Neuronal Basis of V1 Binocular Normalization
Mitchell, Blake Austin
0000-0001-7321-2854
:
2024-05-13
Abstract
This dissertation investigates the spatiotemporal dynamics of binocular integration in the primate primary visual cortex (V1) through a combined approach of laminar neurophysiology and computational modeling. The first study examined how populations of recorded V1 neurons represent a two-fold increase in visual input (i.e., binocular stimulation) across varying levels of visual contrast. We observed that V1 populations transiently increased their firing rates in proportion to contrast, followed by more complex, contrast-dependent interactions. Notably, the binocular response gain at the neuronal level was comparable to performance gains in psychophysical studies of binocular summation. In the second study, we assessed the role of ocular dominance (OD)—each neuron’s response preference for one eye over the other—in V1 binocular integration. Our findings indicate that OD influenced binocular combination of visual contrast by weighting the effectiveness of contrast placed in the neuron's non-dominant eye. We subsequently show that incorporating ocular dominance information into computational models enhances their predictive accuracy. The third and final study evaluated V1 binocular integration within a computational framework of divisive normalization tailored for binocular inputs—termed binocular normalization—across the V1 laminar microcircuit. Results suggest that the observed pattern of sublinear binocular summation can be explained by a single, dynamical process of divisive normalization operating both within and between the eyes. This process was weakest in the geniculate input layers and intensified as signals propagated through the microcircuit. Collectively, these studies deepen our understanding of the neural mechanisms that underpin binocular single vision and, more generally, help establish a critical link between a fundamental neural computation and canonical circuitry.