Model-based characterization of cognitive-emotion brain networks in humans with and without depressive symptoms

About This Grant

Project Summary/Abstract Depression is a leading global cause of disability in the United States and worldwide. Despite the availability of pharmacological and behavioral therapies, approximately 30% of individuals with depression remain treatment- resistant, in part due to persistent cognitive and emotional impairments. These deficits are thought to reflect disrupted interactions across prefrontal, temporal, and limbic brain regions, but the underlying neural mechanisms remain poorly understood. Prior stereotactic EEG (sEEG) studies from our group have shown that increased theta/alpha (4–15 Hz) power and altered frontotemporal connectivity during cognitive control differentiate individuals with and without depressive symptom history. However, it remains unclear whether emotion processing engages these same/overlapping networks, and whether biophysical model-based cognitive-emotion network features are associated with symptom severity. This proposal tests an overall hypothesis that dysregulated communication across fronto-temporo-limbic networks during simultaneous cognitive and emotional demands will distinguish between individuals with or without high (moderate to severe) depressive symptoms. To test this, we will examine large-scale neural dynamics underlying cognitive conflict and emotion processing using sEEG recordings from epilepsy patients with minimal to severe depression symptoms. In Aim 1a, we will assess whether individuals with high depressive symptoms demonstrate altered behavioral performance and fronto-temporo-limbic responses during a Multi-Source Interference Task (MSIT) with negative and neutral emotional distractors. In Aim 1b, we will use Jansen-Rit neural mass models within a Dynamic Causal Modeling (DCM) framework to estimate network interactions encoding cognitive conflict and emotional stimuli. In Aim 2, we will apply machine learning models trained on spectral, connectivity, and DCM-derived network features to determine their predictive validity for distinguishing between participants with and without high depression symptoms. Together, these aims are expected to identify biologically plausible mechanisms of cognitive-emotional dysfunction that map onto depression severity to inform future targeted interventions. The proposed fellowship training plan includes comprehensive mentorship from experts in neural engineering, computational modeling, and neuropsychiatry that will be crucial for developing an interdisciplinary skillset. Through the proposed training plan, the PI will gain expertise in advanced sEEG, biophysical modeling, and neural decoding techniques, as well as dedicated translational research training with external mentors. The PI, an MD/PhD student at the University of Cincinnati, will utilize outstanding institutional resources and mentorship to support their future career goal of advancing treatment for neuropsychiatric disorders as an independent physician-scientist.