AMES WINDOW

Why do we feel that the window appear to switch its direction of its rotation after every half-rotation? In fact the window just like the bar rotates continuously in one direction. The trick is that the window does not have a rectangular shape, as we would expect, but rather its shape is trapezoidal. However, we see it as a normal, rectangular window, and we do not consider the possibility that the window could have some other shape with converging horizontal sides. Instead, the converging sides are attributed to the perspective projection of a normal rectangular window when it is oriented at an oblique angle to our line-of-sight.

As a consequence, the physically longer vertical side of the window is seen as the closer side as it rotates, even though it is actually farther away. As the longer edge of the window gets closer to us in the plane perpendicular to our line-of-sight, its motion appears to slow down, it stops for a moment and then appears to rotate in the opposite direction. The same happens when the window approaches the frontal plane 180 degrees later. Thus the window appears to change direction twice during one physical revolution.

To enhance the paradox of this illusion, a metal bar is inserted through the middle of the Ames window. However, because the metal rod is of uniform thickness (unlike the window itself) we perceive it to be rotating continuously in the same direction unlike the window. As a consequence, it first appears to rotate with the window and then at a critical point, it appears to pass through the window in order to continue its rotation.

The fundamental conditions for seeing this illusion is that the window is trapezoidal in shape and it is viewed at appropriate viewing distance. The effect can be intensified further. For example, by painting false 3-D shadows on the window frame, the impression of the illusion is strengthened. The difference in the length of the vertical edges of the window frame is also important, as well as the difference in the length of the shorter vertical side and the horizontal side. It was experimentally found that the optimum ratio for strong illusory percept is 1.5 : 1 in the case of the long to short vertical side ratio and 1 : 1 in the case of the width to short vertical side ratio. Similarly, the illusory effect can also be affected by the speed of rotation and the viewing distance. Finally the illusion is stronger when viewing with one eye closed.

On the other hand, it is possible to see the actual shape of the window when it is viewed from a position that is deviated from the central axis (eg. while squatting), or from a short distance and using both eyes, so that the magnitude of the perspective distortion is too large to be attributed to the slant of the window and when the binocular perception is better able to reveal the actual shape of the object.

The inventor of this special window, as well as of the unusual room in the third room, is Adelbert Ames.

/ Day, R. H. & Power, R. P. (1965). Apparent reversal (oscillation) of rotary motion in depth: An investigation and a general theory.Psychological Review, 72, 117–127. /