Promotion of Cooperation in a Co-Evolutionary Pragmatic Agent Multigame Environment

Abstract

The promotion of cooperation in a co-evolutionary environment where pragmatic agents participate in a multigame setting that contains Prisoner's Dilemma (PD) and SnowDrift (SD) games is investigated. The pragmatic agent conserves its current perspective when successful; otherwise adopts the opposite perspective. Unlike traditional models, this study introduces a setup in which perception and strategy spaces co-evolve in terms of iterative game payoffs. The players are situated in a 2-D square lattice environment and synchronously update their perceptions and strategies after interacting with their immediate neighbors. The ratios of perceptions and strategy are randomly set based on parameters alpha and /3, indicating initial SD and cooperation (C) percentages, respectively. By the end of various simulations, the system's convergent and stable behavior is shown by means, standard deviations, and confidence intervals. The results show that larger (alpha >= 0.5) initial populations of SD agents promote greater cooperation and lead to a dominance of cooperative strategies even for smaller initial C strategies (/3 similar to 0.2). Conversely, when the environment is initially dominated by PD perspectives and defect (D) strategies (alpha = 0.1, /3 = 0.2), it leads to lower levels of cooperation. Depending on the initial ratios of PD/SD and D/C players, cooperative player clusters are not only formed but are also persistent parts of the environment. Finally, we observed that co-evolving PD/SD and D/C environments coupled with pragmatic players lead to a controllable promotion of cooperation even against small initial SD player ratios.