Promotion of cooperation by payoff noise in a 22 game

A series of numerical simulations of a 22 symmetric game on a network examined whether payoff matrix noise promotes cooperation, as reported initially by Perc [New J. Phys. 8, 22 (2006)]. Agents have no memory (they offer cooperation, C, or defection, D). We assume that the network is time invariable. The effect of payoff matrix noise (PMN) is measured by a simulated payoff difference between a normal network game and a network game with PMN. The effect of PMN appears only when a local strategy adaptation is implemented (for example, a network game with imitation dynamics). The influence of PMN becomes more significant with a larger stochastic deviation, and less significant in a larger degree network. One reason for PMN’s effectiveness is the local strategy adaptation mechanism, which helps both the preservation and fixation of C agents, and not that the payoff matrix noise makes a dilemma game into a Trivial (dilemma-free) game.