Reading focus
Syllabus: How Do Psychotic Experiences Develop and Persist?
Compare the voice-hearing and diagnostic-status contrasts before interpreting the fitted model.
Why you’re reading it. This gives a bounded experimental example of how an explanatory account meets data.
A question to consider. Which finding concerns hallucination status and which concerns clinical status, and why would merging them obscure the question Garety’s broader model is trying to answer?
Why this reading
This experiment asks how learned expectations can influence hearing a signal that is absent. By comparing voice-hearing and clinical status separately, it tests whether a laboratory effect follows an experience rather than a diagnosis alone. It also exposes the steps from behavior to a computational explanation and a brain correlate.
Before you read
A prior is an expectation brought to incoming evidence. Conditioning establishes a predictive relationship between events. Participants learned an association between a visual cue and a tone; a reported tone on an absent-tone trial counted as a conditioned hallucination. The hierarchical model represents beliefs at different levels, including how changeable the cue–tone relationship is.
Read for
- Use the four groups to distinguish voice-hearing status from diagnosed psychotic illness. What comparison would be lost with only patients and healthy controls?
- Follow individual hearing thresholds, tone-present and tone-absent trials, confidence and the changing cue relationship. What alternatives to altered perception do these measures address?
- Separate observed responses from fitted model parameters and imaging associations. Which part of the strong-prior interpretation depends on the chosen model?
Consider afterwards
- If two groups differ in reporting absent tones, what further evidence would distinguish a perceptual difference from a response strategy?
- What would a competing model have to predict to explain the same data without the proposed weighting of expectations?
Editorial context
The four groups contained 59 participants in total. A conditioned tone report is not identical to everyday voice-hearing, and diagnostic groups were not randomly assigned. Imaging responses and model fits strengthen a particular account without uniquely establishing it or a treatment mechanism. The findings also separate hallucination-related effects from belief-updating differences associated with clinical status; those results should not be collapsed into one explanation of all psychosis.