Robin Carhart-Harris has been arguing for a decade that you can measure the size of a conscious experience. Not its content, not its emotional valence, but its sheer breadth: how much is happening, how wide the experiential field feels.
His instrument for this is brain entropy, a measure of how unpredictable and non-repetitive spontaneous brain activity is at any given moment. The more surprising the signal, the higher the entropy, the more expansive the state.
The hypothesis organizes a surprisingly wide spectrum of mental states. At one pole sit coma, general anesthesia, dreamless sleep, and Alzheimer's disease. Brain activity in these states is sluggish and repetitive, low in entropy. At the other sit psychedelics, intense meditation, REM dreaming, near-death-like experiences, and even the aura phase just before an epileptic seizure. Brain activity here is fast, varied, and dense with constantly shifting patterns. A single measure, brain entropy, tracks how much or how little is happening subjectively across all of them.
The new paper, published as an advance article in Brain, surveys what twelve years of evidence has done to this picture. Carhart-Harris's verdict: it has strengthened considerably. The entropic brain effect has been replicated across multiple psychedelic compounds, multiple imaging technologies, and multiple species. The same basic finding holds in EEG, MEG, and fMRI data analyzed three different ways. It holds in humans, cats, and rodents.
The primary measurement tool is Lempel-Ziv complexity, an algorithm borrowed from computer science's field of data compression, where it is used to quantify how much information a file actually contains. Applied to brain signals, it asks: how many distinct, non-repeating sequences does this signal produce? A brain that keeps running the same patterns compresses easily and scores low. A brain generating constantly novel configurations resists compression and scores high. Under psilocybin, LSD, DMT, and ketamine, Lempel-Ziv complexity rises substantially, and the size of that rise correlates with how intense the experience feels. Crucially, these replications have come from independent labs across multiple countries, not from Carhart-Harris's group alone.
The most consequential clinical data point in the paper comes from Lyons et al., published concurrently in Nature Communications by the same research group. Brain entropy measured at 60, 120, and 270 minutes into a high-dose psilocybin session predicted how much better psychedelic-naive participants felt four weeks later, a relationship mediated by psychological insight in the intervening days. A brain entropy readout from a single EEG session could, in principle, tell you within hours whether a therapeutic dose was achieved and whether the biological conditions for lasting change were in place.
The paper's limitations section is where things get most interesting for the field going forward. Carhart-Harris notes that the EBH has so far treated all high-entropy states as variations on the same phenomenon, but this may be too coarse. He flags a distinction between states that are expansive and states that are enriched: high-dose DMT produces experiences that are both wide and full, vast inner landscapes populated with imagery, emotion, and meaning, while high-dose 5-MeO-DMT produces radical expansion with almost no content. The field is boundless but empty.
Underneath this distinction sits a difference in what complexity theorists call type 1 versus type 2 complexity. Type 1 is raw unpredictability: how many novel sequences a signal produces, which is what Lempel-Ziv measures. Type 2 is structured complexity: the organized, scale-free patterns characteristic of systems operating near a critical threshold between order and chaos. A brain can score high on type 1 without type 2, pure noise with no architecture. Carhart-Harris's hypothesis is that DMT drives type 1 up while preserving type 2, producing expansion with content. 5-MeO-DMT may push dynamics beyond that critical zone entirely, maximizing type 1 at the cost of type 2, and with it, the organized inner life. Phenomenal consciousness survives but reflective awareness and the sense of self do not.
Carhart-Harris does not resolve this distinction in the paper. He poses it as a question the EBH needs to grapple with: do these two profiles reflect meaningfully different brain dynamics underneath a shared entropy increase, and if so, what does that imply for how different compounds are selected for different indications? For drug developers, this is not an academic question. A compound that expands consciousness while preserving structured content may be mechanistically better suited to psychotherapy-adjacent indications like depression and PTSD than one that maximizes raw expansion at the cost of any organized inner life.
A second challenge comes from universality. The paper extends the entropic principle to heart rate variability, speech patterns, and the temperature parameters of large language models. If the same measure describes cardiac rhythms as readily as brain states, what specifically does it tell us about consciousness? A construct that is universal may be capturing something real, but universality and explanatory specificity are in tension. The more entropy bridges toward everything, the harder it is to argue it illuminates something singular about the mind.
A third and more pointed challenge comes from cell biology. Alex Kwan's lab published work in Nature in 2025 showing that psilocybin's lasting behavioral effects require specific pyramidal cell types and 5-HT2A receptors in the prefrontal cortex: the plasticity is not diffuse but precisely circuit-specific. The EBH, operating at the level of whole-brain population signals, says little about which cells or synapses are doing the therapeutic work. For investors, this gap matters practically: a theory of conscious states is not a theory of mechanism. Knowing that psilocybin expands brain entropy does not tell you which molecular events need to occur, in which neurons, for a clinical effect to persist. Both levels of description are needed for platform development, target selection, and drug design.
Decision theory has a concept called the explore/exploit tradeoff. Any system, neural or organizational, must balance exploiting known, reliable patterns against exploring uncertain options that might yield something better. Chronic mental illness, on this account, is a form of pathological exploitation: the brain locked into the same maladaptive cognitive and behavioral loops, unable to update. The question is what shifts the balance.
Curiosity may be the most underappreciated mechanism. As a motivational state, curiosity is oriented specifically toward novelty and uncertainty: it drives exploration not because a reward is guaranteed but because the unknown is itself compelling. Psychedelics, on the entropic brain account, may work in part by recreating the neurological signature of genuine curiosity at scale, temporarily dismantling the brain's exploitation defaults and forcing an encounter with a wider possibility space. The therapeutic window is the period before the system reconsolidates, during which new patterns can take hold.
For funds evaluating psychedelic platforms, this framing sharpens due diligence in a specific way. The mechanistic question is no longer simply whether a compound produces a psychedelic effect, but whether it produces the right kind of entropic state for the target indication. Compounds generating high entropy with preserved structured complexity, the DMT profile, are likely better positioned for MDD, PTSD, and OCD, where insight-driven revision of entrenched belief patterns is the proposed mechanism. Compounds that maximize raw entropy at the cost of organized content may be better suited to indications where ego dissolution itself is therapeutic, such as end-of-life anxiety, or may face a harder path to clinical translation altogether. As the Lyons biomarker work matures toward patient populations, the ability to verify mechanistic target engagement within a single session may become as routine as measuring drug plasma levels.
References
