexpectedofhim’’(Luke12:48–49).Asthestewardcaresforthehousehold,soshouldmancarefortheEarthCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n422robertt.pennockanditscreatures(Wright1989).WhileliberalChristiandenomina-tionsmaynothaveneededthiskindofjustification,itsarticulationhashadadocumentableeffectonsomeconservativebelievers.ThenascentenvironmentalmovementamongevangelicalChristianswhohaveadoptedthisperspectiveisapositivedevelopment.Unfortunately,thereisanotherreligiousimpedimenttosolvingourenvironmentalproblems,thisoneespeciallyfoundinthebeliefsofcertainfundamentalistChristiansectsintheUnitedStateswhoholdthatwearealreadylivinginwhattheycallthe‘‘endtimes’’.Theyholdthatbiblicalprophesies,particularlyintheBookofRevelation,thatdiscussthesignsandeventsthatwillheraldtheendoftheworldindicatethatthiswilllikelyhappeninourlifetime.Whyworryaboutlossofbiodiversity,globalwarming,orotherenvironmentalproblemsiftheworldisabouttoendandthetruebelieversaretobetakentoheaventositbyGod?ThisattitudewasexemplifiedbyJamesWatt,secretaryoftheinteriorunderPresidentReagan,whoinpublictestimonybeforeCongresssaidthatiswasunimportanttoprotectnaturalresourcesbecauseoftheimminentreturnofJesus–‘‘[A]fterthelasttreeisfelled,Christwillcomeback.’’Environmentaldestructionmaynotonlybesafelyignored,butactuallywelcomedorevenhastenedforthesebelieversastakingusmorerapidlytowardtheapocalypseand‘‘therapture’’.Giventhatpollsshowthat59percentofAmericansbelievethatthepropheciesofRevelationaregoingtocometrue,thisreligiousimpedimenttoenvironmentalcaremaybealargerproblemthaneventheChristianstewardshipmodelofthegoodSaintFranciscanovercome(Moyers2004).bioethicsand‘‘playinggod’’Thissamekindofviewofthenormativityofnatureisbehindreli-giouswarningsagainst‘‘playingGod’’thatarecommoninreligiousmoralassessmentsinbioethics.WeshouldnotplayGodandmodifytheworld,somehold,becauseGodalreadysetitupandsawtoitthattheworldwasgood.Godhasaplanfortheworldandforeachperson–eveninbirthanddeath–soitwouldbewrongtointerfereinsuchmatters.Muchofthephilosophicalliteratureinbioethics,andcertainlymostofwhatisdiscussedinphilosophycoursesonbio-medicalethics,involvesresponsestoordefensesofreligiousmoralCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion423beliefsofthissort.Wehavespacehereonlytoenumeratebrieflyafewofthese.Beginningwithbioethicalissuesinvolvingtheendoflife,onecouldexamineatlengthhowthiskindofreligiousviewplaysoutintheliteratureonthedefinitionofdeath,oneuthanasiaandphysi-cian-assistedsuicide.Thebiologicalnotionofbraindeath,forinstance,continuestomeetresistancefromreligiousconservatives,asdoesthepracticeofremovinglife-sustainingtechnologyorhas-teningdeathoftheterminallyillbylethalinjection.Indeed,therewereevenearlyobjectionstowhatwascalledthe‘‘Godsquad’’,ahospitalcommitteethatwoulddecidehowtoallocatescarcekidneydialysismachineswhenthesewerefirstdeveloped.OnlyGodshoulddecidewhoshouldliveanddieandwhenalifeisover.Similarly,formanybelievers,onlyGodshoulddecidewhenalifemaybegin.Moralobjectionsagainstcontraception,invitrofertilization(IVF),andreproductivecloninginlargemeasurederivefromthiskindofreligiousviewthathumanbeingsmaynotusurpGod’spowerstocreatenewlife.Therecontinuestobeavigorouspublicdebateabouttheseissues.Bioethicistshavestruggledwiththiskindofreligiouslybasedobjectiontobiomedicaltechnologies,mostrecentlyasarticulatedintheworksofLeonKass,appointedbythesecondPresidentBushtoheadthePresident’sCouncilonBioethicsinlargepartbecauseofKass’ssupportofthesesortsofreligiousviews.Onefindsthesamekindofassumptionbehindreligiousobjec-tionstoothersortsofbiotechnology.Althoughthereareotherkindsofargumentsusedaswell,acommonreasonforoppositiontogeneticengineeringisthatnatureisputtogetherasitisforagoodreasonandthatitishubristopresumetoimproveuponit.ItisnowonderthattheFrankensteinstory,whichdrawsuponthemythofPrometheus’stheftoffirefromthegods–essentiallyacreationstoryoftheoriginofourcontroloffire–isthedominanttropeusedagainstbioengineering.Ofcourseoneshouldnotignoreotherimportantreligiousassumptionsthatunderliecontemporarybioethicalcontroversies.Thecontroversyoverstemcellresearch,forinstance,exemplifieshowmanydifferentreligiousassumptionscancomeintoplay.ThinkerswritingfromaJewishperspective,forinstance,haveheldthattheirreligiousteachingswouldsupportstemcellresearch.JudaismplacesahighvalueuponhumanlifeandholdsthatonehasareligiousdutytoCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n424robertt.pennockcareforone’sbody.Thesickshouldacceptallmedicaltreatments,anditisourdutytodoeverythingreasonablyinourpowertocombattheravagesofthebody.Catholicreligiousthinkers,ontheotherhand,aswellasconservativeProtestants,viewthisparticularissuemoreinlightoftheirreligiousviewsaboutthestatusofthefetus.Ifthefetusisapersonevenintheearliestmomentafterconception,thenhowcanitbeusedtosupplystemcells,evenforagoodpurpose?metaphysicsThislastpointtakesustoissuesofreligiousmetaphysics.Thewayinwhichsuchbelieversjudgethemoralstatusofthefetusisclearlyafunctionoftheirreligiousbeliefsaboutthetruenatureofperson-hood.Itgoeswithoutsayingthatthisisthemainsourceofreligiousoppositiontoabortion.Informationfrombiologyisusedbyallsidesintheabortioncontroversy,butnobiologicaldatacanresolvewhatatbasedependsonthemetaphysicalassumptionthatGodinfusestheconceptuswithahumansoulandthatthatiswhatreallydeterminespersonhood.Suchmetaphysicalreligiousbeliefsaboutthesoulareequallyinplayindebatesaboutthenatureofmindandofthepossibilityoffreewill.Thosewhoholdthatitistheimmaterialsoulthatmakeshumanbeingsuniqueamongtheanimalswillbeunlikelytoacceptbiologicalexplanationsofhumanaction.ThistakesusbacktotheIDcreationists,whoholdjustthiskindofaview.Theirbasicassumptionisthatneitherevolution,noranybiologicalprocess,norevenanynaturalphysicalprocessiscapableinprincipleofaccountingforintelligentaction.Thecreationofinformation,theysay,necessarilyrequiresadesigningintelligenceandthisintelligencecannotbemerelynatural;itmusttranscendthematerialworld.Thisincludeshumanbeings,whoare‘‘embodied’’intelligences,whichistheirwayofspeakingofthesoul.Thisisthekeytotheirargumentagainstevolution,againstscientificnat-uralism,andfortheexistenceofthetranscendentmasterintellectwhoisresponsibleforthecomplexityoftheuniverse.ItisforthisreasonthatIDcreationistsseeChristianmetaphysicsasatstakeintheevolutiondebate.PhilipJohnsonexplainedthesignificanceofthisissueforIDCswhenhewasaskedwhyhefocuseduponDarwinism.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion425IwantedtoknowwhetherthefundamentalsoftheChristianworldviewwerefactorfantasy.Darwinismisalogicalplacetobeginbecause,ifDarwinismistrue,Christianmetaphysicsisfantasy.(QuotedinAnonymous2002)Butthisisaskingtoomuch.Understandingthelimitsofscientificmethodologyoughttohelpushere.Tothedegreethatsuchspiritualpossibilitiesareunderstoodastrulysupernatural,theywillforeverremainoutsidescience.Solvingsuchultimatereligiousmetaphys-icalquestionsismorethanoneshouldaskofbiology.priorityclaims:whichexplainswhich?Attheendoftheday,muchofwhatistakentobeatstaketurnsonthequestionofwhetherpeoplethinkthatreligionexplainsbiologyorbiologyexplainsreligion.Aswehaveseen,creationiststhinkthatifevolutionistrue,thenalloftheirreligiousbeliefsmustbefalse,soitisnowonderthattheyopposeitsovigorously.Thejustificationofamoralcode,writesPhilipJohnson,dependsupongettingthecreationstoryright.TheChristianstoryisoneinwhichGodcreatedhumanbeings,whosesinsseparatethemfromGodandwhomustbesavedfromsintobecomewhole.TheEnlightenmentstoryisofhumanbeingswhosemasteryofscienceenablesthemtoescapefromsuperstitionandeventuallyrealizethattheirancestorscreatedGodratherthanthereverse.Formanyreligiousbelievers,thatprioritydisputeistheultimatephilosophicalquestionforunder-standingtherelationshipbetweenbiologyandreligion.DidGodcreatehumanbeings,ordidhumanbeingsevolveandthenmakeuptheideaofGod?PsychologicalaccountsoftheideaofGodasaprojectionofinfantileimagesofthefatherandmotherarewellknown,andsomebiologistshavegivensimilardeflationaryaccountsofreligion.ThephilosopherDanielDennett(2006)recentlyappliedthebiologistRichardDawkins’sconceptofthememe,aculturalanalogtothebiologicalconceptofthegene,toreligiousideas,analyzingreligionasanaturalphenomenon.Othershavetriedthisinevenmorereductionistevolutionaryterms.InthebookTheBiologyofReligion,forinstance,VernonReynoldsandRalphTannerinvestigatedhowindividuals’religiousfaithormembershipinareligiousgroupaffectedtheirchancesofsurvivalandtheirreproductivesuccess.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n426robertt.pennockThatistosay,theyattemptedtogiveasociobiologicalaccountofreligiousbeliefandreligiouspractices.Marshalinghistoricalandcontemporarycross-culturaldataaboutawiderangeofreligions,theylookedatthebiologicaleffectsofreligiousbeliefsandpracticesinvolvingconceptionandcontraception,abortionandinfanticide,birthandchildhood,marriage,death,disease,andmore.Theycon-cludedthatreligiouspracticeswerebiologicallyadaptive,arisingfrompastsurvivalstrategiesandcontinuingtoenhancereproductivefitness.Religions,astheyratheraudaciouslystatedit,are‘‘cultu-rallyphrasedbiologicalmessages...aprimarysetof‘reproductiverules’’’(ReynoldsandTanner1983,294).ItisnotablethatReynoldsandTannertooktoheartthecritiqueofsociobiologyandwhatthephilosopherofbiologyPhilipKitchercalledits‘‘vaultingambition’’toexplainsuchculturalcomplexities.Theydroppedthetheoreticalframeworkofsociobiologyalmostentirelywhentheyrevisedthebook,retitlingitTheSocialEcologyofReligionandcoiningtheterm‘‘socioecology’’fortheirmoreeclectic(andultimatelyunsatisfying)theoreticalperspective.Theyalsofoundempiricalweaknessesintheirearlierview,recognizingthatreligiouspracticescanexacerbateaswellasreducetheriskofdisease,afactthatshouldnothavebeenthe‘‘surprisingdiscovery’’theyclaimedittobe(ReynoldsandTanner1995,17).Theirmajorrevisedconclusionistheweakeronethatreligions‘‘evolvedtoprovidelegitimatingof‘safe’waysofdealingwiththoseeventsinlifethatbringhumanbeingsintoastateofdangerorfearoranxietyorjustanoverwhelmingfeelingofpointlessness’’(1995,42)and‘‘toendow...lifeeventswithmeaning’’(1995,308).However,thereislittlebiologicalcontenttothenotionof‘‘evolved’’hereanymore.Theybelievethatbiologystillhassomethingtosayaboutwhyreligionsexist,buttheiranalysisisnotaimedasexplainingawayreligiousbelief.TheGodModuleHowever,biologymayyettrytodothis.Biologyhasaddedfueltothedebaterecentlyinadifferentway,inwhathasbecomeknownasthefieldofneurotheology.Somebiologistshaveclaimedtofindevidenceofa‘‘Godmodule’’inthebrain.Inparticular,theyfindanassociationbetweenepilepticseizuresinthelefttemporallobeandCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nBiologyandReligion427feelingsofecstasysometimesdescribedasexperiencesofthepres-enceofGod.Thisworktakesofffromafactthathaslongbeenknown,namely,thatsomesubjectsaffectedbytemporallobeepilepsyreporthavingintensespiritualexperiencesduringtheirseizures;someclaimthatGodspoketothemdirectly.Suchpatientswouldoftenbecomepreoccupiedwithspiritualissuesevenduringseizure-freeperiods.Experimentsusingtranscranialmagneticstimulatorsshowedthatonecouldproducethesekindsofeffectsinsubjectswithnohistoryoftemporallobeseizures.OneresearcherwhostimulatedhisowntemporallobesreportedbeingamazedathavingtheexperienceofGodforthefirsttimeinhislife.Asmaybeexpected,whilesomepeoplehavetakenthisbrainareatobetheseatofaspecialhumanfacultyforexperiencingthedivine,othersseeitasconfirmationthatsuchreligiousexperiencesaredelusionscausedbyelectricaldisturbancesinthebrain.TheneurologistV.S.Ramachandrandiscusseshisresearchontheneuralbasisofreligiousexperienceinthesamewayashedoeshisworkwithpeoplewhofeelphantomlimbsorwhoseecartooncharactersinavisualblindspot.Thisrephrasestheearlierquestionsothatwemaynowask,DidGodcreatethebrain,ordidthebraincreateGod?Inrecentwork,thepsychiatristEugeneD’Aguiliandtheradi-ologistAndrewNewbergusedhigh-techimagingdevicestoobservethebrainsofBuddhistsandnunsduringmeditation.WhenthesesubjectsreportedsubjectivefeelingsofonenesswiththeuniverseorofthepresenceofGodwhileinfocusedmeditation,theresearchersobserveddecreasedactivityinthebrain’s‘‘objectassociationareas’’thatpurportedlyprocessandmediatetheboundarybetweenselfandthemselves.Thedatashow,claimNewbergandD’Aguili(2002),thatmysticalexperienceisnotamerefabricationorasimpleresultofwishfulthinking,butratherhasareal,neurologicalbasis.Moreover,theysaythattheseexperiencesoccuraspartofnormal,healthyneuro-physiologyandshouldnotbedismissedasrandom,pathologicalevents.Mysticalexperience,theysay,isbiologically,observably,andscientificallyreal.TheyarguethathumansseekGodbecauseourbrainsarebiologicallyprogrammedtodoso,hypothesizingthatspiritualexperienceisintimatelyinterwovenwithhumanbiology.Continuingresearchisinvestigatingquestionssuchaswhetherreligiousritualcancreateitsownneurologicalenvironment,andCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n428robertt.pennockwhetherthereisaconnectionbetweenreligiousecstasyandsexualorgasm.BeliefinGodwillnotgoaway,theyconclude,becausethereligiousimpulseishard-wiredinthebiologyofthehumanbrain.Somesuggestthatthisisthecommonbiologicaloriginofallreligions.TheologicalresponsestothiskindofworkrangefromtakingittobesuggestivethatGodisbothrealandreachable,tocriticizingitasaformofscientism.Criticsquestiontheappropriatenessoftryingtomeasuremysticalexperience,suggestingthatitisamistaketothinkthattheologicalnotionsofthetranscendentcouldcorrespondtoempiricalobservations.DidGodcreatenature,ordidnaturecreateGod?Sufficetosay,neitherbiologynorreligionisyetinapositiontoclaimthefinalanswer.note1.Althoughhedidnotspeakofitintheseterms,itisaninterestingcoincidencethatatthesametimethatPhillipJohnsonwaswritinghisinitialarticlesandbookonintelligentdesign,healsobeganwritinginsupportofPeterDuesberg’sdissidentviewthatHIVdoesnotcauseAIDSbutratherthatitistheresultofthehomosexuallifestyle,includinglong-termconsumptionofrecreationaldrugsand/ortheAIDSdrugAZT.Johnsonrecentlywroteofa‘‘racket’’of‘‘AIDScareerists’’whomaybecoveringupa‘‘ghastlymistake’’(2004).HeandotherIDcreationistsmakethesamekindofintimationsofaconspiracyamongscientiststocoverupthepurportedfalseandfraudulentevidenceforevolution.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nrobertj.richards23TheMoralGrammarofNarrativesinHistoryofBiologyTheCaseofHaeckelandNaziBiologyintroduction:scientifichistoryInhisinaugurallectureatCambridgeasRegiusProfessorofModernHistoryin1895,LordJohnActonurgedthatthehistoriandelivermoraljudgmentsonthefiguresofhisresearch.Actondeclaimed:Iexhortyounevertodebasethemoralcurrencyortolowerthestandardofrectitude,buttotryothersbythefinalmaximthatgovernsyourownlivesandtosuffernomanandnocausetoescapetheundyingpenaltywhichhistoryhasthepowertoinflictonwrong.(Acton1906,234)In1902,theyearafterActondied,thepresidentoftheAmericanHistoricalAssociation,HenryLea,indubiouscelebrationofhisBritishcolleague,respondedtotheexordiumwithacontraryclaimaboutthehistorian’sobligation,namely,objectivelytorenderthefactsofhistorywithoutsubjectivemoralizing.ReferringtoActon’slecture,Leadeclared:Imustconfessthattomeallthisseemstobebasedonfalsepremisesandtoleadtounfortunateconclusionsastotheobjectsandpurposesofhistory,howevermuchitmayservetogivepointandpiquancytoanarrative,tostimulatetheinterestsofthecausalreaderbyheighteninglightsanddeepeningshadows,andtosubservethepurposeofpropagatingtheopinionsofthewriter.(Lea1904,234)AsPeterNovickhasdetailedinhisaccountoftheAmericanhistoricalprofession,bytheturnofthecenturyhistoriansintheUnitedStateshadbeguntheirquestforscientificstatus,whichfor429CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n430robertj.richardsmostseemedtoprecludetheleakageofmoralopinionintotheobjectiverecoveryofthepast–atleastinanovertway.Novickalsocataloguesthestumblingfailuresofthisnobledream,whenpoliticalpartisanshipandrampantnationalismsulliedtheideal(Novick1988).Historiansinourowntimecontinuetobewaryofrenderingexplicitmoralpronouncements,thinkingitaderogationoftheirobligations.Onoccasion,somehistorianshavebeenmovedtoembracetheoppositeattitude,especiallywhenconsideringthehorrendouseventsofthetwentiethcentury–theHolocaust,forinstance.Itwouldseeminhumanetodescribesucheventsinmorallyneutralterms.Yetevenaboutoccurrencesofthiskind,mosthistoriansassumethatanymoraljudgmentsoughttobedeliveredasobiterdicta,notreallypartoftheobjectiveaccountoftheseevents.Leathoughtacleandepictionofdespicableindividualsandactionswouldnaturallyprovokereadersintomakingtheirownmoraljudgmentsaboutthepast,withoutthehistorian’scoercingtheiropinions.ThisattitudeofstudiedneutralityhasbecomecodifiedinthecommandmentshandeddownbytheNationalCenterforHistoryintheSchools,whosecommitteehasrecentlyproclaimed:‘‘Teachersshouldnotusehistoricaleventstohammerhometheirownfavoritemorallesson’’(Nash1996).Presumablythisgoesaswellforthehistorianteacher,whosetextsthestudentsstudy.Andonemightsupposethatwhenthenarrativedescribesepisodesinthehistoryofscience,occasionforintrusivemoralassessmentwouldbequitelimited.Ibelievethatthesedemandsthathistoriansdisavowmoralassessmentneglectacrucialaspectofthewritingofhistory,whetheritbepoliticalhistoryorhistoryofscience:thedeepgrammarofnarrativehistoryrequiresthatmoraljudgmentsberendered.AndthatisthethesisIwillargueinthisessay,namely,thatallhistoricalnarrativesmustmakemoralassessments.Iwillbeespeciallycon-cernedwithanevaluationthatmightbecalledthatof‘‘historicalresponsibility.’’Theroleofmoraljudgmentaboutpasthistoricalcharactershas,despitecausalassumptiontothecontrary,beenbroughttoeruptiveboilrecentlyinoneareaofhistoryofbiology–thatofnineteenth-andearlytwentieth-centuryevolutionarythoughtinGermany.TheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology431individualaboutwhomconsiderablehistoricalandmoralcon-troversyswirlsisErnstHaeckel(1839–1919).IwillsaymoreaboutHaeckelinamoment.Heoffersatestcaseformythesis.NowIwillsimplypointoutthatHaeckel,morethananyotherindividual,wasresponsibleforthewarfarethatbrokeoutinthesecondhalfofthenineteenthcenturybetweenevolutionarytheoristsandreligiouslymindedthinkers,awarfarethatcontinuesunabatedinthecon-temporaryculturalstrugglebetweenadvocatesofintelligentdesignandthosedefendingrealbiologicalscience.Mymotivationforconsideringthemoralstructureofnarrativesisencapsulatedinthemaintitleofabookthatwaspublishednotlongago:FromDarwintoHitler.Thepivotalcharacterinthishis-toricaldescent,accordingtotheauthor,isErnstHaeckel.HeandDarwinareimplicitlychargedwithhistoricalresponsibilityforactsthatoccurredaftertheythemselvesdied.Idonotthinkjudgmentsofthiskind,thoseattributingmoralresponsibilityacrossdecades,areunwarrantedinprinciple.Thewarrantliesinthegrammarofhistoricalnarrative.WhetherthisparticularcondemnationofDarwinandHaeckelisappropriateremainsquiteanothermatter.thetemporalandcausalgrammarofnarrativehistoryLetmefocus,foramoment,ontwofeaturesofnarrativehistoryasapreludetomyargumentandasanillustrationofwhatImeanbythegrammarofnarrative.Thisconcernsthewaystimeandcausalityarerepresentedinnarrativehistories.Eachseepsintonarrativesinatleastfourdifferentways(seeRichards1992,19–53).Letmefirstconsider,quitebriefly,thetemporaldimensionsofnarrative.Initially,wemightdistinguishwhatmightbecalledthetimeofevents.Embeddedinthedeepstructureofnarrativeisthetimeduringwhicheventsoccur;thatsortoftimeflowsequitablyonintothefuture,witheachunithavingequalduration.NarrativesprojecteventsasoccurringinaNewtoniantime.Thiskindoftimeallowsthehistoriantoplaceeventsinachronology,tocomparethedura-tionofevents,andtolocatetheminrespecttooneanotherasantecedent,simultaneous,orsuccessive.Butthestructuringoftheseeventsinanarrativealsoexhibitswhatmightbecallednarrativetime,andthisisadifferentsortoftemporalCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n432robertj.richardsmodality.Consider,forinstance,HaroldPinter’splayBetrayal.ThefirstsceneissettemporallytowardtheendoftheNewtoniansequencedramatized,andthenextscene,goingintherightdirection,occursafewdayslater.Butthethirdscenefallsbacktotwoyearsbefore,andthefourthayearbeforethat,withsubsequentscenestakingusbackfinallytoaperiodsixyearsbeforethefinaldayswithwhichtheplaybegins.Theaudience,however,neverlosesitstem-poralbearingsorbelievesthattimestaggersalong,weavingbackandforthlikeanundergraduateleavingthelocalcollegepub.Thehistorianmightstructurehisorhernarrativeinaroughlycomparableway.Heorshemightrelateoneaspectofthehistoryandthenreturntoanearliertimetofollowoutanotherthreadofthestory.Orthehistorianmighthavethenarrativejumpintothefuturetohighlightthesignificanceoftheprecipitatingantecedentevent.Again,whenthisisdonewithmodestdexterity,thereaderisneverconfusedabouttheNewtonianflowoftime.Thetimeofnarrationisalessfamiliardevicebywhichhistoriansrestructurerealtimeaswellasnarrativetime.Oneoftheseveralmodesbywhichhistoriansconstructthiskindoftimeisthroughcontractionandexpansionofsentenceduration.LetmeillustratewhatImeanbyreferencetoahistorywithwhichmostreaderswillbefamiliar–Thucydides’sHistoryofthePeloponnesianWar.Atthebeginningofhishistory,Thucydides–afounder,alongwithHe-rodotus,ofthegenreofnarrativehistory–expendsafewparagraphsoneventsoccurringintheearliestperiodofCretanhegemonythroughthetimeoftheTrojanWartojustbeforetheoutbreakofthewarbetweenthetwogreatpowersofancientGreece,AthensandSparta.Theperiodhesoeconomicallydescribesinafewparagraphsextends,inNewtoniantime,forabouttwothousandyears.ButThucydidesthendevotesseveralhundredpagestotherelativelybrieftwenty-yearperiodofthewar,atleastthatpartofthewarherecorded.Sentencedurationisanindicationoftheimportancethehistorianplacesontheeventsmentioned.Sentenceexpansionorcontraction,however,mayhaveothersustainingcauses.Simplythepacingorrhythmofthehistorian’sprosemightbeone.ThegreatFrenchscientistandhistorianBernarddeFontenellesaidthatifthecadencesofhissentencesdemandedit,theThirtyYearsWarwouldhaveturnedoutdifferently.Somehistorianswilllingeroveranepisode,notbecauseitfillsinasequencevitaltothetale,butCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology433becausethecharactersinvolvedareintrinsicallyinteresting.Maybesomehumorouseventisinsertedinthestorysimplytokeepthereaderturningpages.Inhistories,centuriesmaybecontractedintothespaceofasentence,whilemomentsmaybeexpandedthroughdozensofparagraphs.Afourthtemporaldimensionofnarrativeisthetimeofnarrativeconstruction.Thisisatemporalfeatureespeciallyrelevanttoconsiderationsofthemoralstructureofhistories.Anarrativewillbetemporallylayeredbyreasonofitsconstruction,displaying,asitwere,bothtemporaldepthandatemporalhorizon.Thetemporalhorizonismorepertinentformyconcerns,soletmespeakofthat.Thucydideswrotethefirstpartofhishistorytowardtheendofthewarthathedescribed,whentheawfullatereventsallowedhimtopickoutthoseearlier,antecedenteventsofexplanatoryrelevance–earliereventsthatwouldbeepistemologicallytingedwithAthe-nianfollyyettocome.Onlythebenefitsofhindsight,forexample,couldhaveallowedhimtoputintothemouthoftheSpartanmessengerMelesipus,whowassentonalastdesperatepeacemissionjustbeforethefirstengagementofthewar,thepropheticregret‘‘ThisdaywillbethebeginningofgreatmisfortunestoHellas.’’Bythehorizontalorderingoftime,thehistoriancandescribeeventsinwaysthattheactorsparticipatingintheeventscouldnot:Melesipus’sprophecywaspossibleonlybecauseThu-cydideshadalreadylivedthroughit.Thistemporalperspectiveiscrucialforthehistorian.Onlyfromthevantageofthefuturecanthehistorianpickoutfromaninfinityofantecedenteventsjustthosedeemednecessaryfortheexplanationoftheconsequenteventsofinterest.Differentcausalstructuresofnarrativescorrespondtotheirtemporalmodalities.Iwillnotdetailalloftheiraspects,butletmequicklyrehearsetheirdominantmodes.Themostfundamentalcausalfeatureofnarrativesisthecausalityofevents.Thisissimplythecausalityascribedtoeventsaboutwhichthehistorianwrites.Typicallythehistorianwillarrangeeventssoastoindicatetheircausalsequence,asequenceinwhichthemainantecedentcausesareindicatedsoastoexplainsubsequentevents,ultimatelythecentraleventswhichthehistorywasdesignedtoexplain.Eventsinanarrative,however,displayadifferentcausalgrammarfromeventsinnature.WemaythusspeakofthecausalityofCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n434robertj.richardsnarratedevents.Whenin433B.C.,theAtheniansofThucydides’historyinterferedinaninternalaffairofCorinth,aSpartanally,theycouldnothavepredictedthatwarwouldresult,thoughtheymighthavesuspected;theycertainlycouldnothavepredictedtheirigno-miniousdefeatintheSiciliancampaigntwentyyearslater.FrominsidethescenethatThucydideshasset,thefutureappearsopen;allthingsarepossible,oratleastunforeseeable.YeteachofThucydides’scenesmovesinevitablyandinexorablytothatclimax,namely,tothedestructionofthefleetatSyracuse,thecentraleventofhishistory.Thehistorian,byreasonofhisorhertemporalhorizon,arrangesantecedenteventstomaketheiroutcome,thecen-traleventofinterest,somethingthereader,canexpect–something,intheidealcase,thatwouldberegardedasinevitablegiventheantecedentevents–allthewhilekeepinghisactorsinthedarkuntilthelastminute.Thisisaviewaboutthegrammarofnarrationthatsomehistor-ianswouldnotshare.Sometryassiduouslytoavoidsurfacetermsredolentofcausalityintheirnarratives.ButIthinkthisistobeunawareofthedeepergrammarofnarrative.Theantecedenteventsarechosenbythehistoriantomake,asfarasheorsheisable,theconsequenteventsacausalinevitability.Thatiswhatitistoexplaineventshistorically.Tothedegreethiskindofcausalstructureismissing,tothatdegreethehistorywillfailtoexplainhowitisthatthesubsequenteventsofinterestsoccurredortooktheshapetheydid.Withoutatightcausalgrammarthenarrativewillloosentomerechronicle.Thisgrammaticalfeatureofnarrativehasbearingonanymoralcharacterizationoftheactionsoftheindividualsaboutwhomthehistorianwrites.Andthisintwoways.First,wedothinkthatwhenwemorallyevaluateactions,weassumetheindividualscouldhavechosenotherwise.Therewillthusbeatensionbetweentheactorsrepresentedasregardingthefutureasopen,asfullofpossibilities,andthehistorian’sknowledgethatthefutureoftheactorsisclosed.Theydidwhattheydidbecauseofthenarratedevents,eventscarryingthoseindividualstotheirappointeddestiny.Thesecondwaythecausalityofnarratedeventsbearsonmoralassessmenthastodowiththeconstructionofthesequenceofeventsandtheircausalconnections.Thehistorianwillalsobemakingamoralevaluationoftheactionsofacharacter–implicitlyatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology435least–andwillarrangethatsequenceinwhichthecharacter’sactionsareplacedsoaseithermorallytoindicttheindividual,ormorallytoexculpatetheindividual,or,whatismorefrequentlythecase,tolocatetheindividual’sactionsinamorallyneutralground.Iwillsaymoreaboutthisfeatureofthegrammarofnarrativeinamoment.Athirdcausalmodalitydeployedbyhistoriansmaybecalledthecausalityofnarration.Thisaspectofcausalityhasseveralfeatures,butIwillmentiononlyone:thisisthelocationinanarrativeofvariousscenes.So,forexample,Thucydideswillplaceonescenebeforeanothertoindicatewhathepresumesisanimportantcon-ditionorcauseforasubsequentscene,eventhoughthescenesmaybeatsomerealtemporaldistance.Aspeechmadetomotivateanactionmightbeplacedimmediatelybeforethesceneinwhichtheactionisdescribed,eventhoughthetwoeventsmaybeseparatedbyafairamountofrealtime.Suchjuxtapositioncanhaveaconditioningeffectaswell.ImmediatelyafterThucydidesrelatesPericles’greatfuneraloration,whichextolsthevirtuesofAtheniandemocracyandthegloriesofitslaws,heshovesinadramaticdescriptionoftheAthenianplague,whencitizensignoredthelawsandeachsoughthisownpleasure,thinkingitmightbehislast.Yettheorationandtheplaguewereseparatedbymanymonths.Thiskindofcausalityeffectivelyconditionsthereader’sresponsetotherealitiesofAtheniansociety.Finally,thereisthecausalityofnarrativeconstruction.Therearetwoquitedifferentcausalfeaturesthatwouldfallunderthisrubric.First,onemightdiscriminatethefinalcauseinnarrativeconstruc-tion.Mosthistoriesaimtoexplainsomecentralevent–theoutbreakoftheAmericanCivilWar,Darwin’sdiscoveryandcon-structionofhistheoryofnaturalselection,ortheracialattitudesofHitler.Theantecedenteventsinthehistoryprovidethecausalexplanationofthecentralevent,whichlattermightbethoughtofasthefinalcause,thatis,thegoaloftheconstruction.Historiansintheirresearchusethisfinalcauseasthebeaconinlightofwhichtheyselectoutfromaninfinityofantecedenteventsjustthosethatmightexplainthecentralevent.Nohistorianbegins,asitwere,atthebeginning,ratherattheend.Withoutthefinalcauseasguide–aguidethatmightalter,ofcourse,duringtheresearch–thehistoriancouldnotevenstarttolayoutthoseantecedentcausesthatheorsheCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n436robertj.richardswillfinallyregardastheexplanationfortheconclusionofthehistoricalsequence.Arelatedfeatureofthecausalityofnarrativeconstructionconcernsthemotivesguidingthehistorian,ofwhichtheremaybeseveral.Theproclaimedandstandardmotivationofthegreatnineteenth-centuryhistorianLeopoldvonRankewastodescribeanevent‘‘wieeseigentlichgewesen,’’howitactuallywas;and,insofarashowitwasbecomesinspecificinstancesthecentraleventthatneedsexplanation,thatevent–thefinalcause–becomesthemotiveforconstructingthehistory.Ranke’sgeneralstandardmustbethatofeveryhistorian.Goodhistorianswillwanttoweighpurportedcausesofeventsandemphasizethemostimportant,whilereducingnarrativetimespentonthelessimportant.Yetoftenothermoti-vations,perhapshardlyconsciouseventothehistorian,maygivestructuretohisorherwork.InhissuspiciouslittlebookWhatIsHistory?E.H.Carrurgedthat‘‘whenwetakeupaworkofhistory,ourfirstconcernshouldbenotwiththefactswhichitcontainsbutwiththehistorianwhowroteit’’(Carr1961,24).Ifthereaderknowsinadvancethatthehistorianisofacertaindoctrinalpersuasion,thenajudiciousskepticismmaywellbeinorder.Afterall,ahis-torianmayselecteventsthathaverealbutminorcausalconnectionswithcentraleventsofconcern,whileignoringevenmoreimportantantecedentcauses.Thehistorywouldthenhaveacertainver-isimilitude,yetbeachangeling.Motivationsofauthorsareoftenrevealedbythemoralgrammarofnarratives,anotherstructuralfeaturethatliesatthesyntacticdepthsofhistoricalaccounts.themoralgrammarofnarrativehistoryIamgoingtonowturnspecificallytothefeaturesofthemoralgrammarofhistories,andthenillustratesomeofthewaysthatstructurecharacterizesErnstHaeckel’sstory.Ifnarrativeshavethesegrammaticalstructures,thenitwouldbewellforhistorianstobereflectivelyconsciousofthisandtoformulatetheirreconstruc-tionsinlightofasetofprinciplesthatIbelieveshouldbeoperative.AndinamomentIwillsuggestwhatthoseprinciplesoughttobebywhichwemorallyjudgethebehaviorofindividualswholivedinthepastandbywhichweassesstheirculpabilityforthefutureactionsofothers.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology437Butletmefirstposethequestion,Dohistoriansmakenormativejudgmentsintheirhistoriesandshouldthey?Iwillarguethatnotonlyshouldthey,theymustbyreasonofnarrativegrammar.Atonelevel,itisobviousthathistorians,ofnecessity,domakenormativejudgments.Historicalnarrativesareconstructedonthebasisofevidence:writtendocuments–letters,diaries,publishedworks;alsoartifacts,suchasarchaeologicalfindings;andhigh-techinstru-ments,suchasDNAanalysis;orlow-techoralinterviews.Andhistoriansattributemodesofbehaviortoactorsonthebasisofinferencefromevidenceandinrecognitionofcertainstandards.Evenwhendoingsomethingapparentlyasinnocuousasselectingaverbtocharacterizeapropositionattributedtoanactor,thehistor-ianmustemployanormorstandard.Forexample,ThucydidescouldhavehadMelesipusthinkthatdisasterwasintheoffing,believethatdisasterwasintheoffing,beconvincedthatdisasterwasintheoffing,suspectthatdisasterwasintheoffing,assumethatdisasterwasintheoffing,orprophesythatdisasterwasintheoffing.Whateververbthehistorianselects,heorshewilldosobecausetheactor’sbehavior,assuggestedbytheevidence,hasmetacertainstandardforsuchandsuchmodaldescription–say,beinginastateoffirmconvictionasopposedtovaguesupposition.Alldescriptionsrequiremeasurementagainststandardsornorms–whichisnottosaythatinagiveninstance,thestandardandconsequentdescriptionwouldbethemostappropriate.Thebetterthehistorian,themoreappropriatethenormsemployedinrenderingdescriptions.Virtuallyallofthehistorian’schoicesofdescriptivetermsmustbenormativeinthissense.Butmustsomeofthesenormsalsobemoralnorms?Ibelievetheymust.Theargumentisfairlystraight-forward–atleastasstraightforwardasargumentsofthissorteverget.Humanhistoryisaboutresgestae,thingsdonebyhumanbeings,humanactions.Actionsarenotmerephysiologicalbehav-iors,butbehaviorsthatareintendedandmotivated.Inevitablytheseactionsimpingeonotherindividualsimmediateorremote.Butintentionalbehaviors’impingingonothersispreciselythemoralcontext.Thehistorian,therefore,inordertoassignmotivesandintentionstoindividualswhosebehaviorsaffectothersandtodescribethosemotivesandintentionsadequately–thathistorianmustemploynormsgoverningsuchintentionalbehaviors,thatis,behaviorsinthemoralcontext.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n438robertj.richardsCertainlytheassessmentofmotivesandintentionsmayyieldonlymorallyneutraldescriptions.Butevendecidingthatanintendedbehaviorismorallyneutralis,implicitlyatleast,alsotojudgeitagainststandardsofpositiveornegativemoralvalenceandtodecidethatitconformstoneitherorthatitliessomewhereinbetween.Evenamorallyneutralassessmentisamoralassessment.Thereisnoclaimhere,ofcourse,thatsuchevaluationsaregenerallyself-consciouslyperformedbyhistorians.Mostlytheseevaluationsoccurquitereflexively,insteadofreflectively.Andtheyusuallyexistnotexplicitlyonthesurfaceofthenarrative,butintheinterstices.LetmeofferoneexamplemoreconcretelyofwhatIamarguing,andthisfromahistorianwhomnoonewouldaccuseofcheapmoralizing–hismoralizingisanythingbutcheap.Hisdescriptionsrevealarainbowofshadedmoralevaluations,whichrangesubtlybetweenthepolarcategoriesofshiningvirtueanddarklingvice.ByroncalledhimtheLordofIrony,anditisoftenthroughthattropethathemakeshismoralassessments.Iamspeaking,ofcourse,ofEdwardGibbon.LetmequotejustashortpassagefromtheDeclineandFall,whereGibbonisdescribingwhatmighthavebeenthemotivesofJulian,ashissoldierswereclamoringforhiselevationtoemperor,evenwhileConstantiuswasstillonthethrone.Julianprotestedhecouldnottakethediadem,evenashereluctantlyandsadlyacceptedit.Gibbonwrites:ThegriefofJuliancouldproceedonlyfromhisinnocence;buthisinnocencemustappearextremelydoubtfulintheeyesofthosewhohavelearnedtosuspectthemotivesandtheprofessionsofprinces.Hislivelyandactivemindwassusceptibleofthevariousimpressionsofhopeandfear,ofgrati-tudeandrevenge,ofdutyandofambition,oftheloveoffameandofthefearofreproach.Butitisimpossibleforustoascertaintheprinciplesofactionwhichmightescapetheobservation,whiletheyguided,orratherimpelled,thestepsofJulianhimself...Hesolemnlydeclares,inthepresenceofJupiter,oftheSun,ofMars,ofMinerva,andofalltheotherdeities,thattillthecloseoftheeveningwhichprecededhiselevationhewasutterlyignorantofthedesignsofthesoldiers;anditmayseemungeneroustodis-trustthehonourofahero,andthetruthofaphilosopher.Yetthesuper-stitiousconfidencethatConstantiuswastheenemy,andthathehimselfwasthefavourite,ofthegods,mightprompthimtodesire,tosolicit,andCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology439eventohastentheauspiciousmomentofhisreign,whichwaspredestinedtorestoretheancientreligionofmankind.