RationallyBiasedCognitionAStudyofIdeologicallyMotivatedInformationProcessing
AlexSernyak
AdvisorDanKahan
YaleDepartmentofCognitiveScience
SeniorThesis
April21,2017
Sernyak 1
ABSTRACT
Thisthesisattemptstoaddtotheliteratureonsystematicbiasesthatoccurin
politicallymotivatedreasoning.Bothliberalsandconservativesdisplayideologically
motivatedcognitiononavarietyofissues;inotherwords,theyusereasontofitevidenceto
previouslyheldconvictions.Intelligenceandeducationhasbeenshowntoincrease
polarizationandincreasebiasinpoliticalinformationprocessing,ashigherlevelsof
intelligenceallowforgreaterdexterityinfittingfactstoanarrative.Thiseffectwasfurther
studiedbyusingtheWasonselectiontask,alogicaltestcommonlyusedtoinvestigate
reasoningcapacity.Thetestproveddifficultforthesamplepopulation,andveryfewwere
abletocorrectlysolvetheprompt.Thethesisalsoattemptstodeterminewhetherbias
extendsbeyondthecontextoflogicalreasoningandpolicy.Theconjunctionfallacy,atestof
narrativerichness,wasusedtoseewhethersubjectsviewedmembersoftheopposing
ideologyasimmoral.Whilethereweremodesteffectsintheideologicalconditions,the
referencedisplayedthemostsignificantresults,asconservativesweremorelikelytomake
theconjunctionfallacythanliberalswhentheimmoralbehaviorwasascribedtoanex-
convict.PolarizationincreasedathigherlevelsoftheCognitiveReflectionTest,atestof
propensitytouselogic,suggestingrationalitymaypromoteamorenarrativelyrich
worldview.
Sernyak 2
TABLEOFCONTENTSABSTRACT……………………………………………..………………………………………………………1
INTRODUCTION……………………………………..………………………………………………………3
BACKGROUNDINFORMATION……..………………………………………………....……3
IMPLEMENTS………………………………………………………………………………....….12
METHODS……………………………………………………..………………………………………………16
HYPOTHESES………………………………………………………………………………....……………..24
RESULTS……………………………………………………..………………………………………………..28
DISCUSSION…………………………………………………………………………………………..…….38
FUTUREDIRECTIONS……………………………………………………………………………………41
ACKNOWLEDGEMENTS…………………………………………………………………………………46
REFERENCES…………………………………………………………………………………………………47
APPENDIX………………..…….…...…………………………………………………………………………57
Sernyak 3
INTRODUCTION
BACKGROUNDINFORMATION
ThepopularityoftheAmericanpoliticalsystemhasplummeted,withhistorically
lowapprovalratingsofcongressandwiththemostrecentelectionfeaturingthemost
unpopularpresidentialcandidatesever(Gallup,Enten,2016).Thebuzzwordpolarization
hasbeenthrownaroundtheblogosphereandamongthetalkingheadsastherootofthese
issues.Thereportedproblemisthatthecountryisdivided:DemocratsandRepublicans
cannotagreeonanything,refusetoworktogether,andaremoreconcernedwithwinning
electionsthanhelpingtheAmericanpeople.Prophetically,GeorgeWashingtonwarnedof
theinherentproblematicnatureofatwopartysysteminhisfarewelladdress,imploring
thecountrytostayawayfromthetoday’smodeltoavoidpolarizingdeadlock(Washington,
1776).Butproblemsofpolarizationandbiasliedeeper.Whiletherearecertainlyissues
withthetwopartysystem,theydonotaccountfortheentiretyofobservedpoliticalbias.
Takeclimatechange,anillustrativeexampleoftenusedinstudiesofpoliticalbias.Despite
thefactthat99%ofclimatescientistsagreethatclimatechangeiscausedbyman(Powell,
2015),only48%percentofthegeneralpopulationholdthisviewpoint(PewResearch
Center).Theproblemisnotsolelywithintheconfinesofthesystem,butliesinthevery
waypeoplethink.Researchconcerninginformationprocessinginpoliticalcontextcanlead
tosmarterpolicies,allowingfordebatesthatarenothamperedbythesystematicbiases.
Thisthesisseekstocontinuetoaddtotheunderstandingofthecomplexitiesofpolitical
bias.UsingdatafromastudyintheCulturalCognitionProjectaboutattemptstoimprove
theCognitiveReflectionTest(SeeAppendixAandBfordetailsofthisstudy),thisthesis
aimstoaddtothecurrentliteratureonpolarizationandmotivatedbeliefs.
Sernyak 4
SCIENCECOMPREHENSIONTHESIS
Oftentheblameforpoliticalpolarizationislackofeducation.Manypeoplebelieve
thatwithamoreeducatedpopulaceavarietyofpositionsonpolarizedissues,suchasgun
controlorclimatechange,wouldconvergeinthefaceoffacts.Inthismodel,conservatives
wholackpropereducationsimplydonotrealizethepreponderanceofevidenceinfavorof
thecauseofglobalwarmingbeinghumans.Thisclassictheoryforwhycertainpolitical
disagreementsarise,despiteapreponderanceofevidencepointingtowardsonesolution,is
the“ScienceComprehensionThesis”.Thesametheoryisoftenappliedtoavarietyofissues,
andisoftenusedbyliberalstoclaimthatifonlythepublicwerebettereducatedaboutthe
issues,polarizationwoulddecrease.Yetwhilethistheoryisintuitivelysatisfying,thereis
littleevidentialsupportforitsvalidity(anddirectlycontradictoryevidencethatwillbe
discussedlater).Infact,thesatisfyingnatureofthetheoryisquitedetrimental,asmuchof
thefocusoncombattingpolarizationcentersoneducationalreform(seeNewYorkTimes
EditorialBoard,2015).Polarizationneedstobescientificallystudied,notintuitively
understoodinordertocreatemoreproductivepolicies,scientificcommunicationtactics,
anddebates.Theproblemmaynotbetheunderlyinginformationusedtoengagewiththe
issuesbutamoreinherentissueduetotheverywaypeoplethink.
CONSERVATIVISMASSOCIALCOGNITION
Thereisatrendinsocialpsychologytoconsiderdifferencesincognitivestylebased
onideology,largelyplacingtheblameofpolarizationonconservatives.Thetheoryisthat
conservativesaremorefearful,biased,prejudiced,andlessreliantonrationality.There
definitelymaybesometruthtotheclaimthereisadifferencebetweenconservativesand
liberals.Liberalsareseenasmoreeducatedbecausetheyare,andthedifferenceisgrowing
Sernyak 5
(PewResearchCenter).Thereisalsoarobustbodyofliteraturethatpointstoadifference
incognitionbetweenthoseontheleftandtheright.Asearlyasthe50’snegativeresults
werebeingreportedaboutconservatives,asAdornoshowedconservativestobemore
authoritarian(Adorno,1950).Ina2008metanalysisthisresultwasreiterated,and
furthermorethatconservativeswerelessopenmindedandlessagreeable(Sibley&
Duckitt,2008).Conservativesareoftenseenasmoreindividualistic,andhavingamore
structuredpictureoftheworld,beinglessopentochange,greaterfeardeath,dealin
absolutes,andaremorereliantandtrustingofinstitutions(Jostetal.,2003,Sibley&
Duckitt,2008,Deppeetal.m2015).Lowintelligencehasalsobeenlinkedtoconservatism,
asconservativessenatorswereshowntomakelesscomplexcommentsonpolicyevenafter
controllingforothervariablessuchaseducation(Tetlock,1983).Conservativeswerealso
foundtohavelowercognitiveability(Hodson&Busseri,2012)andtohaveincreased
prejudice(Hodson&Busseri,2012,Jostetal.,2003).Theyhavealsobeenshowntobe
moreimpactedbyemotionalstates,especiallyincasesoffearanddisgust.Whendisgust
wasactivated,participantswereshowntobemorefearfulofimmigrationanddisdainfulof
homosexuals,andthisresultwasfoundmoreprominentlyinconservatives(Terrizzi,
Shook&Ventis,2010,Eskine,Kacinik&Prinz2011).Moregenerallyconservativeswere
linkedtohigherlevelsofdisgustaswellasgreaterdisgustsensitivity(Inbar,Pizarro,Iyer&
Haidt2011).FindingssuchastheseledJosttosuggestinhismetanalysisthatconservatism
shouldbethoughtofasasocialcognition.Ifthistheoryweretrue,itwouldmake
polarizationaninevitableproductoftheconservativemind.
Recentresultshavechallengedthetheoryofconservatismasasocialcognition,as
similarbiaseshavebeenfoundinliberalsandconservatives.Thesebiasestypicallyarisein
Sernyak 6
situationswhenanoutcomedoesnotfitadesiredworldview.Takeharmreductionasa
policytomitigatethenegativeeffectsofdrugaddiction.Theoppositiontoharmreduction
strategieslikeneedleexchangeisoftenseenasblightontheconservativepsyche.
Conservativesaresodisgustedbydrugusagetheyareunabletoputinapolicythathas
beenshowntohelppeople,astheyfeelsomehowcomplicit.Yetarecentstudyshows
similarpatternsinliberals,whoarereticenttosupportharmreductionstrategiesin
similarlymoralizedissuessuchasfemalecircumcision(MacCoun,2013).Infact,inmany
studiesofpoliticalreasoningitseemssimilarpatternsemergeinliberalsand
conservatives.Bothgroupsdistrustedsciencemoreifgivenscientificevidencethatclashed
withtheirworldview(Nisbetetal.,2015,Lewandowsky&Oberauer,2016).Liberalsare
similarlybiasedagainstideologicallydissimilargroupsasconservatives(Brandtetal.,
2014).Thebeliefintheirrationalityofconservativesmaybeenaproductofalackof
diversityinsocialpsychology.Liberalscientistandtheorists,whosepoliticalviewsmay
havecoloredtheirperception,reportedmostoftheseresults.Theyalsowerelookingat
issuesinwhichconservativesarebiased,leadingtounfavorableresultsfortheright
(Duarteetal.,2015).Ultimately,whiletheremaybedifferencesbetweenliberalsand
conservatives,amorenuancedviewoftheissueisnecessary,andpoliticalbiasisclearly
presentonbothsides.
FUNDAMENTALIRRATIONALITY
Anotheranglemaybenecessarytofullyunderstandthephenomenonof
polarization.Essentially,thereisafundamentaldichotomyofhumanthought:intuitive,
automaticSystem1Processes,anddeliberate,rationalSystem2processes(Kahneman,
2011)(ForamoreindepthanalysisofthedifferencebetweenSystem1and2Processes
Sernyak 7
seeAppendixA).PoliticaldecisionswouldhopefullybemadewithSystem2processes,yet
peopledonotalwaysseemtoapproachpoliticsrationally.Takevoting,forexample,which
seemstobeaSystem2process,asoneneedstoevaluatethepoliciesofeachcandidateand
cometoadecision.Yetitseemsintuitivejudgmentsonmeaninglesscharacteristicssuchas
looksfactorheavilyintodecision-making(SeeDruckman,2003forananalysisoftheNixon
Kennedydebate,Todorovetal.,2005).Otheremotionalaffectscanimpactpoliticaldecision
makingaswell(Huang,Sedlovskaya,Ackerman&Bargh,2011,Eskine,Kacinik&Prinz
2011).Thereseemstobeanelementofirrationalityandadisconnectbetweenthecorrect
useofSystem1and2processes,butthisdoesnotfullyexplainissuessuchaspolarization.
Ifthiswerethereason,polarizationwouldbemitigatedbyrelianceonlogic.TheCognitive
ReflectionTest,orCRT(SeeAppendixBformoredetail),hasbeenshowntobeareliable
testofthepropensitytouselogicalSystem2processes,(Toprak,West&Stanovich,2014),
yetthishasnotbeenshowntodecreasepolarization(Kahan,2012).Neitherhavemore
traditionaltestsofintelligencesuchasnumeracy(Kahan,2013).Theoppositehasshown
tobethecase;asinathesestudiesthemeasuresoflogicalabilitywereshowntoincrease
polarization,andtheimplicationsoftheseresultswillbediscussedshortly.Sowhilethere
maybeelementsofirrationalityinpoliticalthought,itcannotexplainthephenomenonof
polarization.Infactrationalityseemsdirectlylinkedtopoliticalpolarization.
MOTIVATEDREASONING
Theremaybeaneedforimprovedscienceeducation,andpeoplemayinherentlybe
biased,butnoneofthesetheoriescanexplainthephenomenonthispaperisinterestedin:
howpeople,inthefaceofoverwhelmingevidence,remainpolarizedonclimatechange,
believeinbirtherism,andbendinformationtofittheirownnarrative.Inrecentyears,
Sernyak 8
politicalbiashasbeenthoughtofasaformofmotivatedreasoning.Motivatedreasoning
occurswhenhumansprocessevidencepresentedtotheminawaythatconfirms
previouslyheldconclusions.PerhapsthemostfamousexampleisthatoftheSeekers,a
groupwhopredictedtheapocalypseinthe1950’s,yetwhenthatdatecame,andno
apocalypseoccurred,thegroupdoubleddownontheirbeliefstatingtheapocalypsehad
beenavoidedduetotheirfaith(Festinger,Riecken&Schachter1959).Thenewpieceof
evidence,thefactthattheapocalypsehadnotoccurred,paradoxicallyledthemtobelieve
evenmorestronglythattheprophecywasreal.
Therearemanywaysinwhichthistypeofself-deceptioncanoccur.Mysidebias,or
confirmationbias,istheprocessbywhichpeoplebolstertheirownopinionbyinterpreting
information,seekingevidence,ortestinghypothesisinwaysthatwillconfirmtheir
previouslyheldbeliefs(Stanovich&West,2007,Nickerson,1998,Chen,1999,Greenetal.,
2002,Westfalletal.,2015,Kunda,1990).Thisbiasisapartofcognitivedissonancetheory,
whichpositsthatpeoplewantalltheirbeliefstobeheldinharmony,andwillworktoget
themtoalign(Festinger,1962).Thisalsoresultsinpeoplecompensatingwheninformation
goesagainsttheirworld-view,bytryingtofindwaystoreconcilethisdissonant
information(Proulxetal.,2012).Sonotonlydopeopleavoidinformationthatchallenges
theirworldview,whentheycomeacrossittheyattempttodismissit.
