The Curse of Knowledge in Reasoning About False Beliefs

PSYCHOLOGICALSCIENCERese1andPaulBloom21UniversityofBritishColumbia,Vancouver,BritishColumbia,Canada,and2YaleUniversityABSTRACT—Assessingwhatotherpeophildrenfinditdiffi,beliefsthatconflictwithreality).Thesourceofthisdiffishowthatifsensitive-enoughmeasuresareused,adultssicular,,adults’ownknowledgeofanevent’soutcomecancompromisetheirabilitytoreasonaboutanotherperson’foundthatadults’findingsshedlightonthefactorsinvolvedinfalse-beliefreasoningandaredis-cussedinlightoftheirimplicationsforbothadults’andchildren’ingaboutwhatotherpeoplhofresearchhasinvestigatedchildren’sappreciationthatthemindcanmisrepresentreality—thatis,searchonchildren’sfa,Baron-Cohen,Leslie,&Frith,1985;Wimmer&Perner,1983).Forexample,subjectsaretoldastoryaboutSally,bsence,llyreturns,wherewillshelookforhercandy?Therightanswer,thatshewilllookinthebox,-year-oldstendtodofairlywellatsuchtasks,butyoungerchildrentendtofail(seeWellman,Cross,&Watson’s,2001,meta-analysis).Youngerchildrentendtoanswerinaccordwiththeirownknowledge,scorrespondencetoSusanBirch,DepartmentofPsychology,UniversityofBritishColumbia,2136WestMall,Vancouver,BC,CanadaV6T1Z4,e-mail:sbirch@rceofchildren’sdiffisearchersinterpretchildren’sdifficultiesonthesetasksasreflectingaconceptualdeficit:Perhapsyoungchildrenlac,Gopnik,1993;Perner,Leekam,&Wimmer,1987;Wellman,1990;Wellmanetal.,2001).Analternativeviewisthatyoungchildren’sproblemsareduetomoregeneralcognitivefactorssuchasmemoryandprocessinglimitations,,Fodor,1992;German&Leslie,2000;Leslie,1987;Onishi&Baillargeon,2005;Roth&Leslie,1998;Zait-chik,1990;foradiscussion,seeBloom&German,2000).Accordingtooneversionofthisalternative,childrenhavethesamebiasinperspectivetakingasadults,ierwork(Birch&Bloom,2003),weshowedthat3-and4-year-oldsaremoresusceptiblethan5-year-oldstoacognitivebiasfoundinadults,thecurseofknowledge(seealsoBernstein,Atance,Loftus,&Meltzoff,2004;Pohl&Haracic,2005).WeadoptedthistermfromCamerer,Loewenstein,andWeber(1989)anduseittorefertothetendencytobebiasedbyone’sowncurrentknowledgestatewhentryingtoappreciateamorenaiveperspective,whetherthatmorenaiveperspectiveisone’sownearlierperspective(asinthehindsightbias)orsomeoneelse’sperspective(seeBirch&Bernstein,2007,foradiscussion).Weproposed(Birch&Bloom,2003)thattheexaggerationofthebiasinyoungerchildrenmaypartiallyaccountforwhyyoungerchildrenexperiencegreaterdifficultythanolderchil-drenonfalse-belieftasks(seealsoBirch,2005;Birch&Bloom,2004).Ifthisiscorrect,thenonemightexpectadultstoalsoexperiencepricmimicsthatofDiamondandKirkham(2005),whoshowedthatwhensensitivemeasuresareused,evenadultshavedifficultyonaproblem(thedimen-sional-changecard-sorttask)thatisparticularlydiffildemphasize,however,thatsuchafindingofadultdifficultyinthedomainoffalsssible,forinstance,thatchildren’sproblemsareduetobothanexagger-atedcurse-of-knowledgebiasandconceptuallimitations.382Copyrightr2007AssociationforPsychologicalScienceVolume18—Number5

ndPaulBloomThereisawell-establishedliteratureonthisbiasinsocialandcognitivepsychology,spurredbytheworkofFischhoff(1975)and,morerecently,byagrowingbodyofliteraturelinkingthisbiastochildren’,Bernsteinetal.,2004;Keysar,Lin,&Barr,2003;Royzman,Cassidy,&Baron,2003).However,ithasnotyemarygoalofthepresentstudywastotestwhetheradults,whoundoubtedlydonothaveaconceptualdeficitinbeliefreasoning,findreasoningabouthis,wegaveadultsadisp,weusedamoresensitivemeasutswereaskedtoreporttheprobability,weusedfourcontainers,insteadoftwo,sowecouldmanipulatethesubjects’subjectsweretoldthatthetargetobject(aviolin),theknowledgeconditions)ortheywereto,theignorancecondition).Third,thecontainerswererearrangedfollowingdisplacementtoallowustomanipulatenipulationwasincludedbecauseoursecondobjectivewastodeterminehowtheperceivedplausibilityoftheprotagonist’sactions(fromthesubject’sperspective)mightinflusresearchhasdemonstratedthatplausibilitycanme-diatethemagnitudeofthecurseofknowledge(seePohl,1998;seealsoPezzo,2003,foradiscussionofthemediatingroleofsurprise).Forinstance,adultsaremorebiasedbytheiroutcomeknowledgewhentheoutcomeisbroughtaboutbecauseofaplausiblyforeseeabmple,Wasserman,Lempert,andHastie(1990)toldsomesubjectsthattheBritish-Gurkawarwasw,plausiblyforeseeable),buttoldot,lessforeseeable).,subjects’estimatesoftheprobabilitytheywouldhaveas-signedtotheoutcomehadtheynotknowntheBritishwon)estofourknowledge,noonehassystematicallyexploredtheinflresentexperiment,wemanipulatedtheperceivedplausibilityoftheoutcomeasfollows:Intheknowledge-plau-siblecondition,subjectsweretoldthattheviolinwasmovedtoadifferentcontainerthat,followingrearrangement,wasinthesamephysicallocationastheviolinhadbeenoriginally;thus,itwnowledge-implausiblecondition,subjectsweretoldthattheviolinwasmovedtoadifferentVolume18—Number5containerthat,followingrearrangement,ictedthatthecurseofknowledgewouldbestrongerforadultsiftheycouldconceiMENTMethodSubjectsOnehundredfifty-fivestudents(69male,82female,4un-specified)enrolledinanintroductorypsychologycourseatYaleUniversityinNewHaven,CT,tscompletedtalsandProcedureSubjectswererandomlyassignedtooneofthreeconditions:ignorance,knowledge-plausible,jectsreceivedthesamestimuli—acolorversionofFigure1(withoutthecolor-wordlabels).Thefirstpicturede-picntainerwasadifferentcolor:blue,purple,red,hthefirstpicturewasanimageofadifferentgirlholdingaviolin;inthispicture,tsinallconditionsread,‘‘fiickiisoutsideplaying,’’Atthispoint,theconditionsdiffered:Ignorance:‘‘movestheviolintoanothercontainer.’’Knowledge-plausible:‘‘movestheviolintotheredcontainer.’’Knowledge-implausible:‘‘movestheviolintothepurplecontainer.’’Allsubjectsthenread,‘‘Then,Deniserearrangesthecon-tainersintheroomuntiltheroomlookslikethepicturebelow.’’Thiswasfollowedby,‘‘WhenVickireturns,ethechancesVickiwillfirstlookforherviolinineachoftheabovecontainers?Writeyouranswersinpercentagesinthespacesprovidedundereachcontainer.’’ResultsIntheignorancecondition,subjectsgaveameanprobabilityratingof71%tothebluecontainer,whereVickioriginallyplacedtheviolin,andameanprobabilityratingof23%totheredcontainer,thecontainerthat,aftertheboxesweremoved,vetheothertwocontainersacombinedprobabilityratingof5%.Thatis,aswesuspected,subjectswhodidnotknowthefinallocationoftheviolinbelieveditwasrelativelyplausible(23%)thatVickiwouldfirstlookintheredcontainer,whichoccupiedthelocationoftheoriginalcontainer,383

ieveditwasrelativelyimplausible(2%)thatVickiwouldfi,tsintheknowledge-plausiblecondition,whoweretoldthattheviolinwasmovedtotheredcontainer,whichwasinalocationwhereVickimightplausiblylook,assignedsignif-icantlyhigherprobabilitiestotheredcontainerthandidsub-jectsintheignorancecondition,t(105)5À2.42,prep5.95,rmore,subjectsintheknowledge-plausiblecon-ditionassignedsignificantlylowerprobabilitiestothebluecontainerthandidsubjectsintheignorancecondition,t(105)52.35,prep5.95,,theirownknowledgeofthelocationoftheviolininfluencedthewthatVickiwasabsentwhentheviolinwasmovedtotheredcontainerandhenceshouldnotknowthisinformation,yettheirjudgmentsoftheprobabilitythatshewouldlookintheredcontainerweresignificantlyhigherthanthejudgmentsofadultswhodidnotpossessspecificknowledgeoftheviolin’rly,theirjudgmentsoftheprobabilitythatshewouldactaccordingtoafalsebelienowledge-implausiblecondition,subjectsknewthattheviolinwasmovedtothepurplecontainer,butthepurplecontainerwasanimprobablecontainerforVickitolookinfiilityjudgmentsthatVickiwouldlookinthatcontainerwerenotsignificantlyhigherinthisconditionthanintheignorancecondition,t(97)5À1.44,prep5.95,er,probabilityjudgmentsthatVickiwouldactrwords,subjectsreportedsimilarprobabilitiesforthe384Volume18—Number5

ndPaulBloomTABLE1MeanProbabilityJudgmentsThatVickiWillLookinEachoftheContainersConditionContainerBlue(wheretheviolinwasoriginally)Red(occupiesthelocationwheretheviolinwasoriginally)Purple(occupiesalocationdifferentfromwheretheviolinwasoriginally)Green(occupiesalocationdifferentfromwheretheviolinwasoriginally)Ignorance71%(26%)23%(22%)2%(5%)3%(7%)Knowledge-plausible59%(27%)34%(25%)3%(5%)4%(7%)Knowledge-implausible73%(29%)19%(21%)6%(16%)3%(5%)ntainerintheknowledge-implausibleandignorancecon-ditions,t(97)5À0.21,prep5.95,n.s.(SeeTable1forasum-maryofsubjects’probabilityjudgmentsinallthreeconditions.)Thus,atleastinthistask,knowledgewasacurseforadultsonlywhentheyhadavailableapotentialexplanationforwhyVickimightactinaccordwiththeirknowledge(insteadofherfalsebelief).Notethatthepotentialexplanationthatsubjectscouldusetojustifyassigningahigherpro,thattheviolinwasinthesamephysicallocationasithadbeenoriginally),itwasthesubjects’knowledgsadultswillsuccumbtothecurseofknowledgeonlywhenthereisaseeminglyjustifiablereasontosupportsuchabiasedresponse,or,toputitdifferently,perhapsknowledgeisacurseunlesstheoutcomeseemssuffiSIONOurfindingsdemonstratethatanadult’sownknowledgecancompromisehisorherabilitytoreasonaboutotherpeople’tswhoknewthespecificoutcomeofthedisplacementeventandhadavailableaplausibleexplanationforVickitoactinaccordwiththeirknowledgeweresignificantlylesslikelytopredictthatVickiwouldactaccorditswhoknewthespecificoutcomeofthedisplacementeventandhadavailableaplausibleexplanationforVickitoactinaccordwiththeirknowledgewerealsosignificantlylesslikelytopredictthatVickiwouldactaccordingtoafalsebeliefthanwerethosewhoknewtheoutcomebutdidnothaveavailabfindingshaveimplicatilts,ourfindingssuggestthatknowledgebecomesamorepotentcursewhenitcanbecombinedwitharationale(evenifonlyanim-plicitone)forinflatingone’pulatedtheplausibilityoftheprotagonist’sactions,butthisislikelyonlyoneofanumberofwayssubjectsmayfallpreytoVolume18—Number5thetandemeffectshildren,give,Birch&Bloom,2003;Pohl&Haracic,2005),itfollowsthatyoungerchildren’sper-formanceonfalse-belieftaskswouldbemorecompromisedthanolderchildren’sandadults’sothatalthoughwetestedadults’curse-of-knowledgebiasonlyinadisplacementtask(oneofthetasksmostcommonlyusedtoassesschildren’sfalse-beliefreasoning),thesamelogiccanapplytoanumberofdifficultiesinmental-statereasoningwhenthesubjecthasspecificknowledge,suchasthediffi,Perneretal.,1987),,Gopnik&Astington,1988),,Taylor,Esbensen,&Bennett,1994).Ourfindingssuggestthatallowingchildrentoremainignorantofthespecificoutcomeoftheeventsinthesetasksormanipulatingtheplausibilityofthevariousoutcomesinthesetaskscouldhelpadvanceunderstandingofchildren’gerchildrenhavealesssophisticatedunderstandingofwhatis,andisnot,plausiblethanolderchildrenandadultsdo,orsimplyfailtoconsidertheplausibilityoftheoutcomeatall,thenthismaycontributetotheirgreatersusceptibilitytothecurseofknowledgeandex-acerbatetheirdiffi,youngchildrenareoftennotedtobemoregullibleandacceptingofcertainimpossibilities,suchasthosepresentedinmagictricks,findingsfromthepresentstudydonotprecludeahybridaccountof3-year-olds’diffissiblethatyoungchildren’sdifficultyonstandardfalse-belieftasksstemsfromfindingsdoshow,however,thatthecurseofknowledgecaninterferewithfalse-beliefrea-soningeveninadults,whoarelesssuscept,one’sownknowledgecanbeacursewhenreasoningaboutbeliefsthatdifferfromone’ldren,greatersus-ceptibilitytothiscursemayleadtomoreblatanterrorsinmental-stateattribution,butknowledgecancontaminateevenadults’abilitytoreasonaboutotherpeople’sactionsandbeliefs.385

TheCurseofKnowledgeAcknowledgments—ThisresearchwassupportedbyaNaturalSciencesandEngineeringResearchCouncilofCanadaDis-coveryGrant(04-4747)andbyanEarlyCareerScholarAwardfromthePeterWallInstituteforAdvancedStudiestothefiNCESBaron-Cohen,S.,Leslie,A.M.,&Frith,U.(1985).Doestheautisticchildhavea‘theoryofmind’?Cognition,21,37–ein,D.M.,Atance,C.,Loftus,G.R.,&Meltzoff,A.N.(2004).Wesawitallalong:logicalScience,15,264–,S.A.J.(2005).Whenknowledgeisacurse:Children’sandadults’tDirectionsinPsychologicalScience,14,25–,S.A.J.,&Bernstein,D.(2007).Whatcanchildrentellusabouthindsightbias:Afundamentalconstraintonperspective-taking?SocialCognition,25,98–,S.A.J.,&Bloom,P.(2003).Childrenarecursed:logicalScience,14,283–,S.A.J.,&Bloom,P.(2004).Understandingchildren’sandadults’inCognitiveSci-ences,8,255–,P.,&German,T.P.(2000).ion,77,B25–r,C.,Loewenstein,G.,&Weber,M.(1989).Thecurseofknowledgeineconomicsettings:lofPoliticalEconomy,97,1232–d,A.,&Kirkham,N.(2005).logicalScience,16,291–off,B.(1975).Hindsightdoesnotequalforesight:lofExperimentalPsychology:HumanPerceptionandPerformance,1,288–,J.(1992).Atheoryofthechild’ion,44,283–,T.P.,&Leslie,A.M.(2000).ll&(Eds.),Children’srea-soningandthemind(pp.229–252).Hove,England:,A.(1993).Howweknowourownminds:TheillusionoffioralandBrainSci-ences,16,1–,A.,&Astington,J.W.(1988).Children’sunderstandingofrepresentationalchangeanditsrelationtotheuevel-opment,59,26–,B.,Lin,S.,&Barr,D.J.(2003).ion,89,25–,A.M.(1987).Pretenseandrepresentation:Theoriginsof‘‘theoryofmind.’’PsychologicalReview,94,412–,K.H.,&Baillargeon,R.(2005).Do15-month-oldinfantsun-derstandfalsebeliefs?Science,308,255–,J.,Leekam,S.R.,&Wimmer,H.(1987).Three-year-olds’difficultywithfalsebelief:hJournalofDevelopmentalPsychology,5,125–,M.V.(2003).Surprise,defence,ormakingsense:Whatremoveshindsightbias?Memory,11,421–,R.(1998).anJournalofCognitivePsychology,10,191–,R.,&Haracic,I.(2005).Derru¨ckschaufehlerbeikindernunderwachsenen[Hindsightbiasinchildrenandadults].Zeitschriftfu¨rEntwicklungspsychologieundPadagogishePsychologie,37,46–,D.,&Leslie,A.M.(1998).Solvingbeliefproblems:ion,66,1–n,E.B.,Cassidy,K.W.,&Baron,J.(2003).‘‘Iknow,youknow’’:ofGeneralPsychology,7,38–,M.,Esbensen,B.M.,&Bennett,R.T.(1994).Children’sun-derstandingofknowledgeacquisition:Thetendencyfoevelopment,65,1581–man,D.,Lempert,R.O.,&Hastie,R.(1990).alityandSocialPsychologyBulletin,17,30–n,H.M.(1990).Thechild’dge,MA:n,H.M.,Cross,D.,&Watson,J.(2001).Meta-analysisoftheoryofminddevelopment:evel-opment,72,655–,H.,&Perner,J.(1983).Beliefsaboutbeliefs:Representationandconstrainingfunctionofwrongbeliefsinyoungchildren’ion,13,103–ik,D.(1990).Whenrepresentationsconflictwithreality:Thepreschooler’sproblemwithfalsebeliefsand‘‘false’’ion,35,41–69.(RECEIVED5/3/06;REVISIONACCEPTED8/25/06;FINALMATERIALSRECEIVED10/30/06)Volume18—Number5