A complex systems perspective on psychedelic brain action

Earlier neuroimaging work on serotonergic psychedelics (for example LSD, psilocybin, and DMT) has repeatedly reported a broad tendency toward reduced functional differentiation of large-scale brain networks — described here as increased global integration — yet attempts to identify consistent region- or network-specific correlates of the acute psychedelic state have produced heterogeneous and often contradictory results. Girn and colleagues note that conventional resting‑state fMRI analyses, which focus on time-averaged, location-specific functional connectivity (FC) within or between predefined networks, have yielded little spatial convergence across drugs, laboratories, or analytic pipelines. This inconsistency is further complicated by observations that some network changes attributed to psychedelics are not unique to them and by methodological factors such as differing parcellation strategies and individual variability in brain topography. To address this gap, the study proposes a ‘‘meta’’ perspective grounded in complexity science. Rather than seeking a canonical spatial fingerprint, the investigators aim to reinterpret existing findings through a whole-brain, dynamical-systems lens: they argue that psychedelics shift the brain into a distinct dynamical mode that is more flexible, diverse, integrated, and tuned for information sharing — a regime closer to criticality. The paper therefore reviews region- and network-focused fMRI results, presents didactic material on complexity approaches, surveys empirical studies that have applied such approaches to psychedelic datasets, and advances a theoretical framework that foregrounds spatiotemporal trajectories and individual variability as central to understanding psychedelic brain action.

The paper is a narrative, conceptual review rather than a systematic meta-analysis. The authors synthesise prior resting-state fMRI investigations of acute psychedelic effects, methodological commentaries (summarised in boxed text), empirical studies that have applied complex-systems metrics to psychedelic data, and computational modelling work. The extracted text does not clearly report a formal search strategy, inclusion/exclusion criteria, or databases searched, so this should be understood as a theory-led synthesis drawing on illustrative empirical examples rather than a comprehensive systematic review. Analyses and data sources discussed span multiple imaging modalities and drugs — chiefly fMRI but also MEG and EEG in cited empirical work — and include studies of LSD, psilocybin, DMT, and ayahuasca. The authors contrast traditional large-scale network approaches (ICA, seed-based FC, interregional FC, LEiDA) with whole-brain and dynamical approaches from complexity science, such as global brain connectivity (global FC), modularity (graph theory), analyses of phase-locking dynamics, information-theoretic metrics (entropy), measures of criticality (power-law and fractal statistics), metastability indices, and computational whole-brain modelling (e.g., global coupling parameter optimisation). Box 1 and Box 2 in the text outline the limitations of network-based FC methods and provide didactic summaries of dynamical systems and information-theoretic characterisations. Where empirical specifics appear in the extracted text, the authors often report reanalyses of previously collected datasets (for example multiple LSD and psilocybin datasets) and cross-study comparisons; they also note that several key complexity-science analyses to date derive primarily from a limited set of research groups. The review therefore integrates methodological critique, empirical findings, and theoretical modelling to motivate a systems-level framework rather than reporting new primary data.

