Mystery of Moving Eyes Solved

The mystery of why eyes in certain paintings and photographs appear to move has been solved: it has to do with how we perceive two and three dimensions, a new study finds.

According to a paper published in a recent issue of the journal Perception by James Todd, a professor of psychology at Ohio State University, and his colleagues, the optical illusion "is in the misleading information provided by the picture," Todd said.
No matter what angle you look at a painting from, the painting itself doesn't change, since it's on a flat surface. The patterns of light and dark remain the same.

But three-dimensional objects, in life, change with the way light falls on them as viewers move around the object.

"When observing real surfaces in the natural environment, the visual information that specifies near and far points varies when we change viewing direction," Todd said.

He added, "When we observe a picture on the wall, on the other hand, the visual information that defines near and far points is unaffected by viewing direction. Still we interpret this perceptually as if it were a real object. That is why the eyes appear to follow you as you change your viewing direction."

Todd and his team found the mechanisms behind the visual effect by analyzing a 3-D picture of a human torso. Their study consisted of two phases.

First, the scientists used dots to mark near points and far points. Todd told Discovery News that those points referred to areas on the image that appear to be closer or farther away. The researchers mapped out different groupings of these points based on their observations of the image from a number of viewing angles.

Next, they placed a gauge figure over the torso — a computerized circle with a needle sticking out of it. At numerous places within the image, the researchers manipulated the gauge so that the needle would appear to be perpendicular to the surface of the torso. This provided information about perceived depth.

While the torso looked "squashed" when viewed from an angle, the researchers discovered that changes in viewing direction had little effect on how the observer saw it.

"When I move my gaze with respect to the torso, I do perceive a slight rotation especially in the region of the buttocks, but to my eye this effect is not nearly as compelling as the one I experience with eyes or a pointing figure," Todd said.

He explained that the position of an individual's pupils provides information about where he or she is looking. Paintings and photographs can manipulate human perception because viewers think they see figures in three dimensions on what is essentially a flat, two-dimensional surface.

"I suspect it might even work with a smiley face, provided that it contained a circular or oval shaped eye with a pupil in the center," he said.

James Cutting, a psychology professor at Cornell University, told Discovery News that the analysis conducted by Todd and his colleagues is "rigorous" and its resulting database of far, near, and 3-D points is "extensive."

Cutting said his own research findings on how people perceive pictures of faces and rotating objects, as if they were in a movie theater sitting in the front row side aisle, were very consistent with the torso study. They also reveal that even a seemingly bad movie seat may not have too much affect on the moviegoer's view.

"Rather than measure apparent slant with gauge probes, I simply asked viewers about what they saw — did something look rigid or nonrigid, did the face look distorted," Cutting said. "Basically, not until one looks at a picture from a fairly steep angle, say 45 degrees where 90 degrees is straight on, is one bothered by the distortions."

From Discovery