Psychedelics distinctly alter brain entropy and complexity compared to psychostimulants

Introduction

The Entropic Brain Hypothesis proposes that psychedelics shift functional brain dynamics towards greater entropy and complexity and that this underlies both their subjective and therapeutic effects. This framework has motivated the application of myriad entropy and complexity metrics to brain imaging data acquired during the psychedelic experience. However, it remains unclear whether such changes are specific to psychedelics or reflect a more general response to psychoactive brain perturbation. Without pharmacological specificity, the interpretation of these metrics as psychedelic brain signatures and their potential utility as clinical biomarkers remains uncertain. In this study, we directly and systematically compared classical psychedelics and psychostimulants across multiple brain entropy and complexity metrics to identify measures indexing psychedelic-specific effects.

Methods

We analysed resting-state fMRI data from 79 healthy participants across three randomised, double-blind, placebo-controlled crossover trials (255 sessions). Each trial included a placebo condition, and participants received a subset of drugs depending on study allocation. Across studies, drugs administered included classical psychedelics (LSD, psilocybin, mescaline) and stimulants (MDMA, d-amphetamine). We evaluated six entropy and complexity metrics, yielding 53 outcomes at global and network levels (Schaefer-200 atlas; Yeo-7 networks). We used linear mixed-effects models to estimate and compare drug class effects, adjusting for sex, head motion, session order, prior drug exposure, and study project. We used the Holm–Bonferroni correction to control the family-wise error rate across all 53 outcomes. Our analysis framework was pre-registered prior to any data evaluation.

Results

Three metrics showed psychedelic-specific effects. Psychedelics significantly increased meta-state complexity compared to stimulants (β = 0.07 [95% CI: 0.04, 0.10]; p_FWER = 0.003). Multiscale sample entropy showed timescale dependence: at short timescales (scale 1; 3.6–5.4 s), psychedelics induced greater entropy than stimulants in transmodal networks, including the default mode, control, and dorsal attention networks (all p_FWER < 0.001), whereas at longer timescales (scale 5; 18–27 s), this pattern reversed (e.g., default mode: β = −0.03 [−0.04, −0.01]; p_FWER = 0.003). Psychedelics significantly increased dynamic conditional correlation entropy within and between transmodal networks compared to stimulants (p_FWER < 0.05 across multiple network pairs), indicating greater temporal variability in functional connectivity. In contrast, Lempel-Ziv complexity, normalised spatial complexity, and absolute modularity were significantly affected by drug class but did not differ significantly between drug classes.

Conclusions

These findings suggest that Lempel-Ziv complexity, normalised spatial complexity, and absolute modularity may index general sensitivity to psychoactive substances. In contrast, meta-state complexity, multiscale sample entropy, and dynamic conditional correlation entropy, particularly within transmodal networks, appear to capture features specific to the psychedelic brain state and may be better neuroimaging endpoints in clinical psychedelic research. Furthermore, the timescale-dependent MSSE profile, with increased entropy at short timescales and reduced at longer timescales, indicates that psychedelics reorganise, rather than uniformly increase, brain dynamics, refining how we understand psychedelics to affect brain entropy.

References

No references