John Miller

Paper #: 89-003

The coevolution of strategies in the repeated Prisoner's Dilemma game is studied under both perfect and imperfect information conditions. Players are required to submit strategies in the form of finite automata in order to participate in the game. By applying recent developments from the study of genetic algorithms in computer science, an explicit environment and selection process are derived. Using this framework, the effect of imperfect information on the development of cooperation and strategic choice is studied. The strategies that emerge are classified and their performances are analyzed. The results of the analysis indicate that information conditions lead to significant differences arising between the evolving strategic environments. Furthermore, they suggest that the general methodology may have much wider applicability.