Академический Документы
Профессиональный Документы
Культура Документы
Thepsychologyofchoice
AssumptionsofNeoclassical
Economics(HomoEconomicus)
Selfishnessanindividualchoosesonthe
basisofhis/herowninterests(notrue,
systematicaltruism)
Stable,exogenouspreferenceswhatthe
individualwantsiswelldefined,availableto
introspection,andstableovertime
Formalrationalityanindividuals
preferences,tastes,etc.areconsistentwith
eachother
RationalChoiceTheoriesfor
Individuals
Utilitytheoryoneagent,choicedependsonly
onstatesofnature
Example:Adecisionthatdependson
statesofnature
Options:
Planpicnicoutdoors
Planpicnicindoors
Possiblestatesofnature
Rain
Norain
Choicedependsonlikelihoodofrain,relative
qualityofpicnicindoors/outdoorswithand
withoutrain
RationalChoiceTheoriesfor
Individuals(VonNeumannand
Morgenstern,1944)
Utilitytheoryoneagent,choicedependsonly
onstatesofnature
Gametheorymorethanoneagent,choice
dependsonwhatotheragentsmaychoose
Example:adecisionthatdependson
whatothersmaydo
Options:
Gotothebeach
Gotothecinema
Yourfriendmaychooseto:
Gotothebeach
Gotothecinema
Youcannotcontrolorknowwhatyourfriendwilldo
Bothofyouknoweachotherspreferences
Choicedependsonwhatyouthinkyourfriendwilldo,which
dependsonwhats/hethinksyouwilldo,andsoon
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby
(revealed)preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby
(revealed)preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
Preferencesarewellordered
i.e.transitive:IfA?BandB?C,thenA?C
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby(revealed)
preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
Preferencesarewellordered
i.e.transitive:IfA?BandB?C,thenA?C
Choicesunderuncertaintyaredeterminedby
expectedutility
Expectedutilityisaprobabilityweightedcombinationofthe
utilitiesofallnpossibleoutcomesO i
AConcaveUtilityCurve
Example:
ApplicationofUtilityTheory
Options:
Gamble(50%chancetowin$100;else$0)
SureThing(100%chancetowin$50)
Expectedvaluesarethesame:
EV(Gamble)=(.5)($100)+(.5)($0)=$50
EV(SureThing)=(1)($50)=$50
Buttheirexpectedutilitiesmaystilldiffer
EU(Gamble)=.5U($100)+.5U($0)
EU(SureThing)=U($50)
Expectedutilitytheorysaysthat
utilitiesare
Notdirectlyobservable(internaltoan
individual)
Notcomparableacrossindividuals
Constrainedbyrevealedpreferences(i.e.
choicesbetweengambles)
Note:Andersongetstherelationshipbetween
utilityandvaluewrong
Dopeopleschoicesobeythetheoryof
expectedutility(i.e.,formalrationality)?
