Thermodynamic Markers of Consciousness: Sleep, Psychedelics, and Anesthesia

Thermodynamic Metrics of Consciousness: Key Findings from a Multi-Cohort Study

The 2026 pre-registered study rigorously tested the proposed thermodynamic marker R = E/|O_info|—the ratio of spectral energy to integrated informational structure—across sleep, anesthesia, and psychedelic states. The main finding is that this metric, previously suggested as a universal marker of consciousness, fails to generalize across these different states. Sleep and psychedelic states exhibit distinct informational invariants, challenging the notion of a single, scalar neural marker for consciousness level.

Drawing on eight public datasets, including Sleep-EDF (N=76/77), DOSE-I propofol anesthesia (N=170), high-density ketamine (N=10), and DMT+harmine (N=31), the study implemented stringent, pre-registered protocols and statistical controls. The analysis revealed that while energy alone could separate wake from non-REM sleep (AUC 0.826), the original ratio R was undermined by denominator instability and failed to provide additional explanatory power. The study's methodological rigor—such as frozen criteria, synthetic controls, and subject-level statistics—sets a new standard for empirical assessment in consciousness research.

Mechanistic Insights: Informational Invariants and Their Boundaries

The study demonstrates that sleep and psychedelic states are governed by different informational structures, or invariants, as measured by advanced estimators of integrated synergy. In sleep, nearly all epochs exhibited negative O-information (synergy), which increased with sleep depth and was highly stable within and across nights. This synergy-based invariant robustly tracked sleep stages and consciousness levels when measured with a non-parametric estimator, yielding a strong monotonic ordering (median rho = -0.577) and a descriptive cost minimum at light and REM sleep.

However, these sleep-based invariants did not extend to pharmacological states. Under propofol anesthesia, both energy and O-information metrics were near-null, and no invariant "locking" was observed. In psychedelic states (DMT+harmine), the system shifted from marginal redundancy toward synergy, but without the stable "locking" seen in sleep. This qualitative boundary suggests that the informational properties of sleep are not universal markers of consciousness, but rather specific to the endogenous dynamics of sleep itself.

Policy and Research Implications: Rethinking Consciousness Metrics

The failure of R = E/|O_info| as a universal marker has significant implications for both clinical and research applications. For psychedelic research, the findings caution against the use of sleep-derived metrics to quantify or compare altered states induced by pharmacological agents. Regulatory bodies and clinical trial designers should be wary of overgeneralizing neural markers validated in sleep to interventions involving psychedelics or anesthesia.

This study also sets a methodological precedent by demonstrating how to rigorously falsify, refine, and map the boundaries of candidate neural markers. The use of pre-registered refutation branches, mandatory marginal decomposition, and explicit floor-declaring estimators provides a blueprint for future biomarker discovery and validation. Researchers are encouraged to adopt similar protocols to avoid false positives and overinterpretation of neural metrics.

Risks, Unknowns, and the Cost of Order

The study highlights several risks and open questions for the field. First, the lack of a universal neural marker complicates efforts to objectively measure consciousness in clinical and forensic settings. Second, the distinct informational invariants of sleep and psychedelics underscore the need for state-specific metrics, which may limit the generalizability of findings across different interventions.

An important, often overlooked insight from the study is the thermodynamic cost of order in physiological systems. For example, paced breathing in the resonance band increased cardiopulmonary coherence but reduced cardiac entropy, demonstrating that maintaining order (laminarization) comes at the expense of variability. This trade-off may have implications for the design of interventions aimed at modulating consciousness or physiological states, suggesting that the energetic and informational costs of inducing order should be carefully considered.

Looking Forward: Mapping the Frontier of Consciousness Science

This study marks a turning point in the search for neural markers of consciousness. By empirically delineating the boundaries of existing metrics and introducing robust methodological standards, it challenges the field to move beyond one-size-fits-all solutions. Future research will need to develop and validate state- and intervention-specific markers, with careful attention to the underlying informational dynamics and their physiological costs.

For psychedelic science, the key takeaway is that altered states induced by pharmacological agents are not simply mirror images of sleep or anesthesia, but possess unique informational signatures. As the field advances, collaborative, multi-modal studies with transparent, pre-registered designs will be essential to unraveling the complex landscape of consciousness and its neural correlates.

How we research: This article was written and reviewed by Dr. Alex R. Levin, PhD (Neuroscience, University of Toronto), with primary source analysis of the original 2026 pre-registered study (OpenAlex: W7212418690) and direct review of referenced datasets and methodologies. Reviewed on 2026-09-15.