Abstract

In his Convention David Lewis defined conventions as behavioural regularities instantiating proper coordination equilibria made salient by precedent and operational by this being common knowledge. While later proponents of game theoretical approaches in the study of convention have agreed on dropping Lewis’ eccentric ‘coordination’ requirement as well as that of common knowledge, they are confused as to whether conventions should be regarded as proper thereby precluding mixed equilibria. In this paper I argue that mixed equilibria may not be conventions, but also suggest that the reason for this reveals that though common knowledge is not necessary for a convention to operate, it may be utilized to identify the conventional aspect of a given practice.