But do these preference reversals also carry over to future choices and ratings To investigate this question we gave participants the task of choosing which of a pair of faces they found most attractive Unknown to them we sometimes used a card trick ID: 38964
Download Pdf The PPT/PDF document "Choice Blindness and Preference Change Y..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
choice,(ii)controlfortheinformationrevealedbythechoice,(iii)removetheinformationfromthechoiceor(iv)manipulatethechoicesthatpeoplemake.Here,wewouldliketofocusonthefourthandnalsuggestion:n:nal]approachthatresearcherscanuseistomanipulatethechoicesthatpeoplemakeandthencalculatespreadforeachparticipantbasedontheirrandomlyassignedcondition/therearetwocentralproblemsthatneedtobesolved.First,researchersmusteffectivelymanipulatechoice.Second,researchersmustavoiddirectlymanipulatingpreferences.(Risen&Chen,2010,p.1159)Webelievethatourchoiceblindness(CB)methodologyisagoodexampleofchoicemanipulationthatmeetsthecriteriaandsolvestheproblemsidentiedbyChenandRisen(2010).ChoiceblindnessTheCBmethodologywasoriginallyinspiredbytechniquesfromthedomainofclose-upcard-magic,whichpermitsasurreptitiousmanipulationoftherelationshipbetweenchoiceandoutcomethattheparticipantsexperience.InJohansson,Hall,Sikström,andOlsson(2005),participantswereshownpairsofpicturesoffemalefacesandweregiventhetaskofchoosingwhichfaceineachpairtheyfoundmoreattractive.Inaddition,immediatelyaftertheirchoice,theywereaskedtoverballydescribethereasonsforchoosingthewaytheydid.Unknowntotheparticipants,oncertaintrials,adouble-cardploywasusedtocovertlyexchangeonefacefortheother.Thus,onthesetrials,theoutcomeofthechoicebecametheoppositeofwhattheyintended.Bothdecisiontheoryandcommonsensestronglysuggestthateveryonewouldnoticesucharadicalchangeintheoutcomeofachoice.Butonthegreatmajorityoftrials,participantsfailedtonoticethemismatchbetweenchoiceandoutcome,whilestillbeingpreparedtoofferintrospec-tivereasonsforwhytheychosethewaytheythoughttheyhad.WehavedemonstratedCBforattractivenessofabstractartisticpatternsandformaleandfemalefaces,bothwhenpresentedinaliveinteraction,asdescribedearlier(Johanssonetal.,2005;Johansson,Hall,Sikström,Tärning,&Lind,2006),andinacomputerizedparadigm(Johansson,Hall,&Sikström,2008).CBhasalsobeendemonstratedindifferentmodalities.Steenfeldt-KristensenandThornton(2013)haveestablishedCBfortactilechoices,andwehaveextendedittothelinguisticdomainforpurelyspokendecisions(Lind,Hall,Breidegard,Balkenius,&Johansson,submitted).WehavealsoshowntheeffectofCBforthetasteofjamandthesmellofteainanecologicallyvalidsupermarketsetting(Hall,Johansson,Tärning,Sikström,&Deutgen,2010).Merkelbach,Jelicic,andPieters(2010)haveappliedthephenomenonofCBtotheproblemofmalingeringintheclinicaldomain,andrecently,wehaveestablishedthatCBhasclinicalrelevanceasadiagnosticinstrumenttostudyobsessive-compulsivedisorder(Aardemaetal.,submitted).MostrelevantforthecurrentdiscussionisthepreferencereversalatthecoreoftheCBparadigm.FollowingtheassumptionsofChenandRisen(2010),choiceandendorse-mentofanalternativeinanon-manipulated(NM)trialareaparadigmaticcaseofastatedpreference.Consequently,CBtrialswhereparticipantsendorseandarguefortheoriginallyrejectedalternativemustbeseenasapreferencereversal.Thisreversalcansometimestakefairlydramaticform,suchaswhenCBcanbefoundformoraljudgmentsinvolvinghotlydebatedtopicsinthecurrentpoliticaldebate(Hall,Johansson,&Strandberg,2012),orwhenitcanbeshowntostronglyinuencevotingintentionsjustaweekbeforeanationalelection(Halletal.,2013).EventhoughthecoreCBphenomenonrepresentsasimpledemonstrationofpreferencereversals,thequestionremainswhetherthesereversalsareephemeralorlasting,andhowthedynamicsofpreferencechangeplaysoutovermultiplechoiceandratingpoints(seeCoppinetal.,2012;Sharot,Fleming,Koster,Yu,&Dolan,2012,fortworecentstudiesconcerningthestabilityofpreferencemodulationovertime).ItalsoneedstobeclearlydemonstratedhowCBasamethodologycanbeusedtoovercometheproblemshighlightedbyChenandRisen.FortheRCRandRRC,ChenandRisen(2010)arguedthatasinglepointpreferenceelici-tation(whetherbyratingorchoice,orotherwise)tendstobelessinformativethanthewholeseriesofinstances.ForCB,doesthismeanthatparticipantswillreverttotheiroriginallyrevealedpreference,orwilltheycontinuetoprefertheoptiontheyendorsedinthemanipulated(M)trials?Toinvestigatethesequestions,wesetupaCBexperimentwithasimilarrepeatedstructureasinthetypicalFCP.Usingthesamecard-trickmethodologyasinJohanssonetal.