Unravelling the neural signatures of dream recall in EEG: a deep learning approach

Anonymous

TL;DR: We investigate the neural basis of dream recall using convolutional neural network and feature visualization techniques, like tSNE and guided-backpropagation.

Keywords: CNN, EEG, sleep, dreamers, tSNE, guided-backpropagation

Abstract: Dreams and our ability to recall them are among the most puzzling questions in sleep research. Specifically, putative differences in brain network dynamics between individuals with high versus low dream recall rates, are still poorly understood. In this study, we addressed this question as a classification problem where we applied deep convolutional networks (CNN) to sleep EEG recordings to predict whether subjects belonged to the high or low dream recall group (HDR and LDR resp.). Our model achieves significant accuracy levels across all the sleep stages, thereby indicating subtle signatures of dream recall in the sleep microstructure. We also visualized the feature space to inspect the subject-specificity of the learned features, thus ensuring that the network captured population level differences. Beyond being the first study to apply deep learning to sleep EEG in order to classify HDR and LDR, guided backpropagation allowed us to visualize the most discriminant features in each sleep stage. The significance of these findings and future directions are discussed.