People simulate scenarios of what-might-have-been and what-might-be as they make sense of everyday experiences (Kahneman and Miller 1986; Kahneman and Tversky 1982b; Taylor and Schneider 1989). Researchers have studied these tendencies to simulate counterfactual and prefactual scenarios, asking why some forms of simulation are more likely than others (Kahneman 1995). One answer to this question is that forms of simulation serve distinctive functions, such as learning from experience (Roese and Olson 1995a). A central argument has been that upward-directed simulations (how an outcome could have gone better) foster learning more than downwarddirected simulations (how it could have gone worse). In the first decade of research on scenario simulation, evidence from several studies converged in support of the proposition that upward and not downward simulations function in learning (see Roese and Olson 1995a).