(Gibbon1777–88,2:319–20)Inthecascadeofrhetoricaldevicesatplay–zeugma,antithesis,irony–GibbonexplicitlyrefusestoattributemorallydemeaningmotivestoJulian,and,ofcourse,atthesametimeimplicitlydoespreciselythat.ThereisanotherelementofjudgmentthatGibbonevinceshere,whichisalsoanimportantfeatureofthemoralgrammarofhistoricalnarrative.Narrativesexplainactionbyallowingustounderstandcharacter,inthiscaseJulian’scharacter.Gibbon,however,hasledustocomprehendJulian’sactionnotonlybycognitivelysuggestingwhatthemotivesofaprincemightbebutalsobyshapingouremotionalresponsetoJulian’scharacterandthusproducinginusafeelingaboutJulian’saction.Wemorallyevaluateindividuals,partlyatleast,throughfeelingsaboutthem.Thehistoriancanorchestrateoutrage–assomedealingwithHaeckelhave–bycuttingquotationsfromanactorintocertainviciousshapes,selectingthosethatappeardamningwhileneglectingthosethatmightbeexculpating.Or,asGibbondoes,thehistoriancanevokefeelingsofmoraldisdainwithlittlemorethanthemagicalmistofantitheticpossibilities.Asaresult,readerswillhave,asitwere,asensible,anolfactoryunder-standing:theinvisibleairofthenarrativewillcarrythesweetsmellofvirtue,theacridstenchofturpitude,orsimplythebittersweetofirony.Thesefeelingswillbecomepartofthedelicatemoralassess-mentrenderedbytheartistryofthehistorian.Thisisjustonesmallexampleofthewaymoraljudgmentexistsintheinterstitialspacesofanarrative,insteadoflyingrightonthesurface.Butsometimessuchjudgmentsdolieclosertotheskinofthehistory.Letmenowfocuspreciselyonacaseofthisandconsidertheprinciplesthat,Ibelieve,shouldbeoperativeinmakingmoraljudg-mentsofhistoricalfigures.ThisisintheinstanceofErnstHaeckel.ernsthaeckel,darwin’schampioningermanyHaeckelwasDarwin’sgreatchampionofevolutionarytheoryinGermany;hewasaprincipalinthetheory’sintroductionthereandaforcefuldefenderofitfromthemid-1860suntil1919,whenhediedCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n440robertj.richards(Richards2004,2005).Haeckel’sworkonevolutionreachedfarbeyondthebordersoftheGermanlands.Hispopularaccountsofevolutionarytheoryweretranslatedintoalltheknownandunknownlanguages–atleastunknowntotheWest–includingArmenian,Chinese,Hebrew,Sanskrit,andEsperanto.MorepeoplelearnedofevolutionarytheorythroughHaeckel’svoluminouswritingsduringthisperiodthanfromanyothersource,includingDarwin’sownwork.Haeckelachievedmanypopularsuccesses,and,aswell,producedmorethantwentylarge,technicalmonographsonvariousaspectsofsystematicbiologyandevolutionarytheory.Intheseworkshedescribedmanyhithertounknownspecies,establishedthescienceofecology,gavecurrencytotheideaofthemissinglink–whichoneofhisprote´ge´s(EugeneDubois)actuallyfound–andpromulgatedthebiogeneticlawthatontogenyrecapitulatesphylogeny.MostofthepromisingyoungbiologistsofthenextgenerationstudiedwithhimatJena.Hisartisticabilitywasconsiderableand,atthebeginningofthetwentiethcentury,heinfluencedthemovementinartcalledJugendstilbyhisbookKunstformenderNatur(Artformsofnature).Haeckelbecameagreatlycelebratedintellectualfigure,oftenmen-tionedforaNobelPrize.Hewasalsothescourgeofreligonists,smitingthepreachersateveryturnwiththejawboneofevolutionarydoctrine.Headvocatedwhathecalleda‘‘monisticreligion’’asasubstituteforthetraditionalorthodoxies,areligionbasedonscience.Asayoungstudent,tryingtofindasubjectforhishabilitation,HaeckelroamedalongthecoastsofItalyandSicilyinsomedespair.HethoughtofgivingupbiologyforthelifeofaBohemian,spendinghistimeinpaintingandpoetizingwithotherGermanexpatriatesontheislandofIschia.Buthefeltthathehadtoaccomplishsomethinginbiology,sothathecouldbecomeaprofessorandmarrythewomanhehadfallendeeplyinlovewith–hisloveletterssentbacktohisfiance´einBerlinaresomethingdelicioustoread.Hefinallyhituponatopic:asystematicdescriptionofalittleknowncreaturethatpopulatedtheseas,theone-celledprotistcalledaradiolarian.Itwaswhilewritinghishabilitationonthesecreaturesin1861thathehappenedtoreadDarwin’sOriginofSpeciesandbecameaconvert.Haeckelproducedamagnificenttwo-volumetomeontheradiolaria,whichhehimselfillustratedwithextraordinaryartisticandscien-tificacumen(Haeckel1862).Theradiolarianmonograph’smostCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology441immediateandsignificanteffectwastosecureHaeckelaprofessor-shipatJena,thusallowinghimtomarryhisbelovedcousin,AnnaSethe.Onhisthirtiethbirthdayin1863,Haeckellearnedhehadwonaprestigiousprizeforhisradiolarianwork.Andonthatsameday,adaythatshouldhavebeenofgreatcelebration,hiswifeofeighteenmonthstragicallydied.Haeckelwascrushed.Hisfamilyfearedhemightcommitsuicide.Asherelatedtohisparents,thisheart-searingblowledhimtorejectallreligionandreplaceitwithsomethingmoresubstantial,somethingthatpromisedakindofprogressivetrans-cendence,namely,Darwiniantheory.Intheyearsfollowingthisupheaval,Haeckelbecameazealousmissionaryforhisnewfaith,andhisownvolatileandcombativepersonalitymadehimacrusaderwhosedemeanorwasinstrikingcontrasttothatofthemodestandretiringEnglishmasterwhomhewouldserve.Thisoutsizedpersonalityhascontinuedtoirritatehistoriansofsmallerimagination.themoralindictmentofhaeckelIn1868,Haeckelproducedapopularworkonthenewtheoryofevolution,entitledNatu¨rlicheScho¨pfungsgeschichte(Naturalhis-toryofcreation).Itwouldgothroughtwelveeditionsuptothetimeofhisdeathin1919andprovetobethemostsuccessfulworkofpopularscienceinthenineteenthandearlytwentiethcenturies.TherearetwofeaturesofthatworkthatincitedsomeofthefiercestintellectualbattlesofthelastpartofthenineteenthcenturyandhaveledsomehistoriansandotherstocomparablyfiercejudgmentsofHaeckel’smoralprobity.Thefirsthastodowithwhatbecamethecardinalprincipleofhisevolutionarydemonstrations,namely,thebiogeneticlawthatontogenyrecapitulatesphylogeny.Thisprincipleholdsthattheembryoofadevelopingorganismgoesthroughthesamemorpholo-gicalstagesthatthephylumwentthroughinitsevolutionaryhistory:so,forexample,thehumanembryobeginsasaone-celledcreature,justaswepresumelifebeganonthisEarthinaone-celledform;itthengoesthroughastageofgastrulation,andHaeckelbelievedthatinthefardistantpast,ourprimitiveancestorspliedtheseasinthatCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n442robertj.richardsFigure23.1Embryosofdog,chicken,andturtleatthe‘‘sandal’’stage(fromErnstHaeckel‘sNatu¨rlicheScho¨pfungsgeschichte,1868)cuplikeform;thentheembryotakesonthemorphologyofanarchaicfish,withgillarches;thenofaprimate;thenaspecifichumanbeing.Thecorollarytothelawisthatcloselyrelatedcreatures–vertebrates,forexample–willgothroughearlyembryologicalstagesthatarequitesimilartooneanother.SomeofHaeckel’senemieschargedthathehadexaggeratedthetailofthehumanembryotomakeitmoreanimal-like–acontroversythatbecameknownasDieSchwanzfrage.Butthedeeper,moredamagingfightarosewithHaeckel’sillustrationofquiteearlyembryosatthesandalstage,astagewhentheylooklikethesoleofasandal.Intheaccompanyingtexttohisillustration(Figure23.1),Haeckelremarks:‘‘IfyouCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology443comparetheyoungembryosofthedog,chicken,andturtle...,youwon’tbeinapositiontoperceiveadifference’’(Haeckel1868,249).OneoftheveryfirstreviewersofHaeckel’sbook,anembryologistwhobecameaswornenemy,pointedoutthatonecertainlywouldnotbeabletodistinguishtheseembryos,sinceHaeckelhadusedthesamewoodcutthreetimes.Hehad,inthewordsofLudwigRu¨ti-meyer,thereviewer,committedagravesinagainstscienceandthepublic’strustinscience(Ru¨timeyer1868).Inthesecondeditionofhisbook,Haeckelretainedonlyoneillustrationofanembryoatthesandalstageandremarkedinthetext:Itmightaswellbetheembryoofadog,chicken,orturtle,sinceyoucannottellthedifference.Thedamage,however,hadbeeninflicted,andtheindictmentoffraudhauntedHaeckelfortherestofhislife.Thechargehasbeenusedbycreationistsinourowndayaspartofabrief,notonlyagainstHaeckel,butagainstevolutionarytheorygen-erally.Yetnotonlycreationists,butseveralhistorianshaveemployeditintheirownmoralevaluationofHaeckelandhisscience.ThesecondfeatureofHaeckel’sworkonwhichIwouldliketofocusreallydidnotcreateastirinhisowntimebuthasbecomeacentralmoralissueinours.Thishastodowiththeassumptionofprogressinevolution,anassumptionthatHaeckelcertainlymade.ThatassumptionisforcefullydisplayedinthetreediagramappendedtohisNatu¨rlicheScho¨pfungsgeschichte.Thediagram(Figure23.2)displaysthevariousspeciesofhumankind,withheightonthever-ticalaxismeanttorepresentmoreadvancedtypes.HeretheCau-casiangroupleadsthepack(seenintheupperrightbranchofthetree),archingabovethedescendingordersofthe‘‘lowerspecies’’–allrootedintheUrmenschorAffenmench,theape-man.Asalientfeatureofthediagramshouldcatchourattention:amongthevari-etiesoftheCaucasianspecies,theBerbersandJewswerethoughtbyHaeckeltobeasadvancedastheGermansandSouthernEuropeans.ThisclassificationshouldhavehadbearingonHaeckel’sassignment1bysomehistorianstotheranksoftheproto-Nazis.naziracehygienistsandtheiruseofhaeckelianideasThatseveralNaziracehygienistsappealedtoHaeckeltojustifytheirviewsisclear.OnepertinentexampleisHeinzBru¨cher’sCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n444robertj.richardsFigure23.2Thestemtreeofthehumanspecies(fromErnstHaeckel‘sNatu¨rlicheScho¨pfungeschichte,1868)CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology445ErnstHaeckelsBluts-undGeistes-Erbe(ErnstHaeckel’sracialandspirituallegacy),publishedin1936.NotonlydidtheauthorlooktoHaeckel’sviewsofracialhierarchyassupportforpoliciesofNationalSocialism,hefirstgavefullaccountofHaeckel’sownimpeccablepedigree.Includedwiththebookwasafive-footchartlayingoutHaeckel’sfamilytree.TheaimofBru¨cher’sracialhygienicanalysiswasbothtodemonstrateanewmethodofshowingtheworthofanintellectualpositionandtousethatmethodtojustifyHaeckel’sowndoctrine.Thatis,onlythebestbloodflowedthroughHaeckel’sveins,andthereforewemaytrusthisideas.TomakethefavorableconnectionbetweenHaeckelandHitler,Bru¨cherfocusedonapassagefromHaeckel’sNatu¨rlicheScho¨pfungsgeschichte,whichreads:‘‘ThedifferenceinrationalitybetweenaGoethe,aKant,aLamarck,aDarwinandthatofthelowernaturalmen–aVeda,aKaffer,anAustralianandaPapuanismuchgreaterthanthegraduateddifferencebetweentherationalityoftheselatterandthatoftheintelligentvertebrates,forinstance,thehigherapes.’’Bru¨cherthencitesaquitesimilarremarkbyHitlerinhisNurembergspeechof1933(Bru¨cher1936,90–91,Hossfeld2005,312–16).InthiswayhehasmadeHaeckelhistoricallyresponsible,atleastinpart,forHitler’sracialattitudes.thejudgmentof‘historicalresponsibility’Bru¨cher’sattributionofmoralresponsibilitytoHaeckelisofatypecommonlyfoundinhistory,thoughthestructureofthesekindsofjudgmentsusuallyisunnoticed,lyingasitdoesinthedeepgrammarofhistoriography.Forexample,historianswilloftencredit,say,Copernicus,inthefifteenthcentury,withthecouragetohavebrokenthroughtherigidityofPtolemaicassumptionandthus,byunshack-lingmen’sminds,tohaveinitiatedthescientificrevolutionofthesixteenthandseventeenthcenturies.This,too,isamoralappraisalofhistoricalresponsibility,though,needlesstosay,Copernicushimselfneveruttered:‘‘Inowintendedtofreemen’smindsandinitiatethescientificrevolution.’’Yet,historiansdoassignhimcreditforthat,moralcreditforgivingsuccessorstheabilitytothinkdifferentlyandproductively.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n446robertj.richardsTheepistemologicalandhistoricaljustificationforthistypeofjudgmentissimplythatthemeaningandvalueofanideaorsetofideascanberealizedonlyinactionsthatthemselvesmaytakesomelongtimetodevelop–thissignalstheineluctableteleologicalfea-tureofhistory.Whilethistypeofjudgmentderivesfromthemoralgrammarofhistory,thisdoesnotmean,ofcourse,thateveryparticularjudgmentofthissortisjustified.thereactionofcontemporaryhistoriansHowhasHaeckelgonedownwithcontemporaryhistorians?Notwell.Hisideas,mixedwithnotionsabouthisaggressiveandcom-bativepersonality,havelodgedinthearteriesfeedingthecriticalfacultiesofmanyhistorians,causingsputteringconvulsions.DanielGasmanhasarguedthatHaeckel’s‘‘socialDarwinismbecameoneofthemostimportantformativecausesfortheriseoftheNazimovement’’(Gasman1971,xxii,and1998).StephenJayGouldandmanyothersconcurthatHaeckel’sbiologicaltheories,supported,asGouldcontends,byan‘‘irrationalmysticism’’andapenchantforcastingallintoinevitablelaws,‘‘contributedtotheriseofNazism’’(Gould1977,77–81).Andmorerecently,inFromDarwintoHitler,RichardWeikarttracesthemetastaticlinehistitledescribes,withthemidcenterofthatlineencirclingErnstHaeckel.Weikartoffershisbookasadisinterestedhistoricalanalysis.Inthatobjectivefashionthatbespeaksthescientifichistorian,hedeclares,‘‘IwillleaveitthereadertodecidehowstraightortwistedthepathisfromDarwinismtoHitlerafterreadingmyaccount’’(Weikart2004,x).Well,afterreadinghisaccount,therecanbelittledoubtnotonlyofthedirectcausalpathfromCharlesDarwinthroughErnstHaeckeltoAdolfHitlerbutofDarwin’sandHaeckel’scomplicityintheatrocitiescommittedbyHitlerandhisparty.Theseevolutionistsbearhistoricalresponsibility.TakingE.H.Carr’sadvicetoheart,wemightinitiallybesus-piciousofWeikart’sdeclarationofobjectivity,proceedingasitdoesfromamemberofanorganizationhavingstrongfundamen-talistmotivation–theDiscoveryInstitute.Nonetheless,otherhistorianshavemadesimilarsuggestions–GouldandGasman,forinstance.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology447Itisyetdisingenuous,Ibelieve,forWeikarttopretendthatmostreadersmightdrawtheirownconclusionsdespitethemoralgram-marofhishistory.Weikart,aswellasGasman,Gould,andmanyotherhistorians,havecreatedahistoricalnarrativeimplicitlyfollowing–theycouldnotdootherwise–theprinciplesofnarrativegrammar:theyhaveconceptualizedanendpoint–Hitler’sbehavior(thefinalcause)regardedhereasethicallyhorrendous–andhavetracedbackcausallinestoantecedentsourcesthatmighthavegivenrisetothoseattitudesofHitler,taintingthosesourcesalongtheway.Itislikeaspreadingoilslickcarriedonanindifferentcurrentandpollutingeverythingittouches.Nowonecancavil,asIcertainlywould,aboutmanydeficienciesintheperformanceofthesehistorians.Theyhavenot,forinstance,properlyweighedthesignificanceofthemanyothercausallinesthatledtoHitler’sbehavior–thesocial,political,cultural,andpsycho-logicalstrandsthatmanyotherhistorianshaveinfactemphasized.Andthusthattheyhaveproducedamonocausalanalysisthatquitedistortsthehistoricalpicture.WhileresponsibilityassignedDarwinandHaeckelmightbemitigatedbyamorerealisticweighingofcausaltrajectories,someculpabilitymight,nonetheless,remain.Yetisthereanyconsidera-tionthatmightmakeussever,notthecausalchain–thatis,thechainlinkingDarwin’swriting,toHaeckel’s,toBru¨cher’s,tomemosofhigh-rankingNazis,andfinallytoHitler’sspeeches–butthechainofmoralresponsibility?Afterall,Haeckel,andofcourse,Darwin,hadbeendeaddecadesbeforetheriseoftheNazis.AndasMontyPythonmighthaveputit,theyarestilldead.LetmesummarizeatthisjuncturethedifferentmodalstructuresofmoraljudgmentinhistoricalnarrativesthatIhavetriedtoiden-tify.Firstistheexplicitappraisalofthehistorian,renderedwhenthehistorianovertlyappliesthelanguageofmoralassessmenttosomedecisionoractiontakenbyahistoricalfigure.Thisbothisrareandrunsagainstthegrainofthecoolersensibilitiesofmosthistorians,LordActonexcepted.Secondistheappraisalofcontemporaries(orlaterindividuals).Partofthehistorian’staskwilloftenbetodescribethejudgmentsmadeonanactorbyhisorherownassociatesorsubsequentindi-viduals.InthecaseofHaeckel,therewerethosewhocondemnedhimofmalfeasance,aswellascolleagueswhodefendedhimagainstCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n448robertj.richardsthecharge.Thismodeofmoralattributionmaybeforthehistorianevidentiary,buthardlydecisive.Thirdistheappraisalbycausalconnection.Thisoccurswhenthehistorianjoinsthedecisionsofanactorwithconsequentialbehaviorofmoralimport.Thebehavioritselfmightbethatoftheactororbehaviordisplacedatsometemporalremovefromtheactor’sovertintentions–Hitler’sactions,forexample,assupposedlypromotedbyHaeckel’sconceptions.ThislatteriswhatIhavecalled‘‘histor-icalresponsibility.’’Thecausaltrajectorymovesfrompasttofuture,butthemoralresponsibilityflowsalongthecausaltracksfromfuturebacktopast.Anditistheguidinghandofthehistorian–fueledbyacomplexofmotives–thatpushesthishistoricalresponsibilitybackalongthecausalrailstothepast.Anditisherethataminorcausalrelationshipcanbemistakenforamajormoralrelationship.Iwill,injustamoment,indicatehowIbelievethehistorianoughtreflectivelytomodulatetheflowofresponsibility.Finally,thereisappraisalbyaestheticcharge.Thisoccurswhenthehistorianthroughartfuldesignevokesafeelingofpositiveornegativeregardfortheactor.InthetreatmentofHaeckelbyGould,Gasman,andWeikart,theodorofsuspicionhasbeendiffusedoverhischaracter,sothathisscentlingersoverHitler’sactions.principlesofmoraljudgmentThistakesmetothefinalpartofmyargument,namely,theprin-ciplesthatoughttogovernourmoraljudgmentsabouthistoricalfigures,especiallyforactionsthatwereatsometemporaldistancefromtheirownhistoricalpositions.Ibelievethatthesamegeneralprinciplesoughttoserveasstandardsforourmoralassessmentsofhistoricalfiguresasservefortheassessmentsofourcontemporaries,includingourselves.Butmuchwilldependonhowthoseprinciplesarespecifiedwhenjudginghistoricallyremoteindividuals.First,thereisthesupremeprincipleofevaluation:itmightbethegoldenrule,thegreatesthappiness,altruism,orthecategoricalimperative.LikelyinthecasesIhaveinmindanyofthesepre-sumptivefirstprincipleswillyieldasimilarassessmentofmoralmotives,sincetheyexpress,Ibelieve,thesamemoralcore.Secondthereistheintentionoftheactor:whatdidheorsheattempttodo?Whatactiondidtheindividualdesiretoexecute,tobedistinguished,CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology449ofcourse,frommereaccidentalbehavior?Thirdisthemotiveforacting–thatis,thegroundfortheintentiontoactinacertainfashion.Themotivewilldeterminemoralvalence.Finally,inassessingmoralbehavior,wemustexaminethebeliefsoftheindividualactorandtrytodeterminewhethertheywerereasonablebeliefs–andthisisthespecialprovenanceofthehistorian.Letmegiveanexample.WhentheHippocraticphysicians,duringthegreatAthenianplaguethatThucydidessodramaticallydescribed,purgedandbledtheafflicted,theirtreatmentactuallyhastenedthedeathsoftheirpatients.Butwecertainlydonotthinkthephysiciansmalignormalfeasant,sincetheyhadareasonablebeliefinthecurativepoweroftheirpractice.Theirintentionwastoapplythebesttherapeutictechniques.Andtheirmotive,wemaypresume,wasaltruistic,sincetheyriskedtheirownlivestocareforthesick.Oneshouldjudgethem,Ibelieve,moralheroes,eventhoughtheconsequenceoftheirbehaviorwasinjury,andeventhedeathoftheirpatients.ThecaseofErnstHaeckelisdecidedlymoreproblematic.Inassessingthemoralprobityofhisreplicationofwoodcuts,thehis-torianwouldhavetoexaminehisintentionsandmotivations.Didheclaimhiswoodcutswereevidenceofhisbiogeneticlaw?Ifso,hemusthavebeenmotivatedtodeceive,andwemaybethusentitledtosuspecthischaracter.Ordidhemerelyintendtoprovideanillustrationofthelawforageneralaudience,maintainingashedidthatatanearlystagevertebrateembryoscannotbemorphologicallydistinguished?