Peripheral Drift Illusion

A still pattern that creeps

The peripheral drift illusion is the general name for still patterns that seem to drift, ripple or rotate, but only out of the corner of your eye. Look straight at one part and it settles down; let your gaze wander around, and the parts you aren't looking at start creeping again. The rotating snakes are the most famous example, but simpler patterns work too, such as the rings of sawtooth brightness ramps in the demo above.

The key ingredient: a lopsided ramp

Every pattern of this kind is built from a repeating unit whose brightness changes unevenly, for example a sudden jump from black to white followed by a slow fade back to black. That lopsidedness is the whole trick. Evenly spaced light and dark bands produce no drift at all, however you arrange them. If you tick "Make the brightness ramp symmetric" in the demo, you'll see the movement stop.

Why the edge of your vision

Where the pattern falls on your retina matters, because different parts of your visual field are wired differently. The fovea, the small central area that gives you sharp detail, is good at resolving fine, still structure. The rest of the retina relies more on fast, motion-sensitive pathways, which are useful for noticing movement out of the corner of your eye, and it's those pathways this pattern fools. Tiny involuntary eye movements, which happen several times a second even when you think you're holding perfectly still, keep refreshing the false signal in the parts of the pattern you aren't looking at directly (Otero-Millan, Macknik & Martinez-Conde, 2012), and that's why those parts never quite settle.

Discovered before the snakes

Akiyoshi Kitaoka's rotating snakes made the effect famous, but he didn't discover it. The first report came from Fraser and Wilcox in 1979, who described illusory rotation in a disc filled with repeating sawtooth ramps, very like the demo, and noted that people disagreed about which way it turned, or whether it moved at all (Fraser & Wilcox, 1979). Twenty years later, Jocelyn Faubert and Andrew Herbert showed that the effect depends on viewing the pattern off-centre and gave it its name. They also proposed the explanation that's still favoured today: different brightness levels are processed with slightly different delays, and eye movements or blinks turn those delays into a false motion signal (Faubert & Herbert, 1999).

What makes it stronger

Kitaoka and Hiroyuki Ashida then worked out which brightness sequences make the effect strongest (Kitaoka & Ashida, 2003), and that work led directly to the rotating snakes. High contrast helps, so true black and white with mid-greys in between will drift more than muted versions. Lots of small repeating units also work better than a few large ones, because there are more edges for each eye movement to set off. If you're curious about other ways to fool the same motion detectors, the motion aftereffect does it with real motion that has stopped.