TY - JOUR
T1 - Prosocial punishment bots breed social punishment in human players
AU - Shen, Chen
AU - He, Zhixue
AU - Shi, Lei
AU - Wang, Zhen
AU - Tanimoto, Jun
N1 - Publisher Copyright:
© 2024 The Author(s) Published by the Royal Society. All rights reserved.
PY - 2024/3/13
Y1 - 2024/3/13
N2 - Prosocial punishment, an important factor to stabilize cooperation in social dilemma games, often faces challenges like second-order free-riders—who cooperate but avoid punishing to save costs—and antisocial punishers, who defect and retaliate against cooperators. Addressing these challenges, our study introduces prosocial punishment bots that consistently cooperate and punish free-riders. Our findings reveal that these bots significantly promote the emergence of prosocial punishment among normal players due to their ‘sticky effect’—an unwavering commitment to cooperation and punishment that magnetically attracts their opponents to emulate this strategy. Additionally, we observe that the prevalence of prosocial punishment is greatly enhanced when normal players exhibit a tendency to follow a ‘copying the majority’ strategy, or when bots are strategically placed in high-degree nodes within scale-free networks. Conversely, bots designed for defection or antisocial punishment diminish overall cooperation levels. This stark contrast underscores the critical role of strategic bot design in enhancing cooperative behaviours in human/AI interactions. Our findings open new avenues in evolutionary game theory, demonstrating the potential of human–machine collaboration in solving the conundrum of punishment.
AB - Prosocial punishment, an important factor to stabilize cooperation in social dilemma games, often faces challenges like second-order free-riders—who cooperate but avoid punishing to save costs—and antisocial punishers, who defect and retaliate against cooperators. Addressing these challenges, our study introduces prosocial punishment bots that consistently cooperate and punish free-riders. Our findings reveal that these bots significantly promote the emergence of prosocial punishment among normal players due to their ‘sticky effect’—an unwavering commitment to cooperation and punishment that magnetically attracts their opponents to emulate this strategy. Additionally, we observe that the prevalence of prosocial punishment is greatly enhanced when normal players exhibit a tendency to follow a ‘copying the majority’ strategy, or when bots are strategically placed in high-degree nodes within scale-free networks. Conversely, bots designed for defection or antisocial punishment diminish overall cooperation levels. This stark contrast underscores the critical role of strategic bot design in enhancing cooperative behaviours in human/AI interactions. Our findings open new avenues in evolutionary game theory, demonstrating the potential of human–machine collaboration in solving the conundrum of punishment.
KW - committed individuals
KW - costly punishment
KW - simple bots
UR - http://www.scopus.com/inward/record.url?scp=85187777306&partnerID=8YFLogxK
U2 - 10.1098/rsif.2024.0019
DO - 10.1098/rsif.2024.0019
M3 - 文章
C2 - 38471533
AN - SCOPUS:85187777306
SN - 1742-5689
VL - 21
JO - Journal of the Royal Society Interface
JF - Journal of the Royal Society Interface
IS - 212
M1 - 019
ER -