Automatic and Controlled Components of Social Cognition

292 SOCIAL PSYCHOLOGY AND THE UNCONSCIOUS personalitymeasures.EuropeanJournal of PsychologicalAssessment,21(2), 100-107. Sherman,S. J., Presson,C. C., Chassin,L., Rose,J. S.,& Koch,K. (2003).Implicitand explicitattitudestowardcigarettesmoking:Theeffectsof contextandmotivation. JournalofSocialandClinicalPsychology,22,13-39. Simon,H. A. (1955).A behavioralmodelof rationalchoice.QuarterlyJournalofEconom- ics,69,99-118. Sriram,N., Greenwald,A. G.,& Nosek,B.A. (2006).Scaleinvariantcontrastsofresponse latencydistributions.Unpublishedmanuscript. Steffens,M. C..(2004).Is theImplicitAssociationTestimmuneto faking?Experimental Psychology,51(3),165-179. Steffens,M.C.,&Buchner,A.(2003).ImplicitAssociationTest:Separatingtranssituation- allystablesandvariablecomponentsof attitudestowardgaymen.Experimental Psychology,50(1),33-48. Steffens,M. C., & Plewe,I. (2001).Items'cross-categoryassociationsasa confounding factorin theImplicitAssociationTest.Zeitschriftfur ExperimentellePsychologie, 48,123-134. Teachman,B.A.,Gregg,A. P.,&Woody,S.R.(2001).Implicitassociationsforfear-relevant stimuliamongindividualswithsnakeandspiderfears.JournalofAbnonnalPsy- chology,110(2),226-235. Teachman,B. A., & Woody,S. R. (2003).Automaticprocessingin spiderphobia:Implicit fearassociationsoverthecourseof treatment.JournalofAbnonnalPsychology, 112(1),100-109. Teachman,B. A, Gapinski,K. D., Brownell,K. D., Rawlins,M., & Jeyaram,S. (2003). Demonstrationsof implicitanti-fatbias:The impactofprovidingcausalinforma- tionandevokingempathy.HealthPsychology.22(1),68-78. Teige,S., Schnabel,K., Banse,R., & Asendorpf,J. B. (2004).Assessmentof multiple implicitself-conceptdimensionsusingtheExtrinsicAffectiveSimonTask(EAST). Unpublishedmanuscript. Wegner,D. M. (2002).Theillusionofconsciouswill.Cambridge,MA: BradfordBooks. Wigboldus,D. (2004).Single-targetimplicitassociations.Unpublishedmanuscript. Wilson,T. D., Lindsey,S.,& Schooler,T. Y. (2000). A modelof dualattitudes.Psychological Review,107(1),101-126. Wittenbrink,B., Judd,C. M., & Park,B. (1997).Evidencefor racialprejudiceat the implicitlevelanditsrelationshipwithquestionnairemeasures.JournalofPersonal- ityandSocialPsychology,72(2),262-274. Wittenbrink,B., Judd, C. M., & Park,B. (2001).Spontaneousprejudicein context: Variabilityin automaticallyactivatedattitudes.Journalof PersonalityandSocial Psychology,81(5),815-827. AutomaticandControlled Componentsof SocialCognition: A ProcessDissociationApproach B.KEITH PAYNE andBRANDON D. STEWART I n a famousdescriptionof unintendedbehavior,WilliamJames(1890)notedthat,"Veryabsent-mindedpersonsin goingto theirbedroomto dressfor dinnerhavebeenknowntotakeoffonegarmentafteranotherandfinallyto getintobed,merelybecausethatwasthehabitualissueofthefirstfewmovements whenperformedata laterhour,"(p. 115).This kindof absent-mindedness,or somethinglikeit,willprobablystrikemostreadersasalltoofamiliar.Comparethis witha descriptionby theneurologistFrancoisLhermitteof a patientnearlya centurylater.Whenshownintoabedroom,thepatient"immediatelybegantoget undressed.He gotintobed,pulledthesheetuptohisneck,andpreparedtogoto sleep,"(Lhermitte,1986,p. 338).We canbe surethatthisbehavioris more exceptionalthanJames'absent"mindedness,becausethebedroombelongedtoLhermitte. Thepatienthadamassivesurgicallesionoftheleftfrontallobe,abrainregion