Abstract
In past work, we have presented and elaborated a model of achromatic color induction in which lightness and darkness induction signals spread spatially from edges, decay with distance, and are summed to produce a total achromatic color signal. Our previous data indicate that multiplicative interactions between edge contrasts also play a role in achromatic color. To account for these edge interaction effects, Rudd and Arrington (2001) proposed that induction signals associated with remote edges are partially “blocked” by nearby edges, by an amount that depends on the nearby edge contrast. Rudd and Popa (VSS 2004, submitted) showed that this blockage effect can also operate in the opposite direction: that is, remote edge contrast can attenuate the effect of local contrast on target lightness. I will show how both of these effects can be produced by a simple neural mechanism by which cortical edge detector neurons interact spatially to control one another's gains prior to color induction and spatial integration. The theory makes the non-intuitive prediction that contrast and assimilation effects can be seen in the same simple disk-and-ring display over different ring luminance ranges, a prediction that I will support with data (Rudd & Zemach, submitted). The theory thus gives a unitary account of two of the most basic spatial phenomena of color vision, phenomena which have historically been thought to result from separate underlying mechanisms (Jameson & Hurvich, 1961), and suggests a mechanism by which contrast and assimilation effects may arise in more complex displays (Bindman & Chubb, 2004).