When reviewed at the level of regions and predefined networks, the psychedelic neuroimaging literature shows a mixed picture. A common broad pattern is reduced functional differentiation, often described as increased global integration, and some within-network decreases in integration have been relatively recurrent for visual, somatomotor, and default mode networks. However, between-network interaction findings have shown negligible consistency across drugs and datasets: one direct comparison of two LSD and one psilocybin dataset that used the same analytic protocol failed to identify any consistent between-network FC changes across all three datasets. Attempts to summarise whole-brain coupling via global FC have also produced divergent spatial results: datasets from Imperial College London reported increased global FC in transmodal association regions (default mode and frontoparietal networks), whereas datasets from the University of Zurich instead found increased global FC in somatomotor and visual regions. Thalamocortical FC has been reported to increase in several studies, but that effect appears nonspecific to psychedelics (also reported with other drugs) and may partly reflect averaging across heterogeneous thalamic subdivisions. In contrast, empirical work applying complexity-science and whole-brain dynamical analyses has revealed a more convergent set of systems-level signatures. Several studies found evidence of reduced modularity and increased prevalence of globally coherent substates during the acute psychedelic state. Computational modelling applied to LSD datasets indicated an increased optimal global coupling parameter when fitting whole-brain FC, consistent with stronger large-scale integration. Analyses of macroscale hierarchical organisation reported a ‘‘flattening’’ of the principal cortical hierarchy under psychedelics, implying increased crosstalk between unimodal and transmodal extremes. Multiple studies also reported increases in indices associated with criticality: improved power‑law fits and higher critical exponents, increased spatial and temporal fractal dimensions, and results from one analysis suggesting dynamics moved closer to a mathematically defined edge-of-chaos. Entropy and metastability measures were among the most consistent findings across modalities and compounds. Increased entropy of neural dynamics appeared across psilocybin, LSD, DMT, and ayahuasca datasets, and these increases were most consistently localised to parietal and occipital cortices. Complementary findings included a larger repertoire of brain states, increased variance in network synchrony, reduced energy barriers for state transitions, and indicators of enhanced long-range correlations and small-world properties. The authors emphasise that these effects are preliminary, that quantitative effect sizes and confidence intervals are not consistently reported in the extracted text, and that many of the reported results require replication across more independent groups and datasets.

Girn and colleagues interpret the heterogeneous regional findings as arising naturally from a fundamentally dynamical, nonergodic perturbation induced by serotonergic psychedelics rather than from a single spatially localised effect. They propose that 5-HT2A receptor agonism in association cortices — receptors that are densely expressed on apical dendrites of layer 5 pyramidal cells — provides a plausible neurobiological mechanism for exerting widespread, idiosyncratic effects on whole-brain dynamics. Under this view, psychedelic administration acts as a perturbation that pushes the brain into a distinct dynamical mode characterised by increased entropy (greater temporal complexity and unpredictability), enhanced metastability (an easeful traversal among many transient states), and tuning closer to criticality (greater sensitivity to inputs and extended dynamic range). These properties are argued to explain key phenomenological features of the psychedelic state, including its strong context dependence (set and setting), inter- and intraindividual variability, reported cognitive looseness or primary-process‑like thinking, and experiences such as ego‑dissolution. The authors position this complexity-based account as complementary to, rather than mutually exclusive with, spatially focused approaches: whole-brain dynamical metrics can capture distributed, temporally evolving features that time-averaged, locationist analyses miss, yet spatial specificity remains important and may be revealed by combining approaches. They also note concrete methodological and empirical limitations highlighted in the review: much of the supportive empirical evidence comes from a limited number of research groups and reanalyses of the same datasets, complexity-science methods require technical expertise and are nontrivial to implement, and system-level analyses risk obscuring meaningful spatially specific effects if used exclusively. The discussion closes by outlining several outstanding research questions raised by the framework, including how different measures of dynamical complexity relate to one another, how to link temporally refined subjective reports to dynamic neural measures, how to connect microscale receptor effects to macroscale dynamics, the influence of individual differences in brain and receptor topography, and the temporal persistence of acute neural effects in relation to therapeutic outcomes. The authors advocate for replication, use of neurophenomenological methods (experience sampling paired with imaging), increased team science collaborations, and movement toward multivariate, temporally sensitive analyses to test and refine their proposal.

In their concluding remarks, the authors argue that the principal effect of serotonergic psychedelics is not a fixed pattern of regional activation but a shift in the way brain regions interact and evolve over time. They contend that conceptualising psychedelic action as a transition to a more flexible, diverse, and information‑propagating dynamical regime can unify previous disparate neuroimaging findings and better account for phenomenology and context sensitivity. Finally, they call for future neuroimaging work to prioritise individualised, spatiotemporal trajectories and dynamic, multivariate methods to build more compelling mechanistic models of psychedelic brain action and to advance understanding of brain function more broadly.