Reinforcement learning explains various conditional cooperation

作者：

Highlights：

• We consider the repeated pairwise prisoner's dilemma game among two groups of agents.

• We found that the mixed strategy-update rule (Q-learning-based update rule and Fermi-function-based update rule) can evaluate cooperation.

• If the proportion of AI is moderate, cooperators among the whole population exhibit conditional behavior and moody conditional behavior.

摘要

•We consider the repeated pairwise prisoner's dilemma game among two groups of agents.•We found that the mixed strategy-update rule (Q-learning-based update rule and Fermi-function-based update rule) can evaluate cooperation.•If the proportion of AI is moderate, cooperators among the whole population exhibit conditional behavior and moody conditional behavior.

论文关键词：Evolutionary games,Q-learning,Conditional cooperation

论文评审过程：Received 19 October 2021, Revised 3 March 2022, Accepted 12 April 2022, Available online 27 April 2022, Version of Record 27 April 2022.

论文官网地址：https://doi.org/10.1016/j.amc.2022.127182