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

As source of sensory information, the body provides a sense of agency and self/non-self-discrimination. The integration of bodily states and sensory inputs with prior beliefs has been linked to the generation of bodily self-consciousness. The ability to detect surprising tactile stimuli is essential for the survival of an organism and for the formation of mental body representations. Despite the relevance for a variety of psychiatric disorders characterized by altered body and self-perception, the neurobiology of these processes is poorly understood. We therefore investigated the effect of psilocybin (Psi), known to induce alterations in self-experience, on tactile mismatch responses by combining pharmacological manipulations with simultaneous electroencephalography-functional magnetic resonance imaging (EEG-fMRI) recording. Psi reduced activity in response to tactile surprising stimuli in frontal regions, the visual cortex, and the cerebellum. Furthermore, Psi reduced tactile mismatch negativity EEG responses at frontal electrodes, associated with alterations of body- and self-experience. This study provides first evidence that Psi alters the integration of tactile sensory inputs through aberrant prediction error processing and highlights the importance of the 5-HT2A system in tactile deviancy processing as well as in the integration of bodily and self-related stimuli. These findings may have important implications for the treatment of psychiatric disorders characterized by aberrant bodily self-awareness.


Introduction
The skin, as the body's largest organ, is our first contact point with the environment and is central to the processing of boundaries and self/non-self-discrimination (Allen et al. 2016;Kahl and Kopp 2018). The body, as a source of sensory information, is considered the starting point of our self-awareness and provides a sense of agency and ownership (Tsakiris 2017). Furthermore, affective and cognitive processes are deeply rooted in the body's interaction with the environment (Wilson 2002;Damasio and Carvalho 2013). However, any disruption in this complex system of multisensory processing and integration of sensory signals has an effect on our bodily self-awareness (Tsakiris 2017). Altered bodily self-perception is a core symptom of many psychiatric disorders, such as schizophrenia (Sakson-Obada et al. 2018), depression (Fuchs and Schlimme 2009), or anorexia nervosa (Gadsby 2017).
The ability to react to novel or surprising environmental stimuli is essential for survival as well as for the mental and physical health of an organism (Riva 2018). In general, surprising stimuli imply higher motivational importance. At the same time, our prior expectations affect our subjective perception (Clark 2013). Being able to detect and discriminate surprising stimuli from habituated ones and to adapt by forming new memory traces or updating mental representations fulfill an important role in the maintenance of a homeostatic level (Craig 2009;Damasio 2012;Riva 2018).
The predictive coding account offers a framework for understanding processes underlying the bodily self and their importance in psychiatric disorders (Friston 2005;Seth 2013;Owens et al. 2018;Allen et al. 2019;Allen 2020;Allen et al. 2020). The brain learns to model and predict incoming sensory input to minimize surprise across different body representations. Discrepancy between the predicted and the actual incoming bottom-up content produces a predictions error (PE) signal. Subsequently, this PE is minimized by updating the mental model (Friston 2005;Seth 2013;Tsakiris 2017). It is suggested that the integration of bodily states and sensory inputs with prior beliefs underlies the generation of self-awareness (Lenggenhager et al. 2007;Seth 2014;Tsakiris 2017).
The mismatch negativity (MMN) is an event-related brain potential (ERP) that provides an index for the neural processes underlying the initial response to unpredicted stimuli (Wacongne et al. 2012) and has been linked to perceptual learning and neuroplasticity (Garrido et al. 2009). The MMN can be elicited by a novel stimulus called the "deviant" after presentation of repeated habituated stimuli called the "standards" (Näätänen et al. 2014). The occurrence of an MMN is independent of the level of attention towards the stimuli and has been reported across different sensory modalities (Kekoni et al. 1997;Pazo-Alvarez et al. 2003;Näätänen et al. 2007;Allen et al. 2016). Clinical studies have linked a reduced MMN amplitude to aberrant perceptual learning, for example, in patients with alterations of sensory information processing such as schizophrenia (Baldeweg et al. 2004;Umbricht and Krljes 2005). Interestingly, these disorders are also characterized by disturbances in body image and self-experiences (Sakson-Obada et al. 2018).
Serotonergic psychedelics, such as psilocybin (Psi) or lysergic acid diethylamide (LSD) exert their psychological effects primarily via 5-HT 2A receptor activation (Vollenweider et al. 1998;Halberstadt and Geyer 2011;Preller et al. 2017) and are valuable tools to study brain mechanisms of consciousness, cognition, and emotion. Furthermore, recent results indicate the therapeutic potential of psychedelic-assisted therapy as effective treatment option for various psychiatric disorders (Grob et al. 2011;Bogenschutz et al. 2015;Rucker et al. 2016;Bogenschutz and Ross 2018;Carhart-Harris et al. 2018). Psi produces a dosedependent altered state of consciousness and induces transient and reversible alterations in body and self-perception which are closely linked to each other (Vollenweider and Kometer 2010;Studerus et al. 2011;Preller and Vollenweider 2018).
Recent studies have investigated the effects of serotonergic psychedelics on the MMN in the auditory domain. However, results have been inconsistent (Umbricht et al. 2002(Umbricht et al. , 2003Schmidt et al. 2012;Bravermanová et al. 2018).
The impact of psychedelics on the processing of tactile mismatch responses has not been investigated so far. Given that Psi induces alterations in self/body boundaries, feelings of oneness, and disembodiment, previously associated with changes in frontal glucose metabolism (Vollenweider 1997), investigating tactile deviancy processing after the administration of Psi offers the unique opportunity to gain valuable insights into the neurobiological processes that give rise to the formation of bodily awareness and self-experience. Leveraging simultaneous electroencephalography-functional magnetic resonance imaging (EEG-fMRI) data acquisition furthermore allows us to investigate neuroanatomical substrates as well as computational mechanisms underlying these processes.
This study therefore investigated the impact of Psi on the processing of tactile mismatch responses induced by a tactile oddball paradigm during simultaneous EEG-fMRI measurement (Allen et al. 2016). We hypothesized that Psi compared with placebo (Pla) 1) induces changes in the blood oxygen leveldependent (BOLD) signal in brain regions previously found to be involved in tactile deviancy processing (Allen et al. 2016;Fardo et al., 2017;Ostwald et al., 2012), 2) reduces the EEG-MMN amplitude, and that 3) these changes are correlated with subjective alterations in self and body experience. Collectively, this pharmacological neuroimaging study demonstrates that Psi-induced alterations in body and self-perception are related to changes in the neural response to surprising tactile versus habituated stimuli in particular in frontal brain regions, indicating alterations in the integration of tactile sensory inputs through aberrant PE processing and potentially reduced memory trace formation of tactile information.

