Feedback is essential for many kinds of learning, but the cognitive processes involved in learning from feedback areunclear. In models of category learning, feedback is typically treated as an error signal without a temporal component. Weconducted two simple category learning experiments that manipulated the duration of feedback (1s vs. 9s) and investigatedthe effect on learning and gaze. In two different category structures, participants in the longer feedback condition learnedfaster. The analysis of gaze data showed several findings. Participants in the 9s condition had longer fixations, and in bothconditions and experiments, participants spent far more time looking at stimulus features than the feedback. Overall, ourfindings provide empirical support for the idea that feedback processes, and temporal factors more generally, have muchto tell us about how people learn categories.