The Neural Basis of Economic Decision-Making in the Ultimatum Game

Author(s): Alan G. Sanfey, James K. Rilling, Jessica A. Aronson, Leigh E. Nystrom, Jonathan
D. Cohen
Source: Science, New Series, Vol. 300, No. 5626 (Jun. 13, 2003), pp. 1755-1758
Published by: American Association for the Advancement of Science
Stable URL: http://www.jstor.org/stable/3834595 .
Accessed: 15/03/2011 04:23

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=aaas. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and
extend access to Science.

http://www.jstor.org

ide-sensitive factor)(33-35) or AP2 (36) may
be requiredfor exportof functionalheteromeric
receptorcomplexesand/orstabilizationof these
complexesat the cell surface.
Our results suggest that phosphorylation
of GluR2 Ser880is necessary for LTD induc-
tion. In Purkinjecells transfectedwith GluR2
K882A, which presumably contain at least
some GluR2 K882A/GluR3 and/or GluR2
K882A/GluR4c heteromeric receptor com-
plexes, the presence of a PKC consensus site
on subunits other than GluR2 appearsto be
insufficient to enable LTD. It is unknown
whether the correspondingserines on GluR3
and GluR4c are indeed phosphorylatedby
PKC or if upstreamsequence differences and
differentialprotein binding renderthese sub-
units incapable of supportingLTD.
Previous attemptsto test the involvement
of cerebellar LTD in motor learning para-
digms have relied on drugsor genetic manip-
ulations that act early in the LTD induction
signaling cascade, either at receptors or sec-
ond messengers (1). These studies have been
limited owing to the nonspecific natureof the
manipulations(e.g., disruptionof mGluRl or
PKC function). A GluR2 K882A knock-in
mouse could provide the first strong test of
the hypothesisthatcerebellarLTD is required
for certain forms of motor learning.
References and Notes
1. M. F. Bear,D. J. Linden,in TheSynapse,W. M. Cowan,
T. Sudhof, C. F. Stevens, Eds. (Johns Hopkins, Balti-
more, MD,2000), pp. 455-517.
2. C. Hansel, D. J. Linden,E. D'Angelo,Nature Neurosci.
4, 467 (2001).
3. F. Crepel,M. Krupa,BrainRes. 458, 397 (1988).
4. D. J. Linden,J. A. Connor,Science 254, 1656 (1991).
5. N. A. Hartell,NeuroReport5, 833 (1994).
6. C. I. De Zeeuw et al., Neuron 20, 495 (1998).
7. J. H. Freeman,T. Shi, B. G. Schreurs,NeuroReport9,
2237 (1998).
8. J. Goossens et al., J. Neurosci. 21, 5813 (2001).
9. M. Ito, M. Sakurai,P. Tongroach,J. Physiol. (London)
324, 113 (1982).
10. D. J. Linden,M. H. Dickinson,M.Smeyne, J. A. Connor,
Neuron 7, 81 (1991).
11. D. J. Linden,Learn.Mem.(ColdSpringHarbor)1, 121
(1994).
12. K. Narasimhan,D. J. Linden,Neuron 17, 333 (1996).
13. D. J. Linden,Proc. Natl. Acad. Sci. U.S.A. 98, 14066
(2001).
14. S. Matsuda,T. Launey,S. Mikawa,H. Hirai,EMBOJ.
19, 2765 (2000).
15. Y.-T.Wang, D. J. Linden,Neuron 25, 635 (2000).
16. J. Xia, H. J. Chung,, C. Wihler, R. L. Huganir, D. J.
Linden,Neuron 28, 499 (2000).
17. S. Matsuda, S. Mikawa, H. Hirai,J. Neurochem. 73,
1765 (1999)-
18. H. J. Chung J. Xia, R. H. Scannevin, X. Zhang, R. L.
Huganir,J. Neurosci. 20, 7258 (2000).
19. J. L. Perez et al., J. Neurosci. 21, 5417 (2001).
20. H. Dong et al., Nature 386, 279 (1997).
21. P. Osten et al., Neuron 27, 313 (2000).
22. B. E. Kemp, R. B. Pearson, TrendsBiochem. Sci. 15,
342 (1990)-
23. Single-letter abbreviations for the amino acid resi-
dues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F,
Phe;G, Gly;H, His;I, Ile;K,Lys;L,Leu;M, Met; N, Asn;
P, Pro;Q, Gln;R, Arg, S, Ser;T, Thr;V, Val;W, Trp;X,
any amino acid; and Y, Tyr.
24. A. Y. Hung, M. Sheng, J. Biol. Chem. 277, 5699
(2001 ).
25. We obtained polyclonalantiserumdirected against a
www.sciencemag.org SCIENCE VOL300
REP O RT S
GluR2 carboxy-terminalpeptide phosphorylated at 34. P. Osten et al., Neuron 21, 99 (1998).
Ser880as previouslydescribed(18). To account for the 35. I. Song et al., Neuron 21, 393 (1998).
possibilitythat mutation of Lys882interfereswith the 36. S. H. Lee,L.Liu,Y.-T.Wang, M.Sheng,Neuron 36, 661
recognitionof phosphowlatedSer880by this antibody, (2002).
we purifiedpolyclonalantiserumon an affinitycolumn 37. Thanksto R. Bock for skillfultechnical assistance, K.
