Dissertation/Thesis Abstract

A model-based approach to the neural coding problem
by Nichols, Zachary, Ph.D., Weill Medical College of Cornell University, 2014, 148; 3578491
Abstract (Summary)

The early visual system is a widely-studied model of parallel processing and neural coding in the brain. In this system, neural populations composed from several distinct cell types perform computations and carry relevant sensory information into other brain areas. While the early visual system, including the retina, shares basic organizational features with other sensory systems, it has several advantages that make it particularly well-suited to studying questions related to population coding. The first of these is the ability to capture the output of ganglion cells in experiments, which form a complete representation of the visual world due to the bottleneck of the optic nerve. The second is the development of models which accurately describe the input/output relationships between stimuli and the cells' spiking output. Here, using the modeling and experimental advantages afforded in this system, we investigate several open questions related to parallel processing: the role of correlated activity, and the role of canonical cell class divisions in representing stimulus information. We find that the information carried by neural populations in the retina is largely done so by the independent responses, and we identify novel asymmetries in the information carried by the responses cells on either side of a canonical cell division – ON vs. OFF – that is explained by an interaction between their linear and nonlinear response components.

Indexing (document details)
Advisor: Nirenberg, Sheila
Commitee:
School: Weill Medical College of Cornell University
School Location: United States -- New York
Source: DAI-B 75/05(E), Dissertation Abstracts International
Source Type: DISSERTATION
Subjects: Neurosciences
Keywords: Neural correlations, Parallel processing, Population coding, Retina
Publication Number: 3578491
ISBN: 9781303722806
Copyright © 2019 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy
ProQuest