ABSTRACT
In an attempt to solve the problem, a single notion of salience has been advocated that supposedly gives agents a reason to choose a given strategy in both one-shot and iterated games. Only public salience is capable of providing an acceptable account of convention. Rationality is not enough, and neither is the traditional formulation of salience. The salient is normally defined as that which is unique and obvious. Salience changes neither payoffs nor expectations and hence does not give rational agents a reason to choose the salient. The functioning of private salience requires relatively small groups with a rather specific psychology. Salience provides the means for agents to make the appropriate reformulation. An asymmetry becomes full fledged salience when the probability judgements it yields allows agents to form strategies that increase their expected utility. In small group interactions, where the parties are not anonymous, private salience combined with a possible tendency to choose the salient facilitates coordination.
