Abstract
Color charts can capture some aspects of the phenomenal structure of color experience in a coordinate system with two bipolar axes, one for redness (positive) vs. greenness (negative) and the other for yellowness vs blueness. This leads to an arrangement in which perceptually unitary (‘unique’) red and green are found in opposite directions from white at the origin, and blue and yellow are similarly opposite, while the blue/yellow and red/green directions are mutually orthogonal. The neural representation of color by color-opponent signals is often viewed as supportive of such schemes. But the pervasive rectifying nonlinearity of neural responses, and the marked asymmetry between excitatory and inhibitor response dictated by the relatively low spontaneous firing rate, are more suggestive of an idealization with multiple monopolar signals for redness, greenness, yellowness and blueness. Wuth mutually orthogonal monopolar coordinates for the four primary signals, the isoluminant colors occupy the surface of a hypercube in the 4D space; white is at one corner. To test this scheme experimentally, subjects first identified examples of unique red, green, yellow, and blue perceived as equidistant from white. We then asked whether these colors are perceptually equidistant from each other, as the hypercube model predicts. One of the set of colors judged equidistant from white, for instance the red, was presented with adjustable purity, and subjects adjusted the purity as needed to make it as different from the green as from the blue or the yellow. On the hypercube model, no adjustment should be needed. But if there were perceptual red/green opponency, one would expect subjects to select a less saturated red than the one chosen in the first phase of the experiment. Results are close to the predictions of the hypercube surface model, with significant but small deviations.