My dissertation examines how emotions induced by prior outcomes and choices affect preferences, and the ways in which individuals anticipate these behavioral affects and adjust their choices ex-ante. The first chapter explores a contradiction in the literature documenting how prior losses affect risk attitudes : some studies find that people become more risk seeking after a loss, whereas others find that they become more risk averse. I show that these seemingly inconsistent findings can be explained by individuals' differential responses to realized versus paper losses. Following a realized loss, individuals are sensitized to further losses and avoid risk; if the loss has not been realized -- a paper loss -- individuals take on greater risk. The second chapter studies the effectiveness of prosocial incentives, where effort is tied directly to charitable contributions. I show that people work harder for charity than for themselves, but only when incentive stakes are low. When stakes are raised, prosocial incentives are no longer superior, suggesting that such incentive schemes should only be used when funds are limited. Additionally, individuals correctly anticipate these effects, choosing to work for charity at low incentives and for themselves at high incentives. These results are consistent with a model of warm-glow giving. The third chapter, co-authored with Uri Gneezy, examines whether individuals use anger strategically in interactions. We first show that in some environments anger helps performance, while in others anger hurts. We then show that individuals anticipate these effects and strategically use the option to anger their opponents in competitions. This finding suggests people understand the effects of emotions on behavior and exploit. The final chapter, co-authored with Uri Gneezy and Kristof Madarasz, presents theory and experiments where people's prosocial attitudes fluctuate over time following norm violations. We show that people who made an immoral choice were more likely to donate to charity, and that immoral behavior increases when people are informed of a future donation opportunity. We propose a model where guilt induced by past immoral actions increases prosocial behavior, and discuss its importance in charitable giving and in the identification of norms in choice behavior through time- inconsistency