Exploring mechanisms of psychedelic action using neuroimaging

This paper is a narrative review that gathers and integrates findings from multiple neuroimaging modalities, principally resting-state and task fMRI, PET molecular imaging, and MEG/EEG electrophysiology, alongside emerging computational and theoretical models (notably hierarchical predictive processing and the REBUS account). The extracted text does not provide a formal literature-search strategy, inclusion/exclusion criteria, date range, or a count of included studies; therefore the review should be understood as a synthesis of representative and influential work rather than as a systematic meta-analysis. The authors consider studies of both acute (on-drug) states and sub-acute/persistent effects following dosing sessions, across healthy volunteers and patient populations. Analyses reviewed include seed-based and parcellation approaches to fMRI connectivity, complexity and entropy metrics, spectral power and event-related paradigms in EEG/MEG, PET receptor-occupancy and receptor-mapping studies, and computational interpretations grounded in hierarchical predictive-processing frameworks. Methodological heterogeneity across molecules, participant populations, routes of administration and image-processing pipelines is noted as a complicating factor for direct comparison and pooling of results.

Resting-state fMRI: Across more than 40 rs-fMRI studies since the first 2012 psilocybin report, a recurring pattern is reduced within-network functional connectivity in high-level cognitive networks (default mode, executive control, dorsal attentional) and in sensory systems (somatomotor, visual). Acute psychedelic states commonly show increased between-network coupling and greater global functional connectivity, interpreted as reduced functional segregation. Analyses of cortical connectivity gradients indicate a partial collapse of the unimodal–transmodal gradient under psychedelics, i.e. increased cross-talk between sensory and higher-order transmodal areas. Complexity measures applied to rs-fMRI signal show increases in the diversity of connectivity states and in entropy-like metrics. Task fMRI: Despite practical challenges of task engagement during acute psychedelic exposure, task studies (largely in healthy participants) report shifts in emotional processing: psilocybin acutely attenuates amygdala responses to emotional stimuli and this attenuation relates to short-term positive mood changes; MDMA augments responses to positive memories while reducing responses to negative ones; ketamine reduces amygdala and hippocampal emotional reactivity. The authors note possible confounds related to task disengagement and recommend within-session controls. Sub-acute and treatment studies: Two independent patient samples treated with psilocybin-assisted psychotherapy (dosing regimens noted in the text as 10 then 25 mg and 25 then 25 mg) showed reductions in brain modularity (a measure of whole-brain functional segregation) at 1 and 21 days post-treatment, and these modularity reductions correlated with symptom improvement. Other post-treatment findings include increased DMN connectivity at short timepoints, transient increases in amygdala reactivity one day after treatment, decreased amygdala cerebral blood flow correlated with clinical improvement, altered vmPFC–amygdala functional connectivity associated with rumination, and increased BOLD responses to music. Sub-acute imaging results are fewer and less consistent than acute findings, but persistent changes in emotional processing and cognitive/neural flexibility up to about four weeks post-treatment are commonly reported. MEG/EEG: Electrophysiological studies consistently find broadband desynchronisation (reduced power across delta–beta) and, most robustly, reductions in alpha-band power (8–13 Hz) following serotonergic psychedelics; alpha reductions are prevented by the 5-HT2A antagonist ketanserin, implicating 5-HT2A receptor involvement. Reports of increased gamma power exist but are noted to be vulnerable to muscle-artifact confounds. Psychedelics increase signal entropy as measured by Lempel–Ziv complexity, in the opposite direction to states of reduced consciousness (for example, NREM sleep or propofol). Event-related paradigms show suppression of mismatch negativity (MMN) under DMT and LSD (psilocybin appears not to influence MMN in the studies cited), and reduced late neural responses consistent with disrupted top-down cognitive processing. One EEG study during DMT reported decreased backward (top-down) traveling waves and increased forward (bottom-up) waves. PET and molecular imaging: PET work has been more limited historically but recent developments include the agonist 5-HT2A ligand [11C]Cimbi-36, which is sensitive to endogenous serotonin release and provides receptor-occupancy estimates with psilocybin; reported subjective effects relate to 5-HT2A occupancy and modelling suggests that so-called microdoses (roughly 10% of a full ~25 mg psilocybin dose) may still yield appreciable receptor occupancy. PET-derived receptor-density maps have been fitted to fMRI data to show that regional 5-HT2A distribution predicts the spatial profile of drug-induced activity changes. Autoradiography in pigs reported higher SV2A (a synaptic-density marker) one week after a single psilocybin dose; human SV2A PET studies are underway. Computational synthesis: The reviewed imaging results support computational accounts framed in hierarchical predictive processing, particularly the REBUS model, which proposes that psychedelics reduce the precision weighting (confidence) of high-level priors, leading to weakened top-down constraints, greater bottom-up influence, increased entropy, and an expanded repertoire of brain states. Empirical anchors for this model include alpha-power reductions, increased signal entropy, diminished backward traveling waves, and dysregulation of high-level networks such as the DMN. Observational links with clinical and personality changes (for example, increased openness) are reported, but many inferences are acknowledged to be indirect or provisional.

Erritzoe and colleagues interpret the convergent multi-modal imaging findings as providing a plausible mechanistic bridge from 5-HT2A receptor pharmacology to large-scale brain dynamics and, ultimately, psychological and therapeutic effects. They argue that psychedelics acutely induce decreased functional segregation, increased global integration and entropy, and dysregulation of high-level networks and top-down signalling—effects that computational models (notably REBUS within hierarchical predictive processing) can coherently link to phenomenological features such as ego dissolution, cognitive flexibility and emotional insight. Persistent post-treatment changes—such as reduced modularity after psilocybin therapy—are tentatively positioned as potential neural correlates of enduring therapeutic benefit, though the authors stress that sub-acute evidence is less abundant and more heterogeneous than acute-state data. Several limitations and uncertainties are acknowledged. Methodological heterogeneity across studies (different molecules, doses, participant samples, routes of administration and diverse analytic pipelines) complicates direct comparison and generalisation. Measurement issues include confounds in electrophysiological gamma-band findings and the indirect nature of some putative links between imaging metrics and computational constructs (for example, treating alpha-power reductions as an index of precision-weighting). The authors also highlight important unanswered clinical questions, most prominently psychedelic interactions with commonly used serotonergic drugs such as SSRIs; they note that drug–drug interactions could occur at peripheral pharmacokinetic, central pharmacokinetic or pharmacodynamic levels and may alter functional and clinical outcomes. In terms of implications, the paper advocates for an integrative, multi-modal research strategy that spans molecular PET measures, functional MRI and electrophysiology, computational modelling and longitudinal clinical assessment. Such a ‘‘molecular–functional–clinical’’ bridge is viewed as essential for translational development, for addressing individual heterogeneity in treatment response (precision psychiatry), and for informing safe and effective therapeutic implementation. The authors call for more temporally refined and entropy-sensitive metrics, independent dataset replication, human PET studies of neuroplasticity markers such as SV2A, and deeper investigation of the conditions under which acute neural changes carry over into lasting clinical benefits.

The authors conclude that modern neuroimaging and computational methods have created a strong translational foundation for understanding psychedelic action, from receptors to large-scale brain dynamics and clinical effects. Despite encouraging convergent findings, they emphasise that many mechanistic and practical questions remain, and that progress will depend on coordinated, multi-modal research that integrates molecular imaging, functional dynamics, computational theory and long-term clinical evaluation to support the safe and effective development of psychedelic therapies.