Does reduced use of prior beliefs in perceptual decisions track delusion proneness and psychotic-like experience?

Why would using prior beliefs less be linked to delusions?

In everyday perception, your brain combines what it expects to see with what your eyes actually detect. If you expect a friend at the door, you may briefly see their face in a stranger's silhouette. This balancing act is often described in Bayesian terms: prior beliefs act as expectations, and sensory information acts as new evidence. Delusions have been proposed to arise when this balance goes wrong, either by overweighting expectations or by overweighting sensory input [3].

A large study of neurotypical adults found that people with more psychotic-like experiences weighted sensory information more heavily than prior expectations during a perceptual decision task [3]. The effect appeared in two independent groups, one with 363 participants and a replication with 782, which makes the pattern more robust than a single small sample. The researchers interpreted this as reduced precision given to prior information, meaning the brain treats its expectations as less reliable [3].

Belief updating after surprise is slower, not faster

If people with PLEs rely less on priors, you might expect them to update their beliefs more readily when new evidence arrives. A separate study using a helicopter-location task found the opposite pattern in some situations. Participants with higher PLEs were slower to update their beliefs after large prediction errors, meaning when the outcome was very different from what they expected [1]. They also showed a smaller increase in belief precision after a change point, the moment when the hidden helicopter's location suddenly shifted [1].

Computational modeling suggested that PLEs were associated with reduced overall belief updating in response to prediction errors and reduced modulation of updating at inferred environmental change points [1]. In plain terms, people with more PLEs did not adjust their learning rate as much when the environment signaled that a change had occurred. The authors propose this could lead to rigid beliefs, because failing to respond to change points may limit flexibility when contradictory evidence appears [1].

The problem may be imprecision in both directions

The picture is not simply 'less prior, more sensory.' The same large perceptual study found that when prior information was more uncertain, participants with greater PLEs encoded sensory information with greater noise [3]. That means their sensory evidence was also less precise, not just their priors. The researchers suggest this may reflect generally heightened perceptions of task instability, as if the world feels less predictable overall [3].

This helps reconcile seemingly conflicting findings. Reduced reliance on priors in one context and slower updating after surprise in another can both arise from a system that assigns unreliable precision to multiple sources of information. The authors of the perceptual study lend empirical support to notions of both weaker bottom-up and weaker top-down processes, rather than a simple overreliance on either [3].

Does this extend beyond perceptual decisions?

The link between PLEs and belief-related phenomena appears outside the lab as well. In a community sample of 438 people during the COVID-19 pandemic, psychotic-like experiences were associated with beliefs in COVID-19 conspiracy theories, particularly perceptual abnormalities and persecutory ideation [2]. That study did not measure prior use in perceptual decisions, so it cannot show that reduced prior weighting causes conspiracy beliefs. It does show that PLEs track with a tendency to endorse illogical or erroneous arguments, which is consistent with altered belief evaluation more broadly [2].

A different line of evidence complicates the idea that PLEs simply mean weaker priors. In a fear-learning study, delusion-prone individuals showed stronger instructed fear learning, where verbal suggestions about contingencies shaped their responses, compared with non-delusion-prone subjects [5]. This suggests that under some conditions, psychosis-related traits are associated with stronger reliance on high-level expectations, not weaker ones. The authors note this is paradoxical alongside evidence of weak classical fear conditioning in psychosis-related states [5].

Group beliefs also matter for PLE proneness. A study of 159 individuals found that people endorsing esoteric and paranormal beliefs had significantly higher PLE scores than religious believers or skeptics, with mean scores roughly twice as high [4]. That study measured PLEs with a questionnaire and did not test perceptual decision-making, so it does not directly address prior use. It does indicate that PLE proneness varies across non-clinical belief groups, which may be relevant to how beliefs are formed and maintained [4].

Sources used in this answer

Related research questions