Participants
Participants were recruited through advertisements placed at local universities. Before admission to the study, participants underwent a screening visit. All included subjects were aged between 20 and 40 years and healthy according to medical history, physical examination, blood analysis, and electrocardiography. The Mini-International Neuropsychiatric Interview (M.I.N.I.; Sheehan et al. 1998), the DSM-IV self-rating questionnaire for Axis-II personality disorders (SCID-II; Fydrich et al. 1997), and the Hopkins Symptom Checklist (SCL-90-R; Franke 1995)w e r e used to exclude subjects with present or previous psychiatric disorders or a history of major psychiatric disorders in firstdegree relatives. Participants were asked to abstain from the use of any prescription or illicit drugs for a minimum of 2 weeks prior to the first test day and for the duration of the entire study, and to abstain from drinking alcohol for at least 24-h prior to test days. To verify the absence of drug and alcohol use, urine tests and a self-report questionnaire were used at the beginning of each test day. Urine tests were also used to exclude pregnancy. Furthermore, participants were required to abstain from drinking caffeine during the test day and to abstain from smoking for at least 60 min before MRI assessment. Further exclusion criteria included history of head injury or of neurological disorders, cardiovascular disease, history of alcohol or drug dependence, left-handedness, poor knowledge of the German language, any exclusion criteria for MRI studies (including claustrophobia), and previous significant adverse reactions to a hallucinogenic drug.
The initial sample consisted of 24 healthy participants. To ensure interpretability of the data, participants with excessive head movement in the fMRI (>3 mm in any direction) or poor EEG data quality (<50% clean segments in the ERP analysis) were excluded from data analysis. Six participants were excluded due to poor EEG or fMRI data quality in at least one of the sessions. Additionally, 3 participants were excluded because of malfunctioning equipment for delivering the electrical stimulation. Therefore, the final sample consisted of 15 participants (n = 10 men and n = 5 women; mean age = 26.86 years).
Before participating, all participants provided written informed consent after having received detailed written and oral descriptions of the study procedures, as well as details regarding the effects and possible risks of Psi administration in accordance with the Declaration of Helsinki. The study was approved by the Cantonal Ethics Committee of Zurich (KEK), and the Swiss Federal office of Public Health (BAG) authorized the use of Psi in humans. The study was registered at clinicaltrials.gov (NCT03736980). No substantial side effects were recorded during the study. One participant reported transient sleep disturbances for 1 night and 3 participants reported mild transient headaches after drug administration. No further side effects were recorded.

