Memory-based prisoner's dilemma game with conditional selection on networks

We investigated the memory-based prisoner's dilemma game with conditional selection on networks. The proposed selection takes the historical information into account, which helps evaluate the recent performance in the history and select neighbors with strong attractiveness. Those neighbors who get more payoffs than average payoff in the memory length are considered in potential growth way. The simulation results shows that memory length M makes a dual impact on the spatial interaction. Defection is benefit from small M while cooperation is promoted by large M. Cooperator can resist the defector's invasion with the increase of memory length M. Discussions for the average payoff and strategy distribution in the spatial game show the effect of the proposed mechanism with conditional selection. The findings may be helpful in understanding cooperative behavior in natural and social systems consisting of conditional selection based on the recent performance in the history.