(Indeed,thehistoriandoesrecognizethatwiththetechniquesthenavailabletheembryoscouldnotbedistinguished.)Andthusatbest,throughafalseeconomy,Haeckelmayhaveonlycommittedaveryminorinfraction,onethatwouldnotrisetotheleveloffraudandmoralcondemnation.ConcerningHaeckel’sconceptionofaracialhierarchy,thehis-torianhasthetaskofexploringtwoquestionsinparticular:whatdidheintendtoaccomplishbyhistheory?Andhowreasonablewerethebeliefsheharboredaboutraces?Totakethefirstquestion:CouldHaeckel’sactionsbereasonablyinterpretedasintendingtosetinmotionsomethinglikethecrimesoftheNazis?Orminimally,didheexhibitacarelessdisregardforthetruthofhisviewsaboutraces,sothatsomemalfeasantactcould,atleast,havebeenvaguelyanticipated.Itisinansweringthisquestionthatthegrammarofnarrativemustbecarefullyobserved.ThehistorianmaylaydownCambridgeCollectionsOnline©CambridgeUniversityPress,2008\n450robertj.richardsthescenesofhisorherhistorysoastoleadcausallytoacentralevent,suchasHitler’sracialbeliefsandtheirresultsintheHolo-caust,butthehistorianneedstokeeptheactorsinthedark–insofarasitisreasonabletodoso–aboutthosefutureconsequences,inthiscase,tokeepHaeckeloblivioustothefutureuseofhiswork.Thehistorianmayeasilyslip,sinceheorsheknowsthefutureoutcomeoftheactor’sdecisions.Itiseasytoassumetheactoralsokneworcouldhaveanticipatedthoseoutcomes,atleastinsomevagueway.Morelikely,though,thehistorianmightsimplyfailtoreflectonthecrucialdifferencebetweenhisorherfirmknowledgeofthepastandtheactor’sdimknowledgeofthefuture.Inadditiontocarefullyassessingintentionsandmotives,thehis-torianmustalsoconsiderthesetofbeliefsharboredbytheactor.Forexample,wasitreasonableforsomeonelikeDarwinorHaeckeltobelievethatevolutionarytheoryledtoahierarchyofspecieswithinagenusorraceswithinaspecies?Ordidtheyholdtheseideasinrecklessdisregardforthetruth?Toassessreasonablenessofbeliefinthisinstance,thehistorianwouldhavetoknowwhatthescientificcon-sensushappenedtobeinthesecondhalfofthenineteenthcentury.Andinthiscase,amodestlydiligenthistorianwoulddiscoverthatthecommunityofevolutionarytheorists–aswellasotherbiologists–didunderstandthehumanracestostandinahierarchy,justasdidotheranimals,whichalsodisplayedscalabletraits.Inthehumancase,thetraitsincludedthoseofintelligence,moralcharacter,andbeauty.Nineteenth-centuryevolutionarytheoryimpliedthatracesstoodinahierarchy,andalloftheavailableevidencesupportedit.Wemightrecognizefromourperspectivecertainsocialfactorsconstrainingthejudgmentsofthosebiologists,butitissafetosaytheydidnot.Thenthehistoriancanfurtherask,Inthisparticularinstance,whatdoescategorizingpeoplesasbranchesofaracialhierarchymeanforthetreatmentofthosesoclassified?Thisquestiondoesnotallowforauniversalanswer,butwilldependmoreparticularlyontheindividualscientist.Weikart,forinstance,indictsDarwinforaccedingtoabeliefinaracialhierarchy,butneglectstomentionthatDarwindidnotthinkanyactionshouldbetakentoreducethewelfareofthoselowerinthescale.Haeckel’sownattitudesabouthowoneshouldtreatthoselowerinthehierarchyarelessclear,butthereishardlyroomformoralcondemnation,giventheobscurityofhisviewsaboutpracticalaction.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nMoralGrammar:HaeckelandNaziBiology451conclusionItcanonlybeatendentiousanddogmaticallydrivenassessmentthatwouldcondemnDarwinforthecrimesoftheNazis.Iwillconfess,though,thatIhavenotyetmadeupmyownmindaboutthehistoricalresponsibilityofHaeckel,withwhomIhaveconsiderablesympathy.Ahistoriancannotwriteanextendedaccountofthelifeofanindividualwithoutsomemeasureofidentification.Ifoneisgoingtorecoverthepastwithanythinglikeverisimilitude,onemust,asR.G.Collingwoodhasmaintained,relivetheideasofthepast,whichisnotonlytounearthlonginterredintellectualstructuresbutalsotofeelagainthepulseoftheirvitality,tosensetheirurgency,toadmiretheiroriginality,andthustoempathizewiththeirauthors.Andyetonehastodoallofthiswhileretainingareflectiveaware-nessofthemoralstructureinwhichactorsconceivedthoseideasandperceivedtheirimport.note1.ThereisdirectevidenceforHaeckel’sattitudeaboutJewsbeyondhisplacementofthemamongtheadvancedraces.Intheearly1890s,hediscussedthephenomenonofanti-SemitismwiththeAustriannovelistandjournalistHermannBahr(1863–1934),whocollectedalmostfortyinterviewswithEuropeannotablesontheissue,suchindividualsasAugustBebel,TheodorMommsen,JamesArthurBalfour,andHenrikIbsen.InhisdiscussionwithBahr,Haeckeldidacknowledgethatlower-classRussianJewswouldberegardedasanoffensetothehighstandardsofGermanculture–butthenheregardedlower-classItalians–especiallytheNeapolitans–inthesameway.YetofeducatedGermanJews,heremarked:‘‘IholdtheserefinedandnobleJewstobeimportantelementsinGermanculture.Oneshouldnotforgetthattheyhavealwaysstoodbravelyforenlightenmentandfreedomagainsttheforcesofreaction,inexhaustibleopponents,asoftenasneeded,againsttheobscurantists[Dunkelma¨nner].Andnowinthedangersoftheseperiloustimes,whenPapismagainrearsupmightilyeverywhere,wecannotdowithouttheirtriedandtruecourage’’(Bahr1894,69).By‘‘darkmen’’HaeckellikelymeanttheJesuits.ThereissimplynoreasontobelieveHaeckeltoberaciallyanti-Semitic,asGasmanandWeikartdo.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nreferencelistActon,J.1906.Onthestudyofhistory.LecturesonModernHistory,526.London:Macmillan.Adcock,G.J.,E.S.Dennis,S.Easteal,G.A.Huttley,L.S.Jermiin,andW.J.Peacock.2001.MitochondrialDNAsequencesinancientAustralians:Implicationsformodernhumanorigins.ProceedingsoftheNationalAcademyofSciences98:53742.Aderem,A.2005.Systemsbiology:Itspracticesandchallenges.Cell121:51113.Alatalo,R.V.1981.Problemsinthemeasurementofevennesinecology.Oikos37:199204.Alberch,P.,andE.A.Gale.1983.Sizedependenceduringthedevelopmentoftheamphibianfoot:Colchicine-induceddigitallossandreduction.JounalofEmbryologyandExperimentalMorphology76:17797.Alexander,J.M.2000.Evolutionaryexplanationsofdistributivejustice.PhilosophyofScience67:490516.Alroy,J.2000.Understandingthedynamicsoftrendswithinevolvinglineages.Paleobiology26,no.3:31929.Amundson,R.2005.TheChangingRoleoftheEmbryoinEvolutionaryBiology:StructureandSynthesis.NewYork:CambridgeUniversityPress.1996.Historicaldevelopmentoftheconceptofadaptation.Adaptation,1153.NewYork:AcademicPress.Ankeny,R.A.2000.Fashioningdescriptivemodelsinbiology:Ofwormsandwiringdiagrams.PhilosophyofScience67:26072.Anonymous.2002.Berkeleysradical:AninterviewwithPhillipE.Johnson.Touchstone:AJournalofMereChristianity.2005.Inpursuitofsystems.Nature435:1.Antonovics,J.,A.D.Bradshaw,andR.G.Turner.1971.Heavymetaltoleranceinplants.AdvancesinEcologicalResearch7:185.453CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n454ReferenceListAriew,A.2002.PlatonicandAristotelianrootsofteleologicalarguments.Functions:NewReadingsinthePhilosophyofPsychologyandBiology.EditorsA.Ariew,M.Perlman,andR.Cummins.NewYork:OxfordUniversityPress.Aristotle.1979.DeGenerationeAnimaliumasGenerationofAnimals.Cambridge,MA:HarvardUniversityPress.1984.Physics.TheCompleteWorksofAristotle.EditorJ.Barnes.Princeton,NJ:PrincetonUniversityPress.Arnold,A.J.,andK.Fristrup.1982.Thetheoryofevolutionbynaturalselection:Ahierarchicalexpansion.Paleobiology8:11329.Arthur,W.2004.BiasedEmbryosandEvolution.Cambridge:CambridgeUniversityPress.Auffray,C.,S.Imbeaud,M.Roux-Rouquie,andL.Hood.2003.Fromfunctionalgenomicstosystemsbiology:Conceptsandpractices.ComptesRendusBiologies326:87992.Axelrod,Robert.1984.TheEvolutionofCooperation.NewYork:BasicBooks.1997.Theevolutionofstrategiesintheiteratedprisonersdilemma.TheDynamicsofNorms.EditorsC.Bicchieri,R.Jeffrey,andB.Skyrms,199220.Cambridge:CambridgeUniversityPress.Baarsma,E.A.,andH.Collewfin.1975.Changesincompensatoryeyemovementafterunilaterallabyrinthectomyintherabbit.ArchivesofOtorhinolaryngology211:21930.Bahr,H.1894.ErnstHaeckel.DerAntisenitismus,6269.Berlin:S.Fischer.Bailey,J.M.,andS.Agyei,Y.Cladue,andB.A.Gaulin.1994.Effectsofgenderandsexualorientationonevolutionaryrelevantaspectsofhumanmatingpsychology.JournalofPersonalityandSocialPsychology66:108193.Bailey,J.M.,andR.C.Pillard.1995.Geneticsofhumansexualorientation.AnnualReviewofSexResearch5:12650.Bailey,J.M.,R.C.Dawood,K.Miller,andM.B.Pillard.1999.Afamilyhistorystudyofmalesexualorientationusingthreeindependentsamples.BehaviorGenetics29:7986.Bailey,J.M.,andK.J.Zucker.1995.Childhoodsex-typedbehaviorandsexualorientation:Aconceptualanalysisandquantitativereview.DevelopmentalPsychology31:4355.Bamshad,M.,andS.P.Wooding.2003.Signaturesofnaturalselectioninthehumangenome.NationalReviewofGenetics4:99111.Barinaga,M.1995.Remappingthemotorcortex.Science268:169698.Beatty,J.1984.ChanceandNaturalSelection.PhilosophyofScience51:183211.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList4551998.Ecology.RoutledgeEncylopediaofPhilosophy.EditorF.Craig.London:Routledge.AccessedOctober8th,2005,fromwww.rep.routledge.com/article/Q030.1995.Theevolutionarycontingencythesis.Concepts,Theories,andRationalityintheBiologicalSciences:TheSecondPittsburgh-KonstanzColloquiuminthePhilosophyofScience.EditorsJ.LennoxandG.Wolters,4581.Pittsburgh:UniversityofPittsburghPress.1981.Whatswrongwiththereceivedviewofevolutionarytheory?PSA1980.Vol.2,397426.EastLansing,MI:PhilosophyofScienceAssociation.Beatty,J.,andS.Finsen.1989.Rethinkingthepropensityinterpretation:ApeekinsidePandorasBox.WhatthePhilosophyofBiologyIs:EssaysDedicatedtoDavidHull.EditorMichaelRuse,1730.Dordrecht:KluwerAcademic.Bechtel,W.2000.Fromimagingtobelieving:Epistemicissuesingeneratingbiologicaldata.EpistemologyandBiology.EditorsR.CreathandJ.Maienschein,13863.Cambridge:CambridgeUniversityPress.Bechtel,W.,andA.Abrahamsen.2005.Explanation:AMechanistAlternative.StudiesinHistoryandPhilosophyofBiologicalandBiomedicalSciences.SpecialIssue:MechanismsinBiology36:42141.Bechtel,W.,P.Mandik,J.Mundale,andR.Stufflebeam,Editors.2001.PhilosophyandtheNeurosciences:AReader.NewYork:Blackwell.Bechtel,W.,andR.C.Richardson.1993.DiscoveringComplexity:DecompositionandLocalizationasStrategiesinScientificResearch.Princeton,NJ:PrincetonUniversityPress.Beckwith,J.1987.TheOperon:Anhistoricalaccount.EscherichiacoliandSalmonellatyphimurium:CellularandMolecularBiology.EditorF.C.Neidhartetal.Vol.2,143943.Washington,DC:AmericanSocietyforMicrobiology.Behe,M.J.1996.Darwin’sBlackBox.Toronto:FreePress.Benton,M.J.Forthcoming.Thehistoryoflife.TheHarvardCompaniontoEvolution.EditorsM.RuseandJ.Travis.Cambridge,MA:HarvardUniversityPress.Bergsrom,T.C.,andOdedStark.1993.Howaltruismcanprevailinanevolu-tionaryenvironment.AmericanEconomicReview83,no.2:14955.Bernheim,D.1984.Rationalizablestrategicbehavior.Econometrica52,no.4:100728.Berthoz,A.1988.Theroleofgazeincompensationofvestibulardysfunction:Thegazesubstitutionhypothesis.ProgressinBrainResearch7:41120.Binmore,K.1998.JustPlaying.Cambridge,MA:MITPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n456ReferenceListBlair,J.E.,andS.BHedges.2005.MolecularclocksdonotsupporttheCambrianexplosion.MolecularBiologyandEvolution22,no.3:38790.Blanchard,R.,andA.F.Bogart.1996.Homosexualityinmenandnumberofolderbrothers.AmericanJournalofPsychiatry153:2731.Blanchard,R.,andP.Klassen.1997.H-Yantigenandhomosexualityinmen.JournalofTheoreticalBiology185:37378.Bobrow,D.,andJ.M.Bailey.2001.Ismalehomosexualitymaintainedviakinselection?EvolutionandHumanBehavior22:36168.Bogen,J.2005.RegularitiesandCausality:GeneralizationsandCausalExplanations.StudiesinHistoryandPhilosophyofBiologicalandBiomedicalSciences.SpecialIssue:MechanismsinBiology36:35167.Bonner,J.T.1998.Originsofmulticellularity.IntegrativeBiology1,no.1:2736.Boorman,S.A.1978.Mathematicaltheoryofgroupselection:Structureofgroupselectioninfounderpopulationsdeterminedfromconvexityoftheextinctionoperator.ProceedingsoftheNationalAcademyofSciences69:190913.BorgerhoffMulder,M.1991.Humanbehaviouralecology.BehoviouralEcology:AnEvolutionaryApproach.EditorsJ.R.KrebsandN.B.Davies,6998.Oxford:Blackwell.Bottger,D.,etal.2000.TheCambriansubstraterevolution.GSAToday10,no.9:17.Bowler,P.1971.Preformationandpre-existenceintheseventeenthcentury:Abriefanalysis.JournaloftheHistoryofBiology4:22144.Boyd,R.,1999.Homeostasis,species,andhighertaxa.Species:NewInterdisciplinaryEssays.EditorR.A.Wilson,14185.Cambridge,MA:MITPress.1991.Realism,anti-foundationalism,andtheenthusiasmfornaturalkinds.PhilosophicalStudies61:12748.Boyd,R.,andP.J.Richerson.1985.CultureandtheEvolutionaryProcess.Chicago:UniversityofChicagoPress.Brady,R.1985.Ontheindependenceofsystematics.Cladistics1:11326.Brandon,R.,1990.AdaptationandEnvironment.Princeton,NJ:PrincetonUniversityPress.1978.AdaptationandEvolutionaryTheory.StudiesinHistoryandPhilosophyofScience9:181206.2005.Thedifferencebetweenselectionanddrift:AreplytoMillstein.BiologyandPhilosophy20:15370.1999.Introduction.BiologyandPhilosophy14:17.1982.Thelevelsofselection.ProceedingsofthePhilosophyofScienceAssociation1:31523.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList4571994.Theoryandexperimentinevolutionarybiology.Synthese99,no.1:5973.Brandon,R.,andS.Carson.1996.Theindeterministiccharacterofevolutionarytheory:Nonohiddenvariablesproofbutnoroomfordeterminismeither.PhilosophyofScience63:31537.Brandon,R.,andF.N.Nijhout.Forthcoming.Theempiricalnon-equivalenceofgenicandgenotypicmodelsofselection:A(decisive)refutationofgenicselectionismandpluralisticgenicselectionism.PhilosophyofScience.Bromham,L.2003.WhatcanDNAtellusabouttheCambrianexplosion?IntegrativeandComparativeBiology43:14856.Bromham,L.,etal.1998.TestingtheCambrianexplosionhypothesisbyusingamoleculardatingtechnique.ProceedingsfortheNationalAcademyofSciences95,no.11:1238689.Brower,L.P.,andS.B.Malcom.1991.Animalmigrations:Endangeredphenomena.AmericanZoologist31:26576.Bru¨cher,H.1936.ErnstHaeckelsBluts-undGeistes-Erbe:EinekulturbiologischeMonographie.Munich:LehmannsVerlag.Brunet,M.,etal.2002.AnewhominidfromtheUpperMioceneofCentralAfrica.Nature418:14551.Budd,G.,andS.Jensen.2000.Acriticalreappraisalofthefossilrecordofthebilaterianphyla.BiologicalReviewsoftheCambridgePhilosophicalSociety75:25395.Buller,D.J.2005.AdaptingMinds:EvolutionaryPsychologyandthePersistentQuestforHumanNature.Cambridge,MA:MITPress.Burian,R.M.2004a.TheEpistemologyofDevelopment,Evolution,andGenetics.NewYork:CambridgeUniversityPress.1993.Howthechoiceofexperimentalorganismmatters:Epistemologicalreflectionsonanaspectofbiologicalpractice.JournaloftheHistoryofBiology26:35167.2004b.Molecularepigenesis,molecularpleiotropy,andmoleculargenedefinitions.HistoryandPhilosophyoftheLifeSciences26,no.1:5980.Buss,D.M.1995.Evolutionarypsychology:Anewparadigmforpsychologicalscience.PsychologicalInquiry6:130.Buss,L.W.1987.TheEvolutionofIndividuality.Princeton,NJ:PrincetonUniversityPress.Butcher,E.C.,E.L.Berg,andE.J.Kunkel.2004.Systemsbiologyindrugdiscovery.NatureBiotechnology10:125359.Cabeza,R.,andL.Nyberg.1997.Imagingcognition:AnempiricalreviewofPETstudieswithnormalsubjects.JournalofCognitiveNeuroscience9:126.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n458ReferenceList2000.ImagingcognitionII:Anempiricalreviewof275PETandfMRIStudies.JournalofCognitiveNeuroscience12:147.Callebaut,W.,andD.Rasskin-Gutman.2005.Modularity:UnderstandingtheDevelopmentandEvolutionofNaturalComplexSystems.Cambridge,MA:MITPress.Camerer,C.,andR.Thaler.1995.Anomalies:Ultimatums,dictators,andmanners.JournalofEconomicPerspectives9,no.2:20919.Canguilhem,G.1979.WissenschaftsgeschichteundEpistemologie.Frankfurt/Main:Suhrkamp.Cantino,P.D.,andK.deQueiroz.2003.PhyloCode:Aphylogeneticcodeofbiologicalnomenclature.Availableatwww.ohiou.edu/phylocode/.Cantor,G.2005.Quakers,Jews,andScience:ReligiousResponsestoModernityandtheSciencesinBritain,1650–1900.Oxford:OxfordUniversityPress.Carr,E.1961.WhatIsHistory?NewYork:VintageBooks.Carroll,S.B.2001.Chanceandnecessity:Theevolutionofmorphologicalcomplexityanddiversity.Nature409:110209.2005.EndlessFormsMostBeautiful:TheNewScienceofEvoDevo.NewYork:W.W.Norton.Carroll,S.B.,etal.2005.FromDNAtoDiversity:MolecularGeneticsandtheEvolutionofAnimalDesign.Malden,MA:Blackwell.Carroll,S.B.,J.K.Grenier,andS.D.Weatherbee.2001.FromDNAtoDiversity:MolecularGeneticsandtheEvolutionofAnimalDesign.Oxford:Blackwell.Cassidy,J.1978.Philosophicalaspectsofthegroupselectioncontroversy.PhilosophyofScience45:57594.Cavalli-Sforza,L.L.,andM.W.Feldman.1981.CulturalTransmissionandEvolution.AQuanitativeApproach.Vol.16.Princeton,NJ:PrincetonUniversityPress.CelaConde,C.J.,andF.J.Ayala.2001.SenderosdelaEvolucionHumana.Madrid:AlianzaEditorial.Chapdelaine,Y.,andL.Bonen.1991.ThewheatmitochondrialgeneforsubunitIoftheNADHdehydrogenasecomplex:Atrans-splicingmodelforthisgene-in-pieces.Cell65,no.3:46572.Charles,D.1995.Teleologicalcausationinthephysics.Aristotle’s‘‘Physics’’:ACollectionofEssays.EditorJ.Lindsay.Oxford:ClarendonPress.Cheney,D.1986.Interactionsandrelationshipsbetweengroups.PrimateSocieties.EditorsB.Smuts,D.Cheney,R.Seyfarth,R.Wrangham,andT.Struhsaker,26781.Chicago:UniversityofChicagoPress.1983.Proximateandultimatefactorsrelatedtothedistributionofmalemigration.PrimateSocialRelationships:AnIntegratedApproach.EditorRobertHinde,24149.Oxford:BlackwellScientific.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList459Cohen,J.,andS.H.Rice.1996.Wheredobiochemicalpathwayslead?IntegrativeApproachestoMolecularBiology.EditorsJ.Collado-Vides,B.Magasanik,andT.F.Smith,23951.Cambridge,MA:MITPress.Collins,F.S.,E.D.Green,A.E.Guttmacher,andM.S.Guyer.2003.Avisionforthefutureofgenomicsresearch.Nature422:83547.Collins,F.,etal.1998.NewgoalsfortheU.S.HumanGenomeProject:19982003.Science282:68289.Colvin,J.1983.Rankinfluencesrhesusmalepeerrelationships.PrimateSocialRelationships:AnIntegratedApproach.EditorR.Hinde,5764.Oxford:BlackwellScientific.Colyvan,M.,andL.R.Ginzburg.2003.Lawsofnatureandlawsofecology.Oikos101:64953.Connell,J.,andE.Orias.1974.Theecologicalregulationofspeciesdiversity.AmericanNaturalist903:399413.Connors,B.W.,andM.J.Gutnick.1990.Intrinsicfiringpatternsofdiverseneocorticalneurons.TrendsinNeuroscience13:99140.Cook-Deegan,R.1994.TheGeneWars.NewYork:W.W.Norton.Cooper,G.1998.Generalizationsinecology:Aphilosophicaltaxonomy.BiologyandPhilosophy13:55586.Cooper,J.1987.Hypotheticalnecessityandnaturalteleology.PhilosophicalIssuesinAristotle’sBiology.EditorsA.GotthelfandJ.Lennox.Cambridge:CambridgeUniversityPress.Cosmides,L.,andJ.Tooby.1997.Themodularnatureofhumanintelligence.TheOriginandEvolutionofIntelligence.EditorsA.B.ScheibelandJ.W.Schopf,71101.Sudbury,MA:Jones&Bartlett.Cracraft,J.1990.Theoriginofevolutionarynovelties:Patternandprocessatdifferenthierarchicallevels.EvolutionaryInnovations.EditorM.Nitecki,2146.Chicago:UniversityofChicagoPress.1983.Speciesconceptsandspeciationanalsis.CurrentOrnithology1:15987.Craver,C.F.Beyondreduction:Mechanisms,mulitfieldintegration,andtheunityofneuroscience.StudiesinHistoryandPhilosophyofBiologicalandBiomedicalSciences.SpecialIssue:MechanismsinBiology.EditorsC.F.CraverandL.Darden.36:37397.2002b.Interlevelexperiments,mulitlevelmechanismsintheneuro-scienceofmemory.PhilosophyofScience(Supplement)69:S83S97.2001.Rolefunctions,mechanisms,andhierarchy.PhilosophyofScience68:5374.2002a.Structuresofscientifictheories.BlackwellGuidetothePhilosophyofScience.EditorsP.K.MachamerandM.Silberstein,5579.Oxford:Blackwell.