Abstract
It has been proposed that color preference is largely governed by the two cone-opponent processes that underlie sensory encoding of color, and that robust sex differences in the weighting of these ‘biological components’ of color preference evolved in line with sex differences in the behavioural use of color vision (Hurlbert & Ling, 2007). In support of this theory, Hurlbert and Ling found that cone-contrast between stimulus and background (L-M, S-(L + M)) explained 70% of the variance in hue preference, with a sex difference in the weighting of L-M cone-contrast for both British and Chinese samples. Here, we further investigate whether cone-contrast effectively summarizes color preference, and whether the sex difference in L-M is indeed robust and ‘universal’. In a series of three experiments, we measured color preference using three different stimulus sets, and tested British, Saudi, Archis (a rural ethnic group in southern Daghestan) and Himba (a semi-nomadic ethnic group in northern Namibia) samples. Regression analyses were run on preference ratings to establish the percentage of variance explained by L-M and S-(L + M) stimulus-background cone-contrast (and CIE-LUV lightness, chroma and saturation), and to identify the weights on these different components. Results indicate that stimulus-background cone-contrast is less effective at summarising color preference the wider the gamut of the stimulus set (see also Palmer & Schloss, 2010). The sex difference in the weighting of L-M was present for the Saudi sample (females weight L-M positively, males negatively), but was not present for the Archis or Himba, and was not reliable across the three experiments for the British. In addition, Himba preference was predominantly explained by chroma (higher chroma, higher preference: 74% of the variance), rather than stimulus background cone-contrast. Overall, the findings challenge the theory that color preference is heavily and universally constrained by the biological components of color vision.