BME PhD Prospectus Defense: Isaac Boyd

Title: "Modeling Cortical Circuits for Solving the Cocktail Party Problem"

Advisory Committee: Dr. Kamal Sen, PhD - BME (advisor) Dr. Brian DePasquale, PhD - BME (chair) Dr. Kayhan Batmanghelich, PhD - CDS Dr. Howard Gritton, PhD - BIOE, UIUC

Abstract: Listening to a target sound source in a scene of competing noises is a task that individuals with healthy hearing can do well. However, populations with autism and ADHD as well as aging and hearing-impaired populations struggle with this task. Disentangling the reason behind this is a major challenge. There exists diversity amongst the neural population and individual neural responses in the auditory cortex making it difficult to infer cortical circuit function from neural recordings alone. This project will develop a data-constrained computational model to identify how excitatory and inhibitory neurons shape sound-evoked activity across increasing complex listening conditions. First, I will model single-cell responses using a spiking network to establish a validated model framework for when a single target is present without any competition. The model will be fit using a gradient-based, eligibility propagation-like approach. Using this, I will investigate the role of parvalbumin (PV) neurons in timing and selectivity in neural responses. Next, I will extend this model to scenes with competing stimuli and characterize how surround inhibition through somatostatin (SOM) neurons is implicated in target-masker discrimination. Finally, I will incorporate top-down modulatory dynamics into the network and will identify how VIP-mediated disinhibition shapes attention.