PROJECT SUMMARY
Psychotic disorders (PD) affect 3.5% of the population. They are impairing, chronic, and result in reduced
disability-adjusted life years and premature death. Advances in assessment and treatment of PD are slowed by
the need for identification and mechanistic understanding of biomarkers by which symptoms arise and persist.
The mismatch negativity (MMN), an event-related potential elicited by expectancy violations, has been proposed
as a biomarker in PD: its reliable reduction is associated with both psychotic and cognitive symptoms. However,
mechanisms through which MMN is associated with such symptoms or indexes their course over time are not
clear, limiting the clinical utility of this effect. Predictive coding theory (PC) attempts to rectify this by establishing
links between neurobiological and clinical phenomena. PC posits a hierarchical organization of brain function
whereby sensory input and prior expectations (priors) are integrated to inform perception. When inputs diverge
from priors, the mismatch gives rise to prediction error (PE), which contributes to belief updating. Deficits in PE
are thought to explain important aspects of psychotic symptoms and cognitive functioning; however, empirical
support is sparse, and whether such PE reductions are associated with overly strong or weak reliance on priors
is not clear. Computational work suggesting that MMN is a neural representation of sensory PE provides a
framework for understanding mechanistic links between MMN and symptoms. Furthermore, oscillatory-based
effective connectivity is a critical neural information-routing mechanism through which top-down priors and
bottom-up PEs are conveyed, and can be quantified using oscillatory activity underlying MMN. PE has not been
derived from MMN in PD, and effective connectivity underlying MMN reduction is not well understood.
Importantly, emotion also plays a critical role in the development and maintenance of psychotic symptoms and
impairs cognition. Quantifying PE from emotion-MMN (eMMN) allows us to elucidate mechanisms through which
emotion exacerbates symptoms, lending explanatory value and utility to research in this area. Finally, though
knowledge regarding course of illness is crucial to informing intervention, the utility of MMN and PE in predicting
illness trajectories is unknown. Thus, this project aims to elucidate the neurobiological and computational
mechanisms through which MMN reduction indexes clinical and cognitive symptoms in PD over time. This study
capitalizes on a large (N=220), transdiagnostic cohort with PD and a never-psychotic group (N=252), followed
over 3 timepoints. Computational models will be used to derive PE from MMN and eMMN, and effective
connectivity to characterize feedforward (PE) and feedback (priors) oscillatory information flow. Overall, this
study uses a well-replicated, biologically-based measure and theoretical framework to elucidate computational
and neurobiological mechanisms underlying psychotic and cognitive symptoms and provide insight into their
trajectories. This project will facilitate training for the applicant in 1) computational modeling, 2) time-frequency
analyses and effective connectivity, 3) longitudinal analyses, and 4) activities for professional development.