Evolution of self-organized systems

In this paper I ask questions about the evolution of self-organized activity cycles that are found in some ant colonies. I use a computer model that generates periodic activity patterns in interacting subunits and explore the parameters of this model using a genetic algorithm in which selecting on one aspect of the system produces the distinctive self-organized pattern. The general point that I explore, using the example of activity cycles, is that the observation of a self-organized pattern does not mean that the pattern is an adaptation. Self-organized patterns can represent nonadaptive correlated responses to selection, exaptations or even selectively disadvantageous traits. Evolution of self-organized patterns requires genetic feedback between the self-organized output and the subunits that produce the pattern. Without this necessary feedback, a self-organized system does not evolve.