Consider a two-player discounted repeated game in which each player optimizes with respect to prior beliefs about his opponent's repeated game strategy. One would like to argue that if beliefs are cautious then players will learn as the game unfolds to predict the continuation path of play. If...