The Mona Lisa's half-smile is famously ambigious. Is she grinning or grimacing? Art historians argue that her mercurial mouth is little more than Renaissance special f/x. Neuroscientists offer newer theories: It's about how your brain processes light - or how your eye perceives detail. Maybe all three explanations are right. Leonardo, as good a scientist as he was an artist, probably would have thought so.
Theory 1: Brush strokes
Art experts credit sfumato, the brush technique da Vinci used to paint Mona's expression. From the Italian for "blended" or "smoky," sfumato refers to the use of featherlight strokes to apply layers of translucent color. The blended tones soften outlines, hazing the boundaries between Mona's lips and cheeks [A]. The shadows around the lips seem to pull up the corners of the mouth, suggesting a smile.
Theory 2: Fields of view
Mona's enigmatic expression may literally be in the eye of the beholder. "Look directly at her mouth - her smile disappears," says Margaret Livingstone, a Harvard neurobiologist. "Central vision does not perceive shadowy components well. Look at the eyes, seeing the mouth only with your peripheral vision, and her glorious grin becomes obvious." That's because the human eye's foveal, or central, vision is set up to best perceive detail (and is connected to a disproportionately larger chunk of the brain's visual cortex); peripheral vision is optimized for broad visual strokes.
To illustrate the effect, Livingstone altered Mona in Photoshop. She used a Gaussian blur filter to emphasize the coarse [B] and medium [C] grains, to mimic how you'd see the painting out of the corner of your eye. Then she applied a high-pass filter to underscore the finer detail [D], isolating the dead-ahead view. The result: As detail increased, Mona's grin goes flat.
Theory 3: Mental images
The human brain is wired to perceive emotion in facial expressions. But with millions of neurons clicking on and off, the brain is also intrinsically noisy. This neural static lends a normally imperceptible instability to what we see. And it can make Mona's face seem to flip from happy to sad, depending on what the noise distorts. The critical features: her lips and cheeks.
Researchers at San Francisco's Smith-Kettlewell Eye Research Institute tested the effect of the noise by overwriting a copy of Mona with random, computer-generated pixelation. In the images where the pixels accidentally enhanced the curvature at the corners of her mouth [E], observers saw Mona as happy. Images where the pixels got in the way of the curve, giving a straighter line [F], had the opposite effect. "It worked with a photo of my girlfriend, too," says vision scientist Leonid Kontsevich. "But she refused to have her picture published in the study."
- Kari Lynn Dean
from Wired
