Abstract
A growing body of evidence indicates that sleep facilitates visual perceptual learning (VPL). However, how this facilitation occurs is controversial: researchers in the field are divided into two groups with two completely different views: use-dependent and learning-consolidation models. The use-dependent model assumes that sleep downscales overly increased synapses in the networks used during wakefulness, irrespective of whether learning is involved in the networks or not, and leads to the survival of only significant synapses. Downscaling is assumed to be involved in long-term depression related to slow-wave activity (SWA). The learning-consolidation model assumes that sleep affects the networks specifically related to learning by replaying activity patterns during training. The replay is assumed to be involved in long-term potentiation related to sigma activity. Although usage and learning components were not separated in previous studies, here we successfully dissociated the usage from learning components using an interference effect and tested which model is correct. There were two conditions. In a learning condition (n=12), participants were trained on one texture discrimination task (TDT), the learning of which is associated with response changes in the region of the early visual cortex that retinotopically corresponds to the trained visual field (trained region). In an interference condition (n=12), two different TDTs were serially trained so that learning should not occur due to mutual interference effects. The use-dependent model predicts increased SWA in the trained region during the post-training sleep in both conditions, whereas learning-consolidation model predicts sigma increase in the trained region during the post-training sleep only in the learning condition. Participants were trained with TDT before sleep and retested with TDT after sleep. Learning occurred only in the learning condition. EEG-source localization analysis indicated significant increase in sigma activity in the trained region during the post-training sleep. These results supports learning-consolidation model in VPL.
Meeting abstract presented at VSS 2015