The Complex Brain Hypothesis and why contentless awareness still shows high brain entropy
Jonas Mago, Edmundo Lopez-Sola, Jakub Vohryzek, Michael Lifshitz, Robin Carhart-Harris, Karl Friston, and Shamil Chandaria posted a paper to arXiv on 15 May 2026 titled “The Complex Brain Hypothesis, Resolving the Entropy-Content Conundrum in Minimal Phenomenal Experience” (arXiv:2605.16146). The hypothesis is a correction to the entropic brain hypothesis. That hypothesis, developed by Carhart-Harris and colleagues, treats the entropy of spontaneous brain activity as a marker of phenomenal richness. The correction is forced by a conundrum the data created. States of minimal phenomenal experience, contentless meditation, and possibly high-dose 5-MEO-DMT, show low content and yet appear with elevated neurophysiological entropy, the same signature the hypothesis associates with rich psychedelic experience.
The Complex Brain Hypothesis resolves the conundrum by distinguishing entropy from complexity. The richness of experience, the hypothesis holds, is not indexed by entropy but by complexity, and complexity is modulated by the grain of inference through which the brain resolves uncertainty. High-content psychedelic experience and contentless meditation both show elevated entropy, but they differ in complexity, and the difference is what the phenomenology tracks.
The conundrum in detail
The entropic brain hypothesis has been one of the most productive frameworks in the consciousness and psychedelics literature, and this site covered Carhart-Harris’s presentation of the framework at the Science of Consciousness 2026 conference. Its core claim is that the entropy of spontaneous brain activity tracks the richness of phenomenal content. Psychedelics, which produce high-content experiences, increase brain entropy. The conundrum emerged when neuroimaging studies of meditation-induced minimal phenomenal experiences, and possibly 5-MEO-DMT states, showed that states defined by their phenomenological simplicity also show increased neurophysiological entropy.
The problem is the opposite of what the hypothesis predicted. If entropy tracks content, then contentless states should show low entropy. They show high entropy. The correlation between entropy and phenomenal richness breaks in the minimal experience direction.
Complexity and the grain of inference
The Complex Brain Hypothesis replaces entropy with complexity as the index of phenomenal richness. The two are not the same. Entropy measures the dispersion of a distribution. Complexity, in the sense the hypothesis uses, measures the structure a system generates in resolving uncertainty, and it depends on the grain of inference.
The hypothesis proposes two regimes. High-content psychedelic experiences exemplify a fine-grained regime. Loosened constraints amplify fluctuations into proliferating content, so the model entertains many distinct contributions to experience. Minimal phenomenal experiences exemplify a coarse-grained regime. The system adopts a simpler model that dissolves variety into contentless awareness, an experience of awareness without objects. Both regimes can show elevated brain entropy, because both involve high dispersion of neural activity, but they diverge in complexity, and they diverge in phenomenology. The perturbational signatures, how the system responds to perturbation, distinguish the two where entropy could not.
The testable prediction
The hypothesis makes a specific prediction about where the conundrum should be resolved. If complexity, not entropy, tracks content, then measures of brain complexity, dynamical complexity rather than simple entropy, should differentiate minimal phenomenal experiences from high-content psychedelic experiences, even when entropy is elevated in both. The paper proposes that minimal phenomenal experiences are a privileged test case for computational theories of consciousness, because they control for content while preserving wakefulness.
For the site’s indicator checklist, the hypothesis supplies a refinement of the integration and complexity indicators. It argues that the relevant complexity is not raw dynamical richness but complexity at the right grain of inference, which is a more precise target than the generic claim that conscious states are complex.
Comparison to The Consciousness AI
The site’s framework treats consciousness as an emergent property of causal organization, and its generative world model work treats the brain’s generative model of the world as the seat of experience. The Complex Brain Hypothesis is directly in that tradition, since the grain of inference is a property of a generative model. The prediction, that coarse-grained inference produces contentless awareness while fine-grained inference produces proliferating content, maps onto the site’s treatment of predictive processing as the mechanism of conscious content.
The connection and the difference are worth stating. Both the site’s position and the hypothesis treat the generative model as central. The hypothesis adds a quantitative claim, that the grain of inference is the variable that separates contentless from contentful experience, which is a prediction the site’s state of the consensus framework would classify as an indicator the project could in principle test on its own generative model.
What the hypothesis does not settle
The paper is a hypothesis paper, and it acknowledges the limits. The conundrum it resolves is defined by the existing neuroimaging evidence, and the resolution is proposed rather than demonstrated. The decisive test, measuring complexity across minimal and high-content states in the same study with perturbational probes, has not yet been run. The hypothesis also inherits the open questions of the entropic brain literature, including whether the entropy reports are confounded by head movement, arousal, or the difficulty of holding a contentless state in the scanner.
The significance is that it repairs a broken correlation without abandoning the framework. The entropic brain hypothesis made a strong claim, entropy tracks content. The data broke the claim in one direction. The Complex Brain Hypothesis restates the claim in a form the data can test, complexity at the right grain tracks content, and identifies the experiment. That is a genuine contribution, and it is exactly the kind of refinement that keeps a theory alive.
*Jonas Mago, Edmundo Lopez-Sola, Jakub Vohryzek, Michael Lifshitz, Robin Carhart-Harris, Karl Friston, and Shamil Chandaria posted “The Complex Brain Hypothesis” to arXiv on 15 May 2026 as arXiv:2605.16146.