In this chapter we propose a new, dual-process model of labor supply, which incorporates both cognitive and affective aspects of decision-making. Consistent with evidence from neuroscience, the worker may experience conflicting cognitive and affective motivations during the workday. In particular, the affective system values effort more highly as long the worker's performance is below a personal goal, or income target, and becomes increasingly aroused as the goal approaches. As a result, affect can distort effort decisions relative to a fully cognitive benchmark, in a way that is consistent with evidence on loss aversion, and with the so-called goal-gradient effect, a tendency for animals and humans to increase effort as a goal approaches. In contrast to a standard model of labor supply, our model can predict a goal gradient, and predicts that workers may actually lower total daily effort in response to a temporary increase in the wage. Also, within-day windfall gains may have an impact on a worker’s effort profile over the workday. The second part of the chapter tests this latter prediction using data from two bicycle messenger firms. At both firms, a windfall gain in the morning has the predicted impact. A lucky messenger works harder than other messengers over the first part of the afternoon, and the difference is increasing, consistent with a goal gradient. Later in the afternoon, a lucky messenger works significantly less hard than the others, consistent with having surpassed a personal earnings goal earlier in the day and having less affective motivation.