Sleep dependent learning in contour integration

Abstract

Aim. A large body of evidence supports the involvement of sleep dependent mechanisms in perceptual learning (Karni et al, 1994; Stickgold et al, 2000). However, sleep dependence has only been studied extensively in the texture discrimination paradigm. We have indicated earlier (Kozma & Kovacs, 2002) that improvement in a contour integration task might also depend on sleep, but our design did not make a clear distinction between time vs. sleep dependent learning. Here we clarify the role of sleep in the contour integration task. Method. The contour integration task consisted of an egg-shaped, closed contour of Gabor patches on a background of randomly positioned and oriented Gabor patches. Subjects had to determine the direction where the egg-shape was pointing to at various levels of difficulty. 40 subjects practiced in the contour integration task over five 15 min sessions (240 trials/session) with 12 hours between sessions. We introduced a circadian phase shift between two groups of subjects. One group of subjects started the 1st session at 8 a.m. (Morning Group/MG), while the second group started at 8 p.m. on the same day (Evening Group/EG). Result. By the 4th session (2 nights of sleep for EG subjects, and only 1 night of sleep for MG subjects), EG performance was significantly better than MG performance. Conclusion. The ‘circadian shift’ design (see also Walker, 2002) allowed us to dissociate the impact of time vs. sleep on perceptual learning. We can now safely say that learning in the contour integration task is dependent on sleep.