consistingof a bovineserumalbumin(BSA)-conjugated Takamiyafor advice on PCRgenotyping, and mem-
GluR2carbo -terminal peptide containingthe K882A bers of the Huganirand Lindenlabs for helpful com-
mutation (KWGIEpS*VAI; pS* = phosphoSer880). ments. Thiswork was supported by HHMIand USPHS
26. N. Burnashev,H. Monyer,P. H. Seeburg, B. Sakmann, NS36715 (R.L.H.),USPHS MH51106 and MH01590
Neuron 8, 189 (1992). (D.J.L.)and the Develbiss Fund (D.J.L.).
27. Z. Jia et al., Neuron 17, 945 (1996).
Supporting Online Material
28. M. I. Daw et al., Neuron 28, 873 (2000).
29. C. H. Kim et al., Proc. Natl. Acad. Sci. U.S.A. 98: www.sciencemag.org/cgi/content/full/300/5626/1 75 1/
11725 (2001). DC1
M. Eto, R. Bock, D. L. Brautigan,D. J. Linden,Neuron Materialsand Methods
30.
36, 1145 (2002). Fig. S1
31. M. Ito, Physiol. Rev. 81, 1143 (2001). References
32. See supportingdata on Science Online.
33. A. Nishimune et al., Neuron 21, 87 (1998). 29 January2003; accepted 6 May 2003
The Neural Basis of Economic
Decision-Making in the
Ultimatum Game
Alan G. Sanfey,1 3* James K. Ritling,* Jessica A. Aronson,2
Leigh E. Nystrom,l 2 Jonathan D. Cohen#2 4
The nascent field of neuroeconomicsseeks to groundeconomic decision-
makingin the biologicalsubstrateof the brain.We used functionalmagnetic
resonanceimagingof UltimatumGameplayersto investigateneuralsubstrates
of cognitiveandemotionalprocessesinvolvedin economicdecision-making. In
thisgame,two playerssplita sumof money;one playerproposesa divisionand
the other can acceptor rejectthis. We scannedplayersas they respondedto
fairandunfairproposals.Unfairofferselicitedactivityin brainareasrelatedto
both emotion (anteriorinsula)and cognition(dorsolateralprefrontalcortex).
Further,significantlyheightenedactivityin anteriorinsulafor rejectedunfair
offers suggests an importantrole for emotionsin decision-making.
Standardeconomicmodels of humandecision- UltimatumGame.In the UltimatumGame,two
making (such as utility theory)have typically playersare given the opportunityto split a sum
minilliizedor ignoredthe influenceof emotions of money. One player is deemed the proposer
on people's decision-makingbehavior,idealiz- and the other, the responder.The proposer
ing the decision-makeras a perfectlyrational makesan offer as to how this money shouldbe
cognitive machine. However, in recent years split between the two. The second player (the
this assumptionhas been challengedby behav- responder)can eitheracceptor rejectthis offer.
ioraleconomists,who have identifiedadditional If it is accepted,the money is split as proposed,
psychologicaland emotionalfactorsthat influ- butif the responderrejectsthe offer,thenneither
ence decision-making(1, 2), and recently re- player receives anything.In either event, the
searchershave begun using neuroimagingto game iS over.
examinebehaviorin economicgames (3). This The standardeconomicsolutionto the Ulti-
study applies functional neuroimagingtech- matum Game is for the proposerto offer the
niquesto investigatethe relativecontributions of smallestsum of moneypossibleto the respond-
cognitive and emotional processes to huma er and for the responderto accept this offer,
social decision-making. on the reasonablegroundsthat any monetary
The limitationsof the standardeconomic amountis preferableto none. However,consid-
model are effectively illustratedby empirical erablebehavioralresearchin industrializedcul-
findings from a simple game known as the turesindicatesthat,irrespectiveof the monetary
sum, modal offers aretypicallyaround50% of
the total amount.Low offers (around20% of
'Center for the Study of Brain,Mindand Behavior, the total) have about a 50% chance of being
2Departmentof Psychology,3Centerfor Healthand
Well-Being, Princeton University, Princeton, NJ
rejected(4-8). This latter,quiterobust,experi-
08544, USA.4Departmentof Psychiatry,Universityof mental finding is particularlyintriguing,dem-
Pittsburgh,Pittsburgh,PA 15260, USA. onstratingthat circumstancesexist in which
*These authors contributed equally to this manu- people are motivated to actively turn down
script. monetaryreward.
13 JUNE2003 1755
 REP O RT S

