Abstract
Whether The Dress is blue and black, white and gold, or just plain ugly, one thing is certain—it is guaranteed to generate passionate discussion. Lafer-Sousa et al. (2015) and Gegenfurtner et al. (2015) demonstrated substantial individual differences in the perceived color of The Dress. Here, we asked whether differences in perception of the ambiguous dress affect neural responses measured using electroencephalography (EEG) in a simple deviant-detection experiment. Our stimuli comprised three images (from Wired magazine), the original ambiguous photo and two whose luminance and color cast were manipulated to force either the blue/black or white/gold percept of the foreground dress. All three images had identical background luminance and color. Subjects viewed a stream of images (displayed for 700 ms; inter-stimulus interval 1000 ms) comprising one standard image (80% of trials), and two randomly-occurring deviants (10% of trials each). Subjects were instructed to identify deviant stimuli via keypress. Each image served as the standard in one block of 500 trials, and as a deviant in two other blocks. We recorded EEG scalp potentials elicited by the onset of each stimulus while subjects performed this task. The visual P300 was reliably evoked by deviant images, consistent with the fact that the deviant detection task was relatively easy (behavioral accuracy >90%). We also observed stimulus-driven effects on the magnitude and latency of the P300. When the ambiguous image served as the standard, white/gold deviants elicited stronger responses than blue/black deviants. This effect did not depend on individuals' percepts of the ambiguous image; the P300 was independent of whether subjects perceived white/gold or blue/black. Our results suggest that early deviant detection processes respond to low-level stimulus characteristics and do not reflect perception of color category.
Meeting abstract presented at VSS 2016