criticalforstrategicplanningandcontrolofactions.Thisandothersimilarpatients sufferedfromwhatLhermittetermed"environmentaldependencysyndrome." Althoughtheirbehaviorswerecoordinatedandcomplex(notsimplereflexes),they wereunderthecontrolof theenvironmenttoastrikingextent.Theybehavedin accordwithwhateverenvironmentalcuescaughttheireye.In onedemonstration, uponnoticingthatasyringehadbeenlaidout,apatientpickeditupandbeganto givetheneurologistaninjection.Thereaderisnottoldwhethertheinjectionwas carriedout.In another,evenmoremacabreinvestigation,thepatientwasledtoa tablewhereapistolandsomebulletshadbeenplaced.Withoutpauseor com- ment,hepickedupthepistol,pulledbackthemagazine,andloadedit. At this pointweread,"Theexperimentwasthenstopped,"(p.338).Thesewerenotthe momentarylapsesof voluntarycontrolthatwe all experience,but profound absences.Lhermittedescribeditasadisorderofautonomy. No lessintriguingarethestrugglesof individualswithalienhandsyndrome. 294 SOCIAL PSYCHOLOGY AND THE UNCONSCIOUS Becauseofdamagetothefrontallobesorthecorpuscallosumconnectingleftand righthemispheres,thesepatientsexperience"autonomous"actionsby oneor morelimbs.The actionsareautonomousin thesensethatthepatientcannot voluntarilycontrolthem,nordotheyexperiencethebehaviors(noreventhelimb itself)astheirmVll.LikeLhermitte'spatients,the"alien"actionsareusuallytrig- geredbyenvironmentalcues.For example,patientsmayexperiencea"struggle betweenthehandsaseachattemptsto answerthetelephone,"or evenmore drastically,"onehandtriedto turnleftwhentheotherhandtriedto turnright whiledrivingacar,"(Doody& Jankovic,1992;p.807).MarchettiandDellaSala (1998)reportapatientwho,"atdinner,muchto herdismaysawherlefthand takingsomefishbonesfromthe leftoversandputtingthemintoher mouth," (p.196). Thesebizarreconditionsseemso strangebecausetheyare dissociations betweenfunctionsthatusuallyworkseamlesslytogether,withintentionsreiningin theautomaticwhenit goesastray.Dissociationsaresoinformativebecauseso muchcanbelearnedaboutthestructureofasystemfromwhereitsfaultlineslie. Whenacrystallineicicleshattersonthefloorit comesapartin remarkablyregular patterns,verydifferentfromtheshatteringof anegg.Their innerorganization revealsitselfin thewaystheycomeapart.Neurologistshavebeenstudyingdis- sociationscausedby anatomicallesionsfor morethana century.Neurological dissociationshavemorerecentlyattractedtheattentionofsocialpsychologistsfor theirpotentialto shedlighton thecontrolof complexthoughtprocessessuch asthoseinvolvedin socialbehavior(e.g.Bargh,2005;BeeretaI.,2003;Wegner, 2002). Althoughanatomicaldissociationsmapout thesefaultlinesin vividdetail, dissociationsinthebehaviorofhealthymenandwomencanbejustasinformative. One way to find the seamsbetweenthe intentionallycontrolledand the uncontrolledis to compareexplicitandimplicittasks.Amnesiacpatients,for example,showprofounddeficitswhentestedusingexplicitmemorytests,which askthepersonto intentionally'retrievea memory.Yetwhentested\vithimplicit memorytests,theirperformanceshowseffectsof pastexperience\vithoutthe intentto rememberor thefeelingof remembering(Shimamura,1986).Butit is notjustamnesiacpatientswhoshowthisdifference.Normalhealthycollegestu- dentsalsoshowdissociationsbetweenimplicitandexplicitmemorytests(Jacoby & Dallas,1981).Factorsthataffectperformanceononekindoftestoftenhaveno