Study Design and Procedure
This study employed a double-blind, randomized, placebocontrolled, and crossover design. At 2 different occasions at minimum 2 weeks apart, each participant received either: 1) Placebo (179-mg mannitol and colloidal silicon dioxide [Aerosil; Evonik Resource Efficiency GmbH] 1-mg orally; Pla condition) 2) Psilocybin (0.2-mg/kg body weight, orally; Psi condition) The roving somatosensory oddball task (RSOT) was conducted 85 min after Psi/Pla administration during the plateau of peak subjective Psi effects. Subjective drug effects were assessed using the 5-Dimensional Altered States of Consciousness Rating Scale (5D-ASC; Dittrich 1998;Studerus et al. 2010) 360 min after each drug treatment to retrospectively assess the subjective experience after drug intake.

Roving Somatosensory Oddball Task
Stimuli of the RSOT consisted of somatosensory electrical stimulation (50-ms pulse duration) on the median nerve of the left forearm at about twice the individual perceptual threshold. To induce tactile mismatch responses, trains of stimuli switched randomly between high and low intensity after a variable number of 3-7 repetitions (Allen et al. 2016). Low intensity trains consisted of single pulses separated by 2000-ms intervals. High intensity trains consisted of 2 pulses delivered in a rapid sequence (100-ms stimulus onset asynchrony) followed by 2000ms interstimulus intervals.
The first stimulus of each new train was modeled as the "deviant (D)" and each third repetition in a train as "standard" (S). For the high intensity condition, the S stimulus was modeled as the onset of the second pulse of the third repetition. Trains of stimuli varying from 3 to 7 repetitions were uniform randomly sampled to generate an unpredictable stimulus sequence. Two test versions (A and B) were developed and administered in a counter-balanced randomized order to the subject on the 2 experimental days. Participants received a total of 320 stimuli in each session of which 69 stimuli were D and 69 stimuli were S. The duration of the task was ∼13 min. All stimuli were delivered using a MR-safe electrode and a constant current stimulator (Digitimer, 7SHVA; for an overview of the experimental setup see Supplementary Methods, Fig. S1).
The individual perceptual threshold was determined immediately prior to scanning in each drug condition using an adaptive staircase procedure (adapted from Allen et al. 2016, Supplementary Methods, Table S2). The staircase procedure consisted of a one-up/3 down procedure.
Step size was reduced every 2 reversals until reaching the individual threshold. After the individual threshold was reached, the intensity was doubled and then reduced until participants did not perceive it as uncomfortable. The participants reported the sensation as a mild "pinching," but not painful. Thresholds and intensities are reported in Supplementary Table S2. The mean of the participant's individual perceptual threshold differed significantly between Pla and Psi conditions (P = 0.017), however, there was no significant difference for the mean of the participant's final intensities between treatments (P > 0.1). Participants were instructed to pay attention (Allen 2020) to each single stimulus. After the thresholding procedure and a short practice version of the oddball task, the main experiment started after each participant confirmed that they had fully understood the task.

