The NMDA antagonist ketamine and the 5-HT agonist psilocybin produce dissociable effects on structural encoding of emotional face expressions

Schmidt and colleagues frame emotional face recognition as a core component of social interaction that can be modulated by neuromodulatory systems, notably serotonin (5-HT) and glutamate. Prior work has shown that serotonergic manipulation (for example, with the SSRI citalopram) can shift processing towards positive emotional information and alter visual evoked responses, while glutamatergic manipulation with the NMDA antagonist ketamine reduces visual responses to fearful faces in functional imaging. Event-related potentials (ERPs) such as the P100 and the face-sensitive N170 provide time-resolved measures of early visual and structural encoding processes and have been used to probe emotional biases in health and affective disorders. This study set out to compare directly how acute manipulation of the NMDA receptor system (via S-ketamine) versus the 5-HT receptor system (via psilocybin) affects conscious and non-conscious processing of emotional faces. Using behavioural discrimination thresholds and ERP measures (P100 and N170), the investigators hypothesised that both drugs would reduce negative face processing and enhance positive face processing, and that drug effects on ERPs might differ between non-conscious and conscious visual presentation. The comparison was designed to probe receptor-specific contributions to early visual signatures of emotional bias and to provide a framework for studying mechanisms relevant to affective disorders.

Healthy student volunteers were recruited and allocated into two independent groups: an S-ketamine group (N = 21; 12 male; mean age 26 ± 5.39 years) and a psilocybin group (N = 21; 13 male; mean age 23 ± 2.22 years). Screening included medical history, clinical examination, ECG, blood tests and structured psychiatric interviews to exclude personal or first-degree family histories of major psychiatric disorders. Substance use histories were recorded and urine drug screens performed; all participants were medication-free for at least 3 weeks. Ethics approval and regulatory authorisation for psilocybin were obtained and written informed consent was provided. A double-blind, within-subject placebo-controlled design was used separately in each group: each subject attended two sessions (placebo and active drug) in counterbalanced order with at least 2 weeks between sessions. S-ketamine was administered intravenously as a bolus (10 mg over 5 min) followed by a continuous infusion (0.006 mg/kg/min for 80 min) with stepwise dose reductions every 10 min as described; placebo infusion matched the procedure with saline/glucose. Psilocybin was given orally at 115 μg/kg in gelatin capsules; a lactose capsule served as placebo. Behavioural and electrophysiological testing occurred during the known plateaus of drug effects (approximately 25 min after ketamine infusion start and 110 min after psilocybin ingestion). Participants completed the Altered State of Consciousness (ASC) questionnaire after acute effects had subsided (≈240 min post-ketamine; ≈360 min post-psilocybin). Stimuli were greyscale faces from the Ekman–Friesen set, standardised for luminance and contrast and presented centrally on a CRT monitor. The study used backward masking to probe non-conscious versus conscious processing. Two discrimination tasks established sensitivity (d') for fearful vs neutral and happy vs neutral faces at target durations of 20, 30, 50, 90 and 170 ms, each followed by a 150 ms neutral mask; five blocks of 40 trials were run per task. For ERP recording, target faces were presented for 10 ms for non-conscious and 200 ms for conscious processing (each followed by the 150 ms neutral mask); no button-press responses were required during ERP trials. EEG was recorded from 64 scalp electrodes (Biosemi ActiveTwo) at 512 Hz with horizontal and vertical EOG channels. Preprocessing included independent component analysis for ocular artifacts, average re-referencing, epoching with a 200 ms prestimulus baseline and 500 ms post-stimulus window, rejection of epochs exceeding ±100 μV, and band-pass filtering (1–30 Hz). P100 and N170 amplitudes were analysed at right (P08/P8/P10/O2) and left (PO7/P7/P9/O1) posterior electrode clusters; the extracted text reports peak windows but the exact latency windows for computation are not fully clear in the extraction. Behavioural sensitivity was analysed using signal detection theory (d') and Student's t tests against chance (d' = 0) to determine thresholds. Repeated-measures ANOVAs assessed effects of treatment, target duration (non-conscious/conscious), valence and laterality, with additional analyses on placebo–drug change scores and linear regressions relating discrimination changes to N170 change scores. The ASC data were analysed by repeated-measures ANOVA across scales with post hoc Fisher's LSD where ANOVA indicated differences.

