Fig. 12.

Questions about the illumination. The first row (a-e) shows the dress scores for DressO (along the y-axis) as a function of the observers’ answer to the respective questions concerning the illumination. The second row (f-i) illustrates the same kind of results for the jacket scores. The first column (“shadow”) refers to the question of whether the dress (a) and the Jacket (f) were in the shadow (“yes”=in the shadow). The second column (“flash”) distinguishes between observers that believed the dress (b) and the Jacket (g) were in the flash of the camera. In the third column (“daylight”) the question was whether observers thought the dress (c) and the Jacket (h) were in daylight. Panel d (“overexposed”) concerns the question of whether the image of DressO was overexposed (there was no such question for the Jacket). The last column refers to the question about the source of gloss (reported in the section on gloss). In this question observers could indicate that they believed the gloss was due to the fabric, the light or both. Numbers at the bottom of the diagram indicate the number of observers that gave each answer option. Error bars correspond to standard errors of mean. Symbols above the bars report significance of t-tests comparing dress and jacket scores between observers who answered yes and No. ^***p < 0.001, ^**p < 0.01, ^*p < 0.05, ^°p < 0.1, ns = not significant. Note the significant results for DressO (first row), but not for the Jacket (last row).