Spatiotemporal Brain Dynamics of Emotional Face Processing Modulations Induced by the Serotonin 1A/2A Receptor Agonist Psilocybin

Facial expressions are central to social behaviour, and accurate emotional face processing is disrupted in mood disorders. Previous work has implicated the brain's serotonergic system in these processes: selective serotonin reuptake inhibitors (SSRIs) and receptor-directed manipulations alter behavioural recognition of emotional faces and neural responses in regions such as the amygdala, parahippocampal gyrus and prefrontal cortex. Acute administration of psilocybin, a serotonin 5-HT1A/2A receptor agonist, has been reported to impair aspects of emotional face processing, particularly affecting the fine-grained, structural encoding of facial features occurring around 170 ms after stimulus onset, but the detailed spatiotemporal brain mechanisms underlying these effects remain unclear. Bernasconi and colleagues set out to characterise those mechanisms by applying electrical neuroimaging to visual evoked potentials (VEPs) elicited by neutral, fearful and happy faces under placebo and psilocybin. The study aimed to disentangle whether psilocybin-induced changes reflect alterations in overall response strength versus changes in topography (generator configuration) and to localise those effects with a distributed source model (LAURA). This approach allows separation of amplitude (Global Field Power), topographic (global dissimilarity) and latency effects across time, providing a temporally precise picture of psilocybin's impact on emotional face processing.

Thirty healthy volunteers (16 males; mean age 25 ± 0.6 years; 27 right-handed) without personal or first-degree family histories of major psychiatric disorder participated after screening with clinical interview instruments. Exclusion included drug dependence and positive urine drug tests; a small number of participants reported minimal prior recreational drug use. The study received ethics approval and national authorisation for psilocybin use. A randomised, double-blind, placebo-controlled crossover design was used: each participant completed two sessions at least two weeks apart, receiving 170 μg/kg psilocybin or lactose placebo in identical gelatin capsules. Subjective acute effects were assessed with the Altered State of Consciousness (ASC) questionnaire at 360 minutes post-treatment. The experimental task was an EEG passive-viewing emotional face paradigm with backward masking. Target faces (neutral, fearful, happy) were sourced from the Ekman–Friesen set, with task-irrelevant features removed so that only eyes, eyebrows, nose and mouth remained visible. Targets were followed by a neutral-face mask of the same identity. Targets were presented for either 10 ms (unconscious condition) or 200 ms (conscious condition); the present paper focuses on conscious processing (the unconscious data are to be reported elsewhere). Each session comprised 240 target–mask trials (40 trials per valence per duration), presented in a sound-attenuated, electrically shielded booth. EEG was recorded with a 64-channel Biosemi ActiveTwo system at 512 Hz. Epochs ranged from 100 ms prestimulus to 500 ms poststimulus. Preprocessing included semiautomated artifact rejection (±60 μV), interpolation of contaminated electrodes with 3D splines, average referencing and band-pass filtering (0.18–40 Hz). The baseline was defined as the 100 ms prestimulus period, but no prestimulus baseline correction was applied because preparatory processes might differ across drug conditions and baseline correction can alter topography. Average accepted trial counts per condition were approximately 30–33 trials and did not differ significantly across drug or valence conditions, reducing concern about signal-to-noise confounds. Analyses followed an electrical neuroimaging workflow. First, traditional electrode-wise voltage waveform analyses were run via timewise 2-way ANOVAs (treatment × face valence) to visualise effects. The primary, reference-independent analyses comprised Global Field Power (GFP) to index instantaneous scalp electric field strength, and global dissimilarity (DISS) to test topographic changes; both were assessed time point by time point with a temporal persistence criterion of ∼15 ms to limit false positives. Significant GFP interaction windows were then submitted to distributed source estimation using the LAURA inverse solution with the SMAC lead-field and a 3005-node gray-matter grid. Source-level statistics used 2-way ANOVAs (treatment × face valence) at each node, with significance constrained to P < 0.05 and clusters of at least 21 contiguous nodes (AlphaSim used to determine the spatial threshold).