Haidtproposedatheoryofmoraljudgmentswheredecisionsaremadeintuitively,
andthenintelligenceandconsciousnessareusedtodefendthem.Asheputsit,human
intelligenceisusedmoreoftenintheroleofalawyerthanofscientist(Haidt,2000).An
illustrativeexampleofthesefallaciesisofthegenerallyover-optimisticnatureofthe
humanpopulation.Whenpredictingthefuture,animportantuseofreasoning,peopleare
Sernyak 9
likelytooverratetheirchildren’slikelyattributes,theirownlikelihoodofsuccess,and
underratetheprobabilityofnegativeeventssuchascanceroraccidents(Sharot,2011).
Theyalsoarelikelytoratethemselvesasbetterthantheyare(Hoorens,1993),aswellas
createself-servingdefinitionsofambiguoustraitstoboosttheirratings(Dunning,
Meyerowitz&Holzberg,1989).
Mysidebiasismitigatedbycertaincognitivetendencies.Whilestudieshaveshown
thatelementaryeducationdoesnothelpinformalreasoning(Perkins,1985),otherstudies
haveshownthatyearsincollegecandecreasebias(Toplak&Stanovich,2003).Motivated
reasoninghasbeenshowntobeunrelatedtointelligence(Stanovich&West,2007,
Klaczynski&Gordon,1996,Klaczynski&Robinson,2000).Intelligencewasshownto
increasebiasblindspots,orthetendencytoviewothersasmorebiasedthanoneself(West,
Meserve&Stanovich,2012).Yetwhileintelligenceisunrelatedtobias,areflective
cognitivestyle,asmeasuredbytheCRT,hasbeenshowntodecreasemysidebias(Toplak,
West&Stanovich,2001).Infact,itseemsasthoughmostpeoplehaveagreaterabilityto
decreasemysidebiasthentheynormallydisplay.Forexampleexperimenterswereableto
decreasemysidebiasthroughinstruction(Evansetal.,1994).Andparticipants
recommendedlessbiasedconclusionstoafictionalexperimenterthanwhenaskedabout
theirownconclusions(Greenhoot,Semb,Colombo&Schreiber,2004).Basically,people
oftenhavetheabilitytocometothecorrectconclusionuponreflection,butwilloftenbe
satisfiedwithamoreself-servinglogicalprocess.
IDENTIY-PROTECTIECOGNITIONTHESIS
AnalternatetotheScienceCognitionThesis(SCT),whichaccountsforPolitically
MotivatedReasoning,wasproposed:theIdentity-protectiveCognitionThesis(ICT).Not
Sernyak 10
onlydopeopleavoidcognitivedissonanceingeneral,butalsothereisextramotivationin
thisinstancetoremainwithaviewpointconsistentwithonesidentity.Thisisarational
response,asinthegrandschemeofthingsonehaslittleimpactonadebatesuchasclimate
change,buttheconsequencesofchangingopinionscanbedisastrous.Aliberalscientist
whocametotheconclusionthatclimatechangewasahoaxwouldbeostracized,aswould
anNRAmemberwhodecidedguncontrolwouldsavelives.Soingeneral,peopleavoid
havingpoliticalopinionsthatwedgethembetweentheirgroup(Kahan,2010).Itisnotthe
casethatpeopleareover-reliantonheuristics,itisthefactthattheysystematicallyfilter
informationtofittheirworldview(Stanovich,2013,Kahan,2015).Forexample,subjects
willratestudiesandargumentsmorefavorablyiftheyconformtopre-conceivedcriteria
(Lord,Ross&Lepper,1979,Fagerlin,Gready&Peterson,2002).
Asintraditionalexamplesofmysidebias,politicallymotivatedcognitiondoesnot
followaBayesianframework.Aperfectlylogicalthinkerwouldhaveaprior,then,upon
receivingnewinformation,wouldattempttoupdatetheirbeliefs.Alikelihoodratioofthe
newinformationshouldbemultipliedbythepriortogetaposteriorunderstandingofthe
subject.Inthismodel,accesstomoreinformationshouldalwaysbringposterior
likelihoodsclosertogether.Yetinstudiesthishasnotbeenshowntobethecase,asbeliefs
willnotconvergeinawaythatwouldbeconsistentwithaBayesianupdatingmodel
(Kahan,2015).Theoppositeiseventruecorrectionsofmisinformationactuallymake
ideologicallymotivatedindividualsmorecertainintheirbeliefs(Nyhan&Reifler,2006).
Thewayinwhichpoliticalthinkersintakeinformationisclearlyinfluencedbyfactorsother
thantheevidenceitself.
Sernyak 11
ThereseemstobeadistinctionbetweentheICTandmysidebias,asintelligence
playsaroleintheformerbutnotthelatter.Peopleusecognitiveresourcestocreatea
worldviewthatfitstheiridentity,andsmarterpeoplearebetteratfilteringinformation
thatkeepsthemincongruencewiththeirpeers(Kahan&Corbin,2016).Theyarebetterat
interpretingitinfavorableways,andbetteratlatchingontoitwhenitconfirmstheir
identity.Thisiswhybeliefscanactuallydivergeinthefaceofmoreeducationoraccessto
information(Kahan,2013).Forexample,participantsinoneconditionwereaskedto
determinefromexperimentalresultswhetheraskincreamhadworked,orinanother
conditionwhetheraguncontrolexperimenthadworked.Whenaskedabouttheskin
cream,intelligencecorrelatedwithbetterperformanceacrosstheboard.Intheguncontrol
condition,increasedintelligenceactuallyincreasedpolarization(Kahanetal.,2013).
Similarresultshavebeenfoundinpoliticallymotivatedreasoning(Kahanetal.,2012),but
notoutsideofthiscontext.Infact,mostmeasuresofcognitivecapacityhavebeenshownto
increasepolarization,demonstratingtheabilityofpeopletouselogictoprotecttheir
worldview.TheCRThasbeenshowntoincreasepolarization(Kahan&Stanovich,2016),as
hastheActivelyOpen-mindedTest(Kahan&Corbin,2016).Thepropensitytouselogic
allowsforbetterprotectionofidentityasthemindattemptstokeepthesociallyexpedient
beliefsinplace.
THEROLEOFCULTUREANDIDENTITY
Ratherthancometopoliticalopinionsindependentlybyweighingtheprosand
cons,groupmembershipplaysalargepartinhowpeoplethink.Culturehasbeenshownto
formmassopinions(Wildavsky,1987).ThesefindingsareconsistentwiththeICT;as
groupidentityisthemostimportantfactorindeterminepoliticalthought.Andcontraryto
Sernyak 12
popularbelief,everythingdoesnotbreakdownuponpartylines.Forexample,inastudy
askingaboutOutpatientCommitmentLaws(OCLs),thebreakdowninopinionwasalong
culturalaffiliation,especiallybetweenegalitariancommunitariantypesandhierarchical
individualists(Kahan,Braman,Monahan,Callahan&Peters,2010).Infact,identityisoften
moreimportantindeterminingworld-viewthanideology(Kahan,2012,Kahan,Braman,
Slovic,Gastil,&Cohen,2007).Peopleusuallyformopinionsandevaluationsthatlineup
withtheirculturalidentityanddefiningvalues(KahanandBraman,2006).Forexample,a
strongbelieverincapitalismandthepowerofinnovationwillbelesslikelytoratetherisks
ofclimatechangehighlyasdoingsowouldrequireacriticalevaluationoftheroleof
businessinsociety(Kahan,Braman,Slovic,Gastil,&Cohen,2007).Identityisdynamic
though,andwhengroupmembershipislesssalientindividualopinionswillnotadhereas
closelytogroup(Conover,1984).Framingclimatechangesolutionsasinnovativeandas
relatedtothefreemarketcandecreasepolarizationasacceptingclimatechangenolonger
presentsafundamentalchallengetoidentity(Kahan,2010).
IMPLEMENTS
COGNITIVEREFLECTIONTEST
Thegoalofthecurrentstudyistolookattheinteractionsbetweenlogical
propensityandideologyandhowthisinformspoliticalinformationprocessing.Three
measuresofcognitionweretakenallmeasuringdifferentthings.Everyparticipanttookthe
CRT,andfromtheretheywereplacedononeoftwotracks.Theyeithertookanumeracy
test,whichisameasureofnumericalintelligence(Welleretal.2012,Kahanetal.,2013),or
tookalongerversionoftheCRT(SeeAppendixA).BothnumeracyandCRThavebeen
showntoincreasepolarization(Kahan&Stanovich,2016)sotheoriginalCRTwastakenas
Sernyak 13
ameasureofcognitivecapacityasthegreatestnumberofsubjectshadtakenthistest.Itis
unclearexactlywhatdistinctionscanbemadebetweenthesemeasuresofcognitionina
politicalcontext,butallseemtocaptureasimilareffect.Allresultswerecheckedwithboth
numeracyandtheinclusiveCRTandthesamepatternsemerged,therewasjustalarger
errorinvolved,asthesamplepopulationwassmaller.
WASONSELCTIONTASK
Forfurtherunderstandingofmotivatedreasoning,thecurrentstudyusedthe
WasonSelectionTask.ThetaskwascreatedbyWasonin1966andisadifficultlogicaltest
aboutconditionalrules.Thetaskisperhapsthe“mostinvestigatedexperimentalparadigm
inthepsychologyofreasoning”(Maktelow,2012).Thegeneralstructureisasfollows,
participantsareaskedtodetermineifthefollowingruleistrue:“IfPthenQ”.Theyarethen
presentedwith4cardstoturnover,“P”,“Not-P”,“Q”,“not-Q”.Theyshouldturnoverthe“P”
cardandthe“not-Q”card,butparticipantswilloftenfollowtheirheuristicanswerand
matchthe“Q”cardwiththeprompt,turningitover,oroftentimestheywillnotturnover
the“not-Q”card(Wason,1966,Cox&Grigss,1982).Itisadifficulttask,withlowbaseline
ratesofcorrectreasoning.Oneofthereasonsithasbeensuchausefultooltostudy
reasoningcapacityisthatitisverycontextdependent.Forexample,peoplearebetter
whencontentevokesrelevantknowledgefrommemory,forexampletransportationuse
(ManktelowandOver,1990).Performanceonthetaskwasalsoimprovedwhenitwas
policingsocialinteractions(Tooby&Cosmides,1992).Thisresulthashadimportant
impactontherealmofevolutionarypsychology.
Motivationhasalsobeenshowntoincreaseperformanceonthetest.Subjectswere
morelikelytocometothecorrectconclusionthatthepromptwasincorrectwhenthe
Sernyak 14
prompt’smessagewasundesirable.Iftherulestatedtheyweremorelikelytohaveanearly
deaththeywouldbemorelikelytoturnoverthecorrectcardstodisprovethisclaim,butif
theruledidnotimplyanearlydeaththeywerelesslikelytotrytodisproveit(Daweson,
Gilovich,Regan,2002).Thisresultwasimportantindemonstratingthewaysinwhich
confirmationbiascomesabout.Forthisreason,theWasonSelectionTaskstandsasa
promisingtooltouseinthestudyofpoliticalbias.
CONJUNCTIONFALLACY
Theothertoolusedinthisthesisistheconjunctionfallacy,whichwasusedtolook
atnewareasinwhichbiasedcognitionmayoccur:whethertheyviewsimilarly
ideologicallymindedpeopleasmorallysuperior.Thereisreasontobelievethiswouldbe
thecase,asthereisarobustliteraturethatshowshumansviewpeoplemoresimilarlyto
themmorefavorably.In-groupout-groupbiasisacommonlyheldphenomenonthathas
beenappliedtopoliticsinthepast.Peoplefavorthosewhotheyviewasintheirgroup,and
aremorelikelytogivemoneytotheminexperimentalsettings.Thishasbeenobservedin
evenminimalgrouppairssuchaswhethertheylikeacertainpainting(Tajfeletal.,1971).
Groupsgivepeopleasenseofself-esteemandprovideaplaceintheworld,andthe
enhancementofone’sowngroupoverothersprovidesself-satisfactionbutalsoleadstothe
in-groupout-groupdynamic.Groupfavoritismhasbeenshowntoapplytobelief(Skitkaet
al.,2005).Yettheconjunctionfallacyprovidesanopportunityforamoreinterestingtest
thansimplyoneofwhetherdemocratslikerepublicans,astheanswertothatwouldmost
likelybethattheypreferthemselves.Theconjunctionfallacyprovidesawindowintothe
narrativerichnessofperson’srepresentations.
Sernyak 15
Theconjunctionfallacyisanexampleofabiasthatoccurswhenpeopleareover
reliantonintuitiveprocessestomakejudgments,orheuristics.Aheuristiccommonlyused
formakingevaluativejudgmentsistherepresentativenessheuristic(Tversky&Kahneman,
1983).Theheuristicisusedtomakeajudgmentbyevaluatingthedegreeof
correspondencebetweenandoutcomeandapreviouslyheldmodel(Tversky&Kahneman,
1983).Forexamplerobinsandpenguinsarebothbirds,butarobinismorerepresentative
(Tversky&Kahneman,1983).Thisheuristicallowsforeasyjudgmentswithoutneedingto
bringincognitivelyexpensiveSystem2processesinplay.(Kahneman,2011).
TverskyandKahneman,intheirfoundational1983study,foundthatratherthanuse
basiclawsofprobability,peoplewillrelyontheseheuristicstodeterminehowlikely
eventsare.Oneofthemostbasictenantsofprobabilityisthattheprobabilityofasingle
eventcannotbelessthantheprobabilityofthatandanotherevent.Subjectsweregiventhe
followingdescription:“Lindais31yearsold,single,outspoken,andverybright.She
majoredinphilosophy.Asastudent,shewasdeeplyconcernedwithissuesof
discriminationandsocialjustice,andalsoparticipatedinanti-nucleardemonstrations”.
Thedescriptionwasmeanttoelicittherepresentationofafeminist.Participantswere
askedwhetheritwasmoreprobablethatshewasabanktellerorabanktelleranda
feminist.Ofcourseitismoreprobablethatsheisabankteller,asthatsetincludesthe
secondoption.Yet85%ratedLindaasmoreprobablyabanktellerandafeminist(Tversky
&Kahneman,1983).Thisisbecausetheevent,Linda,fitstherepresentationoffeminist
betterthanbankteller,andthestrengthofthecausallinkisoverwhelmingprobability
judgments(Thüring&Jungermann,1990).
