Effects of external stimulation on psychedelic state neurodynamics

Recent work has shown that classic psychedelics reliably increase measures of brain entropy, understood here as the informational complexity or diversity of neural signals, and that these increases relate to acute phenomenology and longer-term psychological change. However, a robust decrease in entropy is also a marker of reduced conscious level (for example during sleep), raising questions about how entropy relates to different facets of consciousness. The Entropic Brain Hypothesis (EBH) proposes that the rich altered states induced by psychedelics reflect an enrichment of spontaneous, population-level neural dynamics, but it remains unclear how contextual factors such as sensory environment and eye state (components of "set and setting") modulate this entropy-enhancing effect. Mediano and colleagues set out to test how external stimulation and eye opening interact with LSD to shape neural entropy and its relation to subjective experience. Using MEG recordings and subjective ratings collected while participants received intravenous LSD or placebo, the study compares four environmental conditions (eyes closed, music with eyes closed, eyes open fixation, and watching a silent nature video) to determine whether and how context alters both the magnitude of entropy changes and their correspondence with phenomenology. The work is framed as a proof-of-principle translational investigation in healthy volunteers with implications for the practice of psychedelic-assisted therapy.

The analysis reuses data from a previously reported experiment in which twenty healthy participants attended two sessions each: one with intravenous LSD (75 μg) and one with intravenous saline placebo. Session order was randomised and sessions were separated by two weeks; participants were blind to order (single-blind design). During each session whole-brain magnetoencephalography (MEG) was recorded while participants experienced four conditions: resting with eyes closed (closed), listening to instrumental ambient music with eyes closed (music), resting with eyes open fixating a dot (open), and watching a silent nature documentary (video). Visual Analog Scale (VAS) questionnaires were administered at the end of each session to capture subjective dimensions including intensity, emotional arousal, ego dissolution, positive mood, and simple and complex imagery (the imagery items were only rated in eyes-closed conditions). MEG data were acquired with a 271-gradiometer CTF system and structural MRI scans were obtained for source reconstruction. After excluding three participants for incomplete recording or excessive movement, data were preprocessed using FieldTrip: artefacts were removed by visual inspection and ICA to eliminate muscle and line noise, a second-order low-pass Butterworth filter at 100 Hz was applied, and data were split into 2 s epochs. Source time series (virtual sensors) were estimated using a regularised LCMV beamformer at the centroids of the AAL-90 atlas; Lempel-Ziv complexity (LZ), an algorithmic measure of signal diversity, was computed on these source time series and mapped to a standard template for group analysis. Statistical analysis combined non-parametric tests (Kruskal-Wallis and post-hoc t-tests) with linear mixed-effects (LME) models to assess predictors of LZ and VAS scores. LME models treated average LZ (or VAS items) as the dependent variable, included fixed effects such as drug condition (LSD vs placebo), presence of stimulus (none/music/video), and eye opening, and modelled subject identity as a random effect; model selection used the Bayesian Information Criterion (BIC). Between-subject Pearson correlations (with false discovery rate correction where reported) were used to probe relationships between regional LZ changes and VAS items. Multivariate regression and a conditional predictive power analysis were also employed: the latter builds a directed network indicating whether the predictive information a brain region provides about a VAS item is mediated by another region.

