Effects of the 5-HT2A Agonist Psilocybin on Mismatch Negativity Generation and AX-Continuous Performance Task: Implications for the Neuropharmacology of Cognitive Deficits in Schizophrenia

Neurocognitive impairments are a core feature of schizophrenia and include both higher-order deficits (attention, executive function, working memory) and more basic sensory memory impairments. Umbricht and colleagues describe mismatch negativity (MMN) as an auditory event-related potential (ERP) that indexes echoic sensory memory and context-sensitive preattentive processing, and note that MMN is reduced in most studies of schizophrenia. They contrast MMN with performance on an AX-type continuous performance task (AX-CPT), which taxes use of contextual information and is sensitive to prefrontal dysfunction; schizophrenic patients characteristically make excess BX errors, reflecting failure to use preceding cues to inhibit a prepotent response to a target stimulus. Against this background, previous pharmacological work has implicated NMDA receptor (NMDAR) dysfunction in both MMN and AX-CPT abnormalities: NMDAR antagonists such as ketamine reduce MMN and produce AX-CPT deficits in healthy volunteers that resemble schizophrenia. At the same time, evidence suggests that 5-HT2A receptor dysfunction may contribute to psychotic and cognitive phenomena, because 5-HT2A agonists (for example psilocybin) induce psychotomimetic effects and impair some cognitive functions. The present study therefore set out to test whether psilocybin (a 5-HT2A agonist) affects MMN generation, sensory ERPs and AX-CPT performance in healthy volunteers using the same procedures used previously with ketamine, with the aim of clarifying whether 5-HT2A-related neurotransmission contributes to the deficits observed in schizophrenia and whether shared downstream effects (for example on glutamate or dopamine systems) could account for overlapping cognitive impairments.

The study used a within-subject, randomized and counterbalanced design in which 18 healthy, right-handed volunteers (10 men, 8 women; mean age 25.1 ± 4.3 years) each completed a psilocybin and a placebo session. Subjects were screened with structured clinical interview, physical examination, ECG and laboratory tests; exclusion criteria included past or present Axis I disorders, substance dependence or abuse, family history of Axis I disorders to second-degree relatives, and medical disorders. Intelligence and handedness were assessed; none of the female participants was pregnant. Psilocybin was administered orally at 0.28 mg/kg in 1- and 5-mg capsules prepared by the hospital pharmacy. Subjects were blind to drug order; sessions started around 09:00 after an overnight fast. Baseline ERP recordings were obtained, then psilocybin or placebo was given, and a second ERP recording took place 70 minutes after ingestion. Behavioural ratings (Brief Psychiatric Rating Scale, BPRS) and orientation items were collected at the end of each ERP recording. Participants remained under supervision until effects subsided. Auditory ERPs were elicited by a fixed sequence of 100-ms 1000 Hz standards, interspersed with 100-ms 1500 Hz frequency deviants and 250-ms 1000 Hz duration deviants (SOA 300 ms; 1517 stimuli across four blocks). EEG was recorded from 28 scalp sites with nose reference, sampled at 500 Hz and band-pass filtered at 0.1–100 Hz during acquisition; epochs comprised a 100-ms prestimulus baseline and 500-ms poststimulus interval. After ocular correction and artifact rejection, epochs were averaged and low-pass filtered at 15 Hz. Analyses concentrated on N1 and P2 to standards (at Fz) and MMN to frequency and duration deviants (at FCz); MMN waveforms were computed by subtracting standards from deviants. Due to an insufficient number of sweeps in one subject, MMN analyses included 17 subjects. During auditory stimulation subjects performed a visual AX-CPT presented in four blocks of 280 stimuli each. Letters were shown for 250 ms; subjects pressed a button when X followed A (AX). Sequences were 70% AX, 10% each of BX, AY, BY. Within blocks, 50% of cue-target pairs had short ISI (0.8 s) and 50% long ISI (4 s). One subject ceased task performance during psilocybin; a second had response registration failure, so AX-CPT analyses were based on 16 subjects. Primary dependent measures were N1/P2 amplitudes and latencies, MMN amplitudes and latencies, AX-CPT hit rates, false alarm rates (BX, AY, BY), discrimination index d' and response bias Br. Statistical evaluation used repeated-measures ANOVAs with session (psilocybin vs placebo) and time (baseline vs infusion) as within-subject factors; MMN type and false alarm type were included where appropriate, and paired t-tests were used for post hoc contrasts. Alpha was set at 0.05.