Sernyak 16
Theconjunctionfallacycanbeusedasatestofnarrativerichness,becauseforthe
fallacytooccurthesubjectmustviewthelinkbetweentheconditionanddescriptionas
strongandcausal.Forthisreasonithasoftenbeenusedasatooltodetermineto
robustnessofprejudice.Totestanti-atheistbiasastudywasdoneusingaconjunction
fallacypromptthatdescribeduntrustworthyandimmoralbehavior.Participantswere
muchmorelikelytofallfortheconjunctionfallacywhentheimmoralindividualwas
describedasanatheistasopposedtoothermajorreligions(Gervais,2014).Religious
peoplewerealsomorelikelytomaketheerrorthanatheistswere,althoughtheeffectstill
existed(Giddings&Dunn,2016).Informationpresentedtosubjectscouldalsomitigate
theseeffects,suchaslettingpeopleknowthatmostofthepopulationisatheist(Giddings&
Dunn,2016).Theconjunctionfallacywasalsousedtoshowthatthepublicholdsspecific
negativerepresentationsofscientists,despitetheirhighstatusinsociety.Theywereshown
tobethoughtofasimmoral,andwouldbreaknorms,buttheywerenotthoughtofas
deliberatelyevilorunfair(Rutiens&Heine,2016).Thecurrentstudyseekstousethis
implementandseewhatnegativerepresentationsaboundinapoliticalsetting.Thestudy
willlookattheeffectofideologyandintelligenceontheserepresentations.Theimmediate
goalistobegininvestigatingtheroleofintelligenceinmoralpoliticaljudgments,aswellas
narrativejudgments,lookingatpoliticalcognitionbeyondinformationprocessing.The
moregeneralpurposeofthispaperistoaddtotheliteratureonmotivatedreasoningand
begintoexplorepoliticalbiasinasimilarcontext.
METHODS
Thestudytestedmotivatedreasoningbyaskingavarietyofquestionsusedin
previousstudiesthatcanelicitbothheuristicandrationalresponses.Avarietyof
Sernyak 17
independentvariableswereassessedtotrytoparseoutwhatfactorscanpredictmotivated
reasoning.TasksusedweretheWasonselectiontasktomeasuremotivatedreasoningand
theconjunctionfallacytoassessbiasagainstpoliticalothers.Independentvariableswere
conditions,politicalideology,andcognitivecapacity.
POPULATION
ThesubjectsofthestudywereadiversesampleoftheUnitedStatesPopulation.
2800adultswererecruitedtoparticipatebyYouGov,andthesurveywasadministeredvia
thefirm’son-linetestingfacilities.Thesamplewas55%female,76%white,9.4%black,
and8%Hispanic.Themedianagewas47yearsold(SD=17).Participantswereaskedtheir
familyincomeona16pointscale:407declinedtorespond,andofthosewhodidthe
medianonthescalewasa5(SD=3.2),correspondingto$40,000-$49,00.Similarly
participantswereaskededucationlevelonascalefrom1(NoHighSchool)to6(Post
Grad):themedianscorewasa3(SD=1.4)correspondingtosomecollege.
STIMULUS1
ParticipantswereplacedintooneofthreeconditionstoansweraWasonSelection
Taskquestion.Ineachconditiontheywereaskedtoevaluatealogicalrule“IfP,thenQ.”
Theparticipantwasthenprovidedwithfourcardsasfollows:“P”,“Not-P”,“Q”,“not-Q”.
Thecorrectresponsewouldhavebeentoturnoverthe“P”cardandthe“not-Q”card,and
theywouldthenfindthattheprompttobefalse.
Thethreeconditionsaskedabouttaxbreaksfordifferenttechnologycompanies.In
thefirstconditionparticipantswereaskedtotesttheallegationthat:“Ifacompany
producedNano-Widgets,thentheIRSdeniedita‘novel-technologymanufacturer’tax
exemption.”Theparticipantswerethenaskedtoturnoverasfewcardsaspossibleinorder
Sernyak 18
todeterminewhetherornottheallegationwastrue.Theycardstheyturnedoverwere
recorded.
Figure1:ThecardsfacingtheparticipantintheNano-widgetWasonSelectionTaskcondition.
Figure2:TheundersideofthecardsintheNano-widgetWasonSelectionTaskcondition,whichthesubjectscouldseeiftheychosetoturnoverthecards.
Theallegationisfalseaccordingtothecards,whichbecomesclearafterturningover
card2.Followingalllogicalrulesaparticipantcouldeitherturncard3overfirst,thenturn
card2overandcometotheconclusionthattheclaimisfalse,ortheycouldjustturncard2
overandreachtheconclusion.Thisconditionconcernedanon-politicizedtechnologyto
haveareferencecondition.
Thelasttwoconditionsaskedforparticipantstousethesamelogicalprocesses,
excepttheallegationswereslightlymodified.Participantswereplacedeitherinaskeptical
Nano-widget
(1) (2) (3) (4)
ExemptionGiganto-widget Noexemption
(1) (2) (3) (4)
Exemption Nano-widget Noexemption Nano-widget
Sernyak 19
oralarmeddiscriminationcondition.Intheskepticalconditiontheallegationtheywere
askedtoinvestigatewas:“Ifascience-educationfoundationwas‘skepticalaboutclimate
change,’thentheDepartmentofEducationdeniedita‘science-education-foundation’
grant.”Thecardswereasfollows:
Figure3:Thecardsusedintheskeptical-discriminationconditionoftheWasonselectiontask.
Figure4:Theundersideofthecardsintheskeptical-discriminationcondition.
Inthealarmedconditiontheywereaskedtoinvestigatetheclaimthat:“Ifscience-
educationfoundationwas‘alarmedaboutclimatechange,’theDepartmentofEducation
arbitrarilydeniedita‘science-education-foundation’grant.”Thecardswerethesame
exceptthatalarmedandskepticalwereswitchedateveryspot(includingonthebacks).
Skeptical
(1) (2) (3) (4)
Alarmed Grant Nogrant
(1) (2) (3) (4)
Grant Nogrant SkepticalSkeptical
Sernyak 20
ThesamelogicalprocessesapplyinthealarmedandskepticconditionsasintheNano-
widgetcondition.
Thebreakdownwasasfollows:929subjectswereaskedaseparatequestionnot
discussedinthispaper,943subjectswereintheNano-widgetcondition,463subjectswere
intheskeptic-discriminationcondition,and465subjectswereinthealarmed-
discriminationcondition.Thedesignwasabetweensubjectsone,asparticipantswere
assignedtoonlyonecondition.
STIMULUS2
Participantswereplacedintooneofthreeconditionsandineachcondition
answeredoneconjunctionfallacyquestion.Allweregiventhesameprompt,whichwasas
follows:
Richardis31yearsold.Onhiswaytoworkoneday,heaccidentallybackedhiscarintoa
parkedvan.Becausepedestrianswerewatching,he got out of his car. He pretended to
write down his insuranceinformation.He then tucked the blanknote into the van’s
windowbeforegettingbackintohiscaranddrivingaway.
Laterthesameday,Richardfoundawalletonthesidewalk.Nobodywaslooking,sohetook
allofthemoneyoutofthewallet.Hethenthrewthewalletinatrashcan.
TheywerethengiventwodescriptionsofRichardandaskedwhichwasmorelikely.The
firstoptionremainedconstantthroughthethreeconditions:that“Richardisself-
employed”.InthereferenceconditiontheyweregiventheoptionofRichardasself-
employedandaconvictedfelon(ananswerwhich,accordingtothelawsofprobability,
mustbefalse).
Conditionswerecreatedtoindirectlytestwhetherpoliticalorientationwas
associatedwithdistrustinasimilarmannertoaconvictedfelon.Intheothertwo
conditionsthedescriptionswere:“self-employedandaverystrongsupporterofstrictgun
controllaws”and“self-employedandaverystrongopponentofstrictguncontrollaws”.
Sernyak 21
Thebreakdownwasasfollows:923wereintheex-convictcondition,918wereinthe
supportcondition,and959wereintheopposecondition.Thedesignwasabetween
subjectsone,asparticipantswereassignedtoonlyonecondition.
*Thetestwasadministeredfromthe8thofNovembertothe16thofDecember,2016.
INDEPENDENTVARIABLES
ExperimentalConditionwasanindependentvariablebetweensubjects.Besides
condition,afewvariableswereusedtopredictperformanceonthestimuli.
IDEOLOGY
Thefirstvariablewassubjects’politicalaffiliation.Twoquestionsaboutpolitical
affiliationattheendofthesurveyinthedemographicssectionwereusedtocreatea
measureofpoliticalideology.Forthefirstmeasuresubjectswereaskedtoranktheir
ideologyonafive-pointscale,rangingfromveryliberal(1)toveryconservative(5).Of
thosewhoanswered,themedianwas3(moderate),themeanwas3.1(SD=1.1),between
moderateandconservative.Thesecondmeasurewasaseven-pointscaleasking
participantstoranktheirpartyidentification,fromStrongDemocrattoStrongRepublican.
Ofthosewhoansweredthequestionthemedianwasa4(independent)andthemeanwas
3.8(SD=2.1),betweenleandemocraticandindependent.69subjectsskippedbothofthese
questionsandwereremovedfromtheanalysis.46subjectsansweredthefirstquestionbut
notthesecondquestionsotheywereassignedthemedianscoreonthesecondquestion.
211subjectsansweredthesecondquestion,butnotthefirstquestionsotheywere
assignedthemedianscoreonthefirstquestion.Theanswerstothetwomeasureswere
standardized(median=0,SD=1),andthenaddedtogetherandstandardizedagain.The
Sernyak 22
EntireSample
resultingscorewasusedasameasureforideologyhenceforthcalledideologyscore(for
short:ideo)(Cronbach’sAlpha=.76).
Figure5:Abreakdownofthesamplepopulation’sideologicalscores(calledideo).
CLASSICCRT
ShaneFrederick’soriginalCRT(Frederick,2005)testwasadministeredtoall
participants.Thetestisa3questionnumericinstrument,whichaimstotestcognitivestyle
andrelianceonlogic(forthetestquestionsandamoredetaileddiscussionoftheCRTsee
AppendixB).Itemresponsetheory(IRT)analysiswasusedtoweightthequestionsand
deriveascorefromthethree-questionimplement.IRTanalysisprovidesimproved
accuracyofanimplement,andcandifferentiatemorefinelythansimplyaggregatingthe
score(Primietal.,2015).Thisallowsformorepossiblescoresthantheoriginal3.
Nonetheless,theCRTisnotaperfectinstrumentevenafterIRTanalysis.59%ofsubjects
answeredzeroofthequestionscorrectly.Thisisslightlyhigherthanfoundationalstudies
MoreLiberal MoreConservative
Ideo
Sernyak 23
(Frederick,2005),butaslightlyhigherpercentageofzeroesmakessense,asthesubject
populationwasnottakenfromauniversity.ThescoreafterIRTanalysiswillbeknownas
CRTintheresultssections,andwasthechosenmeasureofcognitivefunctioning
(Cronbach’salpha=.69).Liberals(allparticipantswithideo<0)hadameanof.117(SD=
.80)ontheCRTscale,whileconservatives(ideo>0)hadameanof.085(SD=.76).
Figure6:Ontheleftisahistogramoftheentiresamplepopulation’sCRTscoresafterIRTanalysis.Ontherightisahistogramofliberalscoredensityoverlaidonahistogramofconservativescoredensity.
DATAANALYSIS
AvarietyofmodelswerecreatedfromthedatausingRStudioandGaryKing’sZelig
software.Logisticregressionsweredonewithallofvariablesinthesamemodel,inorder
tokeepthepowerofthemodel(Judd,2000).Foreachofthestimuli,conditionwasusedas
adummyvariable,witheachconditionbeingcodedasonewheninthatcondition,andthe
referenceconditionbeingactivatedwhenallconditionswerezero.Ideologicalscorewas
alsousedinthemodel,aswellasitsinteractionwithCRT.Correlationmatricesarein
EntireSample PoliticalSub-GroupSamples
CRTCRT
Liberals(ParticipantswithIdeo<0)
Conservatives(ideo<0)
Sernyak 24
AppendixCandgraphsweremadewithpredictionsfromthemodels.MarkovChainMonte
Carlosimulationswereusedtodetermineconfidenceintervals(King,2000).
IntheWasonSelectiontasktherewerethreeconditionscodedasdummyvariables,
nano-widget,climate-alarmist,andclimate-skepticdiscrimination.Thefirstoutcome
recordedwascorrectanswer,orwhetherparticipantratedthestatementasfalse.Other
logicalprocessesweretrackedaswell.Modelswerecreatedtotrackwhethercorrectlogic
wasusedtogettherightanswer,orwhetheranycardwasmorelikelytobechosen.
Skippedanswerswerecodedaswrong,asitshowedaninabilitytodealwiththesubject
matter.Fewpeopleskippedineitherconditionsoitdidnotimpacttheresultssignificantly.
Independentvariablesinthemodelwerecondition,ideologyandCRT.
Fortheconjunctionstimulustherewerethreeconditionscodedasdummy
variables,convict,supporter,andopponent.Theoutcomewasrecordedandamodelwas
createdtopredictwhatpercentageofpeoplewouldgetthepromptrightintheactual
population.Theonlydependentvariablelookedatwaswhethertheoutcomewascorrect.
Skippedanswerswerecodedaswrong.Independentvariablesinthemodelwere
condition,ideologyandCRT.
HYPOTHESIS
THEWASONSELECTIONTASK
TheWasonSelectionTaskisstrictlyalogicaltest,soresultsconsistentwiththe
Idenity-protectiveCognitionThesiswoulddemonstratesometypeofideological
motivation.Theexpectationisthatifanorganizationfromaperson’sidentitygroupwere
beingdiscriminatedagainst,theywouldbemotivatedtoprovethatwrongastheydonot
wantthattobethecase.Yettherearesometangentiallyrelatedresultsthatsuggestthe
Sernyak 25
oppositemightoccur.Peopleoftenclaimthattheyarebeingdiscriminatedagainst(seeIRS
targetingcontroversy)butareveryunlikelytoadmitthattheyarediscriminating.People
willalsoviewtheirin-groupasbeingtreatedmoreunfairly(Hastorf&Cantril,1954)and
mightbecomplacentandnotmotivatedtocorrecttheassertionthattheyarebeingtreated
unfairly.Asaresult,itisnotentirelyclearhowpeoplewillbemotivated,butthehypothesis
isthatinsomewaytheywillbe,whichwouldbecomeclearwhenresultsweremodeled.A
nullresultwouldmeaneitherthatthepromptdidnotinstigatemotivatedcognitionor
politicalbiasdidnotoccurforthistopic.Inthistask,participantsarenotnecessarily
motivatedtoprotecttheiridentityinthesamewaytheyarewhenevaluatingevidenceof
climatechange.Forthisreasontheirperformancemaynotdependheavilyonideology.
