Dynamic noise from action errors enhances network reciprocity in the prisoner's dilemma game

Inspired by the fact that people make mistakes in a transient, fluctuating or chaotic environment, we establish a spatial prisoner's dilemma model where an agent commits action errors proportionally varying with the increasing/decreasing rate of the global cooperation fraction. A series of numerical simulations reveal that the cooperation level is enhanced in games in which the stag hunt (SH)-type dilemma is dominant; however, it is slightly diminished in games in which the chicken-type dilemma is dominant, compared with the standard network reciprocity model. Intensive analysis reveals that the noise created by the action error contribute to the spatial expansion of a cooperators' cluster, because a dilemma that is less chicken-type and more SH-type makes it disadvantageous for defectors to neighbor cooperators. Our finding, that errors in behavior in a chaotic environment contribute to the evolution of cooperation, might aim to explain the problem of how network reciprocity works.