资源描述：
《The Status of Moral Emotions in Consequentialist Moral Reasoning英文资料》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
TheStatusofMoralEmotionsinConsequentialistMoralReasoningByRobertH.FrankCornellUniversityPresentedat“FreeEnterprise:ValuesinAction”GruterInstitute/JohnTempletonFoundation/UCLA-SloanResearchProgram4thWorkingConferenceatHarvardBusinessSchoolJune20-21,2006 2TheStatusofMoralEmotionsinConsequentialistMoralReasoningby1RobertH.FrankThephilosopherBernardWilliamsdescribesanexampleinwhichabotanistwandersintoavillageinthejunglewhereteninnocentpeopleareabouttobeshot.Heistoldthatnineofthemwillbesparedifhehimselfwillshootthetenth.Whatshouldthebotanistdo?Althoughmostpeoplewouldprefertoseeonlyoneinnocentpersondieratherthanten,Williamsarguesthatkillingtheinnocentvillagerisnotbyanymeansthe2obviouslyrightthingforJimtodo.Andmostpeopleseemtoagree.Theforceoftheexampleisitsappealtoawidelysharedmoralintuition.Yetsomephilosopherscounterthatitisthepresumedvalidityofmoralintuitionsthatsuchexamplescallintoquestion(Singer,2002).Theseconsequentialistsinsistthatwhetheranactionismorallyrightdependsonlyonitsconsequences.Therightchoice,theyargue,isalwaystheonethatleadstothebestoverallconsequences.Iwillarguethatconsequentialistsmakeapersuasivecasethatmoralintuitionsarebestignoredinatleastsomespecificcases.Butmanyconsequentialistsappeartotakethestrongerpositionthatmoralintuitionsshouldplaynoroleinmoralchoice.Iwillargueagainstthatpositiononthegroundsthatshouldappealtotheirwayofthinking.AsIwillattempttoexplain,ignoringmoralintuitionswouldleadtoundesirableconsequences.Mybroaderaimistoexpandtheconsequentialistframeworktotakeexplicitaccountofmoralsentiments.1HenriettaJohnsonLouisProfessorofManagementandProfessorofEconomics,JohnsonGraduateSchoolofManagement,CornellUniversity,Ithaca,NY14853.2Williams,1973. 3Consequentialistvs.DeontologicalMoralTheoriesConsequentialismdiffersfromtraditional,ordeontological,moraltheories,whichholdthattherightchoicemustbeidentifiedonthebasisofunderlyingmoralprinciples.Theseprinciplesmayspringfromreligioustradition(forexample,theTenCommandments)butneednot(forexample,Kant’scategoricalimperative).Whatevertheirsource,themoralforceoftheprinciplesinvokedbydeontologistsincreaseswiththeextenttowhichtheseprinciplesaccordwithstronglyheldmoralintuitions.Formany,perhapseventheoverwhelmingmajority,ofcases,consequentialistanddeontologicalmoraltheoriesyieldthesameprescriptions.Bothcamps,forexample,holdthatitwaswrongforEnronexecutivestohaveliedabouttheircompany’searningsandwrongforDavidBerkowitztohavemurderedsixinnocentpeople.Evenincasesinwhichtheremightappeartobeampleroomfordisagreementaboutwhatconstitutesmoralbehavior,amajorityofpractitionersfrombothcampsoftentakethesameside.Consider,forexample,avariantofthefamiliartrolley-carproblemdiscussedbyphilosophers.Youarestandingbyarailroadtrackwhenyouseeanout-of-controltrolleycarabouttostrikeagroupoffivepeoplestandingonthetracksahead.Youcanthrowanearbyswitch,divertingthetrolleyontoasidetrack,whichwouldresultinthedeathofonepersonstandingthere.Failuretothrowtheswitchwillresultinallfivepersonsbeingkilledonthemaintrack.Consequentialistsarevirtuallyunanimousinconcludingthatthemorallycorrectchoiceisforyoutothrowtheswitch.Somedeontologistsequivocate,arguingthattheactivestepofthrowingtheswitchwouldmakeyouguiltyofkillingthepersonontheside 4track,whereasyouwouldnotbeguiltyofkillingthefiveonthemaintrackifyoufailedtointervene.Yetevenmostdeontologistsconcludethatthedistinctionbetweenactandomissionisnotmorallyrelevantinthisexample,andthatyourbestavailablechoiceistothrowtheswitch.Buteventhoughthetwomoralframeworksexhibitbroadagreementwithrespecttotheethicalchoicesweconfrontinpractice,manydeontologistsremaindeeplyhostiletotheconsequentialistframework.TheStatusofMoralIntuitionsCriticsoftenattackconsequentialistmoraltheoriesbyconstructingexamplesinwhichthechoicethatconsequentialismseemstoprescribeviolatesstronglyheldmoralintuitions.Inanotherversionofthetrolley-carproblem,forexample,thetrolleyisagainabouttokillfivepeople,butthistimeyouarenotstandingnearthetracksbutonafootbridgeabovethem.Thereisnoswitchyoucanthrowtodivertthetrain.Butthereisalargestrangerstandingnexttoyou,andifyoupushhimoffthebridgeontothetracksbelow,hisbodywillderailthetrolley,intheprocesskillinghimbutsparingthelivesofthefivestrangers.