Facial discrimination thresholds: Under placebo, d' for fearful versus neutral faces was at chance at 20 ms and rose above chance at 30 ms. Under both S-ketamine and psilocybin, performance at 30 ms remained at chance and only reached significance at 50 ms, indicating impaired early discrimination for fearful faces under both drugs. For happy versus neutral discrimination, d' values across durations were generally above chance in all drug conditions. A repeated-measures ANOVA across groups showed increasing d' with longer target durations. Crucially, S-ketamine reduced d' for both fearful (p < 0.001) and happy (p < 0.001) faces relative to placebo, whereas psilocybin reduced d' only for fearful faces (p < 0.001) and not for happy faces (p = 0.87). P100 ERP: Both groups showed larger P100 amplitudes over right than left electrodes. No treatment effects on P100 amplitude were observed for either psilocybin or S-ketamine, and change-score comparisons found no significant differences between the two drugs on the P100, indicating that early fast extraction of emotion-related visual information (as indexed by P100) was not altered by the pharmacological manipulations. N170 ERP: Under placebo, N170 amplitudes were larger over the right than left hemisphere and larger for emotional than neutral faces. Both S-ketamine and psilocybin produced a general reduction of N170 amplitude that was lateralised to right parieto-occipital electrodes and more pronounced during conscious than non-conscious processing. For S-ketamine, the N170 reduction was observed for fearful, neutral and happy faces (overall treatment effect and a treatment × laterality interaction showing the effect only over the right electrodes). Psilocybin also reduced the right-sided N170 overall, but the treatment × valence interaction showed a selective reduction for fearful and neutral faces (p < 0.000001 and p < 0.01, respectively) and no significant effect for happy faces (p = 0.1). Direct comparison of drug change scores found that S-ketamine and psilocybin produced comparable reductions of the N170 for fearful and neutral faces over right electrodes, but differed for happy faces: S-ketamine induced a significantly larger N170 reduction to happy faces than psilocybin (p < 0.05). Linear regression showed no significant relationships between changes in discrimination performance (d' fear or d' happy) and N170 change scores for the corresponding emotions in either group. ASC questionnaire: Both drugs markedly increased global ASC scores relative to placebo. A treatment × scale × group interaction indicated differential subjective profiles: S-ketamine produced higher scores than psilocybin for disembodiment, auditory alterations, and impaired control and cognition, whereas psilocybin produced greater elementary imagery and visual illusions. Both drugs increased most ASC scales relative to placebo, with some scale-specific exceptions noted in the extracted text.

Schmidt and colleagues interpret three principal findings: first, structural encoding of faces as indexed by the N170 is impaired by both S-ketamine and psilocybin, while the earlier P100 component is unaffected; second, the two drugs show dissociable effects on the N170 with both reducing responses to fearful faces but only S-ketamine reducing responses to happy faces; third, drug-induced N170 changes are more pronounced during conscious than non-conscious presentation. On behaviour, the investigators report that both drugs impaired discrimination of fearful faces, whereas only S-ketamine impaired discrimination of happy faces, suggesting a valence-specific behavioural effect of psilocybin. Electrophysiological findings paralleled behaviour: psilocybin reduced N170 amplitudes to fearful and neutral faces but left happy-face N170s unchanged, whereas S-ketamine produced a more general N170 suppression consistent with an overall emotional blunting. The authors link the N170 generators to the fusiform gyrus (FG) and discuss how altered FG–amygdala and FG–prefrontal interactions could underlie the observed ERP changes, citing prior imaging studies that showed ketamine reduces fusiform and amygdala responses to fearful faces. Differences between psilocybin and serotonergic agents such as citalopram are attributed to distinct pharmacological profiles: psilocybin (via its active metabolite psilocin) is a direct 5-HT receptor agonist with affinity for multiple 5-HT receptor subtypes, whereas SSRIs increase extracellular 5-HT via reuptake inhibition. The authors suggest that receptor-subtype-specific actions may explain why psilocybin, but not acute citalopram, modulated the N170 in this study. They also propose that the stronger drug effects during conscious processing may reflect reduced allocation of attentional resources under both drugs, consistent with prior evidence that ketamine and psilocybin impair attention and reduce N170 amplitudes in other object completion tasks. The authors acknowledge unresolved issues and limitations reported in the extracted text. Notably, the lack of a statistically significant relationship between behavioural discrimination changes and N170 changes is highlighted, and they note that the dissociation for happy-face processing between the two drugs cannot be fully explained by the present data. They also discuss concerns about the non-conscious stimulus duration—some evidence suggested conscious perception of happy faces at very short durations—but argue that prior ERP studies support the non-conscious status of presentation times below 20 ms. Finally, the investigators conclude that early visual ERPs, particularly the N170, provide a sensitive window to detect pharmacologically induced changes in emotional processing biases and offer a framework for probing pathophysiological mechanisms underlying dysfunctional emotional biases.