Emotions in a repeatedCournot game

A psychophysiological experiment

Ilkka Leppanen and Raimo P. HamalainenSystems Analysis Laboratory

Aalto University

July 14, 2014

Why study emotions in OR?

I To understand behavior in decision andnegotiation support

I Are some processes or models emotionallybetter/worse; or more acceptable/stimulatingthan others?

I To understand the impact of different ways ofcommunication

I Take into account the role of emotions indeveloping modeling and decision supportapproaches

Do emotions play a role in cooperation?I It is known that subjects cooperate in repeated

interactions; two explanations:I Reputation seeking: entirely self-regardingI Reciprocal fairness: other-regarding (Sobel,

2005, Bowles and Gintis, 2011)I Standard methods in experimental economics cannot

distinguish other-regarding behavior fromself-regarding behavior in repeated interactions (Fehr,2009)

Measuring emotions can reveal whethercooperation is other-regarding orself-regarding

EmotionsI Changes in bodily states , triggered by the

brainI Responses to actual or recalled perceptionsI Operate alongside cognitionI Not necessarily observable from outside

I Can be studied byI Neurophysiology (brain imaging)I Self-reportsI Psychophysiological methods

ArousalI A bodily response that prepares the subject

to act (Bechara and Damasio, 2005)I Activation of the autonomic nervous system

leading to a condition of sensoryalertness, mobility and readiness torespond

I Arousal occurs when a subject experiencesas well as anticipates situations

We measure by the skin conductanceresponse (SCR)

Emotional expressionsI Have both functional and communication rolesI Can accompany both real and abstract stimuli

We measure by facial surface electromyography (EMG)

Anger Disgust Positive affect

Muscles related to emotional expressions

Earlier research on emotions in gamesI Ultimatum games:

I Chapman et al. (2009): the responder’s disgustis higher for unfair offers than for fair offers

I Public goods games:I Joffily et al. (2011): arousal is higher when the

subject does not cooperate and when thesubject learns that he has cooperated less thanothers

I Other:I Ben-Shakhar et al. (2007): arousal is related to

punishments in a power-to-take gameI Cannon et al. (2011): anger, disgust, and

positive affect predict moral judgements

Our experiment: repeated Cournot game

”Example of human conflict between cooperation anddefection” (Fouraker and Siegel, 1963)

I Step 1: indicate your ideal resultI Step 2: choose your production quantityI Step 3: view resultsI Repeated for 20 rounds, duration not known

by the players

Payoff matrixPayoff increases as the other’s production quantity decreases, and vice versa

J = joint-optimum (72,72), N = Cournot-Nash equilibrium (64,64)

Experimental arrangement

Pilot subject demonstratingthe equipment

I 44 subjects (24 female)I Mean age 26.05I One pair at a timeI Complete anonymityI Mean reward 22.26 eur

Scoring and analysis of the signals

I Arousal : integrated SCR (Benedek andKaernbach, 2010) over a 5 s time windowwhen the subject sees the results

I Anticipatory arousal : sum of SCRamplitudes over a varying-length timewindow during decision making, divided bythe length of the time window

I Anger , disgust , positive affect : meanEMG amplitude over a 5 s time windowwhen the subject sees the results

Explanatory variablesI Fair result : own payoff at least as high as the other’s

payoffI Payoff share : own payoff divided by the sum of own and

the other’s payoffI Own ideal payoff difference : own payoff minus own ideal

payoffI Other’s ideal payoff difference : the other’s payoff minus

ideal payoff to the otherI Own choice : if high, the subject is not cooperativeI Other’s previous round choice : if high, the other is not

cooperativeI GenderI Round of play

Cournot game: not much cooperation

Only two pairs out of 22 cooperate for more thantwo consecutive rounds

I Mean payoff 55.24, lower than the Cournotequilibrium payoff (64)

I Less cooperation than in similar Cournotduopoly experiments of Huck et al. (2001)and Potters and Suetens (2013)

I Possibly explained by complete anonymity inthe experiment

Arousal higher in fair than in unfair results

unfair fair

0.0

0.5

1.0

1.5

2.0

p=0.0017ar

ousa

l (lo

g uS

)

0.00

0.04

0.08

payoff share

arou

sal (

log

uS)

low equal high

0.00

0.04

0.08

payoff share

arou

sal (

log

uS)

low equal high

0.0

0.1

0.2

0.3

0.4

payoff share

arou

sal (

log

uS)

low equal high

0.00

0.02

0.04

payoff share

arou

sal (

log

uS)

low equal high

0.0

0.2

0.4

0.6

payoff sharear

ousa

l (lo

g uS

)

low equal high

0.0

0.2

0.4

payoff share

arou

sal (

log

uS)

low equal high

0.0

0.5

1.0

1.5

payoff sharear

ousa

l (lo

g uS

)

low equal high

0.0

0.5

1.0

1.5

own ideal payoff difference

arou

sal (

log

uS)

less ideal more

0.00

0.01

0.02

0.03

0.04

0.05

own choicean

ticip

ator

y ar

ousa

l (uS

)

more cooperative

less cooperative

0.00

0.01

0.02

0.03

0.04

0.05

other´s previous round choice

antic

ipat

ory

arou

sal (

uS)

more cooperative

less cooperative

Arousal and anticipatory arousal

I Arousal is high in fair results and withhigh payoff share, but also when own payoffis less than the ideal payoff

