Intracranial Investigation of Neural Circuity Underlying Human Mood

We apply novel modeling approaches to unique human intracranial data with the goal of understanding the neural basis of depression. Aims 1 and 2 apply complementary approaches to achieve the goal of relating mood states to neural dynamics. Aim 3 then models the effect of stimulation to derive causal understanding of the systems response to modulation. In Aim 1, we decode mood state obtained from subject self report to produce reliable neural correlates of mood. To do so in an informative way, we use a ladder of models to build from conventional AI models, with their known limitations, to novel mechanistically explainable dynamic models. In Aim 2, we use an alternative transdiagnostic approach inspired by the RDoC. Rather than measuring depression as variations in self reported mood and symptoms, we study how depression manifests in behavior. In particular, we examine performance on a suite of tasks targeted to reveal patients characteristics along neurobiologically relevant axes of Positive Valence, Negative Valence, and Cognitive Systems. We apply our novel inverse rational control methodology to infer the subjects internal models of tasks from observed behavior. This process then allows us to estimate neural correlates of relevant (RDoC-based) latent parameters such as reward sensitivity, loss aversiveness, cognitive flexibility, etc. Side by side comparison of results from Aims 1 and 2 will thus allow synergistic understanding of the brain behavior relationships related to mood and depression. To improve our therapeutic interventions, in Aim 3 we will quantify brain responses to electrical stimulation. To model and explain these measurements, we will apply recurrent neural networks to network responses measured from high entropy stimulation patterns to build predictive models of neural responses to stimulation. We then use our novel inception loop strategy to generate optimized open and closed loop stimulation paradigms to coax the network from unhealthy (depressed) to healthier (euthymic) state.