(2005),weletparticipantschoosebetweentwofaces,andforsometrials,wereversedtheirchoices.Butwealsoincludedasecondroundofchoicesusingthesamefacepairs,aswellastwostagesofpost-choiceattractivenessratingsofthefaces(choiceratingchoicerating).Thisway,weareabletomea-surepreferencestrengthbothaschoiceconsistency,thatis,towhatextenttheparticipantspreferthesamefacethersttimeandthesecondtimetheyarepresentedwiththepairs,andasadifferencemeasurebetweenthechosenandrejecteditems.EXPERIMENT1:CHOICEBLINDNESSANDPREFERENCECONSISTENCYFortymainlystudentparticipants(24women)tookpartinexperiment1(meanage24.3years,=4.7),eachreceivingacinematicketaspaymentforparticipation.Fifteenpairsofgrey-scalepicturesoffemalefacesfromastu-dentpopulationwereusedasstimuli.Thepairswereconstructedbytheexperimenters,andanattemptwasmadetokeepphysicalsimilarityconstantatanintermediatelevel(i.e.clearlydifferent,butnotdrasticallyso).Allpictureswereratedby17independentratersforattractivenessonascalefrom1to10(1=notatallat-tractiveand10=veryattractive).Sixofthe15pairswerecho-senastargetpairsforthestudy,withadifferenceinattractivenessbetweenthefacepairsrangingfrommoderate=.23)tolarge(=1.7)(Figure1).Allpictureswereprintedandgluedonredcardboard,size7.5×9.3cm.AnJournalofBehavioralDecisionMakingCopyright©2013JohnWiley&Sons,Ltd.J.Behav.Dec.Making(2013):10.1002/bdm additionalsetofcopiesofthesixtargetpairswereprintedandgluedontoblacklaminateplate.Theparticipantswerepresentedwith15pairsofpicturesoffemalefacesandwereaskedtodecidewhichineachpairtheyfoundmoreattractive.Eachpairwasshownfor4secondsandwasthenputfacedownonthedeskpad.Theparticipantswereinstructedtopointatthefacetheyfoundthemostattractiveassoonasthecardswereturneddown.Forthesixtargetpairs,theparticipantswereaskedtoexplaintheirchoicewhentheypickedupthechosencard.Forthreeofthesesixpairs,amanip-ulationwasintroduced,sothattheparticipantsreceivedtheoppositeoftheirchoice(Figure2).Allpairswerepresentedinarandomizedorder.Themanipulationswererandomlydistributedoverthelast10presentedpairs;however,twomanipulatedpairswereneverpresentedinarow.Aftertheparticipantshadexplainedwhytheypreferredthechosenpicture(i.e.thefacetheydidchoseorinMtrialswereledtobelievetheychose),theywereaskedtoratetheattractivenessofthechosenfaceona10-pointscale.Whentheyhadstatedtheirrating,theparticipantswerealsogiventhenon-chosencardtorateforattractiveness.Aftertherstroundof15choices,allthepairswerepresentedasecondtimeinarandomizedorder,andtheparticipantswereaskedtochoosetheoneineachpairtheynowpreferred.Inthesecondround,nomanipulationstookplace,andtheparticipantsdidnothavetoexplaintheirchoices.Finally,afterthesecondroundofchoices,theparticipantswereaskedtorateallthepicturesagain.Thistime,thepictureswerenotpresentedinpairs,butonebyoneinarandomizedserialorder.Attheendoftheexperiment,theparticipantswereaskedwhattheythoughtabouttheexperimentingeneralandiftheythoughtanythinghadbeenstrangewiththedesignoftheexperiment.Finally,theparticipantsweretoldaboutthemanipulationandthetruepurposeoftheexperiment.Iftheparticipantsatanypointduringthedebriengindicatedthattheyhadfeltthatsomethingwasstrangewiththepictures,theywereaskedtolookthroughallpicturesagainandpickouttheonestheythoughthadbeenmanipulated.ResultsDetectionrateOutof120Mtrials,11%ofthemanipulationswereimmediatelydetected,9%ofthetrialsweredetectedwhentheoriginallychosenimagewaspresentedduringtherstratingprocedureandanal13%ofthetrialswerecategorizedasretrospectivelydetectediftheparticipantsinthedebriengclaimedtohaveexperiencedsomethingbeingstrangeduringtheexperiment(seeJohanssonetal.,2005,2006,formoredetailsondetectioncriteriaanddebriengprocedure).Thismeansthatthelargemajorityofthemanipulationsremainedundetected.PreferencechangeWeanalysedpreferencestrengthforthesixtargetpairsbyusinganumberofdifferentmeasuresandatseveraldifferent Figure1.Thesixtargetpairsusedinthestudy Figure2.Asnapshotofthechoiceprocedureduringamanipulationtrial.(A)Theparticipantisshowntwopicturesoffemalefacesandisaskedtochoosewhichoneheorshendsmoreattractive.Unknowntotheparticipant,asecondcarddepictingtheoppositefaceisconcealedbehindthevisiblealternatives.(B)After4seconds,thepicturesareturneddown,andtheparticipanthastoindicatehisorherchoicebypointingatthepreferredcard.Theexperimenterthenslidesthehiddenpictureovertotheparticipantandrakesthehiddenblackcarddownintoherownlap.