FMRI Data Acquisition and Preprocessing
Magnetic resonance data were acquired on a Philips Achieva 3.0T whole-body scanner. A 32-channel receive head coil and MultiTransmit parallel radio frequency transmission was used. Images were acquired using a whole-brain gradient-echo planar imaging (EPI) sequence (repetition time = 2430 ms; echo time = 27 ms; slice thickness = 3 mm; 45 axial slices; no slice gap; field of view, 240 × 240 mm 2 ; in-plane resolution, 3 × 3 mm; and sensitivity-encoding reduction factor, 2.0). Additionally, highresolution anatomical images (voxel size, 0.7 × 0.7 × 0.7 mm 3 ) were acquired using a standard T1-weighted 3D magnetization prepared rapid-acquisition with gradient echo sequence. Images were analyzed using SPM12 (www.fil.ion.ucl.ac.uk). Preprocessing consisted of slice time correction, realignment, and spatial normalization to the standard EPI template of the Montreal Neurological Institute (MNI), and spatial smoothing using a Gaussian kernel of 8-mm full-width at half-maximum to meet the statistical requirements of the general linear model (GLM). For the detection and repair of artifacts due to movement during scanning the ArtRepair toolbox was used (http://cibsr.sta nford.edu/tools/human-brain-project/artrepair-software.html).

FMRI Data Analysis
fMRI images were analyzed using a GLM implemented in SPM12. To identify BOLD responses to tactile surprising stimuli represented by the deviants we applied a standard summary statistic approach. At the first level, we modeled deviants (D, first Downloaded from https://academic.oup.com/cercor/advance-article/doi/10.1093/cercor/bhab202/6319929 by University of Zurich user on 08 September 2021 stimuli of each new train) and standards (S, the third repetition following each D) as separate event-related regressors convolved with the canonical hemodynamic response function. The remaining repetition trials (S2 und S4-S7) were not modeled, that is, they were left as "implicit baseline." For a second analysis modeling the final stimulus of each train as standard, see Supplementary Fig. S2. The contrast D > S was computed for each participant.
To identify brain regions sensitive to deviancy processing the contrast D > S was entered into a second-level random-effects group analysis using a paired t-test for the comparison between drug treatment conditions (Pla > Psi and Psi > Pla) with a threshold of P < 0.05 cluster level family-wise-error (FWE) corrected with a cluster-defining primary threshold of P < 0.001 to meet the requirements of random field theory. All brain coordinates are reported in the MNI atlas space.

EEG Aquisition and Preprocessing
Simultaneous EEG-fMRI was recorded using an MR-compatible EEG system (64 Channels BrainAmp MR Plus; Brain products GmbH). The Fz electrode served as recording reference, the AFz as ground and 2 electrocardiogram (ECG) electrodes for the cardioballistogram correction (CBC). The EEG signal of all electrodes was recorded with a sampling rate of 5000 Hz (DC). Data were lowpass filtered with a cut-off of 250 Hz for scalp electrodes and 1000 Hz for ECG channels. Impedances were kept below 30 kΩ. To minimize gradient residuals occurring during simultaneous EEG-fMRI recordings the EEG system was synchronized to the scanner clock (Philips Achieva 3.0T; Mandelkow et al. 2006).
Data were analyzed by using Brain Vision Analyzer 2.1 software (Brain Products GmbH). Preprocessing consisted of the following steps: MR gradient artifact removal using implemented sliding average subtraction (Allen et al., 2000), visual inspection and manual exclusion of periods with major artifacts, topographic interpolation, ballistocardiogram correction, ocular, and residual ballistocardiogram artifacts were removed using independent component analysis (ICA; Bell and Sejnowski 1997), rereferencing to the average reference (Lehmann and Skrandies 1980), band-pass filtering between 0.1 and 30 Hz (notch filter 50 Hz), and automatic artifact removal of artifacts exceeding ±100 μV.

