{"paper_id":"6c3dd6fe-809d-4d79-9be3-b8f8740d3609","body_text":"Anomalies in effective connectivity explain different hallucination subtypes in Parkinson’s disease psychosis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Anomalies in effective connectivity explain different hallucination subtypes in Parkinson’s disease psychosis Miriam Vignando, Dominic ffytche, Ndabezinhle Mazibuko, Giulio Palma, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6627999/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Psychosis and visual hallucinations (VH) in Parkinson’s disease (PD) significantly impact patient outcomes, yet underlying neural mechanisms remain unclear, limiting effective treatments. Here, we used dynamic causal modelling (DCM) to leverage the fast temporal dynamics captured with EEG data during a visual mismatch negativity task in PD patients with (N = 20) and without (N = 18) VH to examine effective connectivity. We found reduced top-down and enhanced bottom-up connectivity in ventral visual and prefrontal regions during task performance in PD-VH, suggesting deficits in sensory prediction updating and an over-reliance on visual input. Connectivity patterns differed with hallucination complexity: minor VH related to left hemisphere deficits, complex VH to altered top-down and bottom-up right-hemisphere connectivity, and multimodal hallucinations to widespread bilateral disruption. Increased task activity as computed with source reconstruction correlated positively with cortical 5-HT2A receptor distribution. These findings highlight specific neural targets for early therapeutic interventions, supporting a transdiagnostic computational architecture of hallucinations. Biological sciences/Neuroscience/Diseases of the nervous system/Parkinson's disease Biological sciences/Neuroscience/Computational neuroscience Health sciences/Diseases/Neurological disorders/Movement disorders/Parkinson's disease Biological sciences/Neuroscience/Cognitive ageing Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction Parkinson’s disease (PD) is characterised by motor and non-motor symptoms 1 , among which psychosis and visual hallucinations (VH) notably impact patient outcomes and are associated with cognitive decline 2 – 4 . Although morphological 5 − 3-5 and functional 8 , 9 alterations differentiate PD patients with VH (PD-VH) from those without VH, standard cognitive assessments lack mechanistic insights into VH. A recent review of frameworks 10 converges on the proposal that ‘ascending’ sensory disturbances together with ‘descending’ factors (e.g. expectations, specific object features, attention) contribute to VH. Among the different proposals, predictive coding, a cognitive model of psychosis, suggests hallucinations arise from disrupted balance between top-down expectations and bottom-up sensory inputs, with a stronger influence of expectations observed in people with hallucinations 11 – 13 . The ERP component mismatch negativity (MMN), a robust marker of psychosis 14 , is thought to reflect an attempt to minimise this prediction error and capture its disruption 15 – 18 . We recently demonstrated reduced visual MMN responses in PD-VH 19 , paralleling auditory MMN impairments observed in psychosis 14 . However, traditional ERP analysis cannot uncover the directional neural mechanisms involved. Dynamic causal modelling (DCM), leveraging EEG’s temporal precision, enables exploration of these mechanisms 20 , 21 . DCM has proved instrumental in clarifying the disruption of the MMN in people with schizophrenia-related psychosis 22 and a spectral DCM study (resting-state fMRI) found increased top-down and reduced bottom-up connectivity in the visual network in PD with VH, a pattern was predictive of hallucination severity 8 . Here, we employ DCM with neural microcircuit models across dorsal (V1, IPL) and ventral (V1, ITG) visual pathways interacting with the prefrontal cortex (PFC) during a visual MMN task, to investigate their relation to hallucination complexity. We include the PFC to test the prediction error impairment as top-down connectivity is crucial for integrating sensory information 23 – 25 and we focus on the dorsal and ventral pathways based on the literature 19 , 26 – 30 and our MMN task changing line orientation, a process known to involve specific parietal regions 31 . We hypothesise reduced top-down and enhanced bottom-up connectivity in PD-VH, reflecting impaired prediction updating and reliance on compromised sensory information, as visual input is known to be defective in PDP 32 . Additionally, exploratory analyses relate MMN-derived source activity to cortical distributions of receptors, previously implicated in PD-VH. Methods and Materials Data and code availability statement. All the code developed for the study is available here: https://github.com/VMiri/vMMN-dcm . ERP data after artefact removal and averaging is available in the same repository in the .mat and .dat formats.. The main ERP study was pre-registered ( https://osf.io/q9x7v ). Participants. The study received ethics approval from London Camberwell St Giles REC (18/LO/2144). We enrolled 18 patients with PD without hallucinations (PD-noVH) and 20 patients meeting criteria for PDP (PD-VH in this article) 33 . Participant recruitment and screening details are provided in detail in Supplementary Information S1 and in our ERP study 19 . VH and psychosis in PD-VH were further assessed using the Scale for the Assessment of Positive Symptoms-PD (SAPS-PD) 34 , and an expanded version of the North-East Visual Hallucination Interview (NEVHI) 35 to examine the phenomenology of visual hallucinations and their subtypes in PD-VH. (see SI1 ). Task. We used a visual mismatch negativity task whereby the stimulus frequency variation was the orientation of peripheral bars (SI2). Figure 1a provides a visual summary of the vMMN task; the task design was inspired by vMMN paradigms proposed in Qian et al. 36 ; details are provided in SI2 and the ERP analysis is described in detail in Vignando et al 19 . Statistical analysis. Participant demographics. PD-VH and PD did not differ on age, sex, disease duration, MMSE score, levodopa equivalent daily dose (LEDD) and motor severity ( Supplementary SI1) . None of the participants were on quetiapine. Two PD-VH and two PD patients were on SSRIs. EEG data pre-processing. EEG data was pre-processed with the standard spm12 pipeline (see SI2) for ERP analysis 37 . In brief, we created a custom montage file to exclude eye channels and a trial definition file for the interval –100 to 500 (the duration of the task) and for the two conditions of interest (standard, rare deviant), with shift triggers = 0. We used a high-pass filter with cutoff = 0.3, then we down sampled to 500 Hz and lowpass filter with cutoff = 30. The trial definition file was used for epoching and baseline correction was applied pre-stimulus. After epoching, bad trials and channels were visually inspected, marked and then artefacts were removed with threshold set at 80 37 . ERPs were then averaged for standard and rare deviant conditions (seeSI2). Sensory space and Source reconstruction analysis. We conducted analyses in sensor space to identify the electrodes where the ERP amplitude differed across conditions during the task and to validate the results from our ERP study 19 (see SI3 ). To gain more detailed spatial information regarding the source of our signal, we perform source reconstruction ( SI4 ) for the 60-400ms interval of interest to be investigated with the DCM analyses and we entered them in second level analyses (one-sample t-test; paired t-test) to investigate the neural correlates of task-related activity to guide the choice of the dipoles. Dynamic Causal Modelling: Individual DCM model specification. Based on the results of the ERP study 19 and of the source reconstruction analyses, pointing to the involvement of occipital, occipito-parietal and frontal regions, we decided to investigate the ventral and the dorsal pathways separately. For the ventral pathway, we designed three models: all three models share the same dipoles, to allow for Bayesian model selection, but different neuronal models (Supplementary Figure 3). Since we did not have strong results for the dorsal pathway in the source reconstruction analysis we focussed on one dorsal model, using source reconstruction results, and the literature on the Benton line judgement task and the neural anatomy of the dorsal pathway 31 to define the dipoles (SI11). We used an ERP neural mass model, focussing on the 0 to 400ms interval, rare deviant vs. standard. An equivalent current dipole (ECD) was used for each source, with anterior occipital, inferior temporal and dorsolateral prefrontal regions, bilaterally (Figure 2c). All model specification details are provided in SI5, Supplementary Figure 3 and SI11 and Bayesian model comparisons are detailed in Supplementary Figure 4. 2 nd level analysis: Parametrical empirical Bayes (PEB) . After specifying individual DCMs for each participant , PEB 38,39 was applied to test for group differences in effective connectivity. PEB uses a Bayesian linear regression approach, which allows group-level comparisons of modulatory parameters estimated at the individual level, without losing information about the precision of those estimations. We fitted a general linear model (GLM) to the connection strengths estimated with the first level analyses and with covariates: mean, group (PD-VH (-1) or PD (1)) and age. spm_dcm_bmr and spm_dcm_peb were used to perform Bayesian model reduction and to generate model posteriors, respectively. We used Bayesian model averaging using spm_dcm_bma to investigate group-level differences between patients with and without VH. We analysed the different B “task modulatory” matrix separately and A “baseline” matrix (feedforward, feedback, inter-hemispheric connectivity). To understand which of the parameters contributed best to explaining variance between DCMs, we also created a model space to carry out PEB analyses recursively with specific parameters specified, allowing switching of specific parameters on and off and then comparing the models (SI6, Supplementary Figure 7). Bayesian model comparison at the second level was then explored and the parametrised winning 2 nd level model was further examined with Bayesian model reduction ( BMR ). BMR allows to take a hypothesis-free approach to finding the winning model for effective connectivity within a network. We also explored task connectivity in PD-noVH to isolate the effect of task, as we previously showed that this group had a stronger MMN if compared to PD-VH ( Supplementary Figure 6 ). Hallucination subtypes: individual connections linear regression analyses . We conducted linear regression analyses between NEVHI scores for visual complex (CVH) and minor hallucinations (MH) (for the PD-VH group) and the estimated connection strength (Ep) for each of the connections found (pp>.99) with the PEB analysis. As we also had collected information on multimodality of hallucinations in our patients from the SAPS-PD (see SI1 ), we explored the relationship between connectivity in PD-VH and multimodality of hallucinations. Outliers were identified with Cook’s distance 40 , inspected, and removed. We computed multiple comparisons correction using false discovery rate across all individual models ran. R package Tidyverse 41 and dplyr 42 were used for these analyses (see SI7). Hallucination subtypes: Leave one out multiple regression models. As we found more than one connection correlating with the severity of VH, we performed post-hoc backwards regression analyses with leave one out cross-validation (using R package caret 43 ). We also ran models including only the individual connections found with the individual regression analyses described in the previous paragraph to explore whether the model with the multiple connections was improved for predicting simple and complex VH (severity scores) (separately). We used BIC (Bayesian Information Criterion) to select the best model and avoid overfitting. We also computed the same models with MoCA and then LEDD as covariates to accommodate the potential role of general cognition and medication (SI8). Analysis of neural sources. We explored the estimated activity of the three neural populations described in the microcircuit model (ERP neural mass model: spiny stellate cells, inhibitory interneurons, deep pyramidal neurons) by retrieving the estimated values from DCM.M.dipfit.H and used to better understand the deficit in the PD-VH group for the difference between deviant and standard in relation to the hallucination severity. We used R packages dplyr 42 and caret 43 to perform leave-one-out cross-validation (LOOCV) with linear regression implemented via the caret package 43 in R (method = 'lm') to test if these estimates could significant predict VH severity. We specify VH scores as the dependent variable and activity in a specific neural population and region as an independent predictor. For each iteration, one participant was excluded from model training and used as the test case. We used LOOCV since we had a limited sample size and using the linear regression method implemented in caret allows to maximize data utilization and to minimise overfitting (see SI9). Excitatory/inhibitory (E/I) coupling simulations. We conducted this analysis to probe the network-level mechanisms and to understand more clearly the role of intrinsic coupling in the results pertaining to the previous sections and on the PEB analysis where specific parameter changes were found to be associated to the inhibitory to excitatory coupling. The aims was to modulate the intrinsic coupling parameter and examine how connectivity changes with VH. We did so by adapting and tailoring the methods and procedures described in Rosch et al 44 to our group contrast and our neural mass model (all details in SI10.) Briefly, we modulated the intrinsic coupling parameter to simulate the graded effects of VH on our cortical microcircuits. We started from the grand mean model and individual participants inversions and applied VH-related perturbations estimated from the PEB analysis. We extracted the VH contrast from the design matrix, and scaled the G parameter accordingly across 10 steps (range: 0–0.5) 15 , applying these to all six cortical regions in the model (bilateral V1, ITG, and PFC), with each step incrementally increasing the strength of G proportionally to the magnitude of the PEB-derived group difference. The purpose was to simulate biologically and physiologically plausible perturbations to explore how network dynamics evolve under increasing psychosis load. We then used spm_gen_erp to generate predicted neural responses for each parameter configuration. For each simulation, we extracted excitatory and inhibitory time-series data from all regions and neural populations. We examined both time-domain activity (e.g. peristimulus firing rate heatmaps) and state-space trajectories (excitatory vs. inhibitory activity over time). To quantify possible differences, we computed the area under the curve (AUC) for inhibitory and excitatory activity of the time series using a trapezoidal method. We also computed state-space area as a measure of network flexibility. All details are provided in SI10. These analyses were carried out in MATLAB 2023b. Exploratory receptor binding atlases and MMN source reconstructed signal analysis . The aim of this analysis was to explore a possible relationship between vMMN signal and neurotransmitter systems implicated in PD with VH, namely the serotonergic 45 , dopaminergic 46 and cholinergic 47 systems. First, we computed vMMN difference images by subtracting standard from rare deviant activity for each participant, with each image then registered to the atlas using the imcalc function. We performed PET maps and atlas alignment; figures resulting from this procedure are provided in SI13. A custom script was created to parcellate the source-reconstructed difference images and the PET maps of interest (11c-lsn3172176 48 for muscarnic M1, [ 18 F]altanserin 50 for 5-HT2A, [ 18 F]FEOBV 51,52 for VAChT, [ 18 F]fallypride 53 for dopamine D2 receptors and [ 11 C]SCH23390 54 for dopamine D1 receptors). This allowed for the extraction of regional binding potential (BP ND ) for each PET map and the corresponding reconstructed EEG signal within the regions defined by the atlas. First, we ran individual regression models using regional receptor binding as a predictor and regional MMN signal difference (VH – noVH) as dependent variable. Outliers were identified with Cook’s distance 43 and removed (details and results provided in SI13). P values for these models were corrected for multiple comparisons. We proceeded with further analysis including only the maps surviving multiple comparisons correction. We accounted for spatial autocorrelation using the BrainSMASH toolbox 56 to generate the MNI coordinates for our atlas. We removed the outlier regions we had removed with the regression models, to run the correlational analyses on the exact same regions on which we ran the regressions. By creating centroids from the .nii atlas and running multiple permutations, we obtained a correlation coefficient that is proposed, by Burt and colleagues, to reflect the relationship between the two maps free from spatial autocorrelation. For the maps surviving this ‘actual’ correlational analysis, we carried out linear mixed models using individual-level data, to investigate the effect of age, MoCA score, LEDD and disease onset in this relationship using lme4 57 and lmerTest 58 . Further details on the methods and the models are provided in SI12. Results Participants did not differ in sex (χ² = .01, p =.91, 6 F in the PD-VH group, 7 F in the PD group, 13 M in each group), age, disease duration, motor symptoms, LEDD or MoCA score ( Supplementary Tables 1, 2; Supplementary Figure 1) . Sensor space and source reconstruction analyses identify occipital and temporal sources Sensor space analyses confirmed that the channels involved in the task returned temporal and frontal locations ( p <.001, SI3 ) and parieto-occipital (FWE p <.05; SI3 ). The t-test conducted for source reconstruction returned clusters in the inferior temporal gyrus and in lateral occipital regions ( FWE p <.05) and in the thalamus (p<.001, pFWE = .07 ) (Figure 2, SI4). vMMN differences correlate with decreased top-down connectivity in PD-VH BMS (Supplementary Figure 4) with a fixed effect design determined which of our ventral stream models (fully connected model) best-fit the data. We then carried out 2 nd level analyses with PEB finding that task stimulus condition (rare vs standard) for PD-VH was associated with increased bottom-up effective connectivity and decreased self-inhibition for lV1, and with decreased connectivity, especially top-down (pp>.99; Figure3 ). PEB analyses on latent connectivity (matrix A) show that at baseline for PD-VH there is an overall increase in effective connectivity both top-down and bottom-up, with an overall opposite pattern of that observed to the task from the PFC to V1 bilaterally ( Figure3F, G ). The recursive PEB exploratory analysis on alternative second-level models (SI6, Supplementary Figure 7 ) points to a primary role of feedforward connections, with increased