SPATIAL COLOR CHANNELS AND THEIR ROLE IN COLOR CONSTANCY

Color always occupies a region of space, typically attached to an object, a light, or a surface; however, most of our models of color vision and approaches to understanding color perception treat color as an abstract quality void of a spatial dimension. In this dissertation, I explore the utility of considering separate types of color systems based on spatial information in the environment. In Chapter 1, I review of the study of color perception as it relates to space. Chapter 2 explores the existence of a luminance versus luminance-contrast marker in the cortex using the contrast asynchrony stimulus. Observers were shown disks modulating from bright to dark at low frequencies on either dark, midgray, and bright backgrounds while steady-state visual evoked potentials (SSVEPs) were recorded. A luminance-driven signal would elicit equivalent SSVEPs across all backgrounds, but a luminance-contrast signal would differ depending on the background. I found that SSVEP amplitudes and phase relations were indicative of a contrast-driven signal. The results suggest that early cortical signals, at least from V1, are driven by contrast, rather than luminance, levels. In Chapter 3, I investigate the efficacy of color separation based on spatial scale as it relates to the phenomenon of color constancy. In a series of three studies (the first on a plain background, the second on a Mondrian, and the third on a painting), I investigate the hypothesis that object color can be represented by high spatial frequency (HSF) content while illuminant color can be represented by low spatial frequency (LSF) content. Observers’ object matches in Experiment 1 exhibited less variation across stimuli with similar than different HSF content. In Experiment 2, object matches moved in relation to HSF content and illumination matches moved in relation to LSF content. In Experiment 3, observers perceived greater change in object color across images with different HSF content than different LSF content; however, their perception of illuminant color change was no different than chance across HSF-differing and LSF-different stimuli. Altogether, this work demonstrates that color behaves differently and serves different roles at different spatial scales. Color constancy and other phenomena may be better understood from this point of view.