eMB: New Approaches for Interpreting Neural Responses to Behaviorally-Relevant Sensory Stimuli

Neural information is encoded in spike trains that describe the timing of electrical impulses in neurons. In neural sensory systems, information about the presence and type of stimulus is encoded this way, both at the level of individual neurons and at the level of neural ensembles. Neural spike trains have been studied for many years using various mathematical and statistical techniques. However, there has been a recent explosion of new techniques developed for the analysis of data sets and networks that have potential application to the analysis of neural spike trains. This project will develop new methodology for analyzing neural spike trains using recently developed mathematical techniques. The techniques will be developed for the interpretation of spike trains recorded from the part of the brain responsible for coding information about taste. However, the methodology can be broadly applied to spike trains from other neural sensory systems. The team members of this highly interdisciplinary project include senior investigators with expertise in chemosensory neural processing, computational neuroscience, and data science. They will work together with undergraduate and graduate students on all aspects of the project. Senior team members also participate in workshops such as computer coding boot camps that share their expertise at the interface of mathematics and biology with students and others interested in learning new approaches to data analysis. Extracting information about input stimuli from a neuron's spike train is extremely challenging. This project has two aims that describe approaches for doing this using new mathematical techniques. The focus of Aim 1 is at the single-neuron level, with the goal of determining which neurons code information about the stimulant, and how they code that information. It employs Bayesian analysis, in combination with techniques developed recently from optimal transport and topological analysis of burst dendrograms. The focus of Aim 2 is at the neural ensemble level, taking advantage of experimental technology that allows for the simultaneous recording from tens of neurons in the behaving animal. An analysis of this network-level data will build on techniques used in the first aim and will also use network science techniques such as community detection along with topological data analysis to characterize how stimuli are coded in neural ensembles. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.