Information-theoretic investigation of multi-unit activity properties under different stimulus conditions in mouse primary visual cortex
File(s)
Author(s)
Tolkiehn, Marie
Type
Thesis
Abstract
Primary visual cortex (V1) is the first cortical processing level receiving topographically mapped inputs from the retina, relayed through thalamus. Electrophysiological studies discovered its important role in early sensory processing particularly in edge detection in single cells. To this end, little is investigated how these activities relate on a population level. Orientation tuning in mouse V1 has long been reported as salt-and pepper organised, lacking apparent structure as was found in e.g. cat or primates.
This is a novel synthesis of specially designed in-vivo electrophysiological experiments aiming to make certain information-theoretic data analysis approaches viable. Sophisticated state-of-the-art data analysis techniques are applied to answer questions about stimulus information in mouse V1. Multi-unit electrophysiological experiments were devised, performed and evaluated in the anaesthetised and in left hemisphere V1 of the awake behaving, head-fixed mouse. A detailed laboratory and computational analysis is presented validating the use of Multi-Unit-Activity (MUA) and information-theoretic measures. Our results indicate left forward drifting gratings (moving from the temporal to nasal visual field) elicit consistently highest neuronal responses across cortical layers and columns, challenging the common understanding of random organisation. These directional biasses of MUA were also observable on the population level.
In addition to individual multi-unit analyses, population responses in terms of binary word distributions appear more similar between spontaneous activity and responses to natural movies than either/both to moving gratings, suggesting that mouse V1 processes natural scenes differently from sinusoidal drifting gratings. Response pattern distributions for different gratings emerge to be spatially but not orientationally clustered. Further computational analysis suggests population firing rates can partially account for these differences. Electrophysiological experiments in the awake behaving mouse indicate V1 to contain information about behavioural outcome in a GO/NOGO task. This, along with other statistical measures is examined with statistical models such as the population tracking model, which suggest that population interactions are required to explain these observations.
This is a novel synthesis of specially designed in-vivo electrophysiological experiments aiming to make certain information-theoretic data analysis approaches viable. Sophisticated state-of-the-art data analysis techniques are applied to answer questions about stimulus information in mouse V1. Multi-unit electrophysiological experiments were devised, performed and evaluated in the anaesthetised and in left hemisphere V1 of the awake behaving, head-fixed mouse. A detailed laboratory and computational analysis is presented validating the use of Multi-Unit-Activity (MUA) and information-theoretic measures. Our results indicate left forward drifting gratings (moving from the temporal to nasal visual field) elicit consistently highest neuronal responses across cortical layers and columns, challenging the common understanding of random organisation. These directional biasses of MUA were also observable on the population level.
In addition to individual multi-unit analyses, population responses in terms of binary word distributions appear more similar between spontaneous activity and responses to natural movies than either/both to moving gratings, suggesting that mouse V1 processes natural scenes differently from sinusoidal drifting gratings. Response pattern distributions for different gratings emerge to be spatially but not orientationally clustered. Further computational analysis suggests population firing rates can partially account for these differences. Electrophysiological experiments in the awake behaving mouse indicate V1 to contain information about behavioural outcome in a GO/NOGO task. This, along with other statistical measures is examined with statistical models such as the population tracking model, which suggest that population interactions are required to explain these observations.
Version
Open Access
Date Issued
2017-09
Date Awarded
2018-02
Advisor
Schultz, Simon
Clopath, Claudia
Sponsor
Imperial College London
Publisher Department
Bioengineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)