(Itwon’tworkforyoutojumpdownontothetracksyourself,becauseyouaretoosmalltoderailthetrolley.)Consequentialismseemstoprescribepushingthelargestrangerfromthebridge,sincethiswouldresultinanetsavingsoffourlives.Yetwhenpeopleareaskedwhattheythinkshouldbedoneinthissituation,mostfeelstronglythatitwouldbewrongtopushthestrangertohisdeath.Thosewhosharethisintuitionarenaturallysympathetictothedeontologists’claimthattheexamplesomehowdemonstratesafundamentalflawin 5theconsequentialistposition.ThisversionofthetrolleyproblemthuselicitsessentiallythesamemoraljudgmentasBernardWilliams’sexampleinvolvingthebotanist.Manyconsequentialists,PrincetonphilosopherPeterSingeramongthem,questionthevalidityofthemoralintuitionsevokedbysuchexamples(Singer,2002).Toillustrate,Singerasksustoimagineanothervariantofthetrolleyproblem,onethatisidenticaltothefirstexceptforonedetail.Youcanthrowaswitchthatwilldivertthetrainnotontoasidetrack,butontoaloopthatcirclesbackontothemaintrack.Standingontheloopisalargestrangerwhosebodywouldbringthetrolleytoahaltifitweredivertedontotheloop.Singernotesthatthistimemostpeoplesaythattherightchoiceistodivertthetrolley,justasintheoriginalexampleinwhichtheswitchdivertedthetrolleyontoasidetrackratherthanaloop.Inbothcases,throwingtheswitchcausedthedeathofonestranger,intheprocesssparingthelivesofthefiveothersonthemain3track.Singer’sPrincetoncolleagueJoshuaGreene,acognitiveneuroscientist,hassuggestedthatpeople’sintuitionsdifferinthesetwoexamplesnotbecausethemorallycorrectactiondiffers,butratherbecausetheactionthatresultsinthelargestranger’sdeathissomuchmorevividandpersonalinthefootbridgecasethaninthelooped-trackcase:Becausepeoplehavearobust,negativeemotionalresponsetothepersonalviolationproposedinthefootbridgecasetheyimmediatelysaythatit’swrong…Atthesametime,peoplefailtohaveastrongnegativeemotionalresponsetotherelativelyimpersonalviolationproposedintheoriginaltrolleycase,andthereforereverttothemostobviousmoralprinciple,“minimizeharm,”whichinturnleads3Thelooped-trackexamplesuggeststhatitwasnottheKantianprohibitionagainstusingpeoplemerelyasmeansthatexplainstheearlierreluctancetopushthestrangerfromthefootbridge,sincechoosingtothrowtheswitchinthelooped-trackexamplealsoentailsusingthestrangerasmerelyameanstosavetheotherfive.Intheoriginalexample,divertingthetrolleyontothesidetrackwouldhavesavedtheothersevenifthestrangerhadnotbeenonthesidetrack. 6themtosaythattheactionintheoriginalcaseispermissible.(Greene,2002,p.178.)Totestthisexplanation,Greenusedfunctionalmagneticresonanceimagingtoexamineactivitypatternsinthebrainsofsubjectsconfrontedwiththetwodecisions.Hispredictionwasthatactivitylevelsinbrainregionsassociatedwithemotionwouldbehigherwhensubjectsconsideredpushingthestrangerfromafootbridgethanwhentheyconsidereddivertingthetrolleyontotheloopedtrack.Healsoreasonedthattheminorityofsubjectswhofelttherightactionwastopushthestrangerfromthefootbridgewouldreachthatjudgmentonlyafterovercomingtheirinitialemotionalreactionstothecontrary.Thushealsopredictedthatthedecisionstakenbythesesubjectswouldtakelongerthanthosereachedbythemajoritywhothoughtitwrongtopushthestrangertohisdeath,andlongeraswellthanittookforthemtodecidewhattodointhelooped-trackexample.Eachofthesepredictionswasconfirmed.Isitmorallyrelevantthatthinkingaboutcausingsomeone’sdeathbypushinghimfromafootbridgeelicitsstrongeremotionsthanthinkingaboutcausinghisdeathbythrowingaswitch?PeterSingerarguesthatitisnot—thatthedifferenceisasimple,non-normativeconsequenceofourevolutionarypast.Undertheprimitive,small-groupconditionsunderwhichhumansevolved,heargues,theactofharmingothersalwaysentailedvividpersonalcontactatclosequarters.Onecouldnotcauseanother’sdeathbysimplythrowingaswitch.Soifitwasadaptivetobeemotionallyreluctanttoinflictharmonothers—surelyaplausiblepresumption—therelevantemotionsoughttobemuchmorelikelytobetriggeredbyvividpersonalassaultsthanbyabstractactionslikethrowingaswitch. 7Ahistoricalcaseinpointhelpshighlightthedistinction.ShortlyafterBritishintelligenceofficershadbrokenNaziencryptionschemesinWorldWarII,WinstonChurchillhadanopportunitytosparethelivesofBritishresidentsofCoventrybywarningthemofapendingbombingattack.Todoso,however,wouldhaverevealedtotheNazisthattheircodeshadbeenbroken.InthebeliefthatpreservingthesecretwouldsaveconsiderablymoreBritishlivesinthelongrun,ChurchillgaveCoventrynowarning,resultinginlargenumbersofpreventabledeaths.Itisdifficulttoimagineamorewrenchingdecision,andwecelebrateChurchill’smoralcourageinmakingit.ButitisalsoeasytoimaginethatChurchillwouldhavechosendifferentlyhaditbeennecessaryforhimpersonallytokilltheCoventry’sresidentsatclosequarters,ratherthanmerelytoallowtheirdeathsbyfailingtowarnthem.Singer’sclaimisthatwhilethisdifferenceisapredictableconsequenceofthewayinwhichnaturalselectionforgedouremotions,ithasnomoralsignificance.Insum,thefactthatconsequentialistmoraltheoriessometimesprescribeactionsthatconflictwithmoralintuitionscannot,byitself,betakenasevidenceagainstthesetheories.Moralintuitionsarecontingentreactionsshapedbythedetailsofourevolutionaryhistory.Oftentheywillnotberelevantforthemoralchoicesweconfronttoday.MoralSentimentsasCommitmentDevicesThefactthatitmightsometimesbebesttoignoremoralemotionsdoesnotimplythatitisalways,orevenusually,besttoignorethem.Ifwearetothinkclearlyabouttheroleofmoralemotionsinmoralchoice,wemustconsidertheproblemsthatthese 