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby
(revealed)preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
UtilityversusPreference(Lichtenstein
andSlovic,1971;1973)
Psgiventwooptions:
Pbet:29/36probabilitytowin$2
$bet:7/36probabilitytowin$9
Twoconditions:
Chooseone:MostpreferPbet
Valuethebets:Mostvalue$bethigher
Showsutility(basedoncashvalue)isnot
consistentwithrevealedpreference
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby
(revealed)preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
Contradictedbypreferencereversal
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby
(revealed)preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
Contradictedbypreferencereversal
Preferencesarewellordered
i.e.transitive:IfA?BandB?C,thenA?C
TestsofTransitivity(A.Tversky,1969)
Psshownratingsofcollegeapplicantsonthree
dimensions:
Applicant
Intelligence
Stability
Social
A
B
C
D
E
69
72
75
78
81
84
78
72
66
60
75
65
55
45
35
PschoseAoverB,BoverC,CoverD,DoverE,but E
overA(differenceinintelligenceoutweighed)
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby
(revealed)preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
Contradictedbypreferencereversal
Preferencesarewellordered
i.e.transitive:IfA?BandB?C,thenA?C
Contradictedbythreeoptionintransitivities(and
preferencereversals)
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby(revealed)preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
Contradictedbypreferencereversals
Preferencesarewellordered
i.e.transitive:IfA?BandB?C,thenA?C
Contradictedbythreeoptionintransitivities(andpreferencereversals)
Choicesunderuncertaintyaredeterminedbyexpectedutility
Expectedutilityisaprobabilityweightedcombinationoftheutilitiesof
allnpossibleoutcomesO i
TestingExpectedUtility(Tverskyand
Kahneman,1981)
Choosebetween
A.Surewinof$30
B.80%chancetowin$45
TestingExpectedUtility(Tverskyand
Kahneman,1981)
Choosebetween:
A.Surewinof$30
B.80%chancetowin$45
Choosebetween:
C.25%chancetowin$30
D.20%chancetowin$45
TestingExpectedUtility(Tverskyand
Kahneman,1981)
Choosebetween:
A.Surewinof$30[78percent]
B.80%chancetowin$45[22percent]
Choosebetween:
C.25%chancetowin$30[42percent]
D.20%chancetowin$45[58percent]
TestingExpectedUtility(Tverskyand
Kahneman,1981)
Choosebetween:
A.Surewinof$30[78percent]
B.80%chancetowin$45[22percent]
Choosebetween:
C.25%chancetowin$30[42percent]
D.20%chancetowin$45[58percent]
ButthispatternisinconsistentwithEUT:
EU(A)>EU(B)=>u($30)>.8u($45)
EU(D)>EU(C)=>.25u($30)<.2u($45)
Multiplybothsidesofbottominequalityby4:contradictstop
inequality
TestingExpectedUtility(Tverskyand
Kahneman,1981)
Choosebetween:
A.Surewinof$30[78percent]
B.80%chancetowin$45[22percent]
Choosebetween:
C.25%chancetowin$30[42percent]
D.20%chancetowin$45[58percent]
ButthispatternisinconsistentwithEUT:
EU(A)>EU(B)=>u($30)>.8u($45)
EU(D)>EU(C)=>.25u($30)<.2u($45)
Multiplybothsidesofbottominequalityby4:contradictstopinequality
Thisiscalledacertaintyeffect:certaingainshaveextra
psychologicalvalue
ExpectedUtilityTheory
CrucialFeatures
Utility(degreeofliking)isdefinedby(revealed)preferences
i.e.U(A)>U(B)iffAispreferredto(chosenover)B
Contradictedbypreferencereversals
Preferencesarewellordered
i.e.transitive:IfA?BandB?C,thenA?C
Contradictedbythreeoptionintransitivities(andpreferencereversals)
Choicesunderuncertaintyaredeterminedbyexpectedutility
Expectedutilityisaprobabilityweightedcombinationoftheutilitiesof
allnpossibleoutcomesO i
Contradictedbycertaintyeffect
So,peopleschoicesdonotobey
formalrationality.
Aretheirpreferencesnonetheless
stable?
