Skip navigation
Please use this identifier to cite or link to this item: http://arks.princeton.edu/ark:/88435/dsp0170795b25d
 Title: Adaptation in Coding by Large Populations of Neurons in the Retina Authors: Ioffe, Mark Lev Advisors: Berry, Michael JBialek, William Contributors: Physics Department Keywords: AdaptationClusteringMaximum EntropyNoise CorrelationsPopulation CodingRetina Subjects: NeurosciencesPhysics Issue Date: 2017 Publisher: Princeton, NJ : Princeton University Abstract: A comprehensive theory of neural computation requires an understanding of the statistical properties of the neural population code. The focus of this work is the experimental study and theoretical analysis of the statistical properties of neural activity in the tiger salamander retina. This is an accessible yet complex system, for which we control the visual input and record from a substantial portion - greater than a half - of the ganglion cell population generating the spiking output. Our experiments probe adaptation of the retina to visual statistics: a central feature of sensory systems which have to adjust their limited dynamic range to a far larger space of possible inputs. In Chapter 1 we place our work in context with a brief overview of the relevant background. In Chapter 2 we describe the experimental methodology of recording from 100+ ganglion cells in the tiger salamander retina. In Chapter 3 we first present the measurements of adaptation of individual cells to changes in stimulation statistics and then investigate whether pairwise correlations in fluctuations of ganglion cell activity change across different stimulation conditions. We then transition to a study of the population-level probability distribution of the retinal response captured with maximum-entropy models. Convergence of the model inference is presented in Chapter 4. In Chapter 5 we first test the empirical presence of a phase transition in such models fitting the retinal response to different experimental conditions, and then proceed to develop other characterizations which are sensitive to complexity in the interaction matrix. This includes an analysis of the dynamics of sampling at finite temperature, which demonstrates a range of subtle attractor-like properties in the energy landscape. These are largely conserved when ambient illumination is varied 1000-fold, a result not necessarily apparent from the measured low-order statistics of the distribution. Our results form a consistent picture which is discussed at the end of Chapter 5. We conclude with a few future directions related to this thesis. URI: http://arks.princeton.edu/ark:/88435/dsp0170795b25d Alternate format: The Mudd Manuscript Library retains one bound copy of each dissertation. Search for these copies in the library's main catalog: catalog.princeton.edu Type of Material: Academic dissertations (Ph.D.) Language: en Appears in Collections: Physics

Files in This Item:
File Description SizeFormat
Ioffe_princeton_0181D_12084.pdf24.49 MBAdobe PDF
Ioffe_princeton_0181D_408/recordings.zip193.41 MBUnknown

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.