Abstract
Attempts to develop a universal color preference model have thus far failed to explain individual differences or incorporate physiological factors. In our recent experiments investigating the biological and cultural determinants of color preference (e.g. Ling et al., ECVP 2004), we have shown that the S cone contrast value accounts for the greatest variance in hue preference, with the L-M cone contrast value second. We have therefore proposed a model for hue preference in which each observer's hue preference curve may be predicted by a weighted sum of the S and L-M cone-contrast components of the stimuli.
Here, we extend the hue preference model into the lightness domain. We hypothesize that for color stimuli varying in both hue and lightness, the individual preference curves may be predicted by the weighted sum of three components: the achromatic lightness channel (L+M+S contrast), and the 2 cone-contrast channels above. To test this hypothesis, we carry out a new set of experiments in which observers perform a paired comparison task for color stimuli systematically varied in hue and lightness. The results confirm the hypothesis: for each subject, the color preference curve is better predicted by the weighted sum of 3 channels than the weighted sum of 2 cone-opponent channels. Moreover, a positive weight for the lightness component indicates that the observer prefers brighter colors, and vice versa. By reducing the description of each individual's color preferences to a set of 3 weights, this model readily accounts for individual differences and is extensible across populations.