Evolving cooperation via communication in homogeneous multi-agent systems

The paper describes a simulation of evolving communicating agents. This model does not explicitly force agents to communicate by directly rewarding communication, creating a structured communication protocol, or making a distinction between sender and receiver. Instead, the agents are given limited sensory capabilities and the communicative behavior emerges as a form of cooperation among the agents. The auditory signals are utilized by the agents to maximize performance in an unpredictable environment. Three types of signals emerged as useful in the system-a call for foraging, a general purpose recruitment call, and an alarm call.