Fig. 1. (A)Timeline for a single A B

roundof the UltimatumGame.
Each round lasted 36 s. Each
roundbeganwith a 12-s prep- 12 sec 6 sec
aration interval. The partici-
pant then saw the photograph EES E X
and name of their partnerin + | l 2i K
that trial for 6 seconds. A pic- | | | | |
ture of a computerwas shown
if it was a computertrial,or a t t t
roulettewheel if it was a con-
trol trial. Next, participants Reveal Reveal Revec!al Reveal Reveal
saw the offer proposedby the Fixation Partner Offerr Options Outcome
partnerfor a further6 s, after $6:$5 $7:$3 Se:$2 $o:$t
which they indicatedwhether Ofbr
they accepted or rejectedthe offer by pressingone of two buttons on participantssaw five $5:$5 offers, one $7:$3 offer, two $8:$2 offers,
a button box. (B) Behavioralresults from the UltimatumGame. These and two $9:$1 offers from both human and computer partners (20
are the offer acceptance rates averaged over all trials. Each of 19 offers in total).

Why do people do this? The game is so
simple that it is improbablethat these rejec-
tions are due to a failure to understandthe
rules of the game, or an inability to concep-
tualize a single-shot interaction with a part-
ner (9). On the basis of participantreports,it
appears that low offers are often rejected
after an angry reaction to an offer perceived
as unfair (10). Objecting to unfairness has
been proposed as a fundamental adaptive
mechanismby which we assertand maintaina
social reputation(11), and the negative emo-
tions provokedby unfairtreatmentin the Ul-
timatum Game can lead people to sacrifice
sometimesconsiderablefinancialgain in order
to punish their partnerfor the slight. Unfair
offers in the UltimatumGame induce conflict
in the responderbetween cognitive ("accept")
and emotional("reject")motives, motives that
we might expect to see representedin brain
areas implicated in cognitive and emotional
modes of thought,with additionalregionspos-
sibly mediatingthese competinggoals (12).
To shed light on the neuraland psycholog-
ical processes mediating such behaviors, we
scanned 19 participantsusing functionalmag-
netic resonanceimaging (fMRI), each in the
role of the responderin the UltimatumGame.
We were interestedin neural and behavioral
reactionsto offers which were fair (the money
is split 50:50) orunfair(theproposerofferedan
unequalsplit to his or her advantage).In par-
ticular, we hypothesized that unfair offers
would engage neural structuresinvolved in
both emotionaland cognitive processing,and
that the magnitudeof activationin these struc-
turesmight explainvariancein the subsequent
decisionto accept or rejectthese offers.
Before scanning,each participantwas intro-
ducedto 10 peopletheyweretoldwouldpartner
withthemin thegamesto follow.Theyweretold
that they would play a single iterationof the
game with each partnerand thattheirdecisions
with each partnerwould not be revealedto the
other partnersand, therefore,could not affect
subsequentoffers. The participantswere then
placedinsidethe MRI scannerand beganplay-
ing the UltimatumGamewith theirparMersvia
a computerinterface(Fig. 1A) (13). They com-
pleted30 roundsin all, 10 playingthe gamewith
a human partner(once with each of the 10
partners),10 with a computerpartner,and a
further10 controlroundsin which they simply
receivedmoney for a buttonpress. The rounds
were presentedrandomly,andall involvedsplit-
ting $10. Offersmadeby humanpartnersin fact
a&ered to a predeterniinedalgorithm,which
ensuredthat all participantssaw the same set
(anda fullrange)of offers(14, 15) Halfof these
offers were fair, that is, a proposalto split the
$10 evenly($5:$5),withthe remaininghalfpro-
