by Brian on August 27, 2003

I imagine most readers have seen Edward Adelson’s checkershadow illusion, because it’s done the rounds of a few blogs. If you haven’t seen it it’s worth looking at, because it’s really quite remarkable.

I always find these things fascinating because of what they tell us about the boundary between perception and inference. So I was interested to see how many similar illusions he has posted, many of them as part of the very good HTML version of Lightness Perception and Lightness Illusions.

I had always thought that most of the illumination illusions were all trading in some way or other on either simultaneous contrast or perceived three-dimensionality. But it turns out that this can’t be the whole story, at least if simultaneous contrast is construed fairly narrowly.

In the snake illusion two of the squares that appear to be different shades are (a) the same shade and (b) surrounded by areas that are also the same shade. If there’s a contrast effect, it’s from areas that surround the areas that surround the squares. Either there are many more contrast effects than seemed a priori plausible (not that contrast effects were given much weight back when philosophers tried to do psychology a priori) or something other than contrast effects are active here. (The best Edelman can do with it is to suggest that straight lines are better than curved lines at marking off contrast areas, which seems to describe the phenomenon more than explain it.)

I’m no expert in this area, so I don’t know someone has explained what’s going on with the snake in the 7 or so years since it appeared, but it seems very mysterious to me.



Murray Smigel 08.27.03 at 7:20 pm

I put up a card with holes over all but the A and B parts of the checkshaddow illusion image, and the squares continue to look different in brightness to me. Shouldn’t the masking of the rest of the image abolish the illusion?


Brian Weatherson 08.27.03 at 7:56 pm

That’s surprising. Yes, that should eliminate the illusion.

It could be because you’re at some level remembering what is under the card. Also there’s going to be a mild contrast effect between the squares and the letters inside them, but it’s hard to imagine that could explain everything. (In other words I don’t know how to explain your result and I’m guessing a bit.)

I wasn’t convinced the squares were the same shade until I threw the picture into Photoshop and deleted everything but the squares, which I guess is something like covering the board apart from the squares. They looked the same shade then. (This does seem to undercut what I said last paragraph about the letters causing a contrast effect, doesn’t it?)


Joshua 08.27.03 at 8:11 pm

Try covering up the letters A and B as well–the fact that one is black and the other white is an additional contrast cue. I found that once I did that they did indeed appear to be the same shade of gray (which is what my little web-design color selecter tool insisted despite the evidence of my lying eyes).


Andrew Edwards 08.28.03 at 1:53 am

Save the image and load it into Photoshop or Paint or the GIMP. Use the eyedropper to sample the colour from one, and paint it over the other.

Try using the paintbrush to slowly expand the squares, and see how much you need to do to get them to appear the same.

It took me a while, but the REALLY ARE IDENTICAL.


Dell Adams 08.28.03 at 4:19 am

Covering the intervening space with my finger – making both squares contrast to the same object, in other words – also convinced me they were the same color.

But I now have to suppose that didn’t prove anything, did it? Hmmm.


cerebrocrat 08.28.03 at 4:46 am

I believe in this instance, the illusion may be suffering from the limitations of the medium involved. The illusion is certainly for real; I’ve seen it many times before (it’s understandably a favorite in perceptual psychology). But the image at this link does not reproduce the illusion accurately; the two squares, on my lcd screen, are demonstrably not the same.


Elina 08.28.03 at 8:47 am


I agreed that there was no way the squares were the same colour but copying and pasting into paint and then erasing everything but the two squares I am convinced.

Which all just goes to show how great the illusion is.


bad Jim 08.28.03 at 11:10 am

Edwin Land? Retinex theory? I thought Oliver Sacks gave it a thought. The eye integrates shading across boundaries, or something.

This is the second time I’ve seen this illusion. The first time I envisioned using a hole punch to make two carefully spaced apertures to test it. Both times I wondered why I trusted my monitor to give a calibrated result, and didn’t bother.


Murray Smigel 08.28.03 at 3:30 pm

Ok, I cut uniform blocks of the A and B squares into a blank canvas, and yes they are the same, wow!
I also did the test with small blocks containing the letters A and B, and there was a little difference between the colors, but nothing as large as the actual shadow illusion.



Tim Lambert 08.28.03 at 4:57 pm

It made a great demo in class the other day. “They are the same colour! Who believes me?” No one, it would seem. Using GIMP’s pencil tool, I fill in the area between the squares with the same colour. 300 students go “Oooh”.


Matt McIrvin 08.29.03 at 6:35 pm

Computers can be fooled too, if not so subtly. Some of the people who insist that the squares in the checkershadow illusion are *provably* not the same color just might be viewing them on computers that actually are giving them different colors, because they’re using some sort of error-diffusion dither that can be led astray by local contrast effects.

They’re the same color on my screen, though.

Comments on this entry are closed.