connectivity from rV1 to rITG and to rPFC. The fully connected model was among the top two best models and further analyses confirmed the results presented in Figure 3 (Supplementary Figure 7). We also explored the parameters of the intrinsic coupling only model, where the difference in PD with and without VH is associated to a change in cortical microcircuitry encoded in the intrinsic coupling parameter, finding a reduction in intrinsic connectivity from excitatory cells in right V1, left ITG and right PFC and from inhibitory cells (disinhibition) for all regions. Hallucination severity, subtype and multimodality analysis. The severity of complex visual hallucinations (CVH) correlated with distinct patterns of connectivity: greater severity was associated with reduced top-down connectivity (right PFC to right ITG; left V1 to left ITG) and enhanced bottom-up connectivity (right V1 to right PFC and right ITG). Minor hallucinations (MH) had lower connectivity from left PFC to left V1, although this did not remain significant after correction for multiple comparisons (all stats in Figure 4; details in SI7). ML Multiple regression models with LOOCV Both top-down and bottom-up connectivity right-hemisphere changes predicted CVH severity [ R 2 = 0.49, adjusted R 2 0.4 (3,15) = 4.9, p -value: 0.014 RMSE =2.26, MAE = 177] (SI8; Figure 8), MH were best predicted by left-hemisphere decreases (lV1-rV1, lPFC-lV1) [ R 2 = 0.40, adjusted R 2 = .32 F(2,17) = 5.60, p = 0.01, RMSE = 5.69, MAE = 4.30, BIC = 126.26]. Results were unchanged when introducing MoCA or LEDD as covariates (SI8). At baseline, CVH severity was best predicted by inter-hemispheric connectivity in V1 and top-down connectivity from PFC bilaterally; MH were predicted by left-hemisphere effective connectivity (SI8). Multimodal hallucinations were best predicted by a model using the 7/10 connections found altered (Figure 4D); introducing MoCA as a covariate rendered the top-down, but not the bottom-up model non-significant. Neural microcircuit models and simulations. Leveraging the estimates of neural dynamics featured in the neural mass model used (inhibitory interneurons, spiny stellate cells, pyramidal neurons), we found that decreased inhibitory activity across ventral visual and prefrontal regions significantly predicted the severity of complex visual hallucinations (CVH); this was also the case when combining inhibitory and excitatory estimates, supporting a network-level deficit and a disruption of excitatory-inhibitory (E/I) balance, with the ITG as a particularly significant contributor (SI8). Simulations of perturbation of E/I balance revealed pronounced reductions in both excitatory and inhibitory activity in PD-VH, particularly within the left V1 and bilateral ITG regions. These disruptions occurred predominantly during the 100–250ms period (MMN) and confirmed by our AUC analysis (SI10). When computing state space area in the two groups the difference is statistically significant for every region, confirming the differences observed with the plots (Figure 5; SI10). Dorsal model. All details about model specification are reported in SI11. For what concerns the task connectivity the BMA results show that task effects (rare vs standard) for PD-VH were mostly associated with an altered connectivity between parietal and frontal nodes (pp > .99). For this model PD-VH connectivity was increased from left V1 to left IPL, and reduced IPL to PFC, bilaterally; there were no correlations with severity of hallucinations (Supplementary figures 9-11). Source-reconstructed signal and receptor binding atlases exploratory analysis. Results for the first linear regression analyses with Cook’s distance 43 to inspect outliers are reported in SI13 and in Figure 6B. Using BrainSMASH 56 we confirmed a positive correlation between 5-HT 2A BP ND and MMN signal ( r =.284, p =.037) with higher connectivity in regions of higher BP ND . The opposite pattern was observed for D1 ( r =-.29, p =.017) and VAChT ( r =-2.73, p =.03). D2/D3 was no longer significant after correcting for spatial autocorrelation (p >.05). When investigating these results at the individual level using LMMs ( MMN_signal ~ _receptor * VH + age + LEDD + sex + disease onset +MoCA + (1 | Participant)) , we found for 5-HT 2A a main effect of 5‐HT 2A t =2.13, p =.034, hallucination status (VH) t =2.52, p =.017 and a positive interaction between 5‐HT 2A and VH t =2.06, p =.040. The same model using D1 receptor availability found a significant group effect t = 2.54, p = .015, and a negative D1 * VH interaction t = –2.13, p =.033. For VAChT, using the map from N=18 older participants 52 , we found a positive association with MMN ( t =1.75, p =.08) and a negative VAChT * VH interaction effect ( t =-1.58, p =.1), though neither reached significance; VH status was significant t =2.46, p =.019. In a sensitivity analysis with the N=4 healthy controls map 51 , we observed the same direction of effects but with a larger effect size and formal significance (partial r ≈ –.037 VAChT * VH interaction; see SI13). These convergent patterns suggest a negative relationship between VAChT and MMN in the VH group, though not as robust as D1. No effect of disease duration, LEDD and age was observed (SI13). Discussion We investigated the neural dynamics underlying sensory processing deficits in patients with PD and visual hallucinations (PD-VH). Using dynamic causal modelling applied to EEG data from a visual MMN task, we showed that PD-VH patients respond differently to the sensory environment and that these differences correlate with their visual hallucinations. Decreased top-down and increased bottom-up activity underlies PD-VH reduction of vMMN We found that in PD-VH there was a significant reduction of top-down activity and an increase in bottom-up connectivity. Top-down connectivity is crucial for the integration of sensory information 24-26 and this finding suggests that in PD-VH there is a failure to update predictions based on sensory evidence and an over-reliance on sensory information. This over-reliance can contribute to the difficulties in stimulus detection due to retinal degeneration 28,32 and responding to the changes in the sensory environment. This interpretation resonates with the principles of a recently proposed framework on VH in PD 10 . When taking a hypothesis free approach the feedforward model, with a right-hemisphere pattern of increased V1-ITG and V1-PFC connectivity emerged as the strongest contributor. In line with this result, recent work has shown that people are more likely to hallucinate when sensory information is noisy, especially when combined with strong expectations 60 . Interestingly this was found with an orientation perception task, consistent with our stimuli, and in the input layers of V2, where orientation-specific activity takes place, lending further support to our interpretation and suggesting that spontaneous feedforward activity in the visual cortex can also lead to hallucinations, rather than feedback activity alone 60 . The finding that local source differences correlate positively with 5-HT 2A density suggests that this receptor may be facilitating cortical disinhibition, consistent with previous proposals 45,61 and studies showing 5-HT 2A antagonists reducing psychosis-like symptoms 19,45,61 . Based on our results, we may speculate that this reflects an aberrant or a compensatory excitation. We also find a negative relationship with dopaminergic receptor D1; dopamine is proposed to encode precision weighting 62 ; a possible interpretation is that this could contribute to bias the system to assign undue salience to bottom-up inputs with the result of illusions and/or hallucinations, and the disruption of normal mismatch detection. The analysis of the baseline activity complements these results: top-down connectivity from PFC increased in the PD-VH group, while the same connections were under-responsive during the task, and vice versa for some of the bottom-up connections. These differences in latent connectivity even in the absence of experimental perturbations suggest that PD-VH are not just over-relying on sensory information during the vMMN task, but they seem to do so more generally. The decreased activity from visual to prefrontal at baseline could reflect the low-level visual deficits common in PD-VH and affecting also higher-order processes, which together with the reduced PFC to ITG left connectivity seems to support this and previous models of VH in LBD 63 . Decreased top-down and increased bottom-up connectivity in PD-VH are related to hallucinations severity Hallucinations severity directly correlated with these connectivity patterns, with pathways of increased connectivity showing a positive relationship and those with decreased connectivity showing a negative relationship. Complex visual hallucinations correlated with decreased top-down connectivity from right PFC to ITG, and increased bottom-up connectivity from early visual cortex (V1) to higher-order regions (ITG, PFC) (Figure7). These results underscore a robust right-hemisphere network impairment associated with complex hallucinations. Minor hallucinations showed a similar but weaker trend predominantly involving left-hemisphere regions, though this did not survive multiple comparison correction. A recent DCM fMRI study found that the pattern of altered connectivity observed at rest was predictive of hallucinations severity 8 . When addressing the relationship between multiple connections and hallucinations severity with a similar purpose, we find that CVH were best predicted by right-hemisphere bottom-up and top-down connections, supporting the hypothesis that these regions operate as a network. This right-hemisphere pattern of connections associated with CVH is in line with a recent fMRI study exploring VH severity with resting-state fMRI in PD and DLB with hallucinations, where CVH duration was associated with right hemisphere activity in ventral visual and parieto-occipital regions, and minor visual phenomena to decreased connectivity only in the left hemisphere 29 , as we also found for MH. The result of increased activity from rV1 being associated to VH and VH severity aligns with the finding in schizophrenia of disinhibition in sensory areas (A1) predicting abnormal auditory perception in the aMMN and positive symptoms 22 . Visual hallucinations severity may be related to disinhibition Simulations of intrinsic excitatory-inhibitory (E/I) dynamics demonstrated significant imbalance and instability in the PD-VH network, particularly in visual and ITG regions, extensively involved in object recognition, particularly of human and animal stimuli 63,64-66 , which are often the content of VH in these patients 67 . The exploratory analysis with estimated activity of excitatory cells in the left ITG positively related to CVH specifically suggests that this impairment may be associated with the complex stimuli in the visual modality alone and presents a circuit-level causal hypothesis involving the ITG for the failure to have greater activity for the deviant trials we showed previously 19 . The ITG also showed instability in the simulations, with the analysis of the simulated time series and the state space area analyses suggesting that hallucinations might arise from disrupted cortical inhibition and heightened neural sensitivity to perturbations. Our results lend themselves to more than one potential explanation. First, that of an E/I imbalance similar to what has been proposed in schizophrenia 68 . We can interpret the pattern observed as a reduced adaptability and stability in the network dynamics and in the intrinsic coupling within the regions in our model. This is in line a recent aMMN study where the electrophysiological patterns observed under ketamine administration and in the relative simulations appear compatible with those observed in psychosis 44 . In this view, reduced activity through glutamate receptors is proposed to induce a decrease in the inhibitory activity of the inhibitory interneurons. Lower GABA+/creatine in the ITG in PD-VH 69 supports this view. Disinhibition may lead to excessive excitatory signals, causing the positive symptoms in schizophrenia and we may speculate the hallucinations in PD-VH. We observe at task an excess in excitatory activity in low visual regions and at baseline an overall increase in excitatory activity, in line with this hypothesis. Further supporting the interpretation that hallucinations may arise from disrupted cortical inhibition and heightened neural sensitivity to perturbations, exploratory analyses using PET receptor atlases revealed a positive correlation between cortical 5-HT 2A receptor distribution and abnormal MMN source activity in PD-VH, consistent with the known role of serotonin in cortical disinhibition 70 and psychosis 45 . Another possible interpretation is that our results may be related to thalamic dysfunction. We did not investigate the thalamus here, as with EEG recordings the signal captures postsynaptic potentials in the cortex, making the choice of using the thalamus as a dipole a challenging one. Nevertheless, a large thalamic cluster was found associated to the task in our source reconstruction analysis. Previous research has shown that the thalamus plays a role in VH in Parkinson’s 71,72 and a recent proposal 73 sees the thalamus as driver of unbalanced network recruitment, suggested to induce a decoupling of the default mode network (DMN) and task-positive networks, possibly disrupting the comparison of priors and of sensory percepts 73,74 . A recent spectral DCM fMRI study found an involvement of the LGN of the thalamus, together with ventral visual regions, in PD-VH 8 . Despite not directly comparable to our results due to the different (resting fMRI and task EEG) modalities, nevertheless when examining the latent activity (between trials) we also have increased top-down connectivity, including from left PFC to V1 as reported in the aforementioned study, and both increased and reduced bottom-up activity. Our exploratory analysis with PET atlases and source-reconstructed MMN signal also shows a negative relationship between cholinergic receptors distribution and MMN signal in PD-VH. A recent PET study in PD-VH showed a marked cholinergic deficiency in the left ventral visual stream 47 . We may speculate that a dysregulation in the cholinergic system might affect not only cortical areas but also thalamic processing, thereby contributing to abnormal sensory processing. Global cognitive decline may modulate the presence of multimodal hallucinations Our results also indicate that multimodal hallucinations might reflect more extensive neural network disruptions, modulated by a sub-clinical global cognitive decline, as controlling for cognition left primarily bottom-up sensory pathways significantly associated with multimodal hallucinations. This is consistent with reports of PD-VH declining cognitively more rapidly 2 and showing possible sub-clinical cognitive variations early on in their VH 4 . Limitations While our study primarily focused on cortical regions, exploratory analyses hint at the involvement of subcortical structures like the thalamus and cholinergic systems, consistent with previous research, which may offer promising directions for future studies. Limitations include the inability of EEG to robustly assess subcortical sources, the exploratory nature of our receptor-binding analyses, and that the vAChT map we used was developed on data from both healthy and Alzheimer’s disease participants, thus it might partly reflect AD‐related patterns. We validated our finding with a sensitivity analysis using another VAChT map (N=4), that by itself would have been a less realistic fit for our study. Conclusions Overall, our findings clarify the top-down and bottom-up neural abnormalities underlying PD-VH, strongly supporting the predictive coding account, and pointing to an alteration in updating predictions based on sensory evidence and a deficit in tracking changes in the sensory environment. Our results highlight potential receptor targets mediating this effect and reinforce the critical role of the ventral visual stream in the generation of VH. The hyperconnectivity of visual regions strongly contributes to explaining vMMN differences and correlates with hallucinations severity and complexity, supporting targeted therapeutic approaches focused on normalizing sensory processing dynamics. Finally, we show an increasingly extensive network being related to VH complexity and a possible relationship with sub-clinical cognitive decline. Having provided a mechanistic model of predictive coding deficits in PD-VH, we think future longitudinal studies will be essential to confirm whether observed connectivity changes directly underpin hallucinations alone or reflect a broader underlying cognitive decline. Declarations Acknowledgements We thank the CRISP (patient representation) group for support and advice during the inception of this study and we thank all our participants for taking part in the research and their partners for their support on the study days. We thank Rosalyn Moran for her insight regarding the possible avenues of investigation to follow with DCM and advice on model specification. We also thank Rick Adams for comments on an earlier version of the analyses. We also thank Dag Aarsland for his support with the development of important study ideas. We thank Stephanie Stephenson for training the study researchers with clinical trials procedures and for her support throughout the duration of the study; and we thank Caroline Woolridge for her support with study logistics and ethics applications. We thank Simon Hill for the contribution to the EEG task and recording setup. We thank the KCL Clinical Research Facility staff for their support with the study and the CRF director Elka Giemza. For open access, the author has applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising from this submission. Funding This is independent research funded by the Medical Research Council (grant number MR/R005931/1 awarded to M.A.M., D.F., K.R.C. and Dag Aarsland) and carried out at the NIHR (National Institute for Health and Care Research) Maudsley Biomedical Research Centre (BRC). The views expressed are those of the author(s) and not necessarily those of the Medical Research Council, the National Institute for Health and Care Research or the Department of Health and Social Care. M.V. was supported by Medical Research Council and is currently supported by Alzheimer’s Research UK (grant ARUK-RF2022B-002). C.H.W.G. is supported by the Medical Research Council (MR/W029235/1) and the National Institute for Health and Care Research Cambridge Biomedical Research Centre (NIHR203312). The views expressed are those of the authors and not necessarily those of the National Institute for Health and Care Research or the Department of Health and Social Care. References Schapira, A. H., Chaudhuri, K. R., & Jenner, P. (2017). Non-motor features of Parkinson disease. Nature Reviews Neuroscience , 18 (7), 435-450. Ffytche, D. H., Creese, B., Politis, M., Chaudhuri, K. R., Weintraub, D., Ballard, C., & Aarsland, D. (2017). The psychosis spectrum in Parkinson disease. Nature Reviews Neurology , 13 (2), 