8emotionsweremoldedbynaturalselectiontosolve.Mostinterestingmoralquestionsconcernactionstheindividualwouldprefertotakeexceptforthepossibilityofcausingundueharmtoothers.Unbridledpursuitofself-interestoftenresultsinworseoutcomesforeveryone.Insuchsituations,aneffectivemoralsystemcurbsself-interestforthecommongood.ButasHandy,O’Hara,andSolomonhaveemphasizedelsewhereinthisvolume,moralsystemsmustnotonlyidentifywhichactionisright,theymustalsoprovidemotivesfortakingthataction.Thedifficultyofservingbothgoalsatonceisimmediatelyapparent.Humansevolvedinharshenvironmentsinwhichtheconsequencesoffailuretopursueself-interestwereoftensevere.Faminesandotherthreatstosurvivalwerecommon.Polygynywasalsothenorminearlyhumansocieties,whichformenmeantthatfailuretoachievehighrankensuredfailuretomarry.Underthecircumstances,itwasanobviouschallengetomotivateindividualstoforgoself-interestforthecommongood.Yetinstancesinwhichpeopleforgoself-interestareactuallyquitecommon,eventhoughthehumannervoussystemismuchthesametodayasitwastensofthousandsofyearsago.Forexample,althoughself-interestdictatesleavingnotipinrestaurantsyoudon’texpecttovisitagain,mostpeopletipataboutthesamerateatsuchrestaurantsasin4localrestaurants.Similarly,whensociologistsperformtheexperimentofdroppingwalletscontainingsmallamountsofcashonsidewalksinNewYork,abouthalfare5returnedbymailwiththecashintact.4BodvarssonandGibson,1994.Tippinginlocalrestaurantscanberationalizedasaself-interestedactivity:ifyoudon’ttipwell,youmightnotgetgoodservicethenexttime.Peopleresistthetemptationtostiffthewaiterbecausetheshadowofthefutureisstaringatthem5Hornstein,1976. 9TheFalklandsWarisanothergoodexample.TheBritishcouldhaveboughttheFalklandersout—givingeachfamily,say,acastleinScotlandandagenerouspensionforlife—forfarlessthanthecostofsendingtheirforcestoconfronttheArgentines.Insteadtheyincurredconsiderablecostintreasureandlives.YetfewintheUKopposedthedecisiontofightforthedesolateSouthAtlanticislands.OnecouldsaythatMargaretThatchergainedpoliticalbyrespondingasshedid,butthisbegsthequestionofwhyvoterspreferredretaliationtoinaction.Whenpressed,mostpeoplespeakintermsofthenation’spridebeingatstake.Peoplerescueothersindistressevenatgreatperiltothemselves;theydonatebonemarrowtostrangers.Suchactionsareintensionwiththeself-interestmodelfavoredbyeconomists.Theyseemtobemotivatedbymoralsentiments.Butwheredothesesentimentscomefrom?AdamSmithsaidthatmoralsentimentswereendowedinusbythecreatorforthegoodofsociety.Itistruethatsocietyworksbetterifmoralsentimentsmotivatepeopletoexerciserestraint.ButasDarwinemphasized,selectionpressuresaregenerallyfarweakeratthesocietylevelthanattheleveloftheindividualorganism.Moralsentimentsoftenmotivatepeopletoincurcoststhattheycouldavoid.Onwhatbasismightthesesentimentshavebeenfavoredbynaturalselection?Inmy1988book,PassionsWithinReason,IproposedamechanismbasedonTomSchelling’sworkonthedifficultiespeoplefacewhenconfrontedwithcommitment6problems.Heillustratesthebasicideawithanexampleofakidnapperwhoseizesavictimandthengetscoldfeet.Thekidnapperwantstosetthevictimfreebutknowsthat,oncefreed,thevictimwillrevealthekidnapper’sidentitytothepolice.Sothekidnapper6Schelling,1960. 10reluctantlydecideshemustkillthevictim.Indesperation,thevictimpromisesnottogotothepolice.Theproblemisthatbothknowthatonceheisoutthedoor,hismotiveforkeepingthatpromisewillvanish.Schellingsuggeststhefollowingingenioussolution:Ifthereissomeevidenceofacrimethatthevictimhascommitted,hecansharethatevidencewiththekidnapper,whichwillcreateabondensuringhissilence.Theevidenceofthevictim’scrimeisacommitmentdevicethatmakesanotherwiseemptypromisecredible.Schelling’sbasicinsightcanbeextendedtoshowwhyatrustworthypersonmightbeabletoprospereveninhighlycompetitivemarketsettings.Supposeyouhaveabusinessthatisdoingsowellthatyouknowitalsowouldthriveinasimilartown300milesaway.Theproblemisthatbecauseyoucan’tmonitorthemanagerwhowouldrunthisbusiness,hewouldbefreetocheatyou.Supposeamanagerialcandidatepromisestomanagehonestly.Youmustthendecidewhethertoopenthebranchoutlet.Ifyoudoandyouremployeemanageshonestly,youeachcomeoutverywell—say,$1,000eachbetterthanthestatusquo.Butifthemanagercheats,youwilllose$500onthedeal,andhewillgain$1,500relativetothestatusquo.TherelevantpayoffsforthevariousoptionsarethusassummarizedinFigure1.(NotethatthesepayoffsdefineatrustgameliketheonesdiscussedelsewhereinthisvolumebySchwabandOstromandbyBergstrom,Kerr,andLachman.) 