Evolution of cooperation in a one-shot Prisoner's Dilemma based on recognition of trustworthy and untrustworthy agents

Abstract

This article explores the conditions under which agents will cooperate in one-shot two-player Prisoner's Dilemma games if they are able to withdraw from playing the game and can learn to recognize the trustworthiness of their opponents. When the agents display a number of symbols and they learn which symbols are important to estimate the trustworthiness of others, agents will evolve who cooperate in games in line with experimental observations. These results are robust to significant levels of mutations and errors made by the players.