Umwelten

Roseby & Osborn Moar 2025 — Predictive processing as a lens on psychedelics and autism

🔬ResearchPeer-reviewed

5 min read

Citation

Roseby, W., & Osborn Moar, C. (2025). “A Comparative Neurophenomenology of the Psychedelic State and Autism: Predictive Processing as a Unifying Lens.” Psychoactives, 4, 41. doi:10.3390/psychoactives4040041. Open Access, CC BY 4.0. Sussex Centre for Consciousness Science, University of Sussex.

What this paper is

A 35-page narrative review and conceptual synthesis comparing the psychedelic state and autism from a neurophenomenological perspective, using predictive processing (PP) as a unifying framework. This is a theoretical paper — no original data — but it is the most developed attempt to date to bridge psychedelic neuroscience and autism research through a shared computational framework. It generates testable hypotheses rather than confirming them.

Key arguments

Both autism and psychedelics involve a shift towards sensory information

Under predictive processing, the brain is modelled as continuously generating predictions about incoming sensory data and updating those predictions based on prediction errors. The balance between “top-down” priors (what the brain expects) and “bottom-up” sensory evidence (what actually arrives) can be shifted.

In autism, several influential accounts (Pellicano & Burr 2012, Van de Cruys et al. 2014, Lawson et al. 2014) propose that prediction errors are weighted more heavily — the brain treats incoming sensory information as more important relative to its prior expectations. This maps onto the lived experience of sensory sensitivity: the world is louder, brighter, more detailed, more surprising than it is for neurotypical people, because the brain’s filter is set to let more sensory evidence through.

Under psychedelics, the REBUS model (Relaxed Beliefs Under Psychedelics; Carhart-Harris & Friston 2019) proposes that 5HT₂A stimulation reduces the precision of high-level priors, particularly those related to the self (default mode network). This also shifts the balance towards sensory information — but at a different level of the cortical hierarchy.

The same direction of shift, but at different levels

This is the paper’s sharpest conceptual contribution. Both autism and psychedelics involve “more sensory, less prior” — but:

  • In autism, the alteration may be primarily at lower levels of the cortical hierarchy (sensory cortex, early processing), producing heightened detail sensitivity, sensory overload, and the characteristic presentation of hyper- and hyporesponsivity.
  • Under psychedelics, the alteration is primarily at higher levels (default mode network, self-referential processing), producing ego dissolution, enhanced psychological flexibility, and altered self-world boundaries.

The comparison is not a claim that autism and psychedelic states are the same thing. It is a claim that comparing them through a shared framework illuminates both.

Context sensitivity as a shared feature

Both autistic people and psychedelic users show heightened sensitivity to context — but this is interpreted differently in each field. In psychedelic research, context sensitivity is foregrounded (the concept of “set and setting”). In autism research, context sensitivity is under-theorised despite clear evidence that the same autistic person processes stimuli very differently depending on environment, emotional state, and felt safety (see the SGL synthesis darts-match example in Hypo- and hyperresponsivity).

Psychological flexibility is not one thing

The paper argues that the concept of “psychological flexibility” — central to psychedelic research — needs refinement. Psychedelics enhance some forms of flexibility (associative thinking, ego dissolution, insight) while potentially reducing others (set-shifting, working memory). Autism involves yet another pattern — reduced flexibility in some domains (routine, predictability) combined with extraordinary flexibility in others (pattern recognition, detail processing, deep-domain expertise). Treating “flexibility” as a unitary construct conflates phenomena that are neurobiologically distinct.

Method in brief

Narrative review with targeted searches of PubMed and Google Scholar (July–September 2025). Inclusion was based on relevance to neurophenomenological and theoretical accounts rather than methodological uniformity. Focus on autistic adults without intellectual disability (some child studies included where adult evidence was thin). No formal risk-of-bias assessment.

Relevance to this wiki

This paper provides a theoretical bridge between two things this wiki already articulates:

  1. The phenomenological observation that sensory processing differences in autism are about how the brain weighs incoming information — not about broken senses.
  2. The emerging pharmacological evidence (see Whelan et al. 2024 — PSILAUT) that serotonin pathways, specifically 5HT₂A, are implicated in how that weighing works.

The predictive processing framework also strengthens the wiki’s existing “mechanisms (tentative)” section in Sensory processing in autism and intellectual disability, which lists four competing accounts (modulation, filtering, integration, arousal). Predictive processing offers a meta-framework that can subsume all four: modulation and filtering are about how prediction errors are weighted; integration is about how multi-sensory predictions are combined; arousal is about the overall gain setting on the system.

For practitioners, the most actionable insight is about context sensitivity: if autistic sensory processing is modulated by the precision assigned to predictions (which is itself modulated by context, safety, and attention), then environmental design — making environments predictable, safe, and low-surprise — is not a workaround for a broken system. It is a direct intervention on the computational process that produces sensory experience.

Limitations

  • Narrative review, not systematic. The authors acknowledge this and frame their conclusions as conceptual/testable hypotheses rather than established findings.
  • Focuses on adults without ID. Most predictive processing research in autism studies higher-functioning adults. Whether the framework applies to autistic people with intellectual disability is an open question — one that the wiki should track as an evidence gap.
  • Compares a transient state with a lifelong trait. The psychedelic experience lasts hours; autism is developmental and lifelong. The comparison illuminates mechanisms but cannot be taken as implying the experiences are equivalent.
  • Both “fields” are internally heterogeneous. There is no single autism and no single psychedelic experience. The paper necessarily simplifies both.

Nearby

Backlinks