Noticing the location of an object that causes a change to the mean of a set relies on the ability to determine the mean of the set, and detect that a change has occurred (Rensink, 2002). Previous research suggests that people are able to retain information about the mean emotion of a set of faces even when they are unsure which items changed between the two sets (Haberman & Whitney, 2011). Subjects in that study, however, could use a strategy of localizing the most emotionally extreme face in the set to reliably determine the correct response in the mean discrimination task. In the present study, the utility of this strategy was eliminated. Subjects completed 4 blocks of trials consisting of 48 trials per block. On each trial, subjects viewed two consecutive displays of faces contained within circles. Four items increased (or decreased) in size or emotional intensity. In Experiment 1, subjects first determined whether average size or emotion increased or decreased from the first display to the second, then localized one of the four changed items. In Experiment 2, the order of responding was reversed. The results suggest that when performing both a mean discrimination and localization task, subjects use their knowledge of which stimulus in the set changed to guide their response on the mean discrimination task. Focusing attention to a local region of a display prevents the global distribution of attention necessary for perceptual averaging (Chong & Treisman, 2003). Thus, averaging is not possible when change detection fails.