EVOLUTIONARY ANALYSIS OF PRISONER'S DILEMMA GAMES BASED ON MIXED RANDOM-CONFORMITY SELECTING MODEL

Inspired by the conformity phenomenon in human society, we develop a mixed neighbor selecting model adopting random-conformity rule to explore the evolutionary weak prisoner's dilemma game. The neighbor selection rule of nodes is adjusted based on their fitness and collective influence. Under the degree-normalized payoff framework, the findings derived from Monte Carlo simulations reveal that this mixed selecting model can contribute to an impressive improvement in the Barabasi-Albert network's cooperation. In addition, experimental data obtained by investigating the game-learning skeleton indicate that, in this mixed random-conformity selecting model, normalized collective influence at moderate depth length enables influential nodes to maintain a cooperative strategy for an extended period of time. This can promote the emergence of cooperative strategies at low-degree nodes by facilitating the formation of stable cooperation-clusters centered on high-degree nodes. In addition, the normalized collective influence at excessive depth length increases the likelihood that influential nodes become defectors, thereby inhibiting the growth of cooperation-clusters and limiting cooperation.