EEG-ERP Analysis
Stimulus locked EEG segments were created based on the marker position of the D and S stimuli types (epochs from −100 ms prestimulus to +700 ms poststimulus, averaging typewise) per condition. After artifact rejection, at least 63 S stimuli (mean S = 63.53, i.e., >92%) and 64 D stimuli (mean D = 64.01, i.e., >92%) were available for the Pla condition. For the Psi condition at least 53 S stimuli (mean S = 53.7, >78%) and 55 D stimuli (mean D = 55.3, >80%) were available after artifact rejection. The EEG segments were baseline corrected using the −100 to 0-ms prestimulus interval as baseline.
The time interval for the ERP analysis was defined based on inspection of the global field power (GFP), a measure of global field strength (Lehmann and Skrandies 1980), computed over the grand average of D and S stimuli types for each condition. Visual inspection of the highest GPF mean amplitudes of the grand average for each drug condition and D and S stimuli types defined the interval 216-414 ms as time window for the ERP analyses (see Results section, Fig. 2).
The mean amplitudes of this time window (216-414 ms) were calculated for each condition (Pla and Psi) and stimulus type (S and D) for the frontal electrodes (Fp1, Fp2, and AF2). The selection of the electrodes was based on the visual inspection of the topographical maps of activity for standard and deviant stimuli during this time interval defined by the GFP and the literature based somatosensory MMN electrodes clusters (Strömmer et al. 2014). A repeated measures analyses of variance (ANOVA) was performed to compare mean amplitudes between stimulus type (S and D), electrodes (Fp1, Fp2, and AF2), and condition (Pla and Psi) as within subject factors. Analyses were conducted using Brain Vision Analyzer 2.1 and IBM SPSS Statistics 23 software (IBM).

Subjective Drug Effects: 5D-ASC Questionnaire
The 5D-ASC is a standardized questionnaire that comprises 94 items that are answered on visual analogue scales. Scores were calculated for 11 validated second order scales (Studerus et al. 2010): experience of unity, spiritual experience, blissful state, insightfulness, disembodiment, impaired control and cognition, anxiety, complex imagery, elementary imagery, audiovisual synesthesia, and changed meaning of perception. The 5D-ASC second order scales were analyzed using a repeatedmeasures ANOVA with condition (Pla and Psi) and scale as within-subject factors.

Correlations between Subjective Drug Effects, fMRI, and EEG Effects
An exploratory analysis was conducted to investigate the relationship between subjective drug-induced alterations in body perception and the EEG and fMRI responses to the Psi-induced changes in the processing of tactile surprising stimuli. We therefore correlated the 5D-ASC scores "experience of unity" and "disembodiment" in the Psi condition with the tactile MMN response (i.e., the subtraction of the mean amplitude in response to the S stimulus from the amplitude of the D stimulus) at the frontal electrodes (Fp1, Fp2, and AF2) averaged for the time window 216-414 ms in the Psi condition, and the first eigenvariate of clusters showing a significant difference of D > S in the Pla condition compared with Psi in the fMRI data. Analyses were conducted by using IBM SPSS Statistics 21 Software (IBM) and carried out with a significance level of P < 0.05 (2-tailed).

Psi Changes GFP Mean Amplitude in the Time Window 216-414 ms
We computed the GFP over the grand average per condition and stimuli type to define the time interval for the ERP analysis (see below). Comparison of GFP mean amplitudes between condition and stimuli types in the time window 216-414 ms showed higher amplitudes for D compared with S stimuli in the Pla condition. In the Psi condition the opposite pattern appeared, with S stimuli showing higher GFP mean amplitudes than D (Fig. 2). Visual inspection of the mean amplitudes between the conditions points towards higher overall GFP mean amplitudes in the Psi compared with Pla condition for both stimuli types.

