Neural Mechanisms of Luminance Perception

Abstract

Visual animals evolved to efficiently encode luminance increments and decrements with ON and OFF visual pathways. Recent work from our lab indicates that these ON and OFF pathways segregate in primary visual cortex, function relatively independently from each other and process spatial-temporal contrast differently. In my thesis, I investigate the role of ON and OFF cortical pathways in processing luminance contrast and I apply my research findings to image processing and the eye clinic. In the first chapter, I record neural activity from cat primary visual cortex with multielectrode arrays and human visual cortex with electroencephalography. I use these approaches to measure visually evoked cortical responses from ON and OFF pathways to stimuli with different contrast polarity (light or dark) and luminance range (maximum-minimum or standard deviation of luminance distribution). I demonstrate that ON and OFF pathways have different contrast response functions and the differences increase with luminance range. I also demonstrate that these ON-OFF differences are needed to efficiently sample distributions of light and dark contrast in natural scenes. I use my findings to develop a new algorithm of ON-OFF image processing to reproduce more accurately luminance contrast in image photography. In the second chapter, I develop a new test of ON-OFF perimetry in a head-mounted visual display to measure human visual performance for detecting light and dark stimuli at different contrasts and eccentricities. My measurements demonstrate that the relative dominance of ON and OFF pathways is strongly dependent on contrast. At low contrasts, humans are more accurate and faster at detecting light stimuli (ON pathway dominance) but, as contrast increases, they become more accurate and faster at detecting dark stimuli (OFF pathway dominance). I show that multiple light/dark ratios of visual dominance based on performance and reaction time are strongly correlated with stimulus contrast and eccentricity. My results provide new insights on the neuronal mechanisms underlying the perception of luminance contrast and may help to advance computing strategies for image processing and tools to evaluate visual function in the eye clinic.