UC Irvine

This thesis delves into the mechanisms of visual processing, focusing on salience maps and luminance. The first two chapters examine salience maps. The final chapter explores parvo isoluminance in the central fovea which became an important component in the attempt to measure foveal salience maps. A salience map, proposed by Koch and Ullman in 1985, is a 2D topographic map that combines inputs from various feature maps to represent the combined salience at each x,y location as a real number. A salience map is mostly used to predict processing priority in the literature. Chapter 1 demonstrates that tasks such as computing distances in the frontal plane, calculating centroids (the center of a group of items), determining the numerosity of a collection of items, and identifying alphabetic letters are substance-indifferent and thus utilize salience maps. Feature-salience alone, without relying on luminance information, is sufficient for these visual judgments. Chapter 2 investigates the number of salience maps an observer has by assessing their performance in concurrent centroid judgments. In 11 experiments, subjects viewed flashes of item arrays with 28 to 32 items, each with M different features (M = 3 to 8). Subjects then mouse-clicked the centroid of post-cued feature items. Ideal detector response analyses indicated that subjects utilized at least 12-17 stimulus items. Performance in (M-1)-feature experiments predicting M-feature experiments led to the conclusion that one subject had at least 7 salience maps, while two others had at least 5 each. A computational model identified the primary performance limitations as channel capacity for representing multiple groups and working-memory capacity for maintaining computed centroids. Luminance describes the visual effectiveness of a light and is typically considered the primary source for spatial visual information. Isoluminance, where different wavelength compositions have equal luminance, offers a way to evaluate the importance of luminance by studying alternative visual processing capabilities when luminance information is unavailable. Whereas isoluminance is almost universally determined under conditions that measure pure magno isoluminance, Chapter 3 used fine-gratings in the central fovea to measure parvo isoluminance. Subjects viewed high spatial-frequency yellow/red and yellow/green gratings, judging their orientation while varying red and green intensity. Results showed higher sensitivity to color than luminance in the central fovea, making foveal parvo isoluminance unmeasurable as color sensitivity overshadowed luminance sensitivity.