impactontheother(seeRoediger& McDermott,1993).Withthehelpofimplicit andexplicittasks,researcherscanprobethedistinctionsbetweenmentalpro- cesseswithoutrelyingonrare,unfortunatecasesofbraindamage.Theycaninfer theinnerstructureswithoutwaitingforthemtocrack.Thefactthathealthymen andwomenshowdissociationsbetweenintentionalandunintentionalaspectsof behaviorforcesustoaskaboutthefaultlinesinordinarythought.Doweallhave littledisordersofautonomy?In someways,weshallsee,theanswerisyes. The casesreviewedhere are all dissociationsbetweenintentionaland unintentionalprocesses,eachat differentlevelsof analysis.The dissociations becomeincreasinglysubtle,butnolessintriguingaswezoomin fromthemacro scaletothemicro.Lhermitte'sfrontalpatientsshowedadissociationatthelevelof METHODS: PROCESS DISSOCIATION PROCEDURE 295 thewholeperson.Mostpeopleactwithintentionalcontrolmostof thetime,but thesepatientsdidnot.Casesofalienhandrepresentdissociationswithinaperson, butbetweenlimbs.Onehandisundervoluntarycontrol,andtheotherisnot.At a stillfinergrainaredissociationsbetweentasks.Thesamepersonactsdifferently onthetwotasksbecausethetasksdrawondifferentprocesses. We can zoomin further.This chapterfocuseson a processdissociation approach- a techniquefor separatingintentionalandunintentionalcontributions to thesamebehaviorperformedbythesamepersonatthesametime.Imagine thatamaniswalkingtowardyouonthestreet.Althoughyoudonotrecognizehim, youhaveabadfeelingabouthimanddecidetocrossthestreet.If youhadbeen ableto intentionallyretrievethefactthatyousawhimyesterdayon a wanted poster,youwouldhavehadanevenbetterbasisforyourdecision,andmighthave hurriedfaster,or calledthepoliceafterward.But evenwithoutbeingableto remembertheposter,thevaguesenseofthreatcanalsoguideyourresponse.Here intentionalandunintentionalformsoflearningcouldbothfeedintotheresponse, invaryingdegrees.Thefactthatyoucanhaveonewithouttheotherillustratesthat theyareseparable.Just aslesionstudiesallowdramaticdissociationsbasedon anatomy,theprocessdissociationapproachseeksto separateintentionaland unintentionalinfluences,eventhoughtheynormallyoperatetogether. Thegoalofthischapteristooverviewthelogicbehindtheprocessdissociation approach,whatit measures,andwhatit doesnot.Wewill describesomeof the manydifferenttopicswhereprocessdissociationhasbeenused,andthekindsof insightsit canprovidefor socialpsychology.Alongtheway,wewill discussthe assumptionsthatmustbemettoproperlyusetheprocedure,andwewillseehow thiswayof thinkingabouttheautomatic-controlleddistinctioncomparesand contrastswith other prominentapproaches.Althoughthe procedurewas developedin the contextof memoryresearch(Jacoby,1991;Jacoby,Toth,& Yonelinas,1993)we\villfocusonapplicationsoutsideof purememoryresearch, emphasizinginsteadsocialcognitionandbehavior.Interestedreadersarereferred to Jacoby(1998)for an overviewof theprocedureas developedin memory researchandYonelinas(2002)foratheoreticalreviewofdual-processtheoriesof memoryusingprocessdissociationandrelatedmethods. PROCESSDISSOCIATIONINSOCIALCOGNITION A studybyHense,Penner,andNelson(1995)markedapointof departurefrom pure memoryresearchto socialmemory.We \vill usethisexperimentto illustrate how processdissociationcanbe usedto studysocialmemorydistortions.Partici- pantswere askedto remembera list of traits that describedelderlyor young