Artists have often been ahead of scientists in figuring out how we see the world around us. Leonardo da Vinci, for example, understood how to make pictures look three-dimensional before there was a scientific theory of perspective. According to an article in tomorrow's Science, paintings of a modern-day Leonardo have served as scientific experiments--and countered a fundamental notion about perception.
Chuck Close is famous for painting disconcertingly large portraits, in which a single head spans a 2-meter canvas. Many of these portraits are mosaics of little blobs of color that psychophysicist Denis Pelli of New York University calls "marks." When you stand close to the picture, all you see is an abstract pattern of marks; step back a few feet, though, and a startlingly realistic face appears.
It's natural to assume that this trick works because our eyes blur the marks together, but Pelli found that isn't the case. He asked five volunteers to look at 33 Chuck Close paintings at various distances from the canvases and note how far away they were when they saw each face. If the effect were due to blurring, the brain would create the face at the point when the viewer's eyes can't resolve individual marks. But instead, Pelli found that the critical mark size was about three times larger than the smallest resolvable mark.
At first one of Pelli's colleagues was skeptical. "I was firmly committed to the view that it was just the optics of the eye that blurred the image," says David Williams of the University of Rochester in New York. He pointed out that the retinal image of the painting at 2 meters is not simply twice as small as it is at 1 meter--it's also twice as blurry. So following William's suggestion, Pelli asked the volunteers to view the paintings through two sizes of pinholes, which enabled him to control the blurriness of the image. Williams now agrees that the apparent size of the marks, not their clarity, determines when the face appears.
Vision scientists are still a long way from figuring out how the brain constructs images, says Concetta Morrone of the Consiglio Nazionale delle Ricerche in Pisa, Italy. But the new research, she notes, "is clearly pointing to the solution."