(C)Theparticipantpicksupthepictureandisaskedtoexplainwhythechosenfacewaspreferred.Theparticipantthenratesthefacetheyholdintheirhandforattractiveness,andthentherejectedface(i.e.theoriginalchoice)ispickedupandratedinthesamefashionChoiceBlindnessandPreferenceChangeP.Johanssonetal.Copyright©2013JohnWiley&Sons,Ltd.J.Behav.Dec.Making(2013):10.1002/bdm timepointsduringtheexperiment.Toprovidethemostconservativemeasureofthepreferenceeffects,ourmainanalysiswasperformedonalltrials,includingdetectedMtrials(seeChen&Risen,2010,onthegeneralproblemsoftrialexclusioninchoiceparadigms).Asreportedearlier,twothirdsofthedetectionsweremadealreadyduringtheexposureorrstratingphase,whichmakesitlikelythatmuchsmallerpreferencechangeeffectswouldbeseenindetectedtrials.Thus,tobetterunderstandthedynamicsoftheCBmanipulation,wealsocompareandreporttheeffectsofnon-detectedanddetectedMtrialsseparately.LookingattherstattractivenessratingfortheNMtrials,thechosenfacesareratedmuchhigherthantherejectedones=1.49,=1.1).ThisrelationshipisreversedfortheMtrials;themeanratingdifferencebetweentheoriginallychosenalternativeandtheoriginallyrejectedalternativeisnegative,indicatinganoverallchangeinrelativepreferenceinfavouroftheoriginallyrejectedphotographs(Mdiff.35,=1.2)(Figure3).ThisdifferenceinratingissignicantlyhigherintheNMtrialscomparedwiththeMones,(238)=12.46,.00001,=1.62.InadirectcomparisonbetweentheattractivenessratingsfortheoriginallychosenfacesinNMtrials(=5.8,=1.4)andMtrials(=5.1,=1.6)trials,theratingsweresignicantlyhigherintheNMtrials,(238)=3.40,.001,=.44.Similarly,theratingsfortherejectedfacesweresignicantlyhigherintheMtrials(=5.5,=1.4)whencomparedwiththeNMtrials(=4.3,=1.3),(238)=6.63,.00001,=.86.Thisindicatesthatthemanipulationleadstoanincreaseintheperceivedattractivenessfortheoriginallyrejectedalterna-tives,aswellasadecreaseintheratedattractivenessoftheoriginallypreferredfaces.Ifwecomparetheratingdifferencefornon-detected(Mdiff.5,=.86)anddetectedMtrialsMdiff.04,=1.6),weseethatbotharenegativebutthatthedifferenceissignicantlylargerfornon-detectedtrials,(118)=2.02,.05,=.37.Thesecondandmainpreferencemeasureisconsistencybetweenchoice1andchoice2,thatis,towhatextentthesamefaceineachpairwaspreferredthesecondtimethechoicewasmade.ThelevelofchoiceconsistencybetweenrstandsecondchoiceswasclosetoceilingfortheNMtrials(93.3%).However,fortheMtrials,thisnumberwasjust56.6%.ComparingchoiceconsistencybetweenNMandMtrials,thedifferencewashighlysigni=240)=43.02,=.42).Again,itisinterestingtonotethatfortheMtrials,thechoiceconsis-tencyislowerforthenon-detectedtrials(43.8%)comparedwiththedetectedtrials(82.5%)(=120)=16.31,=.37).Theattractivenessratingperformedattheendoftheexperimentshowsasimilarpattern,despitethisbeinganon-pairedserialratingofthefaces.Comparingtheratingdifferencebetweenthechosenalternativeandtherejectedalternative,weagainndasignicantdifference,withlargerpositivedifferenceforNM(Mdiff=1.2,=1.4)comparedwithMtrials(=.2,=1.4),(238)=5.62,=.73(Figure1).Mostofthiseffectisaccountedforbytheincreasedpreferencefortheinitiallyrejectedfaces:TheratingoftheoriginallychosenfaceswasnotsignilowerintheM(=5.5,=1.5)trialscomparedwiththeNMtrials(=5.7,=1.5),(238)=1.49,=.138,=.19,buttherejectedfacewasagainratedhigherinMtrials(=5.2,=1.6)whencomparedwiththeNMtrials=4.6,=1.4),(238)=3.35,=.43.Whencomparingthenalratingdifferenceofnon-detectedMtrials(=.09,=1.0)anddetectedMtrials(=.5,=2.0),wendtheeffecttobenolongersigni(118)=1.58,=.12,=.29.TheresultoftheratingiscongruentwiththatofSharot,Velasquez,andDolan(2010),inwhichtheinvestigatorsfoundapreferencemodulationforthechosenbutnotfortherejectedalternative.Thenaturalinterpretationofthisisthatthebeliefinhavingmadeachoicehasastrongerimpactonthepreferenceforthechosenobjectcomparedwithadecreaseinpreferencefortheitembelievedtohavebeenrejected.TheanalysisearlierrevealshowCBinuencesfuturepreferences.ParticipantschoseX,butendupendorsingY,andoftenkeepdoingsoinrepeatedchoicesandratings.However,toclearlydemonstratehowthisexperimentfulthemethodologicalcriteriasuggestedbyRisenandChen(2010)anddetailedintheIntroductionsectionofthecurrentarticle,weneedtolookatthedatafromadifferentperspec-RememberthatinRisenandChensdescriptiononhowtoproperlymanipulatetheparticipantschoicesinordertoshowapreferencechangeeffect(i.e.thefourthoptionontheirlist),notonlythechoiceshouldbecontrolledbytheexperimenterbuttheoutcomeshouldalsoberandomlyassignedforeachchoicesotheparticipantsobtainapre-determinedoptionregardlessofwhattheypreferthemselves.Withsucharandomassignment,someparticipantswill Figure3.Thedifferenceinattractivenessratingbetweenthechosenandrejectedalternativesinnon-manipulatedandmanipulatedtrialsfortherstandsecondratings Therewerenogenderdifferencesindetectionrate,(238)=0,=1,choiceconsistency,(238)=.61,=.54,orintherstratingdifference,(238)=.18,=.86,orthesecondratingdifference,(238)=1.35,=.18.Overall,womenratedthefacessignicantlyhigherthanthemendid,bothforthechosen(rating(238)=3.59,.0001;rating2:(238)=3.45,.0001)andtherejected(rating1:(238)=4.55,.0001;rating2:(238)=3.08,.005)images,butastherewasnodifferenceintheratingdifferencebetweenchosenandrejectedfaces,thisgenderdifferencedoesnotinuencethemainresults.Wewouldliketothankananonymousreviewerforsuggestingthispresen-tationofthedata.JournalofBehavioralDecisionMakingCopyright©2013JohnWiley&Sons,Ltd.J.Behav.Dec.Making(2013):10.1002/bdm