Figure7:TwopotentialpredictedresultsaredisplayedfortheWasonselectiontaskprompt.Ontheleftisapredictionifsubjectsdonotundergomotivatedreasoning,asperformancedoesnotincreasedependingonideologyinpolarizedconditions.Ontherightisapredictionforabroadlyapplicablemotivatedreasoningthesis.Subjectsaremuchbetterandworsedependingonideologyandcondition.
CONJUNCTIONFALLACY
Fortheconjunctionfallacyquestions,conditionshouldaffectthecorrectresponse
percentage.Theex-convictconditionshouldprovideagoodbaselineforhowa
PercentCorrect
Ideo
LiberalDesiredOutcome
ReferenceCondition
ConservativeDesiredOutcome
NoMotivatedReasoning MotivatedReasoning
Ideo
Sernyak 26
representativestorybehavesamongthepopulation.Theguncontrolconditionsmayor
maynotbeasrepresentativeastheex-convictconditions.Ideologyshouldinfluencethe
outcomeofthesequestions,asmostlikelypeoplewillviewRichardasmorenegativelyifhe
holdstheoppositepoliticalbeliefsasthem.Ifsothismeanspeoplearemorelikelyto
portrayideologicaldissimilarpeopleasimmoral.Itisveryunlikelythatpeoplewouldbe
morelikelytofallfortheconjunctionfallacyifRichardisdescribedashavingsimilarbeliefs
asthem.Ifthereisnoeffectofideologyonthepopulation’sviewofRicharditcouldeither
betheinstrumentisimprecise,orpeopledonotholdgenerallynegativeviewsofthosein
theoppositeparty.Whileinthisdayandagenationalpoliticshavebecomequitepolarized,
itisnotclearthatthisfervorandanimosityisheldonindividuallevel.Ifthenationhas
becomefullofzealots,theremaybeaverystrongbiasagainstthoseoftheopposite
ideology.Ifnot,lessdramaticresultswouldbeexpected.
Figure8:Ontheleftisapredictionforapopulationwheretherewasnotastronglynegativerepresentationofthosewithdifferingideology.Ontheleftisapredictionofasampleinwhichpartisansheldstronglynegativeviewsofthosewithdifferingideology.
PercentCorrect
NoBias
Ex-Con
IdeologicalBias
OpponentofGunControl
SupporterofGunControl
Ideo Ideo
Sernyak 27
ROLEOFCOGNITIVECAPACITY
Resultsshowingideologicalpolarizationwouldnotbeverysurprising,asthey
wouldfitinmostpoliticalbiasframeworks,includingtheSCT.Resultsconsistentwiththe
ICTwouldshowanincreasedpolarizationwithintelligence.TheWasonSelectionTaskis
veryhardandrequirestheuseoflogic,soitseemslikelyCRTwouldimproveperformance.
InanIdentityProtectionThesisthatindividualsaremotivatedtowilluselogictoaffirm
theirworldview.Ifresultsimprovedacrossallconditionsregardlessofmotivationtheway
inwhichinformationwasprocessedwouldmostlikelynotbeinordertoprotectidentity.
Figure9:Ontheleftmeasuresoflogicaltendencycorrelatewithincreasedpolarization.ThisresultfitswiththeIdentity-protectiveCognitionThesis.Ontherighttheuseoflogicimprovesperformanceacrossallconditionsandpolarizationdecreases.ThisresultfitswiththeScienceCognitionThesis.
Fortheconjunctionfallacy,resultscouldalsofollowsimilarpatternsasinfigure9.If
polarizationincreasesalongwithCRTtherecouldbeafewexplanations.Morelogical
peopleseembetteratfittinginformationtotheirworldview,andsoperhapstheymake
harshermoraljudgments.Theyalsocouldsimplybemorepolarizedingeneral.Whatever
thecase,itwouldbeclearthatthenarrativerichnesswouldincreaseforthosewithhigher
levelsofcognition.Infact,evenanullresultofCRTwouldbesignificant.Inprevious
SCTModelICTModel
PercentCorrect
CRT CRT
*solidlinesare
ideologically
agreeable
conditions
*dashedlinesare
ideologically
disagreeable
conditions
Sernyak 28
EffectofIdeology
studies,CRTwasshowntodecreasetheconjunctionfallacyintheLindaproblem
(Oechssler&Roider,2009).Or,ifmotivatedreasoningisnotafactor,performancemay
simplyincreaseacrosstheboard.
RESULTS
WASONSELECTIONTASK
Inthethreeconditions,nano-widget,alarmed,andskepticdiscrimination,noreal
effectwasseeninregardstocorrectansweroftheprompt,condition,andideology(see
Figure10).As,subjectswereaskedatruefalsequestionandresultshoveredaround50%
regardlessofconditionitseemedasthoughanswerswererandom.Thiswasbackedupby
thefactthat609participantshadpickedananswerwithoutturningoveranycards,and
theyansweredatsimilarratesasthosewhohadactuallyturnedovercards.
Figure10:LogisticregressionofallthreeconditionsfortheWasonselectiontask.Thedotsrepresentthemeanexpectedvalueandtheerrorbarsrepresent95%confidenceintervalsderivedfromanMCMCsimulation.About50%gettheanswercorrectregardlessofideologyorperformance.
Ideo
PercentCorrect
Alarmed-discrimination
Skeptical-discrimination
Nano-widget
Sernyak 29
Skeptical-discrimination Alarmed-discrimination
Liberal(ideo=-1)
TheeffectofCRTwasexaminedaswell,andtheredidseemtobeapositive
correlationbetweenCRTandperformance.Inallthreeconditions(seefigure10),CRT
improvedthechanceofgettingtheanswercorrect,althoughtheimprovementisslightand
therateneverrisessignificantlyabove50%.Thereisnosignificantinteractionbetween
ideologyandperformance,asliberalsandconservativesimprovesimilarlyoverevery
condition(aquicknote:inthesimulationsinfigure11andinallsubsequentsimulations
“liberals”arethepredictedvaluesofthemodelswhenideoissetto-1,whichis1SDleftof
centerinthesurveypopulation.Bluelinescorrespondtothis.Redlines,and
“conservatives”,correspondto1ontheideoscale,oroneSDrightofcenter).
Figure11:ThethreefiguresdemonstratetheeffectofCRToneachcondition.Ineachconditionbothliberals(liberal(ideo=1)inblueinthisgraphandallfuturegraphs)andconservatives(graphedinred)improveslightlyoverCRT,butnointeractioneffectsareobserved.
PercentCorrect
CRT CRT
CRT
Nano-widget
Conservative(ideo=1)
Sernyak 30
EffectofIdeology
Nextparticipant’slogicalabilitywastested:whethertheyturnedovertheright
cardsandgottheanswerright.Veryfewparticipantsusedthecorrectlogictogettothe
answer.Ofthe1210participantsthatansweredthediscriminationquestions,only4%
wereabletoidentifythecorrectcardstoturnover,aswellassubsequentlyanswerthe
questioncorrectly.Amodelwascreatedwiththecorrectansweralongwithcorrectlogical
processesastheoutput.Neitherideologynorconditionseemedtohaveaneffect,as
baselineratesofansweringthequestioncorrectlyweresolow(seefigure12).
Figure12:Noeffectsareobservedofideologywereobservedinpredictedvaluesofgettingtheanswercorrectlythroughtheproperlogicalprocesses.Ineachconditionmeanpredictedratesarebelow10%.
CRTwascorrelatedwithimprovedperformanceinthiscase,buttherewasagain
littledifferencebetweentheconditions(seefigure13).Ratesofsolvingthequestion
correctlywereprettylowupuntilveryhighlevelsofCRT.Themaxscoresomeonereceived
inthestudypopulationwasa1.8,andatthatscorenoconditionwassignificantlyabove
10%.Nosignificantinteractionoccurredbetweenideologyandintelligence,asineach
conditionconservativesandliberalsimprovedsimilarly.
Ideo
PercentLogicalandCorrect
Sernyak 31
Liberal
Figure13:Thesegraphsmodelcorrectanswersarrivedatlogicallyovereachconditionamongconservativesandliberals.Baselineratesareverylow,andwhileCRTimprovesperformance,thereislittledifferencebetweenconditions.
Thenextstepwastotrytoexpandthecriteriaforwhatwasconsideredlogical,
allowingforsomeextraneouscards,butloweringthebardidnotproduceanynew
patterns.Finally,modelswerecreatedforeachcard,toseeifanysystematiclogical
processesemerged.Therewassomerandomvariationbutnosignificanttrendsemerged
surroundinganyofthecards.Allthecardswereturnedoverabout2/3ofthetime,further
reinforcingtheideathatperformanceontheselectiontaskwaslargelyrandom.
CONJUNCTIONFALLACY
Theconjunctionfallacyprompthadhigherratesofsubjectsarrivingattheright
answer,asitwasconsiderablysimpler.36%ofpeoplegottherightanswerintheex-con
CRT
CRT
CRT
PercentLogicalandCorrect
Skeptical-discrimination Alarmed-discrimination
Nano-widget
Conservative
Sernyak 32
EffectofIdeology
condition(andtherefore64%fellforthefallacy),while74%gottherightanswerinthe
guncontrolsupporterconditionand70%gottherightanswerintheguncontrolopponent
condition.Therewereslightideologicaleffectsinthethreeconditions(seefigure14).For
allideologies,ratesofconjunctionfallacywerehighestintheex-convictcondition.Liberals
showedmorebiasedthanconservatives,astheywere8%(SE=3%)morelikelytofallfor
theconjunctionfallacywhenRichardopposedguncontrolthanwhenhesupportedit(see
figure15).Conservativeswerenotmorelikely(0%,SE=2%)toviewasupporterofgun
controlasimmoralbuttheresultwasnotsignificant.Themostdramaticresultwasactually
foundinthereferencecondition.Ratherthanactasabaselineratetocompareagainstthe
ex-convictconditionvariedbetweenliberalsandconservatives,asconservativeswere12%
(SE=3%)morelikelythanliberalstorateRichardasmorelikelytobeanex-convictand
self-employed(seefigure16).
Figure14:Logisticregressionofallthreeconditionsfortheconjunctionfallacy.Thedotsrepresentthemeanexpectedvalueandtheerrorbarsrepresent95%confidenceintervalsfromanMCMCsimulation.Ahigherpercentrightmeantthattheimmoralbehaviordescribedwaslessrepresentativeofthecondition.
Oppose
Support
Ex-con
Ideo
PercentCorrect
Sernyak 33
SimulationsofEx-convictCondition
PercentCorrect
Figure15:Densityofsimulationsofliberalsoverthetwoguncontrolconditions.Basicallyacrosssectionofthegraphinfigure14atis=-1.1000MCMCsimulationswererecordedandthengraphed.Themeanofthesupportconditionwas82%(95%CI=79%to86%).Themeanoftheopposeconditionwas75%(95%CI=70%to78%).
Figure16:Densityofsimulationsoftheex-convictconditionforbothliberalsandconservatives.Thereferenceconditionactuallydisplaysinterestingresults,asconservativesaremorelikelytocommittheconjunctionerrorintheex-convictcondition.Themeanofliberalswas43%(95%CI=39%to49%).Themeanofconservativeswas31%(CI95%=27%to35%).Inthishistogramandinfuturehistogramsliberalsareinblueandconservativesareinred.
SimulationsofLiberals
SupportOppose
PercentCorrect
8±3
12±3
LiberalConservative
Sernyak 34
Oppose
Suppose
Liberal
CRTincreasedpolarizationintwoofthethreeconditions(Seefigure17).Inthe
opposecondition.Intheopposeguncontrolcondition,CRTslightlyincreasedperformance,
buttherewasnosignificantpolarizationineitherloworhighCRTindividuals.Intheother
twoconditionspolarizationincreasedasCRTperformanceonlyimpactedliberals.Infact,
conservativesdidnotsignificantlyincreaseperformancewithCRTinanycondition.
Figure17:AlookattheeffectsofCRTperformanceineachconditiononbothliberalsandconservatives.Themostdramaticresultisfoundintheex-convictcondition.
Eachconditionwasobservedinmoredetail,toseeiftherewereinteractioneffects
betweenCRTperformanceandideology.Thequestionaskedwasasfollows:doliberals
improvemorefromlowCRT(CRT=-.5)tohighCRT(CRT=1)thanconservativesdo.Four
PercentCorrect
Ex-ConCRT
CRT
CRT
Conservative
Sernyak 35
OpposeCondition
EffectofCRTinOpposeCondition
ChangeinPercentCorrect
simulationsweredone1000timesineachcondition.OnesimulationwasdoneforlowCRT
liberals,oneforhighCRTliberals,oneforlowCRTconservatives,andoneforhighCRT
conservatives.Theimprovementamongtheideologicalgroupwastrackedandcompared.
Figure18:Simulationsoffourtypesofindividualsintheopposecondition:highandlowCRTliberalsandconservatives.CRTslightlyimprovesperformanceinbothideologies,andpolarizationdoesnotincrease.LowCRTindividualsarerepresentedbydottedlines,highCRTindividualsbysolidlines.
Figure19:ThehistogramdisplaystheimprovementofpercentcorrectasCRTgoesfromlowtohighinthesimulations.Inthiscondition,bothimproveslightlyonaverage,butthereisnointeractionbetweenideologyandCRT.Liberalsimproveby4%(95%CI=-4%to11%)andconservativesimproveby2%(95%CI=-6%to8%).