I Anticipatory arousal is high when thesubject makes less cooperative decisions

I Anticipatory arousal is high when the other’sprevious round decisions are morecooperative

I Gender effect: higher for malesI Habituation: decrease in time

Disgust and pos. affect are higher in fairresults

unfair fair

0.5

1.5

2.5

3.5

p=0.31

ange

r (lo

g uV

)

unfair fair

12

34

p=0.074

disg

ust (

log

uV)

unfair fair

0.5

1.0

1.5

2.0

2.5

3.0

p=0.078

pos.

affe

ct (

log

uV)

0.5

1.0

1.5

2.0

2.5

3.0

payoff sharean

ger

(log

uV)

low equal high

0.5

1.0

1.5

2.0

2.5

3.0

own ideal payoff diff.

ange

r (lo

g uV

)

less ideal more

0.5

1.0

1.5

2.0

2.5

3.0

other´s ideal payoff diff.

ange

r (lo

g uV

)less ideal more

AngerI Increases as payoff share increases

I i.e. the higher the relative payoff, themore there is anger

I Positive and negative affect (anger) arenot mutually exclusive: the corrugatorsupercilii muscle can measure both (Itoet al. 1998)

I Increases as the subject gets less payoffthan the ideal, and as the other gets morepayoff than the ideal

0.5

1.0

1.5

2.0

2.5

other´s ideal payoff diff.

disg

ust (

log

uV)

less ideal more

0.5

1.0

1.5

2.0

2.5

other´s ideal payoff diff.

posi

tive

affe

ct (

log

uV)

less ideal more

Disgust and positive affect

I Do not depend on payoff shareI There is possibly correlation between

disgust and positive affectI Disgust and positive affect are higher in

fair resultsI Disgust and positive affect are higher

when the other gets less payoff than theideal

Relationship to earlier studies

I Anticipatory arousal : similar results as inJoffily et al. (2011) who find that lesscooperative decisions elicit higheranticipatory arousal

I Disgust in unfair results: we find oppositeresults than Chapman et al. (Science, 2009)I They measure disgust, not positive affectI We do not deceive subjects and use fake

opponents as they do, and therefore ourexperiment is more reliable

ConclusionsI Fair behavior of the other player elicits arousalI Anticipatory arousal is related to own noncooperative

behaviorI Subjects get angry if they get less than their ideal

payoffs, and when the other gets more than their idealpayoffs

I These results imply that subjects are not entirelyself-regarding in cooperative situations

I Emotions should receive more interest in ORI The psychophysiological measurement method suits

well for studying emotions in group decision making

References

I Bechara, A., and Damasio, A.R. 2005. The somatic marker hypothesis: A neuraltheory of economic decision. Games and Economic Behavior 52(2)

I Benedek, M., and Kaernbach, C. 2010. Decomposition of Skin ConductanceData by Means of Nonnegative Deconvolution. Psychophysiology 47.

I Ben-Shakhar, G., Bornstein, G., Hopfensitz, A., and van Winden, F. 2007.Reciprocity and Emotions in Bargaining Using Physiological and Self-ReportMeasures. Journal of Economic Psychology 28(3)

I Bowles, S., and Gintis, H. 2011. A Cooperative Species. Human Reciprocity andits Evolution. Princeton, NJ: Princeton University Press.

I Cannon, P.R., Schnall, S., and White, M. 2011. Transgressions and Expressions:Affective Facial Muscle Activity Predicts Moral Judgments. Social Psychologicaland Personality Science 2(3)

I Chapman, H.A., Kim, D.A., Susskind, J.M., and Anderson, A.K. 2009. In BadTaste: Evidence for the Oral Origins of Moral Disgust. Science 323

I Huck, S., Muller, W., and Normann, H-T. 2001. Stackelberg Beats Cournot: OnCollusion and Efficiency in Experimental Markets. The Economic Journal 111

References

I Fehr, E. 2009. Measuring Social Preferences, in Glimcher, P.W., Camerer C.F.,Fehr, E., and Poldrack, R.A. (eds.), Neuroeconomics: Decision Making and theBrain, London, UK: Academic Press

I Fouraker, L. and Siegel, S. 1963. Bargaining Behavior, New York: McGraw-HillI Ito, T. A., Cacioppo, J. T., and Lang, P. J., 1998. Eliciting affect using the

International Affective Picture System: Bivariate evaluation and ambivalence.Personality and Social Psychology Bulletin 24.

I Joffily, M., Masclet, D., Noussair, C.N., Villeval, M-C. 2011. Emotions, Sanctionsand Cooperation. IZA Discussion Paper No. 5592

I Potters, J., and Suetens, S. 2013. Oligpoly Experiments in the CurrentMillennium. Journal of Economic Surveys 27(3)

I Sobel, J. 2005. Interdependent Preferences and Reciprocity. Journal ofEconomic Literature 43(2)