Consequentialist conditional cooperation in social dilemmas with imperfect information
This work addresses the challenge of fostering cooperation in multi-agent systems where actions are partially unobserved, which is incremental as it builds on existing conditional cooperation methods by focusing on outcomes rather than intentions.
The paper tackles the problem of designing cooperative agents in social dilemmas with imperfect information by proposing strategies that condition behavior solely on past rewards, called consequentialist conditional cooperation. They demonstrate the effectiveness of these strategies using deep reinforcement learning, showing they work in complex games beyond simple matrix games, though they also identify limitations in relying only on consequences.
Social dilemmas, where mutual cooperation can lead to high payoffs but participants face incentives to cheat, are ubiquitous in multi-agent interaction. We wish to construct agents that cooperate with pure cooperators, avoid exploitation by pure defectors, and incentivize cooperation from the rest. However, often the actions taken by a partner are (partially) unobserved or the consequences of individual actions are hard to predict. We show that in a large class of games good strategies can be constructed by conditioning one's behavior solely on outcomes (ie. one's past rewards). We call this consequentialist conditional cooperation. We show how to construct such strategies using deep reinforcement learning techniques and demonstrate, both analytically and experimentally, that they are effective in social dilemmas beyond simple matrix games. We also show the limitations of relying purely on consequences and discuss the need for understanding both the consequences of and the intentions behind an action.