individuals.Each trait was stereotypicalfor either old or youngpeople. After studyingthe traits,participantswere askedto recallthe traitsthatdescribedthe older andyoungertargetpersonsundertwo setsof instructions.In the inclusion condition,participantswere askedto respondwith the trait theyhad studiedor, if they could not rememberthe trait, respondwith the first word thatcameto mind. In thiscondition,responsescanbe drivenby eitherintentionallyretrieved j 296 SOCIAL PSYCHOLOGY AND THE UNCONSCIOUS memoriesorbyautomaticformsofmemorythatcausecertainthoughtstocometo mindmorereadily.In theinclusionconditionintentionalandunintentionalforms ofmemoryworkinconcert.In theexclusioncondition,participantswereaskedto respondwithanewtraitthatwasnotstudied.If theyrememberedhavingstudied a trait,theycouldsuccessfullyavoidreportingit. However,if theyfailedto consciouslyremembera trait,but it unintentionallycameto mind,theywould be likelyto reportit. This conditionpitsintentionalandunintentionalformsof memoryagainsteachother. By comparingperformancein inclusionandexclusionconditions,theinflu- enceof intentionalandunintentionalusesof memorycanbeestimated.To the extentthata personrespondswiththecorrecttraitwhentheytryto, butalso \vithholdsitwhentheytryto,memoryisunderintentionalcontrol.Becauseinthis paradigmsubjectiveawarenessofthememoryisthebasisforcontrollingmemory reports,consciousnessof thememorycanalsobeinferred.Buttotheextentthat pastexperienceinfluencesperformanceregardlessofwhatparticipantsaretrying todo,theyarebeingunintentionallyinfluencedbymemory.Jacobyandcolleagues havetermedthisthe"logicof opposition."The equationsfor estimatingthese influenceswillbediscussedinalatersection. Henseandcolleaguesfoundthatstereotypicaltraitshadaselectiveinfluence on theunintentionaluseof memorywithoutaffectingcontrolledrecollection. Stereotype-consistenttraitssuchasslowandfrailcametomindeasilyandbiased memoryreportswhethertheyweretryingtoretrievethemortryingnottoretrieve them.Consciouslycontrolledmemorywasaffectedbyadividedattentiontask,but thiswasindependentof thestereotypingeffect. This studyrevealedanimportantdissociation.Withinthesingleactivityof rememberingtraits,intentionalandunintentionalformsof memorybothfedinto responses.Automaticinfluencesreflectedparticipants'ownstereotypes.In con- trast,consciouslycontrolledmemorywasaffectedbydividedattention.Thisstudy helpedtocharacterizethemechanismsbehindstereotypicalmemorybiasesthat havebeenknownforyears(fora reviewseeStangor& McMillan,1992).Social psychologyhasa traditionof emphasizingdissociationsbetweenunderlyingpro- cessesthatwouldseeminglygotogether.Devine's(1989)importantdemonstra- tionthattheautomaticactivationof stereotypescouldbedissociatedfromtheir useprovidesa clearexamplein thedomainof stereotyping.WhereasDevine contrastedresultsfromaprimingtaskwithresultsfromself-reportmeasures,the studyby Henseandcolleagues(1995)contrastedtwoaspectsof a singleactof remembering. Evenwithinthesametask,intentionalandunintentionalusesof memory operateverydifferently.Usinga memoryparadigmin whichracestereotypes couldbiasmemory,wehaveexploredthewaysthatcomponentsof memoryare relatedtosubjectiveexperience(Payne,Jacoby,& Lambert,2004).Wefoundthat people'ssubjectivesenseof confidencein theirmemorieswaswell-attunedto recollection.Whentheywereconsciouslyrecollectingthepast,theyexpectedto