Shoutik Mukherjee

Research Summary

My research interests lie at the intersection of signal processing, neural modeling, and machine learning, with the aim of developing methodologies for studying how the sensory stimuli and internal neuronal dynamics map to decision-making and behavior in high-dimensional neuronal recordings.

Previous Research

My graduate research addressed three topics: statistical signal processing algorithms for scalable population-level analyses; characterization of functional network interactions in large neural assemblies; and hierarchical transformation of stimulus representations along the auditory pathway.

Population-level Analyses of Neuronal Ensemble Activity

Multi-electrode arrays (MEA) and calcium imaging enable simultaneous recording from increasingly large neuronal ensembles. Population-level analyses of the resulting high-dimensional neuronal data can provide novel insights into the underlying functional networks that propagate and encode information. To this end, my research involved developing and applying signal processing tools for dynamic synchrony and Granger Causality network analyses of neuronal populations, working on both electrophysiological and calcium imaging data.

Computational Models of Acoustic Signal Representation in the Mammalian Auditory Cortex

The brain is a highly interconnected dynamical system. Behavioral tasks have been shown to induce rapid changes to acoustic signal representation in the auditory cortex of animal models, and more complex behaviors likely require coordination of several cortical areas. Using signal processing tools tailored for high-dimensional statistics, my research also addressed the representation of salient features of acoustic signals throughout the cortex.