An analysis of network reciprocity in Prisoner's Dilemma games using Full Factorial Designs of Experiment.

Despite hundreds of studies on the Prisoner's Dilemma (PD) game, understanding about network reciprocity remains a unsolved puzzle. Thus, we performed a series of Full Factorial Design of Experiments (FFDOE) to evaluate what dominates emerging cooperation in the PD game on various networks. The results qualitatively reveal the influence of each factor and show that some combinations of factors have complicated interactions. Remarkably, the choice of strategy update rule or update dynamics is much more important than the type of network imposed or, at least, the factorial effect of the average degree of the network reported by Nowak (Science 314, 5805, 1560-1563, 2006) and Ohtsuki et al. (Nature 441, 502-505, 2006). Furthermore, the decision of which PD game type to investigate (whether all PD games, PD-Chicken boundary games or Donor & Recipient games) is important for discussing network reciprocity.