Abstract
Introduction. It is well known that color can be induced spatially as the completion of a color boundary (Pinna, 1987) and temporally as an afterimage (especially within a boundary contour; Daw, 1962). Also, figures with sparse color contours can induce local extensions of the contour boundary, as in Neon Color Spreading (Varin, 1971). These observations have invoked the concept of color induction as driven largely by color contours. Consideration that most chromatic cells in primary visual cortex are not contour selective, however, suggests that contourless color induction should be possible. Methods. Fields of sparse random dots of one color on the left side of a black field and a contrasting color on the other side are accurately fixated for 10 s, and assessed for color induction. The dots are then removed and the empty black field assessed for color induction. Results. The colored dots induce color spreading across each side of the field. Rather than being the opposite color, as predicted by the local color border hypothesis, the field color is a darker version of the color of the dots overlying the field. This induced color represents a novel form of color induction from the color of the dots alone. Removal of the dots switches the induced field color to the opponent color. The dots also have afterimages of the opponent color that may appear and disappear with microsaccadic eye movements, while the induced opponent field color remains and fades gradually over about 5s, as though it were an afterimage of the initial induced field color. The two different induced field colors form a color border much sharper than between the sets of sparse random dots. Discussion. Induced colors can form borders in the absence of contour border in the input image, inverting the standard logic of color border processing.
Meeting abstract presented at VSS 2015