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n460ReferenceListCrick,F.H.C.1958.Onproteinsynthesis.SymposiumoftheSocietyforExperimentalBiology12:13663.1959.Thepresentpositionofthecodingproblem.StructureandFunctionofGeneticElements:BrookhavenSymposiainBiology12:3539.Cummins,Robert.2002.Neo-teleology.Functions:NewReadingsinthePhilosophyofPsychologyandBiology.EditorsA.Ariew,M.Perlman,andR.Cummins.Oxford:OxfordUniversityPress.Damuth,J.,andI.L.Heisler.1988.Alternativeformulationsofmultilevelselection.BiologyandPhilosophy3:40730.Darden,L.2000.ReviewofPaulThagardsHowScientistsExplainDisease.PhilosophyofScience67:35254.2006.ReasoninginBiologicalDiscoveries:Mechanisms,InterfieldRelations,andAnomalyResolution.NewYork:CambridgeUniversityPress.Darden,L.,andJ.Tabery.2005.Molecularbiology.TheStanfordEncyclopediaofPhilosophy.EditorE.N.Zalta.Availableathttp://plato.stanford.edu/entries/molecular-biology/.Darwin,C.1859.OntheOriginofSpecies.London:JohnMurray.1964.OntheOriginofSpecies:AFacsimileoftheFirstEdition.Cambridge,MA:HarvardUniversityPress.Davidson,E.H.2001.GenomicRegulatorySystems:DevelopmentandEvolution.SanDiego,CA:AcademicPress.2006.TheRegulatoryGenome:GeneRegulatoryNetworksinDevelop-mentandEvolution.Burlington,MA,andSanDiego:AcademicPress.Dawkins,R.1986.TheBlindWatchmaker.NewYork:W.W.Norton.1996.ClimbingMountImprobable.NewYork:W.W.Norton.1989.Theevolutionofevolvability.ArtificialLife.EditorC.G.Langton,20120.RedwoodCity,CA:Addison-Wesley.1982b.TheExtendedPhenotype:TheGeneastheUnitofSelection.Oxford:W.H.Freeman.1982a.Replicatorsandvehicles.CurrentProblemsinSociobiology.Cambridge:CambridgeUniversityPress.1978.Replicatorselectionandtheextendedphenotype.ZeitschriftFurTierpsychologie47:6176.1976.TheSelfishGene.Oxford:OxfordUniversityPress.1989.TheSelfishGene.2nded.Oxford:OxfordUniversityPress.1983.UniversalDarwinism.MoleculestoMen.EditorD.S.Bendall.Cambridge:CambridgeUniversityPress.DeGelder,B.2000.Moretoseeingthanmeetstheeye.Science289:114849.deLaplante,K.2004.Towardamoreexpansiveconceptionofecologicalscience.BiologyandPhilosophy19:26381.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList461deQueiroz,K.2005.ErnstMayrandthemodernconceptofspecies.ProceedingsoftheNationalAcademyofSciences102:660007.1998.Thegenerallineageconceptofspecies,speciescriteria,andtheprocessofspeciation:Aconceptualunificationandterminologicalrecommendations.EndlessForms:SpeciesandSpeciation.EditorsD.J.HowardandS.H.Berlocher,5775.Oxford:OxfordUniversityPress.1992.Phylogeneticdefinitionsandtaxonomicphilosophy.BiologyandPhilosophy7:295313.1988.SystematicsandtheDarwinianrevolution.PhilosophyofScience55:23859.deQueiroz,K.,andJ.A.Gauthier.1990.Phylogenyasacentralprincipleintaxonomy:Phylogeneticdefinitionsoftaxonnames.SystematicBiology39:2731.1994.Towardaphylogeneticsystemofbiologicalnomenclature.TrendsinEcologyandEvolution9:2731.deQueiroz,K.,andS.Poe.2001.Philosophyandphylogeneticinference:AcomparisonoflikelihoodandparsimonymethodsinthecontextofKarlPopperswritingsoncorroboration.SystematicBiology50:30521.Dembski,W.A.1998.TheDesignInference:EliminatingChancethroughSmallProbabilities.Cambridge:CambridgeUniversityPress.1997.Intelligentdesignasatheoryofinformation.PerspectivesonScienceandChristianFaith3:18090.Dennett,D.C.2006.BreakingtheSpell:ReligionasaNaturalPhenomenon.NewYork:Viking.1995.Darwin’sDangerousIdea.NewYork:Simon&Schuster.DeYoe,E.A.,andD.C.VanEssen.1988.Concurrentprocessingstreamsinmonkeyvisualcortex.TrendsinNeuroscience11:21926.Diamond,J.M.1975.Theislanddilemma:Lessonsofmodernbiogeographicstudiesforthedesignofnaturalreserves.BiologicalConservation7:12946.1986.Overview:Laboratoryexperiments,fieldexperiments,andnaturalexperiments.CommunityEcology.EditorsJ.DiamondandT.J.Case,322.NewYork:Harper&Row.Dick,D.M.,R.J.Viken,R.J.Kaprio,J.Koskenvou,andM.Rose.2001.Exploringgene-environmentinteractions:Socioregionalmoderationofalcoholuse.JournalofAbnormalPsychology110:62532.Dickemann,M.1979.Femaleinfanticideandthereproductivestrategiesofstratifiedhumansocieties.EvolutionarySocietiesandHumanSocialBehavior.EditorsN.A.ChagnonandW.Irons,32167.NorthScituate,MA:Duxbury.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n462ReferenceList1995.WilsonsPanchreston:Theinclusivefitnesshypothesisofsociobiologyre-examined.JournalofHomosexuality28:14783.Dietrich,Michael.2000a.Formhopefulmonsterstohomeoticeffects:RichardGoldschmidtsintegrationofdevelopment,evolutionandgenetics.AmericanZoologist40:73847.2000b.Theproblemofthegene.ComptesRendusDeL’AcademiedesSciences-SeriesIII-SciencesdelaVie323,no.12:113946.Dillon,N.2003.Positions,please....Nature425:457.Dobzhansky,Th.1937.GeneticsandtheOriginofSpecies.NewYork:ColumbiaUniversityPress.1962.MankindEvolving.NewHaven,CT:YaleUniversityPress.1956.Whatisanadaptivetrait?AmericanNaturalist40,no.855:33747.Dohrn,G.1875.DerUrsprungderWirbelthiereunddasPrincipdesFunctionswechsels.Leipzig.Donoghue,M.1985.Acritiqueofthebiologicalspeciesconceptandrecommendationsforaphylogeneticalternative.Bryologist88:17281.Dretske,F.1981.KnowledgeandtheFlowofInformation.Cambridge,MA:MITPress.Dugatkin,L.A.,andM.Alfieri.1991a.Guppiesandthetit-for-tatstrategy:Preferencebasedonpastinteraction.BehavioralEcologyandSociobiology28:24346.1991b.Tit-for-tatinguppies(poeciliareticulata):Therelativenatureofcooperationanddefectionduringpredatorinspection.EvolutionaryEcology5:30009.Dupre,J.1993.TheDisorderofThings:MetaphysicalFoundationsoftheDisunityofScience.Cambridge,MA:HarvardUniversityPress.Dupuis,C.1984.WilliHennigsimpactontaxonomicthought.AnnualReviewofEcologyandSystematics15:124.Edwards,A.W.F.1972.Likelihood.Cambridge:CambridgeUniversityPress.Ehrlich,P.R.2000.HumanNatures:Genes,Cultures,andtheHumanProspect.Washington,D.C.,andCovelo,CA:IslandPress/ShearwaterBooks.Eldredge,N.1995.ReinventingDarwin.NewYork:JohnWiley.1985.UnfinishedSynthesis:BiologicalHierarchiesandModernEvolutionaryThought.NewYork:OxfordUniversityPress.Eldredge,N.,andJ.Cracraft.1980.PhylogeneticPatternsandtheEvolutionaryProcess:MethodandTheoryinComparativeBiology.NewYork:ColumbiaUniversityPress.Eldredge,N.,andS.J.Gould.1972.Punctuatedequilibria:Analternativetophyleticgradualism.ModelsinPaleobiology.EditorT.J.M.Schopf,82115.SanFrancisco:W.H.Freeman.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList463Elliott,K.2004.Errorasmeanstodiscovery.PhilosophyofScience71:17497.Ellis,L.,M.A.Ames,andD.Burke.1987.Sexualorientationasacontinuousvariable:Acomparisonbetweensexes.ArchivesofSexualBehavior16:52629.Elton,C.S.1958.TheEcologyofInvasionsbyAnimalsandPlants.London:Methuen.Enc,B.,andF.Adams.1992.Functionsandgoal-directedness.PhilosophyofScience59:63554.Endler,J.1986.NaturalSelectionintheWild.Princeton,NJ:PrincetonUniversityPress.Endler,J.,andT.McClellan.1988.Theprocessofevolution:Towardsanewersynthesis.AnnualReviewofEcologyandSystematics19:395421.Ernst,Z.2001.Explainingthesocialcontract.BritishJournalforthePhilosophyofScience52,no.1:124.Erwin,D.,andE.Davidson.2002.Thelastcommonbilaterianancestor.Development129:302132.Essock-Vitale,S.,andR.Seyfarth.1986.Intelligenceandsocialcognition.PrimateSocieties.EditorsB.Smuts,D.Cheney,R.Seyfarth,R.Wrangham,andT.Struhsaker,45261.Chicago:UniversityofChicagoPress.Fager,E.W.1972.Diversity:Asamplingstudy.AmericanNaturalist106:293310.Faith,D.P.,andJ.W.H.Trueman.2001.Towardsaninclusivephilosophyforphylogeneticinference.SystematicBiology50:33150.Falk,R.1991.Thedominanceoftraitsingeneticanalysis.JournaloftheHistoryofBiology24:45784.2000.Thegene:Aconceptintension.TheConceptoftheGeneinDevelopmentandEvolution.EditorsP.R.FalkandH.J.RheinbergerBuerton,31748.Cambridge:CambridgeUniversityPress.Geneticanalysis.Inpress.InternationalHandbookofthePhilosophyofBiology.EditorsM.MatthenandC.Stephens.NewYorkandAmsterdam:Elsevier.2005.Genetics.ThePhilosophyofScience:AnEncyclopedia.EditorsJ.PheifferandS.Sarkar,33039.NewYork:RoutledgeReference.2001.Theriseandfallofdominance.BiologyandPhilosophy16,no.3:285323.1995.Thestruggleofgeneticsforindependence.JournaloftheHistoryofBiology28:21946.1986.Whatisagene?StudiesintheHistoryandPhilosophyofScience17:133.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n464ReferenceListFarris,J.S.1979.Theinformationcontentofthephylogeneticsystem.SystematicZoology28:483519.1983.Thelogicalbasisofphylogeneticanalysis.AdvancesinCladistics.EditorsN.I.Platnick,andV.A.Funk,736.NewYork:ColumbiaUniversityPress.Farris,J.S.,A.G.Kluge,andJ.M.Carpenter.2001.PopperandlikelihoodversusPopper*.SystematicBiology50:43843.Felsenstein,J.1978.Casesinwhichparsimonyorcompatibilitymethodswillbepositivelymisleading.SystematicZoology27:40110.Finta,C.,andP.G.Zaphiropoulos.2001.Astatisticalviewofgenometranscription.JournalofMolecularEvolution53:16062.Fisch,U.1973.Thevestibularresponsefollowingunilateralvestibularcompensation.ActaOtlaryngology76:22938.Fisher,D.C.1985.Evolutionarymorphology:Beyondtheanalogous,theanecdotal,andtheadhoc.Paleobiology11:12038.Fisher,R.A.1930.TheGeneticalTheoryofNaturalSelection.Oxford:ClarendonPress.1925.StatisticalMethodsforResearchWorkers.Edinburgh:OliverandBoyd.Fisher,R.A.,A.S.Corbet,andC.B.Williams.1943.Therelationbetweenthenumberofspeciesandthenumberofindividualsinarandomsampleofananimalpopulation.JournalofAnimalEcology12:4258.Fisher,R.A.,andE.B.Ford.1947.ThespreadofageneinnaturalconditionsinacolonyofthemothPanaxiadominula.Heredity1:14374.Fitelson,B.,E.Stephens,andC.Sober.1999.HownottodetectdesignareviewofWilliamDembskisTheDesignInference.PhilosophyofScience66:47288.Flohr,H.,J.Bienfold,W.Abeln,andI.Macskovics.1981.Conceptsofvestibularcompensation.Lesion-InducedNeuronalPlasticityinSensorimotorSystems.EditorsH.FlohrandW.Precht,15372.Amsterdam:Springer.Fogle,T.2001.Thedissolutionofproteincodinggeneseinmolecularbiology.TheConceptoftheGeneinDevelopmentandEvolution.EditorsR.Falk,H.J.Rheinberger,andP.Beurton,325.Cambridge:CambridgeUniversityPress.Fox-Keller,E.2000.TheCenturyoftheGene.Cambridge,MA:HarvardUniversityPress.Francis,R.2003.WhyMenWon’tAskforDirections:TheSeductionsofSociobiology.Princeton,NJ:PrincetonUniversityPress.Frey,B.S.,andI.Bohnet.1980.Institutionsaffectfairness:Experimentalinvestigations.Behavior75:262300.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList465Friboulet,A.,andD.Thomas.2005.Systemsbiologyaninterdisciplinaryapproach.Biosensors&Bioelectronics20:240407.Fujimura,J.H.2005.Postgenomicfutures:Translationsacrossthemachine-natureborderinsystemsbiology.NewGeneticsandSociety24:195225.Futuyma,D.1979.EvolutionaryBiology.Sunderland,MA:Sinauer.1986.EvolutionaryBiology.2nded.Sunderland,MA:Sinauer.Garfinkel,A.1981.FormsofExplanation.NewHaven,CT:YaleUniversityPress.Gasman,D.1998.Haeckel’sMonismandtheBirthofFascistIdeology.Frankfurt:PeterLang.1971.TheScientificOriginsofNationalSocialism.NewYork:ScienceHistory.Gaston,K.J.1996.Whatisbiodiversity?Biodiversity:ABiologyofNumbersandDifference.EditorK.J.Gaston,19.London:Blackwell.Ghiselin,M.T.1997.MetaphysicsandtheOriginofSpecies.Albany,NY:SUNYPress.1974.Aradicalsolutiontothespeciesproblem.SystematicZoology23:53644.Gibbon,E.177788.TheHistoryoftheDeclineandFalloftheRomanEmpire.6Vols.London:StrahanandCadell.Giere,R.1988.ExplainingScience:ACognitiveApproach.Chicago:UniversityofChicagoPress.Gilbert,S.F.,andS.Sarkar.2000.Embracingcomplexity:Organicismforthe21stcentury.DevelopmentalDynamics219:19.Gillespie,J.H.1977.Naturalselectionforvariancesinoffspringnumber:Anewevolutionaryprinciple.AmericanNaturalist111:101014.1974.Naturalselectionforwithin-generationvarianceinoffspringnumber.Genetics76:60106.1973.Polymorphisminrandomenvironments.TheoreticalPopulationBiology4:19395.Glennan,S.S.1996.Mechanismsandthenatureofcausation.Erkenntnis44:4971.2005.Modelingmechanisms.StudiesinHistoryandPhilosophyofBiologicalandBiomedicalSciences.SpecialIssue:MechanismsinBiology.EditorsC.F.CraverandL.Darden.36:44364.2002.Rethinkingmechanisticexplanation.PhilosophyofScience(Supplement)69:S342S353.Godfrey-Smith,P.1992.Additivityandtheunitsofselection.PSA:ProceedingsoftheBiennialMeetingofthePhilosophyofScienceAssociation,19921:31528.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n466ReferenceList1999.Genesandcodes:Lessonsfromthephilosophyofmind?WhereBiologyMeetsPsychology:PhilosophicalEssays.EditorV.Hardcastle.Cambridge,MA:MITPress.2001.Informationandtheargumentfromdesign.IntelligentDesignCreationismanditsCritics:Philosophical,TheologicalandScien-tificPerspectives.EditorR.Pennock,57596.Cambridge,MA:MITPress.2000.OntheTheoreticalRoleofGeneticCoding.PhilosophyofScience67:2644.Forthcoming.Thestrategyofmodel-basedscience.BiologyandPhilosophy.2001.Threekindsofadaptationism.AdaptationismandOptimality.EditorsS.H.OrzackandE.Sober,33557.Cambridge:CambridgeUniversityPress.Goldschimdt,R.1940.TheMaterialBasisofEvolution.NewHaven,CT:YaleUniversityPress.Goode,S.1999.Johnsonchallengesadvocatesofevolution.Insight,October25.Goodman,D.1975.Thetheoryofdiversity-stabilityrelationshipsinecology.QuarterlyReviewofBiology50:23766.Goodrich,E.S.1912.TheEvolutionofLivingOrganisms.London:T.C.andE.C.Jack.Gould,S.J.1983.Hen’sTeethandHorse’sToes.NewYork:W.W.Norton.1977.OntogenyandPhylogeny.Cambridge,MA:BelknapPress.1980.ThePanda’sThumb.NewYork:W.W.Norton.2002.TheStructureofEvolutionaryTheory.Cambridge,MA:HarvardUniversityPress.1989.WonderfulLife:TheBurgessShaleandtheNatureofHistory.NewYork:W.W.Norton.Gould,S.J.,andR.C.Lewontin.1979.ThespandrelsofSanMarcoandthePanglossianparadigm:Acritiqueoftheadaptationistprogram.ProceedingsoftheRoyalSocietyofLondon,SeriesB:BiologicalSciences205:58198.Gould,S.J.,andE.A.Lloyd.1999.Individualityandadaptationacrosslevelsofselection:HowshallwenameandgeneralizetheunitofDarwinism?ProceedingsoftheNationalAcademyofSciencesUSA96:1190409.Gould,S.J.,andE.S.Vrba.1982.Exaptationamissingterminthescienceofform.Paleobiology8:415.Goulson,D.,andD.Owen.1997.Long-termstudiesofthemedionigrapolymorphisminthemothPanaxiadomincula:Acritique.Oikos80:61317.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList467Grafen,A.1991.Modellinginbehaviouralecology.BehaviouralEcology:AnEvolutionaryApproach.EditorsJ.R.KrebsandN.B.Davies,531.Oxford:Blackwell.Grant,S.G.N.2003.Systemsbiologyinneuroscience:Bridginggenestocognition.CurrentOpinioninNeurobiology13:57782.Greenberg,A.S.,andJ.M.Bailey.1993.Dobiologicalexplanationsofhomosexualityhavemoral,legal,orpolicyimplications?JournalofSexResearch30:24551.Grene,M.1990.Evolution,typology,andpopulationthinking.AmericanPhilosophicalQuarterly27:23744.2002.ReplytoDavidHull.ThePhilosophyofMarjorieGrene.EditorsR.E.AuxierandL.E.Hahn,27983.LaSalle,IL:OpenCourt.Griesemer,J.R.2000.Development,culture,andtheunitsofinheritence.PhilosophyofScience67:34868.2005.Theinformationalgeneandthesubstantialbody:Onthegeneralizationofevolutionarytheorybyabstraction.IdealizationXII:CorrectingtheModel:IdealizationandAbstractionintheSciences.EditorsM.JonesandN.Cartwright.Amsterdam/NewYork:Rodopi.2004.Three-dimensionalmodelsinphilosophicalperspective.Models:TheThirdDimensionofScience.EditorsSorayadeChadarevianandNickHopwood,43342.Stanford,CA:StanfordUniversityPress.Griesemer,J.R.,andM.Wade.1988.Laboratorymodels,causalexplanationsandgroupselection.BiologyandPhilosophy3:6796.Griffiths,G.C.D.1974.Onthefoundationsofbiologicalsystematics.ActaBiotheoretica13:85131.Griffiths,P.E.2001.Geneticinformation:Ametaphorinsearchofatheory.PhilosophyofScience68:394412.1996.Thehistoricalturninthestudyofadaptation.BritishJournalforthePhilosophyofScience47:51132.1999.Squaringthecircle:Naturalkindswithhistoricalessences.Species:NewInterdisciplinaryEssays.EditorR.A.Wilson,20928.Cambridge:CambridgeUniversityPress.Griffiths,P.E.,andR.D.Gray.1994.Developmentalsystemsandevolutionaryexplanation.JournalofPhilosophy91,no.6:277305.1997.ReplicatorII:Judgmentday.BiologyandPhilosophy12,no.4:47192.Griffiths,P.,andE.Neumann-Held.1999.Themanyfacesofthegene.Biosciences49:65664.Guston,D.H.,andD.Sarewitz.2002.Real-timetechnologyassessment.TechonologyinSociety24:93109.Gu¨th,W.,R.Schmittberger,andB.Schwarz.1982.Anexperimentalanalysisofultimatumbargaining.JournalofEconomicBehaviorandOrganization3:36788.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n468ReferenceListGuyer,M.S.,andF.S.Collins.1993.Thehumangenomeprojectandthefutureofmedicine.AmericanJournalofDiseasesofChildren147:114552.Haeckel,E.1862.DieRadiolarien(Rhizopodaradiaria).EineMonographie.Berlin:G.Reimer.1868.NaturlicheSchopfungsgeschichte.Berlin:G.Reimer.Haile-Selassie,Y.2001.LateMiocenehominidsfromtheMiddleAwash,Ethiopa.Nature412:17881.Haldane,J.B.S.1932.TheCausesofEvolution.London:Longmans.Haldeman,D.C.1994.Thepracticeandethicsofsexualorientation.JournalofConsultingandClinicalPsychology62:22127.Hall,B.K.2000.Evo-devoordevo-evodoesitmatter?EvolutionandDevelopment2,no.4:17778.1998.EvolutionaryDevelopmentalBiology.LondonandNewYork:Chapman&Hall.Hall,B.K.,andW.M.Olson.2003.KeywordsinEvolutionaryDevelopmentalBiology.Cambridge,MA:HarvardUniversityPress.Halperin,D.M.1990.OneHundredYearsofHomosexuality.NewYork:Routledge.Hamblin,M.T.,E.E.Thompson,andA.DiRienzo.2002.Complexsignaturesofnaturalselectionattheduffybloodgrouplocus.AmericanJournalofHumanGenetics70:36983.Hamburger,V.1988.TheHeritageofExperimentalEmbryology:HansSpemannandtheOrganizer.NewYork:OxfordUniversityPress.Hamer,D.H.,S.Hu,V.L.Magnuson,N.Hu,andA.M.L.Pattatucci.1993.AlinkagebetweenDNAmarkersontheX-chromosomeandmalesexualorientation.Science261:32137.Hamilton,W.D.1996NarrowRoadsofGeneLand:TheCollectedPapersofW.D.Hamilton.NewYork:W.H.FreemanSpektrum.Hampe,M.,andMorgan,S.R.1988.TwoconsequencesofRichardDawkinsviewofgenesandorganisms.StudiesinHistoryandPhilosophyofScience19:11938.Hardcastle,V.1995.HowtoBuildaTheoryinCognitiveScience.Albany,NY:SUNYPress.Hariharan,I.K.,andD.A.Haber.2003.Yeast,flies,worms,andfishinthestudyofhumandisease.NewEnglandJournalofMedicine348:245763.Harlin,M.1999.Thelogicalpriorityofthetreeovercharactersandsomeofitsconsequences.BiologicalJournaloftheLinnaeanSociety68:497503.Harre´,Rom.1970.ThePrinciplesofScientificThinking.Chicago:UniversityofChicagoPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList469Hartl,D.,andA.Clark.1989.TheoreticalPopulationGenetics.Sunderland,MA:SinauerHartwell,L.H.,J.J.Hopfield,S.Leibler,andA.W.Murray.1999.Frommoleculartomodularcellbiology.Nature402:4752.Haught,J.F.2000.GodAfterDarwin:ATheologyofEvolution.Boulder,CO:WestviewPress.Heisler,I.L.,andJ.Damuth.1987.Amethodforanalyzingselectioninhierarchicallystructuredpopulations.AmericanNaturalist130:582602.Hennig,W.1950.GrundzugeeinerTheoriederPhylogenetischenSystematik.Berlin:DeutscherZentralverlag.1966.PhylogeneticSystematics.Urbana:UniversityofIllinoisPress.Henry,C.M.2003.Systemsbiology.ChemicalandEngineeringNews81,no.20:4555.Availableathttp://pubs.acs.org/cen/coverstory/8120/8120biology.html.Herdt,G.1997.SameSexDifferentCultures.Boulder,CO:WestviewPress.Hesse,M.1966.ModelsandAnalogiesinScience.NotreDame,IN:UniversityofNotreDamePress.Hill,M.O.1973.Diversityandevenness:Aunifyingnotationanditsconsequences.Ecology54:42732.Hillis,D.M.,J.P.Huelsenbeck,andD.L.Swofford.1994.Hobgoblinofphylogenetics?Nature369:36364.Hines,W.G.S.1987.ESStheory:Abasicreview.TheoreticalPopulationBiology31:195272.Hodge,M.J.S.1987.Naturalselectionasacausal,empirical,andprobabilistictheory.TheProbabilisticRevolution.EditorL.Kruger.Cambridge,MA:MITPress.Holmes,F.L.2000.SymourBenzerandthedefinitionoftheGene.TheConceptoftheGeneinDevelopmentandEvolution.EditorsR.Falk,H.J.Rheinberger,andP.Beurton,11555.Cambridge:CambridgeUniversityPress.Holyoak,K.J.,andP.Thagard.1995.MentalLeaps:AnalogyinCreativeThought.Cambridge,MA:MITPress.Hood,L.,J.R.Heath,M.E.Phelps,andB.Lin.2004.Systemsbiologyandnewtechnologiesenablepredictiveandpreventativemedicine.Science306:64043.Hopwood,N.2002.EmbryosinWaxModelsfromtheZieglerStudio.UniversityofCambridgeandUniversityofBern:WhippleMuseumoftheHistoryofScienceandtheInstituteoftheHistoryofMedicine.Hossfeld,U.2005.GeschichtederbiologischenAnthropologieinDeutschland.Stuttgart:FranzSteinerVerlag.