Welcome to Psychiatryai.com: Latest Evidence - RAISR4D

Hallucination Proneness is Linked to Over-Reliance on Internal Priors for Noisy Speech

AI Summary
  • Hallucination-proneness linked to reduced ability to distinguish intelligible from unintelligible degraded speech and a higher bias to report speech presence in noise.
  • Greater influence of internally generated priors on perception when incorrect, and reduced reliance on externally provided priors when correct.
  • Supports predictive processing models: imbalance favouring internal over external priors contributes to hallucination-proneness in a non-clinical sample.
Summarise with AI (MRCPsych/FRANZCP)

Schizophr Bull. 2026 Apr 10;52(3):sbag048. doi: 10.1093/schbul/sbag048.

ABSTRACT

BACKGROUND: According to predictive processing models, hallucinations can arise from over-weighting of prior expectations relative to sensory input. Individuals who hallucinate may misattribute internally-generated experiences to external sources. Combining these mechanisms, we hypothesized that hallucination-proneness in a non-clinical sample is associated with a greater reliance on internally-generated priors compared to externally-provided priors during degraded speech perception.

STUDY DESIGN: Two online experiments were conducted with healthy adult participants. Experiment 1 (n = 92) targeted lower-level perceptual processes by assessing sensitivity and bias in distinguishing regular vocoded speech from unintelligible, spectrally-rotated vocoded stimuli at varying levels of sensory clarity. Experiment 2 (n = 100) compared the influence of internally-generated versus externally-provided priors on perception of vocoded speech content. Internal- and external-priors were matched in reliability to test if hallucination-proneness was linked to differential influences of internally-generated priors on speech perception.

STUDY RESULTS: Experiment 1 confirmed that participants who are more hallucination-prone showed a reduced ability to distinguish potentially intelligible speech from unintelligible speech, and a higher bias toward indicating that speech was present in unintelligible stimuli. Experiment 2 confirmed differential effects of internal priors, showing that hallucination-proneness scores were linked to a greater influence on internally-generated priors on perceptual report when these were incorrect, and a lower reliance on externally-provided priors when these were correct.

CONCLUSIONS: These findings suggest that hallucination-proneness is linked to an increased influence of internally-generated compared to externally-provided predictions on perception, suggesting an additional element of predictive processing theories of hallucinations.

PMID:42104803 | DOI:10.1093/schbul/sbag048

Document this CPD

AI Search

Share Evidence Blueprint

QR Code

Search Google Scholar

Save as PDF

close chatgpt icon
ChatGPT

Enter your request.

Psychiatry AI: Real-Time AI Scoping Review