Rubin's Vase (Face/Vase Illusion)

A vase, or two faces

The figure is a dark vase on a light background. Its two curved edges are also the outlines of two faces in profile, looking at each other nose to nose. Which you see depends on which region you treat as the object: the dark one gives you a vase, the light ones give you faces.

You can't see both at once. The boundary between dark and light can only belong to one side at a time, so when the faces appear, the vase doesn't just fade into the background; it stops being a shape at all and becomes the empty space between two heads.

Deciding what's object and what's background

Before the brain can recognise anything, it has to split the scene into figure (the things) and ground (the space behind them). Edgar Rubin was the first to study this step systematically, and his name for it, figure-ground organisation, is still in use. Normally it's settled instantly: a mug on a desk is obviously the figure. Rubin's drawing is built so the contest can't be settled, because the shared edge fits a vase just as well as it fits two profiles.

With nothing else to break the tie, perception commits to one reading, holds it for a few seconds, and then switches, the same alternation you get with the Necker cube. Brain imaging shows that the switch is more than a change of opinion. Uri Hasson and colleagues found that face-selective regions of visual cortex were more active when people saw the faces than when they saw the vase, even though the image was the same (Hasson et al., 2001).

Tipping the balance

Rubin also identified cues that push a region towards being seen as figure: being enclosed, being symmetrical, being convex, and being the smaller of two regions (see Wagemans et al., 2012, for a modern review). The slider in the demo uses size. Narrow the vase and the dark region becomes the smaller one, which favours the vase; widen it and the faces are squeezed into thin slivers, which favours the faces. The profiles themselves don't change shape at all, only how much space each region gets.

Rubin's figure

The Danish psychologist Edgar Rubin published the figure in 1915 in his doctoral thesis, Synsoplevede Figurer ("Visually experienced figures"), with a German edition following in 1921. People had noticed reversible images long before, but Rubin turned figure and ground into a problem that could be studied experimentally, and his work fed directly into the Gestalt psychology that was taking shape in Germany at the same time. The duck-rabbit and the "my wife and my mother-in-law" drawing flip in a different way: there, the same lines get two different identities, rather than the same edge being claimed by two different regions.