Past-performance-driven strategy updating promote cooperation in the spatial prisoner's dilemma game

Strategy update rules play an important role in repeated Prisoner's Dilemma games. This work proposes a modified strategy update rule based on the traditional Fermi function, in which individual past performance is taken into account in strategy update. Then, the consistency aspiration α serves as a benchmark to measure an individual's past performance, and the past performance score is dynamically adjusted during the evolution process according to the BM reinforcement learning rules. The computational results indicate that the proposed modified strategy update rules can significantly improve system cooperation t than the traditional version, and the network reciprocity effect is enhanced as the result of past performance is coupled into the strategy update rule. Moreover, different temptation to defection b exist a corresponding aspiration α result in maximizing system cooperation. Furthermore, an optimal sensitivity level β can also result in a maximizing system cooperation. As a whole, for α − β phase diagram, different a will correspond to an optimal value β that allows the system to achieve the maximum cooperation. Finally, the proposed mechanism is robust. Hopefully this can help to inspire further research on how to deal with social dilemmas.