PercentCorrect
ConservativeHighCRT
ConservativeLowCRTLiberalLowCRT
LiberalHighCRT
LiberalConservative
Sernyak 36
SupportCondition
PercentCorrect
ChangeinPercentCorrect
EffectofCRTinSupportCondition9±5
PolarizationdidincreaseathigherCRTscoresinthesupportcondition(seefigure
20).LiberalsincreasedperformanceinthisconditionathighCRTscores,while
conservativesremainedconstant(seefigure20).Also,polarizationonlyoccursathigh
levelsofCRT.AtlowlevelsofCRTliberalsperformedequallytoconservatives.Athigh
levels,liberalswere9%(SE=4%)morelikelytogetthequestioncorrect.
Figure20:Simulationsoffourtypesofindividualsinthesupportcondition:highandlowCRTliberalsandconservatives.CRTimprovesperformanceinliberals,andpolarizationdoesincrease.
Figure21:ThehistogramsdisplaytheimprovementasoflowtohighCRTsimulations.Inthiscondition,onlyliberalsimprovewithhigherlevelsofCRT.Liberalsimproveby7%(95%CI=0%to13%)andconservativesgetworseby2%(95%CI=-10%to5%).
LiberalHighCRTLiberalLowCRT
ConservativeHighCRT
ConservativeLowCRT
LiberalConservative
Sernyak 37
PercentCorrect
Ex-conCondition
EffectofCRTinEx-conCondition
ChangeinPercentCorrect
Themostdramaticresultswereintheex-convictcondition.Polarizationincreased
athigherlevelsofCRT(seefigure22),ashighscoringliberalswere22%(SE=5%)less
likelytomaketheconjunctionfallacyerrorthanconservatives.Thedifferencebetweenlow
scoringCRTindividualswasonly7%(SE=4%).Increasedpolarizationoccurredasonly
liberalssawanincreaseinperformancewithhighCRTlevels(seefigure23).
Figure22:Simulationsoffourtypesofindividualsintheex-concondition:highandlowCRTliberalsandconservatives.CRTimprovesperformanceinliberals,andpolarizationdoesincrease.
Figure23:ThehistogramsdisplaytheimprovementasoflowtohighCRTsimulations.Inthiscondition,onlyliberalsimprovewithhigherlevelsofCRT.Liberalsimproveby16%(95%CI=6%to22%)andconservativesimproveby1%(95%CI=-7%to8%).
LiberalHighCRT
LiberalLowCRT
ConservativeLowCRT
ConservativeHighCRT
15±8
LiberalConservative
Sernyak 38
DISCUSSION
WASONSELECTIONTASK
PerformanceacrosstheboardintheWasonSelectionTaskwasverylow.Thetask
appearedtoohard,andevenunderconditionsthatnormallyproducebetterperformance,
fewwereabletodothetask.Therewasnodiscernableeffectofideology,anditseemslike
peoplewereguessingmorethansolvingtheproblem.TherewascorrelationwithCRT
performance,butevenathighlevelsofCRTsofewpeoplearegettingtheanswerrightitis
difficulttodrawanyconclusionsfromtheresults.AtalmostalllevelsofCRT,theexpected
percentcorrectintheregressionwaslessthan10%.OnlyatthehighestlevelsofCRTdidit
riseabovethatlevel.Andintermsofsimplygettingtheanswerright,theresultnever
reachedsignificantlyabovechance.
Theresultsmayhavebeendrivendownbythefactthatthequestionwastowards
theendofaverydifficulttest,andpeoplemaynothavegiventhatmucheffortonthe
problem.Manydidnoteventurnoveranyofthecardsandjustansweredatchance.Even
loweringthebarandlookingatlogicalprocesses,todeterminewhetherpeoplewereable
togetclosetotherightanswerwasnotcorrelateddiscernablytoideology.Ultimately,it
wasdifficulttotellwhetherornotthepromptactivatedmotivatedreasoning.
CONJUNCTIONFALLACY
Therewasaslightresultbetweenthetwopoliticizedconditions,butthereisclearly
notastrongrepresentationbetweenimmoralbehavioralandpolicypositionongun
control.Notonlywastherenotmuchdifferencebetweenthetwoconditions,butalsothe
rateoffallingfortheconjunctionfallacywassolowtheredoesnotseemtobearelevant
representation.Noanimosityseemstobecapturedbythisresult,althoughtheslight
Sernyak 39
differencedoespointtosomeeffectofideology.Thismaybebecauseliberalsand
conservativesdonothavestrongnegativerepresentationsofeachother,andareonly
slightlybiased.Whiletheremaybesomein-groupout-groupbias,thereisnotastrong
negativeportrayaloftheotherside.Theviciousnessmaybelefttothepoliticians.
Theotherpossibleexplanationisthattheconditionsweretoounrelatedtothe
prompt.Inthisscenariothereisstillasstronglinkbetweenimmoralbehaviorandopposite
ideology.Inthiscase,ideologywouldbeviewedasamoralcharacteristic.Thestudy
implementwasunabletounearththisrepresentationthough.Perhapsthelinkbetween
onepolicypositionandimmoralbehaviorwastootenuous.Amoresalientandrelevant
representationmaybeneeded.Ideologyisanall-encompassingmoralundertakingandthe
oneelementitselfisnotcloselyrelatedtoimmoralbehavior.Orperhapsguncontrolwas
notthebestfocus,asitmaybeconsideredacharacterflawbutnotlinkedtopettycrime.
Itisunclearastowhythebiaswasonlyfoundinliberals,asconservativesanswered
atsimilarratesinthetwoguncontrolconditions.Mostresultshaveeitherfoundequalor
greaterbiasinconservatives,soitisabitsurprisingtofindtheeffectonlyinliberals.This
maybebecauseliberalshaveastrongerassociationbetweenimmoralbehaviorandgun
control,astheyseethoseopposedtoitascomplicitinmassshootings.Incasessuchas
climatechangethismaynotbethecase.Liberalsalsomightbemorelikelytoview
conservativesasimmoralingeneral.Therealsomaysimplyhavebeenceilingeffectsasthe
rateofconjunctionfallacywassolowtobeginwith.Thistopicshouldbeexploredmore,
andpotentialwaystodosoarementionedinthefuturedirections.
Themostinterestingresultsareintheinteractioneffectsbetweenideologyand
intelligence.Unlikepreviousstudies,polarizationinresultsdoesnotoccurduetobiased
Sernyak 40
informationprocessing.Instead,theconjunctionfallacyisatestofnarrativerichness.CRT
normallyincreasesperformanceonthetest(Oechssler&Roider,2009),assubjectsareless
reliantonintuitivejudgmentsandaremorelikelytouselogic.Ifratesofconjunctionfallacy
increaseorstayconstant,thenmostlikelythelinkbetweentheconditionandimmoralityis
gettingstrongertoovercomethepropensitytouselogic.
Inthesupportsguncontrolcondition,CRTimprovedperformanceinliberals.Very
fewliberalssawasupporterofguncontrolasanimmoralperson,andevenfewer
intelligentliberalssawtheconnectionasrepresentative.Republicansthoughdidnot
increasewithCRT,suggestingthatthenarrativerichnessoftheaccountmayhave
increasedwhichoffsetnaturalimprovementsthatwouldbeexpectedinmorelogical
people.
Intheopposesguncontrolcondition,CRTimprovedperformancemodestlyinboth
liberalsandconservatives.Ifthatconditionhadfollowedthepatternsofthesupportand
referencecondition,onewouldexpectconservativestoimprovewithCRT,butnotliberals,
increasingpolarization.Thiswouldalsofollowpreviousresultsofideologicallymotivated
cognition,asliberalsandconservativeshaveshownsimilarbiasesinmoststudies.Yet
therearedistinctdifferencesbetweenideologicallymotivatedcognitionandthebias
associatedwiththeconjunctionfallacy.Itisunclearwhetherthereisafundamental
differenceinthenarrativeaccountsofconservativesandliberals,orwhetherthis
implementwastoosubtle.Ceilingeffectsmayalsohaveplayedaroleincapping
conservativeimprovementinthecondition.
Themostdramaticinteractionbetweenideologyandintelligencewasfoundinthe
referencecondition.PolarizationincreaseddramaticallywithincreaseinCRTscores,
Sernyak 41
suggestingthatthenarrativerichnessofconservativesincreased.Thiscouldhave
implicationsinbiasandprejudice.Smarterpeoplearebetteratapplyinglogictofita
worldview,andthiscouldallowthemtocreatemorenarrativelyrichaccountsofpeople.In
thisexampleconservativesmaybebiasedagainstex-cons,andviewthemasimmoral,
slightlymorethanliberalsdo.Thiswouldfitpreviousfindingsasconservativesaremore
reliantonsocialinstitutionsandsocietalorder(Jostetal.,2003).AsCRTincreases,liberals
aremorelikelytolookpasttheirlessnarrativelyrichaccount,andcometothecorrect
conclusionthatRichardismorelikelytonotbeanex-convict.Smartconservativesthough,
havecreatedamorerobuststoryinwhichex-convictsareimmoral,andthereforetheir
performancedoesnotincreasedespitetheirnaturalinclinationtouselogic.Inthefuture
thistopicandtheoryshouldbeinvestigated.
FUTURESTUDY
Hereisaproposedtesttobegiventoademographicallysimilargroupofparticipants:
OriginalCRT:
1. Abatandaballcost$1.10intotal.Thebatcosts$1.00morethantheball.Howmuchdoestheballcost?[correctanswer=5cents;heuristicanswer=10
cents]
2. Ifittakes5minutesforfivemachinestomakefivewidgets,howlongwouldittakefor100machinestomake100widgets?[correctanswer=5minutes;
heuristicanswer=100minutes]
3. Inalake,thereisapatchoflilypads.Everyday,thepatchdoublesinsize.Ifittakes48daysforthepatchtocovertheentirelake,howlongwouldittakefor
thepatchtocoverhalfofthelake?[correctanswer=47days;heuristic
answer=24days]
NewCRTQuestions:
1. Thereathreefulldecksofcardsonatableandadealerisdrawingcards.FromDeck1hedrawsthreeblackcards.FromDeck2hedraws2blacksand1redcard.From
Deck3hedrawsthreeredcards.Youwanttopickablackcard.Whichdeckshould
youdrawfrom?[correctanswer=3;heuristicanswer=1]
2. WhenIwas6mysisterwashalfmyage.Iamnow50,howoldisshe?[correctanswer=47;heuristicanswer=25]
3. Youaretoldtodesignnewarmorforthemilitary,andyouneedtodecidewheretoputbulletproofmaterial.Peoplemostlygetshotinthearmsorthehead.Doctors
noticethatmostinjuredpeoplecomingbackfromthewarhavewoundsintheir
Sernyak 42
arms.Whereshouldarmorbereinforced?[correctanswer=head;heuristicanswer
=arms]
ConjunctionFallacyQuestion1:
“Richardis31yearsold.Onhiswaytoworkoneday,heaccidentallybackedhiscar
intoaparkedvan.Becausepedestrianswerewatching,he got out of his car. He
pretended towrite down his insuranceinformation.He then tucked theblank
note into thevan’swindowbeforegettingbackintohiscaranddrivingaway.”
Whichismorelikely?
a) Richardisself-empolyedb) Richardisself-employedand(1ofthreeconditions)
Thethreeconditionswouldbe“ex-convict”,“ex-convictwhowasarrestedforcorporate
fraud”,and“ex-convictwhowasarrestedfordrugdealing”
WasonSelectionTask:
Thepromptwouldasksubjectstoevaluatetheclaimthatafewpoliticianshadsent
outtweetsrecentlyandnoneoftheonessentoutby_______hadbeenfactual.Thethree
politicianswouldbePaulRyan,BarrackObama,andDavidCameronasareference
condition.IntheTrumpcondition,thecardswouldreadObama,fact,Ryan,lies.Thecorrect
cardstoturnoverwouldbefactandRyan,andtheassertionwouldturnouttobefalse.
ConjunctionFallacyQuestion#2:
“Sarahwaswalkingdownthestreetwhenamaninfrontofherdroppedhiswallet.Shewas
abouttocalltohimwhenshesawthatthewallethad$200init.Shekeptthemoney,and
threwthewalletinthetrash.”
Whichismorelikely?
c) Sarahisaclerkd) Sarahisaclerkand(1ofthreeconditions)
Thethreeconditionswouldbe“atheist”,“liberal”,and“conservative”
Base-rateNeglect:
“Ofthe_________population,1in10,000areterrorists.TheFBIhasawatchlisttosurvey
suspectedterrorist.Ifapersonisaterroristthereisa99%chancetheywillbeonthelist.If
thepersonisnotaterroristthereisa99%chancetheywillnotbeonthelist.
Apersonrealizestheirneighborisonthelist,howlikelyaretheytobeaterrorist?”
Pickone:
(0-10%,10-20%,20-30%,30-40%,40-50%,50-60%,60-70%,70-80%,80-90%,90-100%)
Sernyak 43
Thethreeconditionswouldbedifferentpopulations:alt-right,Muslim,andgeneral.
Thecorrectansweris0-10%buttheheuristicansweris90-100%.
Thefuturestudyproposedwillbeinasimilarformatandwillseektoclarifymany
ofthequestionsthatcameupinthisstudy,andtofurtherexaminemotivatedreasoning
andpoliticalbias.AgainthesurveywillbestartedwiththeoriginalCRT.Wearealso
proposingourownquestionsthathavethepotentialtobeaddedtotheCRT.Findingmore
questionsthatcorrelatewiththeoriginaltest,andthatcanbeusedtoaddorreplacethe
testwillbeimportantforfutureresearch,sowewanttotryafewnewonesinthestudy.
ThesequestionsmostlikelyeasierthantheonesonthecurrentCRT,butthehopeisthat
theycorrelatewithincreasedCRTperformance,butprovidetraitdiscriminationatlower
levels.
Nextwewishedtoinvestigateprejudiceandworldvieweffects.Buildingoffofthe
resultsintheex-convictcondition,wehopetolookatrulebreaking,andtheeffectsof
worldview.Theimplementwouldbesimilartotheconjunctionfallacyquestioninthe
study,butallthreeconditionsallaboutrulebreaking.Thefirstreferenceconditionwould
be“ex-convict”,thesecondconditionwouldbe“ex-convictwhowasarrestedforcorporate
fraud”,andthethirdconditionwouldbe“ex-convictwhowasarrestedfordrugdealing”.
Thisresultwouldmostlikelybepolarized,asliberalsaremorecommunitarianandare
waryofcorporategreed,whileconservativesaregenerallymoreanti-drugs.The
hypothesiswouldbethatthispolarizationwouldincreasewithCRTaswell.Thisquestion
wouldcontinuetoinvestigatethenarrativerichnessofintelligentpartisans.
