Hallucinations Under Psychedelics and in the Schizophrenia Spectrum: An Interdisciplinary and Multiscale Comparison

Hallucinations—percepts experienced without corresponding external stimulus—occur across a range of contexts, including schizophrenia-spectrum disorders (SCZs), certain neurological illnesses, the general population, and following administration of psychotomimetic drugs such as classical serotonergic psychedelics. Since the nineteenth century, researchers have debated whether drug-induced states can serve as models for psychosis. The recent resurgence of clinical and neuroscientific work on psychedelics has renewed interest in similarities and differences between pharmacologically induced experiences and those seen in SCZs. Leptourgos and colleagues, reporting from a multidisciplinary working group of the International Consortium on Hallucinations Research, set out to compare hallucinations under classical psychedelics with those observed in the schizophrenia spectrum. Their approach is explicitly multiscale: they examine synaptic pharmacology, network-level brain imaging, first-person phenomenology, anthropological evidence about cultural shaping of experiences, and computational models that might bridge these levels of analysis. The stated aim is to identify commonalities and differences and to suggest directions for future empirical work that can integrate these scales of explanation.

This paper is a narrative, multidisciplinary review produced by a working group convened within the International Consortium on Hallucinations Research. Rather than reporting a formal systematic review or meta-analytic search, the authors synthesise evidence across several domains: pharmacology (synaptic mechanisms and animal assays), brain-imaging studies (fMRI, MEG and effective/connectivity analyses), phenomenological reports (modalities and content), anthropological and cross-cultural literature, and computational psychiatry models. The extracted text does not report a formal literature-search strategy, inclusion/exclusion criteria, databases searched, or a risk-of-bias assessment, so the review should be understood as an expert, integrative synthesis rather than a systematic review. Methodologically, the working group framed their comparison at multiple scales. In the pharmacology section they review animal behavioural paradigms (drug discrimination, head-twitch response, prepulse inhibition) and human pharmacological blockade studies. For brain imaging they juxtapose capture studies that sample phasic hallucinatory ON/OFF states in SCZs with neuroimaging of drug-induced states (which typically measure sustained intoxication rather than momentary hallucinations) and discuss both functional connectivity (FC) and effective connectivity findings. Phenomenology draws on prevalence estimates, descriptive clinical studies, and reports from psychedelic research. The anthropology section surveys ethnographic and cross-cultural findings, and the computational section focuses on Bayesian-inference frameworks (predictive coding, precision-weighting, belief propagation/circular inference) as mechanistic hypotheses linking synaptic changes to altered perceptual inference.