posing unequalsplits (two offers of $9:$1, two
offersof $8:$2, and one offer of $7:$3).The 10
offers fromthe computerpartnerwere identical
to those fromthe humanpartners(halffair,half
unfair).The 10 controltrialswere designedto
controlfor the responseto monetaxyreinforce-
ment,independentof the social interaction.The
distributionof offersgenerallymimicsthe range
of offers typically made in uncontrolledver-
sions of the game (i.e., involving freely acting
humanpartners).
Behavioral results were very similar to
those typically found in Ultimatum Game
experiments (Fig. 1B) (16). Participantsac-
cepted all fair offers, with decreasing accep-
tance rates as the offers became less fair.
Unfair offers of $2 and $1 made by human
partnerswere rejected at a significantlyhigh-
er rate than those offers made by a computer
($9:$1 offer: x2 = 5.28, 1 df, P-0.02; $8:$2
offer: x2 = 8.77, 1 df, P - 0.003), suggesting
that participants had a stronger emotional
reactionto unfair offers from humansthan to
the same offers from a computer(17).
Amongthe areasshowinggreateractivation
forunfaircomparedwith fairoffersfromhuman
partners(Fig.2, A andB; tableS1) werebilateral
anteriorinsula, dorsolateralprefrontalcortex
(DLPFC),and antenorcingulatecortex(ACC).
The magnitudeof activationwas also signifi-
cantlygreaterfor urlfairoXersfromhumanpart-
ners as comparedto both unfair offers from
computerpartners(left insula:t = 2.52, P <
0.02; rightinsula:t - 2.2, P < 0.03) and low
controloffers (left insula:t = 3.46, P <0.001,
rightinsula:t = 2.83, P < 0.05). This suggests
thatthese activationswere not solely a function
ofthe amountof moneyofferedto theparticipant
but ratherwere also uniquely sensitive to the
context,namelyperceivedunfairtreatmentfrom
anotherhuman(Fig. 2, C and D). Further,re-
gions of bilateralanteriorinsula demonstrated
sensitivityto the degreeof unfairnessof an offer,
exhibitingsignificantlygreateractivationfor a
$9:$1 offer than an $8:$2 offer from a human
parter (Fig. 2E) (left insula,P < 0.001; right
insula,P < 0.01), in additionto the aforemen-
tioned greateractivationfor unfairoffers than
fair ($5:$5) offers.
Activationof bilateralanteriorinsulato un-
fair offers from humanpartnersis particularly
interestingin light of this region's oft-noted
associationwith negativeemotionalstates.An-
terior insula activationis consistentlyseen in
neuroimagingstudiesof pain and distress(18-
20), hungerand thirst(21, 22), and autonomic
arousal(23). Thisregionhas also been implicat-
ed in studiesof emotion,in particularinvolve-
ment in the evaluationand representationof
specific negative emotionalstates (24). Chief
amongstthese are anger and disgust, both of
which have been found to engage a distinct
region of the anteriorinsula activatedby an
unfair offer in the present study (25, 26).
Thoughstudiesof disgusthave largelyfocused
on physicalsensationsof tasteand odor(27), it
has been suggestedthat emotion-baseddisgust
(as perhaps induced by an insultinglyunfair
offer)may be conceptuallysimilar.The recruit-
ment of similarneural structures,namely the
anteriorinsula,in both physicaland moraldis-
gust gives some credenceto this notion.
If the activationin the anteriorinsula is a
reflectionof the responders'negativeemotional
response to an unfair offer, we might expect
activityin this regionto correlatewith the sub-
sequentdecision to accept or reject the offer.
Because all fair offers and the vast majorityof
$7:$3 offers were accepted,we focused on the
$8:$2and$9:$1offersfroma humanpartnerfor
the analysisof whetherneuralactivitywas re-
latedto the decisionmade in the game. Indeed,
looking at the participantlevel, those partici-