Psi Induces Changes in the Somatosensory EEG-MMN Response
The ERP analysis was based on the time window (216-414 ms) as identified above. The selection of the electrodes was based on visual inspection of the topographical maps of activity for standard and deviant stimuli during this time interval (216-414 ms) defined by the GFP as well as on previous studies investigating somatosensory MMN (Strömmer et al. 2014). A repeated measures ANOVA (electrode * condition * type) of the mean amplitudes of the frontal electrodes (Fp1, Fp2, and AF2) during the time interval 216-414 ms revealed a significant main effect for electrodes (F[2, 28] = 11.607, P < 0.001) and a trend for the interaction of type * condition (F[1, 14] = 3.636, P = 0.077). Furthermore, there was a significant interaction for type * condition for the electrodes Fp2 (F[1, 14] = 4.824, P = 0.045) and AF2 (F[1, 14] = 5.129, P = 0.040). Simple main effect analyses revealed a significant difference between S and D in the Pla condition for Fp1 (t[14] = 2.328, P < 0.035) and AF2 (t[14] = 2.433, P < 0.029) and a trend for Fp2 (t[14] = 2.138, P = 0.051). There was no significant difference in the Psi condition between S and D (P > 0.28) at these electrodes (Fp1, Fp2, and AF2; see Fig. 3). To investigate the influence of Psi on early sensory components, the mean amplitude during the time interval 0-50 ms was analyzed analogously. This analysis did not reveal any significant main effects or interactions (all P > 0.05).

Correlations between Subjective Alterations in Body Perception and Tactile MMN Responses
To investigate associations between Psi-induced subjective alterations in body and self-perception and tactile MMN responses, we correlated the 5D-ASC scores "experience of unity" and "disembodiment" in the Psi condition with the tactile MMN response (i.e., the subtraction of the mean amplitude in response to the S stimulus from the amplitude of the D stimulus) at the frontal electrodes (Fp1, Fp2, and AF2) for the time window 216-414 ms in the Psi condition, and the first eigenvariates of clusters showing a significant difference of D > S in the Pla condition compared with Psi in the fMRI data. A significant positive Pearson correlation was found between the 5D-ASC scale "disembodiment" and tactile MMN responses in the Psi condition (r = 0.630, P = 0.012, Fig. 4A.). Furthermore, we found a positive relationship between the 5D-ASC scale "experience of unity" and tactile MMN responses at AF2 in the Psi condition (r = 0.578, P = 0.024, Fig. 4B.). Both 5D-ASC scales "disembodiment" and "experience of unity" were positively correlated (r = 0.698, P = 0.004). No significant correlations were observed for other electrodes (all P > 0.05). Furthermore, the first eigenvariates of clusters showing a significant decrease of BOLD signal in the Psi condition compared with Pla for the contrast D > S ( Fig. 1D) did not correlate significantly with Psi-induced disembodiment, experience of unity, or tactile MMN responses at frontal electrodes in the Psi condition (all P > 0.05).

Discussion
This study provides first evidence that stimulation of the serotonin (5-HT) receptor system with Psi alters the processing of tactile mismatch responses by combining pharmacological manipulation with simultaneous EEG-fMRI recording. For this our results showcase the advantage of combining fMRI for the spatial resolution with the temporal resolution of the EEG. Our results show that Psi compared with Pla 1) decreases the BOLD signal in response to surprising tactile stimuli versus habituated stimuli in brain regions previously found to be involved in tactile deviancy processing (Allen et al. 2016), 2) reduces the EEG-MMN amplitude, and 3) produces robust perceptual alterations of bodily awareness and self-experience, which are associated with tactile MMN responses at the frontal AF2 electrode in the Psi condition.

