On Evolving Fixed Pattern Strategies for Iterated Prisoner's Dilemma

Jang, D., Whigham, P. and Dick, G.

    This paper describes the social evolution of an environment where all individuals are repeating patterns of behaviour. The paper follows Axelrod's work [1] of computer simulations of Iterated Prisoner's Dilemma (IPD), which is widely regarded as a standard model for the evolution of cooperation. Previous studies by Axelrod [2], Hirshleifer and Coll [3], Lindgren [4], Fogel [5], Darwen and Yao [6] focused on strategies that are history dependent. In other words, these strategies use the outcome of the opponent's past game history in making a decision on a given move. This includes the most wellknown strategy, tit-for-tat. The way strategies are encoded in the computer program reflects the model's assumption concerning individual decision-making. In this paper, we study environments where all players are simply repeating patterns of behaviour without using past game history. In doing so, a genetic algorithm is used to evolve such strategies in a coevolution environment. Simulations indicate that such an environment is harmful to the evolution of cooperation
Cite as: Jang, D., Whigham, P. and Dick, G. (2004). On Evolving Fixed Pattern Strategies for Iterated Prisoner's Dilemma. In Proc. Twenty-Seventh Australasian Computer Science Conference (ACSC2004), Dunedin, New Zealand. CRPIT, 26. Estivill-Castro, V., Ed. ACS. 241-247.
pdf (from crpit.com) pdf (local if available) BibTeX EndNote GS