Neural Signatures of Conscious Experience During Sleep: A Serial Awakening Study Using High-Density EEG
Atakan Selte, Steven E. Haworth, Thomas J. Vanasse, Tariq Alauddin, Klevest Gjini, Santiago Philibert‐Rosas, Cameron Brace, Brinda Sevak, Brady A. Riedner, Mariel Kalkach Aparicio, Giulio Tononi, Melanie Boly, Aaron F. Struck
bioRxiv (Cold Spring Harbor Laboratory) August 20, 2026 preprint DOI: 10.64898/2026.08.16.745040 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Observational cohort |
|---|---|
| Sample size | 140 |
| Population | Participants undergoing overnight high-density EEG with serial awakenings |
| Key findings | Dreaming during NREM sleep was associated with reduced posterior alpha and delta power and increased gamma-related measures, alongside altered large-scale network organization; classifiers distinguished dreaming from no-experience awakenings with ROC-AUC of 0.80. |
Abstract
Abstract Identifying neural signatures of consciousness remains a central challenge in neuroscience. Sleep offers a tractable model for comparing brain activity in the presence or absence of subjective experience while minimizing behavioral responsiveness confounds. Using overnight 256-electrode high-density EEG in 140 participants and a serial-awakening paradigm, we analyzed 699 non-rapid eye movement (NREM) sleep stage 2 and 3 awakenings (351 dreaming experience, 348 no experience). Features from the 60s preceding awakening included regional spectral power, lagged-coherence connectivity, graph-theoretic metrics and gamma-to-alpha power ratios. Dreaming experiences were associated with shifts in posterior spectral balance, particularly reduced alpha and delta power and increased gamma-related measures, together with altered large-scale network organization. In participant-level cross-validated machine-learning analyses, all classifiers performed above chance, with the best ensemble model reaching an ROC-AUC of 0.80 and average precision of 0.80. These findings identify reproducible posterior electrophysiological and network-level signatures of conscious states during NREM sleep.