Abstract
It has been pointed out that dichromats can categorize colors using color names in fair agreement with normal trichromats. Some possible mechanisms, such as rod contribution, an anomalous cone pigment or a parallel nonlinear channel, have been proposed to account for this ability. In this study we aimed to clarify whether two dimensions of cone responses (L and S or M and S) are sufficient for showing their categorical color regions without overlaps in a color space. We carried out categorical color naming experiments using protanopes and deuteranopes as observers as well as normal trichromats as references. They named 424 OSA uniform color samples twice with the B&K eleven basic color terms under a 6500K illuminant and other illuminants. We plotted the stimuli divided into different color categories in the M+S vs. M−S cone space or in the L+S vs. L−S cone space for protanopes or deuteranopes, respectively. The results show that, for a dichromat who showed trichromat-like categorical color regions, most categorical color regions were slightly shifted in the two dimension color space so that they could be named different categories. This suggests that dichromats might use barely visible differences in chromatic and/or lightness appearance to categorize colors.