Spinning objects (balls) must certainly be of biased occurrence in modern times of human civilization. Assuming the illusion is a cognitive artifact of a learned phenomena, I wonder how a caveman would recognize this demonstration. Would the illusion be lost? Probably the experience would be quite rad, some way or another, however.
I'd guess that it is not an artifact of learned phenomena. The rotation has the pixels of the ball moving from right to left, and the illusion is that the ball is moving right to left.
I think that this would hold up for cavemen - that if they see part of an object moving from right to left (in their peripheral vision) that they'd also see the whole object moving from right to left.
I agree, I think it has to do with the motion filtering of your retina. If the rotational speed is faster than the linear movement, the motion detection circuits on your retina will detect movement predominantly in the rotation direction. On the other hand, when you fixate on the ball, the brain instead judges motion by eyeball movement, and the illusion disappears.