11Figure1.TheBranch-OutletProblemIfyouopentheoutlet,themanagerfindshimselfonthetopbranchofthedecisiontree,wherehefacesachoicebetweencheatingandnot.Ifhecheatshispayoffis$1,500;ifnot,hispayoffisonly$1,000.Standardrationalchoicemodelsassumethatmanagersinthesesituationswillbeself-interestedinthenarrowsense.Ifthatisyourbelief,youpredictthatthemanagerwillcheat,whichmeansyourpayoffwillbe-$500.Andsincethatisworsethanthepayoffofzeroyouwouldgetifyoudidn’topenthebranchoutlet,yourbestbetisnottoopenit.Thepityisthatthismeansalosstobothyouandthemanagerrelativetowhatcouldhavebeenachievedhadyouopenedthebranchoutletandthemanagerrunithonestly.Nowsupposeyoucanidentifyamanagerialcandidatewhowouldbewillingtopay$10,000toavoidtheguilthewouldfeelifhecheatedyou.Needlesstosay,usingafinancialpenaltyasaproxyforguiltfeelingswouldbeinappropriateinnormativediscourse.Wewouldnotsay,forexample,thatit’sOKtocheataslongasyougain 12enoughtocompensatefortheresultingfeelingsofguilt.Buttheformulationdoesnonethelesscaptureanimportantelementofbehavior.Incentivesmatter,andpeoplearelesslikelytocheatwhenthepenaltiesarehigher.Inanyevent,itisclearhowthissimplechangetransformstheoutcomeofthegame.Ifthemanagercheats,hispayoffisnot$1,500but–$8,500(afterthe$10,000psychologicalburdenofcheatingisdeducted).Soifyouopenthebranchoutlet,themanagerwillchoosetomanagehonestly,andbothheandyoucomeoutahead.Ifyoucouldidentifyatrustworthymanagerinthissituation,heorshewouldnotbeatadisadvantage.Onthecontrary,bothyouandthatmanagerwouldclearlyprofit.Figure2.TheBranch-OutletProblemwithanHonestManagerNote,however,thatthemanagerialcandidatewon’tbehiredunlesshistasteforhonestyisobservable.Thusanhonestcandidatewhoisnotbelievedtobehonestfares 13worsethanadishonestcandidateisbelievedtobehonest.Thefirstdoesn’tevengethired.Thesecondnotonlygetsthejobbutalsothefruitsofcheatingtheowner.Imagineamutationthatcausedtrustworthypeopletobebornwithanidentifyingmark,suchasa‘c’ontheirforeheads(for‘cooperator’).Thentheproblemwouldbesolved.Insuchaworld,thetrustworthytypeswoulddrivetheuntrustworthytypestoextinction.Ifthetwotypeswerecostlesslydistinguishable,theonlyequilibriumwouldbe7onewithpuretrustworthytypesinthepopulation.Ingeneral,though,itiscostlytofigureoutwhoistrustworthy.Sopeoplewouldnotvigilantintheirchoiceoftradingpartnersinanenvironmentinwhicheveryonewastrustworthy.Beingvigilantwouldnotpay,justasbuyinganexpensivesecuritysystemforyourapartmentwouldnotpayifyoulivedinaneighbourhoodinwhichtherehadneverbeenaburglary.Thusapopulationconsistingexclusivelyoftrustworthytypescouldnotbeanevolutionarilystableequilibrium.Givenreducedlevelsvigilance,untrustworthytypescouldeasilyinvadesuchapopulation.Soifcharactertraitsarecostlytoobserve,theonlysustainableequilibriumisoneinwhichthereisamixedpopulationconsistingofbothhonestanddishonestindividuals.Howmightasignaloftrustworthinesshaveemergedinthefirstplace?Evenifthefirsttrustworthypersonboresomeobservablemarker,nooneelsewouldhavehadanyideawhatitmeant.NicoTinbergenarguedthatasignalofanytraitmustoriginate8completelybyhappenstance.Thatis,ifatraitisaccompaniedbyanobservablemarker,thelinkbetweenthetraitandthemarkerhadtohaveoriginatedbychance.Forexample,thedungbeetleescapespredatorsbyresemblingthedungonwhichitfeeds.Howdidit7Frank,1988,chapter3.8Tinbergen,1952. 14gettolooklikethis?Unlessitjusthappenedtolookenoughlikeafragmentofdungtohavefooledthemostnear-sightedpredator,thefirststeptowardamoredunglikeappearancecouldn’thavebeenfavouredbynaturalselection.AsStephenJayGould9asked,“Cantherebeanyadvantageinlooking5percentlikeaturd?”Theproblemisthatnopredatorwouldbefooled.Sohowisthethresholdlevelofresemblancereached?Itmustbeginwithapurelyaccidentallinkbetweenappearanceandsurroundings.Butoncesuchalinkexists,thenselectioncanbegintoshapeappearancesystematically.Similarly,wemayask,“Howcouldamoralsentimenthaveemergedifnooneinitiallyknewthesignificanceofitsaccompanyingmarker?”Onehypothesisissuggestedbythelogicoftheiteratedprisoner’sdilemma.Thereisnodifficultyexplainingwhya10self-interestedpersonwouldcooperateinaniteratedprisoner’sdilemma.Ifyouareatit-for-tatplayer,forexample,andhappentopairwithanothersuchplayeronthefirst11round,youandthatotherplayerwillenjoyanendlessstringofmutualcooperation.Forthisreason,evenAttilatheHun,lackinganymoralsentiments,wouldwanttocooperateonthefirstmoveofaniteratedprisoner’sdilemma.Theproblemisthatifyoucooperateonthefirstmove,youforgosomegaininthepresentmoment,sincedefectiononanyiterationalwaysyieldsahigherpayoffthancooperation.Itiswellknownthatbothhumansandotheranimalstendtofavoursmallimmediaterewardsoverevenmuchlarger12long-termrewards.So,eventhoughyouexpecttomorethanrecouptheimmediatesacrificeassociatedwithcooperation,youmaydiscountthosefuturegainsexcessively.Successfulcooperation,inshort,requiresself-control.9Gould,1977,p.104.10Frank,1988,chapter4.11RapoportandChammah,1965.12Ainslie,1992. 