Subjective effects: Psilocybin robustly increased ASC scores. A treatment × scale interaction (F1,319 = 11.263; P < 0.001) was followed by post hoc tests showing increases on most ASC subscales (all P < 0.001, Bonferroni-corrected), with the exceptions of anxiety and spiritual experience, which did not differ from placebo. Voltage waveform observations: Visual inspection at an occipital electrode suggested psilocybin reduced the N170 component and produced a latency shift at about 200 ms, largely independent of valence. Timewise electrode-level 2-way ANOVAs revealed a significant treatment × face interaction in several windows including 152–183 ms and 193–265 ms. Main effects of treatment and face valence were also observed across multiple poststimulus intervals; these electrode-based analyses were presented mainly for contextualisation, since the authors relied primarily on reference-independent measures for interpretation. Global Field Power (GFP): The key GFP analysis identified two time windows showing a significant treatment × face valence interaction: 168–189 ms (F2,51 = 8.673; P = 0.00051) and 211–242 ms (F2,51 = 7.365; P = 0.0014). Post hoc tests on GFP reported in the extracted text indicated a significant decrease in GFP for neutral and happy faces during 168–189 ms (P < 0.01, Bonferroni-corrected), and a significant decrease only for happy faces during 211–242 ms (P < 0.01, Bonferroni-corrected). The nonparametric DISS statistics did not reveal a significant treatment × face valence interaction across the 0–500 ms epoch; main effects of treatment and of face valence were observed in separate windows. Source estimations (LAURA): Distributed inverse solutions were computed over the two GFP interaction windows. Over 168–189 ms, a significant treatment × face valence interaction emerged in right fusiform gyrus, bilateral parahippocampal gyrus, middle–posterior cingulate cortex, amygdala, right insula, left middle–superior temporal gyrus and right inferior frontal cortex (cluster criterion kE = 21). Over 211–242 ms, significant interactions were located in right lingual and fusiform gyri, middle–inferior occipital gyrus, bilateral hippocampus and right parahippocampal gyrus, middle–inferior temporal gyrus and right precentral gyrus. Cluster-level follow-up: The authors subdivided significant regions into clusters and examined mean source scalar values per cluster. For the first interval (168–189 ms) four clusters (left superior temporal gyrus; bilateral cingulate cortex; left parahippocampal gyrus; right insula/parahippocampal region) showed significant reductions in activity for neutral and fearful faces (P < 0.01, Bonferroni-corrected) but not for happy faces. For the second interval (211–242 ms) clusters in right prefrontal cortex, right occipito-temporal cortex and left superior temporal gyrus showed significant reductions exclusively for happy faces (P < 0.01, Bonferroni-corrected). Note on an apparent inconsistency in the extracted text: the GFP post hoc results for the first interval are reported as decreased GFP for neutral and happy faces, whereas the source-cluster analyses report reductions for neutral and fearful faces. The extraction does not clearly resolve this discrepancy.

Bernasconi and colleagues interpret their findings as evidence that psilocybin produces selective, temporally dissociable modulations of the neuronal correlates of emotional face processing. Two distinct poststimulus windows were emphasised: an early/moderate-latency modulation around 168–189 ms, aligning with the structural encoding or fine-grained categorisation of faces, and a later modulation around 211–242 ms, consistent with subsequent stages of expression decoding. The authors argue these latencies fit an established sequence in which an initial coarse categorisation occurs at about 100 ms and is followed by finer structural encoding roughly 50 ms later. Source localisations implicated both core visual face-processing areas (fusiform, occipito-temporal cortex, lingual gyrus) and limbic regions (amygdala, parahippocampal gyrus, hippocampus), with effects tending to lateralise to the right hemisphere. From these patterns the investigators propose that psilocybin reduces neural responses within this network, and they frame this as compatible with an increase in top-down prefrontal control over limbic regions mediated by 5-HT2A receptor stimulation on cortical pyramidal cells. This mechanistic interpretation is presented cautiously: the authors note it is speculative because the present study did not assess functional connectivity directly. The Discussion positions the results relative to prior serotonergic manipulations. Differences between psilocybin and SSRIs are highlighted: SSRIs have produced mixed effects on happy-face processing and, in some studies, modulations at later latencies (∼250 ms), whereas the current data suggest a receptor-subtype-specific action of psilocybin on earlier structural encoding. The authors also note that faster processing of negative stimuli observed in some event-related potential (ERP) literature is broadly consistent with their temporal pattern, though some ERP studies report the opposite and the literature is heterogeneous. Limitations acknowledged in the extracted text include the absence of direct connectivity measures to test the top-down control hypothesis, and that unconscious-face processing data were not included in this report. The authors also concede pharmacological differences between psilocybin and SSRIs complicate direct comparisons. They conclude by emphasising that this study is the first to map the spatiotemporal brain dynamics of psilocybin-induced modulations of emotional face processing in healthy volunteers and that the pattern of reduced neural responses is consistent with enhanced prefrontal top-down modulation affecting fine-grained facial categorisation.