Title: Emergence of cooperation in a mixed population: the strength of irrationality

Abstract: How cooperation emerges in our society remains largely unknown. One reason is that most previous studies strictly follow the assumption of \emph{Homo economicus} that individuals are rational and maximizing the earnings is the only incentive of their actions. We human beings, however, are complex and our behaviors are multiple-motive driven. In this study, we investigate the evolution of cooperation within a population on the $2d$ square lattice with mixed rational and irrational individuals, or specifically, players respectively with the Fermi and Tit-for-tat updating rules. While the former rule is to imitate the strategies of those neighbors who are doing well, the latter is like emotion-driven behaviors where the players repeat what their neighbors did to them. In a structural mixing implementation, where the updating rule is fixed for each individual, we find that a moderate fraction of irrational behaviors unexpectedly boosts the overall cooperation level of the population. The boost is even more pronounced in the probabilistic mixing, where each individual randomly chooses the two rules at each step, and full cooperation is seen in a wide range. These findings also hold in complex structured populations, such as Erd\H{o}s-R\'enyi random networks and Barab\'asi-Albert scale-free networks. Our mean-field treatment reveals that the cooperation prevalence within the rational players linearly increases with the introduced irrational fraction and explains the non-monotonic dependence in the structural mixing. Our study indicates that the role of irrationality is highly non-trivial and is indispensable to understand the emergence of cooperation in realistic contexts.