NeoclassicalAssumptionsAbout
Preferences
Thechosenoptioninadecisionproblem
shouldremainthesameevenifthesurface
descriptionoftheproblemchanges
(descriptiveinvariance)
ATestofDescriptiveInvariance
(TverskyandKahneman,1981)
Consideratwostagegame.Inthefirststage,there
isa75%chancetoendthegamewithoutwinning
anything,anda25%chancetomoveintothesecond
stage.Ifyoureachthesecondstage,youhavea
choicebetween
Surewinof$30
80%chancetowin$45
Yourchoicemustbemadebeforethegamestarts,
i.e.beforetheoutcomeofthefirststageisknown
ATestofDescriptiveInvariance
(TverskyandKahneman,1981)
Consideratwostagegame.Inthefirststage,there
isa75%chancetoendthegamewithoutwinning
anything,anda25%chancetomoveintothesecond
stage.Ifyoureachthesecondstage,youhavea
choicebetween
Surewinof$30[74percent]
80%chancetowin$45[26percent]
Yourchoicemustbemadebeforethegamestarts,
i.e.beforetheoutcomeofthefirststageisknown
ATestofDescriptiveInvariance
(continued)
Butthisgambleisformallyidenticaltoa
problemwesawearlier,namely:
Choosebetween:
C.25%chancetowin$30[42percent]
D.20%chancetowin$45[58percent]
ATestofDescriptiveInvariance
(continued)
Butthisgambleisformallyidenticaltoaproblemwe
sawearlier,namely:
Choosebetween:
C.25%chancetowin$30[42percent]
D.20%chancetowin$45[58percent]
Compare:
Consideratwostagegame.Inthefirststage,thereisa
75%chancetoendthegamewithoutwinninganything,and
a25%chancetomoveintothesecondstage.Ifyoureach
thesecondstage,youhaveachoicebetween
Surewinof$30[74percent]
80%chancetowin$45[26percent]
ATestofDescriptiveInvariance
(continued)
Butthisgambleisformallyidenticaltoaproblemwesawearlier,namely:
Choosebetween:
C.25% chancetowin$30[42percent]
D.20% chancetowin$45[58percent]
Compare:
Consideratwostagegame.Inthefirststage,thereisa75%chancetoendthe
gamewithoutwinninganything,anda25%chancetomoveintothesecond
stage.Ifyoureachthesecondstage,youhaveachoicebetween
Surewinof$30[74percent]
80% chancetowin$45[26percent]
Aviolationofdescriptiveinvariance
Thisisknownasapseudocertaintyeffect:Whenastageoftheproblem
ispresentedasinvolvingacertaingain,itcarriesextraweightevenif
gettingtothatstageisitselfuncertain.
FramingEffects(Tverskyand
Kahneman,1981)
Problem1:ImaginethattheU.S.ispreparingfortheoutbreakofan
unusualAsiandisease,whichisexpectedtokill600people.Two
alternativeprogramstocombatthediseasehavebeenproposed.Assume
thattheexactscientificestimateoftheconsequencesoftheprogramsare
asfollows:
IfProgramAisadopted,200peoplewillbesaved[72percent]
IfProgramBisadopted,thereis1/3probabilitythat600peoplewillbe
saved,and2/3probabilitythatnopeoplewillbesaved.[28percent]
Problem2:
IfProgramCisadopted400peoplewilldie[22percent]
IfProgramDisadoptedthereis1/3probabilitythatnobodywilldie,
and2/3probabilitythat600peoplewilldie.[78percent]
Buttheprogramsareidentical!Thisexamplealsoviolatesdescriptive
invariance.
Showsreflectioneffect:Riskaversioninthedomainofgains;riskseeking
inthedomainoflosses
NeoclassicalAssumptionsAbout
Preferences
Thechosenoptioninadecisionproblem
shouldremainthesameevenifthesurface
descriptionoftheproblemchanges
(descriptiveinvariance)
Contradictedbypseudocertaintyandframing
effects
NeoclassicalAssumptionsAbout
Preferences
Thechosenoptioninadecisionproblemshould
remainthesameevenifthesurfacedescriptionofthe
problemchanges(descriptiveinvariance)
Contradictedbypseudocertaintyandframingeffects
Thechosenoptionshoulddependonlyonthe
outcomesthatwillobtainafterthedecisionismade,
notondifferencesbetweenthoseoutcomesand
thestatusquo
whatoneexpects
theoverallmagnitudeofthedecision
StatusQuoBias(Kahnemen,Knetsch,
andThaler,1990)
Sellerseachgivencoffeemug,askedhow
muchtheywouldselliffor
Buyersnotgivenmug,askedhowmuchthey
wouldpayforone
Medianvalues:
Sellers:$7.12
Buyers:$2.87
StatusQuoBias(Kahnemen,Knetsch,
andThaler,1990)
Sellerseachgivencoffeemug,askedhowmuchtheywould
selliffor
Buyersnotgivenmug,askedhowmuchtheywouldpayfor
one
Medianvalues:
Sellers:$7.12
Buyers:$2.87
Choosersaskedtochoosebetweenmugandcashpreferred
mugifcashamountwas$3.12orlower,onaverage
Showsendowmenteffectwevaluewhatwehave;andloss
aversionwedontwanttoloseit
Mentalaccountsandexpectations
(TverskyandKahneman,1981)
Imaginethatyouhavedecidedtoseeaplay
whereadmissionis$20perticket.Asyou
enterthetheateryoudiscoverthatyouhave
losttheticket.Theseatwasnotmarkedand
theticketcannotberecovered.Wouldyoupay
$20foranotherticket?
Mentalaccountsandexpectations
(TverskyandKahneman,1981)
Imaginethatyouhavedecidedtoseeaplay
whereadmissionis$20perticket.Asyou
enterthetheateryoudiscoverthatyouhave
losttheticket.Theseatwasnotmarkedand
theticketcannotberecovered.Wouldyoupay
$20foranotherticket?
Mentalaccountsandexpectations
(TverskyandKahneman,1981)
Imaginethatyouhavedecidedtoseeaplay
whereadmissionis$20perticket.Asyou
enterthetheateryoudiscoverthatyouhave
losta$20bill.Wouldyoustillpay$20fora
tickettotheplay?
Mentalaccountsandexpectations
(TverskyandKahneman,1981)
Imaginethatyouhavedecidedtoseeaplaywhereadmission
is$20perticket.Asyouenterthetheateryoudiscoverthatyou
havelosttheticket.Theseatwasnotmarkedandtheticket
cannotberecovered.Wouldyoupay$20foranotherticket?
[No:54%]
Imaginethatyouhavedecidedtoseeaplaywhereadmission
is$20perticket.Asyouenterthetheateryoudiscoverthatyou
havelosta$20bill.Wouldyoustillpay$20foratickettothe
play?[Yes:88%]
Butinbothproblems,thefinaloutcomeisthesameifyoubuy
theticket:youhavethesameamountofmoneyandyousee
theplay.Whyshouldthesecasesdiffer?
DependenceonRatios(Tverskyand
Kahneman,1981)
Imaginethatyouareabouttopurchaseajacketfor
$250,andacalculatorfor$30.Thecalculator
salesmaninformsyouthatthecalculator[jacket]you
wishtobuyisonsalefor$20[$240]attheother
branchofthestore,located20minutesdriveaway.
Wouldyoumakethetriptotheotherstore?
Results:
68%willingtomakeextratripfor$30calculator
29%willingtomakeextratripfor$250jacket
Note:savesameamountinbothcases:$10.Why
thediscrepency?
NeoclassicalAssumptionsAbout
Preferences
Thechosenoptioninadecisionproblemshould
remainthesameevenifthesurfacedescriptionofthe
problemchanges(descriptiveinvariance)
Contradictedbypseudocertaintyandframingeffects
Thechosenoptionshoulddependonlyonthe
outcomesthatwillobtainafterthedecisionismade,
notondifferencesbetweenthoseoutcomesand
thestatusquo:Contradictedbyendowmenteffect
whatoneexpects:Contradictedbymentalaccounts
theoverallmagnitudeofthedecision:Contradictedbyratio
effect
MoreNeoclassicalAssumptionsAbout
Preferences
Preferencesoverfutureoptionsshouldnot
dependonthetransientemotionalstateofthe
decisionmakeratthetimeofthechoice(state
independence)
EndowmentEffectRevisited(Van
Boven,Dunning,andLoewenstein,
2000)
Replicatedcoffeemugendowmenteffect
Avg.sellingprice:$6.37
Avg.buyingprice:$1.85
Sellers[Buyers]askedtoestimatehowmuchbuyers[sellers]
wouldpay,andrewardedforaccuratepredictions
Sellersestimateofbuyingprice:$3.93
Buyersestimateofsellingprice:$4.39
Resultshowsprojectionbias:estimatesarebiasedtowardPs
emotionalstateatthetimeofestimate(attachedorunattached
tomug)
Validatedforpredictingonesownsellingpricebeforeowninga
mug(Loewenstein&Adler,1995)
Whyyoushouldntshoponanempty
stomach(Read&VanLeeuwen,1998)
Officeworkerschoosebetweenhealthyandunhealthy
snackstobereceivedinaweek
Decisiontimesandprojectedsnackreceptiontimes
eitherwhen
hungry(lateinafternoon)
satiated(rightafterlunch)
Future
Hunger
Results:%choosing
unhealthysnack:
Current
Hunger
Hungry
Satiated
Hungry
78%
42%
Satiated
56%
26%
MoreNeoclassicalAssumptionsAbout
Preferences
Preferencesoverfutureoptionsshouldnot
dependonthetransientemotionalstateofthe
decisionmakeratthetimeofthechoice(state
independence)
Contradictedbyprojectionbias
