Abstract
Several reports suggest that perceptual learning derives critically from consolidation processes that are sleep-dependent. We examined the extent to which the performance gains that are typically evidenced with practice on a 10-AFC face-identification task depend on sleep between sessions. The stimuli were faces embedded in one of three levels of external noise, and presented at one of seven different contrasts using the method of constant stimuli. Two separate groups of observers performed the task on a training and test session that was separated by a 12-hour interval. The ‘No sleep’ group performed the training (9am) and test (9pm) session on the same day, and did not nap between sessions. The ‘Sleep’ group performed the training session at 9pm and rested overnight before performing the test session at 9am the next day. We assessed the time-course of learning by splitting the sessions into quartiles and estimating proportion correct at each quarter within each session. Both groups performed comparably on the training session and evidenced substantial within-session gains in performance that did not differ as a function of time of day. There was a slight but significant advantage for the ‘sleep’ group relative to the ‘no sleep’ group on the test session, but the time-course of learning reveals that gains from within the first session account for a larger proportion of the total improvement found in both groups. Thus, we find that perceptual learning occurs in the absence of sleep and conclude that sleep-related consolidation processes comprise only a small component of learning to identify faces.