Modeling perception and behavior in individuals at clinical high risk for psychosis: Support for the predictive...

Area of Science:

• Neuroscience
• Computational Psychiatry
• Clinical Psychology

Background:

• Early intervention is crucial for psychotic spectrum disorders.
• Risk stratification requires reliable biomarkers.
• Computational modeling of behavioral tasks offers a novel approach to understanding psychosis pathophysiology.

Purpose of the Study:

• To investigate the utility of computational tasks in identifying individuals at clinical high risk for psychosis (CHR).
• To examine differences in perceptual inference between CHR individuals and healthy controls (HC).
• To explore the potential of these tasks as biomarkers for psychosis risk.

Main Methods:

• Administered the Conditioned Hallucinations (CH) task and Sine-Vocoded Speech (SVS) task to CHR and HC participants.
• Utilized computational modeling to analyze task performance, focusing on Bayesian inference principles.
• Assessed hallucination propensity and speech perception abilities.

Main Results:

• CHR participants exhibited more conditioned hallucinations and higher pre-training SVS detection compared to HC.
• Computational modeling revealed that CHR participants showed poorer recognition of task volatility and a trend toward higher weighting of priors on the CH task.
• Performance on both tasks was interrelated, suggesting shared underlying perceptual inference mechanisms.

Conclusions:

• The CH and SVS tasks, analyzed computationally, may capture similar latent factors in perceptual inference.
• These tasks show potential as easily obtainable biomarkers for identifying individuals at clinical high risk for psychosis.
• Computational approaches to behavioral tasks can enhance risk stratification and inform early intervention strategies.