MoreNeoclassicalAssumptionsAbout
Preferences
Preferencesoverfutureoptionsshouldnot
dependonthetransientemotionalstateofthe
decisionmakeratthetimeofthechoice(state
independence)
Contradictedbyprojectionbias
Preferencesbetweenfutureoutcomesshould
notvarysystematicallyasafunctionofthe
timeuntiltheoutcomes(delayindependence)
TestingDelayIndependence(Ainslie
andHaendel,1983)
Pschosebetweentwoprizestobepaidby
reputablecompany:
1.$50todayversus$100in6months
2.$50in12monthsversus$100in18months
TestingDelayIndependence(Ainslie
andHaendel,1983)
Pschosebetweentwoprizestobepaidby
reputablecompany:
1.$50todayversus$100in6months
2.$50in12monthsversus$100in18months
Mostchose$50todayinproblem1,but$100
in18monthsinproblem2
Violatesdelayindependenceillustrates
hyperbolicdiscounting
TemporalDiscounting
NormativeTheory:exponentialdiscounting
(constantdiscountrate)
TemporalDiscounting
NormativeTheory:exponentialdiscounting
(constantdiscountrate)
DescriptiveTheory:hyperbolic
discounting(temporal
myopia/impulsiveness)
MoreNeoclassicalAssumptionsAbout
Preferences
Preferencesoverfutureoptionsshouldnotdependon
thetransientemotionalstateofthedecisionmakerat
thetimeofthechoice(stateindependence)
Contradictedbyprojectionbias
Preferencesbetweenfutureoutcomesshouldnotvary
systematicallyasafunctionofthetimeuntilthe
outcomes(delayindependence)
Contradictedbyhyperbolicdiscounting/impulsiveness
MoreNeoclassicalAssumptionsAbout
Preferences
Preferencesoverfutureoptionsshouldnotdependonthe
transientemotionalstateofthedecisionmakeratthetimeof
thechoice(stateindependence)
Contradictedbyprojectionbias
Preferencesbetweenfutureoutcomesshouldnotvary
systematicallyasafunctionofthetimeuntiltheoutcomes
(delayindependence)
Contradictedbyhyperbolicdiscounting/impulsiveness
Experiencedutilityshouldnotdiffersystematicallyfrom
decisionutility
predictedutility
retrospectiveutility
TheHarvard/YaleAssistantProfessors
Problem(Anecdotal)
HarvardandYalegranttenuretoveryfewjuniorfaculty
Butprestigeconsiderationsoftencauseacceptanceofjob
offersoverschoolsmorelikelytogranttenure(e.g.Michigan)
Resultcanbeamiserableexperience:dropinstatus,feelingof
failurewhenassistantprofessorshipisover
Possiblyaninstanceofdecisionutility(revealedbychoice)
beinginconsistentwithexperienced(andevenpredicted)utility
AnticipatedbyAdamSmith:peopleexaggerateimportanceof
socialstatus
FailuresofHedonicPrediction
Peopleneglecteffectsofadaptationtosurroundings
inpredictingfutureutility
Misprediction,afterinitial(unpleasant)exposure,of(non)
enjoymentofplainyogurtafter8dailyepisodesof
consumption(Kahneman&Snell,1992)
Changeinsocialcomparisongroup(e.g.teachingat
Harvard/Yale,movingtoanewneighborhood)
Wearinesswithtravelplanningoverlylongvacations,too
muchtimeatthebeach
Assistantprofessorsoverestimateeffectsoftenure
decisiononhappinessoneyearlater(Gilbertand
Wilson,2000)
ATestofHedonicMemory(Kahneman
etal.,1993)
Psgiventwounpleasantexperiences:
Shorttrial:Holdhandin14Cwaterfor60s
Longtrial:Holdhandinwaterfor90s;14Cfor60s,followed
bygradualriseto15Covernext30s
Aftersecondtrial,Pscalledintorepeatoneofthe
twotrialsexactly
65%chosetorepeatthelongtrial
Interpretation:durationneglectpeopleremember
andoverweighttheendoftheexperience(agradual