81-95. Pisani, S., Gosse, L., Aarsland, D., Chaudhuri, K. R., Ballard, C., Ffytche, D., ... & Bhattacharyya, S. (2024). Parkinson’s disease psychosis associated with accelerated multidomain cognitive decline. BMJ Ment Health , 27 (1). Stampacchia, S., Bernasconi, F., Van De Ville, D., Amico, E., Pagonabarraga, J., Kulisevsky, J., & Blanke, O. (2025). Connectome-based brain fingerprints predict early cognitive decline in Parkinson's patients with minor hallucinations. bioRxiv . Vignando, M., Ffytche, D., Lewis, S. J., Lee, P. H., Chung, S. J., Weil, R. S.,... & Mehta, M. A. (2022). Mapping brain structural differences and neuroreceptor correlates in Parkinson’s disease visual hallucinations. Nature communications , 13 (1), 519. Weil, R. S., Hsu, J. K., Darby, R. R., Soussand, L., & Fox, M. D. (2019). Neuroimaging in Parkinson’s disease dementia: connecting the dots. Brain communications , 1 (1), fcz006. Lenka, A., Jhunjhunwala, K. R., Saini, J., & Pal, P. K. (2015). Structural and functional neuroimaging in patients with Parkinson's disease and visual hallucinations: a critical review. Parkinsonism & related disorders , 21 (7), 683-691. Thomas, G. E., Zeidman, P., Sultana, T., Zarkali, A., Razi, A., & Weil, R. S. (2023). Changes in both top-down and bottom-up effective connectivity drive visual hallucinations in Parkinson’s disease. Brain Communications , 5 (1), fcac329. Shine, J. M., Muller, A. J., O’Callaghan, C., Hornberger, M., Halliday, G. M., & Lewis, S. J. (2015). Abnormal connectivity between the default mode and the visual system underlies the manifestation of visual hallucinations in Parkinson’s disease: a task-based fMRI study. npj Parkinson's Disease , 1 (1), 1-8. Collerton, D., Barnes, J., Diederich, N. J., Dudley, R., Friston, K., Goetz, C. G., ... & Weil, R. S. (2023). Understanding visual hallucinations: a new synthesis. Neuroscience & Biobehavioral Reviews , 150 , 105208. Sterzer, P., Adams, R. A., Fletcher, P., Frith, C., Lawrie, S. M., Muckli, L., ... & Corlett, P. R. (2018). The predictive coding account of psychosis. Biological psychiatry , 84 (9), 634-643 Bastos, A. M., Usrey, W. M., Adams, R. A., Mangun, G. R., Fries, P., & Friston, K. J. (2012). Canonical microcircuits for predictive coding. Neuron , 76 (4), 695-711. Haarsma, J., Knolle, F., Griffin, J. D., Taverne, H., Mada, M., Goodyer, I. M.,... & Nspn Consortium. (2020). Influence of prior beliefs on perception in early psychosis: Effects of illness stage and hierarchical level of belief. Journal of abnormal psychology , 129 (6), 581. Bhat, A., Irizar, H., Thygesen, J. H., Kuchenbaecker, K., Pain, O., Adams, R. A., ... & Bramon, E. (2021). Transcriptome-wide association study reveals two genes that influence mismatch negativity. Cell reports , 34 (11), 108868 Friston, K. J., & Dolan, R. J. (2010). Computational and dynamic models in neuroimaging. Neuroimage , 52 (3), 752-765. Feldman, H., & Friston, K. J. (2010). Attention, uncertainty, and free-energy. Frontiers in human neuroscience , 4 , 215. Friston, K. J. (2005). Hallucinations and perceptual inference. Behavioral and Brain Sciences , 28 (6), 764-766. Rao, R. P., & Ballard, D. H. (1999). Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. Nature neuroscience , 2 (1), 79-87. Vignando, M., Ffytche, D., Mazibuko, N., Palma, G., Montagnese, M., Dave, S., ... & Mehta, M. A. (2024). Visual mismatch negativity in Parkinson's psychosis and potential for testing treatment mechanisms. Brain Communications , 6 (5), fcae291. Kiebel, S. J., Garrido, M. I., Moran, R. J., & Friston, K. J. (2008). Dynamic causal modelling for EEG and MEG. Cognitive neurodynamics , 2 , 121-136. Kiebel, S. J., David, O., & Friston, K. J. (2006). Dynamic causal modelling of evoked responses in EEG/MEG with lead field parameterization. NeuroImage , 30 (4), 1273-1284. Adams, R. A., Pinotsis, D., Tsirlis, K., Unruh, L., Mahajan, A., Horas, A. M., ... & Anticevic, A. (2022). Computational modeling of electroencephalography and functional magnetic resonance imaging paradigms indicates a consistent loss of pyramidal cell synaptic gain in schizophrenia. Biological psychiatry , 91 (2), 202-215. Friston, K. (2005). A theory of cortical responses. Philosophical Transactions of the Royal Society B: Biological Sciences, 360 (1456), 815-836. Summerfield, C., & Egner, T. (2009). Expectation (and attention) in visual cognition. Trends in Cognitive Sciences, 13 (9), 403-409. Kveraga, K., & Bar, M. (2014). Top-down predictions in the cognitive brain. Brain and Cognition, 94 , 65-67. Shine, J. M., O’Callaghan, C., Halliday, G. M., & Lewis, S. J. (2014). Tricks of the mind: visual hallucinations as disorders of attention. Progress in neurobiology , 116 , 58-65. Pagonabarraga, J., Soriano-Mas, C., Llebaria, G., López-Solà, M., Pujol, J., & Kulisevsky, J. (2014). Neural correlates of minor hallucinations in non-demented patients with Parkinson's disease. Parkinsonism & related disorders , 20 (3), 290-296. Diederich, N. J., Fénelon, G., Stebbins, G., & Goetz, C. G. (2009). Hallucinations in Parkinson disease. Nature Reviews Neurology , 5 (6), 331-342. D’Antonio, F., Boccia, M., Di Vita, A., Suppa, A., Fabbrini, A., Canevelli, M., ... & Ffytche, D. (2022). Visual hallucinations in Lewy body disease: pathophysiological insights from phenomenology. Journal of Neurology , 269 (7), 3636-3652. Montagnese, M., Vignando, M., & Mehta, M. A. .. (2022). Cognitive and visual processing performance in Parkinson's disease patients with vs without visual hallucinations: A meta-analysis. Cortex , 146 , 161-172. Garcia-Diaz, A. I., Segura, B., Baggio, H. C., Marti, M. J., Valldeoriola, F., Compta, Y., ... & Junque, C. (2018). Structural brain correlations of visuospatial and visuoperceptual tests in Parkinson’s disease. Journal of the International Neuropsychological Society , 24 (1), 33-44. Weil, R. S., Schrag, A. E., Warren, J. D., Crutch, S. J., Lees, A. J., & Morris, H. R. (2016). Visual dysfunction in Parkinson’s disease. Brain , 139 (11), 2827-2843. Ravina, B., Marder, K., Fernandez, H. H., Friedman, J. H., McDonald, W., Murphy, D., ... & Goetz, C. (2007). Diagnostic criteria for psychosis in Parkinson's disease: report of an NINDS, NIMH work group. Movement Disorders , 22 (8), 1061-106 Voss, T., Bahr, D., Cummings, J., Mills, R., Ravina, B. & Williams, H. 2013. Performance of a shortened Scale for Assessment of Positive Symptoms for Parkinson's disease psychosis. Parkinsonism Relat Disord, 19, 295-9. Mosimann, U. P., Collerton, D., Dudley, R., Meyer, T. D., Graham, G., Dean, J. L., ... & McKeith, I. G. (2008). A semi‐structured interview to assess visual hallucinations in older people. International Journal of Geriatric Psychiatry: A journal of the psychiatry of late life and allied sciences , 23 (7), 712-718. Qian, X., Liu, Y. I., Xiao, B., Gao, L. I., Li, S., Dang, L., ... & Zhao, L. (2014). The visual mismatch negativity (vMMN): Toward the optimal paradigm. International Journal of Psychophysiology , 93 (3), 311-315. Garrido, M. I., Kilner, J. M., Stephan, K. E., & Friston, K. J. (2009). The mismatch negativity: a review of underlying mechanisms. Clinical neurophysiology , 120 (3), 453-463. Friston, K., FitzGerald, T., Rigoli, F., Schwartenbeck, P., & Pezzulo, G. (2016). Active inference and learning. Neuroscience & Biobehavioral Reviews , 68 , 862-879. Zeidman, P., Jafarian, A., Seghier, M. L., Litvak, V., Cagnan, H., Price, C. J., & Friston, K. J. (2019). A guide to group effective connectivity analysis, part 2: Second level analysis with PEB. Neuroimage , 200 , 12-25. Venables, W. N. & Ripley, B. D. (2002) Modern Applied Statistics with S. Fourth Edition. Springer, New York. ISBN 0-387-95457-0 Wickham et al., (2019). Welcome to the tidyverse. Journal of Open Source Software,4(43), 1686, https://doi.org/10.21105/joss.01686 Wickham H, François R, Henry L, Müller K, Vaughan D (2023). dplyr: A Grammar of Data Manipulation. R package version 1.1.4, https://github.com/tidyverse/dplyr, https://dplyr.tidyverse.org. Kuhn, Max (2008). “Building Predictive Models in R Using the caret Package.” Journal of Statistical Software, 28(5), 1–26. doi:10.18637/jss.v028.i05, https://www.jstatsoft.org/index.php/jss/article/view/v028i05. Rosch, R. E., Auksztulewicz, R., Leung, P. D., Friston, K. J., & Baldeweg, T. (2019). Selective prefrontal disinhibition in a roving auditory oddball paradigm under N-methyl-D-aspartate receptor blockade. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging , 4 (2), 140-150. Ballanger, B., Strafella, A. P., van Eimeren, T., Zurowski, M., Rusjan, P. M., Houle, S., & Fox, S. H. (2010). Serotonin 2A receptors and visual hallucinations in Parkinson disease. Archives of neurology , 67 (4), 416-421. Ramesh, S., & Arachchige, A. S. P. M. (2023). Depletion of dopamine in Parkinson's disease and relevant therapeutic options: A review of the literature. AIMS neuroscience , 10 (3), 200. d’Angremont, E., van der Zee, S., Slingerland, S., Slomp, A. C., de Vries, E. F., van Laar, T., & Sommer, I. E. (2024). Cholinergic deficiency in Parkinson’s disease patients with visual hallucinations. Brain , 147 (10), 3370-3378. Mika Naganawa, Nabeel Nabulsi, Shannan Henry, David Matuskey, Shu-Fei Lin, Lawrence Slieker, Adam J Schwarz, Nancy Kant, Cynthia Jesudason, Kevin Ruley, and others. First-in-human assessment of 11c-lsn3172176, an m1 muscarinic acetylcholine receptor pet radiotracer. Journal of Nuclear Medicine, 62(4):553–560, 2021. Markello, R. D., Hansen, J. Y., Liu, Z. Q., Bazinet, V., Shafiei, G., Suárez, L. E., ... & Misic, B. (2022). Neuromaps: structural and functional interpretation of brain maps. Nature Methods , 19 (11), 1472-1479. Savli M, Bauer A, Mitterhauser M, Ding Y (2012) Normative database of theserotonergic system in healthy subjects using multi-tracer PET. Neuroimage 63:447– 459. Justine Y Hansen, Golia Shafiei, Ross D Markello, Kelly Smart, Sylvia ML Cox, Martin Nørgaard, Vincent Beliveau, Yanjun Wu, Jean-Dominique Gallezot, Étienne Aumont, and others. Mapping neurotransmitter systems to the structural and functional organization of the human neocortex. Nature neuroscience, 25(11):1569–1581, 2022. M Aghourian, C Legault-Denis, JP Soucy, P Rosa-Neto, S Gauthier, A Kostikov, P Gravel, and MA Bedard. Quantification of brain cholinergic denervation in alzheimer’s disease using pet imaging with [18 f]-feobv. Molecular psychiatry, 22(11):1531–1538, 2017 Natalia Jaworska, Sylvia ML Cox, Maria Tippler, Natalie Castellanos-Ryan, Chawki Benkelfat, Sophie Parent, Alain Dagher, Frank Vitaro, Michel Boivin, Robert O Pihl, and others. Extra-striatal d 2/3 receptor availability in youth at risk for addiction. Neuropsychopharmacology, 45(9):1498–1505, 2020. Kaller, S., Rullmann, M., Patt, M., Becker, G. A., Luthardt, J., Girbardt, J., ... & Sabri, O. (2017). Test–retest measurements of dopamine d 1-type receptors using simultaneous pet/mri imaging. European journal of nuclear medicine and molecular imaging , 44 , 1025-1032. Dukart J, Holiga S, Rullmann M, Lanzenberger R, Hawkins PCT, Mehta MA, Hesse S, Barthel H, Sabri O, Jech R, Eickhoff SB. JuSpace: A tool for spatial correlation analyses of magnetic resonance imaging data with nuclear imaging derived neurotransmitter maps. Hum Brain Mapp. 2021 Feb 15;42(3):555-566. doi: 10.1002/hbm.25244. Epub 2020 Oct 20. Burt, J.B., Helmer, M., Shinn, M.W., Anticevic, A., Murray, J.D. Generative modeling of brain maps with spatial autocorrelation. Neuroimage, 220 (2020). Bates D, Mächler M, Bolker B, Walker S (2015). “Fitting Linear Mixed-Effects Models Using lme4.” Journal of Statistical Software , 67(1), 1–48. doi:10.18637/jss.v067.i01. Kuznetsova A, Brockhoff PB, Christensen RHB (2017). “lmerTest Package: Tests in Linear Mixed Effects Models.” Journal of Statistical Software , 82(13), 1–26. doi:10.18637/jss.v082.i13. Moran, R., Pinotsis, D. A., & Friston, K. (2013). Neural masses and fields in dynamic causal modeling. Frontiers in computational neuroscience , 7 , 57. Haarsma, J., Deveci, N., Corbin, N., Callaghan, M. F., & Kok, P. (2023). Expectation cues and false percepts generate stimulus-specific activity in distinct layers of the early visual cortex. Journal of Neuroscience , 43 (47), 7946-7957. Huot, P., Johnston, T. H., Darr, T., Hazrati, L. N., Visanji, N. P., Pires, D., ... & Fox, S. H. (2010). Increased 5‐HT2A receptors in the temporal cortex of parkinsonian patients with visual hallucinations. Movement disorders , 25 (10), 1399-1408. Parr, T., Benrimoh, D. A., Vincent, P., & Friston, K. J. (2018). Precision and false perceptual inference. Frontiers in integrative neuroscience , 12 , 39. Tsukada, H., Fujii, H., Aihara, K., & Tsuda, I. (2015). Computational model of visual hallucination in dementia with Lewy bodies. Neural networks , 62 , 73-82. Kanwisher, N.,McDermott, J., & Chun, M. M. (2002). The fusiform face area: a module inhuman extrastriate cortex specialized for face perception. N.,McDermott, J., & Chun, M. M. (2002). The fusiform face area: a module inhuman extrastriate cortex specialized for face perception. Giussani, C., Riva, M., Gallucci, M., Boukhatem, L., Sganzerla, E. P., Demonet, J. F., & Roux, F. E. (2011). Anatomical correlates for category-specific naming of living and non-living things. Neuroimage , 56 (1), 323-329. Yang, X., Yan, J., Wang, W., Li, S., Hu, B., & Lin, J. (2022). Brain-inspired models for visual object recognition: an overview. Artificial Intelligence Review , 55 (7), 5263-5311. Mosimann, U. P., Rowan, E. N., Partington, C. E., Collerton, D., Littlewood, E., O'Brien, J. T., ... & McKeith, I. G. (2006). Characteristics of visual hallucinations in Parkinson disease dementia and dementia with Lewy bodies. The American Journal of Geriatric Psychiatry , 14 (2), 153-160. Liu, Y., Ouyang, P., Zheng, Y., Mi, L., Zhao, J., Ning, Y., & Guo, W. (2021). A selective review of the excitatory-inhibitory imbalance in schizophrenia: underlying biology, genetics, microcircuits, and symptoms. Frontiers in cell and developmental biology , 9 , 664535. Firbank, M. J., Parikh, J., Murphy, N., Killen, A., Allan, C. L., Collerton, D., ... & Taylor, J. P. (2018). Reduced occipital GABA in Parkinson disease with visual hallucinations. Neurology , 91 (7), e675-e685. Carhart-Harris, R. L., & Nutt, D. J. (2017). Serotonin and brain function: a tale of two receptors. Journal of psychopharmacology , 31 (9), 1091-1120. Ignatavicius, A., Matar, E., & Lewis, S. J. (2024). Visual hallucinations in Parkinson's disease: spotlight on central cholinergic dysfunction. Brain, awae289. Zarkali, A., McColgan, P., Leyland, L. A., Lees, A. J., & Weil, R. S. (2022). Longitudinal thalamic white and grey matter changes associated with visual hallucinations in Parkinson’s disease. Journal of Neurology, Neurosurgery & Psychiatry, 93(2), 169-179. Onofrj, M., Russo, M., Delli Pizzi, S., De Gregorio, D., Inserra, A., Gobbi, G., & Sensi, S. L. (2023). The central role of the Thalamus in psychosis, lessons from neurodegenerative diseases and psychedelics. Translational Psychiatry , 13 (1), 384. Onofrj, M., Espay, A. J., Bonanni, L., Delli Pizzi, S., & Sensi, S. L. (2019). Hallucinations, somatic‐functional disorders of PD‐DLB as expressions of thalamic dysfunction. Movement Disorders , 34 (8), 1100-1111. Additional Declarations There is NO Competing Interest. Supplementary Files dcmarticlesupplement.docx Supplementary Material \"Anomalies in effective connectivity explain different hallucination subtypes in Parkinson’s disease psychosis \" Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {\"props\":{\"pageProps\":{\"initialData\":{\"identity\":\"rs-6627999\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":false,\"archivedVersions\":[],\"articleType\":\"Article\",\"associatedPublications\":[],\"authors\":[{\"id\":467660908,\"identity\":\"2b3c4d92-8d8b-4dfa-8032-845123577c0a\",\"order_by\":0,\"name\":\"Miriam Vignando\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBklEQVRIiWNgGAWjYDACZgaGAyCajQFCykkwMDeA2RLEajGWYGAkoAUVsDEkziCkRbed9+GBnzkMiX38aw9+LiizSZ/ZfrCB4UcNQ+LMBuxazA6zGxzs3caQ2CbxLll6xrm03Nk8iQ2MPccYEmfjsMXsMBvDAd5tDMZsEmcMpHnbDufOkwA6jLeBIXEeHi0H/0K0GP8GakmXA2ph/EtAy2GgLXJs/D1mIFsSpIFamEG24HPYYdltEnJsEjxm1jzn0gxn9iQ2HJY5JmGM0/vnjzF/fLvNhke+/4zxbZ4yG3mJ44cPPnxTYyM74wAOayAAGAcSCQjuAeIikh+/oaNgFIyCUTCCAQAA11PMU2fUtAAAAABJRU5ErkJggg==\",\"orcid\":\"https://orcid.org/0000-0002-7542-3089\",\"institution\":\"King's College London\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Miriam\",\"middleName\":\"\",\"lastName\":\"Vignando\",\"suffix\":\"\"},{\"id\":467660909,\"identity\":\"6bb20813-1759-494c-af32-a138b4def14f\",\"order_by\":1,\"name\":\"Dominic ffytche\",\"email\":\"\",\"orcid\":\"https://orcid.org/0000-0002-4214-9642\",\"institution\":\"Department of Old Age Psychiatry, Institute of Psychiatry, Psychology \\u0026 Neuroscience, King's College London\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Dominic\",\"middleName\":\"\",\"lastName\":\"ffytche\",\"suffix\":\"\"},{\"id\":467660910,\"identity\":\"65ad32bc-aabe-4955-8a3b-e4ea66b59083\",\"order_by\":2,\"name\":\"Ndabezinhle Mazibuko\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"King's College London\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Ndabezinhle\",\"middleName\":\"\",\"lastName\":\"Mazibuko\",\"suffix\":\"\"},{\"id\":467660911,\"identity\":\"d0502b88-b506-406e-8495-c6219d009cfc\",\"order_by\":3,\"name\":\"Giulio Palma\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"University of Southampton\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Giulio\",\"middleName\":\"\",\"lastName\":\"Palma\",\"suffix\":\"\"},{\"id\":467660912,\"identity\":\"5f3ebc43-a1d2-4a2c-8481-4b8e06008ad9\",\"order_by\":4,\"name\":\"Anjali Bhat\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"King's College London\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Anjali\",\"middleName\":\"\",\"lastName\":\"Bhat\",\"suffix\":\"\"},{\"id\":467660913,\"identity\":\"0f3323d4-1605-44c4-b598-3c4de685643d\",\"order_by\":5,\"name\":\"Marcella Montagnese\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"University of Cambridge\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Marcella\",\"middleName\":\"\",\"lastName\":\"Montagnese\",\"suffix\":\"\"},{\"id\":467660914,\"identity\":\"d6c65239-681e-4fe8-8152-483da2a3119c\",\"order_by\":6,\"name\":\"Sonali Dave\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"City, University of London,\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Sonali\",\"middleName\":\"\",\"lastName\":\"Dave\",\"suffix\":\"\"},{\"id\":467660915,\"identity\":\"d8e93f27-29b4-41dd-85e3-bcb847b965ab\",\"order_by\":7,\"name\":\"Yen Tai\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Imperial College London\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yen\",\"middleName\":\"\",\"lastName\":\"Tai\",\"suffix\":\"\"},{\"id\":467660916,\"identity\":\"e0903896-5184-4fd3-9e68-af331952c764\",\"order_by\":8,\"name\":\"Lucia Batzu\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"King's College London\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Lucia\",\"middleName\":\"\",\"lastName\":\"Batzu\",\"suffix\":\"\"},{\"id\":467660917,\"identity\":\"93ee6f71-00a8-410a-8b59-3ce330d83b67\",\"order_by\":9,\"name\":\"Valentina Leta\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Istituto Besta\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Valentina\",\"middleName\":\"\",\"lastName\":\"Leta\",\"suffix\":\"\"},{\"id\":467660918,\"identity\":\"7a6312ee-3856-49ff-b427-0163b5ac4795\",\"order_by\":10,\"name\":\"K. Chaudhuri\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"King's College London\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"K.