A research-focused tour of the state of the art in Neurophysics

James Johnson, Department of Physics, Washington University
February 23, 2018 at 4:00 pm
204 Crow
Event Description 

Neurophysics is the application of abstract quantitative modeling and insights from the fundamental behavior of matter and energy to neuroscience at large. It deals with atomic and molecular features, as well as collective phenomena recognizable to a condensed matter physicist and the nonlinear dynamics and chaos associated with "complexity science". This talk will be an introduction to neurophysics anchored around three projects in the Wessel lab. The first uncovers how the fact that neurons act as coarse-graining and rescaling operators may directly imply that brain networks are critical branching networks. Building on this interpretation of neuronal action, the second project shows how population-level information can be gleaned from recording just one single neuron for only a few milliseconds. Lastly, we tackle the problem of degenerate states in localized neuronal activity. Functions are degenerate. The same motion of a hand has many different possible representations in neuronal activity. At present neuroscientists use the precise firing patterns of neurons as a description of state. By using algorithms to detect over-lapping communities we can describe the state of a local group of neurons (~200) based on which communities are most active. This description is degenerate with respect to precise firing patterns, just like functions are. Most importantly, an unsupervised learning algorithm can predict behavior better and more efficiently with community activity patterns than with neuronal activity patterns.