Do Psychedelics Change Beliefs?

The paper opens by noting renewed scientific and public interest in psychedelics alongside longstanding debates about whether and how these substances change human beliefs. The authors frame belief change as a multifaceted question: reports of profound, sometimes lasting changes in religious, political and self-related beliefs exist, but such changes could arise through multiple pathways including expectancy effects, social suggestion, and selection biases among research participants. Mcgovern and colleagues set out to unpack existing findings on belief change after psychedelic consumption and to clarify the mechanisms that might underlie those changes. They aim to define what counts as a belief, distinguish alternative conceptualisations (propositional attitudes, associative representations, and inferential/learning-based models), and to assess, against those definitions, whether acute or longer-term psychedelic exposure produces genuine belief change. The authors emphasise caution: apparent belief changes may reflect pre-existing expectations or social influences rather than pharmacologically induced formation of new propositional beliefs.

The authors first consider competing definitions of belief because whether psychedelics “change beliefs” depends on which definition is used. Three formulations are reviewed: propositional beliefs (linguistically expressible attitudes about truth), associative accounts (strengths of links between representational features, akin to associative learning), and inferential/learning frameworks (Bayesian-style priors, likelihoods and posterior distributions that encode both expected values and precision/confidence). The paper emphasises the distinction between inference (transient explanations of sensory data) and learning (longer-term updating of priors), arguing this distinction is central to interpreting psychedelic effects. Evidence about propositional beliefs is mixed. Some studies and surveys associate psychedelic use with more positive outlooks, reductions in authoritarianism, more liberal political attitudes and increased pro-environmental stances; however, the authors note that these effects are contested and susceptible to confounds. Self-report and longitudinal data suggest that weaker or already-leaning attitudes are more likely to shift under psychedelics, whereas strongly held beliefs appear resistant to change. Social and environmental factors feature prominently in the empirical picture. Propositional beliefs depend heavily on language and social transmission, and experimental work shows that social suggestion and group norms can shape attitudes during psychedelic exposure. For example, under LSD participants shifted their aesthetic judgements closer to a presented group norm, but this conformity occurred mainly for less extreme, already proximate attitudes. Such findings suggest psychedelics may amplify deference to others' input for tentative beliefs rather than creating wholly new convictions. Person-level variables and selection effects are emphasised as important moderators. Volunteers for psychedelic studies often have baseline openness to alternative or spiritual ideas, and those baseline expectations appear to bias both acute experiences and subsequent attributions of meaning. Laboratory work reports increased suggestibility under LSD, with the largest effects in participants higher in trait conscientiousness. The authors also highlight strong placebo and expectation effects in this literature and note that ritual and cultural framing in anthropological settings can prime both the acute experience and later belief attributions. Two conceptual ‘‘boxes’’ synthesise mechanistic debates. Box 1 addresses whether psychedelics weaken or strengthen priors in an inference-based account. The authors distinguish spatial/hierarchical priors (top-down, high-level structure) from temporal priors (shorter-timescale sequence statistics), and argue that acute and longer-term psychedelic effects are compatible with a strong-prior account in some domains: for instance, increased persistence in bistable perception under psychedelics is interpreted as overcounted priors. Box 2 summarises correlations between the phenomenology of the psychedelic experience (mysticality, meaningfulness) and clinical or behavioural outcomes: higher scores on instruments such as the MEQ30 and persisting effects questionnaires predict reductions in anxiety and depression and behavioural changes in addiction samples, and intensity of acute experience has been linked to sustained personality changes (e.g. increases in openness) in some cohorts. Mechanistic evidence implicates 5-HT2A receptor activity and downstream cortical effects in both acute perceptual phenomena and possible longer-term changes. Acute agonism at 5-HT2A receptors on pyramidal neurons is proposed to alter cortical dynamics and visual processing. In human experimental work, pre-treatment with the 5-HT2A antagonist ketanserin blocked certain LSD-induced conformity effects, implicating this receptor in social-norm sensitivity under psychedelics. Preclinical studies additionally identify medial prefrontal cortex 5-HT2A signalling as a locus for memory reactivation and reconsolidation, offering a potential route by which psychedelic exposure might strengthen or reweight existing priors via reconsolidation mechanisms. Adverse and non-propositional persistent effects are discussed. Persistent perceptual changes (HPPD) occur in some recreational users and manifest as simple visual phenomena; these often preserve insight and do not typically induce new propositional metaphysical beliefs. The authors interpret such phenomena as possible evidence of altered associative priors rather than wholesale formation of new propositional beliefs. Overall, the synthesis leads to a tempered conclusion: there is equivocal evidence that psychedelics directly change propositional beliefs. The stronger, more consistent evidence supports effects on associative processes, on inference and on learning strategies—particularly in social contexts—often moderated by expectation, set and setting, and individual differences. The authors therefore caution against straightforward claims that psychedelics induce lasting, content-specific belief changes without accounting for these converging non-pharmacological influences.

Mcgovern and colleagues interpret the assembled evidence as indicating that psychedelics are more likely to alter the weighting and updating of associative and inferential processes than to implant new propositional beliefs de novo. They propose that acute changes in inference (how sensory data are explained) can be dissociated from longer-term learning (how priors are updated), and that many reported long-term shifts in attitudes or behaviour may arise through interactions between pharmacology, suggestion, and pre-existing psychological predispositions. The authors place their synthesis in relation to earlier work by noting convergent findings—correlations between phenomenological intensity and clinical outcomes, increased suggestibility, and 5-HT2A involvement—but they stress that these correlations do not establish causality. Methodological issues they acknowledge include sample selection bias (volunteers often hold prior spiritual or environmental orientations), strong placebo effects, the challenge of adequate blinding, and difficulty separating pharmacological effects from cultural and ritual framing. They also raise conceptual uncertainty about which level of description (associative links, propositional attitudes, or probabilistic belief distributions) best captures belief change. Limitations and open questions identified by the authors include the equivocal nature of evidence for propositional belief change, incomplete understanding of whether psychedelics first alter non-social processes which then affect social learning (or vice versa), and the need to clarify whether phenomena such as ego-dissolution cause, follow from, or simply correlate with changes in social weighting. The authors recommend experimental priorities: direct tests of social versus non-social learning under psychedelics, mechanistic studies of 5-HT2A-mediated reconsolidation, and attempts to disentangle the causal role of subjective mystical experiences—for example, by probing effects when conscious experience is attenuated (noting ethical and safety constraints). For future research and practice, the paper advocates adopting formal computational frameworks that model beliefs and belief-updating at algorithmic and implementational levels. Such frameworks, the authors argue, can guide empirical designs that collect neural and behavioural data before, during and after psychedelic exposure, thereby clarifying whether changes reflect probabilistic reweighting of priors, associative learning, or shifts in propositional attitudes. They also flag applied concerns: understanding individual differences in social sensitivity may be important for predicting therapeutic benefits versus adverse social or paranoid reactions, and attention should be paid to potential phenomena such as spiritual narcissism among some users.

The authors conclude that psychedelics may facilitate lasting change, but that whether these changes amount to new propositional beliefs remains unresolved. They call for systematic, model-driven assessments of belief and belief updating, leveraging formal mathematical and computational models to interrogate neural and behavioural data collected before, during and after psychedelic exposure. Mcgovern and colleagues express hope that psychedelic research will inform fundamental questions about how brains represent and update beliefs, and that such work may illuminate both pathological and healthy belief formation.