\",\"middleName\":\"\",\"lastName\":\"Chaudhuri\",\"suffix\":\"\"},{\"id\":467660919,\"identity\":\"cb9ef12f-e531-4cb1-ac70-1571234987db\",\"order_by\":11,\"name\":\"Caroline Williams Gray\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"University of Cambridge\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Caroline\",\"middleName\":\"Williams\",\"lastName\":\"Gray\",\"suffix\":\"\"},{\"id\":467660920,\"identity\":\"a79be3d5-06ac-4518-a4f1-0020207e3e5c\",\"order_by\":12,\"name\":\"Mitul Mehta\",\"email\":\"\",\"orcid\":\"https://orcid.org/0000-0003-1152-5323\",\"institution\":\"King's College London\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Mitul\",\"middleName\":\"\",\"lastName\":\"Mehta\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2025-05-09 11:21:10\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-6627999/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-6627999/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":84392161,\"identity\":\"f6a279e8-f59a-42b7-a8f0-c021781bed53\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 11:43:19\",\"extension\":\"png\",\"order_by\":1,\"title\":\"Figure 1\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":407449,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003e\\u003cstrong\\u003eSummary of the vMMN task and findings. \\u003c/strong\\u003e\\u003c/em\\u003e\\u003cem\\u003eThe overt task required participants to press a button when the fixation cross became bigger and a different button when the fixation cross became smaller. A) Visual summary of a trial (600ms): the fixation cross remained on the screen for the whole duration of the experiment; changes in cross size lasted 200ms; bars flashed peripherally every 50ms, changing in orientation (0, 30 and 60 degrees). B) vMMN ERP for PD with and without psychosis (POZ shown); C) density plots showing distribution of amplitude of MMN in PD and PD-VH at frontal and parieto-occipital channels (PD patients showed overall greater negativity at POZ and mismatch ‘positivity’ at frontal sites at 100-180ms, here shown is electrode F2). PD (bottom row) show a marked negativity at POZ at 100-125ms, whereas PDVH (upper row) show almost no negativity in this time frame (p\\u0026lt;.05). The blue bar in the legend indicates the latencies at which MMN was found; the red bar indicates where frontal positivity was observed.\\u003c/em\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/89b1e59eab1fa7b72a69b049.png\"},{\"id\":84392162,\"identity\":\"9c5e74aa-1170-4851-b79c-65a4dddf19a7\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 11:43:19\",\"extension\":\"png\",\"order_by\":2,\"title\":\"Figure 2\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":640638,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003e\\u003cstrong\\u003eSource reconstruction methods, results and ventral model specification.\\u003c/strong\\u003e\\u003c/em\\u003e\\u003cstrong\\u003e \\u003c/strong\\u003eA) Source\\u003cem\\u003e reconstruction main steps: cortical mesh creation, projection on structural template, co-registration of EEG signal to template, MIP (maximum intensity projection) for each participant, allowing to conduct analyses in sensor space (SI3) and to generate a 3D image for the time window of interest (here 60-400ms). B) Source reconstruction results for the rare deviant trials are shown (p \\u0026lt;.001 uncorrected in the top row and pFWE corrected \\u0026lt;.05 in the bottom row). The one sample t-test for vMMN in the whole sample for signal source reconstruction for the rare deviant, on images generated for the 60-400ms interval (p \\u0026lt;.001, height threshold F = 11.74, df (1,74) k \\u0026gt; 200) returned clusters in the inferior and anterior temporal gyrus, bilaterally and in lateral occipital regions (the occipital-calcarine cluster remains significant after FWE correction, p =.026 cluster , p =.031 peak). We also found a large medial thalamic cluster (k = 725, p\\u0026lt;. 001 uncorrected, however it did not survive multiple comparisons correction pFWE = .07). We also conducted paired-samples t-test in each group separately comparing deviant with standard, which returned a significant peak (k=145, p \\u0026lt;.001 uncorrected, pFWE =.014) in the bilateral calcarine cortex, mirroring the results of the standard and deviant source reconstruction analysis (Figure 3B; see SI4 for coordinates). For the PD-VH group, a significant peak (k=83, p \\u0026lt;.001 uncorrected, pFWE =.021) was found in the left ITG (fusiform region) (See SI4). C) Dipole locations selected based on the results presented in (A) here presented on a template of the brain. Dipole coordinates: (V1 left –10, –82, –0.5, right 10, –82, –0.5; left ITG –44, -4, –34; right 44, –4, –34; left PFC –38, 33, 35, right 38, 33, 35). D) Schematic of model connections: AF forward connections in blue, AB backward connections in fuchsia, AL cross-hemispheric connections in orange; input (C matrix). Connections included in the task matrix (B matrix) for winning model v3: forward connections from V1 to ITG, V1 to PFC, ITG to PFC; backward connections from ITG to V1, PFC to ITG, PFC-V1 were specified. Lateral (intra-hemispheric) connections between regions were also specified between left and right V1, ITG and PFC. Green circles represent connectivity on the same region, testing for self-connectivity during task condition. E) Visual representation of microcircuit underlying the model as described in Moran et al\\u003c/em\\u003e\\u003csup\\u003e\\u003cem\\u003e59\\u003c/em\\u003e\\u003c/sup\\u003e\\u003cem\\u003e.\\u003c/em\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/67ed23a9feef08f986ddaee4.png\"},{\"id\":84392165,\"identity\":\"8ee19b38-c7a9-4a42-af53-7d6745951e31\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 11:43:19\",\"extension\":\"png\",\"order_by\":3,\"title\":\"Figure 3\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":466833,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003e\\u003cstrong\\u003ePEB results.\\u003c/strong\\u003e\\u003c/em\\u003e\\u003cem\\u003e A-E. Condition specific effects (vMMN): group differences. A) bar plot of estimated connection strength (Ep) for each connection surviving the posterior probability pp\\u0026gt;.99 (free energy) threshold. B) Connectivity matrix of results presented in (A). The matrix shows how top-down effective connectivity is reduced vs increased bottom up in VH in this task. C) Schematic of BMA results with PEB values. Yellow = increased strength in PD-VH; purple = decreased strength in PD-VH (negative values correspond to increased connectivity in PD-VH as the PD-VH group was classified with -1 in the design matrix, opposed to 1 for PD). The increased strength in the self-connection in left V1 indicates less self-inhibition. D) Graphical representation of the connections, colour coded as in a) and b). The brain and nodes have been generated with BrainNet viewer. E) box and whisker and point plots showing the distribution of Ep in PD with and without VH for each connection found differing in the model (the plots are colour coded with the colour of the node from which the connection starts; noVH patients = black dots; VH patients = white dots; the boxplots indicate the mean and the quantiles). F,G): Baseline A matrix: F, a graphical representation of the connections, colour coded as in A) and B). The brain and nodes have been generated with BrainNet viewer; G, connectivity differences in PD-VH and PD: bar plot of connection strength (Ep) for each connection surviving the pp\\u0026gt;.99 (free energy) threshold. Yellow = increased connectivity in PD-VH; purple = decreased connectivity in PD-VH. Connectivity is overall increased in PD-VH in the connections modelled. The BMA for matrix A (forward, feedback and inter-hemispheric latent connectivity) shows that at baseline PD psychosis brain activity is positively correlated with connectivity from PFC to V1 bilaterally, from ITG to V1 bilaterally, between V1 (left to right and right to left), from left to right ITG, and from left V1 to left PFC. In addition, PD-VH activity negatively correlated with connectivity from left PFC to left ITG and from right V1 to right PFC (all with pp \\u0026gt;.99).\\u003c/em\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"3.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/36a5d905a7cabc9bb8b136dd.png\"},{\"id\":84392164,\"identity\":\"f5212e50-18c2-44f0-ae01-6031836f246a\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 11:43:19\",\"extension\":\"png\",\"order_by\":4,\"title\":\"Figure 4\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":367376,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003e\\u003cstrong\\u003eHallucinations phenomenology regression models.\\u003c/strong\\u003e\\u003c/em\\u003e\\u003cem\\u003e\\u0026nbsp;A). Sankey plot showing the proportion of\\u0026nbsp;participants (out of 20 PD-VH) having each specific hallucination subtype. Complex, Presence, Passage,Pareidolia and Simple VH scores are derived from the NEVHI (SI1); multimodal hallucination scores are\\u0026nbsp;derived from the SAPS-PD (SI1) and the modality of multimodal hallucination is further subtyped in D in this figure. B) Linear regression analyses on CVH for the Matrix B. C) Matrix B linear regression for MH score (passage, presence and illusions summed). Yellow= increased connectivity in the PEB in PD-VH; purple = decreased connectivity; Ep = estimated connection strengths. D) Multimodal score (derived from SAPS-PD information) regression analyses. Left: Matrix A and multimodal hallucinations individual regression. Right: connections significantly predicting multimodal hallucination score in the LOOCV analysis. Bottom row: below the multimodal scatter plot, a mosaic plot shows the proportion of participants with hallucinations in each specific modality (light blue = symptom present; dark blue = not present). Reported in all the plots are the t values and the individual regression uncorrected and corrected (FDR) p values as described. \\u0026nbsp;Complete details are provided in SI7; SI8.\\u003c/em\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"4.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/de27697834e06026b8e7984f.png\"},{\"id\":84392168,\"identity\":\"402c8d60-2af7-4257-a3c2-a2a379109b60\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 11:43:20\",\"extension\":\"png\",\"order_by\":5,\"title\":\"Figure 5\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":536587,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003e\\u003cstrong\\u003eIntrinsic coupling modulation simulations for effect of VH results. \\u003c/strong\\u003e\\u003c/em\\u003e\\u003cem\\u003eA) Schematic of the procedure used: after running the recursive PEB analysis described above, we operationalised VH effect\\u003c/em\\u003e\\u003cem\\u003e\\u003cstrong\\u003e \\u003c/strong\\u003e\\u003c/em\\u003e\\u003cem\\u003eas the contrast between Parkinson’s patients with and without visual hallucinations (PD-VH vs. PD-noVH), encoded as the second column of the design matrix used in the PEB model. We used the corresponding group-level effect estimate from the PEB to scale the perturbation of inhibitory-excitatory coupling parameters in the simulations. B) Estimated intrinsic coupling parameter changes with posterior probability pp\\u0026gt;.99 of the parameter being affected by group; results of the recursive PEB intrinsic parameters exploration. Significant changes were observed in some but not all the coupling parameters, in particular for excitatory to inhibitory cells (blue) and inhibitory to excitatory (pink). PD-VH had reduced excitatory and inhibitory activity in these regions if compared to PD. The regions with the largest reduction in intrinsic coupling are V1 and ITG for inhibitory, and left ITG for excitatory activity. c) Time series plots of excitatory and inhibitory activity resulting from the simulation of increase in the parameter regulating the strength of the E/I coupling (pink = inhibitory; blue = excitatory; dashed line = PD, solid line = PD-VH). The AUC analysis reveals significant differences for all regions in both excitatory and inhibitory activity in all regions during the MMN latencies (100-250ms); PD-VH show a particularly smaller AUC in inhibitory activity in left V1 and left IFG\\u003c/em\\u003e\\u003cem\\u003e\\u003cstrong\\u003e. \\u003c/strong\\u003e\\u003c/em\\u003e\\u003cem\\u003e\\u003cstrong\\u003eD)\\u003c/strong\\u003e\\u003c/em\\u003e\\u003cem\\u003eSimulated activity heatmaps for simulated excitatory and inhibitory cell populations in each region, x-axis = time; y-axis = incremental modulation of inhibitory–excitatory coupling. The colour indicates population-level firing amplitude, yellow = high, blue = low. Shown is the effect of VH, where inhibitory activity in V1 and ITG shows early and sustained attenuation, particularly around MMN latencies (100–250 ms). Excitatory responses follow a similar pattern, especially in left V1, suggesting a possible dampened responsiveness to input\\u003c/em\\u003e. \\u003cem\\u003eE) State space plots: blue = low perturbation, red = high perturbation; top row, PD-VH, bottom row PD (noVH). In PD without VH the trajectories follow smooth and non-convoluted loops, with changes occurring only at higher values of the coupling parameter modulation, whereas PD-VH show a very complex trajectory with multiple loops. When the parameter regulating the strength of the inhibitory/excitatory coupling is increased in a stepwise manner, PD-VH state space plots show a higher variability (loops), suggesting that the system is more sensitive to changes. E/I balance appears particularly disrupted in the left ITG in all analyses, consistent with the deviant-standard estimated neural activity in the ITG ~ complex VH severity analysis presented in the previous section.\\u003c/em\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"5.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/3af0d9d4ace315ea1ba41fe1.png\"},{\"id\":84392171,\"identity\":\"b13ce0d2-2574-41a5-b211-01ba6d92f839\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 11:43:20\",\"extension\":\"png\",\"order_by\":6,\"title\":\"Figure 6\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":471651,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003e\\u003cstrong\\u003eSource-reconstructed signal ~ receptor binding atlases exploratory analysis\\u003c/strong\\u003e\\u003c/em\\u003e.\\u003cem\\u003e a) Methods. As described in the methods section and in SI13, we parcellated the PET atlases and the source reconstructed EEG signal with the Desikan-Killiany atlas (converted to MNI template space); atlases and individual participant data was aligned to the brain atlas and registration was checked examining PET atlas and brain atlas size and transformation matrix and visually with checkreg prior to data extraction. Parcellated data was then used to run linear regression models (one per receptor) on a subset of regions of the Desikan-Killiany atlas (cerebellar, white matter and ventricular regions were removed) to identify and remove outliers using Cook’s distance. P values from the regressions\\u003c/em\\u003e \\u003cem\\u003ewere corrected for multiple comparisons. Then the BrainSMASH toolbox\\u003c/em\\u003e\\u003csup\\u003e\\u003cem\\u003e56\\u003c/em\\u003e\\u003c/sup\\u003e\\u003cem\\u003e was used to generate MNI coordinates for the Desikan-Killiany atlas; we removed the cerebellar and ventricular regions and the regions that were marked as outliers from the regression models before proceeding to compute the Pearson’s actual correlations of the two maps.\\u0026nbsp; Only the maps surviving this threshold were finally entered in linear mixed models, which had the purpose of examining individual-level data allowing to take age, LEDD, MoCA score, disease onset into account. \\u0026nbsp;b) Linear regressions results; dark blue = models with negative t value, pink = models with positive t value; in the box on the top right of each scatter plot, the statistics from the linear regression model are reported, together with the actual correlation coefficient from the spatial autocorrelation analysis and the final FDR corrected p value. The PET map images were rendered by overlapping the Desikan-Killiany atlas and the relevant PET map; the colours were arbitrarily chosen to highlight the binding distribution in the different maps. c) LMM results. x axis: mean centred receptor density; y axis: mean centred MMN signal; we mean centred the data as the scale of the different predictors varied greatly. Blue line: PD-VH; red line: PD-noVH.\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"6.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/92bf4d27d9652e2a1744134a.png\"},{\"id\":84392172,\"identity\":\"4d504e7c-4c39-44d1-92b6-1155b36c7b8d\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 11:43:20\",\"extension\":\"png\",\"order_by\":7,\"title\":\"Figure 7\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":231742,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cem\\u003e\\u003cstrong\\u003eSummary of results of individual regression models and LOOCV machine learning models relating VH severity to effective connectivity in PD-VH.\\u003c/strong\\u003e\\u003c/em\\u003e\\u003cem\\u003e CVH were best predicted by all the right-hemisphere connections found related to CVH severity (rV1-rPFC (increased), rPFC-rITG (decreased), supporting the view that these regions operate as a network and the best contributors of the LOOCV model for task activity are the increased connectivity between V1 (in both directions) and rPFC to rV1. We found that minorVH were related to top-down left and cross-hemispheric (left to right) connectivity. Multimodal hallucinations only correlated with baseline/latent widespread changed in connectivity.\\u003c/em\\u003e\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"7.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/ee7c0a6bc16d7acede1b3ccc.png\"},{\"id\":84393616,\"identity\":\"bec4b1f1-aed1-4f8f-a97d-03d13e19fb6e\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 12:07:21\",\"extension\":\"pdf\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"manuscript-pdf\",\"size\":4302961,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"manuscript.pdf\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/d64bbffc-4701-484c-a801-e14bbbcde1ae.pdf\"},{\"id\":84392179,\"identity\":\"88f47ec2-6113-4006-9687-a60df40dd479\",\"added_by\":\"auto\",\"created_at\":\"2025-06-11 11:43:20\",\"extension\":\"docx\",\"order_by\":1,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":29669287,\"visible\":true,\"origin\":\"\",\"legend\":\"Supplementary Material \\\"Anomalies in effective connectivity explain different hallucination subtypes in Parkinson\\u0026#x2019;s disease psychosis \\\"\",\"description\":\"\",\"filename\":\"dcmarticlesupplement.docx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6627999/v1/00e996483355519ab7f74027.docx\"}],\"financialInterests\":\"There is \\u003cb\\u003eNO\\u003c/b\\u003e Competing Interest.