Psilocybin Induces Aberrant Prediction Error Processing of Tactile Mismatch Responses-A Simultaneous EEG-FMRI Study

This was a double-blind, randomised, placebo-controlled, crossover study. Healthy volunteers aged 20–40 were screened for medical and psychiatric exclusion criteria; the initial sample was 24 participants. Each participant completed two sessions at least two weeks apart and received either oral psilocybin (0.2 mg/kg) or an oral placebo (179 mg mannitol + 1 mg colloidal silicon dioxide). The roving somatosensory oddball task (RSOT) was administered 85 minutes after dosing, during the reported plateau of subjective drug effects. Subjective experience was measured retrospectively with the 5-Dimensional Altered States of Consciousness (5D-ASC) questionnaire at 360 minutes post-dose. The RSOT delivered electrical pulses to the left forearm median nerve at roughly twice the individual perceptual threshold. Trains of stimuli randomly switched between low and high intensity after 3–7 repetitions; the first stimulus of a new train was modelled as the deviant (D) and each third repetition as the standard (S). Each session comprised 320 stimuli, including 69 D and 69 S stimuli, and lasted about 13 minutes. Individual perceptual thresholds were determined before scanning using an adaptive staircase and intensities were adjusted to be perceptible but not painful. Participants were instructed to attend to each stimulus. Simultaneous EEG (64-channel, MR-compatible) and whole-brain fMRI (3T, EPI) were acquired. fMRI preprocessing used standard steps (slice-timing, realignment, normalization to MNI space, 8 mm smoothing) and ArtRepair for movement artefacts. First-level GLMs modelled D and S events convolved with a canonical hemodynamic response; contrast D > S was carried forward to second-level paired t-tests comparing Pla > Psi and Psi > Pla using cluster-level FWE correction (cluster-defining threshold P < 0.001, cluster-level P < 0.05 FWE). EEG preprocessing included gradient and ballistocardiogram removal, ICA for ocular and residual artefacts, band-pass filtering (0.1–30 Hz), re-referencing and automatic artifact rejection. Epochs ranged -100 to +700 ms around D and S events, baseline-corrected to -100–0 ms. For ERP analyses the global field power (GFP) of grand averages defined the MMN time window as 216–414 ms. Mean amplitudes in that window were extracted at frontal electrodes (Fp1, Fp2, AF2) and analysed with repeated-measures ANOVA with factors stimulus type (S, D), electrode, and condition (Pla, Psi). Exploratory correlations were computed between selected 5D-ASC scales ("experience of unity" and "disembodiment") and tactile MMN amplitudes (D minus S) at frontal electrodes, and with the first eigenvariate of fMRI clusters showing significant D > S differences between conditions. Correlations used Pearson's r with a two-tailed significance threshold of P < 0.05.

Sample and tolerability: After excluding six participants for excessive head motion or poor EEG quality and three for stimulation equipment malfunction, the final analysed sample comprised 15 participants (10 men, 5 women; mean age 26.86 years). No substantial side effects were reported; transient sleep disturbance (one participant) and mild transient headaches (three participants) were recorded. Subjective effects: The 5D-ASC showed a strong overall drug effect. Repeated-measures ANOVA revealed a main effect of treatment (F[1,14] = 37.3, P < 0.001), a main effect of scale (F[10,140] = 7.821, P < 0.001), and a condition × scale interaction (F[10,140] = 7.103, P < 0.001). Bonferroni-corrected tests indicated that Psi increased scores on all second-order scales except "spiritual experience" and "anxiety". fMRI: The D > S contrast under placebo (reported in supplementary material) identified typical deviance-sensitive regions. Direct comparison Pla > Psi for D > S revealed significantly reduced BOLD responses under psilocybin in medial prefrontal regions (ventromedial and dorsomedial PFC), primary visual cortex (V1), and the cerebellum. The extracted text does not provide peak coordinates or cluster sizes in full, but these regions are reported as the main loci of reduced deviant-related activation under Psi. EEG and MMN (GFP): GFP analysis defined the 216–414 ms interval for ERP analyses. In the placebo condition GFP showed higher amplitudes to D versus S stimuli. Under Psi this pattern inverted, with S stimuli producing higher GFP than D. Visual inspection suggested an overall increase in GFP in the Psi condition across stimuli. EEG-MMN (frontal electrodes): Repeated-measures ANOVA on mean amplitudes (216–414 ms) at Fp1, Fp2, AF2 yielded a main effect of electrode (F[2,28] = 11.607, P < 0.001) and a trend for a type × condition interaction (F[1,14] = 3.636, P = 0.077). Significant type × condition interactions were observed at Fp2 (F[1,14] = 4.824, P = 0.045) and AF2 (F[1,14] = 5.129, P = 0.040). Simple effects showed that under placebo D versus S differed significantly at Fp1 (t[14] = 2.328, P < 0.035) and AF2 (t[14] = 2.433, P < 0.029), with a trend at Fp2 (t[14] = 2.138, P = 0.051). Under psilocybin there were no significant differences between D and S at these electrodes (all P > 0.28). Analysis of an early sensory window (0–50 ms) showed no treatment effects (all P > 0.05). Correlations with subjective experience: In the Psi condition tactile MMN amplitudes (D minus S) at frontal electrodes correlated positively with subjective alterations. Disembodiment correlated with tactile MMN (r = 0.630, P = 0.012). Experience of unity correlated with tactile MMN at AF2 (r = 0.578, P = 0.024). Disembodiment and experience of unity were themselves positively correlated (r = 0.698, P = 0.004). First eigenvariates from fMRI clusters that showed decreased D > S under Psi did not correlate significantly with disembodiment, experience of unity, or the tactile MMN measures (all P > 0.05).