experiencetheirchoice,andsomewillbepresentedwithamanipulatedreversedoutcome.Wedividedthedatasetoftheexperimentonthebasisofwhethertheparticipantswerepresentedwiththeleftface(L-group)ortherightface(R-group)ineachpairashavingbeentheirchoice.SomeoftheparticipantsintheL-groupwerethuspresentedwiththeirtruechoice(i.e.theNMtrials),whereassomepartici-pantswerepresentedwiththeoppositeoftheirchoice(i.e.theMtrials),andviceversafortheR-group.Thispost-hocdivisionisstatisticallyequivalenttohavingmadearandom-izedgroupassignmentbeforethestartoftheexperiment.Whencomparingthesecondchoicedistributionsofthetwogroups,aspredicted,wendthattherightfaceisstronglypreferredintheR-group,andtheleftfaceisstronglypreferredintheL-group(=240)=31.3,.00001).Similarly,creatingadifferencescorebysubtractingtheattractivenessratingoftherightfacefromthatoftheleftface,wendthatthetwogroupsdiffersignicantlyinbothrstrating(L-group,=1.3,=1.3;R-group,=1.1),(238)=12.7,=1.65,andthesecondrating(L-group,=.9,=1.5;R-group,=1.4),(238)=5.6,=.73.ThiscomparisonshowsthatitispossibletousetheCBmethodologytofullthecriteriasetoutbyRisenandChen.Ourresultindicatesthatthemanipulationstronglyinsubsequentchoicesandattractivenessratings.Theinitiallyrejectedfaceischosenmorefrequentlyinasecondchoice,andtheperceivedattractivenessisincreasedeveninratingsperformedwhenthefacesarepresentedoutsidethemanipu-latedpairingwithanotherface.Thenaturalinterpretationofthisresultisthat,intheMtrials,theparticipantscometopreferthefacetheywereledtobelievetheyliked.Apossibleobjectiontothisinterpretationisthatthechangeinratingsandchoiceconsistencyisnotduetothebeliefinhavingmadethechoicebutratheraresultoftheincreasedpresentationtimeoftheinitiallyrejectedalternative.WhentheparticipantsexplainwhytheypreferredthechosenalternativeinanMtrial,theyalsoexperiencethisoriginallyrejectedfaceforalongerperiod.Priorresearchhassuggestedthatprolongedexposuremayuencetheperceivedattractivenessofvisuallypresentedobjects,theso-calledmereexposureeffect(Zajonc,1968,2001).Toruleoutthispossibility,weranasimpleCBconditionwithasecondroundofchoicesbutwiththeratingprocedureremoved.Thiswascomparedwithamere-exposureconditioninwhicheitherthechosenorrejectedpicturewasshowntotheparticipantsforanextendedperiod,butwithoutanymanipulation,toseeiftheextraexposurewouldhaveaneffectonthesecondchoice.EXPERIMENT2:MEREEXPOSUREASAPOSSIBLEMECHANISMFORTHECHOICEBLINDNESSEFFECT?Fortymainlystudentparticipants(24women)tookpartinexperiment2(meanage24.0years,=3.6),eachreceivingacinematicketaspaymentforparticipation.Thesame15pairsofprintedimagesoffaceswereusedasinexperiment1,withthesamesixpairsusedastargetpairs.ProcedureExperiment2consistedoftwoconditions,asimpleCBcondition(=20)andamere-exposurecondition(=20).InthesimpleCBcondition,theprocedurewasthesameasinexperiment1regardingpresentationtimeandchoicepro-cedurebutwiththeratingprocedureoftheimagesremoved.Forthesixtargetpairs,theparticipantswereeithergiventhechosenfaceor,throughmanipulation,therejectedfaceandwereaskedtoexplaintheirchoice.Thechosen/rejectedimagewasvisibleonaverage8.9seconds(=4.2)duringthemotivation.Inthemere-exposurecondition,asinpreviousexperi-ments,theparticipantswerepresentedwiththe15pairsoffaces,eachpairshownfor4secondsandthenturnedfacedownonthetableuntiltheparticipantshaddecidedwhichonetheythoughtthemoreattractive.Forthreeofthesixtargetpairs,thechosenimagewasgiventotheparticipanttoholdandsimplylookatfor10seconds;fortheotherthreepairs,thenon-chosenimagewasgiventotheparticipantstolookatfor10secondsbutwithoutanyattempttoconcealthefactthatitwasthefacenotpreferred(i.e.itwasgiventotheparticipantsdirectlyfromthehandthatheldthenon-preferredface).Thisway,theparticipantsviewedthechosenfaceandtherejectedfaceforanequalamountoftimebutwithoutbeingledtobelievethattheyhadchosenthenon-preferredface.Inbothconditions,aftertherstroundof15choices,allthepairswerepresentedonemoretimeinarandomizedor-der,andtheparticipantswereaskedtochoosetheoneineachpairtheynowpreferred.ResultsThedetectionratefortheMtrialswassimilartothatinexperiment1,with10%ofthemanipulationsdetectedconcurrentlyandanother16%retrospectivelydetectedaftertheexperiment.ComparingthelevelofchoiceconsistencybetweenandsecondchoicesintheCBcondition,theparticipantswereconsistentin83.3%oftheNMtrialsandin61.6%oftheMtrials.Thisamountstoasignicantdifferenceinlevelofchoiceconsistency(=120)=7.06,.01).Inthemere-exposurecondition,theparticipantswerealsoexposedtothreeofthechosenalternativesandtothreeoftherejectedalternativesanequalamountoftimebutwithoutbeingmisledinrelationtowhichalternativetheypreferredineachpair.IfmereexposurewasthesolemechanismatplayintheCBexperiment,thereshouldbethesamepreferenceeffectinthisconditiontheparticipantsshouldbe