Dynamic asymmetric reflective decision-making supports the evolution of cooperation

Abstract

Up to now, most previous studies have assumed that agents employed symmetric decision-updating mechanisms to investigate how cooperation evolved in selfish populations. However, dynamic and asymmetric decision-making update is overlooked. In this paper, we try our best to construct a reputation-driven dynamic asymmetric reflective decision-making model to fill in the blanks. Initially, each agent is randomly assigned a reputation value within the range [1,100], which will increase or decrease if it makes cooperative or defective decision in the process of evolution. Then, we distinguish agents as standouts or laymen by the pivotal reputation threshold $R_c$. The standouts with high-reputation are cautious in updating their decisions and therefore may think more times. In sharp contrast, the laymen with low-reputation, appear too reckless and just only think once in the similar process. Simulation results show that this simple model can support the evolution of cooperation, especially expanding the threshold for cooperation annihilation. Through qualitative micro-analysis and quantitative statistical analysis, we find that standouts with high reputations play an irreplaceable role in facilitating cooperation. The robustness test further confirms that the promoting effect of this mechanism on cooperation is independent of network structures, decision update patterns, and social dilemma types. We hope that our research can deepen the understanding and cognition of the evolution of cooperation.