declineinpain)
ApplicationinClinicalSetting
(RedelmeierandKahneman,1996)
Patientsundergoingcolonoscopyreportedintensityof
painevery60s
Laterprovidedseveralmeasuresofremembered
utilityforthewholeexperience
Rememberedutilityratingsreflectednottotalutility
(additionofpainratings)butapeakandendrule:
highestandlastpainratingsdominatedmemory
Failuretointegratemomentutilities:mayaccountfor
differenceinreportedhappinessbetweenFrenchand
U.S.surveytakers
MoreNeoclassicalAssumptionsAbout
Preferences
Preferencesoverfutureoptionsshouldnotdependonthe
transientemotionalstateofthedecisionmakeratthetimeof
thechoice(stateindependence)
Contradictedbyprojectionbias
Preferencesbetweenfutureoutcomesshouldnotvary
systematicallyasafunctionofthetimeuntiltheoutcomes
(delayindependence)
Contradictedbyimpulsiveness
Experiencedutilityshouldnotdiffersystematicallyfrom
decisionutility:Harvard/Yalejuniorfacultyproblem
predictedutility:Contradictedbyfailuretopredictadaptation
retrospectiveutility:Contradictedbydurationneglectandfailureto
integratemomentutilities
ProspectTheory(Kahnemanand
Tversky,1979;1992)
Prospectsareevaluatedaccordingtoavalue
functionthatexhibits
referencedependence(subjectivelyoriented
aroundazeropoint,defininggainsandlosses)
diminishingsensitivitytodifferencesasonemoves
awayfromthereferencepoint
lossaversion:steeperforlossesthanforgains
TheValueFunction
ProspectTheory(Kahnemanand
Tversky,1979;1992)
Prospectsareevaluatedaccordingtoavaluefunctionthat
exhibits
referencedependence(subjectivelyorientedaroundazeropoint,
defininggainsandlosses)
diminishingsensitivitytodifferencesasonemovesawayfromthe
referencepoint
lossaversion:steeperforlossesthanforgains
Probabilitiesaretransformedbyaweightingfunctionthat
exhibitsdiminishedsensitivitytoprobabilitydifferencesasone
movesfromeithercertainty(1.0)orimpossibility(0.0)toward
themiddleoftheprobabilityscale(0.5)
Refinementofreflectioneffect:riskaversionformediumtohigh
probabilitygainsandlowprobabilitylosses;riskseekingformediumto
highprobabilitylossesandlowprobabilitygains
Someeveryday,observed
consequencesofprospecttheory
(Camerer,2000)
Lossaversion:
Reflectioneffect:
Equitypremiuminstockmarket:stockreturnstoohighrelativetobondreturns
Cabdriversquitarounddailyincometargetinsteadofmakinghaywhilesun
shines
Mostemployeesdonotswitchoutofdefaulthealth/benefitplans
Peopleatquarterbasedschoolspreferquarters,atsemesterbasedschools
prefersemesters
Horseracing:favoritesunderbet,longshotsoverbet(overweightlowprobability
loss);switchtolongshotsatendoftheday
Peopleholdlosingstockstoolong,sellwinnerstooearly
Customersbuyoverpricedphonewireinsurance(overweightlowprobability
loss)
Lotteryticketsalesgoupastopprizerises(overweightlowprobabilitywin)
Moreseriousconsequences
Lossaversionmakesindividuals/societiesunwillingto
switchtohealthierliving(fearlossofincome,
unsustainableluxuries)
Riskseekingforlikelylossescancauseprolonged
pursuitofdoomedpolicies,e.g.warsthatarenot
likelytobewon,choosingcourttrialinsteadof
bargaining
Riskseekingforunlikelygainscanleadtoexcessive
gamblinginindividuals,quixoticpolicieswhenleaders
gettoopowerful
Okay,peoplearerationalanddont
havestablepreferences,butarentthey
atleastbasicallyselfish?
