Psychophysics of Color Vision

All visual information available to us is the result of combining signals from photoreceptors in the retina differentially sensitive to three wavelengths. Such information processing occurs at different stages, and current models have varied levels of success explaining sometimes contradicting psychophysical data for each stage (e.g. in the case of the existence of ``half-axis'' mechanisms); notably, in the case of color perception, as we move from the retina to higher order processing, the uncertainty of what mechanisms account for a large variety of results increases. In this dissertation, I use sophisticated psychophysical methods, combined with mathematical models of perception, to investigate long-held assumptions about performance-based luminance. Specifically, I test the plausibility of the assumption of co-planarity for colored lights, which its luminance is made equal to some given achromatic light using a task based on first order motion. I use the motion-equiluminant lights so obtained to investigate the mechanisms used by the human visual system to make judgments about motion-equiluminant colored textures.In chapter one I describe a new method, based on minimum motion, to obtain motion-equiluminant lights, and I test the planarity of the motion equiluminant surface. This technique reveals properties of the motion-equiluminant plane that were previously undetected using other methods. In particular, I find that the motion-equiluminant lights deviate from planarity, and do so following a pattern. In chapter two I compare this new method to one used to derive photopic luminance, Heterochromatic Flicker Photometry (HFP). Using HFP I find that deviations of equiluminant lights from planarity are small, suggesting that the assumption of planarity is plausible. Yet, I argue that the experimental data from both techniques shows that the reason for this discrepancy is that minimum motion yields finer-grained measurements compared to the measurements obtained using HFP.Finally, in chapter three I model higher order visual mechanisms sensitive to color, rooted on the differential activation of each of the three cone classes. Using the minimum motion method presented in chapter one to acquire individualized motion-equiluminant color palettes, I apply the seed expansion method to create colored white noise textures. I find that performance in a detection task can be explained by two pairs of complementary full axis mechanisms: one with sensitivity that increases linearly across the gamut from red to green and one that increases linearly across the gamut from green to red, forming one complementary pair, and one with sensitivity that increases evenly with increased red-green saturation, and one that increases evenly with decreased red-green saturation, to form the other complementary pair. However, such explanation does not strongly preclude the existence of half-axis mechanisms