Duerler and colleagues interpret their findings as evidence that psilocybin alters tactile deviancy processing at both neural and experiential levels. Functionally, psilocybin reduced deviant-related BOLD responses in frontal (ventromedial and dorsomedial prefrontal cortex), visual cortex (V1), and cerebellar regions, and diminished the somatosensory MMN over frontal electrodes. These effects occurred without change in very early sensory components, suggesting altered processing at later stages of PE signalling rather than primary afferent transduction. The authors suggest that frontal reductions reflect disrupted salience and self-related integration processes: the medial prefrontal cortex is implicated in integrating self-relevant information and in schema-related memory, and its downregulation under 5-HT2A stimulation may impair discrimination between surprising and habituated tactile inputs. Reduced V1 activation during tactile deviancy may reflect altered top-down visual contributions to tactile localisation and ownership, consistent with known psilocybin effects on visual processing. Cerebellar involvement is interpreted in light of its role in generating sensory predictions. On the EEG side, the pattern of results points to reduced divergence between responses to deviant and standard stimuli under psilocybin: whereas placebo showed the expected more-negative response to deviants (the MMN), psilocybin abolished this difference. The authors note an apparent overall increase in GFP under psilocybin, particularly to standard stimuli, which they interpret as a reduction in adaptation or memory-trace formation for repeated stimuli and a possible hypersensitivity to incoming inputs. This altered balance could produce aberrant salience processing and make it difficult to update internal models based on PE signals. Framing these observations in predictive coding terms, the investigators propose that 5-HT2A stimulation may increase bottom-up thalamo-cortical excitation and concurrently relax the precision weighting of priors, yielding heightened sensory input and weakened top-down constraints. Such changes would produce a sensory environment in which priors are less able to be updated effectively, thereby perturbing bodily self-representations. The positive correlations between tactile MMN amplitude and subjective scales of disembodiment and experience of unity are offered as supporting a link between altered tactile PE processing and changes in bodily self-experience. The authors acknowledge limitations: the small final sample (n = 15) limits power and generalisability, and somatosensory MMN characteristics can vary with stimulated body part and stimulation parameters. They also note the possibility of increased inter-individual variability in EEG responses under psilocybin and recommend larger, well-powered studies and exploration of other sensory modalities and stimulation protocols to determine the specificity of the reported effects.

The study concludes that psilocybin, acting preferentially at 5-HT2A receptors, disrupts the integration of tactile sensory inputs by altering prediction error processing and potentially reducing memory-trace formation for tactile information. These neurophysiological changes—reduced deviant-related BOLD in frontal, visual and cerebellar regions and diminished somatosensory MMN—are associated with subjective alterations in body and self-experience. The authors propose that these findings implicate serotonergic modulation of multisensory and self-related processing and may be relevant to psychiatric disorders characterised by disturbed bodily self-representation.