ThenextquestiononthetestwouldbeathreeconditionWasonSelectionTaskas
usedinthesurvey,butwithsimplerconditionsfocusedonCheaterDetection.The
Sernyak 44
hypothesiswouldbethatsubjectswouldundergomotivatedreasoningtohelptheir
candidate,andpolarizationwouldincreasewithCRT.Theeffectmaybeweakerthoughas
inthisinstancetheyareprotectingacandidatetheymayagreewithratherthanprotecting
anopiniontheyfeelisnecessarytotheirgroupidentity.Cheaterdetectioncouldprovide
someinterestingresultsinthissituation,andcouldcontinuetodemonstratethe
relationshipbetweenconfirmationbiasandthetask.Whenapartisanhearsthatthe
opposingcandidatelies,theymaybelikelytoacceptthatresultandwillnotsearchfarther.
Thissimplerversionisproposedtoseeifthetaskisevenaviabletoolforbiased
informationprocessing.Thiswouldbeahelpfuldiscoveryasthetaskhasbeenusedinthe
pasttoexaminebiasinthepopulationandinjudges,sotheusefulnessofthetestshouldbe
examined.
Therewouldthenbeanotherconjunctionfallacyquestion,againaboutimmoral
behavior,andthistimetryingtodeterminewhetherpeopleviewothersofdiffering
ideologyasimmoral.ThedescriptionwouldbeofanimmoralSara,andsubjectswould
thenbeaskedifshewasmorelikelytobeasalesclerk,orasalesclerkandoneofthree
conditions:aliberal,aconservative,oranatheist.Astheindirectmethoddidnotworktoo
wellinproducingrichrepresentationswethoughtwewouldtrybeingmoredirectinthe
implement.TheeffectofCRTandofintelligencewouldbecomeclearerifthemoredirect
promptismorerepresentative.Thisisnotnecessarilythecase,butwouldbeinterestingto
check.
Finally,wewantedtotestlogicalprocessesinanothercontext,sowecreatedabase
rateneglect(Bar-Hillel,1980)problemtotestmotivatedreasoning.Peopleoftenneglect
thebase-rateofthepopulationandreportahighprobabilityoftheeventoccurring,inthis
Sernyak 45
casethatthepersonisaterrorist.Thegoalwouldbetoseeiflookingatdifferentgroups
wouldalterpeople’sabilitytoreasonproperly,andcontinuetolookatmotivatedreasoning
injudgmentsofothers.Thepredictionwouldbethatconservativeswouldbemorelikelyto
ignorethebase-rateintheMuslimcondition,asthatwouldfitwiththeirworld-view.
Liberalswouldbemorelikelytoignorethebase-rateinalt-rightconditions,asthatresult
wouldfitwiththeirworld-view.
Thestudyaimstoexploreotherareaswheremotivatedreasoningmaybeafactor
besidesinformationprocessing.Itislogicaltoassumeitwouldberelevantincheater
detection,asthisinvolveslogic,butthesenewinstrumentswouldalsocontinuetoexplore
whethermotivatedreasoningisrelevantinhowotherpoliticalgroupsareviewed.The
questionremainswhethertheAmericancitizenviewsotherpoliticalviewsas
fundamentallylessmoral,andismoreexpectantofbadbehavior.Itwillalsobeinteresting
toseetheroleofintelligenceinavarietyofcontextsoutsideoflogicalprocesses.
Intelligencemayholisticallyincreasepolarizationbyallowingpeopletobemoreconfident
intheiropinions.
Sernyak 46
ACKNOWLEDGMENTS
IamverygratefulformyadvisorDanKahan,andforhisguidanceandmentorship
throughouttheprocess.Professorwasincrediblyreceptive,helpful,andIfeelasthoughI
havelearnedsomuchfromhimthroughouttheentireprocess.Iamalsogratefulforthe
supportoftheCognitiveScienceDepartment,especiallyMarkSheskin.Iwouldliketothank
myclassmatesandprofessorsovertheyears,withoutwhomthiswouldnotbepossible.
Andofcourse,asalways,myfamilyandfriends.
Sernyak 47
REFERENCES
“AWiderIdeologicalGapBetweenMoreandLessEducatedAdults”PewResearchCenter.
26April2017.Online.
Adorno,T.W.,Frenkel-Brunswik,E.,Levinson,D.J.,&Sanford,R.N.(1950).The
authoritarianpersonality.
Bar-Hillel,M.(1980).Thebase-ratefallacyinprobabilityjudgments.ActaPsychologica,44(3),211-233.
Baron,J.,Scott,S.,Fincher,K.,&Metz,S.E.(2015).Whydoesthecognitivereflectiontest
(sometimes)predictutilitarianmoraljudgment(andotherthings)?.JournalofAppliedResearchinMemoryandCognition,4(3),265-284.
Brandt,M.J.,Reyna,C.,Chambers,J.R.,Crawford,J.T.,&Wetherell,G.(2014).The
ideological-conflicthypothesisintoleranceamongbothliberalsand
conservatives.CurrentDirectionsinPsychologicalScience,23(1),27-34.
Campitelli,G.,&Gerrans,P.(2014).Doesthecognitivereflectiontestmeasurecognitive
reflection?Amathematicalmodelingapproach.Memory&Cognition,42(3),434-447.
Campitelli,G.,&Labollita,M.(2010).Correlationsofcognitivereflectionwithjudgments
andchoices.JudgmentandDecisionMaking,5(3),182.
Chaiken,S.,&Trope,Y.(Eds.).(1999).Dual-processtheoriesinsocialpsychology.Guilford.
Chambers,J.R.,Schlenker,B.R.,&Collisson,B.(2013).Ideologyandprejudice:Theroleof
valueconflicts.Psychologicalscience,24(2),140-149.
Chen,S.,Duckworth,K.,&Chaiken,S.(1999).Motivatedheuristicandsystematic
processing.PsychologicalInquiry,10(1),44-49.
Cohen,G.L.(2003).Partyoverpolicy:Thedominatingimpactofgroupinfluenceon
politicalbeliefs.Journalofpersonalityandsocialpsychology,85(5),808.
Cokely,E.T.,Galesic,M.,Schulz,E.,Ghazal,S.,&Garcia-Retamero,R.(2012).Measuringrisk
literacy:TheBerlinnumeracytest.JudgmentandDecisionMaking,7(1),25.
“CongressandthePublic.”Gallup.5March2017.Online.
Conover,P.J.(1984).Theinfluenceofgroupidentificationsonpoliticalperceptionand
evaluation.TheJournalofPolitics,46(3),760-785.
Sernyak 48
Cosmides,L.,&Tooby,J.(1992).Cognitiveadaptationsforsocialexchange.Theadapted
mind,163-228.
Dawson,E.,Gilovich,T.,&Regan,D.T.(2002).MotivatedReasoningandPerformanceon
thewasonSelectionTask.PersonalityandSocialPsychologyBulletin,28(10),1379-1387.
DeNeys,W.,Rossi,S.,&Houdé,O.(2013).Bats,balls,andsubstitutionsensitivity:Cognitive
misersarenohappyfools.PsychonomicBulletin&Review,20(2),269-273.
Deppe,K.D.,Gonzalez,F.J.,Neiman,J.L.,Jacobs,C.,Pahlke,J.,Smith,K.B.,&Hibbing,J.R.
(2015).Reflectiveliberalsandintuitiveconservatives:AlookattheCognitive
ReflectionTestandideology.JudgmentandDecisionMaking,10(4),314.
Druckman,J.N.(2003).Thepoweroftelevisionimages:ThefirstKennedy-Nixondebate
revisited.TheJournalofPolitics,65(2),559-571.
Duarte,J.L.,Crawford,J.T.,Stern,C.,Haidt,J.,Jussim,L.,&Tetlock,P.E.(2015).Political
diversitywillimprovesocialpsychologicalscience.BehavioralandBrainSciences,38,e130.
Dunning,D.,Meyerowitz,J.A.,&Holzberg,A.D.(1989).Ambiguityandself-evaluation:The
roleofidiosyncratictraitdefinitionsinself-servingassessmentsofability.Journalofpersonalityandsocialpsychology,57(6),1082.
Evans,J.S.B.(2003).Intwominds:dual-processaccountsofreasoning.Trendsincognitivesciences,7(10),454-459.
Enten,Harry.“America’sDistasteforBothTrumpandClintonisRecordBreaking.”538.5
May2016.Online.
Eskine,K.J.,Kacinik,N.A.,&Prinz,J.J.(2011).Abadtasteinthemouthgustatorydisgust
influencesmoraljudgment.PsychologicalScience,22(3),295-299.
Fantino,E.,Kulik,J.,Stolarz-Fantino,S.,&Wright,W.(1997).Theconjunctionfallacy:Atest
ofaveraginghypotheses.PsychonomicBulletin&Review,4(1),96-101.
Fernbach,P.M.,Sloman,S.A.,Louis,R.S.,&Shube,J.N.(2013).Explanationfiendsandfoes:
Howmechanisticdetaildeterminesunderstandingandpreference.JournalofConsumerResearch,39(5),1115-1131.
Festinger,L.(1956).WhenProphecyFails:ASocialandPsychologicalStudyofaModernGroupthatPredictedtheDestinationoftheWorld.Harper.
Sernyak 49
Festinger,L.(1962).Atheoryofcognitivedissonance(Vol.2).Stanforduniversitypress.
Fiedler,K.(1988).Thedependenceoftheconjunctionfallacyonsubtlelinguistic
factors.Psychologicalresearch,50(2),123-129.
Fiske,S.T.,&Taylor,S.E.(1991).Socialcognition,2nd.NY:McGraw-Hill,16-15.
Frederick,S.(2005).Cognitivereflectionanddecisionmaking.TheJournalofEconomicPerspectives,19(4),25-42.
Gastil,J.,Braman,D.,Kahan,D.M.,&Slovic,P.(2005).The'WildavskyHeuristic':The
CulturalOrientationofMassPoliticalOpinion.
Gervais,W.M.(2014).Everythingispermitted?Peopleintuitivelyjudgeimmoralityas
representativeofatheists.PloSone,9(4),e92302.
Giddings,L.,&Dunn,T.J.(2016).TheRobustnessofAnti-AtheistPrejudiceasMeasuredby
WayofCognitiveErrors.TheInternationalJournalforthePsychologyofReligion,26(2),124-135.
Giner-Sorolla,R.,&Chapman,H.A.(2016).BeyondPurityMoralDisgustTowardBad
Character.PsychologicalScience,0956797616673193.
Green,D.,Palmquist,B.,&Schickler,E.(2002).Partisanheartsandminds.
Greenhoot,A.F.,Semb,G.,Colombo,J.,&Schreiber,T.(2004).Priorbeliefsand
methodologicalconceptsinscientificreasoning.AppliedCognitivePsychology,18(2),203-221.
Griggs,R.A.,&Cox,J.R.(1982).Theelusivethematic-materialseffectinWason'sselection
task.BritishJournalofPsychology,73(3),407-420.
Haidt,J.(2001).Theemotionaldoganditsrationaltail:asocialintuitionistapproachto
moraljudgment.Psychologicalreview,108(4),814.
Hart,P.S.,&Nisbet,E.C.(2012).Boomerangeffectsinsciencecommunication:How
motivatedreasoningandidentitycuesamplifyopinionpolarizationaboutclimate
mitigationpolicies.CommunicationResearch,39(6),701-723.
Hastorf,A.H.,&Cantril,H.(1954).Theysawagame;acasestudy.TheJournalofAbnormalandSocialPsychology,49(1),129.
Hertwig,R.,&Gigerenzer,G.(1999).The'conjunctionfallacy'revisited:Howintelligent
inferenceslooklikereasoningerrors.Journalofbehavioraldecisionmaking,12(4),275.
Sernyak 50
Hodson,G.,&Busseri,M.A.(2012).Brightmindsanddarkattitudeslowercognitiveability
predictsgreaterprejudicethroughright-wingideologyandlowintergroup
contact.PsychologicalScience,23(2),187-195.
Hoorens,V.(1993).Self-enhancementandsuperioritybiasesinsocial
comparison.Europeanreviewofsocialpsychology,4(1),113-139.
Huang,J.Y.,Sedlovskaya,A.,Ackerman,J.M.,&Bargh,J.A.(2011).Immunizingagainst
prejudice:Effectsofdiseaseprotectiononattitudestowardout-
groups.PsychologicalScience,22(12),1550-1556.
Imai,Kosuke,GaryKing,andOliviaLau.(2008).“TowardACommonFrameworkfor
StatisticalAnalysisandDevelopment.”JournalofComputationalandGraphical
Statistics,Vol.17,No.4(December),pp.892-913.
Inbar,Y.,Pizarro,D.,Iyer,R.,&Haidt,J.(2012).Disgustsensitivity,politicalconservatism,
andvoting.SocialPsychologicalandPersonalityScience,3(5),537-544.
Jost,J.T.,Glaser,J.,Kruglanski,A.W.,&Sulloway,F.J.(2003).Politicalconservatismas
motivatedsocialcognition.
Judd,CharlesM.(2000).EverydayDataAnalysisinSocialPsychologyComparisonsof
LinearModels.Handbookofresearchmethodsinsocialandpersonalitypsychology,370.
Kahan,D.(2010).Fixingthecommunicationsfailure.Nature,463(7279),296-297.
Kahan,D.M.(2012)."Ideology,motivatedreasoning,andcognitivereflection:An
experimentalstudy."
Kahan,D.M.(2015).Theexpressiverationalityofinaccurateperceptions.
Kahan,D.M.(2015).Thepoliticallymotivatedreasoningparadigm.BrowserDownloadThisPaper.
Kahan,D.M.,&Corbin,J.C.(2016).Anoteontheperverseeffectsofactivelyopen-minded
thinkingonclimate-changepolarization.Research&Politics,3(4),2053168016676705.
Kahan,D.M.,&Stanovich,K.E.(2016).RationalityandBeliefinHumanEvolution.
Kahan,D.M.,Braman,D.,Monahan,J.,Callahan,L.,&Peters,E.(2010).Culturalcognition
andpublicpolicy:Thecaseofoutpatientcommitmentlaws.LawandHumanBehavior,34(2),118-140.