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n470ReferenceListHu,S.,A.Pattatucci,C.Patterson,L.Li,D.Fulker,S.Cherny,L.Kruglyac,andD.Hamer.1995.LinkagebetweensexualorientationandchromosomeXq28inmalesbutnotfemales.NatureGenetics11:24856.Hubbell,S.P.2001.TheUnifiedNeutralTheoryofBiodiversityandBiogeography.Princeton,NJ:PrincetonUniversityPress.Huelsenbeck,J.P.,F.Ronquiest,R.Nielson,andJ.P.Bollback.2001.Bayesianinferenceofphylogenyanditsimpactonevolutionarybiology.Science294:231014.Hull,D.1976a.Arespeciesreallyindividuals?SystematicZoology25:17491.1994.Contemporarysystematicphilosophies.ConceptualIssuesinEvolutionaryBiology.EditorE.Sober,295330.Cambridge,MA:MITPress.1965.Theeffectofessentialismontaxonomy:Twothousandyearsofstasis.BritishJournalforthePhilosophyofScience15:31426;16:118.1980.Individualityandselection.AnnualReviewofEcologyandSystematics11:31132.1976b.Informalaspectsoftheoryreduction.PSA1974.EditorR.S.Cohen,65370.Dordrecht:Reidel.1988.ScienceasaProcess:AnEvolutionaryAccountoftheSocialandConceptualDevelopmentofScience.Chicago:UniversityofChicagoPress.1999.TheuseandabuseofSirKarlPopper.BiologyandPhilosophy14:481504.Hull,D.,andM.V.H.VanRegenmortel,Editors.2002.PromisesandLimitsofReductionismintheBiomedicalSciences.Chichester,England:JohnWiley.Hume,D.1990.DialoguesConcerningNaturalReligion.EditorMartinBell.London:Penguin.Hurlbert,S.H.1971.Thenonconceptsofspeciesdiversity:Acritiqueandalternativeparameters.Ecology52:57786.Hutcheson,K.1970.TheMomentsandDistributionforanEstimateoftheShannonInformationMeasureandItsApplicationtoEcology.Ph.D.dissertation.VirginiaPolytechnicInstitute.Huxley,J.S.1942.Evolution:TheModernSynthesis.London:AllenandUnwin.Ideker,T.,T.Galitski,andL.Hood.2001.Anewapproachtodecodinglife:Systemsbiology.AnnualReviewofGenomicsandHumanGenetics2:34372.Jablonka,E.2002.Information:Itsinterpretaion,itsinheritanceanditssharing.PhilosophyofScience69:578605.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList471Jacob,F.1998OfFlies,MiceandMen:OntheRevolutioninModernBiology,byOneoftheScientistsWhoHelpedMakeIt.Cambridge,MA:HarvardUniversityPress.Janzen,D.H.1986.Thefutureoftropicalecology.AnnualReviewofEcology17:30524.Johansen,T.K.2004.Plato’sNaturalPhilosophy:AStudyoftheTimaeus-Critias.Cambridge:CambridgeUniversityPress.Johnson,P.E.2004.OverestimatingAIDS.TouchstoneMagazine,October.1996.Startingaconversationaboutevolution:AreviewofTheBattleoftheBeginnings:WhyNeitherSideIsWinningtheCreation-EvolutionDebatebyDelRatzsch[Internet].AccessResearchNetwork,8/31/96.Availablefromwww.arn.org/docs/johnson/ratzsch.htm.Judson,H.F.1979.TheEightDayofCreation:MakersoftheRevolutioninBiology.NewYork:Simon&Schuster.Juengst,E.T.1991.TheHumanGenomeProjectandbioethics.KennedyInstituteofEthicsJournal1:7174.Justus,J.,andS.Sarkar.2002.Theprincipleofcomplementarityinthedesignofreservenetworkstoconservebiodiversity:Apreliminaryhistory.JournalofBiosciences27:42135.Karp,P.1989.HypothesisFormationandQualitativeReasoninginMolecularBiology.Ph.D.dissertation.StanfordUniversity.Kauffman,S.A.1993.TheOriginsofOrder:Self-OrganizationandSelectioninEvolution.Oxford:OxfordUniversityPress.Kay,L.2000.WhoWrotetheBookofLife?AHistoryoftheGeneticCode.PaloAlto,CA:StanfordUniversityPress.Kearney,M.,andO.Rieppel.2006.Rejectingthegiveninsystematics.Cladistics22:36977.Keller,E.F.2002.MakingSenseofLife:ExplainingBiologicalDevelopmentwithModels,Metaphors,andMachines.Cambridge,MA:HarvardUniversityPress.Keller,R.A.,R.N.Boyd,andQ.D.Wheeler.2003.Theillogicalbasisofphylogeneticnomenclature.BotanicalReview69:93110.Kendler,K.S.,L.M.Thornton,S.E.Gilman,andR.C.Kessler.2000.SexualorientationinaU.S.nationalsampleoftwinandnontwinsiblingpairs.AmericanJournalofPsychiatry157:184346.Kettlewell,H.B.D.1956.FurtherselectionexperimentsonindustrialmelanismintheLepidoptera.Heredity10:287301.1955.SelectionexperimentsonindustrialmelanismintheLepidoptera.Heredity9:32342.Kilpatrick,A.M.,andA.R.Ives.2003.SpeciesinteractionscanexplainTaylorspowerlawforecologicaltimeseries.Nature422:6568.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n472ReferenceListKincaid,H.1996.PhilosophicalFoundationsoftheSocialSciences:AnalyzingControversiesinSocialResearch.NewYork:CambridgeUniversityPress.Kingsland,S.2002.Creatingascienceofnaturereservedesign:Perspectivesfromhistory.EnvironmentalModelingandAssessment7:6169.Kingslover,J.G.,H.E.Hoekstra,J.M.Hoekstra,D.Berrigan,S.N.Vignieris,C.E.Hoang,A.Hill,P.Gibert,andP.Beerli.2001.Thestrengthofphenotypicselectioninnaturalpopulations.AmericanNaturalist157,no.3:24561.Kinsey,A.,C.Pomeroy,B.Wardell,andC.E.Martin.1948.SexualBehaviourinTheHumanMale.Philadelphia:W.B.Saunders.Kinzig,A.P.,S.W.Pacala,andD.Tilman,Editors.2002.TheFunctionalConsequencesofBiodiversity:EmpiricalProgressandTheoreticalExtensions.Princeton,NJ:PrincetonUniversityPress.Kirk,K.M.,J.M.Dunne,M.P.Martin,andN.G.Bailey.2000.Measurementmodelsforsexualorientationinacommunitytwinsample.BehaviorGenetics30:34556.Kirkpatrick,J.B.,andC.E.Harwood.1983.ConservationofTasmanianmacrophyticwetlandvegetation.ProceedingsoftheRoyalSocietyofTasmania117:520.Kirschner,M.,andJ.Gerhart.2005.ThePlausibilityofLife:GreatLeapsofEvolution.NewHaven,CT:YaleUniversityPress.Kirschner,M.W.2005.Themeaningofsystemsbiology.Cell121:50304.Kitano,H.2002a.Lookingbeyondthedetails:Ariseinsystem-orientedapproachesingeneticsandmolecularbiology.CurrentGenetics41:110.2002b.Systemsbiology:Abriefoverview.Science295:166264.Kitcher,P.1984.1953andallthat:Ataleoftwosciences.PhilosophicalReview93:33573.2001.Battlingtheundead:How(andhownot)toresistgeneticdeterminism.ThinkingAboutEvolution:Historical,PhilosophicalandPoliticalPerspectives.EditorsR.Singh,D.Paul,J.Beatty,andC.Krimbas.Cambridge:CambridgeUniversityPress.1988.Explanatoryunification.TheoriesofExplanation.EditorJ.Pitt,16787.Oxford:OxfordUniversityPress.1999.Gamessocialanimalsplay:CommentaryonBrianSkyrmsEvolutionoftheSocialContract.PhilosophyandPhenomenologicalResearch59,no.1:22128.1999.Thehegemonyofmolecularbiology.BiologyandPhilosophy14,no.2:195210.1996.TheLivestoCome:TheGeneticRevolutionandHumanPossibilities.NewYork:Simon&Schuster.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList4732001.ScienceTruthandDemocracy.NewYork:OxfordUniversityPress.Kluge,A.G.2001.Philosophicalconjecturesandtheirrefutation.SystematicBiology50:32230.2003.Therepugnantandthematureinphylogeneticinference:Atemporalsimilarityandhistoricalidentity.Cladistics19:35668.1999.Thescienceofphylogeneticsystematics:Explanation,prediction,andtest.Cladistics15:42936.1997.Testabilityandtherefutationandcorroborationofcladistichypotheses.Cladistics13:8196.Kluge,A.G.,andJ.S.Farris.1969.Quantitativephyleticsandtheevolutionofanurans.SystematicZoology19:35668.Knight,R.,S.Freeland,andL.Landweber.1999.Selection,historyandchemistry:Thethreefacesofthegeneticcode.TrendsinBiochemicalSciences24:24147.Knoll,A.H.2003.LifeonaYoungPlanet.Princeton,NJ:PrincetonUniversityPress.Knoll,A.,andS.B.Carroll.1999.Earlyanimalevolution:Emergingviewsfromcomparativebiologyandgeology.Science284:212937.Koleff,P.,K.J.Gaston,andJ.J.Lennon.2003.Measuringbetadiversityforpresence-absencedata.JournalofAnimalEcology72:36782.Kowalevsky,A.1867.EntwicklungsgeschichtedesAmphioxuslanceolatus.MemoirsoftheAcademyofSciences,St.Petersburg11:117.Kreitman,M.,andR.R.Hudson.1991.InferringtheevolutionarytheoriesoftheAdhandAdh-duplociinDrosophilamelanogasterfrompatternsofpolymorphismanddivergence.Genetics127:56582.Kreps,D.1990.GameTheoryandEconomicModeling.Oxford:ClarendonPress.Ku¨hn,A.1955.VorlesungenuberEntwicklungsphysiologie.Berlin:Springer.Laland,K.,andG.R.Brown.2002.SenseandNonsense:EvolutionaryPerspectivesonHumanBehaviour.Oxford:OxfordUniversityPress.Landecker,H.L.2004.Buildinganewtypeofbodyinwhichtogrowacell:Theoriginsoftissueculture.CreatingaTraditionofBiomedicalResearch:ContibutionstotheHistoryoftheRockefellerUniversity.EditorDarwinStapleton,15174.NewYork:RockefellerUniversityPress.Laubichler,M.D.2003.CarlGegenbaur(18261903):Integratingcomparativeanatomyandembryology.JournalofExperimentalZoology;PartB:MolecularandDevelopmentalEvolution300,no.1:2331.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n474ReferenceList2005.EvolutionareEntwicklungsbiologie.PhilosophiederBiologie.EditorsU.KrohsandG.Toepfer,32237.Frankfurt/Main:Suhrkamp.2000.Theorganismisdead.Longlivetheorganism!PerspectivesonScience8:286315.Laubichler,M.,andJ.Maienschein.2007.FromEmbryologytoEvo-Devo.Cambridge,MA:MITPress.Laubichler,M.D.,andJ.Maienschein.2003.Ontogeny,anatomyandtheproblemofhomology:CarlGegenbaurandtheAmericantraditionofcelllineagestudies.TheoryinBiosciences112:194203.Lauder,G.V.1996.Theargumentfromdesign.Adaptation.EditorsM.R.RoseandG.V.Lauder,5591.SanDiego:AcademicPress.Laumann,E.O.,J.H.Michael,R.T.Michaels,andS.Gagnon.1994.TheSocialOrganizationofSexuality:SexualPracticesintheUnitedStates.Chicago:UniversityofChicagoPress.Lea,H.1904.Ethicalvaluesinhistory.AmericanHistoricalReview9:23346.Lee,M.1999.Molecularclockcalibrationsandmetazoandivergencedates.JournalofMolecularEvolution49,no.3:38591.Lehman,C.L.,andD.Tilman.2000.Biodiversity,stability,andproductivityincompetitivecommunities.AmericanNaturalist156:53452.Leibold,M.A.1996.Agraphicalmodelofkeystonepredatorsinfoodwebs:Trophicregulationofabundance,incidence,anddiversitypatternsincommunities.AmericanNaturalist147:784812.Leibold,M.A.,J.M.Chase,J.B.Shurin,andA.L.Downing.1997.Speciesturnoverandtheregulationoftrophicstructure.AnnualReviewofEcologyandSystematics28:46794.Lennox,J.G.1993.Darwinwasateleologist.BiologyandPhilosophy8:40921.1985.Platosunnaturalteleology.PlatonicInvestigations.EditorD.OMeara,195218.Pittsburgh:MathesisPublications.Lenoir,T.1982.TheStrategyofLife:TeleologyandMechanicsinNineteenth-CenturyGermanBiology.Chicago:UniversityofChicagoPress.Leroi,A.M.2000.Thescaleindependenceofevolution.Evolution&Development2,no.2:6777.Lewens,T.2002.Adaptationismandengineering.BiologyandPhilosophy17:131.2004.OrganismsandArtifacts:DesigninNatureandElsewhere.Cambridge,MA:MITPress.Forthcoming.Seventypesofadaptationism.Twenty-FiveYearsofSpandrels.EditorD.M.Walsh.Oxford:OxfordUniversityPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList475Lewicki,M.S.1998.Areviewofmethodsforspikesorting:Thedetectionandclassificationofneuralactionpotentials.Network:ComputationalNeuralSystems9:R53R78.Lewis,D.1973b.Causation.JournalofPhilosophy70:55667.1969.Convention:APhilosophicalStudy.Cambridge,MA:HarvardUniversityPress.1973a.Counterfactuals.Oxford:OxfordUniversityPress.Lewontin,R.C.1978.Adaptation.ScientificAmerican239,no.3:21330.1984.Adaptation.ConceptualIssuesinEvolutionaryBiology.EditorE.Sober.Cambridge,MA:MITPress.1985.Adaptation.TheDialecticalBiologist.EditorsR.LevinsandR.Lewontin,6584.Cambridge,MA:HarvardUniversityPress.1992.Thedreamofthehumangenome.NewYorkReviewofBooks39,no.10(May28):3140.1958.Ageneralmethodforinvestigatingtheequilibriumofgenefrequencyinapopulation.Genetics43:42133.1969a.Themeaningofstability.BrookhavenSymposiainBiology22:1324.1969b.Theorganismassubjectandobjectofevolution.TheDialecticalBiologist.EditorsR.C.LewontinandR.Levins.Cambridge,MA:HarvardUniversityPress.1985.Populationgenetics.AnnualReviewofGenetics19:81102.2000a.TheTripleHelix:Gene,Organism,andEnvironment.Cambridge,MA:HarvardUniversityPress.1970.Theunitsofselection.AnnualReviewofEcologyandSystematics1:118.2000b.Whatdopopulationgeneticistsknowandhowdotheyknowit?BiologyandEpistemology.EditorsR.CreathandJaneMaienschein,191214.Cambridge:CambridgeUniversityPress.Lewontin,R.,andR.Levins.1985.TheDialecticalBiologist.Cambridge:CambridgeUniversityPress.Likens,G.E.,F.H.Bormann,N.M.Johnson,D.W.Fisher,andR.S.Pierce.1970.EffectsofforestcuttingandherbicidetreatmentonnutrientbudgetsintheHubbardBrookwatershed-ecosystem.EcologicalMonographs40:2347.Lippman,A.1992.Led(astray)bygeneticmaps.SocialScienceandMedicine35:146976.Lipton,P.2004.InferencetotheBestExplanation.2nded.London:Routledge.Livingstone,D.N.1987.Darwin’sForgottenDefenders:TheEncounterBetweenEvangelicalTheologyandEvolutionaryThought.GrandRapids,MI:WilliamB.Eerdmans.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n476ReferenceListLloyd,E.1987.Confirmationofecologicalandevolutionarymodels.BiologyandPhilosophy2:27793.1988.TheStructureandConfirmationofEvolutionaryTheory.NewYork:Greenwood.1992.Unitofselection.KeywordsinEvolutionaryBiology.EditorsE.F.KellerandE.A.Lloyd,33440.Cambridge,MA:HarvardUniversityPress.2001.Unitsandlevelsofselection:Ananatomyoftheunitsofselectiondebates.ThinkingAboutEvolution:Historical,PhilosophicalandPoliticalPerspectives.EditorsR.Singh,C.Krimbas,D.Paul,andJ.Beatty,26791.NewYork:CambridgeUniversityPress.Lloyd,E.A.,andS.J.Gould.1993.Speciesselectiononvariability.ProceedingsoftheNationalAcademyofSciences90:59599.Logan,G.A.,etal.1995.Terminalprotoerozoicreorganizationofbiogeochemicalcycles.Nature376:5356.Love,A.C.,andR.A.Raff.2003.Knowingyourancestors:Themesinthehistoryofevo-devo.EvolutionandDevelopment5,no.4:32730.Lovejoy,A.O.1965.TheGreatChainofBeing:AStudyoftheHistoryofanIdea.NewYork:Harper&Row.MacArthur,R.H.1957.Ontherelativeabundanceofbirdspecies.ProceedingsoftheNationalAcademyofSciences43:29395.MacArthur,R.H.1965.Patternsofspeciesdiversity.BiologicaReview40:51033.MacArthur,R.H.,andE.O.Wilson.1967.TheTheoryofIslandBiogeography.Princeton,NJ:PrincetonUniversityPress.Machamer,P.2004.Activitiesandcausation:Themetaphysicsandepistemologyofmechanisms.InternationalStudiesinthePhilosophyofScience18:2739.Machamer,P.,C.Craver,andL.Darden.2000.Thinkingaboutmechanisms.PhilosophyofScience67:125.Magurran,A.E.1988.EcologicalDiversityandItsMeasurements.Princeton,NJ:PrincetonUniversityPress.2004.MeasuringBiologicalDiversity.Oxford:Blackwell.Maienschein,J.1991.TransformingTraditionsinAmericanBiology:1880–1915.Baltimore:JohnsHopkinsUniversityPress.2005.WhoseViewofLife?Embryos,CloningandStemCells.Cambridge,MA:HarvardUniversityPress.Maienschein,J.,M.Glitz,andG.E.Allen,Editors.2004.CentennielHistoryoftheCarnegieInstitutionofWashington.Vol.5,TheDepartmentofEmbryology.Cambridge:CambridgeUniversityPress.Marcus,G.2004.TheBirthoftheMind:HowaTinyNumberofGenesCreatestheComplexitiesofHumanThought.NewYork:BasicBooks.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList477Marcus,S.J.,Editor.2004.Neuroethics:ConferenceProceedings.NewYork:DanaPress.Margalef,R.1958.Informationtheoryinecology.GeneralSystemsYearbook3:3671.Margules,C.R.,A.O.Nicholls,andR.L.Pressey.1988.Selectingnetworksofreservestomaximizebiologicaldiversity.BiologicalConservation43:6376.Margules,C.R.,andR.L.Pressey.2000.Systematicconservationplanning.Nature405:24253.Marquet,P.A.,R.A.Quinones,S.Abades,F.Labra,M.Tognelli,M.Arim,andM.Rivadeneira.2005.Scalingandpower-lawsinecologicalsytems.JournalofExperimentalBiology208:174969.Matessi,C.,andS.D.Jayakar.1976.ConditionsfortheevolutionofaltruismunderDarwinianselection.TheoreticalPopulationBiology9:36087.Matthen,M.,andA.Ariew.2002.Twowaysofthinkingaboutfitnessandnaturalselection.JournalofPhilosophy99,no.2:5583.May,R.M.1973.StabilityandComplexityinModelEcosystems.Princeton,NJ:PrincetonUniversityPress.MaynardSmith,J.2000.Theconceptofinformationinbiology.PhilosophyofScience67:17794.1982.EvolutionandtheTheoryofGames.Cambridge:CambridgeUniversityPress.1976.Groupselection.QuarterlyReviewofBiology51:27783.1964.Groupselectionandkinselection.Nature201,no.4924:114547.1987.Howtomodelevolution.TheLatestontheBest:EssaysonEvolutionandOptimality.EditorJ.Dupre,11931.Cambridge,MA:MITPress.2001.ReconcilingMarxandDarwin.Evolution55,no.7:14998.1969.Thestatusofneo-Darwinism.TowardsaTheoreticalBiology.EditorC.H.Waddington.Edinburgh:EdinburghUniversityPress.1974.Thetheoryofgamesandtheevolutionofanimalconflicts.JournalofTheoreticalBiology47:20921.MaynardSmith,J.,R.Burian,S.Kauffman,P.Alberch,J.Campbell,B.Goodwin,R.Lande,D.Raup,andL.Wolpert.1985.Developmentalconstraintsandevolution.QuarterlyReviewofBiology60:26587.MaynardSmith,J.,andG.R.Price.1973.Thelogicofanimalconflict.Nature246:1518.MaynardSmith,J.,andE.Szathmary.1995.TheMajorTransitionsinEvolution.NewYork:W.H.Freeman.1999.TheOriginsofLife:FromtheBirthofLifetotheOriginofLanguage.Oxford:OxfordUniversityPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n478ReferenceListMayr,E.1974.Behaviorprogramsandevolutionarystrategies.AmericanScientist62:65059.1961.Causeandeffectinbiology.Science134:150106.1976.EvolutionandtheDiversityofLife.Cambridge,MA:HarvardUniversityPress.1982.TheGrowthofBiologicalThought:Diversity,Evolution,andInheritance.Cambridge,MA:BelknapPress.1986.Naturalselection:Thephilosopherandthebiologist.Paleobiology12:23339.1965.Numericalpheneticsandtaxonomictheory.SystematicZoology14:7397.1987.Theontologicalstatusofspecies:Scientificprogressandphilosophicalterminology.BiologyandPhilosophy2:14566.1969.PrinciplesofSystematicZoology.NewYork:McGraw-Hill.1942.SystematicsandtheOriginofSpecies.NewYork:ColumbiaUniversityPress.1988.TowardsaNewPhilosophyofBiology:ObservationsofanEvolutionist.Cambridge,MA:BelknapPress.1959.Typologicalversuspopulationthinking.EvolutionandAnthropology:ACentennialAppraisal.EditorB.J.Meggers,40912.Washington,DC:AnthropologicalSocietyofWashington.Mayr,E.,andW.Provine,Editors.1980.TheEvolutionarySynthesis:PerspectivesontheUnificationofBiology.Cambridge,MA:HarvardUniversityPress.McCann,K.S.2000.Thediversity-stabilitydebate.Nature405:22833.McGue,M.1999.Thebehavioralgeneticsofalcoholism.CurrentDirectionsinPsychologicalScience8:10915.McKelvey,R.D.,andT.R.Palfrey.1992.AnExperimentalStudyoftheCentipedeGame.Econometrica60,no.4:80336.McKnight,J.,andJ.Malcom.2000.Ismalehomosexualitymaternallylinked?Psychology,Evolution&Gender2:22939.McMenamin,M.,andD.McMenamin.1990.TheEmergenceofAnimals:TheCambrianBreakthrough.NewYork:ColumbiaUniversityPress.McMullin,E.2000.LifeandintelligencefarfromEarth:Formulatingtheologicalissues.ManyWorlds.EditorStevenDick,15175.Philadelphia:TempletonPress.McShea,D.W.1996.Metazoancomplexityandevolution:Isthereatrend?Evolution50:47792.1998a.Possiblelargest-scaletrendsinorganismalevolution:Eightlivehypotheses.AnnualReviewofEcologyandSystematics29:293318.2000.Trends,ToolsandTerminology.Paleobiology26,no.3:33033.