Entropic Brain Hypothesis Gains Support as a Model of Conscious States
A decade-long review finds brain entropy tracks the breadth of conscious experience, rising in psychedelic states and falling in ageing, anaesthesia and dementia.
Summary
This review by Robin Carhart-Harris updates the entropic brain hypothesis, first proposed in 2014 and revised in 2018. It holds that the felt breadth of conscious experience tracks the unpredictability, or entropy, of spontaneous brain activity. Entropy is higher in expansive states such as psychedelic experiences, expert meditation, REM sleep and early psychosis. It is lower in deep sleep, anaesthesia, seizure, disorders of consciousness, ageing, cognitive impairment and neurodegeneration. The author argues the idea has convergent, predictive, discriminative and external validity. For example, greater entropy under psilocybin in a supportive setting predicted better well-being one month later. EEG/MEG Lempel-Ziv complexity and spatially sensitive fMRI metrics work best for detecting the effect. The model is descriptive rather than explanatory, and it mainly captures simple (type 1) complexity.
Detailed Summary
Why it matters: Understanding what distinguishes rich, expansive conscious states from diminished ones could inform how we diagnose and treat disorders ranging from depression to dementia and disorders of consciousness. The entropic brain hypothesis (EBH) offers a deliberately simple bridge between brain physiology and subjective experience. This review, written by the hypothesis's originator, takes stock of the evidence since its 2014 introduction and 2018 revision.
What was covered: This is a narrative review, not a new experiment. The EBH states that, within a functionally relevant range, the entropy (information-theoretic unpredictability) of spontaneous brain activity encodes the felt breadth of a conscious state. The article distinguishes type 1 complexity (entropy, which rises toward pure randomness) from type 2 complexity (structured, scale-free or critical organisation). The EBH mainly concerns type 1, and the author acknowledges that type 2 properties may be needed for a fuller account. The hypothesis grew out of early psilocybin neuroimaging. Reduced perfusion on arterial spin labelling was first read as a thermodynamic effect but is now seen as likely vasoconstriction and transient neurovascular decoupling. Psychedelics are now better described as causing excitatory dysregulation, especially in cortex. Resting-state fMRI repeatedly shows default-mode network desynchronisation and reduced segregation between networks.
Key results: Expansive states with higher brain entropy include expert meditation, flicker-light stimulation, near-death-like experiences, atypical breathing, REM sleep, the pre-ictal aura, unmedicated early psychosis and psychedelic states. States of reduced or absent consciousness with lower entropy include disorders of consciousness, deep sleep, anaesthesia, seizure, post-stroke states, ageing, cognitive impairment and neurodegenerative illness. EEG and MEG, analysed with Lempel-Ziv complexity (LZc), are especially sensitive in the temporal domain. fMRI has low temporal resolution but high spatial resolution, so spatial metrics suit it better. Three such metrics are connectome harmonic decomposition, normalised global signal complexity and cortical spiral waves. Inconsistent fMRI entropy findings are attributed to differences in metric sensitivity. The author claims convergent, correlative, predictive, discriminative and external validity. Increased entropy under psilocybin in a supportive context predicted improved well-being at one month. Entropy changes selectively index breadth of experience rather than arousal. An entropy-related function is also used in generative AI, which speaks to portability.
Implications: The author proposes the EBH as a basic principle of conscious states that applies across imaging modalities, biological systems and possibly substrates. He discusses its relevance to the nature of consciousness, psychedelic therapy, mental illness, the action of effective treatments, and how artificial general intelligence relates to human intelligence. The link between low entropy and ageing or neurodegeneration suggests brain entropy might serve as a marker of brain health.
Caveats: The EBH is descriptive, not a causal theory of consciousness. It is simple and may underfit the phenomenon, and no equally powerful measure of type 2 complexity exists yet. Some fMRI results have been inconsistent, and the review is written by the hypothesis's originator.
Key Findings
- Brain entropy rises in expansive states (psychedelics, expert meditation, REM sleep, early psychosis) and falls in anaesthesia, deep sleep, seizure and disorders of consciousness.
- Ageing, cognitive impairment and neurodegenerative illness are linked to low brain entropy, suggesting a possible marker of brain health.
- Higher entropy under psilocybin in a supportive context predicted improved well-being one month after dosing.
- Entropy indexes breadth of subjective experience rather than arousal, supporting discriminative validity.
- Lempel-Ziv complexity on EEG/MEG, and spatial fMRI metrics such as connectome harmonics, detect the effect best.
Methodology
Narrative review article in Brain (2026) by the originator of the hypothesis. It synthesises human neuroimaging (fMRI, EEG, MEG) and subjective-report evidence across many conscious states and drugs, and compares the EBH with other consciousness models. It does not report a systematic search or quantitative meta-analysis. The text available for this summary was truncated partway through the fMRI section, so later sections were not reviewed in detail.
Study Limitations
The review is by the hypothesis's originator and is not systematic, so selection and confirmation bias are possible. The EBH mainly captures simple (type 1) complexity, is descriptive rather than explanatory, and has no validated type 2 complexity counterpart. fMRI entropy results are inconsistent across metrics. The text provided was cut off mid-review, so this summary relies on the abstract, introduction and early sections, and the later discussion of challenges, AI and treatment implications is not covered in detail.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: