Title:On the effect of spatial structure on the evolution of cooperation

Abstract: The positive effect on cooperation of the spatial structure of a population is currently regarded as a well established principle in evolutionary game theory. Since Nowak and May's report of a promotion of cooperation in the Prisoner's Dilemma, when players are located on a regular lattice [Nature 359, 1992], many subsequent studies have supported this conclusion. Some recent results, however, seem to refute it. The present situation is that this question remains unclear, because important differences in details of the studied models prevent a thorough comparison. To contribute to this issue we have investigated systematically the effect of spatial structure, by means of a unified simulation framework that allows to consider in a comparable manner different social dilemmas, networks models and rules for the update of strategies. We have found that, in general, spatial structure promotes cooperation specially in coordination games, like Stag Hunt, and that important positive effects on Prisoner's Dilemma only occur when using a particular non-stochastic update rule. We show that the clustering is the key topological feature of spatial networks, as well as the crucial importance of the rules for the update of strategies. We explain all these effects in terms of the local densities of each strategy and the structure of equilibria of the game. In addition, the asymmetry between the effects on coordination and anti-coordination games arises as a fundamental feature of these evolutionary models, in the strong selection limit.