Abstract
We believe that communication in multi-agent system has two major meanings. One of them is to transmit one agent's observed information to the other. The other meaning is to transmit what an agent is intending. In such communication, dynamic communication with a communication loop is required. Here we focus the latter, and aim to the emergence of the autonomous and decentralized arbitration through communication among some agents. The communication contents and representation of them are not prescribed and are acquired by learning using a reinforcement signal, which is given to the agent after its action. The reinforcement signal is not shared with the other agents. Since the agent often has to make a decision from the past communication signals, the architecture using recurrent type (Elman) neural network is proposed. The ability of this architecture was examined by two and four agent negotiation problems. A variety of negotiation strategies (individuality) to avoid the conflicts after their decisions emerged among them through the simple learning with the simple architecture.
