Perceptual distance between colored stimuli in the lizard Anolis sagrei: comparing visual system models to empirical results

Abstract

In the study of the evolution of animal colors, the conspicuousness of a pair of colors (e.g., stimulus vs. background) is often modeled by determining the distance between them in perceptual color space. For similar colors, receptor noise models have been demonstrated to be an effective way to estimate discrimination thresholds. However, the best way to quantify conspicuousness of color pairs that are distant in perceptual space is not widely agreed upon. In experiments where an animal is presented with different colors that it can easily discriminate, its response may be strongly influenced by the biological significance of a color, in addition to its conspicuousness, a phenomenon known as color categorization. This has made it difficult to test and confirm the validity of different models of perceptual distance. We tested the relative conspicuousness of different stimulus/background color combinations in the lizard Anolis sagrei using a visual attention reflex, which has been shown in earlier experiments to be less influenced by color categorization than other commonly used behavioral assays. We compared the results to predictions based on two different visual system-based models. The Euclidian distance between pairs of points plotted in a lizard chromaticity diagram effectively predicted the relative responses. A receptor noise model, in which color space distance was estimated in units of “just noticeable difference,” yielded a similarly accurate prediction of the results. We concluded that for studies of color signal evolution, either of these methods may be effectively employed to make behaviorally-relevant predictions of perceptual distance among colors that are widely separated in visual space.