Length of information-based bidirectional choice in spatial prisoner’s dilemma

We present a computational model of bidirectional choice based on the length of information of previous payoff to study the cooperative dynamics of iterated prisoner’s dilemma game on directed weighted lattices. For paired two individuals in network, they can attend game under the condition that they both successfully choose the other party with the intensity of unidirectional interaction, and the changing of the intensity of unidirectional interaction according to the relation between individual’s current payoff and its previous payoff which are both gained from the other party. The simulation results show that this evolutionary rule can availably remove the stumbling blocks of cooperators for existence and hugely boost the cooperation in population. Interestingly, for a smaller or moderate cost-to-benefit ratio (0 < r < 0.235), the cooperation level ascends monotonically as the length of information increases; however, for a bigger cost-to-benefit ratio (0.235 ≤ r < 0.249), there are some moderate values of length of information, resulting in the status of full cooperation. Surprisingly, we find that there is an optimal length of information which not just leads to the ultimately full cooperation in population but benefits the learning of cooperation strategy for individual, regardless of the value of cost-to-benefit ratio (0 < r < 0.249). Our results may contribute to the understanding and the analysis of cooperative dynamics of bidirectional choice in network.