Tag
This paper examines why evolutionarily stable cooperative outcomes in multi-agent systems may not be reachable through decentralized reinforcement learning algorithms, demonstrating distinct properties between stability and learning accessibility.