Abstract
Cooperation in social dilemmas can be sustained if individuals are effectively rewarded or punished from peers within the group. However, as group size increases, we inevitably face localization, in which a global group is divided into several localized groups. In such societies, members can reward and punish only neighbors within the same localized group, while cooperation for social dilemmas should be solved through global group involvement. In this situation, the global group and the local group are not always equal in terms of welfare, and situations can arise in which cooperation is beneficial for the global group but not for the local group. We predict that in such a locally inefficient situation, peer reward and punishment cannot function to sustain global cooperation. We conducted an experiment in which 16 group members played a public goods game incorporating peer reward and punishment. We manipulated the range of peer reward and punishment (only local members/all global members) and payoff structure (locally efficient/locally inefficient). We found that high cooperation was not achieved and that peer reward and punishment did not function when, and only when, the group was divided into localized groups and the payoff structure was locally inefficient.
Highlights
Cooperation in social dilemmas can be sustained if individuals are effectively rewarded or punished from peers within the group
In a public goods game (PGG), members of a group decide to contribute to a common project, and the total contributions are multiplied and shared among group members
The average PGG contribution, average amounts of reward and punishment, and average profit of players were calculated for each period (Fig. 2)
Summary
The average PGG contribution, average amounts of reward and punishment, and average profit of players were calculated for each period (Fig. 2). Reward and punishment might function to facilitate cooperation in the PGG This tendency was weaker for reward in the Local0.1 than the three other conditions. To test this statistical significance, the difference between reward for a cooperator and that for a non-cooperator was calculated and analyzed using the Mann–Whitney U-test with Bonferroni’s correction. We examined how players changed their cooperation level in PGG after receiving reward and punishment in the previous period—that is, the effectiveness of reward and punishment in facilitating cooperation—using multilevel regression analysis. The results indicated that both reward and punishment functioned to change others’ behavior to cooperation; this function was weaker in Local0.1 than in the other three conditions (see SI Appendix, section 2 for details)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
