Abstract
It has long been known that a particular red, green, blue and yellow is seen as being entirely free of any other color, and that the four color categories defined in this way are arranged in a circular manner. Most textbook accounts suppose that these features of color experience are an incidental consequence of color opponency. An alternative possibility is that these aspects of color experience represent the solution of a fundamental problem in topology, namely insuring that no two areas separated by a common boundary in a 2-dimensional array will appear the same if they are actually different (the four color map problem). However, unlike the cartographer, whose task is simply to distinguish the surfaces in a map as being the ‘same’ or ‘different’, the visual system must distinguish surfaces and at the same time maintain the full range of spectral relationships. Simply differentiating surfaces in color experience would provide little behavioral advantage if the relative similarities and differences among different spectra were not also preserved in perception. If this argument is correct, then structure of subjective color space (i.e., the circular organization of the four color categories and their unique members), should reflect an analogous ordering of spectra. Here we use multidimensional scaling of a spectral data set to show that arranging spectra according to their relative similarities and differences defines a space that is similar to subjective color space. These results are consistent with the conclusion that the major features of subjective color experience represent a simultaneous solution of the four color map problem while maintaining the relative similarities and differences among the full range of light spectra.