13 JUNE 2003 VOL 300 SCIENCE www.sciencemag.org

1756
 o-u@d * a s a a B E 5 B s a a s 5 a 5 b b E s E s a B Fair $9:$1
Person
$5:$5

R EPO RT S

Fig. 2. Activationrelated A B v UnfairPerson

to the presentationof an C O.5.
- - - UnfairComputer
unfair offer. (A) Map of a)
- -A- FairComputer
the t statistic for the - 0.4
s
contrast [unfair human
offer- fair humanoffer] S 0.3
showingactivationof bi- -

lateralanteriorinsulaand - 0.2
._

anteriorcingulatecortex. tn
z 0.1
Areas in orange showed
greateractivationfollow-
ing unfair as compared w. o

with fair offers (P < 3

- -0.1
0.001). (B) Map of the t
CZ
statistic for the contrast
-0.2 . . . . . . . . . . . . . . . . . . . . . . . . .

[unfairhumanoffer- fair -6 0 6
human offer] showing D o.5. + UnfairPerson 12 18
activationof rightdorso- _ Fair Person 0.5
a
lateral prefrontalcortex. - - - UnfairComputer
(C) Event-relatedplot for = 0.4 - -A- FairComputer X 0.4-
unfair and fair offers in S - 4- $8:$2
right anteriorinsula.The A 0.3 8 0.3
o
-

offerwas revealedat t = -
-

0.2 c, 0.2-
0 on the x axis. (D) X
._

Event- related plot for

._

n -

unfair and fair offers in z 0.1 CK0.1

left anterior insula. (E) . .

Event-related plot for -

-
X 0s

different human unfair n

and fair offers in subset = -0.1 --0.1

i i
of left anteriorinsula. - - - - - - - - - - - - - - - - -
-u .g . | . B . z . ^ ^ . . . . . . .
12
-6 0 6 12 18 -3 0 3 6
Scan time (seconds) Scan time (seconds)
9

pantswith strongeranteriorinsulaactivationto thatthisregionmay be competingwith emotion-

unfair offers rejected a higher proportionof al areasin influencingthe decision,we examined A 2

these offers(rightinsula:correlationcoefficient the balancebetweenactivationin anteriorinsula X <,s,

1.5
r = -0.45, P = 0.025, one-tailed;left insula: and DLPFCfor unfairoffers.Unfairoffers that 3@
C _

r = -0.39, P = 0.05, one-tailed)(Fig. 3A). Of are subsequentlyrejectedhave greateranterior .oU

.; 1

particularinterestis whetherthesedifferencesin insulathanDLPFCactivation,whereasaccepted < c 0.5 * * -

anteriorinsular activationextend to the trial offersexhibitgreaterDLPFCthananteriorinsula sr Q

level. Lookingacrossparticipants,an examina- (Fig. 3B). The contrastin activationbetween o

.

tion of individualtrialsalso revealeda relation these two areasis significantlydifferentfor ac- -U.D

0 20 40 60 80 100

between right anteriorinsularactivity and the cepted and rejected offers (P = 0.033, one- Acceptance rates (%)

decisionto acceptor reject(Fig. 3B). Activation tailed),consistentwith the hypothesisthatcom-

in this areawas significantlygreaterin response petition between these two regions influences B 1.2.

