We consider a fully cooperative multi-agent system where agents cooperate to
maximize a system's utility in a partial-observable environment. We propose
that multi-agent systems must have the ability to (1) communicate and
understand the inter-plays between agents and (2) correctly dis