\",\"formattedTitle\":\"Anomalies in effective connectivity explain different hallucination subtypes in Parkinson’s disease psychosis\",\"fulltext\":[{\"header\":\"Introduction\",\"content\":\"\\u003cp\\u003eParkinson\\u0026rsquo;s disease (PD) is characterised by motor and non-motor symptoms\\u003csup\\u003e\\u003cspan citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e1\\u003c/span\\u003e\\u003c/sup\\u003e, among which psychosis and visual hallucinations (VH) notably impact patient outcomes and are associated with cognitive decline\\u003csup\\u003e\\u003cspan additionalcitationids=\\\"CR3\\\" citationid=\\\"CR2\\\" class=\\\"CitationRef\\\"\\u003e2\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e4\\u003c/span\\u003e\\u003c/sup\\u003e. Although morphological\\u003csup\\u003e5\\u0026thinsp;\\u0026minus;\\u0026thinsp;3-5\\u003c/sup\\u003e and functional \\u003csup\\u003e\\u003cspan citationid=\\\"CR8\\\" class=\\\"CitationRef\\\"\\u003e8\\u003c/span\\u003e,\\u003cspan citationid=\\\"CR9\\\" class=\\\"CitationRef\\\"\\u003e9\\u003c/span\\u003e\\u003c/sup\\u003e alterations differentiate PD patients with VH (PD-VH) from those without VH, standard cognitive assessments lack mechanistic insights into VH. A recent review of frameworks\\u003csup\\u003e\\u003cspan citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e10\\u003c/span\\u003e\\u003c/sup\\u003e converges on the proposal that \\u0026lsquo;ascending\\u0026rsquo; sensory disturbances together with \\u0026lsquo;descending\\u0026rsquo; factors (e.g. expectations, specific object features, attention) contribute to VH. Among the different proposals, predictive coding, a cognitive model of psychosis, suggests hallucinations arise from disrupted balance between top-down expectations and bottom-up sensory inputs, with a stronger influence of expectations observed in people with hallucinations \\u003csup\\u003e\\u003cspan additionalcitationids=\\\"CR12\\\" citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e11\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR13\\\" class=\\\"CitationRef\\\"\\u003e13\\u003c/span\\u003e\\u003c/sup\\u003e. The ERP component mismatch negativity (MMN), a robust marker of psychosis\\u003csup\\u003e\\u003cspan citationid=\\\"CR14\\\" class=\\\"CitationRef\\\"\\u003e14\\u003c/span\\u003e\\u003c/sup\\u003e, is thought to reflect an attempt to minimise this prediction error and capture its disruption\\u003csup\\u003e\\u003cspan additionalcitationids=\\\"CR16 CR17\\\" citationid=\\\"CR15\\\" class=\\\"CitationRef\\\"\\u003e15\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR18\\\" class=\\\"CitationRef\\\"\\u003e18\\u003c/span\\u003e\\u003c/sup\\u003e. We recently demonstrated reduced visual MMN responses in PD-VH \\u003csup\\u003e\\u003cspan citationid=\\\"CR19\\\" class=\\\"CitationRef\\\"\\u003e19\\u003c/span\\u003e\\u003c/sup\\u003e, paralleling auditory MMN impairments observed in psychosis\\u003csup\\u003e\\u003cspan citationid=\\\"CR14\\\" class=\\\"CitationRef\\\"\\u003e14\\u003c/span\\u003e\\u003c/sup\\u003e. However, traditional ERP analysis cannot uncover the directional neural mechanisms involved. Dynamic causal modelling (DCM), leveraging EEG\\u0026rsquo;s temporal precision, enables exploration of these mechanisms \\u003csup\\u003e\\u003cspan citationid=\\\"CR20\\\" class=\\\"CitationRef\\\"\\u003e20\\u003c/span\\u003e,\\u003cspan citationid=\\\"CR21\\\" class=\\\"CitationRef\\\"\\u003e21\\u003c/span\\u003e\\u003c/sup\\u003e. DCM has proved instrumental in clarifying the disruption of the MMN in people with schizophrenia-related psychosis\\u003csup\\u003e\\u003cspan citationid=\\\"CR22\\\" class=\\\"CitationRef\\\"\\u003e22\\u003c/span\\u003e\\u003c/sup\\u003e and a spectral DCM study (resting-state fMRI) found increased top-down and reduced bottom-up connectivity in the visual network in PD with VH, a pattern was predictive of hallucination severity\\u003csup\\u003e\\u003cspan citationid=\\\"CR8\\\" class=\\\"CitationRef\\\"\\u003e8\\u003c/span\\u003e\\u003c/sup\\u003e. Here, we employ DCM with neural microcircuit models across dorsal (V1, IPL) and ventral (V1, ITG) visual pathways interacting with the prefrontal cortex (PFC) during a visual MMN task, to investigate their relation to hallucination complexity. We include the PFC to test the prediction error impairment as top-down connectivity is crucial for integrating sensory information\\u003csup\\u003e\\u003cspan additionalcitationids=\\\"CR24\\\" citationid=\\\"CR23\\\" class=\\\"CitationRef\\\"\\u003e23\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR25\\\" class=\\\"CitationRef\\\"\\u003e25\\u003c/span\\u003e\\u003c/sup\\u003e and we focus on the dorsal and ventral pathways based on the literature\\u003csup\\u003e\\u003cspan citationid=\\\"CR19\\\" class=\\\"CitationRef\\\"\\u003e19\\u003c/span\\u003e,\\u003cspan additionalcitationids=\\\"CR27 CR28 CR29\\\" citationid=\\\"CR26\\\" class=\\\"CitationRef\\\"\\u003e26\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR30\\\" class=\\\"CitationRef\\\"\\u003e30\\u003c/span\\u003e\\u003c/sup\\u003e and our MMN task changing line orientation, a process known to involve specific parietal regions \\u003csup\\u003e\\u003cspan citationid=\\\"CR31\\\" class=\\\"CitationRef\\\"\\u003e31\\u003c/span\\u003e\\u003c/sup\\u003e. We hypothesise reduced top-down and enhanced bottom-up connectivity in PD-VH, reflecting impaired prediction updating and reliance on compromised sensory information, as visual input is known to be defective in PDP\\u003csup\\u003e\\u003cspan citationid=\\\"CR32\\\" class=\\\"CitationRef\\\"\\u003e32\\u003c/span\\u003e\\u003c/sup\\u003e. Additionally, exploratory analyses relate MMN-derived source activity to cortical distributions of receptors, previously implicated in PD-VH.\\u003c/p\\u003e\"},{\"header\":\"Methods and Materials\",\"content\":\"\\u003cp\\u003e\\u003cstrong\\u003e\\u003cem\\u003eData and code availability statement.\\u003c/em\\u003e\\u003c/strong\\u003e\\u003cem\\u003e\\u0026nbsp;All the code developed for the study is available here:\\u0026nbsp;\\u003c/em\\u003e\\u003cem\\u003ehttps://github.com/VMiri/vMMN-dcm\\u003c/em\\u003e\\u003cem\\u003e. ERP data after artefact removal and averaging is available in the same repository in the .mat and .dat formats.. The main ERP study was pre-registered ( https://osf.io/q9x7v\\u003c/em\\u003e\\u003cem\\u003e).\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eParticipants.\\u0026nbsp;\\u003c/em\\u003eThe study received ethics approval from London Camberwell St Giles REC (18/LO/2144). We enrolled 18 patients with PD without hallucinations (PD-noVH) and 20 patients meeting criteria for PDP (PD-VH in this article)\\u003csup\\u003e33\\u003c/sup\\u003e. Participant recruitment and screening details are provided in detail in \\u003cstrong\\u003eSupplementary Information S1\\u0026nbsp;\\u003c/strong\\u003eand in our ERP study\\u003csup\\u003e19\\u003c/sup\\u003e.\\u003c/p\\u003e\\n\\u003cp\\u003eVH and psychosis in PD-VH were further assessed using the Scale for the Assessment of Positive Symptoms-PD (SAPS-PD)\\u003csup\\u003e\\u0026nbsp;34\\u003c/sup\\u003e, and an expanded version of the North-East Visual Hallucination Interview (NEVHI) \\u003csup\\u003e35\\u003c/sup\\u003e to examine the phenomenology of visual hallucinations and their subtypes in PD-VH. (see \\u003cstrong\\u003eSI1\\u003c/strong\\u003e).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eTask.\\u0026nbsp;\\u003c/em\\u003eWe used a visual mismatch negativity task whereby the stimulus frequency variation was the orientation of peripheral bars (SI2). \\u003cstrong\\u003eFigure 1a\\u0026nbsp;\\u003c/strong\\u003eprovides a visual summary of the vMMN task; the task design was inspired by vMMN paradigms proposed in Qian et al.\\u003csup\\u003e36\\u003c/sup\\u003e; details are provided in SI2 and the ERP analysis is described in detail in Vignando et al\\u003csup\\u003e19\\u003c/sup\\u003e.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003e\\u003cstrong\\u003eStatistical analysis.\\u0026nbsp;\\u003c/strong\\u003e\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eParticipant demographics.\\u003c/em\\u003e PD-VH and PD did not differ on age, sex, disease duration, MMSE score, levodopa equivalent daily dose (LEDD) and motor severity \\u0026nbsp;(\\u003cstrong\\u003eSupplementary SI1)\\u003c/strong\\u003e. None of the participants were on quetiapine. Two PD-VH and two PD patients were on SSRIs.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eEEG data pre-processing.\\u0026nbsp;\\u003c/em\\u003eEEG data was pre-processed with the standard spm12 pipeline (see SI2) for ERP analysis\\u003csup\\u003e37\\u003c/sup\\u003e\\u003cem\\u003e.\\u0026nbsp;\\u003c/em\\u003eIn brief, we created a custom montage file to exclude eye channels and a\\u003cem\\u003e\\u0026nbsp;\\u003c/em\\u003etrial definition file for the interval \\u0026ndash;100 to 500 (the duration of the task) and for the two conditions of interest (standard, rare deviant), with shift triggers = 0. We used a high-pass filter with cutoff = 0.3, then we down sampled to 500 Hz and lowpass filter with cutoff = 30. \\u0026nbsp;The trial definition file was used for epoching and baseline correction was applied pre-stimulus. After epoching, bad trials and channels were visually inspected, marked and then artefacts were removed with threshold set at 80\\u003csup\\u003e37\\u003c/sup\\u003e. ERPs were then averaged for standard and rare deviant conditions (seeSI2).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eSensory space and Source reconstruction analysis. \\u0026nbsp;\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eWe conducted analyses in sensor space to identify the electrodes where the ERP amplitude differed across conditions during the task and to validate the results from our ERP study\\u003csup\\u003e19\\u003c/sup\\u003e (see \\u003cstrong\\u003eSI3\\u003c/strong\\u003e). To gain more detailed spatial information regarding the source of our signal, we perform source reconstruction (\\u003cstrong\\u003eSI4\\u003c/strong\\u003e) for the 60-400ms interval of interest to be investigated with the DCM analyses and we entered them in second level analyses (one-sample t-test; paired t-test) to investigate the neural correlates of task-related activity to guide the choice of the dipoles.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eDynamic Causal Modelling: Individual DCM model specification.\\u003c/em\\u003e Based on the results of the ERP study\\u003csup\\u003e19\\u003c/sup\\u003e and of the source reconstruction analyses, pointing to the involvement of occipital, occipito-parietal and frontal regions, we decided to investigate the ventral and the dorsal pathways separately. \\u0026nbsp;For the ventral pathway, we designed three models: all three models share the same dipoles, to allow for Bayesian model selection, but different neuronal models (Supplementary Figure 3). Since we did not have strong results for the dorsal pathway in the source reconstruction analysis we focussed on one dorsal model, using source reconstruction results, and the literature on the Benton line judgement task and the neural anatomy of the dorsal pathway\\u003csup\\u003e31\\u003c/sup\\u003e to define the dipoles (SI11). We used an \\u003cem\\u003eERP\\u003c/em\\u003e neural mass model, focussing on the 0 to 400ms interval, rare deviant vs. standard. An equivalent current dipole (ECD) was used for each source, with anterior occipital, inferior temporal and dorsolateral prefrontal regions, bilaterally (Figure 2c). All model specification details are provided in SI5, Supplementary Figure 3 and SI11 and Bayesian model comparisons are detailed in Supplementary Figure 4. \\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003e2\\u003csup\\u003end\\u003c/sup\\u003e level analysis: Parametrical empirical Bayes (PEB)\\u003c/em\\u003e. After specifying individual DCMs for each \\u003cem\\u003eparticipant\\u003c/em\\u003e, PEB \\u003csup\\u003e38,39\\u0026nbsp;\\u003c/sup\\u003ewas applied to test for group differences in effective connectivity. PEB uses a Bayesian linear regression approach, which allows group-level comparisons of modulatory parameters estimated at the individual level, without losing information about the precision of those estimations. We fitted a general linear model (GLM) to the connection strengths estimated with the first level analyses and with covariates: mean, group (PD-VH (-1) or PD (1)) and age. \\u003cem\\u003espm_dcm_bmr\\u003c/em\\u003e and \\u003cem\\u003espm_dcm_peb\\u003c/em\\u003e were used to perform Bayesian model reduction and to generate model posteriors, respectively. We used Bayesian model averaging using \\u003cem\\u003espm_dcm_bma\\u003c/em\\u003e to investigate group-level differences between patients with and without VH. We analysed the different B \\u0026ldquo;task modulatory\\u0026rdquo; matrix separately and A \\u0026ldquo;baseline\\u0026rdquo; matrix (feedforward, feedback, inter-hemispheric connectivity). To understand which of the parameters contributed best to explaining variance between DCMs, we also created a model space to carry out PEB analyses recursively with specific parameters specified, allowing switching of specific parameters on and off and then comparing the models (SI6, Supplementary Figure 7). Bayesian model comparison at the second level was then explored and the parametrised winning 2\\u003csup\\u003end\\u003c/sup\\u003e level model was further examined with Bayesian model reduction (\\u003cem\\u003eBMR\\u003c/em\\u003e).\\u0026nbsp;BMR allows to take a hypothesis-free approach to finding the winning model for effective connectivity within a network.\\u0026nbsp;We also explored task connectivity in PD-noVH to isolate the effect of task, as we previously showed that this group had a stronger MMN if compared to PD-VH (\\u003cstrong\\u003eSupplementary Figure 6\\u003c/strong\\u003e).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eHallucination subtypes: individual connections\\u0026nbsp;\\u003c/em\\u003e\\u003cem\\u003elinear regression\\u003c/em\\u003e\\u003cem\\u003e\\u0026nbsp;analyses\\u003c/em\\u003e. We conducted linear regression analyses between NEVHI scores for visual complex (CVH) and minor hallucinations (MH) (for the PD-VH group) and the estimated connection strength (Ep) for each of the connections found (pp\\u0026gt;.99) with the PEB analysis. As we also had collected information on multimodality of hallucinations in our patients from the SAPS-PD (see \\u003cstrong\\u003eSI1\\u003c/strong\\u003e), we explored the relationship between connectivity in PD-VH and multimodality of hallucinations. Outliers were identified with Cook\\u0026rsquo;s distance\\u003csup\\u003e40\\u003c/sup\\u003e, inspected, and removed. We computed multiple comparisons correction using false discovery rate across all individual models ran. R package \\u003cem\\u003eTidyverse\\u003c/em\\u003e\\u003csup\\u003e41\\u003c/sup\\u003e and \\u003cem\\u003edplyr\\u003c/em\\u003e\\u003csup\\u003e42\\u0026nbsp;\\u003c/sup\\u003ewere used for these analyses (see SI7).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eHallucination subtypes:\\u0026nbsp;\\u003c/em\\u003e\\u003cem\\u003eLeave one out multiple regression models.\\u0026nbsp;\\u003c/em\\u003eAs we found more than one connection correlating with the severity of VH, we performed post-hoc backwards regression analyses with leave one out cross-validation (using R package \\u003cem\\u003ecaret\\u003c/em\\u003e\\u003csup\\u003e43\\u003c/sup\\u003e). We also ran models including only the individual connections found with the individual regression analyses described in the previous paragraph to explore whether the model with the multiple connections was improved for predicting simple and complex VH (severity scores) (separately). We used BIC (Bayesian Information Criterion) to select the best model and avoid overfitting. We also computed the same models with MoCA and then LEDD as covariates to accommodate the potential role of general cognition and medication (SI8).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eAnalysis of neural sources.\\u003c/em\\u003e We explored the estimated activity of the three neural populations described in the microcircuit model (ERP neural mass model: spiny stellate cells, inhibitory interneurons, deep pyramidal neurons) by retrieving the estimated values from \\u003cem\\u003eDCM.M.dipfit.H\\u003c/em\\u003e and used to better understand the deficit in the PD-VH group for the difference between deviant and standard in relation to the hallucination severity. We used R packages dplyr\\u003csup\\u003e42\\u003c/sup\\u003e and caret\\u003csup\\u003e43\\u003c/sup\\u003e to perform leave-one-out cross-validation (LOOCV) with linear regression implemented via the caret package\\u003csup\\u003e43\\u003c/sup\\u003e in R (method = \\u0026apos;lm\\u0026apos;) to test if these estimates could significant predict VH severity. We specify VH scores as the dependent variable and activity in a specific neural population and region as an independent predictor. For each iteration, one participant was excluded from model training and used as the test case. We used LOOCV since we had a limited sample size and using the linear regression method implemented in caret allows to maximize data utilization and to minimise overfitting (see SI9).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eExcitatory/inhibitory (E/I) coupling simulations.\\u003c/em\\u003e We conducted this analysis to probe the network-level mechanisms and to understand more clearly the role of intrinsic coupling in the results pertaining to the previous sections and on the PEB analysis where specific parameter changes were found to be associated to the inhibitory to excitatory coupling. \\u0026nbsp;The aims was to modulate the intrinsic coupling parameter and examine how connectivity changes with VH. We did so by adapting and tailoring the methods and procedures described in Rosch et al\\u003csup\\u003e44\\u003c/sup\\u003e to our group contrast and our neural mass model (all details in SI10.) \\u0026nbsp;Briefly, we modulated the intrinsic coupling parameter to simulate the graded effects of VH on our cortical microcircuits. We started from the grand mean model and individual participants inversions and applied VH-related perturbations estimated from the PEB analysis. We extracted the VH contrast from the design matrix, and scaled the \\u003cstrong\\u003eG\\u003c/strong\\u003e parameter accordingly across 10 steps (range: 0\\u0026ndash;0.5)\\u003csup\\u003e15\\u003c/sup\\u003e, applying these to all six cortical regions in the model (bilateral V1, ITG, and PFC), with each step incrementally increasing the strength of G proportionally to the magnitude of the PEB-derived group difference. The purpose was to simulate biologically and physiologically plausible perturbations to explore how network dynamics evolve under increasing psychosis load.\\u003c/p\\u003e\\n\\u003cp\\u003eWe then used \\u003cem\\u003espm_gen_erp\\u003c/em\\u003e to generate predicted neural responses for each parameter configuration. For each simulation, we extracted excitatory and inhibitory time-series data from all regions and neural populations. We examined both time-domain activity (e.g. peristimulus firing rate heatmaps) and state-space trajectories (excitatory vs. inhibitory activity over time). To quantify possible differences, we computed the area under the curve (AUC) for inhibitory and excitatory activity of the time series using a trapezoidal method. \\u0026nbsp; We also computed state-space area as a measure of network flexibility. All details are provided in SI10. These analyses were carried out in MATLAB 2023b.