The Oriented Difference-of-Gaussians Model of Brightness Perception

Abstract

The Oriented Difference of Gaussians (ODOG) model (6) was developed to gauge the degree to which “early” visual processes such as spatial filtering and response normalization could account for human brightness percepts in a set of canonical stimuli including the White effect (44-46), classical simultaneous brightness contrast (SBC) (25), and grating induction (GI) (3, 5, 20, 34, 37, 48). The ODOG model successfully predicts changes in the magnitude of the White effect (9) and GI (11) as a function of inducing grating spatial frequency and test patch height, as well as the relative magnitude of brightness variations in many other stimuli including the Hermann Grid (8), the Gelb Staircase (16,17), the Wertheimer-Benary Cross (4, 7, 8), Howe's variations on White’s stimulus (15, 28), Todorovic’s (43) and Williams, McCoy, & Purves’ (47) variations on the SBC stimulus (6, 12), the checkerboard induction stimulus (9, 19), the shifted White stimulus (9, 45), Adelson’s Checker-Shadow (1, 12), Snake stimulus (2, 8, 12, 41), and Corrugated Mondrian stimuli (1, 8), including Todorovic’s variation (7, 8, 43), Hillis & Brainard’s Paint/Shadow stimulus (12, 27), “remote” induction stimuli (8, 10, 32, 40), and in the probe discs placed in Cartier-Bresson photographs (12, 22).