Christopher Tyler and Leonid Kontsevich at the Smith-Kettlewell Eye Research Institute in San Francisco manipulated a digital image of the painting by introducing random visual noise- the equivalent of the snow seen on a badly tuned TV set- and asked 12 observers how they rated the resulting expression on a four-point scale from sad to happy.
As would be expected, noise that lifted the edges of her mouth made Mona Lisa seem happier, and those that flattened her lips made her seem sadder (Vision Research, vol 44, p 1493).
More surprising though, was how readily the visual noise changed people's perception of the Mona Lisa's expression.
Tyler says our visual system contains many sources of noise: fluctuations in the number of photons hitting light-receiving cells in the eye, spontaneous false activation of photon absorbing pigments, and randomness in the firing of neurons that carry the visual signals to the brain.
Tyler thinks this natural noise makes people observing the picture believe its expression is subtly changing, rather than thinking they are seeing a single ambiguous expression.
"That may be part of what makes the painting so powerful," he says, something Leonardo must have instinctively realised.
This article appears in New Scientist issue: 26 JUNE 2004.
PLEASE MENTION NEW SCIENTIST AS THE SOURCE OF THIS STORY AND, IF PUBLISHING ONLINE, PLEASE CARRY A HYPERLINK TO: http://www.newscientist.com.