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eExploratory receptor binding atlases and MMN source reconstructed signal analysis\\u003c/em\\u003e. The aim of this analysis was to explore a possible relationship between vMMN signal and neurotransmitter systems implicated in PD with VH, namely the serotonergic\\u003csup\\u003e45\\u003c/sup\\u003e, dopaminergic\\u003csup\\u003e46\\u003c/sup\\u003e and cholinergic\\u003csup\\u003e47\\u003c/sup\\u003e systems. First, we computed vMMN difference images by subtracting standard from rare deviant activity for each participant, with each image then registered to the atlas using the \\u003cem\\u003eimcalc\\u003c/em\\u003e function. We performed PET maps and atlas alignment; figures resulting from this procedure are provided in SI13. A custom script was created to parcellate the source-reconstructed difference images and the PET maps of interest (11c-lsn3172176\\u003csup\\u003e48\\u003c/sup\\u003e for muscarnic M1, [\\u003csup\\u003e18\\u003c/sup\\u003eF]altanserin\\u003csup\\u003e50\\u0026nbsp;\\u003c/sup\\u003efor 5-HT2A, [\\u003csup\\u003e18\\u003c/sup\\u003eF]FEOBV \\u003csup\\u003e51,52\\u003c/sup\\u003e for VAChT, [\\u003csup\\u003e18\\u003c/sup\\u003eF]fallypride\\u003csup\\u003e\\u0026nbsp;53\\u0026nbsp;\\u003c/sup\\u003efor dopamine D2 receptors and\\u003csup\\u003e\\u0026nbsp;\\u003c/sup\\u003e[\\u003csup\\u003e11\\u003c/sup\\u003eC]SCH23390\\u003csup\\u003e\\u0026nbsp;54\\u0026nbsp;\\u003c/sup\\u003efor dopamine D1 receptors). This allowed for the extraction of regional binding potential (BP\\u003csub\\u003eND\\u003c/sub\\u003e) for each PET map and the corresponding reconstructed EEG signal within the regions defined by the atlas. First, we ran individual regression models using regional receptor binding as a predictor and regional MMN signal difference (VH \\u0026ndash; noVH) as dependent variable. Outliers were identified with Cook\\u0026rsquo;s distance\\u003csup\\u003e43\\u0026nbsp;\\u003c/sup\\u003eand removed (details and results provided in SI13). \\u003cem\\u003eP\\u003c/em\\u003e values for these models were corrected for multiple comparisons. We proceeded with further analysis including only the maps surviving multiple comparisons correction. We accounted for spatial autocorrelation using the BrainSMASH toolbox\\u003csup\\u003e56\\u003c/sup\\u003e to generate the MNI coordinates for our atlas. We removed the outlier regions we had removed with the regression models, to run the correlational analyses on the exact same regions on which we ran the regressions. By creating centroids from the .nii atlas and running multiple permutations, we obtained a correlation coefficient that is proposed, by Burt and colleagues, to reflect the relationship between the two maps free from spatial autocorrelation. For the maps surviving this \\u0026lsquo;actual\\u0026rsquo; correlational analysis, we carried out linear mixed models using individual-level data, to investigate the effect of age, MoCA score, LEDD and disease onset in this relationship using lme4\\u003csup\\u003e57\\u003c/sup\\u003e and lmerTest\\u003csup\\u003e58\\u003c/sup\\u003e. Further details on the methods and the models are provided in SI12.\\u003c/p\\u003e\"},{\"header\":\"Results\",\"content\":\"\\u003cp\\u003eParticipants did not differ in sex\\u0026nbsp;(\\u0026chi;\\u0026sup2; = .01, p =.91, 6 F in the PD-VH group, 7 F in the PD group, 13 M in each group),\\u0026nbsp;age,\\u0026nbsp;disease duration, motor symptoms, LEDD or MoCA score (\\u003cstrong\\u003eSupplementary Tables 1, 2;\\u0026nbsp;\\u003c/strong\\u003e\\u003cstrong\\u003eSupplementary Figure 1)\\u003c/strong\\u003e.\\u003cstrong\\u003e\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eSensor space and source reconstruction analyses identify occipital and temporal sources\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eSensor space analyses confirmed that the channels involved in the task returned temporal and frontal locations (\\u003cem\\u003ep\\u003c/em\\u003e \\u0026lt;.001, \\u003cstrong\\u003eSI3\\u003c/strong\\u003e) and parieto-occipital\\u003cem\\u003e\\u0026nbsp;(FWE\\u003c/em\\u003e p \\u0026lt;.05; \\u003cstrong\\u003eSI3\\u003c/strong\\u003e). The t-test conducted for source reconstruction returned clusters in the inferior temporal gyrus and in lateral occipital regions (\\u003cem\\u003eFWE\\u003c/em\\u003e p \\u0026lt;.05) and in the thalamus (p\\u0026lt;.001, pFWE = .07\\u003cem\\u003e)\\u003c/em\\u003e (Figure 2, SI4).\\u003cem\\u003e\\u0026nbsp;\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003evMMN differences correlate with decreased top-down connectivity in PD-VH\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eBMS (Supplementary Figure 4) with a fixed effect design determined which of our ventral stream models (fully connected model) best-fit the data. We then carried out 2\\u003csup\\u003end\\u003c/sup\\u003e level analyses with PEB finding that task stimulus condition (rare vs standard) for PD-VH was associated with increased bottom-up effective connectivity and decreased self-inhibition for lV1, and with decreased connectivity, especially top-down (pp\\u0026gt;.99;\\u003cstrong\\u003e\\u0026nbsp;Figure3\\u003c/strong\\u003e).\\u003c/p\\u003e\\n\\u003cp\\u003ePEB analyses on latent connectivity (matrix A) show that at baseline for PD-VH there is an overall increase in effective connectivity both top-down and bottom-up, with an overall opposite pattern of that observed to the task from the PFC to V1 bilaterally (\\u003cstrong\\u003eFigure3F, G\\u003c/strong\\u003e).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eThe recursive PEB exploratory analysis on alternative second-level models (SI6, \\u003cstrong\\u003eSupplementary Figure 7\\u003c/strong\\u003e) points to a primary role of feedforward connections, with increased connectivity from rV1 to rITG and to rPFC. The fully connected model was among the top two best models and further analyses confirmed the results presented in Figure 3 (Supplementary Figure 7). \\u0026nbsp;We also explored the parameters of the intrinsic coupling only model, where the difference in PD with and without VH is associated to a change in cortical microcircuitry encoded in the intrinsic coupling parameter, finding a reduction in intrinsic connectivity from excitatory cells in right V1, left ITG and right PFC and from inhibitory cells (disinhibition) for all regions. \\u0026nbsp; \\u003cem\\u003e\\u0026nbsp;\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eHallucination severity, subtype and multimodality analysis.\\u003c/strong\\u003e\\u003cem\\u003e\\u0026nbsp;\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThe severity of complex visual hallucinations (CVH) correlated with distinct patterns of connectivity: greater severity was associated with reduced top-down connectivity (right PFC to right ITG; left V1 to left ITG) and enhanced bottom-up connectivity (right V1 to right PFC and right ITG). Minor hallucinations (MH) had lower connectivity from left PFC to left V1, although this did not remain significant after correction for multiple comparisons (all stats in Figure 4; details in SI7).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eML Multiple regression models with LOOCV\\u003c/strong\\u003e\\u003cem\\u003e\\u0026nbsp;\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eBoth top-down and bottom-up connectivity right-hemisphere changes predicted CVH severity [\\u003cem\\u003eR\\u003c/em\\u003e\\u003csup\\u003e2\\u003c/sup\\u003e = \\u0026nbsp;0.49, adjusted \\u003cem\\u003eR\\u003c/em\\u003e\\u003csup\\u003e2\\u003c/sup\\u003e 0.4\\u003csup\\u003e\\u0026nbsp;\\u003c/sup\\u003e(3,15) = 4.9, \\u003cem\\u003ep\\u003c/em\\u003e-value: 0.014 RMSE =2.26, MAE = 177] (SI8; Figure 8), \\u003cem\\u003eMH\\u003c/em\\u003e were best predicted by left-hemisphere decreases (lV1-rV1, lPFC-lV1) [\\u003cem\\u003eR\\u003c/em\\u003e\\u003csup\\u003e2\\u003c/sup\\u003e = 0.40, adjusted \\u003cem\\u003eR\\u003c/em\\u003e\\u003csup\\u003e2\\u003c/sup\\u003e = .32 F(2,17) = 5.60, \\u003cem\\u003ep\\u003c/em\\u003e= 0.01, RMSE = 5.69, MAE = 4.30, BIC = 126.26]. \\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eResults were unchanged when introducing MoCA or LEDD as covariates (SI8).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eAt baseline, CVH severity was best predicted by inter-hemispheric connectivity in V1 and top-down connectivity from PFC bilaterally; MH were predicted by left-hemisphere effective connectivity (SI8).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eMultimodal hallucinations were best predicted by a model using the 7/10 connections found altered (Figure 4D); introducing MoCA as a covariate rendered the top-down, but not the bottom-up model non-significant. \\u003cstrong\\u003e\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e\\u003cem\\u003eNeural microcircuit models and simulations.\\u0026nbsp;\\u003c/em\\u003e\\u003c/strong\\u003eLeveraging the estimates of neural dynamics featured in the neural mass model used (inhibitory interneurons, spiny stellate cells, pyramidal neurons), we found that decreased inhibitory activity across ventral visual and prefrontal regions significantly predicted the severity of complex visual hallucinations (CVH); this was also the case when combining inhibitory and excitatory estimates, supporting a network-level deficit and a disruption of excitatory-inhibitory (E/I) balance, with the ITG as a particularly significant contributor (SI8).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eSimulations of perturbation of E/I balance revealed pronounced reductions in both excitatory and inhibitory activity in PD-VH, particularly within the left V1 and bilateral ITG regions. These disruptions occurred predominantly during the 100\\u0026ndash;250ms period (MMN) and confirmed by our AUC analysis (SI10). When computing state space area in the two groups the difference is statistically significant for every region, confirming the differences observed with the plots (Figure 5; SI10).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eDorsal model.\\u0026nbsp;\\u003c/strong\\u003eAll details about model specification are reported in SI11. For what concerns the task connectivity the BMA results show that task effects (rare vs standard) for PD-VH were mostly associated with an altered connectivity between parietal and frontal nodes (pp \\u0026gt; .99). For this model PD-VH connectivity was increased from left V1 to left IPL, and reduced IPL to PFC, bilaterally; there were no correlations with severity of hallucinations (Supplementary figures 9-11).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eSource-reconstructed signal and receptor binding atlases exploratory analysis.\\u003c/em\\u003e Results for the first linear regression analyses with Cook\\u0026rsquo;s distance\\u003csup\\u003e43\\u003c/sup\\u003e to inspect outliers are reported in SI13 and in Figure 6B. Using BrainSMASH\\u003csup\\u003e56\\u0026nbsp;\\u003c/sup\\u003ewe confirmed a positive correlation between 5-HT\\u003csub\\u003e2A\\u003c/sub\\u003e BP\\u003csub\\u003eND\\u003c/sub\\u003e\\u003cem\\u003e\\u0026nbsp;\\u003c/em\\u003eand MMN signal (\\u003cem\\u003er\\u003c/em\\u003e=.284, \\u003cem\\u003ep\\u003c/em\\u003e=.037) with higher connectivity in regions of higher BP\\u003csub\\u003eND\\u003c/sub\\u003e. The opposite pattern was observed for D1 (\\u003cem\\u003er\\u003c/em\\u003e=-.29, \\u003cem\\u003ep\\u003c/em\\u003e =.017) and VAChT (\\u003cem\\u003er\\u003c/em\\u003e=-2.73, \\u003cem\\u003ep\\u003c/em\\u003e =.03). D2/D3 was no longer significant after correcting for spatial autocorrelation (p \\u0026gt;.05).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eWhen investigating these results at the individual level using LMMs (\\u003cem\\u003eMMN_signal ~ _receptor * VH + age + LEDD + sex + disease onset +MoCA + (1 | Participant))\\u003c/em\\u003e, we found for 5-HT\\u003csub\\u003e2A\\u0026nbsp;\\u003c/sub\\u003ea \\u003cstrong\\u003emain effect\\u003c/strong\\u003e of 5‐HT\\u003csub\\u003e2A\\u003c/sub\\u003e \\u003cem\\u003et\\u0026nbsp;\\u003c/em\\u003e=2.13,\\u003cem\\u003ep\\u003c/em\\u003e =.034, hallucination status (VH) \\u003cem\\u003et\\u0026nbsp;\\u003c/em\\u003e=2.52, \\u003cem\\u003ep\\u003c/em\\u003e =.017 and a \\u003cstrong\\u003epositive interaction\\u003c/strong\\u003e between 5‐HT\\u003csub\\u003e2A\\u003c/sub\\u003e and VH \\u003cem\\u003et\\u003c/em\\u003e =2.06, \\u003cem\\u003ep\\u003c/em\\u003e =.040. The same model using D1 receptor availability found a \\u003cstrong\\u003esignificant group effect\\u003c/strong\\u003e \\u003cem\\u003et\\u003c/em\\u003e = 2.54, \\u003cem\\u003ep\\u003c/em\\u003e = .015, and a \\u003cstrong\\u003enegative D1\\u0026nbsp;\\u003c/strong\\u003e*\\u003cstrong\\u003eVH interaction\\u003c/strong\\u003e \\u003cem\\u003et\\u003c/em\\u003e = \\u0026ndash;2.13, \\u003cem\\u003ep\\u003c/em\\u003e =.033. For VAChT, using the map from \\u003cstrong\\u003eN=18\\u003c/strong\\u003e older participants\\u003csup\\u003e52\\u003c/sup\\u003e, we found a positive association with MMN (\\u003cem\\u003et\\u003c/em\\u003e =1.75, \\u003cem\\u003ep\\u003c/em\\u003e =.08) and a negative VAChT * VH interaction effect (\\u003cem\\u003et\\u003c/em\\u003e =-1.58, \\u003cem\\u003ep\\u003c/em\\u003e =.1), though neither reached significance; VH status was significant \\u003cem\\u003et\\u003c/em\\u003e =2.46, \\u003cem\\u003ep\\u003c/em\\u003e =.019. In a \\u003cstrong\\u003esensitivity analysis\\u003c/strong\\u003e with the \\u003cstrong\\u003eN=4\\u003c/strong\\u003e healthy controls map\\u003csup\\u003e51\\u003c/sup\\u003e, we observed the \\u003cstrong\\u003esame direction\\u003c/strong\\u003e of effects but with a \\u003cstrong\\u003elarger effect size\\u003c/strong\\u003e and formal significance (partial \\u003cem\\u003er\\u003c/em\\u003e \\u0026asymp; \\u0026ndash;.037 VAChT * VH interaction; see SI13). These convergent patterns suggest a negative relationship between VAChT and MMN in the VH group, though not as robust as D1.\\u003c/p\\u003e\\n\\u003cp\\u003eNo effect of disease duration, LEDD and age was observed (SI13).\\u0026nbsp;\\u003c/p\\u003e\"},{\"header\":\"Discussion\",\"content\":\"\\u003cp\\u003eWe investigated the neural dynamics underlying sensory processing deficits in patients with PD and visual hallucinations (PD-VH). Using dynamic causal modelling applied to EEG data from a visual MMN task, we showed that PD-VH patients respond differently to the sensory environment and that these differences correlate with their visual hallucinations.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eDecreased top-down and increased bottom-up activity underlies PD-VH reduction of vMMN\\u0026nbsp;\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eWe found that in PD-VH there was a significant reduction of top-down activity and an increase in bottom-up connectivity. Top-down connectivity is crucial for the integration of sensory information\\u003csup\\u003e24-26\\u003c/sup\\u003e and this finding suggests that in PD-VH there is a failure to update predictions based on sensory evidence and an over-reliance on sensory information. This over-reliance can contribute to the difficulties in stimulus detection due to retinal degeneration\\u003csup\\u003e28,32\\u003c/sup\\u003e and responding to the changes in the sensory environment. This interpretation resonates with the principles of a recently proposed framework on VH in PD\\u003csup\\u003e10\\u003c/sup\\u003e. When taking a hypothesis free approach the feedforward model, with a right-hemisphere pattern of increased V1-ITG and V1-PFC connectivity emerged as the strongest contributor. In line with this result, recent work has shown that people are more likely to hallucinate when sensory information is noisy, especially when combined with strong expectations\\u003csup\\u003e60\\u003c/sup\\u003e. Interestingly this was found with an orientation perception task, consistent with our stimuli, and in the input layers of V2, where orientation-specific activity takes place, lending further support to our interpretation and suggesting that spontaneous feedforward activity in the visual cortex can also lead to hallucinations, rather than feedback activity alone\\u003csup\\u003e60\\u003c/sup\\u003e. \\u0026nbsp;The finding that local source differences correlate positively with 5-HT\\u003csub\\u003e2A\\u0026nbsp;\\u003c/sub\\u003edensity suggests that this receptor may be facilitating cortical disinhibition, consistent with previous proposals\\u003csup\\u003e45,61\\u003c/sup\\u003e and studies showing 5-HT\\u003csub\\u003e2A\\u003c/sub\\u003e antagonists reducing psychosis-like symptoms\\u003csup\\u003e19,45,61\\u003c/sup\\u003e. Based on our results, we may speculate that this reflects an aberrant or a compensatory excitation. \\u0026nbsp;We also find a negative relationship with dopaminergic receptor D1; dopamine is proposed to encode precision weighting\\u003csup\\u003e62\\u003c/sup\\u003e; a possible interpretation is that this could contribute to bias the system to assign undue salience to bottom-up inputs with the result of illusions and/or hallucinations, and the disruption of normal mismatch detection.\\u003c/p\\u003e\\n\\u003cp\\u003eThe analysis of the baseline activity complements these results: top-down connectivity from PFC increased in the PD-VH group, while the same connections were under-responsive during the task, and vice versa for some of the bottom-up connections. These differences in latent connectivity even in the absence of experimental perturbations suggest that PD-VH are not just over-relying on sensory information during the vMMN task, but they seem to do so more generally. The decreased activity from visual to prefrontal at baseline could reflect the low-level visual deficits common in PD-VH and affecting also higher-order processes, which together with the reduced PFC to ITG left connectivity seems to support this and previous models of VH in LBD\\u003csup\\u003e63\\u003c/sup\\u003e.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eDecreased top-down and increased bottom-up connectivity in PD-VH are related to hallucinations severity\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eHallucinations severity directly correlated with these connectivity patterns,\\u0026nbsp;with pathways of increased connectivity showing a positive relationship and those with decreased connectivity showing a negative relationship. Complex visual hallucinations correlated with decreased top-down connectivity from right PFC to ITG, and increased bottom-up connectivity from early visual cortex (V1) to higher-order regions (ITG, PFC) (Figure7). These results underscore a robust right-hemisphere network impairment associated with complex hallucinations. Minor hallucinations showed a similar but weaker trend predominantly involving left-hemisphere regions, though this did not survive multiple comparison correction.