The Evolution of Signals
Daniel Cloud · Columbia University Press eBooks · 2014
This chapter examines the Brian Skyrms' signaling convention. Skyrms realized that a version of Lewis' simple model of a signaling system could be formalized and turned into an evolutionary game. Here there is no requirement that players “know” anything or have expectations of any kind about the other players' behavior. If successful coordination leads to faster replication, an arbitrary signaling convention will become established in the population anyway. Or the players may blindly imitate other players that have been successful in coordinating with their own partners, leading to a snowballing of imitations of imitations that will take the whole population to one of the two possible conventions. Indeed, as Skyrms argues, simple signaling systems of this kind are ubiquitous in nature, as seen in the way bacteria communicate with other bacteria. The concept of “convention” at work in these cases is considerably pared down even from Lewis' version, which itself pared down what W. V. Quine thought was required.