You have to distinguish two questions: 1) what color(s) is the represented dress? 2) what absolute pigments are present in the physical image? These are very

You have to distinguish two questions: 1) what color(s) is the represented dress? 2) what absolute pigments are present in the physical image? These are very different questions. Painters know this distinction, at least implicitly, since a big part of painting is producing illusions with the clever application of color. The checkershadow illusion is a similar demonstration of the distinction between medium and message (or sense and reference, to borrow Fregean language). Phenomena like this reveal that color isn't something known purely by sense perception and that some reasoning is required to conclude what the color is. (A superficial analogy may be made with the theory-ladenness of observation perhaps, though that is a different phenomenon). I wouldn't be surprised if there is a relation between the differences in colors perceived and the ability to reason contextually. Indeed, autism and Asperger's have been characterized as entailing "context blindness".