Across all analyses the administration of LSD produced a clear increase in Lempel-Ziv complexity relative to placebo. In the placebo sessions external stimuli and eye opening themselves modulated LZ: richer stimuli produced higher LZ and eye opening increased LZ, with a Kruskal-Wallis test showing significant differences across conditions and post-hoc tests indicating consistent increases with richer stimuli. In an LME model using stimulus presence and eye opening as predictors (with subject as a random effect), stimulus had a positive effect on LZ (β = 0.013, SE = 0.005, p = 0.017) and eye opening had a larger positive effect (β = 0.025, SE = 0.005, p < 0.001). Adding drug as a fixed effect revealed a substantial further increase in LZ under LSD, larger than those associated with stimulus or eye opening. Crucially, the interaction between drug and eye opening was significant and negative: increased external stimulation (in particular opening the eyes or viewing richer stimuli) reduced the magnitude of the drug-induced LZ increase. The authors interpret this as a form of competition between endogenous, drug-induced neural dynamics and exogenous, stimulus-driven dynamics. This negative interaction was robust to alternative analyses (for example using stricter signal filters) and was spatially widespread across brain regions. Subjective VAS measures were strongly affected by LSD, with the drug producing much larger effects than stimulus or eye opening across the assessed dimensions. Correlations between LSD-induced changes in regional LZ and VAS ratings were present in several conditions but differed markedly by context. In the eyes-closed condition significant positive correlations were observed, for example between ego dissolution and posterior default mode network (pDMN) entropy, positive mood and amygdala entropy, and simple/complex imagery and visual/auditory ROIs. These neural–psychometric correlations persisted in eyes-closed, music, and eyes-open fixation conditions but largely disappeared when participants watched the video: correlations fell to near zero (none exceeding |r| > 0.1). A multivariate regression showed that the interaction of stimulus and eye opening was associated with smaller VAS–LZ correlation values (β = -0.21, SE = 0.08, p = 0.006). Analyses of inter-regional LZ correlations showed that external stimulation increased the correlation between ROIs, which the authors characterise as potential "complexity matching" whereby neural dynamics become more entrained by the stimulus and thus more similar across regions. Regression models of ROI–ROI correlation yielded eye opening associated with smaller values (β = -0.10, SE = 0.04, p = 0.011), while stimulus presence (β = 0.15, SE = 0.04, p < 0.001) and the stimulus-by-eye-open interaction (β = 0.18, SE = 0.05, p = 0.001) predicted larger correlations. Conditional predictive power analyses indicated that lower-level sensory regions tend to mediate the predictive relationships between higher-level areas and subjective reports: once changes in visual and auditory region entropy are known, the entropy changes in pDMN or insula add little extra predictive information about reported complex imagery. Ego dissolution was a notable exception, where pDMN, auditory cortex, and insula each provided complementary, unmediated predictive information about the subjective report.

Mediano and colleagues interpret the findings as quantitative evidence that environmental context substantially shapes both the neural dynamics and subjective phenomenology of the psychedelic state. The data show that while LSD robustly increases neural signal diversity, this effect is attenuated by external stimulation and eye opening, and the close correspondence between entropy and subjective experience is strongest with eyes closed. The authors propose a competition model in which exogenous stimuli can overwrite or mask endogenous, drug-driven dynamics, reducing the capacity to detect relationships between entropy measures and specific aspects of reported experience. In relation to the Entropic Brain Hypothesis, the results qualify a straightforward reading that higher brain entropy uniformly indexes richer conscious content. The authors note that Lempel-Ziv complexity, computed at the level of individual source time series, may preferentially reflect lower-level, sensory aspects of the brain–mind relation rather than higher-order cognitive phenomena. They therefore advocate for complementary approaches that consider network-level entropy and other multidimensional metrics of brain dynamics, while recognising that increasing model complexity can reduce interpretability and statistical power. Implications for psychedelic-assisted therapy are emphasised: the stronger drug–entropy effects and tighter LZ–phenomenology correlations when eyes are closed support therapeutic practices that encourage eyes-closed, introspective states and the selective use of music rather than visual stimulation. The work is presented as a proof-of-concept in healthy volunteers that establishes LZ as a neuroimaging marker sensitive to interactions between drug and context, paving the way for clinical studies. The authors acknowledge limitations and avenues for future research. The sample comprised healthy participants and the design was single blind; three participants were excluded due to incomplete or artefact-laden recordings. Several analyses and controls are reported in supplementary material, and the authors call for replication and extension in clinical cohorts, with other pharmacological agents, dosages, and stimulus modalities to further disentangle the relationships between entropy, phenomenology, and therapeutic outcomes.