to unfair offers that were later rejected(P = behavior. DLPFC activity remains relatively x Accepted
0.028, one-tailed).These resultsprovide addi- constantacrossunfairoffers,perhapsreflecting a Rejected
tional supportfor the hypothesis that neural the steadytaskrepresentation of moneymaximi-
zation,with anteriorinsulascalingmonotonical- tn 0.8
representations of emotionalstatesguidehuman -
T -
decision-maliing. ly to the degree of unfairness,reflectingthe
,x, 0.6
In contrastto the insula,DLPFCusuallyhas emotionalresponseto the offer.Cautionis need- c

been linkedto cognitiveprocessessuch as goal ed when comparingthe magnitudeof the fMRI

maintenanceand executive control (28, 29). signalacrossbrainregions.However,it is inter- 0.4 T
Thus, the DLPFC activationwe observed in esting to note thatthe outcomeof the decision
0.2
responseto unfairoffers may relateto the rep- may reflect the relative engagementof these
resentationandactivemaintenanceof the cogni- regions, with greateranteriorinsula activation
O
tive demandsof the task, namely the goal of biasingtowardrejectionand greaterDLPFCbi-
R. Anterior Insula R. DLPFC
accumulatingas much money as possible. An asing towardacceptance.Finally,unfairoffers
unfairoffer is more difficultto accept,as indi- were also associatedwith increasedactivityin Fig. 3. (A) Acceptance rates of unfairoffers
plotted against rightanteriorinsulaactivation
catedby thehigherrejectionratesof theseoffers, ACC. ACC has been implicatedin detectionof for each participant.(B) Right anteriorinsula
and hence higher cognitive demandsmay be cognitive conflict (30, 31), and activationhere and right DLPFC activationfor all unfairoffer
placed on the participantin orderto overcome may reflectthe conflictbetween cognitiveand trials,categorizedby subsequentacceptanceor
the strongemotionaltendencyto rejectthe offer. emotionalmotivationsin the UltimatumGame. rejection.
AlthoughDLPFCactivatedto unfairoffers,this This studysoughtto identifythe neuralcor-
activationdidnot correlatewith acceptancerates relatesof fairnessand unfairness,and in partic- sentialto many aspectsof societalandpersonal
(r = 0.04, P > 0.05), suggestingthatactivation ular the relativecontributionsof cognitive and decision-makingand underliesnotions as di-
of this region alone is not sufficientto predict emotionalprocessesto humandecision-making. verse as ethics,socialpolicy, legal practice,and
behavior.However,motivatedby the hypothesis A basic sense of fairnessand unfairnessis es- personalmorality.Ourresultsareconsistentwith

