05/15/2008 07:00:00

Illusions of the Present

PASADENA, Calif.--Have you ever noticed that signposts and trees on the side of the road seem to whoosh by faster right as you drive past them, or that a door frame seems to curve outward as you approach it? These are just two examples of real-life movements that underlie more than 50 types of illusions, now systematically organized and explained by scientists at the California Institute of Technology.

The systematization also lends a glimpse into how illusions are not simply tricks your brain likes to play on you; they are manifestations of how the visual system evolved to keep up with real-life motion. These illusions now fall into 28 predictable categories defined by Mark Changizi during a fellowship in the Sloan-Swartz Center for Theoretical Neurobiology at Caltech and appearing May 28 in the journal Cognitive Science.

"I had been reflecting on the classical geometrical illusions always shown in Psychology 101 classes--the ones involving lines and vanishing points--and it struck me that I can explain them," Changizi says.

To picture a geometrical illusion, imagine a spoked bike wheel with two squares superimposed on it in different places. The square closer to the center, where the spokes meet--called the vanishing point--will always seem larger than the square toward the rim of the wheel. In other words, the closer an object is to the vanishing point, the larger it appears.

Your brain thinks that you are physically moving forward. In real life, forward motion would generate "spokes" on the eye's retina, "like in Star Trek, when they go into warp speed," Changizi describes. It would also bring you closer to that square near the wheel's center, naturally making it seem larger than the other object. "Your brain generates a perception of what the world will be like in the next moment because by the time that perception finally occurs--it takes about a tenth of a second--that object will be larger," explains Changizi.

"Later I realized that my same old idea could be radically generalized, so that it made predictions not just about geometrical illusions, but about 27 other illusion classes as well," Changizi says. "I realized that I could make a massive pattern of predictions about the kinds of illusions humans are subject to."

Changizi built a table, a matrix that distributes the different kinds of illusions into four columns distinguishing what visual feature is misperceived (size, speed, luminance, and distance), and seven rows indicating the different kinds of optical features that occur when an observer is moving forward. "Each spot in this table makes a prediction about perception," he says.

From there, Changizi culled a century's worth of papers reporting what people see when they look at different kinds of illusions. "There are hundreds of illusions collected like butterflies over the years," he notes, "with no real systematics behind them. Just a massive heap of illusions." He wanted to see if each individual case would fall into one of the 28 classes he had designated. "I found that the disordered pile of illusions followed the predicted pattern, and I was able to arrange the illusions in an orderly fashion inside the unifying matrix."

Changizi believes these illusions arise from the way the visual system evolved to process and react to visual cues. Called "perceiving the present," the theory explains, for example, why your hand is ready to hit the ball in a game of tennis: your brain translates the ball's motion into where it will be when you hit it.

"Motion is crucial to the story of illusions. What you perceive is a premonition, not present reality!" Changizi notes. Shinsuke Shimojo, a biology professor at Caltech and a coauthor on the report, explains, "We have evidence from other studies showing that when you perceive a moving object, you localize it in the current position because your brain normalizes it. The brain has been trained via genetic and learning processes to compensate in dynamic situations. This paper says even more--that the brain applies the same algorithm to perceive a static image. Nobody had come up with this theory to explain all illusions in this framework."

The potential applications that the new organization of illusions presents are dizzying. Movies or video games, for example, could incorporate illusions to "make someone perceive motion when the object is actually motionless," Changizi suggests. Another possibility: street signs or other visual warnings could incorporate visual tricks to grab attention, by having a pattern that seems to bulge, spiral, or turn redder as the viewer approaches.

Changizi is now an assistant professor of cognitive science at Rensselaer Polytechnic Institute. Other authors on the study are Andrew Hsieh, a former Caltech undergraduate student now at the University of Southern California; Romi Nijhawan, a psychologist at the University of Sussex in England; and Ryota Kanai, a former postdoc in Shimojo's lab now at the Institute of Cognitive Neuroscience in London. 

Written by elisabeth nadin