The entropic brain today

Carhart-Harris presents the entropic brain hypothesis (EBH) as a model of conscious states grounded in information-theoretic entropy, rather than thermodynamic entropy. In the Introduction, he explains that the hypothesis treats expansive states of consciousness, such as psychedelic states, as states in which spontaneous brain activity becomes less predictable and carries more independent bits of potential information. He also contrasts this with states of reduced consciousness, where brain entropy tends to fall. The paper argues that the literature since the earlier 2018 version of the EBH has expanded enough to justify an update, and it places the model in relation to alternative theories of consciousness, complexity science, and clinical psychedelic research. The purpose of the article is to review and synthesise supportive evidence for the EBH across multiple domains, including psychedelics, sleep and dreaming, meditation, psychosis, disorders of consciousness, and other altered states. The authors also aim to discuss how the hypothesis relates to mental health, therapeutic action, and artificial intelligence, while considering challenges and limitations. The paper is therefore a broad integrative review and conceptual update rather than an original empirical study.

This is a narrative review and conceptual synthesis rather than a new experimental study. The author draws on earlier work from neuroimaging, electrophysiology, pharmacology, clinical studies, animal studies, and computational modelling to evaluate whether different measures of brain entropy track the breadth of conscious experience. The review organises evidence by state or domain. It considers psychedelic drug states, sleep and dreaming, meditation, flicker stimulation, atypical breathing, stress, psychosis, disorders of consciousness, ageing and neurodegenerative illness, and other physiological or behavioural correlates. Across these domains, the authors focus on measures such as Lempel-Ziv complexity (LZc), CSER, Shannon entropy, sample entropy, NGSC, connectome harmonics, regional homogeneity, perturbational complexity index, and related fMRI, EEG and MEG metrics. The paper also re-analyses and reinterprets prior findings, particularly the psilocybin and LSD MEG/fMRI literature. It compares how entropy measures behave under different sensory contexts, discusses which metrics are sensitive to temporal versus spatial effects, and uses these patterns to argue that not all complexity measures are equivalent. No new participants, interventions, or statistical analyses are reported in the extracted text for this article itself; instead, the authors synthesise and compare previously published results.

The central result reported by the paper is that a wide range of prior studies are consistent with the EBH: brain entropy tends to increase in states the authors regard as expansive, and decrease in states of reduced consciousness. Under classic psychedelics such as psilocybin, LSD and DMT, multiple studies are said to show increased entropy on EEG, MEG and fMRI-derived measures, especially LZc, CSER, NGSC, connectome harmonics and regional homogeneity. Carhart-Harris notes that in some studies these increases correlated with subjective intensity, richness of experience, or insight. He also cites evidence that increased entropy under high-dose psilocybin predicted improved well-being one month later, and that this relationship was mediated by psychological insight. For fMRI, the paper highlights replicated findings of default mode network desynchronisation, increased coupling with other brain systems, increased spatial entropy, and broader repertoires of connectome harmonics under psychedelics. It also notes that ketamine at psychedelic-like doses often shows similar effects, whereas non-psychedelic drugs such as methylphenidate and methamphetamine generally do not. Some exceptions are described: low-dose or non-psychedelic contexts sometimes fail to produce a clear entropy increase, and very high ketamine doses, when they induce unconsciousness, can reduce entropy. The author also mentions that certain measures are more sensitive than others, with fMRI appearing more suited to spatial than temporal entropy effects. Beyond drugs, the paper reports increased entropy in REM sleep compared with NREM sleep, in intense meditative states, during flicker-light experiments, after atypical breathing exercises, under experimentally induced stress, and in some cases of early unmedicated psychosis, mania, awe, near-death-like experiences, improvised music listening, and pre-ictal epileptic aura. In contrast, low entropy is repeatedly reported in disorders of consciousness, deep sleep, anaesthesia, seizure, delirium, ageing, Alzheimer disease, mild cognitive impairment, and related neurodegenerative conditions. Carhart-Harris also notes peripheral or behavioural parallels, including more entropic cardiac activity, breathing, and language under psychedelics. Overall, the results are presented as converging evidence that entropy rises with expanded conscious states and falls with diminished conscious states.

The paper interprets the accumulated literature as supporting the EBH as a useful descriptive model of conscious states. Carhart-Harris argues that the strongest evidence comes from convergence across modalities and measures, the repeated observation of entropy increases under psychedelics, and the opposite pattern in disorders of consciousness and sleep-related unconsciousness. He emphasises that the EBH is not intended as a full explanatory theory of consciousness, but as a bridging neurophysiological and phenomenological account of how subjective breadth maps onto spontaneous brain dynamics. In his view, the psychedelic state is especially informative because it makes the relation between brain entropy and subjective experience more visible. Relative to earlier research, the author says the literature now extends beyond the original 2014 and 2018 formulations. He notes that later work has clarified that the most reliable effects are often found with high-frequency EEG/MEG measures and with fMRI metrics that capture spatial organisation rather than temporal resolution. He also suggests that the findings have helped generate related models, such as REBUS and the canalisation model of psychopathology, which use the EBH as a conceptual foundation for understanding plasticity and therapeutic change. The main limitations he acknowledges are conceptual and methodological. First, he argues that the field sometimes conflates different kinds of complexity, especially type 1 complexity, meaning information-theoretic entropy, and type 2 complexity, meaning structured complexity or criticality. He cautions that not all metrics used in the literature are equally valid indicators of the entropic brain effect. Second, he notes that some phenomena may involve expanded consciousness without necessarily being enriched in content, making it important to distinguish expansiveness from richness. He also stresses the need for better metrics, more careful filtering and recording choices, and more work across non-human species and non-drug states. In terms of implications, Carhart-Harris suggests that the EBH may help explain why psychedelics can be therapeutically useful, particularly through a state of increased plasticity and insight that may facilitate psychological change when combined with supportive context. He also implies possible relevance to disorders of consciousness, though he acknowledges that improvements in subjective breadth may not translate into measurable access consciousness or behavioural recovery. He presents the model as useful but incomplete, and as one that will need refinement if future work can better separate different aspects of consciousness and complexity.

Carhart-Harris concludes that the EBH has remained useful and adaptable since its initial formulation, and that the current literature supports its validity across several domains. He presents it as a parsimonious model that continues to generate testable ideas about consciousness, mental health, and psychedelic therapy. He also suggests that future work should refine the model, identify its weaknesses, and determine which aspects of complexity or criticality are most relevant to different kinds of conscious experience.