15Ifyouwereendowedwithamoralsentimentthatmadeyoufeelbadwhenyoucheatedyourpartner,evenifnoonecouldseethatyouhadthatsentiment,thiswouldmakeyoubetterabletoresistthetemptationtocheatinthefirstround.Andthat,inturn,wouldenableyougenerateareputationforbeingacooperativeperson,whichwouldbeclearlytoyouradvantage.Moralemotionsmaythushaveoriginatedasimpulse-controldevices.ThisinterpretationaccordswithobservationselsewhereinthisvolumebyBergstrom,Kerr,andLachman,whoargue,ineffect,thataperson’swillingnessto“waste”timeinsocialrelationshipsmayserveasacommitmentdevice.Intheiraccount,willingnesstowastetimewouldbearelativelycostlystepfordefectors,whowouldbeforcedtoseekotherrelationshipsanewifdiscoveredcheating.Myargumentsuggestsacomplementaryinterpretation:Aninclinationtospendseeminglyunproductivetimeinsocialrelationshipsmaybealsobeproductivebecauseitsignalscapacitiestoexperiencesympathyorempathy,whichalsomakecheatingmorecostly.Inanyevent,theactivationoftheseemotions,likeotherformsofbrainactivation,maybeaccompaniedbyinvoluntaryexternalsymptomsthatareobservable.Ifso,theobservablesymptomsovertimecouldhavebecomeassociatedinothers’mindswiththepresenceofthesemoralsentiments.Andoncethatassociationwasrecognized,themoralemotionswouldbeabletoplayasecondrole—namely,thatofhelpingpeoplesolveone-shotprisoner’sdilemmasandothercommitmentproblems.Thesymptomsthemselvescanthenbefurtherrefinedbynaturalselectionbecauseoftheircapacitytohelpidentifyreliablepartnersinone-shotdilemmas. 16Howdoyoucommunicatesomethingtoanotherindividualwhohasreasontobeskepticalofwhatyouaresaying?Suppose,forexample,thatatoadmeetsarivalandbothwantthesamemate.Amonganimalsgenerally,thesmalleroftworivalsdeferstothelarger,therebyavoidingacostlyfightthathewouldhavebeenlikelytoloseanyway.Rivaltoads,however,oftenencounteroneanotheratnight,makingvisualassessmentdifficult.Whattheydoiscroakatoneanother,andthetoadwiththehigher-pitchedcroakdefers.Theideaisthattheloweryourcroak,thebiggeryouareonaverage.Soitisprudenttodefertothelowercroaker.Thisexampleillustratesthecostly-to-fakeprinciple:“I’llbelieveyounotbecauseyousayyouareabigtoad,butratherbecauseyou13areusingasignalthatisdifficulttopresentunlessyoureallyareabigtoad.”Itisthesamewhendogsfaceoff:theyseemtofollowanalgorithmofdeferringtothelargerdog.ConsiderthedrawingsinFigure3,takenfromCharlesDarwin’s1872book,TheExpressionofEmotioninManandAnimals.Theleftpanelportraysadogthatisconfrontingarival.Darwinarguedthatwereliablyinferwhatisgoingonemotionallyinthisdog’sbrainbyobservingthenumerouselementsofitsposturearesoserviceableinthecombatmode:Thedog’shacklesareraised,itsfangsarebared,itsearsarepricked,itseyeswideopenandalert,itsbodypoisedtospringforward.Darwinreasonedthatanydogthathadtogothroughaconsciouschecklisttomanifesttheseposturalelementsonebyonewouldbetooslowonthedrawtocompeteeffectivelyagainstarivalinwhomtheentireprocesswasactivatedautonomouslybytherelevantemotionalarousal.Thatautonomouslink,heconcluded,providesawindowintothedog’sbrain.13Frank,1988,chapter6. 17Figure3.ObservableCorrelatesofEmotionalStates14Darwinarguedthattherearesimilaremotionalsymptomsinhumans.Certainexpressions,forexample,springunbiddentothehumanfaceinthepresenceoftriggeringemotions,yetareextremelydifficulttopresentwhenthoseemotionsareabsent.PeopleraisedindifferentculturaltraditionsaroundtheworldcanreadilyidentifytheschematicexpressionportrayedinFigure4asonecorrespondingtoemotionssuchassadnessorconcern.AsPaulEkmanandhiscolleagueshaveshown,mostpeopleareunableto15reproducethisexpressiononcommand.Variousotheremotionsalsohavetheircharacteristicsignatures.14Darwin,1872.15Ekman,1985. 18Figure4.TheCharacteristicExpressionofSadnessorConcernIntheargumentIamattemptingtoadvance,emotionslikesympathyandempathyplayacentralrole.Otherauthorsinthisvolumehavealsostressedasimilarroletheseemotions.BrosnananddeWaal,forexample,reportthatprecursorsoftheseemotionsareclearlyvisibleinsomeprimatesandappeartomotivatesharing.Solomondiscussestheillustrioushistoryoftheseemotionsinmoraldiscourse,beginningwithDavidHumeandAdamSmith.AndZaknotestherelationshipbetweentheneuroactivehormoneoxytocinlevelsandtheexperienceofempathyandsympathy.Thoseofuswhofocusontheseemotionsstresstheirroleinmotivatingindividualstoforgogainindeferencetotheinterestsofothers.Myparticularconcerniswiththequestionofhowapersonmightidentifythepresenceoftheseemotionsinothers.How,inotherwords,canyoupredictwhethersomeone’ssympathyforyouwillpreventhimfromcheatingyou,evenwhenhehasanopportunitytodosowithno 