Abstract
Extended practice — usually consisting of several hundred trials — substantially improves performance on a face-identification task. In this study we examined how many trials are necessary for this improvement, or perceptual learning, to occur. On consecutive days, observers performed a ten-alternative forced choice face identification task. Faces were presented at one of seven different contrast levels in three levels of external noise for a total of 21 stimulus conditions. The critical manipulation was the number of trials presented in each condition on Day 1. Separate groups completed either 1, 5, 10, 20, or 40 trials per stimulus condition, for a total number of trials that ranged from 21 to 840. The amount of time spent in the lab on Day 1 was equated across subjects by requiring everyone who completed fewer than 840 trials to perform an unrelated visual (filler) task. On Day 2, all observers completed 840 trials. Observers in a zero-trials condition (i.e. the control group) did only the filler task during the first session. Comparisons of face identification thresholds measured on Days 1 and 2, using only subjects that received at least 210 trials on Day 1, showed that thresholds decreased significantly with practice, indicating that perceptual learning occurred. Surprisingly, with the exception of the control group, thresholds did not differ across groups on Day 2. In other words, thresholds for subjects receiving 21–840 trials of practice did not differ significantly, and were significantly lower than thresholds for control subjects. These results suggest that relatively few trials are needed to produce perceptual learning. We currently are conducting experiments to determine what aspects of the brief exposure to our stimuli and/or task are necessary for learning to occur.