be right,andtheyusuallywere.Whentheyhadnorecollection,theyexpected tobewrong,andtheyusuallywere.Butconfidencewasnotatalltunedin tothe automaticinfluencesofstereotypesonmemory.Whenconsciousmemoryfailed, METHODS: PROCESS DISSOCIATION PROCEDURE 297 feelingsofconfidencegavenocluesaboutwhetheranautomaticbiaswasatwork. Thisasymmetryofsubjectiveawarenesshasimportantconsequencesforavoiding stereotypicalbiases. In oneconditionof thestudywerequiredparticipantstoanswereverymem- oryquestion,whethertheyrememberedthecorrectansweror not.In theother condition,weinstructedparticipantstoansweronlyif theybelievedtheiranswer tobecorrect.Whateffectdidthefreedomtochoosehaveonmemoryreports? Thatdependedonwhataspectof memoryis lookedat.Whenit cametooverall accuracy,thefreedomtochoosehelpedmemory.Becausesubjectiveconfidence waswell-tunedtorecollection,participantswereabletoavoidansweringquestions theywouldgetwrong,andtheychosetoanswerthoseitemstheywerelikelytogetcorrect. Butthestorywasverydifferentwhenit cametotheinfluenceofstereotypes. Memoryreportswerebiasedtowardstereotype-consistentmemoryerrors.But critically,thepatternofstereotype-consistentbiaswasjustasstrongwhenpartici- pantswereallowedtochooseaswhentheywererequiredtoanswereveryques- tion.Whenit cametorecollection,asin manyareasof life,awarenessbestowed control.Butwhenstereotypescamereadilytomind,Jamalwaslikelytobecomean athlete,andWalterwaslikelyto becomea politicianindependentof intent, independentof subjectiveexperience,andindependentof thechoiceto keep quiet. Thesestudiesmovedfrombasicmemoryresearchtosocialmemorybiases. But theprocessdissociationprocedurecanalsobeusedin contextscompletely unrelatedto memory.Imaginenowthatthepersonapproachingon thestreet lookslikeasuspectandyouareapoliceofficer.Thesuspectpullsanobjectfrom hispocket.Whatshouldyoudo?Thiswasthedecisionfacingfourpoliceofficers onFebruary4,1999whentheyconfrontedAmadouDiallooutsidehisapartment in NewYorkCity.Theirdecisionturnedouttobewrong.Diallowaskilleddespite beingunarmedandhavingnothingtodowiththecrimetheofficerswereinvesti- gating.Thecasesparkedpublicoutrageandchargesof racismbecauseDiallowas Black.However,asinsomanyinstancesindailylife,therewasno"controlgroup" togaugetheimpactof race.Sincethatincident,manysimilarc;aseshavebeen reportedin thenationalpress.Manyinsideandoutsideof lawenforcementhave askedthemselves,"WhatwouldI doin thatsituation?"Wehaveusedtheprocess dissociationproceduretostudywhatpeopleactuallydointhatkindofsplit-second decisionsituation. To seehowprocessdissociationcanbeinformativehere,considerthediffer- entpossiblescenariosthatmightconfronttheofficer.In onescenario,a Black suspectpullsagun.Here,the"correct"response(atleastforthepurposesofour thoughtexperiment)is to "shoot."Thatresponsemightcomeaboutfromtwo routes.Oneis anintentionallycontrolledresponse,in whichyoushootbecause youmeantoshoot.Thesecondrouteis anunintentionalor automaticresponse evokedby race stereotypesabout the suspect.Becauseintentionaland unintentionalprocessesareworkingtogetherhere,wecannottell themapart. Nowconsideradifferentscenario,inwhichtheBlacksuspectholdsonlyawallet (as,infact,Dialloheld).Hereanintentionallycontrolledresponsewouldbetonot 298 SOCIAL PSYCHOLOGY AND THE UNCONSCIOUS shoot.Butanautomaticresponsebasedonracialstereotypesmightstillleadone topull thetrigger.To