From Anstis & Casco, 2006, Movie 1, p. 1088
OK, here’s a really, really cool illusion published last year, and that I learned about only recently. To see it, go to Stuart Anstis’ page here, watch the first movie only, and then come back here.
You should have seen two flies moving in circles with the same radius. The flies’ rotations are offset so that one is at 6 o’clock when the other is at 12, but otherwise, the circles they’re tracing are identical. Now go back and watch the second movie.
As the caption notes, the two flies are still moving on identical circles, except that they’re out of phase as in the first movie. However, moving the background, which is in phase with the left fly (and thus out of phase with the right fly) makes the left flies orbit look much smaller than the one on the right. In the published version of this illusion(1), Anstis and Casco found that people perceived the right fly’s orbit to be 2.3 times larger than the left fly’s, on average. Again, this is despite the orbits are, in fact, the same size!
Now watch the third video. Again, same to flies with identically-sized, but out of phase orbits, but this time with the background in phase with the “horizontal component” of the left fly’s orbit (that is, the background’s at 3 and 9 o’clock at the same time as the left fly), and with the “vertical component” of the right fly’s orbit (meaning it’s at 6 and 12 o’clock at the same time as the right fly). Now the left fly’s orbit looks really wide and short, while the right fly’s orbit looks really tall and thin. In Anstis and Casco’s study, the left fly’s orbit looked 2.2 times as wide and 6.3 times shorter than it actually is, while the right fly’s orbit was 3.5 times thinner and 2.1 times taller than it actually was.
Anstis conducted a third experiment as well, which unfortunately is not represented on his demos page (at least I can’t find it), but if you have access to the Journal of Vision, you can watch it here (it’s Movie 4), which basically replicated the second experiment with “interrupted motion.”
Anstis and Casco don’t offer a detailed explanation for the illusion, but they does provide this suggestive real-world example of an analogous illusion of motion against a moving background:
Johansson (1950) pointed out that when a friend waves to you from a train, his or her hand traces out a horizontally extended sine wave relative to the earth. However, that is not what you see. The visual system decomposes the movement into the linear motion of the train plus an up-and-down movement of the hand. (p. 1087)
In this case, and likely in the bluebottle illusion, your visual system is trying to separate the different sources of motion so that it can represent each with the effects of the other excluded. In the waving-from-a-train example, this makes it look like the person’s waving hand is moving just as it would if the person were standing on the train station platform with you. In the bluebottle illusion, then, the visual system tries to subtract the background motion from the flies’ motion, but because the background motion is in full or partial phase with the fly’s motion, the visual system doesn’t subtract the background’s from the fly’s motion, and you get the illusion.
1Anstis, S., & Casco, C. (2006). Induced movement: The flying bluebottle illusion. Journal of Vision, 6, 1087-1092.