Theviewingtimewasestimatedfromthelengthoftherecordingofthever-balreport.Inthemereexposurecondition,wethereforeaddedasecondtoaneven10seconds.ChoiceBlindnessandPreferenceChangeP.Johanssonetal.Copyright©2013JohnWiley&Sons,Ltd.J.Behav.Dec.Making(2013):10.1002/bdm signicantlylessconsistentforthepairsinwhichtheyhadhadaprolongedinteractionwiththenon-preferredalternative.However,thereisnoevidencethatthisisthecase:intrialsinwhichtheparticipantsweregiventhepreferredimageaftertherstchoice,thechoiceconsistencywas88.3%,whereasfortrialsinwhichtheparticipantsweregiventherejectedalternative,thelevelofchoiceconsistencywas78.3%,anon-signicantdifference((1,=120)=2.16,=.14).ComparingthechoiceconsistencyfortheMtrialsinthesimpleCBcondition(61.3%)withthetrialsinwhichtheparticipantsreceivedtheoppositeoftheirchoiceinthemere-exposurecondition(78.3%),wedondthattheMtrialsarecantlylessconsistent(=120)=3.97,Giventhisresult,mereexposurecannotbethemechanismbehindthepreferencechangefoundintheCBexperimentsreportedhere.GeneraldiscussionSummarizingtheresults,wendthatourCBmethodologyfulllsthecriteriasetoutbyRisenandChenandthatCBmanipulationscanstronglyinuencesubsequentchoicesandattractivenessratingsinasetupsimilartotheclassicFCBprocedure.Theinitiallyrejectedfaceischosenmorefre-quentlyintherepeatedchoicetrials,andtheperceivedattractivenessoftheinitiallyrejectedfaceisincreasedeveninuncoupledindividualratingsattheendoftheexperiment.ThisdemonstratesthattheparticipantscometopreferthefacetheywereledtobelievetheylikedandthattheeffectsofCBarenotonlyvisibleinsnapshotmeasures(whetherthesearerelativelyinconsequential,asthejudgingofabstractpatternsinJohanssonetal.,2008,orconcernchoicesofgreatpersonalandsocietalimportance,asthevotingdecisionsinHalletal.,2013)butrathercanmanifestthemselvesovermultiplechoiceandratingpoints,potentiallywithlong-lastingconsequences.Anobjectiontoourndingwouldbethatthepreferencechangemightbeconfoundedbymereexposureofthealter-natives(Zajonc,1968,2001),butexperiment2showedthatthisismostunlikelytobethecase.Anotherobjectionwouldbethattheparticipantsweclassiedasnothavingdetectedthemanipulationsactuallydidso,butrefrainedfromtellingus,thendeducedthepurposeoftheexperimentandalteredtheirchoicesandratingstopleaseus.However,wendthisobjectionveryunlikely.OnethingwehaveconsistentlyfoundinourCBstudiesisthatthereareremarkablyfewdifferencesbetweenhowtheparticipantsbehaveinNMandnon-detectedMtrials,whetherthisisshowninlinguisticbehavior,suchasemotionality,specicityandcertaintyintheverbalreportsmotivatingtheirchoices(Johanssonetal.,2005,2006),inexpressedcondenceinthechoicesmade(Halletal.,2010)orinlackofdifferentia-tiononstandardcomplianceandsocialdesirabilityscales(Aardemaetal.,submitted;Merkelbachetal.,2010).Inarecentstudy,participantsweregivenacomputerizedCBtaskwhiletheireyemovementsandpupillaryresponseswererecorded(Pärnamets,Hall,Strandberg,Balkenius,&Johansson,inpreparation).Firstofall,thesimplefactthattheparticipantslookatandfullyattendtothemanipulatedimagesafterpresentationrulesoutinattentionasapossibleexplanationforCB.Inlinewithpreviousstudies,wealsofoundnodifferencesinviewingpatternsafterpresentationbetweenNMandnon-detectedMtrials,whereaswedistinctdifferenceswhencomparedwithdetectedMtrials.Inaddition,wefoundasignicantincreaseinpupildilationinthedetectedMtrialscomparedwithbothnon-detectedMtrialsandNMtrialsand,atthesametime,nodifferencesbetweennon-detectedMtrialsandNMtrials.Pupildilationhasbeenusedasarobustmeasureofsurpriseandcognitiveload(Porter,Troscianko,&Gilchrist,2007;Preuschoff,Hart,&Einhäuser,2011),andthisnewresultthuslendsstrongsupporttotheclaimthattheparticipantsarenotconsciousofthemanipulation.Inthecurrentstudy,theobjectionisfurtherunderminedbythefactthattheprefer-encechangeeffectsareconsiderablysmallerforthedetectedtrials.Together,thispatternofresultsclearlyindicatesthatintrialsclassiedasbeingnon-detected,theparticipantsaretrulyunawareofthemanipulationmade.InlinewiththemethodologicalimprovementssuggestedbyRisenandChen(2010),threenewversionsoftheFCPhaverecentlybeenintroduced(Coppin,Delplanque,Porcherot,Cayeux,&Sander,2012;Egan,Bloom,&Santos,2010;Sharotetal.,2010).Thersttwooftheseprocedurestrytocontroltheimpactofpriorpreferencesbylettingtheparticipantsmakeachoicebetweentwooptionswithoutknowingwhattheoptionsare,aso-calledblindchoice(correspondingtooption3intheintroductorylistingofChenandRisenssuggestedsolutions).Forexample,inSharotetal.