State-Dependent Dissociation Between Neural Differentiation and Temporal Organization in Human Deep Sleep (Sleep-EDF)

This project investigates whether neural signal differentiation and temporal organization constitute empirically separable dimensions of human brain dynamics within a single neurophysiological state. Contemporary approaches in consciousness science often assume that reduced neural differentiation during low-arousal states (e.g., deep sleep or anesthesia) implies a global collapse of temporal structure. However, this assumption has rarely been tested directly using within-state analyses that control for arousal-related confounds. Using human sleep EEG data from the Sleep-EDF Database Expanded, this study implements a pre-registered analysis pipeline comparing wakefulness and deep NREM sleep (N3). Two families of EEG metrics are defined a priori: (A) signal differentiation metrics (Lempel–Ziv complexity, spectral entropy, and aperiodic spectral slope) and (B) temporal organization metrics (slow oscillation temporal variability, delta phase concentration stability, and spectral shape variability). All preprocessing steps, metric definitions, aggregation procedures, and statistical decision criteria are specified in advance using the ECLIPSE v4.0 preregistration framework, with cryptographic hashing to ensure analytic immutability. The primary objective is to test whether temporal organization exhibits meaningful inter-individual variability within N3 sleep even when neural differentiation is globally reduced, and whether the coupling between differentiation and temporal organization differs between wakefulness and N3. Crucially, the study focuses on within-state correlations rather than between-state contrasts, allowing a direct test of the commonly assumed equivalence between these two dimensions. The project is structured as a two-stage design. A small, pre-registered development cohort (N = 10 subjects) is used to evaluate feasibility and preregistered GO/NO-GO criteria. An independent holdout cohort (N = 43 subjects), reserved prior to analysis, is designated for confirmatory validation using an identical pipeline. Robustness is further assessed across multiple preregistered preprocessing and normalization configurations. This work does not claim to measure conscious experience directly, nor does it propose these metrics as diagnostic tools for consciousness. Instead, it aims to provide an operational, reproducible test of whether neural differentiation and temporal organization can dissociate within a low-arousal physiological state. Expected outcomes include either confirmation or falsification of preregistered hypotheses, with all results—positive or negative—reported transparently to inform future theoretical and methodological work in consciousness research.