Preferential selection based on adaptive attractiveness induce by reinforcement learning promotes cooperation

The preferential selection of role models for strategy imitation has been found to have a significant impact on the cooperation evolution. In this paper, we introduce a preferential selection mechanism based on reinforcement learning into the spatial prisoner's dilemma game and explore how the preferential selection of role model affects cooperation. The individuals choose role models based on the adaptive attractiveness of their neighbors. The attractiveness is dynamically updated during the evolution process and is determined by the BM reinforcement learning rule. The results demonstrate that incorporating the preferential selection mechanism effectively enhances the cooperation. Moreover, an optimal individual's sensitivity to stimulus β can result in maximizing the level of cooperation. Furthermore, we observe no substantial variations in the level of cooperation across different initial attractiveness of individuals. Besides, we have explored the effect of noise intensity K on cooperation, and the results reveal that K manifests different effects on the cooperation in the spatial prisoner's dilemma games with and without preferential selection mechanisms.