According to psychologists and neuroscientists, a key source of intrinsic motivation is learning. An economic model of this is presented. Learning may make work less onerous, or the employee may value it in and of itself. Multitasking generates learning: performing one task increases productivity on related tasks. Intrinsic motivation generates a new multitask incentive problem if the rate of learning varies across tasks. With no incentive pay, employee autonomy complements learning because expected output increases as the employee uses his or her knowledge to enhance learning. The second part of the paper adds a simple incentive. Incentive pay does not "crowd out" intrinsic motivation, but it does rebalance effort away from learning and towards output. Learning has complex interactions with performance measurement. A higher rate of learning tends to reduce performance measure distortion, especially for a very distorted measure. It also tends to reduce the potential for manipulation, as does multitasking. However, if job design is strongly imbalanced towards a few key tasks, or learning varies significantly across tasks, a higher rate of learning may increase the employee's ability to manipulate the measure. In that case the firm might prefer an incentive with no autonomy, or autonomy with no incentive.