Sernyak 51
Kahan,D.M.,Braman,D.,Slovic,P.,Gastil,J.,&Cohen,G.L.(2007).Thesecondnationalrisk
andculturestudy:Makingsenseof-andmakingprogressin-theamericanculture
waroffact.
Kahan,D.M.,Landrum,A.R.,Carpenter,K.,Helft,L.,&Jamieson,K.H.(2016).Science
curiosityandpoliticalinformationprocessing.
Kahan,D.M.,Peters,E.,Dawson,E.C.,&Slovic,P.(2013).Motivatednumeracyand
enlightenedself-government.
Kahan,D.M.,Peters,E.,Wittlin,M.,Slovic,P.,Ouellette,L.L.,Braman,D.,&Mandel,G.
(2012).Thepolarizingimpactofscienceliteracyandnumeracyonperceivedclimate
changerisks.Natureclimatechange,2(10),732-735.
Kahan,DanM.,andDonaldBraman."Culturalcognitionandpublicpolicy."YaleLaw&PolicyReview24.1(2006):149-172.
Kahneman,D.(2011).Thinking,fastandslow.Macmillan.
Kahneman,D.,&Frederick,S.(2002).Representativenessrevisited:Attributesubstitution
inintuitivejudgment.Heuristicsandbiases:Thepsychologyofintuitivejudgment,49.
King,G.,Tomz,M.,&Wittenberg,J.(2000).Makingthemostofstatisticalanalyses:
Improvinginterpretationandpresentation.Americanjournalofpoliticalscience,347-361.
Klaczynski,P.A.,&Gordon,D.H.(1996).Self-servinginfluencesonadolescents’evaluations
ofbelief-relevantevidence.JournalofExperimentalChildPsychology,62(3),317-339.
Klaczynski,P.A.andRobinson,B.2000.Personaltheories,intellectualability,and
epistemologicalbeliefs:Adultagedifferencesineverydayreasoning
tasks.PsychologyandAging,15:400–416.
Knowlton,B.J.,Ramus,S.J.,&Squire,L.R.(1992).Intactartificialgrammarlearningin
amnesia:Dissociationofclassificationlearningandexplicitmemoryforspecific
instances.PsychologicalScience,3(3),172-179.
KosukeImai,GaryKing,andOliviaLau.2007.“Zelig:Everyone’sStatisticalSoftware,”
http://GKing.harvard.edu/zelig.
Kunda,Z.(1990).Thecaseformotivatedreasoning.Psychologicalbulletin,108(3),480.
Lewandowsky,S.,&Oberauer,K.(2016).Motivatedrejectionofscience.CurrentDirectionsinPsychologicalScience,25(4),217-222.
Lewis,M.(2004).Moneyball:Theartofwinninganunfairgame.WWNorton&Company.
Sernyak 52
Liberali,J.M.,Reyna,V.F.,Furlan,S.,Stein,L.M.,&Pardo,S.T.(2012).Individual
differencesinnumeracyandcognitivereflection,withimplicationsforbiasesand
fallaciesinprobabilityjudgment.Journalofbehavioraldecisionmaking,25(4),361-381.
Lord,C.G.,Ross,L.,&Lepper,M.R.(1979).Biasedassimilationandattitudepolarization:
Theeffectsofpriortheoriesonsubsequentlyconsideredevidence.Journalofpersonalityandsocialpsychology,37(11),2098.
MacCoun,R.J.(2013).Moraloutrageandoppositiontoharmreduction.CriminalLawandPhilosophy,7(1),83-98.
Manktelow,K.I.(2012).Thinkingandreasoning:Anintroductiontothepsychologyofreason,judgmentanddecisionmaking(Vol.360).PsychologyPress.
Manktelow,K.I.,&Over,D.E.(1990).Deonticthoughtandtheselectiontask.Linesofthinking,1,153-164.
Munro,G.D.,Ditto,P.H.,Lockhart,L.K.,Fagerlin,A.,Gready,M.,&Peterson,E.(2002).
Biasedassimilationofsociopoliticalarguments:Evaluatingthe1996USpresidential
debate.BasicandAppliedSocialPsychology,24(1),15-26.
Morgan,G.S.,Mullen,E.,&Skitka,L.J.(2010).Whenvaluesandattributionscollide:
Liberals’andconservatives’valuesmotivateattributionsforalleged
misdeeds.PersonalityandSocialPsychologyBulletin,36(9),1241-1254.
Morsanyi,K.,Busdraghi,C.,&Primi,C.(2014).Mathematicalanxietyislinkedtoreduced
cognitivereflection:apotentialroadfromdiscomfortinthemathematicsclassroom
tosusceptibilitytobiases.BehavioralandBrainFunctions,10(1),31.
Nickerson,R.S.(1998).Confirmationbias:Aubiquitousphenomenoninmany
guises.Reviewofgeneralpsychology,2(2),175.
Nisbet,ErikC.,KathrynE.Cooper,andR.KellyGarrett."Thepartisanbrain:Howdissonant
sciencemessagesleadconservativesandliberalsto(dis)trustscience."TheANNALSoftheAmericanAcademyofPoliticalandSocialScience658.1(2015):36-66.
Nyhan,B.,&Reifler,J.(2010).Whencorrectionsfail:Thepersistenceofpolitical
misperceptions.PoliticalBehavior,32(2),303-330.
Oechssler,J.,Roider,A.,&Schmitz,P.W.(2009).Cognitiveabilitiesandbehavioral
biases.JournalofEconomicBehavior&Organization,72(1),147-152.
Paxton,J.M.,Ungar,L.,&Greene,J.D.(2012).Reflectionandreasoninginmoral
judgment.CognitiveScience,36(1),163-177.
Sernyak 53
Pennycook,G.,Cheyne,J.A.,Barr,N.,Koehler,D.J.,&Fugelsang,J.A.(2014).Theroleof
analyticthinkinginmoraljudgementsandvalues.Thinking&Reasoning,20(2),188-214.
Pennycook,G.,Cheyne,J.A.,Seli,P.,Koehler,D.J.,&Fugelsang,J.A.(2012).Analytic
cognitivestylepredictsreligiousandparanormalbelief.Cognition,123(3),335-346.
Perkins,D.N.(1985).Postprimaryeducationhaslittleimpactoninformal
reasoning.JournalofEducationalPsychology,77(5),562.
Perkins,D.N.,Farady,M.,&Bushey,B.(1991).Everydayreasoningandtherootsofintelligence.LawrenceErlbaumAssociates,Inc.
“PoliticsofClimate”PewResearchCenter.4October2016.Online.
Powell,J.L.(2015).ClimateScientistsVirtuallyUnanimous:AnthropogenicGlobal
WarmingIsTrue.BulletinofScience,Technology&Society,35(5-6),121-124.
Primi,C.,Morsanyi,K.,Chiesi,F.,Donati,M.A.,&Hamilton,J.(2015).Thedevelopmentand
testingofanewversionofthecognitivereflectiontestapplyingitemresponse
theory(IRT).JournalofBehavioralDecisionMaking.
Proulx,T.,Inzlicht,M.,&Harmon-Jones,E.(2012).Understandingallinconsistency
compensationasapalliativeresponsetoviolatedexpectations.Trendsincognitivesciences,16(5),285-291.
Reilly,D.(2012).Gender,culture,andsex-typedcognitiveabilities.PloSone,7(7),e39904.
Rutjens,B.T.,&Heine,S.J.(2016).Theimmorallandscape?Scientistsareassociatedwith
violationsofmorality.PloSone,11(4),e0152798.
Schwartz,John."ScienceTeacher’sGraspofClimateChangeisFoundLacking."NewYork
Times[NewYork]11February2016.Online.
Scoville,W.B.,&Milner,B.(1957).Lossofrecentmemoryafterbilateralhippocampal
lesions.JournalofNeurology,Neurosurgery&Psychiatry,20(1),11-21.
Sharot,T.(2011).Theoptimismbias.Currentbiology,21(23),R941-R945.
Sibley,C.G.,&Duckitt,J.(2008).Personalityandprejudice:Ameta-analysisandtheoretical
review.PersonalityandSocialPsychologyReview,12(3),248-279.
Silver,Nate.“Education,notIncome,PredictedWhoWouldVoteforTrump.”538.22
November2016.Online.
Sernyak 54
Skitka,L.J.,Bauman,C.W.,&Sargis,E.G.(2005).Moralconviction:Anothercontributorto
attitudestrengthorsomethingmore?.Journalofpersonalityandsocialpsychology,88(6),895.
Skitka,L.J.,Mullen,E.,Griffin,T.,Hutchinson,S.,&Chamberlin,B.(2002).Dispositions,
scripts,ormotivatedcorrection?:Understandingideologicaldifferencesin
explanationsforsocialproblems.Journalofpersonalityandsocialpsychology,83(2),470.
Sloman,S.A.(2002).Twosystemsofreasoning.
Smith,E.R.,&DeCoster,J.(2000).Dual-processmodelsinsocialandcognitivepsychology:
Conceptualintegrationandlinkstounderlyingmemorysystems.Personalityandsocialpsychologyreview,4(2),108-131.
Spranca,M.,Minsk,E.,&Baron,J.(1991).Omissionandcommissioninjudgmentand
choice.Journalofexperimentalsocialpsychology,27(1),76-105.
Stanovich,K.E.(1999).Whoisrational?:Studiesofindividualdifferencesinreasoning.PsychologyPress.
Stanovich,K.E.(2004).Metarepresentationandthegreatcognitivedivide:Acommentary
onHenriques'“PsychologyDefined”.Journalofclinicalpsychology,60(12),1263-1266.
Stanovich,K.E.(2013).Whyhumansare(sometimes)lessrationalthanotheranimals:
Cognitivecomplexityandtheaxiomsofrationalchoice.Thinking&Reasoning,19(1),1-26.
Stanovich,K.E.,&Toplak,M.E.(2012).Definingfeaturesversusincidentalcorrelatesof
Type1andType2processing.Mind&Society,11(1),3-13.
Stanovich,K.E.,&West,R.F.(2007).Naturalmysidebiasisindependentofcognitive
ability.Thinking&Reasoning,13(3),225-247.
Tajfel,H.,&Turner,J.C.(1979).Anintegrativetheoryofintergroupconflict.Thesocialpsychologyofintergrouprelations,33(47),74.
Tajfel,H.,Billig,M.G.,Bundy,R.P.,&Flament,C.(1971).Socialcategorizationand
intergroupbehaviour.Europeanjournalofsocialpsychology,1(2),149-178.
Tetlock,P.E.(1983).Cognitivestyleandpoliticalideology.JournalofPersonalityandSocialPsychology,45(1),118.
Sernyak 55
Terrizzi,J.A.,Shook,N.J.,&Ventis,W.L.(2010).Disgust:Apredictorofsocialconservatism
andprejudicialattitudestowardhomosexuals.PersonalityandIndividualDifferences,49(6),587-592.
TheEditorialBoard.“TeachingtheTruthAboutClimateChange.”NewYorkTimes[New
York]10October2015.Online.
Todorov,A.,Mandisodza,A.N.,Goren,A.,&Hall,C.C.(2005).Inferencesofcompetence
fromfacespredictelectionoutcomes.Science,308(5728),1623-1626.
“TopVotingIssuesinthe2016Election.”PewResearchCenter.7July2016.Online.
Toplak,M.E.,&Stanovich,K.E.(2003).Associationsbetweenmysidebiasonaninformal
reasoningtaskandamountofpost-secondaryeducation.AppliedCognitivePsychology,17(7),851-860.
Toplak,M.E.,West,R.F.,&Stanovich,K.E.(2011).TheCognitiveReflectionTestasa
predictorofperformanceonheuristics-and-biasestasks.Memory&Cognition,39(7),1275.
Toplak,M.E.,West,R.F.,&Stanovich,K.E.(2014).Assessingmiserlyinformation
processing:AnexpansionoftheCognitiveReflectionTest.Thinking&Reasoning,20(2),147-168.
Thüring,M.,&Jungermann,H.(1990).Theconjunctionfallacy:Causalityvs.event
probability.JournalofBehavioralDecisionMaking,3(1),61-74.
Tversky,A.,&Kahneman,D.(1983).Extensionalversusintuitivereasoning:The
conjunctionfallacyinprobabilityjudgment.Psychologicalreview,90(4),293.
Washington,G.(1796).Farewelladdress.May,15,1796.
Wason,P.C.(1966).Reasoning.
Wasserman,E.A.,Dorner,W.W.,&Kao,S.F.(1990).Contributionsofspecificcell
informationtojudgmentsofintereventcontingency.JournalofExperimentalPsychology:Learning,Memory,andCognition,16(3),509.
Weller,J.A.,Dieckmann,N.F.,Tusler,M.,Mertz,C.K.,Burns,W.J.,&Peters,E.(2013).
Developmentandtestingofanabbreviatednumeracyscale:ARaschanalysis
approach.JournalofBehavioralDecisionMaking,26(2),198-212.
West,R.F.,Meserve,R.J.,&Stanovich,K.E.(2012).Cognitivesophisticationdoesnot
attenuatethebiasblindspot.Journalofpersonalityandsocialpsychology,103(3),506.
Sernyak 56
Westen,D.,Blagov,P.S.,Harenski,K.,Kilts,C.,&Hamann,S.(2006).Neuralbasesof
motivatedreasoning:AnfMRIstudyofemotionalconstraintsonpartisanpolitical
judgmentinthe2004USpresidentialelection.Journalofcognitiveneuroscience,18(11),1947-1958.
Westfall,J.,VanBoven,L.,Chambers,J.R.,&Judd,C.M.(2015).Perceivingpolitical
polarizationintheUnitedStates:Partyidentitystrengthandattitudeextremity
exacerbatetheperceivedpartisandivide.PerspectivesonPsychologicalScience,10(2),145-158.
Wildavsky,A.(1987).Choosingpreferencesbyconstructinginstitutions:Aculturaltheory
ofpreferenceformation.AmericanPoliticalScienceReview,81(01),3-21.
Allcitationsfromgooglescholar
Sernyak 57
APPENDIX
APPENDIXA
SYSTEM1VSSYSTEM2
Peoplethinkindifferentwaysdependingonthesituation.Recentliteraturehas
soughttocharacterizethinkingintoafundamentaldichotomybetweenfastautomatic
processesandslowermoredeliberatemodes.Theyfirsttypeofthinkingisknownas
System1thought,andthesecondisSystem2.
