Rubinstein (1988, 2003) and Leland (1994, 1998, 2001, 2002) have shown that choices based on similarity judgments will account for the vast majority of observed violations of expected and discounted utility. In this paper, I show that such judgments also explain which equilibria will be selected...