(2010),theparticipantsrstratealonglistofnamesofholidaydestinationsandarethenaskedtomakechoicesbetweentwoequallyrankeddestina-tionsthatareclaimedtobesubliminallypresentedonthescreen(thatis,theyareunseen,butparticipantsareforcedanyway).Afterthechoice,thetwomaskedalternativesaremadevisibleandmarkedaccordingtowhattheparticipantshadindicatedastheirchoice.Finally,theparticipantsrateallthealternativesonemoretime.Theoriginaltwisthereisthatonlynonsensesymbolsareshownduringthesubliminalpresentation,sotheparticipantshavenoinformationtobasetheirforcedchoiceson.Still,thisproce-dureledtoanincreaseintheratingofthechosenalternative(butnoreductioninvalueforthealternative).Thisresultwasfurtherextendedinafollow-upstudy,showingthatsomeofthesechangesinpreferencewerepresentupto3yearsaftertheblindchoicewasmade(Sharotetal.,2012).Anotherattempttoshowbothshort-andlong-termprefer-enceeffectsofchoicewhilestillavoidingtheobjectionsbyChenandRisenwasintroducedbyCoppinetal.(2012).Inthisexperiment,theparticipantshadtosmellandthenrateindividualodoursforpleasantness.Beforethestartofeachtrial,theparticipantshadtomakeachoicewhethertopayfortheupcomingtrial,withoutknowingwhatsmelltheywouldreceive.Theparticipantsweregivenmoneytopayforexactlyhalfofthetrials.Thestudyfoundanincreasedpleasantnessratingforodoursprecededbythechoicetopay, Thisdescriptionisconfusinglysimilartochoiceblindnessbutisnototh-erwiserelated.OurtermderivesitsnamefromtheparentphenomenonofchangeblindnessandwasnamedinJohanssonetal.(2005).JournalofBehavioralDecisionMakingCopyright©2013JohnWiley&Sons,Ltd.J.Behav.Dec.Making(2013):10.1002/bdm andreducedforodourswithnocost,aneffectfoundbothimmediatelyandaweekaftertheexperiment.Asthechoicewasmadebeforethepresentationoftheodours,itwasalsoblindwithrespecttotheparticipantsunderlyingpreferences.Thesestudieshavegonesomewaytowardsre-establishingthenotionthatchoicescaninuencefuturepreferences.Buttheblindchoiceapproachcanbeseenassomethingofapyrrhicvictory.Holden(2013)arguedthatbycomparingobjectsparticipantsarenearlyindifferentbetween,andbyremovingtheactualcomparisonoftheobjectsfromthechoice,itistoofarremovedfromreal-worlddecisionmakingtobeconsideredachoiceatall.IthasalsobeennotedthattheeffectsizesintheblindchoiceexperimentareconsiderablylowerthanintheoriginalFCPexperiments,againtothepointofquestioningthereal-worldrelevanceofndings(Izuma&Murayama,2013).ComparingtheaverageeffectsizeofstudiesaddressingtheChenandRisencriticism(=4,)=.26;fromthemeta-analysisinIzuma&Murayama,2013),with,forexample,theeffectsizeofthenalnon-pairedratingdifferencescorefortheleftandrightchoicesinexperiment1(=.73),wendthattheeffectisconsiderablystrongerinthecurrentstudy.ItisalsounclearhowfartheresultofCoppinetal.(2012)speakstotheissueathand.Theparticipantschosewhethertopaytosmellanodour,andthisinuencedfuturepreferences.ThisdesignavoidsthecriticismbyChenandRisenbymakingthechoice,buttheeffectobtainedisbestexplainedinrelationtotheeffortjusticationratherthantheFCP.Theparticipantschoicesarenotmadebetweentwooptionswhosepreferencesaretobealtered(ornot);itisjustachoicetospendsomemoneyonthistrialorthenext(withoutthisinanywaybeingrelatedtowhatodourtheparticipantswillreceive).SeeIzumaandMurayama(2013)forasimilardiscussionofCoppinetal.(2012).IfweapplytheprecedingdiscussiontotheCBmethodol-ogy,wearguethatitevadestheproblemsoftheblindchoiceparadigm.Inthecurrentexperiment,allparticipantsweredeeplyfamiliarwiththetaskofevaluatingfacialattractive-ness,andonlytwosimpleoptionswerecomparedbeforetheymadetheirchoice.Mostimportant,ourprimaryout-comemeasurewerenotjustnearlyequalratingssomewhatpushedapartfromtheirinitiallypositions,butachoicecompletelyreversed,andasubsequentchoiceconsistencyinthenon-detectedtrialsaslowas43.8%.Thefacepairsusedwerepre-ratedandselectedtodifferinattractiveness(frommoderatelytostronglydifferent),insteadofmatchedtothepointofindifferenceasintheoriginalFCP,ortheblindchoicestudies,butdespitethis,westillfoundastrongpreferencechangeeffect.However,asweseeit,thecomparativelyeasypartistosatisfytherequirementsofChenandRisen(2010)andtodemonstrateeffectsofchoiceonfuturepreferences.Thecultpartistoevaluatetheunderlyingassumptionsaboutpreferencesthatfuelledtheiroriginalefforttore-describetheFCP.Couchedintermsofstandarddecisiontheory(vonNeumann&Morgenstern,1947),andfollowingtheassumptionslaidoutbyChenandRisen(2010),CBbecomesveryparadoxical.Thehighconsistencybetweenrstandsecondchoices(andratings)inNMtrialsseemstoindicateastrongandenduringtruepreference.However,thisnotioniscontradictedbythedramaticreversalfoundintheMtrials,whereparticipantsstatedtheirpreferencebychoosing,andthenimmediatelycreatedmaximalpotentialdifferencebyacceptingtheoppositealternative.SomeproponentsoftheconceptoftrueunderlyingpreferencesmightwanttoarguethatCBrevealsthatparticipantsneverhadanypreferencetobeginwith(notwithstandingthestrongconsistencyoftheNMtrials),butthisisabarbedbitterbullettobite,asitrendersagreatmanypreferenceelicitationspotentiallymeaningless(e.g.seeHall,Johansson&Strandberg,2012).ThisbecomesmostapparentwhenweconsiderstudieswhereCBhasbeenshownwithecologicallyvalidstimuli,andwithdecisionsinvolvingreal-worldconsequences.Wouldproponentsoftrueprefer-encesbewillingtoarguethatthe50%detectionrateinHalletal.