2014 年 12 巻 2 号 p. 181-189
Human societies are organized around cooperative and altruistic interactions. Natural selection, however, favors selfish and strong individuals who maximize their own resources at the expense of others. Although many previous studies with average payoffs have developed mechanisms for resolving the cooperation dilemma, they have a severe problem. Estimating the average values requires sufficient knowledge of the payoffs for all players in all public goods games (PGGs), which is difficult to achieve in practice. People make estimates every day based on insufficient knowledge. The transition probabilities ought to be therefore calculated based on known payoffs rather than on the average. Through this individual learning, we show that pool-punishers can overcome free-riders to establish stable cooperation within a society without help from non-participants or players using other strategies, even with a small multiplier (r) of public goods. This scenario requires the punishment that the free-riders receive to be greater than the cost to the pool-punishers. We also demonstrate that smaller population sizes and higher participation rates engender greater fixation probabilities for cooperation.