Incremental reinforcement learning in multiagent systems