Abstract # 16:

Scheduled for Thursday, June 21, 2012 02:45 PM-03:00 PM: Session 4 (Camellia ) Oral Presentation


S. A. Price1, M. J. Beran1, B. J. Wilson2 and S. F. Brosnan1
1Georgia State University, Language Research Center & Department of Psychology, Atlanta, GA, USA, 2Economic Science Institute, Chapman University, Orange, CA , USA
     We investigated how capuchins (n = 8) and humans (n = 52) made decisions in an anti-coordination game, the Hawk-Dove model of conflict over a shared resource. In a one-shot game, there are two (uncoordinated) Nash equilibria (Hawk-Dove and Dove-Hawk). When played repeatedly, this game may be more challenging to solve than a coordination game with a Pareto dominant Nash equilibrium. Pairs were tested using a common exchange task, in which both subjects could choose which of two tokens (one representing Hawk, the other representing Dove) to return to the experimenter. Thus, both the individual’s and partner’s choices determined the payoff outcome for both participants. Following nearly equivalent procedures, capuchins received ten 40-trial sessions, while humans received only a single session. Across sessions, none of the capuchin pairs’ choices deviated from chance, although one pair alternated playing the two Nash equilibria by the final session (chi-squared=4.95, df=1, p<0.05). Eleven out of 26 human pairs deviated from chance; nine pairs alternated between the two Nash equilibria (all chi-squared >14.83, df=1, all p<0.001), one pair played the mutually destructive Hawk-Hawk (p< 0.001, Fisher’s exact test) and one repeatedly played a single Nash equilibrium, Hawk-Dove (p<0.05, Fisher’s exact test). Thus, humans outperformed the capuchins overall, though the humans frequently did not settle on repeated Nash equilibria play. Such solutions may be challenging to find without explicit instruction.