Abstract
Lightness illusions are part of the larger category of errors in perceived surface lightness. Such an error can be defined as a discrepancy between the physical reflectance of a surface and the reflectance of the Munsell chip chosen as a perceptual match. These errors are not random, but fall into a particular pattern. This pattern must be the signature of the visual software used by the brain. The general pattern is this: lightness illusions are composed of frameworks, which are driven by gestalt grouping principles. Within each framework, the value of white is assigned to the element with the highest luminance. Darker shades are assigned based on their relationship to the highest luminance, according to the Wallach ratio principle. There is crosstalk between frameworks, however, in the form of the Kardos principle of co-determination. Thus, for a given surface, lightness is a weighted average of its value relative to the highest luminance within the local framework and its value relative to the highest luminance in the global field. I will report a series of experiments in which this account of lightness illusions is pitted against a more conventional spatial filtering account.
National Science Foundation (BCS-0236701) National Institute of Health (BM 60826-02)