SelfInterestAssumptionin
GameTheory
Choicesingamesshouldalwaysreflectwhat
isbestforthedecisionmaker,i.e.whatwill
maximizethedecisionmakerspayoff
PrisonersDilemma(Tucker,1955)
PrisonersDilemmaLabeling
Experiment(RossandSamuels,1993)
WhenPDislabeledtheWallStreetGame,
only1/3cooperate
WhenitislabeledtheCommunityGame,2/3
cooperate
Showspresenceofbothtendencies
defectionandcooperationwhichcanbe
evokedbysocialsignals
TheUltimatumGame(Guthetal.,
1982)
$100inonedollarbillsavailabletobedivided
betweentwoplayers
Proposerchoosesadivision
Receivercaneither
accept:bothreceiveproposedamounts
reject:bothreceivenothing
HowmuchshouldtheProposeroffer?
TheUltimatumGame(Guthetal.,
1982)
$10inonedollarbillsavailabletobedivided
betweentwoplayers
Proposerchoosesadivision
Receivercaneither
accept:bothreceiveproposedamounts
reject:bothreceivenothing
HowmuchshouldtheProposeroffer?
Ultimatumgameexperiment(Thaler,
1988)
Mostproposersoffer$5(evensplit),oralittle
less,tothereceiver
altruism
Lowoffers($1)usuallyrejected
altruisticpunishment
DictatorGame(Kahnemanetal.,1986)
OneP(studentinclass)askedtodivide$20
betweenselfandotherP.OtherPhasno
choicetoaccept/reject.
Twopossibilities:
Evensplit($10each)
Unevensplit($18forself,$2forother)
Gametheorypredictsunevensplit
76%choseanevensplit
UltimatumandDictatorGamesin
TraditionalSocieties(Henrichetal.,
2001)
Pstested15smallscalesocieties
Ultimatumgame:
Meanoffervariedfrom0.26to0.58(0.44inindustrial
societies)
Rejectionratealsoquitevaried:lowoffersrarelyrejectedin
somegroups,inothershighoffersareoftenrejected
Greatvariationinbehaviorevenamongnearby
groups;dependsondeepaspectsofculture,
experience:
e.g.meatsharingAche(Paraguay)andvillageminded
Orma(Kenya)madegenerousoffers,familyfocused
Machiguenga(Peru)showedlowcooperation
UltimatumandDictatorGamesin
TraditionalSocieties(Henrichetal.,
2001)
Pstested15smallscalesocieties
Ultimatumgame:
Meanoffervariedfrom0.26to0.58(0.44inindustrialsocieties)
Rejectionratealsoquitevaried:lowoffersrarelyrejectedinsome
groups,inothershighoffersareoftenrejected
Greatvariationinbehaviorevenamongnearbygroups;
dependsondeepaspectsofculture,experience:
e.g.meatsharingAche(Paraguay)andvillagemindedOrma(Kenya)
madegenerousoffers,familyfocusedMachiguenga(Peru)showedlow
cooperation
Generalconclusion:thereisnosuchthingashomo
economicus;cooperationbehaviorishighlyvariable,heavily
determinedbyculturalnorms
EffectsofaNormofSelfInterest
Peopledescribeinvolvementinsocialcauseasbeing
moreselfinterestmotivatedthanitis(Miller,1999)
VotingbehaviorinU.S.becomingmoreselfinterest
driven(McCarty,inpress),mayreflectshifttoward
greaternormofselfinterestinpolitics
Economictheoriesandlanguagecanbecomeself
fulfilling(Ferraro,Pfeffer,andSutton,2003)
Finally
Cognitivepsychologycaninformmanyareas
ofresearch,theory,andpracticethatdonot
nowproperlyincorporateitsinsights
Theoriesaboutpsychology,whethergrounded
indataornot,canhaveprofound
consequencesforthewaywelive