Neurocomputational approaches for scalable psychosis detection: From lab‑based neuroimaging to real‑world digital tasks

This talk advances cognitive network modeling - a translational framework for scalable evaluation of psychosis symptoms - by aligning a neurocomputational account of learning with the temporal unfolding of clinical symptoms. In this talk, I outline a framework in which precision‑weighted prediction errors and belief stability provide mechanistic bridges from cellular/neurobiological processes to behavior, emphasizing the opportunity for pre‑onset intervention. Methodologically, we integrate neuroimaging methods, including EEG and fMRI, with validated, gamified tasks (MindMetrics) spanning social cognition, sensory learning, and probabilistic reasoning. Individualized computational parameters derived from these tasks enable reliable longitudinal tracking of across cognitive domains, supporting deployment at scale. Together, these components specify a pipeline, from mechanistic modeling to low‑burden digital tools, designed to enhance early identification, monitoring, and care pathways for youth at elevated risk for psychosis.

Alternate locations:

Online: https://utoronto.zoom.us/j/88523051719