著者
井寄 幸平 鳩野 逸生 小田 宗兵衛 上田 完次
出版者
一般社団法人 システム制御情報学会
雑誌
システム制御情報学会論文誌 (ISSN:13425668)
巻号頁・発行日
vol.16, no.9, pp.468-475, 2003-09-15

This paper describes how people establish or fail to establish long-run cooperation in the Prisoner's Dilemma Network (PDN), where subjects are allowed to nominate a subject with whom they want to play the Prisoner's Dilemma (PD) game at the beginning of each round. We have done a series of experiments with undergraduates and computer simulations of the PDN games. In the experiments most subjects either continued to play the PD game cooperatively with the same partner or never played the PD game cooperatively in the long run, and those who were more cooperative earned more. Since the simulations reproduced the results of the experiments, we can guess the subjects' strategies, which were not as apparent and controllable as the programs of agents.