Behavioural effects: Psilocybin produced robust subjective and behavioural changes on the BPRS. The total BPRS score increased from 18.9 ± 1.5 at baseline to 28.1 ± 9.9 during psilocybin (session × time interaction F1,17 = 16.00, p = 0.001). The psychosis factor rose (baseline 4.0 ± 0.0 to 7.5 ± 2.9; F1,17 = 24.79, p < 0.001) and smaller but significant increases were seen on anergia, activation and hostility factors. Orientation (autobiographic, temporal, spatial) was not affected. Auditory ERPs: For sensory ERPs, psilocybin significantly decreased N1 peak amplitude at Fz (session × time interaction F1,17 = 6.03, p < 0.05); post hoc paired t-test showed a smaller N1 during psilocybin versus its baseline (t = −2.09, df = 17, p = 0.05). N1 latency was unchanged. Topographic analysis confirmed a general N1 amplitude reduction across electrode sites (infusion × time interaction F1,17 = 6.14, p = 0.03). Psilocybin did not significantly affect P2 measures as reported. Mismatch negativity (MMN): Robust MMN to both frequency and duration deviants was present at baseline, with maxima at FCz and larger responses to frequency deviants. Although MMN amplitudes decreased slightly during both placebo and psilocybin phases and the decrease was numerically more marked for frequency deviants under psilocybin, repeated-measures ANOVA did not show a significant effect of psilocybin on MMN peak amplitudes (session × time interaction F1,16 = 2.22, p = 0.16; session × time × MMN type interaction F1,16 = 0.14, p = 0.7). Separate analyses showed a trend for greater reduction in the frequency condition (session × time F1,16 = 3.20, p = 0.09) but no effect in the duration condition (F1,16 = 0.36, p = 0.6). MMN peak latencies were unaffected. Note that MMN analyses excluded one subject because of insufficient sweeps (n = 17); sweep counts were slightly reduced during psilocybin for some conditions but remained adequate for signal-to-noise. AX-CPT performance: Psilocybin impaired AX-CPT performance markedly. Hit rates declined at both ISIs (session × time interaction F1,15 = 23.92, p < 0.001), with no significant interaction with ISI. False alarms increased, predominately for BX sequences; a three-way infusion × time × false alarm type interaction was significant (F2,14 = 13.51, p = 0.001). Contrasts showed a significantly greater increase of BX errors than AY (F1,15 = 16.03, p = 0.001) and BY (F1,15 = 28.30, p < 0.001) errors when comparing psilocybin to placebo and baseline. Paired t-tests confirmed higher BX error rates during psilocybin versus baseline for both short ISI (t = 5.95, df = 15, p < 0.001) and long ISI (t = 5.23, df = 15, p < 0.001). Discrimination index d' decreased significantly under psilocybin (session × phase F1,16 = 52.49, p < 0.001); response bias Br decreased non-significantly (F1,16 = 2.34, p = 0.13). AX-CPT analyses were based on 16 subjects due to task non-completion or technical failure in two participants.

Umbricht and colleagues interpret the pattern of findings as evidence that psilocybin—acting as a 5-HT2A agonist—selectively reduces N1 amplitude and impairs context-dependent use of information on the AX-CPT (notably producing increased BX errors), but does not produce the marked MMN reductions previously observed with NMDAR antagonism. The dissociation between effects on N1 and MMN differs from the ketamine profile (ketamine reduced MMN but left N1 intact) while the AX-CPT deficits are similar for both drugs and resemble deficits seen in schizophrenia. From this, the investigators infer that MMN generation depends strongly on NMDAR functioning, whereas 5-HT2A receptor signalling is less important for MMN. Conversely, overlapping AX-CPT impairments after ketamine and psilocybin may arise from shared downstream pharmacological effects on glutamatergic and dopaminergic dynamics rather than from identical primary receptor actions. The discussion highlights mechanistic evidence supporting these claims: both NMDAR antagonists and 5-HT2A agonists can increase dopamine release in humans and induce excessive glutamate release in animals, and group II metabotropic receptor agonists or agents that inhibit glutamate release (for example lamotrigine) can attenuate cognitive effects of NMDAR antagonists. The authors note differences in pharmacological modulation—ketamine-induced deficits can be ameliorated by haloperidol and lamotrigine, whereas psilocybin-induced deficits are reversed by 5-HT2A antagonists but not by D2 blockers—yet both drug classes may converge on abnormal glutamatergic tone as a common denominator. Several limitations and alternative explanations are acknowledged. The authors consider the possibility that the study lacked power to detect a small psilocybin effect on MMN, but argue that subject numbers and the presence of a significant N1 effect make a strong MMN reduction unlikely. They also note that MMN elicited by frequency deviants can be confounded with an N1 to the deviant and that the modest numerical change in frequency MMN may reflect the N1 reduction. Dose comparability to the prior ketamine study is discussed: although behavioural and cognitive effects were similar in magnitude, the processes underlying MMN may be more resistant to disruption by 5-HT2A agonism or might require higher doses. Finally, the authors draw clinical implications cautiously. They suggest that MMN and other early ERP markers may provide indices closer to specific receptor-related abnormalities (notably NMDAR dysfunction), while deficits in tasks like AX-CPT may arise from heterogeneous pathophysiology converging on similar network-level disturbances. This distinction has potential treatment relevance: atypical antipsychotics (with 5-HT2A antagonism) might ameliorate cognitive deficits related to altered glutamatergic dynamics, but would not be expected to correct deficits that are a direct consequence of NMDAR hypofunction such as MMN reduction. The investigators present these conclusions as supportive but circumspect—pointing to independent abnormalities of NMDAR and 5-HT2A systems as contributors to cognitive deficits in schizophrenia, with disrupted glutamatergic signalling a plausible shared mechanism underlying overlapping effects.