How not to murder your neighbor: Using synthetic behavioral ecology to study aggressive signalling

The scientific account of intraspecific aggressive signaling is incomplete. In part, this is because it is difficult to identify the consequences (for the nature of signaling systems) of the different ecological contexts in which signaling has developed. The goal of our investigation is to complement the ongoing work in the biological sciences on this issue. Using a theoretical framework that we call synthetic behavioral ecology, we perform a series of experiments involving populations of simulated animals (animats) that, in their simulated world, must compete for food. When animats pick up reliable information about the "strength " of other animats within sensory range, a coherent form of collective behavior develops, which we call minimal territoriality. We then introduce the signaling of aggressive intentions. Each individual has a bluffing strategy that is determined by a form of artificial evolution in which there is no explicit fitness function. By varying, in energy terms, the cost of producing aggressive signals, and by analyzing the population dynamics at different costs of signaling, we are able to provide evidence that the handicap principle (according to which higher costs enforce honesty) can apply in multiagent ecologies. We then suggest how variations in the cost of signaling affect the territorial behavior. The article ends with a discussion of the current model and identification of some directions for future research. Key Words: aggression; animal signaling; behavioral ecology, communication; handicap principle; simulation of adaptive behavior; territorialit