The perception of color has traditionally been characterized by the subjective dimensions of hue, brightness, and saturation. In the present study we reexamined this view by investigating whether the dimensions of color stimuli are psychologically independent in dissimilarity judgment, spontaneous classification, and instructed classification tasks. Dissimilarity judgments analyzed within the framework of the additive difference measurement model (Beals, Krantz, & Tversky, 1968;Krantz & Tversky, 1975;Tversky & Krantz, 1969 reflected violations of psychological independence for hue-chroma, hue-value, and value-chroma stimulus sets. Spontaneous classifications of each color set revealed that subjects were not sensitive to shared dimensional relations of color stimuli, but rather responded to the holistic, overall similarity relations of the stimuli. In the instructed classification task, both untutored undergraduates and "color experts" (artists specially tutored in the Munsell color system), instructed to classify according to shared dimensional relations, could extract dimensional information about either value or chroma when each was varied with hue, but could not extract dimensional information about hue. Color experts were superior to nonexperts in the extraction of dimensional information about chroma only with moderately or highly saturated stimuli. The implications of these results are considered in relation to current thinking about the perceptual organization of color and current thinking about the identification of appropriate diagnostics for independent psychological dimensions.The precise notion of a psychological dimension continues to elude psychologists interested in the perception and organization of multidimensional stimuli. Defining a psychological dimension has been particularly problematic for investigators interested in color. A common assumption has been that the perception of color is organized in terms of the dimensions hue, brightness, and saturation.1 "There is no doubt, in the case of color differences, as to the number of independent perceptual attributes .... The number is three" (Krantz, 1972, p. 683). This assumption is based on scaling procedures (i.e., confusability scaling, direct estimation, multidimensional scaling, etc.) that have supported the idea that hue, brightness, and saturation are the dimensions of color space (Helm, 1964; Indow & Kanazawa, 1960; Indow & Uchizono, 1960;Newhall, 1939;Newhall, Nickerson, & Judd, 1943;Wright, 1965). One difficulty with this assumption that hue, brightness, and saturationare the three dimensionsorganizing the perception of color is related to the multiple uses of the term dimension. As Tversky and Krantz (1970) and Gati and Tversky (1982) indicated, the term has been given several meanings in the literature. Psychologists have used the term dimension simply to refer to a variable that can be physically manipulated. The term has also been used to mean an organizing principle that structures and orders perception in a consistent way. A t...