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList479Merzenich,M.M.,J.H.Kaas,M.Sur,R.J.Nelson,andD.J.Fellemen.1983.Progressionofchangefollowingmediannervesectioninthecorticalrepresentationofthehandinareas3band1inadultowlandsquirrelmonkeys.Neuroscience10:63965.Michod,R.W.1999.DarwinianDynamics:EvolutionaryTransitionsinFitnessandIndividuality.Princeton,NJ:PrincetonUniversityPress.Mikkelson,G.M.2004.Biologicaldiversity,ecologicalstability,anddownwardcausation.PhilosophyandBiodiversity.EditorsM.OksanenandJ.Pietarinen,11929.NewYork:CambridgeUniversityPress.2003.Ecologicalkindsandecologicallaws.PhilosophyofScience70:13901400.InPress.Realismvs.instrumentalisminanewstatisticalframework.PhilosophyofScience.Miles,F.A.,andS.G.Lisberger.1981.Plasticityinthevestibulo-ocularreflex:Anewhypothesis.AnnualReviewofNeuroscience4:27999.Millikan,R.1984.Language,ThoughtandOtherBiologicalCategories.Cambridge,MA:MITPress.Mills,S.,andJ.Beatty.1979.Thepropensityinterpretationoffitness.PhilosophyofScience46:26386.Millstein,R.L.2002.Arerandomdriftandnaturalselectionconceptuallydistinct?BiologyandPhilosophy17:3353.2005.Selectionvs.drift:AresponsetoBrandonsreply.BiologyandPhilosophy20:17175.Mishkin,M.,L.G.Ungerleider,andK.A.Macko.1983.Objectvisionandspatialvision:Twocorticalpathways.TrendsinNeuroscience6:27399.Mishler,B.,andM.Donoghue.1982.Speciesconcepts:Acaseforpluralism.SystematicZoology31:491503.Moore,J.H.,E.M.Boczko,andM.L.Summar.2005.Connectingthedotsbetweengenes,biochemistry,anddiseasesusceptibility:Systemsbiologymodelinginhumangenetics.MolecularGeneticsandMetabolism84:10411.Morange,M.1998.AHistoryofMolecularBiology.Cambridge,MA:HarvardUniversityPress.Morowitz,H.1985.ModelsforBiomedicalResearch:ANewPerspective.ReportoftheCommitteeonModelsforBiomedicalResearch.Washington,DC:NationalAcademyPress.Moss,L.1992.Akerneloftruth?Ontherealityofthegeneticprogram.PSA1992.Vol.1.EditorsD.Hull,M.Forbes,andK.Okruhlik,33548.EastLansing,MI:PhilosophyofScienceAssociation.2003.WhatGenesCan’tDo.Cambridge,MA:MITPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n480ReferenceListMoyers,B.2004.OnreceivingHarvardMedicalSchoolsGlobalEnviron-mentCitizenAward.Availableathttp://www.commondreams.org.Mueller,L.D.,andA.Joshi.2000.StabilityinModelPopulations.Princeton,NJ:PrincetonUniversityPress.Muller,G.1989.Ancestralpatternsinbirddevelopment.JournalofEvolutionaryBiology2:3147.2005.Evolutionarydevelopmentalbiology.HandbookofEvolution.EditorsF.M.Wuketits,andF.J.Ayala,Vol.2,87115.Weinheim:Wiley-VCH.Muller,G.,andS.Newman.1999.Generation,integration,autonomy:Thestepsintheevolutionofhomology.Homology.EditorsG.CardewandG.R.Bock,6573.Chichester,England:JohnWiley.2003.Originationoforganismalform:Beyondthegeneindevelopmentalandevolutionarybiology.Cambridge,MA:MITPress.Muller,G.,andG.P.Wagner.1991.Noveltyinevolution:Restructuringtheconcept.AnnualReviewofEcologicalSystems22:22956.Mustanski,B.S.,M.L.Bailey,andJ.M.Chivers.2002.Acriticalreviewofrecentbiologicalresearchonhumansexualorientation.AnnualReviewofSexualResearch12:89140.Naeem,S.2002.Biodiversityequalsinstability.Nature416:2324.2002.Ecosystemconsequencesofbiodiversityloss:Theevolutionofaparadigm.Ecology83:153752.Nagel,E.1961.TheStructureofScience:ProblemsintheLogicofScientificExplanation.NewYork:Harcourt,Brace&World.Narbonne,G.2005.TheEdiacarabiota:Neoproterozoicoriginofanimalsandtheirecosystems.AnnualReviewofEarthandPlantScience33:42142.Nash,G.,andC.Crabtree,Supervisors.1996.NationalStandardsforHistory,BasicEdition.LosAngeles:NationalCenterforHistoryintheSchools.Nash,J.1950.Thebargainingproblem.Econometrica18:15562.1951.Non-cooperativegames.AnnalsofMathematics54,no.2:28695.Neander,K.1995.Pruningthetreeoflife.BritishJournalforthePhilosophyofScience46:5980.1991.Theteleologicalnotionoffunction.AusralasianJournalofPhilosophy69:45468.Nelkin,D.,andM.S.Lindee.1995.TheDNAMystique.NewYork:W.H.Freeman.Nelson,G.,andN.Platnick.1981.SystematicsandBiogeography:CladisticsandVicariance.NewYork:ColumbiaUniversityPress.Neumann-Held,E.M.1998.Thegeneisdeadlonglivethegene:Conceptualisingthegenetheconstructionistway.SociobiologyandCambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList481Bioeconomics:TheTheoryofEvolutioninBiologicalandEconomicTheory.EditorP.Koslowski,10537.Berlin:Springer-Verlag.Newberg,A.B.,etal.2002.WhyGodWon’tGoAway:BrainScienceandtheBiologyofBelief.NewYork:BallantineBooks.Newman,S.A.2003.Thefallandriseofsystemsbiology.Availableathttp://www.gene-watch.org/genewatch/articles/164newman.html.Nicholson,J.K.,andI.D.Wilson.2003.Understandingglobalsystemsbiology:Metabonomicsandthecontinuumofmetabolism.NatureReviewsDrugDiscovery2:66876.Nixon,K.C.,andJ.M.Carpenter.2000.Ontheotherphylogeneticsystematics.Cladistics16:298318.Nixon,K.C.,andQ.D.Wheeler.1990.Anamplificationofthephylogeneticspeciesconcept.Cladistics6:21123.Norton,B.G.1987.WhyPreserveNaturalVariety?Princeton,NJ:PrincetonUniversityPress.Novick,P.1988.ThatNobleDream:The‘‘ObjectivityQuestion’’andtheAmericanHistoricalProfession.Cambridge:CambridgeUniversityPress.Nursall,J.R.1959.OxygenasaprerequisitetotheoriginoftheMetazoa.Nature183:117072.Nyhart,L.K.1995.BiologyTakesForm:AnimalMorphologyandtheGermanUniversities,1800–1900.Chicago:UniversityofChicagoPress.2002.Learningfromhistory:MorphologyschallengesinGermanyca.1900.JournalofMorphology252,no.1:214.Odenbaugh,J.2001.Ecologicalstability,modelbuilding,andenvironmentalpolicy:Areplytosomeofthepessimism.PhilosophyofScience68:S493S505.Odling-Smee,J.,etal.2003.Nicheconstruction:TheNeglectedProcessinEvolution.Princeton,NJ:PrincetonUniversityPress.Okasha,S.2004a.Theaveragingfallacyandthelevelsofselection.BiologyandPhilosophy19:16784.2003.Theconceptofgroupheritability.BiologyandPhilosophy18,no.3:44561.2004b.Multilevelselection,covarianceandcontextualanalysis.BritishJournalforthePhilosophyofScience55:481504.2004c.Multilevelselectionandthepartitioningofcovariance:Acomparisonofthreeapproaches.Evolution58,no.3:48694.Olby,R.1985.TheOriginsofMendelism.2nded.Chicago:UniversityofChicagoPress.1974.ThePathtotheDoubleHelix:TheDiscoveryofDNA.Seattle:UniversityofWashingtonPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n482ReferenceListOppenheim,P.,andH.Putnam.1958.Unityofscienceasaworkinghypothesis.Concepts,Theories,andtheMind-BodyProblem.EditorsH.Feigl,M.Scriven,andG.Maxwell.MinnesotaStudiesinthePhilosophyofScience.Vol.2,336.Minneapolis:UniversityofMinnesotaPress.Orr,H.2005.Masterplanned:Whyintelligentdesignisnt.NewYorker,May30.Ortona,G.1991.Theultimatumgame.EconomicNotes20,no.2:32434.Orzack,S.H.,andE.Sober.1994.Optimalitymodelsandthetestofadaptationism.AmericanNaturalist143:36180.Oyama,S.2001.CyclesofContingency:DevelopmentalSystemsandEvolution.EditorsS.Oyama,P.E.Griffiths,andR.D.Gray.Cambridge,MA:MITPress.1985.TheOntogenyofInformation.Cambridge:CambridgeUniversityPress.Paine,R.T.Foodwebcomplexityandspeciesdiversity.TheAmericanNaturalist100:6575.Paley,W.1828.NaturalTheology.2nded.Oxford:J.Vincent.Parker,A.2003.IntheBlinkofanEye.Cambridge:Perseus.Patil,G.P.,andC.Taillie.1976.Biologicaldiversity:Concepts,indices,andapplications.Proceedingsofthe9thInternationalBiometricConference,Raleigh,NC,383411.1982.Diversityasaconceptanditsmeasurement.JournaloftheAmericanStatisticalAssociation77:54861.1979.Anoverviewofdiversity.EcologicalDiversityinTheoryandPractice.EditorsJ.F.Grassel,G.Patil,P.Smith,andC.Taillie,327.Fairland,MD:InternationalCooperativePublishingHouse.Patil,G.P.,andC.Taillie,Editors.1994.HandbookofStatistics12:EnvironmentalStatistics.Amsterdam:ElsevierScience.Pattatucci,A.M.L.,andD.H.Hamer.1995.Developmentandfamiliarityofsexualorientationinfemales.BehaviorGenetics25:40720.Patten,B.C.1962.SpeciesdiversityinnetphytoplanktonofRaritanBay.JournalofMarineResearch20:5775.Patterson,C.1982.Morphologicalcharactersandhomology.ProblemsofPhylogeneticReconstruction.EditorsK.A.JoyseyandA.E.Friday,2174.London:AcademicPress.Peet,R.K.1974.Themeasurementofspeciesdiversity.AnnualReviewofEcologyandSystematics5:285307.Pennisi,E.2003.Tracinglifescircuitry.Science302:164649.Pennock,R.T.1999.TowerofBabel:TheEvidenceAgainsttheNewCreationism.Cambridge,MA:MITPress.Peters,T.2003.PlayingGod?NewYork:Routledge.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList483Peterson,K.,andN.Butterfield.2005.OriginoftheEumetazoa:TestingecologicalpredictionsofmolecularclocksagainsttheProterozoicfossilrecord.ProceedingsoftheNationalAcademyofSciences102,no.27:954752.PetersonK.,andE.Davidson.2000.RegulatoryevolutionandtheoriginofBilaterians.ProceedingsoftheNationalAcademyofScience97,no.9:443033.Peterson,K.,etal.2000.Bilateranorigins:Significanceofnewexperimentalobservations.DevelopmentalBiology219:117.2004.Estimatingmetazoandivergencetimeswithamolecularclock.ProceedingsoftheNationalAcademyofScience101,no.17:653641.Pielou,E.C.1975.EcologicalDiversity.London:JohnWiley.1969.AnIntroductiontoMathematicalEcology.London:WileyInterscience.Pietilainen,K.H.,J.Rissanen,A.Winter,T.Rimpela,A.Viken,R.J.Rose,andR.J.Kaprio.1999.DistributionandheritabilityofBMIinFinnishadolescentsaged16yand17y:Astudyof4884twinsand2509singletons.InternationalJournalofObesity23:10715.Pimental,D.1961.Speciesdiversityandinsectpopulationoutbreaks.AnnalsoftheEntomologicalSocietyofAmerica54:7686.Pimm,S.L.1991.TheBalanceofNature?EcologicalIssuesintheConservationofSpeciesandCommunities.Chicago:UniversityofChicagoPress.Pinker,S.,andP.Bloom.1990.Naturallanguageandnaturalselection.BehavioralandBrainSciences13.Pinto-Correia,C.1997.TheOvaryofEve:EggandSpermandPreformation.Chicago:UniversityofChicagoPress.Platnick,N.I.1979.Philosophyandthetransformationofcladistics.SystematicZoology28:53746.Plato.2000.Timeaus.TranslatorD.Zeyl.Indianapolis:Hackett.Poole,R.W.1974.AnIntroductiontoQuantitativeEcology.NewYork:McGraw-Hill.Popper,K.R.1962.ConjecturesandRefutations.NewYork:BasicBooks.1959.TheLogicofScientificDiscovery.London:Hutchinson.Portin,P.1993.Theconceptofthegene:Shorthistoryandpresentstatus.TheQuarterlyReviewofBiology68,no.2:173223.Preston,F.E.1962.Thecanonicaldistributionofcommonnessandrarity.Ecology43:185215,41032.1948.Thecommonness,andrarity,ofspecies.Ecology29,no.3:25483.Przibram,H.1907.Experimental-Zoologie.Teil1.Embryogenese.LeipzigandWien:FranzDeuticke.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n484ReferenceListPusey,A.,andC.Packer.1986.Dispersalandphilopatry.PrimateSocieties.EditorsB.Smuts,D.Cheney,M.Seyfarth,R.Wrangham,andT.Struhsaker,25066.Chicago:UniversityofChicagoPress.Putnam,H.2002.TheCollapseoftheFact-ValueDichotemy.Cambridge,MA:HarvardUniversityPress.Rabin,M.1993.Incorporatingfairnessintogametheory.AmericanEconomicReview83:12811302.Ramsey,G.2006.Blockfitness.StudiesinHistoryandPhilosophyofScienceC37:48498.Rao,C.R.1982.Diversityanddissimilarityindices:Aunifiedapproach.TheoreticalPopulationBiology21:2443.Reeve,H.K.,andP.W.Sherman.1993.Adaptationandthegoalsofevolutionaryresearch.QuarterlyReviewofBiology68:132.Reynolds,V.,andR.Tanner.1983.TheBiologyofReligion.London:Longman.1995.TheSocialEcologyofReligion.NewYork:OxfordUniversityPress.Rheinberger,H.-J.1997.TowardsaHistoryofEpistemicThings:SynthesizingProteinsintheTestTube.Stanford,CA:StanfordUniversityPress.Rice,G.,C.Risch,N.Ebers,andG.Anderson.1999.Malehomosexuality:AbsenceoflinkagetomicrosatellitemarkersatXq28.Science284:66567.Rice,W.R.,andG.W.Salt.1990.Theevolutionofreproductiveisolationasacorrelatedcharacterundersympatricconditions:Experimentalevidence.Evolution44:114052.Richards,R.2005.TheaestheticandmorphologicalfoundationsofErnstHaeckelsevolutionaryproject.TheManyFacesofEvolutioninEurope,1860–1914.EditorsM.KemperinkandP.Dassen,116.Amsterdam:Peeters.2003.Characterindividuationinphylogeneticinference.PhilosophyofScience70:26479.2004.Ifthisbeheresy:HaeckelsconversiontoDarwinism.DarwinianHeresies.EditorsA.Lusting,R.Richards,andM.Ruse,10130.Cambridge:CambridgeUniversityPress.2002.Kuhnian,valuesandcladisticparsimony.PerspectivesonScience10:127.1992.Thestructureofnarrativeexplanationinhistoryandbiology.HistoryandEvolution.EditorsM.NiteckiandD.Nitecki,1953.Albany,NY:SUNYPress.Richerson,P.G.,andR.Boyd.2005.NotbyGenesAlone:HowCultureTransformedHumanEvolution.Chicago:UniversityofChicagoPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList485Ricotta,C.2005.Throughthejungleofbiologicaldiversity.ActaBiotheoretica53:2938.Ridley,M.2000.Mendel’sDemon:GeneJusticeandtheComplexityofLife.London:WeidenfeldandNicolson.Riedl,R.1975.DieOrdnungdesLebendigen:Systembedingungend.Evolution.Hamburg:Parey.Rieppel,O.2006.ThePhylocode:Acriticaldiscussionofitstheoreticalfoundation.Cladistics22:18697.2003.Popperandsystematics.SystematicBiology52:25971.Rieppel,O.,andM.Kearney.2002.Similarity.BiologicalJournaloftheLinnaeanSociety75:5982.Robert,J.S.2004.Embryology,Epigenesis,andEvolution:TakingDevelopmentSeriously.NewYork:CambridgeUniversityPress.Robert,J.S.,andF.Baylis.2003.Crossingspeciesboundaries.TheAmericanJournalofBioethics3,no.3:113.Robert,J.S.,J.Maienschein,andM.Laubichler.2006.Systemsbioethicsandstemcellbiology.JournalofBioethicalInquiry3:1931.Robert,J.S.,andA.Smith.2004.Toxicethics:Environmentalgenomicsandthehealthofpopulations.Bioethics18:493514.Roe,S.1981.Matter,Life,andGeneration.Cambridge:CambridgeUniversityPress.Rohwer,Y.Forthcoming.Evolutionaryoriginsofaltruisticpunishment.PhilosophyofScience(Supplemental).Rolston,H.1999.Genes,GenesisandGod:ValuesandTheirOriginsinNaturalandHumanHistory.Cambridge:CambridgeUniversityPress.Rosenberg,A.2000.Reductionisminahistoricalscience.PhilosophyofScience68:13563.1997.Reductionismredux:Computingtheembryo.BiologyandPhilosophy12,no.4:44570.1985.TheStructureofBiologicalScience.Cambridge:CambridgeUniversityPress.1978.Thesupervenienceofbiologicalconcepts.PhilosophyofScience45:36886.Rosenzweig,M.L.2003.Reconciliationecologyandthefutureofspeciesdiversity.Oryx37:194205.1995.SpeciesDiversityinSpaceandTime.NewYork:CambridgeUniversityPress.Rousseau,R.,P.vanHecke,D.Nijssen,andJ.Bogaer.1999.Therelationshipbetweendiversityprofiles,evenness,andspeciesrichnessbasedonpartialordering.EnvironmentalandEcologicalStatistics6:21123.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n486ReferenceListRudge,D.W.1998.ABayesiananalysisofstrategiesinevolutionarybiology.PerspectivesonScience6:34160.1999.Takingthepepperedmothwithagrainofsalt.BiologyandPhilosophy14:937.Rudwick,M.J.S.1964.Theinferenceoffunctionfromstructureinfossils.BritishJournalforthePhilosophyofScience15:2740.Ruse,M.1981.Aretheregaygenes?Sociobiologyandhomosexuality.JournalofHomosexuality6,no.4:534.1970.Aretherelawsinbiology?AustralasianJournalofPhilosophy48:23446.1987.Biologicalspecies:Naturalkinds,individuals,orwhat?TheBritishJournalforthePhilosophyofScience38:22542.2003.DarwinandDesign:DoesEvolutionHaveaPurpose?Cambridge,MA:HarvardUniversityPress.1979.TheDarwinianRevolution:ScienceRedinToothandClaw.Chicago:UniversityofChicagoPress.1990.Homosexuality:APhilosophicalInquiry.Oxford:Blackwell.1977.Isbiologydifferentfromphysics?Logic,LawsandLife.EditorR.G.Colodny,89127.Pittsburgh:UniversityofPittsburghPress.1996.MonadtoMan:TheConceptofProgressinEvolutionaryBiology.Cambridge,MA:HarvardUniversityPress.1988.PhilosophyofBiologyToday.Albany,NY:SUNYPress.1976.Reductioningenetics.PSA1974.EditorR.S.Cohen,63351.Dordrect:Reidel.Russell,R.J.,W.R.Stoeger,andF.J.Ayala,Editors.1999.EvolutionaryandMolecularBiology:ScientificPerspectivesonDivineAction.Rome:VaticanObservatory.Rutimeyer,L.1868.ReviewofErnstHaeckel,UeberdieEnstehungunddenStammbaumdesMenschengeschlechtsundNaturlicheSchopfungsgeschichte.Archivfu¨rAnthropologie3:30102.Salais,D.,andR.B.Fischer.1995.Sexualpreferenceandaltruism.JournalofHomosexuality40:5177.Salmon,W.1984.ScientificExplanationandtheCausalStructureoftheWorld.Princeton,NJ:PrincetonUniversityPress.Samuelson,L.1997.EvolutionaryGamesandEquilibriumSelection.Cambridge,MA:MITPress.Sanders,A.R.1998.Posterofpresentation149,annualmeetingoftheAmericanPsychiatricAssociation,Toronto,OntarioCanada.CitedinHamer,D.1999.Geneticsandmalesexualorientation.Science285:803.Sarkar,S.2005.BiodiversityandEnvironmentalPhilosophy:AnIntroductiontotheIssues.NewYork:CambridgeUniversityPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList4872004.ConservationBiology.Availableathttp://plato.stanford.edu/archives/win2004/entrie/conservation-biology.1996.DecodingcodinginformationandDNA.BioScience46:85764.2002.Definingbiodiversity:Assessingbiodiversity.Monist85:13155.1998.GeneticsandReductionism.Cambridge:CambridgeUniversityPress.2005.MolecularModelsofLife:PhilosophicalPapersonMolecularBiology.Cambridge,MA:MITPress.1994.Theselectionofallelesandtheadditivityofvariance.PSA:ProceedingsoftheBiennialMeetingofthePhilosophyofScienceAssociation1:312.Sarkar,S.,C.Pappas,J.Garson,A.Aggarwal,andS.Cameron.2004.Placeprioritizationforbiodiversityconservationusingprobabilisticsurrogatedistributiondata.DiversityandDistribution10:12533.Schaffner,K.1993.DiscoveryandExplanationinBiologyandMedicine.Chicago:UniversityofChicagoPress.2000.Behaviorattheorganismalandmolecularlevels:ThecaseofC-elegans.PhilosophyofScience67:27388.2001.Extrapolationfromanimalmodels:Sociallife,sex,andsupermodels.TheoryandMethodintheNeurosciences.EditorsPeterMachamer,P.McLaughlin,andR.Grush,20030.Pittsburgh:UniversityofPittsburghPress.1988.Modelorganismsandbehavioralgenetics:Arejoinder.PhilosophyofScience65:27688.1976.Reductionisminbiology:Prospectsandproblems.PSA1974.EditorR.S.Cohen,61332.Dordrecht:Reidel.Schlosser,G.,andG.P.Wagner,Editors.2004.ModularityinDevelopmentandEvolution.Chicago:UniversityofChicagoPress.Schmalz,J.1993.Pollfindsanevensplitonhomosexualityscause.NewYorkTimes,A1,March5.Schoener,T.W.1986.Mechanisticapproachestoecology:Anewreductionism?AmericanZoologist26:81106.Schrodinger,E.19441992.WhatIsLife?Cambridge,MA:MITPress.Schweber,S.1977.TheoriginoftheOriginrevisited.JournaloftheHistoryofBiology10:229316.Scott,J.M.,F.Davis,B.Csuti,R.Noss,B.Butterfield,C.Groves,H.Anderson,S.Caicco,F.DErchia,T.Edwards,J.Ulliman,andG.Wright.1993.Gapanalysis:Ageographicapproachtoprotectionofbiologicaldiversity.JournalofWildlifeManagement57,no.1:123.Selten,R.1975.Reexaminationoftheperfectnessconceptforequilibriumpointsinextensivegames.InternationalJournalofGameTheory4:2555.