(2012)demonstratesthathalftheSwedishpopulationholdsnoarticulatedattitudesaboutthemostvisiblemoralissuesinthecurrentsocietaldebate?Forexample,thisisamovethatwouldthreatentomakethemeticulouscritiqueofFCPbyChenandRisen(2010)entirelysuperuous,asmostlikelyCBcouldbedemonstratedforthedifferentchoiceandratingpointsintheFCP.Forexample,theoriginalFCPstud-iescomparedchoicesbetweentheseventhandninthrankedoptionsforaestheticevaluationsofpaintings,andconsiderablelevelsofCBhavealreadybeendemonstratedforclassicalandmodernpaintingsbyMasuda,Seiko,andWatanabe(2010).IfnoneoftheapparentpreferencesinFCPcountasreal(choice-inducedornot),thenthisconstitutesafarmoresevereindictmentoftheeldthanthepossibilityofstatisticalerrorsinthewayspreadiscalculated.Atthesametime,thegistofthecriticismmadebyChenandRisen(2010)makesagreatdealofsense.Often,decisionresearchisseverelyimpoverishedintheelicitationofprefer-encesacrosstimeandcontext(Chater,Johansson,&Hall,2011;Johansson,Hall,&Chater,2011).Ineverydaylife,unlessforcedto,whowouldrelyonasinglestatementorchoicetofullyexplaintheneedsanddesiresofourfellowbeings(Dennett,1991a)?Evolutionarilyspeaking,itmakessensetofocusbothonpatternsofstability(unsurprisingly,afamishedpersonwillnearlyalwayshaveapreferenceforfood)andontheabilityforconstantcontextualre-evaluationsthatwehavehadtodisplay(todecidewhethertocontinueforagingforfood,tosleepandconserveenergy,toescalateaconict,tostanddownandsoon;Davies,Krebs,&West,2012).Butunlessoneassumesthatpeoplehaveperfectintrospectiveaccesstotheirownpreferences,amostcontentiousassumptionabsentfromthelistofChenandRisen(2010),itfollowsfromtheirownargumentsthatchoicesrevealinformationaboutpreferencesnotonlytousasobservingresearchersbutalsototheparticipantsthem-selves(Carruthers,2011;Dennett,1991b;Johanssonetal.,2005,2006;Nisbett&Wilson,1977). Decisionsmightalsobemadefromhabits,whichmightexhibitrewardin-sensitivityafterovertraining,orevenasareexiveaction,operatingoutsidetherealmofinstrumentalactions(Morsella,2009),butthiswouldnottypi-callybethecaseinthetypeofchoicesstudiedbysocialpsychologists.ChoiceBlindnessandPreferenceChangeP.Johanssonetal.Copyright©2013JohnWiley&Sons,Ltd.J.Behav.Dec.Making(2013):10.1002/bdm Mostpreviousexperimentsonchoice-inducedpreferencechangehaveinterpretedtheeffectinrelationtocognitivedissonance(e.g.Brehm,1956;Festinger,1957;Gerard&White,1983;Shultz&Lepper,1996),butinthecurrentcontext,wearguethatsomeformofself-perceptiontheoryisthemorelikelycandidate(e.g.Ariely&Norton,2008;Bem,1967;Chater&Vlaev,2011;Dennett,1987).Itisnotoriouslyhardtodifferentiatebetweencognitivedissonanceandself-perceptionmodelsastheymakealmostidenticalpre-dictions(Bem,1967;Harmon-Jones&Mills,1999).Themainreasonwefavourself-perceptiontheoryisthepreviouslymen-tionedresultsonCBandpupildilation,asthisindicatesthattheparticipantsdonotexperienceanydissonantemotionsofcognitionsthatwoulddrivethechangeinpreference.Eveniffurtherexperimentsareneededtormlydecidethisissue,itisclearthattheCBmethodologyiswellsuitedasatooltodifferentiatebetweenthesetwomodels.Similarly,inthecurrentstudy,itisnotpossibletomea-suretherelativeimpactofbelievingthatachoicehasbeenmade,orbelievingthatachoicehasbeenmadeexplainedthechoice.Askingwhythechoicewasmadeisanaturalwaytomakesurethattheparticipantsinteractwithitemaftermanipulation,butpost-choiceatten-tiontothemanipulatedstimulicouldinfuturestudiesbemeasuredindependentlyby,forexample,eyetracking.Butregardlessoftheexactnatureoftheunderlyingmechanism,thecombinedresultofCBandblindchoiceindicatesthatitisnotthechoicepersethatdrivesthepreferencechangebutratherthebeliefthatacertainchoicehasbeenmade.Wecanthusbegintoseparatetheactofchoosingfromthebeliefinhavingmadeacertainchoice.GiventhesmallbodyofworkthathasbeencarriedoutonCB,itisdifculttoestimateitspervasiveness.Ifweassumesomeunderlyingpreferences,andacertainamountofnoiseintheprocessofratingandchoosing,butstillthinkthatchoicesmayinuencepreference,wewillndourselveswithnaturallyoccurringCB-likesituations,whereparticipantsgetwhattheychoosebutnotwhatthey.IfsomeoneactuallyprefersAoverB,buthappenstochooseBthroughrandomvariation,theymightthenshifttheirpreferencetowardsBthroughself-inference.WealsohaveregularCBsituations,wherepeopleforsomereasondogetwhattheychoose,butfailtonoticeit,andthensetthemselvesupforpotentialpreferentialloops,asinourexperiments.Logically,wecannotknowanythingaboutthefrequencyofmismatcheswedonotnotice(i.e.wemighthaveunknowinglybeenshort-changedthousandsoftimesinourlivesorwalkedformonthsinoddlypairedsocks),buttakentogether,thesetwoalternativeshighlighthoweverypreferencewemeasuremightchangeasanoutcomeofusmeasuringit.ACKNOWLEDGEMENTSP.J.wouldliketothanktheBankofSwedenTercentenaryFoundation,andL.H.wouldliketothanktheSwedishRe-searchCouncilforfundingthisresearch.N.C.wassupportedbyERCgrant295917-RATIONALITY,theESRCNetworkforIntegratedBehaviouralScience,theLeverhulmeTrustandtheTempletonFoundation.REFERENCESAardema,F.,Johansson,P.,Hall,L.,Paradisis,S-M.