Figure 14.

When the dress is embedded in scenes with overt cues to the illumination, the lighting is perceived in a predictable way. Subjects' ratings of the illumination in the simulated scenes from Figure 13 (N = 1,127; online = 1,074; in-lab = 53): warm illumination (top row) and cool illumination (bottom row). Conventions are as in Figure 11; data are grouped by the color terms that subjects used to describe the original dress image. Subjects' ratings of the foreground light did not differ as a function of initial percept (p = 0.9), nor did their ratings of the background light (p = 0.8; two-sample t tests comparing the ratings provided by subjects who originally reported B/K vs. those who originally reported W/G). Rating variance was higher for the original image (Figure 11) than for either test (cool: p < 0.001; warm: p < 0.001; F test), but similar for the tests (p = 0.08).