Multi-metric evaluations of acute psychedelic effects on fMRI brain entropy
Drummond E-Wen Mcculloch, Anders S. Olsen, Brice Ozenne, K O Larsen, Dea Siggaard Stenbæk, Sophia Armand, M. Madsen, Gitte M. Knudsen, Patrick M. Fisher
Nature Communications June 24, 2026 DOI: 10.1038/s41467-026-74215-5 via OpenAlex
Summary
AI-generated from the abstractA prominent theory holds that psychedelics increase brain entropy, but past studies have used many different entropy measures. This work analyzed 121 fMRI scans from 28 healthy adults before and after psilocybin, testing 14 entropy metrics with two brain-parcellation methods and seven denoising pipelines. Five metrics—including Shannon entropy of spatial eigendistribution, path-length, instantaneous correlations, brain-state switching, and sample entropy at short time-scales—consistently showed positive associations with psychedelic effects. However, eight metrics showed no significant effects, and Lempel-Ziv complexity gave inconsistent positive results. The entropy measures correlated poorly with each other, indicating that brain entropy is not a single, unified phenomenon.
Study at a glance
| Characteristics | Observational cohort Peer reviewed |
|---|---|
| Sample size | 28 |
| Population | Healthy adults |
| Intervention | Psilocybin |
| Keywords | Sample entropy Voxel Neuroimaging Pattern recognition psychology Brain mapping |
| Key finding | Psychedelic effects on brain entropy are nuanced: only some entropy metrics consistently increase, while others show no effect, and the metrics are poorly correlated. |
Abstract
A prominent theory of psychedelics is that they increase brain entropy. Thirteen studies have evaluated psychedelic effects on fMRI brain entropy, each applying a distinct measure. Here we evaluated these metrics in an independent 28-participant healthy cohort with 121 pre- and post-psilocybin fMRI scans. We assessed relations between brain entropy and objective and subjective psychedelic drug effects using linear mixed-effects models. All metrics were evaluated using two parcellation strategies and 7 denoising pipelines. We observed consistent significant positive associations for Shannon entropy of the spatial eigendistribution of the time by voxel matrix, path-length, instantaneous correlations, brain-state switching, and sample entropy at short time-scales. We consistently did not observe significant effects for 8 of 14 entropy metrics and observe inconsistent positive effects for Lempel-Ziv complexity of the BOLD signal. Brain entropy quantifications showed limited inter-measure correlations. Our observations support a nuanced acute psychedelic effect on brain entropy, empirically demonstrating that these metrics do not reflect a singular construct.