How do we see things in color? How do we know objects stay the same color when the color of the light they reflect changes as the lighting changes? We see this effect most dramatically in the theater, where the stage lights cover every color of the rainbow, yet we still know the heroine is wearing a purple dress and our hero has majestic blonde hair.
In today’s reading (“Surface-Illuminant Ambiguity and Color Constancy: Effects of Scene Complexity and Depth Cues” by James M. Kraft, Shannon I. Maloney, and David H. Brainard of UC-Santa Barbara [Perception, 2002]), the authors explore some of the implications of this issue.
The problem of color vision is very complex—so complex that there’s certainly no way I can explain it in a single blog post. Nonetheless, much of how we perceive color is well understood. What I’ll be discussing here is a second-order problem. Seeing color is one thing, but adjusting for different lighting conditions—color constancy—is yet another. Somehow, we’re able to acheive color constancy without even thinking about it. But how? Do we unconsciously detect the type of lighting in a particular scene and then adjust accordingly? This seems unlikely; otherwise the art of photography wouldn’t be difficult at all: we’d all intuit how to adjust our cameras to different lighting conditions and take good pictures. Yet the proponderance of ghastly orange photos murking up cheesy family websites and dust-covered-photo albums suggests we possess no such intuition.
Kraft et al. suggest that we might use the other objects in our field of view to help us determine the accurate color of another object. For example, if we see two people in a poorly illuminated room, we could compare the person we know (say, Jenna Bush) with someone we don’t (her new boyfriend, Henry Hagar) and conclude that Henry has dark hair and pinkish skin, simply by comparing what we know with what we don’t know.
Kraft and his colleagues designed an experiment using a simpler setting: a “room” made of cardboard, with a gray patch on the wall whose color could be changed by adjusting a spotlight focused precisely on the patch. Theater lights were used to vary the overall lighting in the room, and then observers were asked to change the lighting on the gray patch so it looked “perfectly gray.” Later, the same observers were asked to do the same task in a more “complex” scene (the same room, but with a couple extra objects added: a board covered with patches of varying colors and textures, and a tall rectangular cardboard column). They suspected that the “complex” scene would help observers maintain color constancy—with many other cues to help them see how the room was lit, they would be better able to recognize a “truly” gray object. You can see their “room” here.
What they found surprised them. Observers were no better at attaining color constancy in a “simple” room than they were in a “complex” room. Perhaps even the simple room was complex enough to help observers attain color constancy. They tried a different tack: they showed viewers the same rooms, but added additional “invalid cues”—they lit some parts of the room differently in order to confuse observers. In this case, observers were better at maintaining color constancy in the “complex” room, suggesting that complexity helps us attain color constancy in confusing lighting situations.
In short, we’re very good at determining the color of objects in most situations. We probably wouldn’t need to see Jenna next to Henry to know he has dark hair. It’s only when things get really confusing that we must resort to crutches to help us accurately see color. It may also explain why it’s so difficult to portray “confusion” in a theater. When there’s supposed to be something like a storm or a battle on stage, the actors have to rely on the audience to use its imagination and suspend disbelief. Our perceptual system is simply too good to be fooled that easily.