Visual structures for achromatic color perceptions.

Accuracy of achromatic perceptions (correctly seeing blacks, grays, and whites) was investigated for three-dimensional rooms, painted matte black or white, presented empty or containing matte or specular, and black, middle gray, or white objects. The rooms were illuminated by a point source at various locations or by extended sources. Observers judged the achromatic color of four walls in the room, and in one experiment the objects in the room, with wide-angle binocular viewing, or judged the far wall under reduction conditions. At issue were questions about absolute and relative achromatic (lightness) constancy, specification of the proximal stimulus for achromatic perceptions, and the relevance of Gilchrist's, Land's, Wallach's, and Helson's theoretical concepts. Despite the three-dimensional spatial arrangements and many variations in experimental conditions, achromatic perceptions were tied to ratios of luminances (contrast ratios or CRs) among surfaces. Moreover, the larger the CR's magnitude, the greater the apparent achromatic difference between surfaces. These achromatic distributions for the perceived colors of the walls were approximately centered on middle gray, when intrasurface luminances were heterogeneous, and on white, when intrasurface luminances were very homogeneous. Despite wide-angle binocular viewing, absolute achromatic constancy occurred only if the CRs were caused by reflectance differences between matte surfaces or within a specular surface. (The stimulus for apparent white could sometimes be an exception to this.) Constancy failed when CRs were caused by illumination gradients on surfaces, even though the scene was three-dimensional and complexly structured. Achromatic constancy depended, therefore, on CRs and on the probability of occurrence of specific physical events. A theory of achromatic perception is described, based on CRs, Helsonian averages, a simple proposition about the illumination, and a concept of nonspatial achromatic segregation.