Evolution of Conditional Cooperation in Prisoner's Dilemma

Conditional cooperation has been a common explanation for the observed cooperation, and its decline in social dilemma experiments. Numerous studies showed that most of the experimental subjects can be categorized into three types: conditional cooperators, self-maximizers and hump-shaped (triangle) cooperators. In this study, I investigate conditional strategy types and their role on the emergence of cooperation and their evolutionary success. For this purpose, I use an extension of the Iterated Prisoner's Dilemma Game. The agents are characterized by their initial move and their conditional responses to each level of cooperation. By using simulations, I estimate the likelihood of cooperation for different probability of continuations.I show that, when the continuation probability is sufficiently large, high levels cooperation is achieved. In this case, the most successful strategies are those who employ an all-or-none type of conditional cooperation, followed by perfect conditional cooperators. In intermediate levels of continuation probabilities, however, hump-shaped contributor types are the ones that are most likely to thrive, followed by imperfect conditional cooperators. Those agents cooperate in a medium level of cooperation within themselves and each other. The results explain the existence of hump-shaped type of cooperators with a purely payoff-based reasoning, as opposed to previous attempts to explain this strategy with psychological mechanisms.