Protocols for Cooperation Cultural Diversity of Strategies for the Alternating Prisoner's Dilemma

The Prisoner's Dilemma has long been the paradigm for modeling cooperation in a competitive environment. The classical version, in which both agents simultaneously choose to cooperate or to defect, is unnecessarily artificial. The stochastic Alternating Prisoner's Dilemma in which choices are made sequentially and mistakes are possible is more faithful to examples in the real world. Intelligent k-ply agents, who base their action on k previous outcomes, can achieve mutual cooperation in the stochastic Alternating Prisoner's Dilemma. Among the 4-ply agents that remember the previous two choices of the agent and of his opponent, several mutually evolutionarily stable clans will emerge in an evolutionary model based on Darwinian natural selection. In particular, the stochastic 4-ply strategy dubbed Tough Love maintains a stable cooperative environment by punishing misbehavior while quickly forgiving the repentant.