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n488ReferenceListSenut,B.,M.Pickford,D.Gommery,P.Mein,K.Cheboi,andY.Coppens.2001.FirsthominidfromtheMiocene(LukeinoFormation,Kenya).ComptesRendusdel’AcademieJesSciences332:13744.Shaefer,K.P.,andD.L.Meyer.1974.Compensationofvestibularlesions.HandbookofSensoryPhysiology.EditorH.H.Kornhuber,Vol.6.46290.NewYork:Plenum.Shannon,C.1948.Amathematicaltheoryofcommunication.BellSystemsTechnicalJournal27:279423,62356.Shea,N.Forthcoming.RepresentationintheGenome.Simberloff,D.S.,andE.O.Wilson.1969.Experimentalzoogeographyofislands:Thecolonizationofemptyislands.Ecology50:27896.Simpson,E.H.1949.Measurementofdiversity.Nature163:688.Simpson,G.G.1953.TheMajorFeaturesofEvolution.NewYork:ColumbiaUniversityPress.1961.PrinciplesofAnimalTaxonomy.NewYork:ColumbiaUniversityPress.Sirkin,D.W.,W.Precht,andJ.H.Courjon.1984.Intitial,rapidphaseofrecoveryfromunilateralvestibularlesioninratnotdependentonsurvivalofcentralportionofvestibularnerve.BrainResearch302:24556.Skipper,R.A.Jr.2004.Calibrationoflaboratorymodelsinpopulationgenetics.PerspectivesonScience12:36993.2002.ThepersistenceoftheR.A.Fisher-SewallWrightcontroversy.BiologyandPhilosophy17:34167.Skipper,R.A.,andR.L.Millstein.2005.Thinkingaboutevolutionarymechanisms:Naturalselection.StudiesintheHistoryandPhilosophyofBiologicalandBiomedicalSciences.SpecialIssue:MechanismsinBiology36:32747.Skipper,R.A.Jr.,C.Allen,R.Ankeny,C.F.Craver,L.Darden,G.M.Mikkelson,andR.C.Richardson,Editors.Inpress.PhilosophyAcrosstheLifeSciences.Cambridge,MA:MITPress.Sklar,L.1999.Thereduction(?)ofthermodynamicstostatisticalmechanics.PhilosophicalStudies95:187202.Skyrms,B.1994.Darwinmeetsthelogicofdecision:Correlationinevolutionarygametheory.PhilosophyofScience61:50328.1996.EvolutionoftheSocialContract.Cambridge:CambridgeUniversityPress.2004.TheStagHuntandtheEvolutionoftheSocialContract.Cambridge:CambridgeUniversityPress.Skyrms,B.,andJ.Alexander.1999.Bargainingwithneighbors:Isjusticecontagious?JournalofPhilosophy96:58898.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList489Slatkin,M.,andM.J.Wade.1978.Groupselectiononaquantitativecharacter.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica75:353134.Slobodkin,L.B.,andA.Rapoport.1974.Anoptimalstrategyofevolution.QuarterlyReviewofBiology49:181200.Smith,B.2002.Thefoundationsofcomputing.Computationalism:NewDirections.EditorM.Scheutz.Cambridge,MA:MITPress.Smith,B.,andJ.B.Wilson.1996.Aconsumersguidetoevennessindices.Oikas76:7082.Smith,E.A.2000.Threestylesintheevolutionaryanalysisofhumanbehavior.AdaptationandHumanBehavior:AnAnthropologicalPerspective.EditorsL.Cronk,N.Chagnon,andW.Irons,2746.NewYork:AldinedeGruyter.Smith,E.A.,MoniqueBergerhoffMulder,andKimHill.2001.Controversiesintheevolutionarysocialsciences:Aguidefortheperplexed.TrendsinEcologyandEvolution16:12835.Sneath,P.H.A.,andR.R.Sokal.1973.NumericalTaxonomy.SanFransisco:W.H.Freeman.Snyder,M.,andM.Gerstein.2003.Defininggenesinthegenomicsera.Science300,no.5617:25860.Sober,E.2001.CoreQuestionsinPhilosophy.3rded.Englewood-Cliffs,NJ:Prentice-Hall.1999.Instrumentalismrevisited.Critica31:339.1999.Themultiplerealizabilityargumentagainstreductionism.PhilosophyofScience66:54264.1995.Naturalselectionanddistributiveexplanation:AreplytoNeander.BritishJournalforthePhilosophyofScience46:38487.1984.TheNatureofSelection.Cambridge,MA:MIT.Press.2000.PhilosophyofBiology.2nded.Boulder,CO:WestviewPress.1988.ReconstructingthePast:Parsimony,Evolution,andInference.Cambridge,MA:MITPress.1998.Sixsayingsaboutadaptationism.ThePhilosophyofBiology.EditorsD.HullandM.Ruse.Oxford:OxfordUniversityPress.2001.Thetwofacesoffitness.ThinkingAboutEvolution:Historical,PhilosophicalandPoliticalPerspectives.EditorsR.S.Singh,C.B.Krimbas,D.B.Paus,andJ.Beatty,30921.NewYork:CambridgeUniversityPress.Sober,E.,andD.S.Wilson.1997.UntoOthers:TheEvolutionofAltruism.Cambridge,MA:HarvardUniversityPress.Sokal,R.R.,andP.H.A.Sneath.1963.PrinciplesofNumericalTaxonomy.SanFrancisco:W.H.Freeman.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n490ReferenceListSopher,B.1993.Alaboratoryanalysisofbargainingpowerinarandomultimatumgame.JournalofEconomicBehaviorandOrganization21,no.1:32434.Sorger,P.K.2005.Areductionistssystemsbiology.CurrentOpinioninCellBiology17:911.Soule,M.E.1985.Whatisconservationbiology?Bioscience35:72734.Stegman,U.2004.Thearbitrarinessofthegeneticcode.BiologyandPhilosophy19:20522.Stein,E.1999.TheMismeasureofDesire:TheScience,Theory,andEthicsofSexualOrientation.Oxford:OxfordUniversityPress.Stent,G.1968.Thatwasthemolecularbiologythatwas.Science160:39095.Stephens,C.2004.Selection,drift,andtheforcesofevolution.PhilosophyofScience71,no.4:55070.Sterelny,K.1999a.Bacteriaatthehightable.BiologyandPhilosophy14,no.3:45970.2003.Lastwillandtestament:StevenJ.GouldsTheStructureofEvolutionaryTheory.PhilosophyofScience70:25563.1999b.Speciesasecologicalmosaics.Species:NewInterdisciplinaryEssays.EditorR.A.Wilson,11938.Cambridge,MA:MITPress.2004.Symbiosis,evolvabilityandmodularity.ModularityinDevelopmentandEvolution.EditorsG.SchlosserandG.Wagner,490516.Chicago:UniversityofChicagoPress.Sterelny,K.,andP.E.Griffiths.1999.SexandDeath:AnIntroductiontoPhilosophyofBiology.Chicago:UniversityofChicagoPress.Sterelny,K.,andP.Kitcher.1988.Thereturnofthegene.JournalofPhilosophy85,no.7:33962.Sterelny,K.,K.Smith,andM.Dickison.1996.Theextendedreplicator.BiologyandPhilosophy11:377403.Stotz,K.2006.Withgeneslikethatwhoneedsanenvironment:Postgenomicsargumentfortheontogenyofinformation.PhilosophyofScience73,no.5.Stotz,K.,A.Bostanci,andP.E.Griffiths.2006.Trackingtheshifttopostgenomics.CommunityGenetics9,no.3:19096.Stotz,K.,andP.E.Griffiths.2004.Genes:Philosophicalanalysesputtothetest.HistoryandPhilosophyoftheLifeSciences26,no.1:528.Stotz,K.,P.E.Griffiths,andR.Knight.2004.Howscientistsconceptualisegenes:Anempiricalstudy.StudiesinHistory&PhilosophyofBiologicalandBiomedicalSciences35,no.4:64773.Strange,K.2005.Theendofnaivereductionism:Riseofsystemsbiologyorrenaissancephysiology?AmericanJournalofCellPhysiology288:96874.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList491Suppe,F.1989.TheSemanticConceptionofTheoriesandScientificRealism.Chicago:UniversityofIllinoisPress.Swenson,W.,D.S.Wilson,andR.Elias.2000.Artificialecosystemselection.ProceedingsoftheNationalAcademyofSciences97:911014.Symons,D.1992.OntheuseandmisuseofDarwinisminthestudyofhumanbehavior.TheAdaptedMind:EvolutionaryPsychologyandtheGenerationofCulture.EditorsJ.H.Barkow,L.Cosmides,andJ.Tooby,13759.NewYork:OxfordUniversityPress.Tabery,J.G.2004.Synthesizingactivitiesandinteractionsintheconceptofamechanism.PhilosophyofScience71:115.Takacs,D.1996.TheIdeaofBiodiversity:PhilosophiesofParadise.Baltimore:JohnsHopkinsUniversityPress.Tauber,A.,andS.Sarkar.1992.TheHumanGenomeProject:Hasblindreductionismgonetoofar?PerspectivesinBiologyandMedicine35:22035.Taylor,P.,andL.Jonker.1978.Evolutionarystablestrategiesandgamedynamics.MathematicalBiosciences40:14556.Templeton,A.R.2002.OutofAfricaagainandagain.Nature416:4551.Thagard,P.1999.HowScientistsExplainDisease.Princeton,NJ:PrincetonUniversityPress.2003.Pathwaystobiomedicaldiscovery.PhilosophyofScience70:23554.Thaler,R.1988.Anomolies:Theultimatumgame.JournalofEconomicPerspectives2:195206.Thoday,J.M.1953.Componentsoffitness.SymposiafortheSocietyforExperimentalBiology,96114.Cambridge:CambridgeUniversityPress.Tilman,D.1999.DiversityandproductioninEuropeangrasslands.Science286:10991100.Tilman,D.,P.B.Reich,J.Knops,D.Wedin,T.Mielke,andC.Lehman.2001.Diversityandproductivityinalong-termgrasslandexperiment.Science294:84345.Tooby,J.,andL.Cosmides.1992.Thepsychologicalfoundationsofculture.TheAdaptedMind:EvolutionaryPsychologyandtheGenerationofCulture.EditorsJ.H.Barkow,L.Cosmides,andJ.Tooby,19136.NewYork:OxfordUniversityPress.Turner,D.2000.Thefunctionsoffossils:Inferenceandexplanationinfunctionalmorphology.StudiesinHistoryandPhilosophyofBiologicalandBiomedicalSciences31:193212.Tygart,C.E.2000.Geneticcausationattributionandpublicsupportofgayrights.InternationalJournalofPublicOpinionResearch12:25975.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n492ReferenceListUyenoyama,M.K.1979.Evolutionofaltruismundergroupselectioninlargeandsmallpopulationsinfluctuatingenvironments.TheoreticalPopulationBiology15:5885.Valentine,J.2004.OntheOriginofPhyla.Chicago:UniversityofChicagoPress.vanderWeele,C.1999.ImagesofDevelopment:EnvironmentalCausesinOntogeny.Albany,NY:SUNYPress.Vane-Wright,R.I.,C.J.Humphries,andP.H.Williams.1991.Whattoprotect?Systematicsandtheagonyofchoice.BiologicalConservation55:23554.vanOmmen,G.J.B.,E.Bakker,andJ.T.Dunnen.1999.TheHumanGenomeProjectandthefutureofdiagnostics,treatment,andprevention.Lancet(Supplement)354:S5S10.Vermeig,G.J.1995.Economics,volcanoes,andPhanerozoicrevolutions.Paleobiology21,no.2:12552.1987.EvolutionandEscalation.Princeton,NJ:PrincetonUniversityPress.1999.Inequalityandthedirectionalityofhistory.AmericanNaturalist153,no.3:24353.Vignaud,P.,etal.2002.GeologyandpalaeontologyoftheUpperMioceneToros-Menallahominidlocality,Chad.Nature418:15255.Vrba,E.1989.Levelsofselectionandsortingwithspecialreferencetothespeciesproblem.OxfordSurveysinEvolutionaryBiology6:11168.1983.Macroevolutionarytrends:Newperspectivesontherolesofadaptationandincidentaleffect.Science221:38789.1984.Whatisspeciesselection?SystematicZoology33:31828.Vrba,E.S.,andS.J.Gould.1986.Thehierarchicalexpansionofsortingandselection:Sortingandselectioncannotbeequated.Paleobiology12:21728.Wachtel,S.S.1983.H-YAntigenandtheBiologyofSexDetermination.NewYork:Grune&Stratton.Waddington,C.H.1940.OrganisersandGenes.Cambridge:CambridgeUniversityPress.Wade,M.J.1978.Acriticalreviewofthemodelsofgroupselection.QuarterlyReviewofBiology53:10114.1977.Anexperimentalstudyofgroupselection.Evolution31:13453.1980.Kinselection:Itscomponents.Science210:66567.1985.Softselection,hardselection,kinselection,andgroupselection.AmericanNaturalist125:6173.Wade,M.J.,andD.E.McCauley.1980.Groupselection:Thephenotypicandgenotypicdifferentiationofsmallpopulations.Evolution34:799812.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList493Wagner,G.P.2000.Whatisthepromiseofdevelopmentalevolution?PartI.Whyisdevelopmentalbiologynecessarytoexplainevolutionaryinnovations?JournalofExperimentalZoology288,no.2:9598.2001.Whatisthepromiseofdevolopmentalevolution?PartII.Acausalexplanationofevolutionaryinnovationsmaybeimpossible.JournalofExperimentalZoology291,no.4:30509.Wagner,G.P.,etal.2000.Developmentalevolutionasamechanisticscience:Theinferencefromdevelopmentalmechanismstoevolutionaryprocesses.AmericanZoologist40:81931.Wagner,G.P.,andL.Altenberg.1996.Complexadaptationsandtheevolutionofevolvability.Evolution50:96776.Wagner,G.P.,andH.C.Larsson.2003.Whatisthepromiseofdevelopmentalevolution?PartIII.Thecrucibleofdevelopmentalevolution.JournalofExperimentalZoology;PartB:MolecularandDevelopmentalEvolution300,no.1:14.Wagner,G.P.,andM.D.Laubichler.2004.RupertRiedlandthere-synthesisofevolutionarydevelopmentalbiology:Bodyplansandevolvability.JournalofExperimentalZoology;PartB:MolecularandDevelopmentalEvolution302,no.1:92102.Walsh,D.M.2000.Chasingshadows:naturalselectionandadaptation.StudiesinHistoryandPhilosophyofBiologicalandBiomedicalSciences31:13553.Walsh,D.M.,T.Lewens,andA.Ariew.2002.Thetrialsoflife:Naturalselectionandrandomdrift.PhilosophyofScience69,no.3:45273.Waters,C.K.2003.TheargumentsintheOriginofSpecies.CambridgeCompaniontoDarwin.EditorsJ.HodgeandG.Radick,11642.Cambridge:CambridgeUniversityPress.1998.Causalregularitiesinthebiologicalworldofcontingentdistributions.BiologyandPhilosophy13:536.1994.Genesmademolecular.PhilosophyofScience61:16385.2000.Moleculesmadebiological.RevueInternationaledePhilosophie4,no.214:53964.2004.Whatwasclassicalgenetics?StudiesinHistoryandPhilosophyofScience35,no.4:783909.1990.Whytheanti-reductionistconsensuswontsurvive:ThecaseofclassicalMendeliangenetics.PSA1990.Vol.1,12539.EastLansing,MI:PhilosophyofScienceAssociation.Waters,M.D.,andJ.M.Fostel.2004.Toxicogenomicsandsystemstoxicology:Aimsandprospects.NatureReviewsGenetics5:93648.Watson,J.D.,andF.H.C.Crick1953.Astructurefordeoxyribosenucleicacid.Nature171:73738.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n494ReferenceListWeber,M.2005.PhilosophyofExperimentalBiology.Cambridge:CambridgeUniversityPress.Weibull,J.1995.EvolutionaryGameTheory.Cambridge,MA:MITPress.Weikart,R.2004.FromDarwintoHitler:EvolutionaryEthics,Eugenics,andRacism.NewYork:PalgraveMacmillan.Weinrich,J.D.1987.Anewsociobiologicaltheoryofhomosexualityapplicabletosocietieswithuniversalmarriage.EthologyandSociobiology8:3747.Weisberg,M.Forthcoming.Whoisamodeler?BritishJournalforthePhilosophyofScience.Weitzman,M.L.1992.Ondiversity.QuarterlyJournalofEconomics107:363405.West-Eberhard,M.J.2003.DevelopmentalPlasticityandEvolution.Oxford:OxfordUniversityPress.Weston,A.D.,andL.Hood.2004.Systemsbiology,proteomics,andthefutureofhealthcare:Towardpredictive,preventative,andpersonalizedmedicine.JournalofProteomeResearch3:17996.Whewell,W.,andM.Ruse.2001.OfthePluralityofWorlds:AFacsimileoftheFirstEditionof1853;PlusPreviouslyUnpublishedMaterialExcisedbytheAuthorJustBeforetheBookWenttoPress;andWhewell’sDialogueRebuttingHisCritics,ReprintedfromtheSecondEdition.EditorM.Ruse.WithintroductorymaterialbyMichaelRuse.Chicago:UniversityofChicagoPress.White,L.Jr.1967.Thehistoricalrootsofourecologiccrisis.Science155,no.3767:120307.Whiten,A.2005.Thesecondinheritancesystemofchimpanzeesandhumans.Nature437:5255.Whiten,A.,J.Goodall,W.C.McGrew,T.Nishida,V.Reynolds,Y.Sugiyama,G.E.G.Tutin,R.W.Wrangham,andC.Boesch.1999.Culturesinchimpanzees.Nature399:68285.Whittaker,R.H.1975.CommunitiesandEcosystems.NewYork:Macmillan.1960.VegetationoftheSiskiyouMountains,OregonandCalifornia.EcologicalMonographs30:279338.Wilkins,A.S.2002.TheEvolutionofDevelopmentalPathways.Sunderland,MA:Sinauer.Williams,G.C.1966.AdaptationandNaturalSelection.Princeton,NJ:PrincetonUniversityPress.1985.Adefenseofreductionisminevolutionarybiology.OxfordSurveysinEvolutionaryBiology2:127.1992.NaturalSelection:Domains,Levels,andChallenges.Oxford:OxfordUniversityPress.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\nReferenceList495Williams,P.A.2001.DoingWithoutAdamandEve:SociobiologyandOriginalSin.Minneapolis:AugsburgFortressPublishers.Wilmut,I.,K.Campbell,andC.Tudge.2000.TheSecondCreation:DollyandtheAgeofBiologicalControl.NewYork:Farrar,Straus,Giroux.Wilson,D.S.,andR.K.Colwell.1981.Evolutionofsexratioinstructureddemes.Evolution35:88297.Wilson,E.O.1997.Introduction.BiodiversityII.EditorsM.L.Reaka-Kudla,D.E.Wilson,andE.O.Wilson,13.Washington,DC:JosephHenryPress.1978.OnHumanNature.Cambridge,MA:HarvardUniversityPress.1975.Sociobiology:TheNewSynthesis.Cambridge,MA:HarvardUniversityPress.Wilson,E.O.,Editor.1988.BioDiversity.Washington,DC:NationalAcademyPress.Wilson,E.O.,andF.M.Peter,Editors.1986.Biodiversity.Washington,DC:NationalAcademyPress.Wimsatt,W.1987.Falsemodelsasmeanstotruertheories.NeutralModelsinBiology.EditorsM.NiteckiandA.Hoffman,2355.Oxford:OxfordUniversityPress.1980a.Reductionisticresearchstrategiesandtheirbiasesintheunitsofselectioncontroversy.ScientificDiscovery:CaseStudies.EditorT.Nickles,21859.Dordrecht:Reidel.1981.Robustness,reliability,andoverdetermination.ScientificInquiryandtheSocialSciences.EditorsM.BrewerandB.Collins,12463.SanFrancisco:Jossey-Bass.1980b.Unitsofselectionandthestructureofthemulti-levelgenome.ProceedingsofthePhilosophyofScienceAssociation2:12283.Wolpoff,M.H.,J.Hawks,D.W.Frayer,andK.Hunley.2001.Modernhumanancestryattheperipheries:Atestofthereplacementtheory.Science291:29397.Woodward,J.1989.Dataandphenomena.Synthese79:393472.Wright,L.1973.Functions.PhilosophicalReview82:13968.1976.TeleologicalExplanations:AnEtiologicalAnalysisofGoalsandFunctions.Berkeley:UniversityofCaliforniaPress.Wright,R.T.1989.BiologyThroughtheEyesofFaith.SanFrancisco:HarperPublishers.Wright,S.1968.EvolutionandtheGeneticsofPopulations:ATreatise.4Vols.Chicago:UniversityofChicagoPress.1929.EvolutioninaMendelianPopulation.AnatomicalRecord44:287.1931.EvolutioninMendelianPopulations.Genetics16:97159.1980.Genicandorganismicevolution.Evolution34:82543.1943.Isolationbydistance.Genetics28:11438.CambridgeCollectionsOnline©CambridgeUniversityPress,2008\n496ReferenceList1948.Ontherolesofdirectedandrandomchangesingenefrequencyinthegeneticsofpopulations.Evolution2:27994.1945.Tempoandmodeinevolution:Acriticalreview.Ecology26:41519.Wulff,P.,andW.Wisden.2005.Dissectingneuralcircuitrybycombininggeneticsandpharmacology.TrendsinNeurosciences28:4450.Yang,Z.,andJ.P.Bielawski.2000.Statisticalmethodsfordetectingmolecularadaptation.TrendsinEcologyandAdaptation15:496503.Zeyl,D.2006.PlatosTimeaus.Availableathttp://plato.stanford.edu/entries/plato-timaeus/.CambridgeCollectionsOnline©CambridgeUniversityPress,2008