cognitive science
and more
Café Wall

Another very striking illusion is the famous Café Wall illusion, first described by Gregory and Heard in 1979. The story goes that a colleague of Gregory and Heard felt that the black and white tiling on the wall of a local bar (they worked in Bristol) conveyed a wedge-like impression. They brought this initial observation into the laboratory, refined it and came up with something like the image below. As you will no doubt agree, the illusion that the horizontal lines (which, of course, are perfectly parallel) form a wedge-like pattern is very strong (you can download a script to generate these type of images here).

To the best of my knowledge, this illusion has never been fully explained. There are, however, some clues. One important factor appears to be that the grey line (all tiles are separated by grey “mortar”) is interpreted differently depending on whether it separates two tiles of the same colour or two tiles of a different colour. If the grey line separates two tiles of the same colour it is interpreted for what it is: A grey line. However, if the grey line separates a black and a white tile it is interpreted as a slight “blurring” of both tiles and not as an actual line (this also explains why the illusion only works if the luminance of the “mortar” is intermediate between the luminance of the tiles).

However, the question remains why defective “border-locking” (as Gregory and Heard call it) results in the particular illusion that we observe. One possibility is that a grey line separating a black and a white tile is taken to be part of the black tile, rather than neutral blur (see (a) in the figure below). Therefore, when a black tile is above a white tile, we observe the boundary between the tiles to be lower than when a black tile is below a white tile (b). In turn, this may cause us to perceive the boundary between tiles as slightly tilted.

Again, even if this explains why we perceive small wedges at level of individual tiles, it does not quite explain why entire rows of tiles appear wedge-like (or does it?). My suspicion is that “border-locking” is only one of many factors that play a role in the Café Wall illusion. If you are interested you can take a look at an elaborate analysis of the Café Wall illusion by Schwartz (see references), but don't expect a clear cut explanation!

References

Gregory, R. L., & Heard, P. (1979). Border locking and the Café Wall illusion. Perception, 8(4), 365-380. [PDF]

Schwartz, J. Studies in Visual Perception, III. An Analysis of the Cafe Wall Illusion. Retrieved October 3rd, from http://www.settheory.com/Glass_paper/cafe_wall_study.html