www.sciencemag.org SCIENCE VOL300 13 JUNE2003 1 757

 R EP O RT S
the idea that the areas of anteriorinsula and
DLPFCrepresentthe twin demandsof the Ulti-
matumGametask,the emotionalgoal of resist-
ing unfairnessarldie cognitivegoal of accumu-
latingmoney, respectively.Further,our finding
that activityin a region well known for its in-
volvementin negativeemotionis predictiveof
subsequentbehaviorsupportsthe importanceof
emotionalinfluencesin humandecision-making.
We believe that these findings,and work that
proceeds from them, will provide a more de-
tailed characterizationof specific emotionalre-
sponses, their neuralsubstrates,and the social
circumstarlcesunder which they are elicited.
Therefore,not only do ourresultsprovidedirect
empiricalsupportfor economicmodels thatac-
knowledgethe influenceof emotionalfactorson
decision-making behavior,but they also provide
the firststep towardthe developmentof quanti-
tativemeasuresiat may be usefill in constain-
ing ie social uiilityfinction in economicmod-
els (32, 33). Models of decision-makingcannot
affordto ignoreemotionas a vital arlddynarnic
componentof our decisionsand choices in ie
realworld.
References and Notes
1. C. Camerer,G. Loewenstein,in Advancesin Behavioral
Economics, C. Camerer, G. Loewenstein, M. Rabin,
Eds. (PrincetonUniv. Press, Princeton,NJ), in press.
2. G. Loewenstein,J. Lerner,in The Handbookof Affec-
tive Science, R. J. Davidson, H. H. Goldsmith, K. R.
Scherer,Eds. (Oxford Univ. Press, Oxford, 2003).
3. K. McCabe,D. Houser, L. Ryan,V. Smith, T. Trouard,
Proc. Natl. Acad. Sci. U.S.A. 98, 11832 (2001).
4. W. Guth, R. Schmittberger, B. Schwarze, J. Econ.
Behav. Organ. 3, 376 (1982).
5. R. H. Thaler,J. Econ. Perspect. 2, 195 (1988).
6. G. E.Bolton,R.Zwicic,GameEcon.Behav.10, 95 (1995).
7. A. E. Roth, in Handbookof ExperimentalEconomics,
J. H. Kagel, A. E. Roth, Eds. (Princeton Univ. Press,
Princeton,NJ, 1995).
8. See J. Henrichet al. [Am. Econ. Rev. 91, 73 (2001)]
for Ultimatum Game research in simple societies.
9. C. Camerer, R. H. Thaler,J. Econ. Perspect. 9, 209
(1995).
10. M. M. Pillutla,J. K. Murnighan,Organ. Behav. Hum.
Dec. Proc. 68, 208 (1996).
11. M. A. Nowak, K. M. Page, K. Sigmund,Science 289,
1773 (2000).
12. We use the term "cognitive"here, in place of the terrn
"rational" (as commonlyused in the traditionaleconom-
ic literature),in recognitionof the fact that emotional
responsesmay also have a rationalbasis (e.g.,to punish
unfairoffers).Theterm "cognitive"is perhapsalso prob-
lematic, for similarreasons.Termssuch as "proximal"
and "distal"may be more accurate,respectivelyindicat-
ing the immediate and longer-term sources of gain
associated with the behavior.Idowever,until the field
convergeson a new set of acceptedterms for designat-
ing these classes of motivation,we use the terms cog-
nitive and emotional as intuitivelyaccessible, if not
techni0Uy accurate.
13. Materials and methods are available as supporting
material on Science Online.
14. Thismethodologydeviatessomewhatfromthe standards
of experimentaleconomics, a fleld that generallypro-
scribesthe use of deception[see (34) for a summaw of
the issues,though there are some exceptions(35)]. We
choseto use a limitedamountof deceptionin the current
study primarilybecauseof the heavylogisticdemandsof
an fMRIstudy,requiringa full distributionof offers in a
constrainednumberof participants.Practicalissues not-
withstanding,we believethe use of deceptionhadlittleif
any impacton ourresults,andanyeffect was not likelyto
confound their interpretation.Duringthe post-experi-
ment debriefing,no subject gave any suggestion that
1 758 13 JUNE 2003
they had been suspiciousof the offers they received. suggeststhat subjectstreatedthe UltimatumGameas a
Further,the behavioralresults in the human partner single-shotgame, as instructed.
conditionreplicatethose found in versionsof the game 17. We asked our participantsas part of the debriefing
usingno deception,with aproximately half of offers of process what they considered a "fair"offer to be
20%or less of the total beingrejected(9). Perhapsmost irrespectiveof their decision to accept or reject, thus
importantly,if subjectssuspecteddeception,this should providingan indicationof their standardsof fairness.
Of our participants, 58% considered any offer less
have diminishedemotional responses (i.e., if subjects
suspected the offers to be fictitious, their emotional than $5:$5 as unfair,with the remaining42% deem-
reactionsto these offers,particularlyunfairoffers,should ing anything less than S7:$3 to be an unfairdivision.
have been muted).The fact that we observedsignificant 18. S. W. Derbyshireet al., Pain 73, 431 (1997).
effects consistent with emotional responses suggests, 19. M. J. Iadarolaet al., Brain121, 931 (1998).
