Success-rate-driven strategy updating enhances cooperation in social dilemmas

Abstract

Abstract Strategy updates play an important role in the evolution of cooperation. In traditional imitation rules, it is assumed that strategic learning is driven by payoff bias. In this study, a modified Fermi rule was proposed, in which payoff and success-rate are synthetically considered to determine players’ strategy updates. It is assumed that individuals have an immediate memory length M, and that success can be measured by using an identical aspiration level a as a benchmark. Success-rate is defined in the frequency that an individual’s payoff equals or exceeds the benchmark a within the memory length M. Numerical simulations conducted by considering the two factors in the strategy update show that cooperation can be facilitated compared to the traditional version. Using the identical aspiration level a as benchmark, the results show that there is a clear phase shift phenomenon in enhancing cooperation. This implies that different b will have an appropriate aspiration interval to achieve optimal cooperation. Moreover, only a moderate immediate memory length can facilitate cooperation. The proposed mechanism is highly robust and can provide valuable insights into the emergence and maintenance of cooperation.