Thefirstsystemreliesonheuristics,whicharequickwaystoformjudgments,easily
retrievableknowledgestructuresthatcanbecalledtomindquickly.Thissystemis
relativelyeffortless(Sloman,2002,Kahneman&Frederick2002).Thissystemisslowly
andgenerallytrainedovertime,bycreatingunconsciousgeneralrepresentationsthrough
association(Smith&DeCoster,2000,Chaikan&Trope,1999).Thesegeneralknowledge
structurescreateanautonomoussetofsystemsthatareusedformostofdaytoday
thinking.Theautonomousandeffortlessnatureisthedefiningfeatureofthesesystems
(Stanovich,2004,Kahneman,2011).
Thesecondsystemisslowerandmoreeffortful,andisusedlessoftenforthis
reason.Itisoftencalledinwhensomethingviolatesexpectation,ormorereflectionis
necessaryoncertaintask.Whilethishelpskeepcontrolinunknownsituations,ittakes
effort(Kahneman,2011).System2processesusespecificrules,andlogic,thatcanthenbe
appliedtosituations.Italsoiswhatallowsforcomplexprocessessuchcognitive
decoupling,ortheabilitytosustainsecondaryrepresentations,andthereforesimulation
(Stanovich,2004).Italsoheavilyrelatedtoconsciousnessandtheunderstandingofself
(Kahneman,2011).System2isassociatedwithwhatisgenerallyconsideredhuman
cognition,asopposedtoSystem1,whichsharesmorecognitivefeatureswithother
animals(Evans,2003).
Type1Processes Type2ProcessesHolistic Analytic
Automatic Controlled
Relativelyundemandingofcognitive
capacity
Capacitydemanding
Relativelyfast Relativelyslow
Acquisitionbybiology,exposure,and
personalexperience
Acquisitionbycultureandformaltuition
Parallel Sequential
Evolutionarilyold Evolutionarilyrecent
Implicit Explicit
Oftenunconsciousorpreconscious Oftenconscious
Lowercorrelationswithintelligence Highercorrelationswithintelligence
Short-leashedgeneticgoals Long-leashedgoalsthattendtoward
personalutilitymaximization
Figure24:AchartinitiallyfromStanovich1999outliningvariousSystem1andSystem2processes.
Sernyak 58
Thetwosystemsworktogetherinawaythatmaximizescognitiveefficiency.Dueto
thehigheffortassociatedwiththinking,peoplearecognitivemisers(Taylor,1991),and
willavoidusingeffortfulprocessesifpossible.System1isoftennotoptimalthough,as
heuristicsaregeneralrulesthatcanbemisappliedandleadtobias.Duetoenergy
constraintspeoplewilloftenusethelesseffortful,butlessexactsystem.Thismakessense,
astomakeday-to-daydecisionspeopledon’twanttohavetoalwayscalculate
probabilities,ordocomplicatedcostbenefitanalysis,andwilloftenuseheuristicstomake
naturalassessments(Tversky&Kahneman,1983).Theseheuristicsmayoversimplifythe
worldthoughandleadtobias.
APPENDIXB
COGNITIVEREFLECTIONTEST
TheCognitiveReflectionTest(CRT),developedbyShaneFrederickin2005,isa
studyinstrumentusedtodeterminehowreliantpeopleareonSystem1processing.Itisa
seriesofthreequestions,eachwithananswerthateasilyspringstomind,butisactually
wrong.TheideaisthatSystem2processesshouldoverridetheintuitiveresponsein
individualswhorelylessheavilyonintuitionandheuristics.Eachquestionisquiteeasy
onceexplained,sothereasonpeoplegetthequestionwrongisduetoalackofreflection,
notaninabilitytosolveit.Thetestisasfollows:
1. Abatandaballcost$1.10intotal.Thebatcosts$1.00morethantheball.Howmuchdoesthe
ballcost?[correctanswer=5cents;heuristicanswer=10cents]
2. Ifittakes5minutesforfivemachinestomakefivewidgets,howlongwouldittakefor100
machinestomake100widgets?[correctanswer=5minutes;heuristicanswer=100
minutes]
3. Inalake,thereisapatchoflilypads.Everyday,thepatchdoublesinsize.Ifittakes48days
forthepatchtocovertheentirelake,howlongwouldittakeforthepatchtocoverhalfofthe
lake?[correctanswer=47days;heuristicanswer=24days]
Thetestisunderstandablycorrelatedwithintelligence,numeracy,self-reportedSAT
scores,andwonderlicpersonalitytests.Thismakessense,asbothwouldrequirecritical
reasoning,aswellasreadingcomprehensionskills,evenifthemathisnotdifficult.Yetthe
CRTismorethanasimpleintelligencetest(Gerrans,2014),asitmoreatestofrelianceon
typeofcognitiveprocesses,whereasintelligencetestsaretestsofabilitytousethesecond
typeofprocesses(Toprak,West&Stanovich,2014,Pennycook,etal.,2012,Paxton,Ungar
&Green,2012,Paxton,Ungar&Green,2012,Pennycook,Cheyne,Barr,Koehler&
Fugelsang,2014,Pennycook,Cheyne,Barr,Koehler&Fugelsang,2014,Campitelli&
Labollita2010).
WhiletheCRTmaybeastrongpredictorofcognitivestyle,itisnotwithoutitsflaws.
Ithasbeenwidelyusedandoncepeopleknowofthecorrectanswersitisobviouslyno
longeruseful.ThisisespeciallytrueonsurveysiteslikeMechanicalTurkwhere
participantswilloftenhavealreadytakenthetest.Thereisalsoaflooreffectinnon-well-
educatedpopulations,asthetestisveryhardandmorethan50%ofpeoplegetnoneofthe
questionsright(Frederick,2005).Thismakesitimpossibletodifferentiatebetweenhalfof
thepopulation.Inourstudy,59%ofparticipantsanswered0outofthe3questionsright,
sothereisdefinitelyaneedforlowertraitdiscriminationproblems.
Sernyak 59
TherehavebeenafewprominenteffortstoaddquestionstotheCRT.Onesuch
effortwasundertakenbyToplak,West,andStanovichin2004.Theyadded4new
questionswiththehopeofaddingquestionstotheoriginalmeasure,whilealsoproviding
analternative.Theyfoundthenew7-questionmeasurewasabetterpredictorofrational
thinkingtasksthantheoriginalCRT.Italsodiscriminatedamonglowertraitindividuals
moresothantheoriginalduetotheincreasednumberofquestionsandwaseasierthanthe
CRT.ThemeanpercentofansweringaquestioncorrectlyontheoriginalCRTwas17,while
onthefournewquestionsthemeanwas24.Onedrawbackisthatonequestionismultiple
choicesoacorrectanswercouldbereachedbychance.Theproposedfourquestionswere
asfollows:
1. IfJohncandrinkonebarrelofwaterin6days,andMarycandrinkonebarrelofwaterin12
days,howlongwouldittakethemtodrinkonebarrelofwatertogether?________days[correct
answer=4days;heuristicanswer=9]
2. Jerryreceivedboththe15thhighestandthe15thlowestmarkintheclass.Howmanystudents
arethereintheclass?_____students[correctanswer=29students;heuristicanswer=30
students]
3. Amanbuysapigfor$60,sellsitfor$70,buysitbackfor80$,andsellsitfinallyfor$90.How
muchhashemade?_____dollars[correctanswer=$20;heuristicansweranswer=$10]
4. Simondecidedtoinvest$8,000inthestockmarketonedayearlyin2008.Sixmonthsafterhe
invested,onJuly17,thestockshehadpurchasedweredown%50.FortunatelyforSimon,
fromJuly17toOctober17,thestockshehadpurchasedwentup75%.AtthispointSimon
has:a.brokeneveninthestockmarket,b.isaheadofwherehebegan,c.haslostmoney
[correctanswer=c,becausethevalueatthispointis$7,000;heuristicanswer=b]
AnotherattempttoincreasethelengthoftheCRTwasundertakenbyPrimi,
Morsanyi,Chiesi,Donati,andHamiltonin2015.Theytestedamultitudeofquestionsand
reportedanadditionalthreetheybelievedwouldbeagoodadditiontothetest.Theirnew
testshowedagreaterabilitytodifferentiateamonglowertraitindividuals.Theyalso
suggestedusingItemResponseTheory(IRT)analysisinfuturestudiesinordertogain
moredifferentiationusingthesamenumberofquestions.Basically,itisawaytoweight
questionsandprovidegreaterlinkagebetweenitemsandlatencycharacteristics.
Ultimatelytheyfoundtheirnewquestions,alongwiththeoriginalquestions,couldcreatea
scorethroughIRTanalysisthatwasabetterpredictorofriskseekingbehaviorthan
intelligence,demonstratingtheirtestwasagainatestthatmeasuredmethodofthinking.
1. Ifthreeelvescanwrapthreetoysinanhour,howmanyelvesareneededtowrapsixtoysin2
hours?[correctanswer=3elves;heuristicanswer=6elves]
2. Jerryreceivedboththe15thhighestandthe15thlowestmarkintheclass.Howmanystudents
arethereintheclass?[correctanswer=29students;heuristicanswer=30students]
3. Inanathleticsteam,tallmembersarethreetimesmorelikelytowinamedalthanshort
members.Thisyeartheteamhaswon60medalssofar.Howmanyofthesehavebeenwonby
shortathletes?[correctanswer=15medals;heuristicanswer=20medals]
ThefinalrelevantattempttoexpandtheCRTwasundertakenbyBaronetal.in
2015.Thegroupattemptedtoaddwordproblemsinanefforttoexpandthescopeofthe
CRTaswellascorrectforthegenderednatureoftheoriginaltest,andonthenewtest
womenscoredaswellasmen.Theyalsofoundtheirtestwasavalidpredictorofmoral
judgment.Thetesthad11wordproblems,alongwiththeoriginalCRT,and3new
Sernyak 60
arithmeticproblems.Someexamplesofthewordproblemsareasfollows(toseethefull
testseeendofappendixA):
Allflowershavepetals.Roseshavepetals.Ifthesetwostatementsaretrue,canweconclude
fromthemthatrosesareflowers.YesNoAllmammalswalk.Whalesaremammals.Ifthesetwostatementsaretrue,canweconclude
fromthemthatwhaleswalk.YesNoInabox,someredthingsaresquare,andsomesquarethingsarelarge.Whatcanwe
conclude?[a.Somethingsarelargeb.Allthingsarelargec.Wecan’tconcludeanythingaboutredthingsandlargethings]
ThispapertakesdatafromastudythatseekstoexpandtheCRT.Participantswere
askedtheoriginalCRTquestionsaswellasthebestperformingonesfrompreviousCRT
expansionpapers.ProfessorKahanaddedsomequestionsofhisowninanefforttoanalyze
thepredictivepowerofthevariousCRTtests.Whathasshownthemostpromiseisan
inclusivemeasureofallofthetests,andithasbeenshowntobebothinternallyconsistent
anddiscriminativeacrossthelatencyvariable.Theproblemwiththetestisthatunlikethe
originalCRTitisverylong.
Baronetal.,2015-PROPOSALFORNEWCRT
Beliefbiasitemswithlures
1. Allflowershavepetals.
Roseshavepetals.
Ifthesetwostatementsaretrue,canweconcludefromthemthatrosesareflowers(no).
2. Allmammalswalk.
Whalesaremammals.
Ifthesetwostatementsaretrue,canweconcludefromthemthatwhaleswalk(yes).
3. Allthingsthathaveamotorneedoil.
Automobilesneedoil.
Ifthesetwostatementsaretrue,canweconcludefromthemthatautomobileshaveamotor(no).
4. Alllivingthingsneedwater.
Rosesneedwater.
…,canweconcludefromthemthatrosesarelivingthings(no).
5. Allvehicleshavewheels.
Boatsarevehicles.
…,canweconcludefromthemthatboatshavewheels(yes).
Syllogisms
1. Inabox,someredthingsaresquare,andsomesquarethingsarelarge.Whatcanweconclude?[a.
Someredthingsarelarge.b.Allredthingsarelarge.c.Wecan’tconcludeanythingaboutredthingsandlargethings.
2. Inabox,nogreenthingsareround,andallroundthingsarelarge.Whatcanweconclude?[a.No
greenthingsarelarge.b.Somegreenthingsarenotlarge.c.Wecan’tconcludeanythingaboutgreenthingsandlargethings.
3. Inabox,nobluethingsaretriangular,andnotriangularthingsarelarge.Whatcanweconclude?
[a.Nobluethingsarelarge.b.Somebluethingsarenotlarge.c.Wecan’tconcludeanythingaboutbluethingsandlargethings.]
OriginalCRTQuestions
…
NewArithmeticQuestions
Sernyak 61
1. Ifittakes2nurses2minutestomeasurethebloodpressureof2patients,howlongwouldittake200
nursestomeasurethebloodpressureof200patients?[correctanswer=2minutes;heuristicanswer
=200minutes]
2. Soupandsaladcost$5.50intotal.Thesoupcostsadollarmorethanthesalad.Howmuchdoesthe
saladcost?[correctanswer=$3.25and$2.25]
3. Sallyismakingsuntea.Everyhour,theconcentrationoftheteadoubles.Ifittakes6hoursforthetea
tobeready,howlongwouldittakefortheteatoreachhalfofthefinalconcentration?[correct
answer=5hours;heuristicanswer=3hours]
Otheritems
1. JackislookingatAnnebutAnneislookingatGeorge.JackismarriedbutGeorgeisnot.Isamarried
personlookingatanunmarriedperson?[a.Yesc.Noc.Cannotbedetermined]2. Ann’sfatherhasatotaloffivedaughters:Lala,Lele,Lili,Lolo,and______.Whatisthenameofthefifth
daughter?[Ann]
3. Onthesideofaboathangsaladderwithsixrungs.Eachrungisonefootfromthenextone,andthe
bottomrungisrestingonthesurfaceofthewater.Thetiderisesatarateofonefootanhour.How
longwilltakethewatertoreachthetoprung?[a.5hoursb.6hoursc.never]
APPENDIXC
MODELSUMMARIES
ModelWasonSelectiontaskcorrectanswer
Sernyak 62
ModelWasonSelectiontaskuseoflogic
ModelofconjunctionfallacyusingCRT