once again,that the effects of deceptionwere minimal 20. K.C. Evanset al., J. Neurophysiol. 88, 1500 (2002).
and, if they were present,have simplycausedan under- 21. D. Denton et al., Proc. Natl. Acad. Sci. lJ.S.A. 96,
estimate of the observedeffects.Althoughwe are sensi- 5304 (1999)
tive to the issue of deception,we believethat the meth- 22. P. A. Tataranniet al., Proc. Natl. Acad. Sci. U.S.A.96,
odologicalconstraintsof fldRIjustifiedour practiceand 4569 (1999)-
that the findingsdo not appearto be taintedby subjects' 23. H. D. Critchley, R. Elliott, C. J. Mathias, R. J. Dolan,
possibleperceptionsof the deceptionused. J. Neurosci. 20, 3033 (2000).
15. A common concem regardingthe use of deception 24. A. J. Calder,A. D. Lawrence,A. W. Young,Nature Rev.
Neurosci. 2, 352 (2001).
involvespossiblecontaminationof the participantpooL
As mentioned previously,rejectionrates in the current 25. A. R.Damasio et al., Nature Neurosci.3, 1049 (2000).
study replicate those typically reported from uncon- 26. M. L. Phillipset al., Nature 389, 495 (1997).
trolled UltimatumGame studies;therefore,we do not 27. P. Rozin,A. E. Fallon,Psychol. Rev. 94, 23 (1987).
believe we suffered unduly from this. Furthermore,a 28. E. K.Miller,J. D. Cohen,Annu. Rev. Neurosci. 24, 167
comparisonof rejection rates over the course of the (2001).
experiment (i.e., longitudinallyover participants)indi- 29. A. D. Wagner, A. Maril, R. A. BJork,D. L. Schacter,
cates no systematic trends in these rates (mean rejec- Neurolmage 14, 1337 (2001).
tion rate of offers for first six participantswas 32%; 30. A. W. MacDonaldIII,J. D. Cohen, V. A. Stenger, C. S.
mean rate for last six participantswas 35%). Carter,Science 288, 1835 (2000).
16. After the condusion of the UltimatumGame with all 31. M. Botvinick,L.E.Nystrom, K.Fissell,C. S. Carter,J. D.
Cohen, Nature 40Z, 179 (1999).
partners,subjects then played a single round of the
32. E. Fehr,K. M. Schmidt, Q. J. Econ. 114, 817 (1999).
Prisoner'sDilemma(PD)game with each of the partners. 33. G. E.Bolton,A Ockenfels,Am.Econ.Rev.90, 166 (2000).
Thisraisesthe possibilitythat subjectsdid not treat the
34. R Hertwig, A. Ortmann, Behav. Brain Sci. 24, 383
UltimatumGameas a true single-shotgame. We do not
(2001).
believe playingthe PD game affected their play in the
35. S. Bonetti,J. Econ. Psychol. 19, 377 (1998).
UltimatumGame in this study for severalreasons.First,
36. We would like to thank D. Kahneman,A. Scheres,and
our behavioralresultssupportthe notion that the Ulti-
three anonymous reviewers for their helpful com-
matumGamewas playedas a single-shotgame.As noted
ments. This work was supported in part by grants
above, the proportionof rejected offers in our study
from the Seaver Institute and the Mind,Brain,Body,
matches proportionsreportedin the experimentaleco-
and Health Initiative.
nomic literaturewhen the game is strictlycontrolledas
single-shot.We wouldhaveexpectedmuchhigherrejec- Supporting Online Material
tion rates in an iteratedUltimatumGame.Second,un- www.sciencemag.org/cgi/content/fulU300/5626/1 755/
puished data of ours using a single-shot Ultimatum DC1
Game(withno subsequenttask)producedrejectionrates Materialsand Methods
of unfairoffers that are virtuallyidenticalto those re Table S1
ported here (S8:S2 split, 47% versus 49%; $9:$1 split,
61% versus 60%). We believe this evidence strongty 31 January2003; accepted 15 April2003
V1 Neurons Signal Acquisition of
an Internal Representation of
Stimulus Location
Jitendra Sharma,l 2* Valentin Dragoi,1 2
Joshua B. Tenenbaum,l Earl K. Miller,12 3 Mriganka Surl 2*
A fundamental aspect of visuomotor behavioris deciding where to look or move
next. Under certain conditions, the brain constructs an internal representation
of stimulus location on the basis of previous knowledge and uses it to move
the eyes or to make other movements. Neuronal responses in primaryvisual
cortex were modulated when such an internal representation was acquired:
Responses to a stimulus were affected progressively by sequential presentation
of the stimulus at one location but not when the location was varied randomly.
Responses of individual neurons were spatially tuned for gaze direction and
tracked the Bayesian probabilityof stimulus appearance. We propose that the
representation arises in a distributed cortical network and is associated with
systematic changes in response selectivity and dynamics at the earliest stages
of cortical visual processing.
To assess whethermonkeys(Macacamu- information about future stimulus locations
of stim-
latta)formaninternalrepresentation could be acquiredprogressively with succes-
ulus location,we devised a task in which sive trials in one experimentalcondition but
VOL 300 SCIENCE www.sciencemag.org