\\u003c/p\\u003e\\n\\u003cp\\u003eA recent DCM fMRI study found that the pattern of altered connectivity observed at rest was predictive of hallucinations severity\\u003csup\\u003e8\\u003c/sup\\u003e. \\u0026nbsp;When addressing the relationship between multiple connections and hallucinations severity with a similar purpose, we find that CVH\\u003cem\\u003e\\u0026nbsp;\\u003c/em\\u003ewere best predicted by right-hemisphere bottom-up and top-down connections, supporting the hypothesis that these regions operate as a network. \\u0026nbsp;This right-hemisphere pattern of connections associated with CVH is in line with a recent fMRI study exploring VH severity with resting-state fMRI in PD and DLB with hallucinations, where CVH duration was associated with right hemisphere activity in ventral visual and parieto-occipital regions, and minor visual phenomena to decreased connectivity only in the left hemisphere\\u003csup\\u003e29\\u003c/sup\\u003e, as we also found for MH. The result of increased activity from rV1 being associated to VH and VH severity aligns with the finding in schizophrenia of disinhibition in sensory areas (A1) predicting abnormal auditory perception in the aMMN and positive symptoms\\u003csup\\u003e22\\u003c/sup\\u003e. \\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eVisual hallucinations severity may be related to disinhibition\\u0026nbsp;\\u003c/em\\u003e\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eSimulations of intrinsic excitatory-inhibitory (E/I) dynamics demonstrated significant imbalance and instability in the PD-VH network, particularly in visual and ITG regions,\\u0026nbsp;extensively involved in object recognition, particularly of human and animal stimuli\\u003csup\\u003e63,64-66\\u003c/sup\\u003e, which are often the content of VH in these patients\\u003csup\\u003e67\\u003c/sup\\u003e. The exploratory analysis with estimated activity of excitatory cells in the left ITG positively related to CVH specifically suggests that this impairment may be associated with the complex stimuli in the visual modality alone and presents a circuit-level causal hypothesis involving the ITG for the failure to have greater activity for the deviant trials we showed previously\\u003csup\\u003e19\\u003c/sup\\u003e. The ITG also showed instability in the simulations, with the analysis of the simulated time series and the state space area analyses suggesting that hallucinations might arise from disrupted cortical inhibition and heightened neural sensitivity to perturbations.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eOur results lend themselves to more than one potential explanation. First, that of an E/I imbalance similar to what has been proposed in schizophrenia\\u003csup\\u003e68\\u003c/sup\\u003e. We can interpret the pattern observed as a reduced adaptability and stability in the network dynamics and in the intrinsic coupling within the regions in our model. This is in line a recent aMMN study where the electrophysiological patterns observed under ketamine administration and in the relative simulations appear compatible with those observed in psychosis\\u003csup\\u003e44\\u003c/sup\\u003e. \\u0026nbsp;In this view, reduced activity through glutamate receptors is proposed to induce a decrease in the inhibitory activity of the inhibitory interneurons. Lower GABA+/creatine in the ITG in PD-VH\\u003csup\\u003e69\\u003c/sup\\u003e supports this view. Disinhibition may lead to excessive excitatory signals, causing the positive symptoms in schizophrenia and we may speculate the hallucinations in PD-VH. We observe at task an excess in excitatory activity in low visual regions and at baseline an overall increase in excitatory activity, in line with this hypothesis.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eFurther supporting the interpretation that hallucinations may arise from disrupted cortical inhibition and heightened neural sensitivity to perturbations, exploratory analyses using PET receptor atlases revealed a positive correlation between cortical 5-HT\\u003csub\\u003e2A\\u003c/sub\\u003e receptor distribution and abnormal MMN source activity in PD-VH, consistent with the known role of serotonin in cortical disinhibition\\u003csup\\u003e70\\u003c/sup\\u003e and psychosis\\u003csup\\u003e45\\u003c/sup\\u003e.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eAnother possible interpretation is that our results may be related to thalamic dysfunction. We did not investigate the thalamus here, as with EEG recordings the signal captures postsynaptic potentials in the cortex, making the choice of using the thalamus as a dipole a challenging one. Nevertheless, a large thalamic cluster was found associated to the task in our source reconstruction analysis. Previous research has shown that the thalamus plays a role in VH in Parkinson\\u0026rsquo;s\\u003csup\\u003e71,72\\u003c/sup\\u003e and a recent proposal\\u003csup\\u003e73\\u0026nbsp;\\u003c/sup\\u003esees the thalamus as driver of unbalanced network recruitment, suggested to induce a decoupling of the default mode network (DMN) and task-positive networks, possibly disrupting the comparison of priors and of sensory percepts\\u003csup\\u003e73,74\\u003c/sup\\u003e. A recent spectral DCM fMRI study found an involvement of the LGN of the thalamus, together with ventral visual regions, in PD-VH\\u003csup\\u003e8\\u003c/sup\\u003e. Despite not directly comparable to our results due to the different (resting fMRI and task EEG) modalities, nevertheless when examining the latent activity (between trials) we also have increased top-down connectivity, including from left PFC to V1 as reported in the aforementioned study, and both increased and reduced bottom-up activity. \\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eOur exploratory analysis with PET atlases and source-reconstructed MMN signal also shows a negative relationship between cholinergic receptors distribution and MMN signal in PD-VH. A recent PET study in PD-VH showed a marked cholinergic deficiency in the left ventral visual stream\\u003csup\\u003e47\\u003c/sup\\u003e. We may speculate that a dysregulation in the cholinergic system might affect not only cortical areas but also thalamic processing, thereby contributing to abnormal sensory processing. \\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eGlobal cognitive decline may modulate the presence of multimodal hallucinations\\u0026nbsp;\\u003c/em\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eOur results also indicate that multimodal hallucinations might reflect more extensive neural network disruptions, modulated by a sub-clinical global cognitive decline, as controlling for cognition left primarily bottom-up sensory pathways significantly associated with multimodal hallucinations. This is consistent with reports of PD-VH declining cognitively more rapidly\\u003csup\\u003e2\\u003c/sup\\u003e and showing possible sub-clinical cognitive variations early on in their VH\\u003csup\\u003e4\\u003c/sup\\u003e.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eLimitations\\u003c/strong\\u003e\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eWhile our study primarily focused on cortical regions, exploratory analyses hint at the involvement of subcortical structures like the thalamus and cholinergic systems, consistent with previous research, which may offer promising directions for future studies. Limitations include the inability of EEG to robustly assess subcortical sources, the exploratory nature of our receptor-binding analyses, and that the vAChT map we used was developed on data from both healthy and Alzheimer\\u0026rsquo;s disease participants, thus it might partly reflect AD‐related patterns. We validated our finding with a sensitivity analysis using another VAChT map (N=4), that by itself would have been a less realistic fit for our study.\\u003c/p\\u003e\"},{\"header\":\"Conclusions\",\"content\":\"\\u003cp\\u003eOverall, our findings clarify the top-down and bottom-up neural abnormalities underlying PD-VH, strongly supporting the predictive coding account, and pointing to an alteration in updating predictions based on sensory evidence and a deficit in tracking changes in the sensory environment. Our results highlight potential receptor targets mediating this effect and reinforce the critical role of the ventral visual stream in the generation of VH.\\u003c/p\\u003e \\u003cp\\u003eThe hyperconnectivity of visual regions strongly contributes to explaining vMMN differences and correlates with hallucinations severity and complexity, supporting targeted therapeutic approaches focused on normalizing sensory processing dynamics. Finally, we show an increasingly extensive network being related to VH complexity and a possible relationship with sub-clinical cognitive decline. Having provided a mechanistic model of predictive coding deficits in PD-VH, we think future longitudinal studies will be essential to confirm whether observed connectivity changes directly underpin hallucinations alone or reflect a broader underlying cognitive decline.\\u003c/p\\u003e\"},{\"header\":\"Declarations\",\"content\":\"\\u003cp\\u003e\\u003cstrong\\u003e\\u003cem\\u003eAcknowledgements\\u0026nbsp;\\u003c/em\\u003e\\u003c/strong\\u003e\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eWe thank the CRISP (patient representation) group for support and advice during the inception of this study and we thank all our participants for taking part in the research and their partners for their support on the study days. We thank Rosalyn Moran for her insight regarding the possible avenues of investigation to follow with DCM and advice on model specification. We also thank Rick Adams for comments on an earlier version of the analyses. We also thank Dag Aarsland for his support with the development of important study ideas. We thank Stephanie Stephenson for training the study researchers with clinical trials procedures and for her support throughout the duration of the study; and we thank Caroline Woolridge for her support with study logistics and ethics applications. We thank Simon Hill for the contribution to the EEG task and recording setup. We thank the KCL Clinical Research Facility staff for their support with the study and the CRF director Elka Giemza. For open access, the author has applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising from this submission.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e\\u003cem\\u003eFunding\\u003c/em\\u003e\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThis is independent research funded by the Medical Research Council (grant number MR/R005931/1 awarded to M.A.M., D.F., K.R.C. and Dag Aarsland) and carried out at the NIHR (National Institute for Health and Care Research) Maudsley Biomedical Research Centre (BRC). The views expressed are those of the author(s) and not necessarily those of the Medical Research Council, the National Institute for Health and Care Research or the Department of Health and Social Care. M.V. was supported by Medical Research Council and is currently supported by Alzheimer\\u0026rsquo;s Research UK (grant ARUK-RF2022B-002). C.H.W.G. is supported by the Medical Research Council (MR/W029235/1) and the National Institute for Health and Care Research Cambridge Biomedical Research Centre (NIHR203312). The views expressed are those of the authors and not necessarily those of the National Institute for Health and Care Research or the Department of Health and Social Care.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\n\\u003cli\\u003eSchapira, A. H., Chaudhuri, K. R., \\u0026amp; Jenner, P. (2017). Non-motor features of Parkinson disease. \\u003cem\\u003eNature Reviews Neuroscience\\u003c/em\\u003e, \\u003cem\\u003e18\\u003c/em\\u003e(7), 435-450.\\u003c/li\\u003e\\n\\u003cli\\u003eFfytche, D. H., Creese, B., Politis, M., Chaudhuri, K. R., Weintraub, D., Ballard, C., \\u0026amp; Aarsland, D. (2017). The psychosis spectrum in Parkinson disease. \\u003cem\\u003eNature Reviews Neurology\\u003c/em\\u003e, \\u003cem\\u003e13\\u003c/em\\u003e(2), 81-95.\\u003c/li\\u003e\\n\\u003cli\\u003ePisani, S., Gosse, L., Aarsland, D., Chaudhuri, K. R., Ballard, C., Ffytche, D., ... \\u0026amp; Bhattacharyya, S. (2024). Parkinson\\u0026rsquo;s disease psychosis associated with accelerated multidomain cognitive decline. \\u003cem\\u003eBMJ Ment Health\\u003c/em\\u003e, \\u003cem\\u003e27\\u003c/em\\u003e(1).\\u003c/li\\u003e\\n\\u003cli\\u003eStampacchia, S., Bernasconi, F., Van De Ville, D., Amico, E., Pagonabarraga, J., Kulisevsky, J., \\u0026amp; Blanke, O. (2025). Connectome-based brain fingerprints predict early cognitive decline in Parkinson\\u0026apos;s patients with minor hallucinations. \\u003cem\\u003ebioRxiv\\u003c/em\\u003e.\\u003c/li\\u003e\\n\\u003cli\\u003eVignando, M., Ffytche, D., Lewis, S. J., Lee, P. H., Chung, S. J., Weil, R. S.,... \\u0026amp; Mehta, M. A. (2022). Mapping brain structural differences and neuroreceptor correlates in Parkinson\\u0026rsquo;s disease visual hallucinations. \\u003cem\\u003eNature communications\\u003c/em\\u003e, \\u003cem\\u003e13\\u003c/em\\u003e(1), 519.\\u003c/li\\u003e\\n\\u003cli\\u003eWeil, R. S., Hsu, J. K., Darby, R. R., Soussand, L., \\u0026amp; Fox, M. D. (2019). Neuroimaging in Parkinson\\u0026rsquo;s disease dementia: connecting the dots. \\u003cem\\u003eBrain communications\\u003c/em\\u003e, \\u003cem\\u003e1\\u003c/em\\u003e(1), fcz006.\\u003c/li\\u003e\\n\\u003cli\\u003eLenka, A., Jhunjhunwala, K. R., Saini, J., \\u0026amp; Pal, P. K. (2015). Structural and functional neuroimaging in patients with Parkinson\\u0026apos;s disease and visual hallucinations: a critical review. \\u003cem\\u003eParkinsonism \\u0026amp; related disorders\\u003c/em\\u003e, \\u003cem\\u003e21\\u003c/em\\u003e(7), 683-691.\\u003c/li\\u003e\\n\\u003cli\\u003eThomas, G. E., Zeidman, P., Sultana, T., Zarkali, A., Razi, A., \\u0026amp; Weil, R. S. (2023). Changes in both top-down and bottom-up effective connectivity drive visual hallucinations in Parkinson\\u0026rsquo;s disease. \\u003cem\\u003eBrain Communications\\u003c/em\\u003e, \\u003cem\\u003e5\\u003c/em\\u003e(1), fcac329.\\u003c/li\\u003e\\n\\u003cli\\u003eShine, J. M., Muller, A. J., O\\u0026rsquo;Callaghan, C., Hornberger, M., Halliday, G. M., \\u0026amp; Lewis, S. J. (2015). Abnormal connectivity between the default mode and the visual system underlies the manifestation of visual hallucinations in Parkinson\\u0026rsquo;s disease: a task-based fMRI study. \\u003cem\\u003enpj Parkinson\\u0026apos;s Disease\\u003c/em\\u003e, \\u003cem\\u003e1\\u003c/em\\u003e(1), 1-8.\\u003c/li\\u003e\\n\\u003cli\\u003eCollerton, D., Barnes, J., Diederich, N. J., Dudley, R., Friston, K., Goetz, C. G., ... \\u0026amp; Weil, R. S. (2023). Understanding visual hallucinations: a new synthesis. \\u003cem\\u003eNeuroscience \\u0026amp; Biobehavioral Reviews\\u003c/em\\u003e, \\u003cem\\u003e150\\u003c/em\\u003e, 105208.\\u003c/li\\u003e\\n\\u003cli\\u003eSterzer, P., Adams, R. A., Fletcher, P., Frith, C., Lawrie, S. M., Muckli, L., ... \\u0026amp; Corlett, P. R. (2018). The predictive coding account of psychosis. \\u003cem\\u003eBiological psychiatry\\u003c/em\\u003e, \\u003cem\\u003e84\\u003c/em\\u003e(9), 634-643\\u003c/li\\u003e\\n\\u003cli\\u003eBastos, A. M., Usrey, W. M., Adams, R. A., Mangun, G. R., Fries, P., \\u0026amp; Friston, K. J. (2012). Canonical microcircuits for predictive coding. \\u003cem\\u003eNeuron\\u003c/em\\u003e, \\u003cem\\u003e76\\u003c/em\\u003e(4), 695-711.\\u003c/li\\u003e\\n\\u003cli\\u003eHaarsma, J., Knolle, F., Griffin, J. D., Taverne, H., Mada, M., Goodyer, I. M.,... \\u0026amp; Nspn Consortium. (2020). Influence of prior beliefs on perception in early psychosis: Effects of illness stage and hierarchical level of belief. \\u003cem\\u003eJournal of abnormal psychology\\u003c/em\\u003e, \\u003cem\\u003e129\\u003c/em\\u003e(6), 581.\\u003c/li\\u003e\\n\\u003cli\\u003eBhat, A., Irizar, H., Thygesen, J. H., Kuchenbaecker, K., Pain, O., Adams, R. A., ... \\u0026amp; Bramon, E. (2021). Transcriptome-wide association study reveals two genes that influence mismatch negativity. \\u003cem\\u003eCell reports\\u003c/em\\u003e, \\u003cem\\u003e34\\u003c/em\\u003e(11), 108868\\u003c/li\\u003e\\n\\u003cli\\u003eFriston, K. J., \\u0026amp; Dolan, R. J. (2010). Computational and dynamic models in neuroimaging. \\u003cem\\u003eNeuroimage\\u003c/em\\u003e, \\u003cem\\u003e52\\u003c/em\\u003e(3), 752-765.\\u003c/li\\u003e\\n\\u003cli\\u003eFeldman, H., \\u0026amp; Friston, K. J. (2010). Attention, uncertainty, and free-energy. \\u003cem\\u003eFrontiers in human neuroscience\\u003c/em\\u003e, \\u003cem\\u003e4\\u003c/em\\u003e, 215.\\u003c/li\\u003e\\n\\u003cli\\u003eFriston, K. J. (2005). Hallucinations and perceptual inference. \\u003cem\\u003eBehavioral and Brain Sciences\\u003c/em\\u003e, \\u003cem\\u003e28\\u003c/em\\u003e(6), 764-766.\\u003c/li\\u003e\\n\\u003cli\\u003eRao, R. P., \\u0026amp; Ballard, D. H. (1999). Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. \\u003cem\\u003eNature neuroscience\\u003c/em\\u003e, \\u003cem\\u003e2\\u003c/em\\u003e(1), 79-87.\\u003c/li\\u003e\\n\\u003cli\\u003eVignando, M., Ffytche, D., Mazibuko, N., Palma, G., Montagnese, M., Dave, S., ... \\u0026amp; Mehta, M. A. (2024). Visual mismatch negativity in Parkinson\\u0026apos;s psychosis and potential for testing treatment mechanisms. \\u003cem\\u003eBrain Communications\\u003c/em\\u003e, \\u003cem\\u003e6\\u003c/em\\u003e(5), fcae291.\\u003c/li\\u003e\\n\\u003cli\\u003eKiebel, S. J., Garrido, M. I., Moran, R. J., \\u0026amp; Friston, K. J. (2008). Dynamic causal modelling for EEG and MEG. \\u003cem\\u003eCognitive neurodynamics\\u003c/em\\u003e, \\u003cem\\u003e2\\u003c/em\\u003e, 121-136.\\u003c/li\\u003e\\n\\u003cli\\u003eKiebel, S. J., David, O., \\u0026amp; Friston, K. J. (2006). Dynamic causal modelling of evoked responses in EEG/MEG with lead field parameterization. \\u003cem\\u003eNeuroImage\\u003c/em\\u003e, \\u003cem\\u003e30\\u003c/em\\u003e(4), 1273-1284.\\u003c/li\\u003e\\n\\u003cli\\u003eAdams, R. A., Pinotsis, D., Tsirlis, K., Unruh, L., Mahajan, A., Horas, A. M., ... \\u0026amp; Anticevic, A. (2022). Computational modeling of electroencephalography and functional magnetic resonance imaging paradigms indicates a consistent loss of pyramidal cell synaptic gain in schizophrenia. \\u003cem\\u003eBiological psychiatry\\u003c/em\\u003e, \\u003cem\\u003e91\\u003c/em\\u003e(2), 202-215.\\u003c/li\\u003e\\n\\u003cli\\u003eFriston, K. (2005). A theory of cortical responses. \\u003cem\\u003ePhilosophical Transactions of the Royal Society B:\\u003cem\\u003eBiological Sciences, 360\\u003c/em\\u003e(1456), 815-836.\\u003c/em\\u003e\\u003c/li\\u003e\\n\\u003cli\\u003eSummerfield, C., \\u0026amp; Egner, T. (2009). Expectation (and attention) in visual cognition. \\u003cem\\u003eTrends in Cognitive Sciences, 13\\u003c/em\\u003e(9), 403-409.\\u003c/li\\u003e\\n\\u003cli\\u003eKveraga, K., \\u0026amp; Bar, M. (2014). Top-down predictions in the cognitive brain. \\u003cem\\u003eBrain and Cognition, 94\\u003c/em\\u003e, 65-67.\\u003c/li\\u003e\\n\\u003cli\\u003eShine, J. M., O\\u0026rsquo;Callaghan, C., Halliday, G. M., \\u0026amp; Lewis, S. J. (2014). Tricks of the mind: visual hallucinations as disorders of attention. \\u003cem\\u003eProgress in neurobiology\\u003c/em\\u003e, \\u003cem\\u003e116\\u003c/em\\u003e, 58-65.