Mutual Adaptation Enhanced by Social Laws

We investigate the issue of multiagent learning with focus on the interactions among the learning agents. Each agent learns directly from the other agents that operate in the same environment. Each agent computes a function that expresses its satisfaction level after it performs an action, and after the other agents have expressed their degree of approval of the agent's behavior. Each agent adjusts its behavior based on an adaptation criterion applied to its satisfaction level. Our aim is that the agents learn to adapt their behaviors. In this work, we impose a social law on the agents' behavior while they learn to adapt to the other agents in the environment. Each agent has attached a level of cooperation that induces it to follow the rule to diierent degrees. We show that the level of mutual adaptation of the agents can increase when they follow a social law.