I just submitted my article and so I’m left with plenty of time to waste on answering “silly” questions. For instance, take a look at this. Now take a few steps back and shake your head while looking at it again.

I was really amazed by this optical illusion when I first came across it.

How does it work and what does it tell us about the visual system?

At first glance it’s just a high spatial frequency grating, when you look really close at the picture you only see a couple of faint grey pixels, but shake your head and the purkinje cell clearly emerges.

As a proper scientist, I immediately started googling for the answer but without much success. Then I started thinking and got the entire lab to ponder over the subject. At some point someone even suggested: “Have you considered  that it could be magic?”.

Here is our first attempt at a serious hypothesis:

You have your two main types of cells in the visual system: P-cells and M-cells.

P-cells are good for perception of high spatial frequencies, but they are bad at low contrast and have poor temporal resolution. Whereas the M-cells fail when it comes to fine details but they are much more sensitive to low contrast and temporal changes.

Since the black bars have such a high spatial frequency and contrast, it involves mostly the P-cells. As soon as you start shaking your head they loose track and the M-cells become more dominant.

When you take an increasing distance from the image, you start increasing the spatial frequency to a point where even P-cells start to fail to encode it properly. And as you might have noticed looking at it from an angle also makes the figure pop out, again this seems to fit since peripheral vision is dominated by M-cells over P-cells.

If our hypothesis were true, the effect should be gone when we use equiluminant colours, since M-cells are insensitive to colour. As real scientists we couldn’t help ourselves from testing it by replacing the black and white with red and green:

And… although the effect seems less, it’s still there… We’re back to the drawing board, any suggestions are welcome!

2012-02-27

As someone correctly mentioned the above green and red image doesn’t have equiluminance and so it’s not a proper test. Here’s a true equiluminant figure with a grating on top. Can you see the hidden text?