\\u003c/li\\u003e\\n\\u003cli\\u003ePagonabarraga, J., Soriano-Mas, C., Llebaria, G., L\\u0026oacute;pez-Sol\\u0026agrave;, M., Pujol, J., \\u0026amp; Kulisevsky, J. (2014). Neural correlates of minor hallucinations in non-demented patients with Parkinson\\u0026apos;s disease. \\u003cem\\u003eParkinsonism \\u0026amp; related disorders\\u003c/em\\u003e, \\u003cem\\u003e20\\u003c/em\\u003e(3), 290-296.\\u003c/li\\u003e\\n\\u003cli\\u003eDiederich, N. J., F\\u0026eacute;nelon, G., Stebbins, G., \\u0026amp; Goetz, C. G. (2009). Hallucinations in Parkinson disease. \\u003cem\\u003eNature Reviews Neurology\\u003c/em\\u003e, \\u003cem\\u003e5\\u003c/em\\u003e(6), 331-342.\\u003c/li\\u003e\\n\\u003cli\\u003eD\\u0026rsquo;Antonio, F., Boccia, M., Di Vita, A., Suppa, A., Fabbrini, A., Canevelli, M., ... \\u0026amp; Ffytche, D. (2022). Visual hallucinations in Lewy body disease: pathophysiological insights from phenomenology. \\u003cem\\u003eJournal of Neurology\\u003c/em\\u003e, \\u003cem\\u003e269\\u003c/em\\u003e(7), 3636-3652.\\u003c/li\\u003e\\n\\u003cli\\u003eMontagnese, M., Vignando, M., \\u0026amp; Mehta, M. A. .. (2022). Cognitive and visual processing performance in Parkinson\\u0026apos;s disease patients with vs without visual hallucinations: A meta-analysis. \\u003cem\\u003eCortex\\u003c/em\\u003e, \\u003cem\\u003e146\\u003c/em\\u003e, 161-172.\\u003c/li\\u003e\\n\\u003cli\\u003eGarcia-Diaz, A. I., Segura, B., Baggio, H. C., Marti, M. J., Valldeoriola, F., Compta, Y., ... \\u0026amp; Junque, C. (2018). Structural brain correlations of visuospatial and visuoperceptual tests in Parkinson\\u0026rsquo;s disease. \\u003cem\\u003eJournal of the International Neuropsychological Society\\u003c/em\\u003e, \\u003cem\\u003e24\\u003c/em\\u003e(1), 33-44.\\u003c/li\\u003e\\n\\u003cli\\u003eWeil, R. S., Schrag, A. E., Warren, J. D., Crutch, S. J., Lees, A. J., \\u0026amp; Morris, H. R. (2016). Visual dysfunction in Parkinson\\u0026rsquo;s disease. \\u003cem\\u003eBrain\\u003c/em\\u003e, \\u003cem\\u003e139\\u003c/em\\u003e(11), 2827-2843.\\u003c/li\\u003e\\n\\u003cli\\u003eRavina, B., Marder, K., Fernandez, H. H., Friedman, J. H., McDonald, W., Murphy, D., ... \\u0026amp; Goetz, C. (2007). Diagnostic criteria for psychosis in Parkinson\\u0026apos;s disease: report of an NINDS, NIMH work group. \\u003cem\\u003eMovement Disorders\\u003c/em\\u003e, \\u003cem\\u003e22\\u003c/em\\u003e(8), 1061-106\\u003c/li\\u003e\\n\\u003cli\\u003eVoss, T., Bahr, D., Cummings, J., Mills, R., Ravina, B. \\u0026amp; Williams, H. 2013. Performance of a shortened Scale for Assessment of Positive Symptoms for Parkinson\\u0026apos;s disease psychosis. \\u003cem\\u003eParkinsonism Relat Disord,\\u003c/em\\u003e 19, 295-9.\\u003c/li\\u003e\\n\\u003cli\\u003eMosimann, U. P., Collerton, D., Dudley, R., Meyer, T. D., Graham, G., Dean, J. L., ... \\u0026amp; McKeith, I. G. (2008). A semi‐structured interview to assess visual hallucinations in older people. \\u003cem\\u003eInternational Journal of Geriatric Psychiatry: A journal of the psychiatry of late life and allied sciences\\u003c/em\\u003e, \\u003cem\\u003e23\\u003c/em\\u003e(7), 712-718.\\u003c/li\\u003e\\n\\u003cli\\u003eQian, X., Liu, Y. I., Xiao, B., Gao, L. I., Li, S., Dang, L., ... \\u0026amp; Zhao, L. (2014). The visual mismatch negativity (vMMN): Toward the optimal paradigm. \\u003cem\\u003eInternational Journal of Psychophysiology\\u003c/em\\u003e, \\u003cem\\u003e93\\u003c/em\\u003e(3), 311-315.\\u003c/li\\u003e\\n\\u003cli\\u003eGarrido, M. I., Kilner, J. M., Stephan, K. E., \\u0026amp; Friston, K. J. (2009). The mismatch negativity: a review of underlying mechanisms. \\u003cem\\u003eClinical neurophysiology\\u003c/em\\u003e, \\u003cem\\u003e120\\u003c/em\\u003e(3), 453-463.\\u003c/li\\u003e\\n\\u003cli\\u003eFriston, K., FitzGerald, T., Rigoli, F., Schwartenbeck, P., \\u0026amp; Pezzulo, G. (2016). Active inference and learning. \\u003cem\\u003eNeuroscience \\u0026amp; Biobehavioral Reviews\\u003c/em\\u003e, \\u003cem\\u003e68\\u003c/em\\u003e, 862-879.\\u003c/li\\u003e\\n\\u003cli\\u003eZeidman, P., Jafarian, A., Seghier, M. L., Litvak, V., Cagnan, H., Price, C. J., \\u0026amp; Friston, K. J. (2019). A guide to group effective connectivity analysis, part 2: Second level analysis with PEB. \\u003cem\\u003eNeuroimage\\u003c/em\\u003e, \\u003cem\\u003e200\\u003c/em\\u003e, 12-25.\\u003c/li\\u003e\\n\\u003cli\\u003eVenables, W. N. \\u0026amp; Ripley, B. D. (2002) Modern Applied Statistics with S. Fourth Edition. Springer, New York. ISBN 0-387-95457-0\\u003c/li\\u003e\\n\\u003cli\\u003eWickham et al., (2019). Welcome to the tidyverse. Journal of Open Source Software,4(43), 1686, https://doi.org/10.21105/joss.01686\\u003c/li\\u003e\\n\\u003cli\\u003eWickham H, Fran\\u0026ccedil;ois R, Henry L, M\\u0026uuml;ller K, Vaughan D (2023). dplyr: A Grammar of Data Manipulation. R package version 1.1.4, https://github.com/tidyverse/dplyr, https://dplyr.tidyverse.org.\\u003c/li\\u003e\\n\\u003cli\\u003eKuhn, Max (2008). \\u0026ldquo;Building Predictive Models in R Using the caret Package.\\u0026rdquo; Journal of Statistical Software, 28(5), 1\\u0026ndash;26. doi:10.18637/jss.v028.i05, https://www.jstatsoft.org/index.php/jss/article/view/v028i05.\\u003c/li\\u003e\\n\\u003cli\\u003eRosch, R. E., Auksztulewicz, R., Leung, P. D., Friston, K. J., \\u0026amp; Baldeweg, T. (2019). Selective prefrontal disinhibition in a roving auditory oddball paradigm under N-methyl-D-aspartate receptor blockade. \\u003cem\\u003eBiological Psychiatry: Cognitive Neuroscience and\\u003c/em\\u003e \\u003cem\\u003eNeuroimaging\\u003c/em\\u003e, \\u003cem\\u003e4\\u003c/em\\u003e(2), 140-150.\\u003c/li\\u003e\\n\\u003cli\\u003eBallanger, B., Strafella, A. P., van Eimeren, T., Zurowski, M., Rusjan, P. M., Houle, S., \\u0026amp; Fox, S. H. (2010). Serotonin 2A receptors and visual hallucinations in Parkinson disease. \\u003cem\\u003eArchives of neurology\\u003c/em\\u003e, \\u003cem\\u003e67\\u003c/em\\u003e(4), 416-421.\\u003c/li\\u003e\\n\\u003cli\\u003eRamesh, S., \\u0026amp; Arachchige, A. S. P. M. (2023). Depletion of dopamine in Parkinson\\u0026apos;s disease and relevant therapeutic options: A review of the literature. \\u003cem\\u003eAIMS neuroscience\\u003c/em\\u003e, \\u003cem\\u003e10\\u003c/em\\u003e(3), 200.\\u003c/li\\u003e\\n\\u003cli\\u003ed\\u0026rsquo;Angremont, E., van der Zee, S., Slingerland, S., Slomp, A. C., de Vries, E. F., van Laar, T., \\u0026amp; Sommer, I. E. (2024). Cholinergic deficiency in Parkinson\\u0026rsquo;s disease patients with visual hallucinations. \\u003cem\\u003eBrain\\u003c/em\\u003e, \\u003cem\\u003e147\\u003c/em\\u003e(10), 3370-3378.\\u003c/li\\u003e\\n\\u003cli\\u003eMika Naganawa, Nabeel Nabulsi, Shannan Henry, David Matuskey, Shu-Fei Lin, Lawrence Slieker, Adam J Schwarz, Nancy Kant, Cynthia Jesudason, Kevin Ruley, and others. First-in-human assessment of 11c-lsn3172176, an m1 muscarinic acetylcholine receptor pet radiotracer. Journal of Nuclear Medicine, 62(4):553\\u0026ndash;560, 2021.\\u003c/li\\u003e\\n\\u003cli\\u003eMarkello, R. D., Hansen, J. Y., Liu, Z. Q., Bazinet, V., Shafiei, G., Su\\u0026aacute;rez, L. E., ... \\u0026amp; Misic, B. (2022). Neuromaps: structural and functional interpretation of brain maps. \\u003cem\\u003eNature Methods\\u003c/em\\u003e, \\u003cem\\u003e19\\u003c/em\\u003e(11), 1472-1479.\\u003c/li\\u003e\\n\\u003cli\\u003eSavli M, Bauer A, Mitterhauser M, Ding Y (2012) Normative database of theserotonergic system in healthy subjects using multi-tracer PET. Neuroimage 63:447\\u0026ndash; 459.\\u003c/li\\u003e\\n\\u003cli\\u003eJustine Y Hansen, Golia Shafiei, Ross D Markello, Kelly Smart, Sylvia ML Cox, Martin N\\u0026oslash;rgaard, Vincent Beliveau, Yanjun Wu, Jean-Dominique Gallezot, \\u0026Eacute;tienne Aumont, and others. Mapping neurotransmitter systems to the structural and functional organization of the human neocortex. Nature neuroscience, 25(11):1569\\u0026ndash;1581, 2022.\\u003c/li\\u003e\\n\\u003cli\\u003eM Aghourian, C Legault-Denis, JP Soucy, P Rosa-Neto, S Gauthier, A Kostikov, P Gravel, and MA Bedard. Quantification of brain cholinergic denervation in alzheimer\\u0026rsquo;s disease using pet imaging with [18 f]-feobv. Molecular psychiatry, 22(11):1531\\u0026ndash;1538, 2017\\u003c/li\\u003e\\n\\u003cli\\u003eNatalia Jaworska, Sylvia ML Cox, Maria Tippler, Natalie Castellanos-Ryan, Chawki Benkelfat, Sophie Parent, Alain Dagher, Frank Vitaro, Michel Boivin, Robert O Pihl, and others. Extra-striatal d 2/3 receptor availability in youth at risk for addiction. Neuropsychopharmacology, 45(9):1498\\u0026ndash;1505, 2020.\\u003c/li\\u003e\\n\\u003cli\\u003eKaller, S., Rullmann, M., Patt, M., Becker, G. A., Luthardt, J., Girbardt, J., ... \\u0026amp; Sabri, O. (2017). Test\\u0026ndash;retest measurements of dopamine d 1-type receptors using simultaneous pet/mri imaging. \\u003cem\\u003eEuropean journal of nuclear medicine and molecular imaging\\u003c/em\\u003e, \\u003cem\\u003e44\\u003c/em\\u003e, 1025-1032.\\u003c/li\\u003e\\n\\u003cli\\u003eDukart J, Holiga S, Rullmann M, Lanzenberger R, Hawkins PCT, Mehta MA, Hesse S, Barthel H, Sabri O, Jech R, Eickhoff SB. JuSpace: A tool for spatial correlation analyses of magnetic resonance imaging data with nuclear imaging derived neurotransmitter maps. Hum Brain Mapp. 2021 Feb 15;42(3):555-566. doi: 10.1002/hbm.25244. Epub 2020 Oct 20.\\u003c/li\\u003e\\n\\u003cli\\u003eBurt, J.B., Helmer, M., Shinn, M.W., Anticevic, A., Murray, J.D. Generative modeling of brain maps with spatial\\u003cu\\u003e \\u003c/u\\u003eautocorrelation. Neuroimage, 220 (2020).\\u003c/li\\u003e\\n\\u003cli\\u003eBates D, M\\u0026auml;chler M, Bolker B, Walker S (2015). \\u0026ldquo;Fitting Linear Mixed-Effects Models Using lme4.\\u0026rdquo; \\u003cem\\u003eJournal of Statistical Software\\u003c/em\\u003e, 67(1), 1\\u0026ndash;48. doi:10.18637/jss.v067.i01.\\u003c/li\\u003e\\n\\u003cli\\u003eKuznetsova A, Brockhoff PB, Christensen RHB (2017). \\u0026ldquo;lmerTest Package: Tests in Linear Mixed Effects Models.\\u0026rdquo; \\u003cem\\u003eJournal of Statistical Software\\u003c/em\\u003e, 82(13), 1\\u0026ndash;26. doi:10.18637/jss.v082.i13.\\u003c/li\\u003e\\n\\u003cli\\u003eMoran, R., Pinotsis, D. A., \\u0026amp; Friston, K. (2013). Neural masses and fields in dynamic causal modeling. \\u003cem\\u003eFrontiers in computational neuroscience\\u003c/em\\u003e, \\u003cem\\u003e7\\u003c/em\\u003e, 57.\\u003c/li\\u003e\\n\\u003cli\\u003eHaarsma, J., Deveci, N., Corbin, N., Callaghan, M. F., \\u0026amp; Kok, P. (2023). Expectation cues and false percepts generate stimulus-specific activity in distinct layers of the early visual cortex. \\u003cem\\u003eJournal of Neuroscience\\u003c/em\\u003e, \\u003cem\\u003e43\\u003c/em\\u003e(47), 7946-7957.\\u003c/li\\u003e\\n\\u003cli\\u003eHuot, P., Johnston, T. H., Darr, T., Hazrati, L. N., Visanji, N. P., Pires, D., ... \\u0026amp; Fox, S. H. (2010). Increased 5‐HT2A receptors in the temporal cortex of parkinsonian patients with visual hallucinations. \\u003cem\\u003eMovement disorders\\u003c/em\\u003e, \\u003cem\\u003e25\\u003c/em\\u003e(10), 1399-1408.\\u003c/li\\u003e\\n\\u003cli\\u003eParr, T., Benrimoh, D. A., Vincent, P., \\u0026amp; Friston, K. J. (2018). Precision and false perceptual inference. \\u003cem\\u003eFrontiers in integrative neuroscience\\u003c/em\\u003e, \\u003cem\\u003e12\\u003c/em\\u003e, 39.\\u003c/li\\u003e\\n\\u003cli\\u003eTsukada, H., Fujii, H., Aihara, K., \\u0026amp; Tsuda, I. (2015). Computational model of visual hallucination in dementia with Lewy bodies. \\u003cem\\u003eNeural networks\\u003c/em\\u003e, \\u003cem\\u003e62\\u003c/em\\u003e, 73-82.\\u003c/li\\u003e\\n\\u003cli\\u003eKanwisher, N.,McDermott, J., \\u0026amp; Chun, M. M. (2002). The fusiform face area: a module inhuman extrastriate cortex specialized for face perception. N.,McDermott, J., \\u0026amp; Chun, M. M. (2002). The fusiform face area: a module inhuman extrastriate cortex specialized for face perception.\\u003c/li\\u003e\\n\\u003cli\\u003eGiussani, C., Riva, M., Gallucci, M., Boukhatem, L., Sganzerla, E. P., Demonet, J. F., \\u0026amp; Roux, F. E. (2011). Anatomical correlates for category-specific naming of living and non-living things. \\u003cem\\u003eNeuroimage\\u003c/em\\u003e, \\u003cem\\u003e56\\u003c/em\\u003e(1), 323-329.\\u003c/li\\u003e\\n\\u003cli\\u003eYang, X., Yan, J., Wang, W., Li, S., Hu, B., \\u0026amp; Lin, J. (2022). Brain-inspired models for visual object recognition: an overview. \\u003cem\\u003eArtificial Intelligence Review\\u003c/em\\u003e, \\u003cem\\u003e55\\u003c/em\\u003e(7), 5263-5311.\\u003c/li\\u003e\\n\\u003cli\\u003eMosimann, U. P., Rowan, E. N., Partington, C. E., Collerton, D., Littlewood, E., O\\u0026apos;Brien, J. T., ... \\u0026amp; McKeith, I. G. (2006). Characteristics of visual hallucinations in Parkinson disease dementia and dementia with Lewy bodies. \\u003cem\\u003eThe American Journal of Geriatric Psychiatry\\u003c/em\\u003e, \\u003cem\\u003e14\\u003c/em\\u003e(2), 153-160.\\u003c/li\\u003e\\n\\u003cli\\u003eLiu, Y., Ouyang, P., Zheng, Y., Mi, L., Zhao, J., Ning, Y., \\u0026amp; Guo, W. (2021). A selective review of the excitatory-inhibitory imbalance in schizophrenia: underlying biology, genetics, microcircuits, and symptoms. \\u003cem\\u003eFrontiers in cell and developmental biology\\u003c/em\\u003e, \\u003cem\\u003e9\\u003c/em\\u003e, 664535.\\u003c/li\\u003e\\n\\u003cli\\u003eFirbank, M. J., Parikh, J., Murphy, N., Killen, A., Allan, C. L., Collerton, D., ... \\u0026amp; Taylor, J. P. (2018). Reduced occipital GABA in Parkinson disease with visual hallucinations. \\u003cem\\u003eNeurology\\u003c/em\\u003e, \\u003cem\\u003e91\\u003c/em\\u003e(7), e675-e685.\\u003c/li\\u003e\\n\\u003cli\\u003eCarhart-Harris, R. L., \\u0026amp; Nutt, D. J. (2017). Serotonin and brain function: a tale of two receptors. \\u003cem\\u003eJournal of psychopharmacology\\u003c/em\\u003e, \\u003cem\\u003e31\\u003c/em\\u003e(9), 1091-1120.\\u003c/li\\u003e\\n\\u003cli\\u003eIgnatavicius, A., Matar, E., \\u0026amp; Lewis, S. J. (2024). Visual hallucinations in Parkinson\\u0026apos;s disease: spotlight on central cholinergic dysfunction. Brain, awae289.\\u003c/li\\u003e\\n\\u003cli\\u003eZarkali, A., McColgan, P., Leyland, L. A., Lees, A. J., \\u0026amp; Weil, R. S. (2022). Longitudinal thalamic white and grey matter changes associated with visual hallucinations in Parkinson\\u0026rsquo;s disease. Journal of Neurology, Neurosurgery \\u0026amp; Psychiatry, 93(2), 169-179.\\u003c/li\\u003e\\n\\u003cli\\u003eOnofrj, M., Russo, M., Delli Pizzi, S., De Gregorio, D., Inserra, A., Gobbi, G., \\u0026amp; Sensi, S. L. (2023). The central role of the Thalamus in psychosis, lessons from neurodegenerative diseases and psychedelics. \\u003cem\\u003eTranslational Psychiatry\\u003c/em\\u003e, \\u003cem\\u003e13\\u003c/em\\u003e(1), 384.\\u003c/li\\u003e\\n\\u003cli\\u003eOnofrj, M., Espay, A. J., Bonanni, L., Delli Pizzi, S., \\u0026amp; Sensi, S. L. (2019). Hallucinations, somatic‐functional disorders of PD‐DLB as expressions of thalamic dysfunction. \\u003cem\\u003eMovement Disorders\\u003c/em\\u003e, \\u003cem\\u003e34\\u003c/em\\u003e(8), 1100-1111.\\u003c/li\\u003e\\n\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":true,\"hideJournal\":false,\"highlight\":\"\",\"institution\":\"\",\"isAcceptedByJournal\":true,\"isAuthorSuppliedPdf\":false,\"isDeskRejected\":\"\",\"isHiddenFromSearch\":false,\"isInQc\":false,\"isInWorkflow\":false,\"isPdf\":false,\"isPdfUpToDate\":true,\"isWithdrawnOrRetracted\":false,\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"nature-portfolio\",\"isNatureJournal\":true,\"hasQc\":false,\"allowDirectSubmit\":false,\"externalIdentity\":\"\",\"sideBox\":\"\",\"snPcode\":\"\",\"submissionUrl\":\"\",\"title\":\"Nature Portfolio\",\"twitterHandle\":\"\",\"acdcEnabled\":false,\"dfaEnabled\":false,\"editorialSystem\":\"ejp\",\"reportingPortfolio\":\"\",\"inReviewEnabled\":true,\"inReviewRevisionsEnabled\":false},\"keywords\":\"\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-6627999/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-6627999/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"\\u003cp\\u003ePsychosis and visual hallucinations (VH) in Parkinson\\u0026rsquo;s disease (PD) significantly impact patient outcomes, yet underlying neural mechanisms remain unclear, limiting effective treatments. Here, we used dynamic causal modelling (DCM) to leverage the fast temporal dynamics captured with EEG data during a visual mismatch negativity task in PD patients with (N\\u0026thinsp;=\\u0026thinsp;20) and without (N\\u0026thinsp;=\\u0026thinsp;18) VH to examine effective connectivity. We found reduced top-down and enhanced bottom-up connectivity in ventral visual and prefrontal regions during task performance in PD-VH, suggesting deficits in sensory prediction updating and an over-reliance on visual input. Connectivity patterns differed with hallucination complexity: minor VH related to left hemisphere deficits, complex VH to altered top-down and bottom-up right-hemisphere connectivity, and multimodal hallucinations to widespread bilateral disruption. Increased task activity as computed with source reconstruction correlated positively with cortical 5-HT2A receptor distribution. These findings highlight specific neural targets for early therapeutic interventions, supporting a transdiagnostic computational architecture of hallucinations.\\u003c/p\\u003e\",\"manuscriptTitle\":\"Anomalies in effective connectivity explain different hallucination subtypes in Parkinson’s disease psychosis\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2025-06-11 11:43:15\",\"doi\":\"10.21203/rs.3.rs-6627999/v1\",\"editorialEvents\":[],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"nature-mental-health\",\"isNatureJournal\":true,\"hasQc\":false,\"allowDirectSubmit\":false,\"externalIdentity\":\"natmentalhealth\",\"sideBox\":\"Learn more about [Nature Mental Health](https://www.nature.com/natmentalhealth/)\",\"snPcode\":\"44220\",\"submissionUrl\":\"https://mts-natmentalhealth.nature.com/cgi-bin/main.plex\",\"title\":\"Nature Mental Health\",\"twitterHandle\":\"\",\"acdcEnabled\":true,\"dfaEnabled\":true,\"editorialSystem\":\"ejp\",\"reportingPortfolio\":\"Nature Research\",\"inReviewEnabled\":true,\"inReviewRevisionsEnabled\":false}}],\"origin\":\"\",\"ownerIdentity\":\"2b63a1df-c9b5-4db4-88a0-48eccccfc27f\",\"owner\":[],\"postedDate\":\"June 11th, 2025\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"under-review\",\"subjectAreas\":[{\"id\":49658472,\"name\":\"Biological sciences/Neuroscience/Diseases of the nervous system/Parkinson's disease\"},{\"id\":49658473,\"name\":\"Biological sciences/Neuroscience/Computational neuroscience\"},{\"id\":49658474,\"name\":\"Health sciences/Diseases/Neurological disorders/Movement disorders/Parkinson's disease\"},{\"id\":49658475,\"name\":\"Biological sciences/Neuroscience/Cognitive ageing\"}],\"tags\":[],\"updatedAt\":\"2026-05-11T14:15:19+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2025-06-11 11:43:15\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-6627999\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-6627999\",\"identity\":\"rs-6627999\",\"version\":[\"v1\"]},\"buildId\":\"8U1c8b4HqxoKbykW_rLl7\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}