19possibilityofbeingpunished?Clearlywouldnotsufficeforhimmerelytosaythathewassympathetictoyourinterests.Youneedamorereliablesignal.Inthebranch-outletproblemdiscussedearlier,forexample,mostpeoplewouldfeelmorecomfortablehiringanoldfriendthanaperfectstranger.Ihavebeentalkingthusfarasifthereweregoodpeopleandbadpeople,withsomemixtureofthesetwopuretypescomprisingtheentirepopulation.Lifeisofcoursemorecomplicated:wehavealldonesomethingdecent,andwehaveallcheatedatonepointoranother.Thequestionis,underwhatconditionsdowecheat?Evidencesuggeststhatwearefarlesslikelytocheatotherswithwhomweenjoystrongsympatheticbonds.Suchbondsappeartoformasaresultofacomplexdancethatplaysoutamong16peopleovertime.Whenyouhaveacommitmentproblemtosolve,youpicksomebodywhoyouthinkcaresaboutyou.PeoplearecriticalofGeorgeW.Bushforgivingjobstocronies.Butallleadersdothat,andforgoodreason.Bush’sparticularproblemhasbeenthatmanyofhiscronieswerenotcompetent.Theideathatyouwouldpicksomeonewellknowntoyouisintelligible:thisisasensiblethingtodo.Intheend,thequestioniswhetherwecanidentifywhowillcheatandwhowon’t.17TomGilovich,DennisReganandIhavedonesomeexperimentsonthis.Oursubjectshadconversationsingroupsofthreefor30minutes,attheendofwhichtimetheyplayedprisoner’sdilemmagameswitheachoftheirconversationpartners.Subjectsweresenttoseparateroomstofilloutformsonwhichtheyindicated,foreachpartner,whethertheyweregoingtocooperateordefect.Theyalsorecordedtheirpredictionsofwhateachpartnerwoulddowhenplayingwiththem.Eachsubject’spayoffwasthe16Forarichdescription,seeSally,2000.17Frank,Gilovich,andRegan,1993. 20calculatedasthesumofthepayoffsfromtherelevantcellsofthetwogames,plusarandomterm,sonooneknewafterthefactwhohaddonewhat.Almost74percentofthepeoplecooperatedinthesepureone-shotprisoner’sdilemmas.Thisfindingiscompletelyunpredictedbythestandardself-interestmodel.Butotherempiricalstudieshavealsofoundhighcooperationratesindilemmaswhen18subjectswereallowedtocommunicate.Ourparticularconcernwaswithwhethersubjectscouldpredicthoweachoftheirspecificpartnerswouldplay.Whensomeonepredictedthatapartnerwouldcooperatetherewasan81percentlikelihoodofcooperation(asopposedtothe74percentbaserate).Onthedefectionside,thebaseratewasjustover26percent,butpartnerswhowerepredictedtodefecthadadefectionrateofalmost57percent.Thisseemsanastonishinglygoodpredictiononthebasisofjust30minutesofinformalconversation.Thefollowingthoughtexperimentalsospeakstothequestionofwhetherpeoplecanmakeaccuratepredictionsaboutwhowillcheatthem.Imaginethatyouhavejustreturnedfromacrowdedconcerttodiscoverthatyouhavelost$1,000incash.Themoney,whichhadbeeninanenvelopewithyournameandaddressonit,apparentlyfellfromyourcoatpocketwhileyouwereattheconcert.Doyouknowanyonenotrelatedtoyoubybloodormarriagewhoyoufeelcertainwouldreturnyourmoney?Mostpeoplesaytheydo.Whatmakesthemfeelsoconfident?Notethatitisextremelyunlikelythattheyhaveexperiencedthissituationbefore.Buteveniftheyhad,ifthepersonnamedhadfoundtheirmoneyandkeptit,theywouldnothaveknownthat.Underthecircumstances,returningthemoneyisastrictcontradictionofthenarrowself-interestmodelfavoredbyeconomists.18SeeSally,1995. 21Manypeoplefinditnaturaltosaythattheactofreturningthemoneyinasituationlikethismustbemotivatedbysomesortofmoralemotion.Thus,peoplemightpredictthatafriendwouldreturntheirmoneybecausethefriendwouldfeelbadabouttheprospectofkeepingit.Howdidyoupickthepersonwhoyouthoughtwouldreturnyourmoneyifshefoundit?Typicallyitissomeonewithwhomyouhavedevelopedsympatheticbondsoveranextendedperiod.Thefeelingisthatyouknowenoughaboutthispersontosaythatifshefoundyourmoney,shewouldn’tfeelrightaboutkeepingit.Tosaythattrustworthinesscouldbeanevolutionarilystablestrategyisnottosaythateveryoneisprimedtocooperateallthetime.Opportunismofthesortpredictedbyself-interestmodelsisinabundantsupply.Yettheprospectsforsustainingcooperationarenotasbleakasmanyeconomistsseemtothink.Manypeoplearewillingtosetasideself-interesttopromotethecommongood.Evenifmoralemotionsareunobservablebyothers,theycanstillhelpyoutobepatientinrepeatedprisoner’sdilemmas.Butifothersrecognizeyoutobeadecentperson,thereareallsortsofwaysinwhichyouarevaluable.Ifyouareinbusiness,yourbossislikelytohaveafirmopinionaboutwhetheryou’dbethesortofpersontoreturnthelost$1,000ifyoufoundit.You’dlikehimtothinkthatyou’dreturnit.Thebestwaytogethimtothinkthat,itappears,istoactuallybethekindofpersonwhowouldreturnit.DoOurModelsofHumanNatureMatter?AsGintisandKhaurananoteelsewhereinthisvolume,neoclassicaleconomicmodelstypicallyassumethatpeopleareself-interestedinthenarrowsense.Yetabundant 