&Roberts,S.ChoiceBlindnessandConfabulatoryIntrospectioninObsessive-CompulsiveDisorder:ANewAreaofClinical.Firstauthorafliation:UniversityofMontreal,FernandSeguinResearchCenter.Ariely,D.,&Norton,M.I.(2008).Howactionscreatenotjustre-TrendsinCognitiveSciences(1),13Bem,D.J.(1967).Self-perception:Analternativeinterpretationofcognitivedissonancephenomena.PsychologicalReviewBrehm,J.W.(1956).PostdecisionchangesinthedesirabilityofJournalofAbnormalPsychology(3),384Carruthers,P.(2011).Theopacityofmind.Oxford:OxfordUniver-sityPress.Chater,N.,&Vlaev,I.(2011).Theinstabilityofvalue.InM.Delgado,E.A.Phelps,&T.W.Robbins(Eds.),Decisionmak-ing:AttentionandperformanceXXIII(pp.81100).Oxford:Ox-fordUniversityPress.Chater,N.,Johansson,P.,&Hall,L.(2011).Thenon-existenceofriskattitude.FrontiersinPsychology3.DOI:10.3389/Chen,M.K.(2008).Rationalizationandcognitivedissonance:Dochoicesaffectorreectpreferences?Workingpaper.NewHaven,CT:YaleUniversity.Chen,M.K.,&Risen,J.(2010).Howchoiceaffectsandreectspref-erences:Revisitingthefree-choiceparadigm.JournalofPersonalityandSocialPsychology(4),573594.DOI:10.1037/a0020217Coppin,G.,Delplanque,S.,Cayeux,I.,Porcherot,C.,&Sander,D.(2010).Imnolongertornafterchoice:Howexplicitchoicesimplicitlyshapepreferencesofodors.PsychologicalScience(4),489Coppin,G.,Delplanque,S.,Porcherot,C.,Cayeux,I.,&Sander,D.(2012).Whenexibilityisstable:Implicitlong-termshapingofolfactorypreferences.PLoSONE(6),113.e37857.Coppin,G.,Delplanque,S.,&Sander,D.(submitted).Theimpactofdecision-makingprocessesonpreferencesinthefree-choiceparadigm:Classicalresearch,recentmethodologicalissues,andemergentconceptualinsights.FirstauthorafmentofPsychologyUniversityofGenevaDavies,N.,Krebs,J.,&West,S.(2012).Anintroductiontobehav-iouralecology.Chichester:JohnWiley&Sons.Dennett,D.C.(1987).Theintentionalstance.Cambridge,MA:MITPress.DOI:10.2307/2026682Dennett,D.C.(1991a).Consciousnessexplained.Boston:Little,Brown&Company.Dennett,D.C.(1991b).Realpatterns.JournalofPhilosophy(1),27Egan,L.,Bloom,P.,&Santos,L.(2010).Choice-inducedpreferencesintheabsenceofchoice:Evidencefromablindtwo-choiceparadigmwithyoungchildrenandCapuchinmonkeys.JournalofExperimentalSocialPsychology(1),204207.Egan,L.,Santos,L.,&Bloom,P.(2007).Theoriginsofcognitivedissonance:Evidencefromchildrenandmonkeys.calScience(11),978Festinger,L.(1957).Atheoryofcognitivedissonance.Stanford,CA:StanfordUniversityPress.Gerard,H.B.,&White,G.L.(1983).Post-decisionalreevaluationofchoicealternatives.PersonalityandSocialPsychologyBulletin(3),365369.DOI:10.1177/0146167283093006Hall,L.,Johansson,P.,&Strandberg,T.(2012).Liftingtheveilofmorality:Choiceblindnessandattitudereversalsonaself-transformingsurvey.PLoSONE(9),18.e45457.Hall,L.,Johansson,P.,Tärning,B.,Sikström,S.,&Deutgen,T.(2010).Magicatthemarketplace:Choiceblindnessforthetasteofjamandthesmelloftea.(1),5461.DOI:Hall,L.,Strandberg,T.,Pärnamets,P.,Lind,A.,Tärning,B.,&Johansson,P.(2013).Howthepollscanbebothspotonanddeadwrong:UsingchoiceblindnesstoshiftpoliticalattitudesJournalofBehavioralDecisionMakingCopyright©2013JohnWiley&Sons,Ltd.J.Behav.Dec.Making(2013):10.1002/bdm ChoiceBlindnessandPreferenceChange:YouWillLikeThisPaperBetterIfYou(BelieveYou)ChosetoReadIt!PETTERJOHANSSON*,LARSHALL,BETTYTÄRNING,SVERKERSIKSTRÖMandNICKCHATERLundUniversityCognitiveScience,LundUniversity,SwedenSwedishCollegiumforAdvancedStudy,UppsalaUniversity,SwedenDepartmentofPsychology,LundUniversity,SwedenBehaviouralScienceGroup,WarwickBusinessSchool,UniversityofWarwick,UKChoiceblindnessisthendingthatparticipantsbothoftenfailtonoticemismatchesbetweentheirdecisionsandtheoutcomeoftheirchoiceand,inaddition,endorsetheoppositeoftheirchosenalternative.Butdothesepreferencereversalsalsocarryovertofuturechoicesand *Correspondenceto:PetterJohansson,LundUniversityCognitiveScience,LundUniversity,22222Lund,Sweden.E-mail:petter.johansson@lucs.lu.seCopyright©2013JohnWiley&Sons,Ltd.JournalofBehavioralDecisionMakingJ.Behav.Dec.MakingPublishedonlineinWileyOnlineLibrary(wileyonlinelibrary.com)10.1002/bdm.1807 onlineJohansson,Chater,ChoiceChange:LikePaperYou)Chose