Pharmacology: The review outlines a clear pharmacological distinction between SCZs and classical psychedelics at the synaptic level. Schizophrenia-spectrum disorders have long been linked with dopaminergic alterations, whereas classical psychedelics (LSD, mescaline, psilocybin) are principally serotonergic agonists with the 5-HT2A receptor recognised as the primary target for hallucinogenesis. Evidence cited includes correlations between 5-HT2A affinity and potency in animal drug-discrimination paradigms, blockade of the head-twitch response (HTR) and other behaviours by selective 5-HT2A antagonists, and loss of HTR in 5-HT2A knockout mice. Human pharmacology studies are reported in which ketanserin (a 5-HT2A antagonist) blocks psilocybin effects, risperidone (5-HT2A/D2 antagonist) can block psilocybin, whereas the D2 antagonist haloperidol was not effective; intensity of psilocybin effects has been correlated with central 5-HT2A receptor occupancy. Brain imaging: At the network level, both overlaps and divergences emerge. In SCZs, fMRI capture studies of hallucinatory ON states implicate overactivation of modality-specific associative cortices, with primary sensory cortex recruitment associated with more vivid experiences; hippocampal hyperactivity and varied patterns of activation/deactivation have also been reported. By contrast, imaging under psychedelics typically finds increased activity in primary sensory cortex (for visual hallucinations, increased blood flow in visual cortex) accompanied by reduced processing in higher-order associative visual areas, suggesting enhanced early sensory input with diminished associative constraining. Both conditions show reduced internal integration of canonical resting-state networks and altered antagonism between the default-mode network (DMN) and central-executive network (CEN), interpreted as blurring between internally and externally oriented processing. Psychedelic states additionally show disintegration of resting-state network coherence, expanded functional connectivity of primary visual cortex, increased undirected FC alongside decreased directed FC (LSD MEG findings), and altered thalamocortical connectivity; LSD in one report increased effective connectivity from thalamus to certain DMN areas and increased thalamic connectivity with fusiform gyrus and anterior insula that correlated with visual and auditory hallucinations respectively. In SCZs, prefrontal-thalamic FC tends to be reduced while thalamic FC with sensorimotor areas can be strengthened, although relationships with symptoms are mixed. Phenomenology: The modal profile of hallucinations differs markedly. In SCZs auditory hallucinations (AH) are most frequent (reported prevalence around 79%), about three times more common than visual hallucinations (VH; mean prevalence ~27%). Other modalities (olfactory, gustatory, somatic) show widely varying prevalence estimates across studies (olfactory 6–26%, gustatory 1–31%, somatic/tactile 4–19%). Hallucinations in SCZs are often detailed, concrete, life-sized, and poorly discriminated from veridical perception; reality testing is often impaired. By contrast, 5-HT2A agonists predominantly induce visual phenomena—both elementary geometric "form constants" and complex imagery—with synesthesia-like experiences common and olfactory/gustatory phenomena rare. Drug-induced VH increase in complexity and frequency with dose and time, but insight about their drug origin is typically preserved. Multimodal or fused hallucinations (MMH) are variably reported in SCZs and under psychedelics; the prevalence of MMH in both contexts requires clearer empirical study. Both types of experience can be highly meaningful and emotionally salient, with metaphysical or spiritual qualities frequently reported. Anthropology: Cross-cultural and ethnographic evidence emphasises significant contextual shaping of hallucinatory content. Some phenomenological features (eg, geometric VH) recur across cultures, while other aspects (tone, meaning, valence) vary widely. Traditional uses of serotonergic plants and fungi span divinatory, healing, initiation and recreational contexts, and ritual/setting variables are proposed as important determinants of experience. Laboratory analyses cited (Studerus et al.) indicate that personality trait absorption, pre-drug state of mind, and recent psychological stability predict more positive psychedelic experiences, whereas emotional excitability, younger age, and equipment-heavy settings predict more negative reactions. Cross-cultural work on hallucinations in SCZs is limited and heterogeneous, with one cited cross-country prevalence study showing wide variability but not analysing cultural correlates in depth. Computational modelling: Bayesian frameworks are offered as bridges between synaptic/neural changes and altered perception. Predictive-coding models posit hierarchical inference in which sensory inputs are compared with top-down predictions; discrepancies produce prediction errors (PE) that update beliefs. Precision-weighting (the reliability assigned to inputs versus priors) governs how strongly PE influence updating. Two contrasting accounts are discussed: Corlett and colleagues propose that psychedelics may produce over-weighted priors (strong priors dominating perception), whereas Carhart-Harris and Friston's REBUS model argues that psychedelics relax high-level priors (increasing reliance on sensory input). The authors note these accounts may be reconciled if priors are differentially weighted at distinct hierarchical levels. Alternative computational schemes such as belief propagation and circular inference are also described: ascending loops (overcounted inputs) have been linked empirically to unimodal hallucinations like AH in SCZs, while descending loops (overcounted priors) could theoretically produce multisensory/synesthetic phenomena seen under some psychedelics, though the latter remains speculative.

Leptourgos and colleagues interpret their synthesis as indicating both shared and distinct mechanisms underlying hallucinations in SCZs and under classical psychedelics. Shared features include reduced integration and stability of large-scale functional networks and a breakdown in the normal opposition between internally and externally oriented networks, phenomena that may blur the boundary between self-generated and perceived contents. Both types of experience are also noted to carry strong personal and metaphysical meaning. Important distinctions are emphasised. Pharmacologically, classical psychedelics act principally at 5-HT2A receptors whereas SCZs have been most consistently associated with dopaminergic abnormalities. Neuroimaging differences include the tendency for SCZ-related hallucinations to implicate associative cortical overactivation, while psychedelics more often over-engage primary sensory cortices. Phenomenologically, auditory-dominant, detailed, and reality-imperceptible hallucinations typify SCZs, whereas drug-induced phenomena are predominantly visual, often geometric or synesthetic, and typically experienced with preserved insight. The authors position computational models as promising integrative frameworks but caution that competing formulations (strong priors versus relaxed priors) and hierarchical nuances complicate a single simple account. The review acknowledges several limitations and uncertainties. It does not present a systematic literature search or quantitative synthesis, and many empirical gaps remain: the role of serotonin in SCZs is not well established; the true prevalence and character of multimodal hallucinations across contexts are unclear; experimental tests of descending-loop circular inference in psychedelics are lacking; and cross-cultural research on hallucinations in SCZs is sparse and methodologically heterogeneous. The authors therefore call for targeted empirical work that more directly compares phasic hallucinatory states across pharmacological and clinical contexts, uses designs able to link synaptic mechanisms to network dynamics and phenomenology, and systematically examines cultural and extra-pharmacological moderators ("set and setting").

The working group concludes that hallucinations under psychedelics and in the schizophrenia spectrum share several neurofunctional signatures—reduced network integration and altered DMN/CEN relationships—and a tendency to be experienced as highly meaningful, yet they differ in key ways: psychedelics preferentially engage primary sensory cortices and produce predominantly visual, often synesthesia-like phenomena with preserved insight, while SCZs-related hallucinations are commonly auditory, tied to associative networks, and associated with impaired reality monitoring. The authors recommend further research to clarify the role of serotonin in SCZs, to determine the prevalence and mechanisms of multimodal hallucinations, to test computational hypotheses (eg, ascending versus descending loops), and to explore cultural moderators that may shape hallucinatory content and clinical response.