22evidencesuggeststhatmotivesotherthanself-interestarealsoimportant.Anobviousconsequenceofinaccuratebehavioralassumptionsisthattheyoftenleadtoinaccuratepredictions.Butintheirpapersinthisvolume,CasebeerandStoutnoteanotherworrisomepossibility—namely,thatinaccurateportrayalsofhumannaturemayproveself-reinforcing.Theself-interestmodelofrationalchoicepredictsthatpeoplewilldefectinone-shotprisoner’sdilemmas.Doesworkingwiththatmodeloverthecourseofmanyyears,asprofessionaleconomistsdo,altertheirexpectationsaboutwhatotherswilldoinsocialdilemmas?Andifso,doesthisalterhoweconomiststhemselvesbehavewhenconfrontedwithsuchdilemmas?TomGilovichandDennisReganandIfoundthateconomicstraining—bothitsdurationandcontent—affectsthelikelihoodthat19undergraduatestudentswilldefectinprisoner’sdilemmagames.Inoneversionofourexperiments,economicsmajorswerealmosttwiceaslikelytodefectasnon-economicsmajors.Thisdifferencecouldsteminpartfromthefactthatpeoplewhoelecttostudyeconomicsweredifferentfromothersinthefirstplace.Butwealsofoundatleastweakevidencefortheexistenceofatrainingeffect.Thedifferencesincooperationratesbetweeneconomicsmajorsandnon-majorsincreasedwiththelengthoftimethatstudentshadbeenenrolledinthemajor.Wealsofoundthat,relativetoacontrolgroupoffreshmenastronomystudents,studentsinanintroductorymicroeconomicscourseweremorelikelyinDecemberthaninSeptembertoexpectopportunisticbehavioronthepartofothers.19Frank,Gilovich,andRegan,1993a;MarwellandAmes(1981)andCarterandIrons(1991)reportsimilarfindings. 23Mypointisnotthatmyfelloweconomistsarewrongtostresstheimportanceofself-interest.Butthosewhoinsistthatitistheonlyimportanthumanmotivearemissingsomethingimportant.Evenmoretroubling,thenarrowself-interestmodel,whichencouragesustoexpecttheworstinothers,maybringouttheworstinusaswell.DifficultiesConfrontingEmotion-FreeConsequentialismConsequentialistmoralsystemsthatignoremoralemotionsfacemultiplechallenges.Itisonethingtosaythatwewouldallenjoygreaterprosperityifwerefrainedfromcheatingoneanother.Butitisquiteanothertopersuadeindividualsnottocheatwhencheatingcannotbedetectedandpunished.Evenforpersonsstronglymotivatedtodotherightthing,consequentialistmoralsystemscansometimesmakeimpossibledemandsonindividuals.Imagine,forexample,thatfivestrangersareabouttobekilledbyarunawaytrolley,whichattheflipofaswitchyoucoulddivertontoasidetrackwhereitwouldkillfourofyourclosestfriends.Manyconsequentialistswouldarguethatitisyourmoraldutytofliptheswitch,sinceitisbetterthatonlyfourdieinsteadoffive.Butapersoncapableofheedingsuchadvicewouldhavebeenunlikelytohavehadanyclosefriendsinthefirstplace.Indeed,itiseasytoimaginethatmostpeoplewouldbecomemorereluctanttoformclosefriendshipsiftheybelievedittheirdutytoignoretheemotionalbondsthatsuchfriendshipsinevitablyentail. 24Thecapacitytoformdeepbondsofsympathyandaffectionisimportantforsolvingavarietyofcommitmentproblems.Itisnotacapacityeasilyabandoned.Andevenifwecouldabandonit,theemotionalandmaterialcostswouldbesubstantial.DoMoralEmotionsDefineRightConduct?Sinceourcurrentenvironmentdiffersinmanyimportantwaysfromtheenvironmentsinwhichourancestorsevolved,weshouldnotbesurprisedthatourintuitionssometimesmisleadusabouttoday’smoralquestions.ThuswehavenotjustSinger’sexampleofaninhibitionthatistoostrong(ourreluctancetopushthefatmanfromthebridgetosavethefivestrangers),butalsomanyothersinwhichourinhibitionsaretooweak(suchasthoseagainststealingfromcorporateemployers,filingoverstatedinsuranceclaims,orunderstatingourincomesfortaxpurposes).Inthelatterexamples,theweaknessofinhibitionisplausiblyexplainedbythefactthatintheenvironmentsinwhichweevolved,cheatingalwaysvictimizedspecificpersonsratherthanfacelessinstitutions.Butourmoralintuitionsdon’talwaysmisleadus.Inthelost-envelopethoughtexperiment,forexample,mymisgivingsaboutkeepingmyfriend’scashwouldpushmetodowhatanindependentconsequentialistmoralanalysissaysIoughttodounderthecircumstances—namely,returnthecash.Indeed,ourintuitionsappeartoprovidesoundmoralguidancemoreoftenthannot.Forthisreason,takingthematfacevalueseemslikeareasonabledefaultoption,providedweremainopentothepossibilitythattheymaybemisleadinginspecificcases.Thatsaid,Imustemphasizethatmyargumentaboutthemoralemotionsin 25PassionsWithinReasonwasintendedtoserveonepurposeonly—toexplainhowpeoplewhoevolvedunderthepressuresofnaturalselectionmightnonethelesshaveinheritedmotivationstodotherightthingundersomecircumstancesinwhichsuchconductentailedavoidablecosts.Ineverclaimedthatourintuitionsdefinerightconduct.Asnoted,theonlyequilibriathataresustainableintheevolutionarygamesIdiscussentailpopulationscontainingatleastsomeindividualswholacktheusualmoralinhibitions.Wemustincurcoststoengageincharacterassessment,anditwouldmakenosensetoincurthesecostsifeverybodywereinclinedtodotherightthingallthetime.Butifpeoplewerenevervigilantwhenchoosingtheirtradingpartners,thenmutantcheaterscouldinvadeanhonestpopulationatwill.Andsinceanypopulationmustthereforecontainatleastsomecheatersinequilibrium,agivenindividual’smoralintuitionssimplycannotbeusedtodefinewhatconstitutesrightconduct.Thatassessmentrequiresanindependentanalysisbasedonsomesortofmoraltheory.Areourmoralintuitionsrelevanttothechoiceofwhichmoraltheorytoemploy?Insomecases,absolutelyyes.Butinatleastsomeothers,IagreewithSingerthatwemustbepreparedtoembraceamoraltheoryeventhoughitmightconflictwithaspecificmoralintuitionweholddear.IfIreadhimcorrectly,however,hegoestoofarinclaimingthatmoralintuitionsshouldplaynoroleatallinmoraljudgment—eitherinchoosingamongmoraltheoriesorinperformingmoralanalysiswithintheframeworkofanygiventheory.Sinceourmoralintuitionsareinharmonywithourmoraltheoriesmostofthetime,thisclaimseemsstrangeonitsface.