Abstract

Three young observers judged the differences they perceived between all pairs of twenty-five colors in twelve different oblique planes formed by the OSA uniform color scales. Each observer repeated each judgment at least three times. The rms variances of the judgments ranged from 5 to 15%. For each set of colors, a 3-D configuration of points was determined for each observer, in which distances between pairs of points are, as nearly as possible, proportional to the judgments raised to a power P. The optimal value of P differed from observer to observer and from set to set of colors. For two of the observers, P was always >1.0 (as great as 1.61). For the third observer, P was 1.29 or less (as low as 0.76). The resulting rms errors of fit in the configurations were 13.8 ± 4.8%. The most successful configurations indicate (with some remarkable agreements between observers) strong curvatures of the surfaces on which the optimal configurations of the sets of twenty-five points are located. These results indicate that non-Euclidean space is needed to represent perceived color differences by geometric distances within the attainable precision of judgments.

© 1987 Optical Society of America