This award funds research that uses decision-making experiments to investigate the importance of strategic complexity in repeated games. The project will test hypotheses about the possible cognitive costs of strategic complexity and how these costs affect decisions to cooperate in a repeated game in which the efficient outcome requires cooperation but each individual involved may benefit from selfish behavior. Previous literature in game theory indicates that cooperation in repeated games may be limited by the complexity of the strategic environment, as cooperative strategies are generally associated with greater cognitive costs and less certain outcomes than selfish strategies. The experiments funded by this award observe the cooperation rate in a standard indefinitely repeated prisoner's dilemma and compare it with cooperation in a similar but more complex strategic environment. The design allows an investigation of whether the impact on cooperation depends on the source of complexity in the strategic environment. Broader impacts include contributing to graduate education and the possible interdisciplinary impact of the research, which crosses boundaries between game theory, decision science, and behavioral economics.
The research studies strategic decision making by human subjects in a laboratory experiment. The experiment is designed to test whether the propensity to cooperate in repeated interactions is sensitive to the complexity of the strategic environment. An important issue in repeated games is that standard theory does not provide a unique predicted outcome but many possible outcomes, some cooperative and some in which players behave selfishly. Strategic complexity is a criterion which may provide a more refined prediction, but it has not yet been tested as such in a laboratory experiment. Analysis of the data collected indicates that subject behavior is sensitive to the complexity of strategic implementation, as hypothesized. Cooperation in the baseline treatment is consistent with results found in previous literature, but we find that subjects are less likely to adopt cooperative strategies in treatments which increase the complexity of such strategies. The observed effect has not been previously documented in the experimental economics literature, but it is consistent with results found in the theoretical literature on strategic complexity in repeated games. The lower cooperation rate observed in the more complex environment represents evidence that the level of cooperation in repeated interactions is sensitive to constraints on subjects' ability to use complex strategies. This result has important implications for the study of cooperative behavior in market competition, politics, and the social sciences, where strategic complexity is not formally considered to be a binding constraint on behavior.