Psi Reduces the BOLD Signal in Frontal and Visual Areas in Response Tactile Surprising Stimuli
Psi significantly reduced the BOLD signal in response to tactile surprising versus habituated stimuli in the vMPFC, dMPFC, V1, and the cerebellum. Therefore, our fMRI data reveal that Psi alters deviancy processing and points towards an important role of the serotonin system in perceptual tactile processing and the ability to discriminate tactile deviant stimuli from habituated ones. Our results are in line with previous studies showing that the medial prefrontal cortex (MPFC) is involved in salience processing (Seeley et al. 2007) and represents a key region for the integration of self-related information (Schmitz and Johnson 2007). Stimulation of the 5-HT 2A receptor has been shown to induce alterations in self/other boundaries (Vollenweider et al. 1998;Kometer et al. 2012;Quednow et al. 2012)a n d self-relevance processing associated with altered activity of the MPFC (Preller et al. 2017;Preller and Vollenweider 2018). Furthermore, the MPFC is crucial in the construction and maintenance of a coherent self (Vollenweider 2001). Our results showing that Psi alters activity primarily in frontal brain areas but not somatosensory brain regions during tactile deviancy processing are also in line with recent formulations of the Global Neuronal Workspace Theory suggesting that higher-level areas such as the PFC play a key role for global broadcasting of information and amplifying and sustaining relevant stimuli (Liu et al. 2019;Mashour et al. 2020;Whyte and Smith 2021).
Activation of V1 during deviancy processing in the Pla compared with Psi condition may reflect the impact of visual processing on tactile perception (Kuehn and Pleger 2018). Vision can exert a top-down influence on the integration and localization of tactile information. Previous studies showed that the ability to discriminate tactile stimuli is enhanced when the body part is viewed. Furthermore, other studies using the rubber hand illusion demonstrated that visual illusions can affect the experience of body ownership (Welch and Warren 1980;Botvinick and Cohen 1998;Press et al. 2008;Tsakiris 2017). The visual top-down influence on tactile information processing may be altered in the Psi condition where we found reduced activation in the V1 compared with Pla in the D > S contrast. Psi has been shown to induce aberrant visual sensory integration and to alter the visual perceptual experience, potentially due to an increase in internaldriven excitability of the visual network via serotonergic modulation (Kometer et al. 2013;Preller and Vollenweider 2018). Furthermore, Psi-induced visual illusions have been linked to the emotion system and changes in the meaning of percepts (Preller and Vollenweider 2018). Psi-induced visual experiences and aberrant visual integration may therefore change top-down processing of tactile sensory information as well as bodily selfawareness (Tsakiris 2017). Additionally, the cerebellum has been proposed to play a critical role in generating predictions concerning upcoming sensory information (Courchesne and Allen 1997) and is also involved in perceptual and cognitive processes (Schmahmann 1997).

Psi Induces a Reduction of Tactile MMN Responses Associated with Subjective Alterations in Body Perception
In the Pla condition, we found the expected MMN response in line with previous findings on the somatosensory MMN, that is, negativity after D compared with S stimuli over frontal electrodes (Strömmer et al. 2014). The mean amplitudes in the Pla condition differed significantly between D and S stimuli, showing more negativity in response to D at frontal electrodesa result that is in line with significantly increased BOLD signal in frontal brain regions in the D > S contrast in the Pla condition. In the Psi condition, however, mean amplitudes at frontal electrodes between D and S did not differ significantly. Psi therefore reduced MMN responses to tactile surprising stimuli compared with Pla. This is also in line with our fMRI results showing that Psi decreased the differential activation of frontal brain regions in response to D versus S stimuli. Furthermore, comparison of GFP mean amplitudes between conditions indicates higher mean amplitudes for S stimuli in the Psi compared with the Pla condition, whereas within the Pla condition D stimuli revealed higher mean amplitudes compared with S stimuli. Additionally, visual inspection of the GFP mean amplitudes between the conditions points towards an increase in overall GFP activity in response to all stimuli in the Psi condition indicating a heightened sensitivity in response to all stimuli regardless of habituation. This suggests a reduced adaptation mechanism that could be caused by difficulties in forming new memory traces, potentially due to a hypersensitivity to all incoming inputs. This may lead to aberrant salience processing making it difficult to discriminate between D and S stimuli. This finding is in line with another recent study showing increased neural response to S stimuli and less divergence between S an D stimuli in an auditory oddball task under LSD (Timmermann et al. 2018). Furthermore, changes in subjective alterations in disembodiment and experience of unity were positively correlated with the tactile MMN amplitude at the frontal AF2 electrode in the Psi condition. This indicates an association between altered self and body perception with changes in the negativity response after a tactile mismatch stimuli corroborating the hypothesis that tactile sensory processing may underlie bodily self-perception (Tsakiris 2017).
Contrary to previous studies reporting no significant reduction of the auditory MMN amplitude after Psi administration (Umbricht et al. 2002;Umbricht et al. 2003;Schmidt et al. 2012;Bravermanová et al. 2018), we found a significant reduction of tactile MMN responses under Psi but no evidence of an impact on early sensory components before the expected MMN. Tactile deviancy processing is potentially more directly related to the sensory integration of bodily and self-related stimuli in the body's multisensory system to construct our sense of self (Tsakiris 2017). Stimulation of the 5-HT 2A receptor therefore seems to play an important role in the disruption of the integration of self-related stimuli and interferes with the formation of a coherent self-experience.

