Serotonergic psychedelics temporarily modify information transfer in humans

Psychedelic drugs produce profound alterations in perception, selfhood and the experience of reality, and renewed interest in their therapeutic and mechanistic potential has stimulated research into their neural actions. Earlier neuroimaging and electrophysiological work has implicated medial frontal regions (including the anterior cingulate cortex) and parietal/posterior midline regions (including the posterior cingulate and dorsolateral parietal cortex) in the acute effects of serotonergic psychedelics. Radiotracer studies typically report increased frontal metabolism or perfusion after drugs such as mescaline, psilocybin and ayahuasca, while EEG and MEG studies have shown reductions in oscillatory power—often most prominent in parieto-occipital regions. Recent resting-state MRI work has additionally suggested altered coupling between frontal and parietal midline structures, but the directionality of information transfer between anterior and posterior areas during the psychedelic state has remained unclear. Alonso and colleagues set out to test whether psychedelics modify the directionality of information flow across the human brain. Specifically, the investigators assessed directed functional connectivity of EEG oscillations using transfer entropy (TE), a nonlinear informationtheory measure that can indicate causal, directional dependencies between signals. The study tested the hypothesis that a serotonergic psychedelic would perturb neural hierarchies and alter top-down versus bottom-up information transfer between frontal and posterior regions during the acute drug effects.

Ten healthy male volunteers with prior psychedelic experience participated in a double-blind, randomised, balanced crossover experiment. Each participant completed two sessions one week apart, receiving placebo in one session and an oral, freeze-dried ayahuasca formulation in the other; the ayahuasca dose was equivalent to 0.75 mg DMT/kg body weight. Medical screening (anamnesis, ECG, laboratory tests) was performed and written informed consent obtained; ethical approval was granted by the hospital committee and national authorities. EEG was recorded from 19 standard scalp leads (Fp1/2, F3/4, Fz, F7/8, C3/4, Cz, T3/4, T5/6, P3/4, Pz, O1/2) with eyes closed at baseline and at multiple post-dose time points (15, 30, 45 minutes; 1, 1.5, 2, 2.5, 3 and 4 hours). Signals were referenced to averaged mastoids, analog band-pass filtered 0.1–45 Hz, digitised at 250 Hz then resampled to 100 Hz. Artifact reduction used a blind source separation method; a 60-second segment was extracted and zero-phase filtered between 0.5 and 35 Hz using a type-II Chebyshev filter (order 8) for subsequent analyses. Directed connectivity was quantified using transfer entropy (TE), which measures the reduction in uncertainty about the future of one signal when the past of another is taken into account. TE was estimated nonparametrically via equiquantal binning (histogram-based marginal equiquantization), a method the authors considered appropriate given the ~6000 samples in each 60-second segment. For robustness, a reanalysis at the 2-hour time point used Granger causality (GC) implemented with a multivariate GC toolbox; an optimal vector autoregressive model order of 8 was selected via the Akaike Information Criterion, and model orders 4 and 16 were also tested. Subjective effects were tracked with a 100-mm visual analogue scale at the same time points as EEG. At 4 hours post-dose participants completed the Spanish versions of the Hallucinogen Rating Scale (HRS; six subscales: somaesthesia, affect, volition, cognition, perception, intensity) and the Altered States of Consciousness Questionnaire (APZ; three subscales: oceanic boundlessness OSE, dread of ego-dissolution AIA, visionary restructuralization VUS). Blood samples for DMT plasma concentrations were taken at baseline and at 0.5, 1, 1.5, 2, 2.5, 3 and 4 hours; DMT quantification followed previously published methods. Statistical analysis treated the 19*18 = 342 directed electrode pairs individually. Baseline-corrected TE values (post-dose minus pre-dose) for ayahuasca were compared to placebo using two-sided Wilcoxon signed-rank tests at each time-point. To address multiple comparisons, the authors used an omnibus binomial test counting the number of above-threshold connections; connection maps were reported only when the count exceeded a binomial threshold set at 44. Connection maps used colour and line thickness to indicate direction and uncorrected P-value ranges (P < .01, < .05, < .1). Directionality maps summarised incoming (sinks) and outgoing (sources) connections per electrode when the corresponding connection map met the omnibus threshold. HRS and APZ scores were compared with Student's t tests and corrected for multiple comparisons using the false discovery rate (FDR); Pearson correlations explored relationships between TE changes (ayahuasca minus placebo), subjective scores and plasma DMT levels without correction for multiple comparisons.

