Theodore LaBonte-Clark¹, Teruaki Kido¹, Takeo Watanabe¹, Yuka Sasaki¹; ¹Brown University

While visual perceptual learning (VPL) — long-term improvements on visual tasks through experience — is known to be enhanced by post-training sleep, the underlying neural mechanisms of sleep-dependent improvement remain unclear. A leading hypothesis is that sleep facilitates learning through reactivation, the re-emergence of training-related neural activity during sleep. VPL is characterized as specific to the trained feature. Here, we tested a novel model proposing that reactivation contributes directly to such feature specificity. In the experiment, participants were trained to detect Gabor patches with a single orientation. During the post-training sleep, we recorded fMRI activity from early visual areas simultaneously with polysomnography to determine sleep stages. We used an fMRI decoder, constructed in a previous session and capable of classifying four Gabor orientations, including the trained orientation, to evaluate representations in visual cortex during sleep. In the event that VPL occurred, the decoder identified neural activity corresponding to the trained orientation above chance, suggesting selective reactivation of the trained orientation during sleep. Furthermore, greater decoding performance for the trained orientation appeared to accompany greater performance improvement after sleep. This pattern suggests that neural reactivation during NREM sleep may reflect the trained orientation and support the proposal that reactivation plays a key role in mediating the orientation specificity characteristic of VPL.

Acknowledgements: NIH R01EY019466, R01EY027841, R01EY031705, NSF-BSF BCS2241417