Psi Induces Aberrant Prediction Error Processing
Adaptation of bodily representations is a constantly ongoing process during the processing of sensory inputs. These representations remain plastic and are constantly shaped through the integration of our experiences with our expectations (Apps and Tsakiris 2014). In terms of predictive coding these representations and their predictions depend on top-down prior expectations that are constantly updated based on PE signals that are produced by unexpected sensory information (Friston 2005).
Psi induced effects on the bodily self-experience can be explained in terms of predictive coding (Friston 2005;Apps and Tsakiris 2014), specifically its effects on bottom-up and topdown processing. Psychedelics have been suggested to alter bottom-up processing via increased thalamo-cortical connectivity . Increased excitatory connections from the thalamus following 5-HT 2A stimulation could lead to a sensory overload resulting in a heightened bottom-up "surprise" signal (Preller et al. 2019). This sensory overload in the cortex affects top-down processing and may led to a break-down of sensory integration (Vollenweider 2001). Top-down predictions and the updating of internal models may not be possible as the incoming information is not predictable. It has also been suggested that the brain may relax the precision weighting of prior beliefs in the psychedelic state while the bottom-up flow of sensory information is increased (Carhart-Harris and Friston 2019). A previous auditory oddball study (Timmermann et al. 2018) found that the presentation of deviant tones elicits an increase in intrinsic connectivity which represents the strength of memory formation due to discrepancy between predicted and actual sensory input. After administration of LSD, this intrinsic connectivity was reduced. In line with this, our study showed less divergence between the D and S stimuli responses in the Psi condition potentially resulting from reduced adaptation and maybe aberrant salience processing. Furthermore, aberrant salience processing and alterations in matching incoming tactile stimuli with the sensory memory under Psi could affect schemarelated learning in the vMPFC, which has been proposed to be a critical node for schema memory (Gilboa and Marlatte 2017). Future studies investigating different sensory modalities are needed to determine if psychedelics specifically impact tactile processing, or if the effects reported here represent a more generally altered mechanism of saliency detection, adaption, and learning.

Limitations
A limitation of the study is the small sample size. Further studies with larger sample sizes are needed to extend our knowledge about the serotonergic neurochemical mechanisms that underlie tactile deviancy processing as well as its association with bodily awareness and self-experience. Furthermore, it needs to be noted that the somatosensory MMN amplitude as well as the time window of its occurrence can vary depending on body parts stimulated and the type, repetition frequency, and interstimulus interval of stimulation (Kekoni et al. 1997;Shinozaki et al. 1998;Spackman et al. 2007;Shen et al. 2018aShen et al. , 2018b. Future studies should therefore extend the current results by including stimulation of other body parts and different stimulation protocols. Additionally, it is possible that Psi induced greater inter-individual variability in the EEG-responses compared with Pla. Further studies that are well powered to investigate interindividual variability are needed to test this hypothesis.

Conclusions
This study investigated the impact of the preferential 5-HT2A agonist Psi on the processing of tactile deviancy processing and its relation to the formation of bodily and self-awareness. The sense of touch is not raw and direct but rather constructed with reference to internal body representations that contain prior expectations (Haggard et al. 2003). We show that Psi alters the integration of tactile sensory inputs via aberrant PE processing and potentially reduced memory trace formation of tactile information. Furthermore, our results point towards an association between Psi-induced reduced responses to surprising stimuli and alterations in subjective body and self-experience.
Our findings therefore highlight the role of the serotonin and in particular the 5-HT 2A system in the disruption of multisensory processing of self-and body-related sensory inputs and perceptual tactile learning. This findings may be important for the treatment of many psychiatric disorders which involve aberrant recall or integration mechanisms of bodily selfrepresentations, such as body dysmorphic disorder (Hrabosky et al. 2009), anorexia nervosa (Gadsby 2017), or depression (Fuchs and Schlimme 2009).