TE-based analysis revealed time-specific alterations in directed EEG coupling after ayahuasca. Significant changes compared with placebo occurred primarily at 1.5, 2.0 and 2.5 hours post-administration. At 2 hours the dominant effect was widespread increases in TE for many electrode pairs; directionality maps indicated these increases were driven by posterior sources exerting greater influence over more anterior (frontal) locations. In contrast, at 1.5 and 2.5 hours significant decreases in TE were observed, arising from a reduced influence of anterior leads over posterior sites. The pattern of decreased anterior-to-posterior TE and increased posterior-to-anterior TE was present across time-points from 45 minutes to 4 hours, but reached statistical significance only at the above-mentioned times. The authors note that supplementary directionality maps (provided irrespective of statistical threshold) showed a consistent pattern throughout the drug effect window. Subjective measures and plasma DMT concentrations followed a parallel time course to the TE changes, with peak values for each observed between 1.5 and 2.5 hours. Compared with placebo, ayahuasca produced statistically significant increases on all six HRS subscales after FDR correction. On the APZ, the visionary restructuralization (VUS) subscale showed a significant increase, while nonsignificant elevations were reported for OSE and AIA. Correlation analyses focused on mean TE values across the 1.5–2.5 hour window for electrode pairs that differed from placebo. Significant negative correlations were reported between decreases in anterior-to-posterior TE (for example, the FP2→P3 pair) and multiple subjective measures: HRS perception (r = -0.640, P = .046), cognition (r = -0.665, P = .036) and intensity (r = -0.736, P = .015); and APZ subscales OSE (r = -0.705, P = .023), AIA (r = -0.684, P = .029) and VUS (r = -0.678, P = .031). No significant correlations were found for posterior-to-anterior TE increases. When the imbalance between anterior-to-posterior and posterior-to-anterior TE was computed (difference scores for pairs such as (Fp2→P3) minus (Pz→F8) or (Pz→F7)), these difference values correlated with the HRS volition scale [(Fp2→P3)-(Pz→F8); r = -0.666, P = .036; (Fp2→P3)-(Pz→F7); r = -0.654, P = .040], indicating an association between directional imbalance and reported incapacitation. A reanalysis using Granger causality at the 2-hour time point produced qualitatively similar results: significant increases in posterior-to-anterior connections were observed, albeit in smaller numbers than with TE. The count of significant GC connections varied slightly with model order (tested orders 4, 8 and 16), but the overall postero-anterior bias at 2 hours was consistent with the TE findings. In supplementary current density analyses, the authors report broadband reductions in estimated current density within medial posterior regions (precuneus and cuneus) and medial anterior regions (cingulate), consistent with earlier EEG/MEG findings.

Alonso and colleagues interpret their directed connectivity results as evidence that acute ayahuasca transiently disrupts neural hierarchies by reducing top-down (anterior-to-posterior) information transfer while enhancing bottom-up (posterior-to-anterior) transfer. The temporal profile of TE changes paralleled both DMT plasma levels and the time course of subjective effects, supporting a pharmacologically driven modulation of large-scale brain dynamics. Reanalysis with Granger causality produced an analogous pattern at the 2-hour peak, lending some robustness to the TE-based conclusions, although the number of significant GC connections was smaller. These findings are positioned relative to prior imaging and electrophysiological studies: radiotracer work tends to emphasise frontal increases in metabolism or perfusion, while EEG and MEG source analyses frequently show broadband power decreases centred on parietal/posterior midline regions. The authors suggest these seemingly paradoxical observations (reduced oscillatory power yet enhanced metabolic or BOLD activity) can be reconciled by established inverse relationships between rhythms such as alpha and metabolic/BOLD signals. Within this context, the TE results link frontal and posterior involvement by showing altered directionality of coupling between higher-order frontal regions and sensory-selective posterior areas. The investigators discuss theoretical implications using hierarchical Bayesian models of brain function, in which perception arises from the interaction between bottom-up sensory signals and top-down predictions. They propose that a weakening of top-down constraints together with increased excitability in posterior cortical layers—possibly involving deep-layer pyramidal neurons rich in 5-HT2A receptors—would allow endogenous posterior activity to propagate upward and produce the vivid perceptual and associative phenomena typical of psychedelics, especially with eyes closed. The authors also note that individual factors such as mood, personality, expectations and prior experience will modulate these interactions and contribute to variability across and within experiences. Limitations acknowledged by the authors include the pairwise TE approach, which does not account for multivariate influences from all other recorded nodes and therefore may miss or misattribute some dependencies; the GC reanalysis produced fewer significant connections possibly because of signal length and local non-stationarities to which TE estimation via equiquantal binning is relatively robust. The authors further remark that if drug-induced disruptions of hierarchical coupling persisted beyond acute effects, pathological consequences could arise—citing hallucinogen persistent perception disorder as a possible example—though they present this as a speculative concern rather than an empirical finding. Overall, the study team concludes that acute administration of a serotonergic psychedelic temporarily modifies the directionality of information transfer in the human brain, reducing frontal top-down influence and increasing posterior-to-frontal information flow, changes that relate to the intensity and time course of subjective effects.