The Dynamics of Information Transfer

Abstract

Philosophers and scientists have long debated how communication can arise in circumstances in which it is not already present. This dissertation uses the techniques of evolutionary game theory to address this puzzle. Following David Lewis (1969), communication is envisioned as occurring between players in a game. One player, who has private knowledge about the state of the world, sends a signal to the second player, who then performs an action. Lewis assumed that the players in this game share a perfect common interest in communicating; i.e., he assumed that when they game ends they both receive identical payoffs. In this dissertation the assumption of common interest is abandoned. Three case studies presented here demonstrate that information transfer robustly arises even when the interests of the sender and receiver diverge. But there's a twist. These same three case studies also indicate that in situations of divergent interests, the communication schemes that emerge only transmit partial information about the state of the world. In other words, although some information is conveyed by the signal, it does not perfectly identify the state. This form of partial communication is sometimes overlooked by philosophers, and is not a phenomenon predicted by Lewis's original framework. Since many real-life interactions are not best modeled by perfect common interest games, these three case studies jointly suggest that partial information transfer may be ubiquitous in actual social systems.

These three studies also reveal a limitation of static equilibrium analysis, which is a dominant methodology in game theory. The systems investigated here to do not converge to the equilibria that this style of analysis traditionally identifies as the rational choices or most likely outcomes of play. In fact, in one case the system does not convert to an equilibrium at all. This suggests that a single-minded focus on Nash equilibria (or worse, a refinement thereof) may yield a misleading picture of the prospects for the emergence of communication.