theextentthatapersonsystematicallyshootsdespitenot intendingto,weknowthatanunintentionalprocessis atwork.Buttotheextent thatapersonshootswhentheymeanto,andnototherwise,weknowthattheyare incontrol. Our laboratoryhasconducteda numberof studiesexaminingthesesortsof scenarios.WehaveusedasimpleprocedureinwhichthefacesofBlackandWhite individualsareflashedonacomputerscreenjustbeforepicturesofhandgunsand handtoolsareshown.All ofthepicturesarepresentedlongenoughtoseeclearly andalloftheitemsareeasytoidentifY.Weshowpairsofobjectsonthescreenand askparticipantsto respondbypressinga"gun"keyor a "tool"keyasquicklyas possible.Acrossmanystudies,wehavefoundaclearandconsistenttendencyto mistakenlyrespond"gun"whenaBlackfaceisflashed.Thattendencyincreasesas participantsarerushedtorespondfasterandfaster(Payne,2001;Payne,Lambert, & Jacoby,2002). Thesimilarityofthistasktootherimplicitmeasuresmakesittemptingtothink ofthebiasasan"automaticeffect."Thatwayofthinkingaboutit isconsistent\vith thetaskdissociationapproach,inwhichimplicittasksareidentified\vithautomatic processesandexplicittasksareidentified\vithcontrolledprocessing.Butthatway of thinkingoverlooksthepossibilityof bothautomaticandcontrolledprocesses feedingintoresponses. To revealthedistinctprocessesseamlesslyguidingresponses,wecompared responseswhenautomaticandcontrolledprocesseswereactinginconcertversus whentheywereopposed.Whenbothautomaticandcontrolledprocessespointed to the"gun"response(aninclusioncondition),theprobabilityof agunresponse was.75.Thiscanbeformalizedasthesumofcontrolledprocessingandautomatic processingwhencontrolfails:Control+Automaticx (1- Control).Becausethe valueshereareprobabilities,theabsenceof aneventcanbeeasilyexpressedas (1- theprobabilityof thatevent).On Black-tooltrialswhereautomaticstereo- typingwouldleadtoa"gun"responsebutcontrolledrespondingwouldleadtoa "tool"response,participantsstillresponded"gun"\vithaprobabilityof .37.This representsthetendencyforanautomaticbiastodriveresponsesin theabsenceof control:Automaticx (1- Control). Toestimatehowmuchof thisbehaviorwasduetointentionalcontrol,wetook thedifferencein performancewhenbothautomaticandcontrolledprocesses favoredaresponse,versuswhenautomaticbutnotcontrolledprocessesfavoredit. Thisgivesa controlestimateof .38(=.75- .37).1It is importantto noticethat control,asconceptualizedhere,is nota reactiontoastereotypicalthought.It is notthoughtsuppressionoranafter-the-facteditingofresponses.It istheabilityto focusattention,thought,andactionon goal-relevantbehaviorsindependentof automaticdistractions.If controlwereaperfect1.0,actionswouldbedetermined completelybyintentions.In thisexperimentthevaluewasmuchlower,allmving forotherfactorstounintentionallyinfluencebehavior.Withsimpleconstraintslike speededresponding,it is notdifficultto seemomentarydisordersof autonomy cripplinggoodintentions. To estimatethe automaticeffectof stereotypingwe lookedat howoften METHODS: PROCESS DISSOCIATION PROCEDURE 299 participantsrespondedin line\viththestereotypeevenwhentheyintendedto respondotherwise(.37).Althoughthisvaluereflectsanunintentionalprocess,it is anunderestimatebecauseit isthejointprobabilitythatanautomaticprocesswas atworkandthatcontrolfailed.Themorecontrolapersonexerts,themorethis valueunderestimatestheautomaticstereotypingeffect.Undertheassumption thatautomaticandcon