Grouping illumination frameworks

According to Koffka (1935), the lightness of a target surface is determined by the relationship between the target and the illumination frame of reference to which it belongs. However, each scene contains numerous illumination frames, and judging each one separately would lead to an enormous amount of computing. Grouping those frames that are in the same illumination would simplify the computation. We report a series of experiments demonstrating that nonadjacent regions of the visual field under the same illumination level are perceptually grouped together and function, to some extent, as a single framework. A small coplanar group of patches under its own illumination exhibits compression of perceived range of gray shades. We obtained the reduction in compression in the presence of an identically illuminated 25-patch Mondrian tableau mounted nearby the coplanar group. The influence of the Mondrian display was reduced when it was (a) moved laterally away from the test display, (b) moved farther back in depth from the test display, or (c) rotated to a different orientation.