(Itisofcourseconsistentwithhiscontrariannature!)Singer’spoint,though,isthattheapparentharmonyislessinformativethanitseems,becausethe 26authorsofmoraltheoriesconsciouslystrivetomakethemconsistentwithourintuitions.Fairenough,butthatclearlydoesnotimplythatmoralintuitionsaregenerallyirrelevant.Onthecontrary,sincetheyappeartoprovideusefulguidancemoreoftenthannot,weshouldbepreparedtoofferacoherentaccountforwhyagivenintuitionismisleadingbeforeproposingtodismissit.Thatstrategyworksjustfineinspecificcases.Forinstance,itseemsplausibletoexplainourrelativelackofinhibitionagainstusingweaponsthatkillatgreatdistances(and,byextension,ourlackofinhibitionaboutkillingastrangerbyflippingatrolleyswitch)bysayingthatkillinginsuchremotewayssimplywasn’tpossibleintheancestralenvironment.Buttosayweshoulddisregardmoralemotionsgenerally,onewouldhavetoofferasimilarargumentagainsteachofthem.Andthis,Ibelieve,Singercannotdo.ItisforthisreasonthatIfindhimunpersuasivewhenheinsiststhatitisnecessarilybettertosavetwostrangersthanasinglefriend.Hisclaimviolatesastronglyheldintuition,butthistimeitisonethatcannoteasilybeshowntobemisleading.ConcludingRemarksInbrief,IhavearguedforamiddlegroundbetweenandSinger’spositionandthatofJohnRawls,whoarguedthatprogressinmoraltheoryresultsfromeffortstoreconcile20ourmoraltheorieswithourmoralintuitions.InsofarasIbelievethatamoraltheoryislikelytobejudgedunacceptableifitsystematicallyviolatesourmoralintuitions,IammoreorlessonRawls’sside.ButwithSinger,Iampreparedtoembraceamoraltheorythatviolatesaspecificmoralintuitionifaplausibleaccountcanbegivenforwhythatintuitionismisleading.20Rawls,1971. 27ReferencesAinslie,George(1992).Picoeconomics,NewYork:CambridgeUniversityPress.Bodvarsson,O.B.andW.A.Gibson(1994).Gratuitiesandcustomerappraisalofservice:evidencefromMinnesotarestaurants,”JournalofSocioeconomics,23,1994:287-302.Carter,JohnandMichaelIrons(1991)."AreEconomistsDifferent,andIfSo,Why?"JournalofEconomicPerspective,5,Spring.Darwin,Charles(1965;1872).TheExpressionofEmotionsinManandAnimals.Chicago:UniversityofChicagoPress.Ekman,Paul(1985).TellingLies.NewYork:W.W.Norton.Frank,RobertH.(1988).PassionsWithinReason.NewYork:W.W.Norton.Frank,RobertH.,Gilovich,Thomas,andRegan,Dennis(1993a).Doesstudyingeconomicsinhibitcooperation?JournalofEconomicPerspectives,7,Spring,159-171.________________________(1993b).Theevolutionofone-shotcooperation.EthologyandSociobiology,14,July,247-256.Gould,StephenJay(1977).EverSinceDarwin.NewYork:W.W.Norton.Greene,Joshua.TheTerrible,Horrible,NoGood,VeryBadTruthaboutMorality,andWhattoDoAboutIt,DepartmentofPhilosophy,PrincetonUniversity,2002.GreeneJ.D.,R.B.Sommerville,L.E.Nystrom,J.M.Darley,andJ.D.Cohen(2001).“AnfMRIInvestigationofEmotionalEngagementinMoralJudgment”Science,293(5537),2105-8.Hornstein,Harvey(1976).CrueltyandKindness,EnglewoodCliffs,NJ:PrenticeHall.Hume,David(1978;1740).ATreatiseofHumanNature.Oxford:OxfordUniversityPress.Marwell,GeraldandRuthAmes(1981)."EconomistsFreeRide,DoesAnyoneElse?"JournalofPublicEconomics15:295-310.Rapoport,AnatolandA.Chammah(1965).Prisoner’sDilemma.AnnArbor:UniversityofMichiganPress.Rawls,John(1971).ATheoryofJustice,Cambridge,Massachusetts:HarvardBelknap,1971. 28Sally,David(1995).Conversationandcooperationinsocialdilemmas:Ameta-analysisofexperimentsfrom1958to1972.RationalityandSociety,7,58-92.____________(2000).Ageneraltheoryofsympathy,mind-reading,andsocialinteraction,withanapplicationtotheprisoners’dilemma.SocialScienceInformation,39(4),567-634.Schelling,ThomasC.(1960).TheStrategyofConflict,NewYork:OxfordUniversityPress.Singer,Peter.“TheNormativeSignificanceofOurGrowingUnderstandingofEthics,”paperpresentedattheOntologyConference,SanSebastian,Spain,October3,2002.Smith,Adam(1966;1759).TheTheoryofMoralSentiments.NewYork:Kelley.Tinbergen,Niko(1952).Derivedactivities:theircausation,biologicalsignificance,andemancipationduringevolution.”QuarterlyReviewofBiology27:1-32.Williams,Bernard(1973).“ACritiqueofUtilitarianism,”inJ.J.C.SmartandBernardWilliams,eds.,Utilitarianism:ForandAgainst,Cambridge,CambridgeUniversityPress.
