Behavioral diversity, choices and noise in the iterated prisoner's dilemma

摘要: Real-world dilemmas rarely involve just two choices and perfect interactions without mistakes. In the iterated prisoner's dilemma (IPD) game, intermediate or mistakes (noise) have been introduced to extend its realism. This paper studies IPD game with both noise multiple levels of cooperation (intermediate choices) in a coevolutionary environment, where players can learn adapt their strategies through an evolutionary algorithm. The impact on evolution is first examined. It shown that models presented this are robust against low (when occur probability). That is, little cooperation. On other hand, high probability) creates misunderstandings discourages However, more learning setting also depends behavioral diversity. further investigates issue diversity coevolution for noise. difficult achieve if model used games higher population resistant strategy representations significant outcomes because different diversities they generate. results show importance learning.