I came across a question on Quora a few days ago which I was excited to write an answer for. Basically, it was a question along the lines of “what are some cool things about color that can result in ‘freaking illusions’?”

The question got me reminiscing about a class on Waves and Vibrations I took with Walter Lewin back as an undergrad (more-or-less the same curriculum as this OCW course).

Anyways, there was one lecture where he demo’d a couple really neat color experiments illustrating how three-color theory is woefully incomplete. His two demos were:

1. Benham’s Top.(wiki). First observed in the 19th century: spinning an entirely black-and white disk/top results in the perception of brown, yellow, and/or grayish green and blue colors!
Scientific Explanation: the phenomenon isn’t entirely understood, but a leading theory is that differential phase delays in the response of color cells to flickering light results in your brain being fooled into seeing colors.

2. Land’s Color Experiment. My personal favorite color experiment (the video is about 6 minutes long, but well worth the watch!!). Basically, with just a gray-level image + a red-channel image, you can (somehow) see many other colors!

I’ve done a digital quasi-emulation of this experiment by making two copies of a color picture.
- For the first copy, I extract out just the red channel.
- For the second copy, I convert the image to graylevels.
Then, I interleave the two images to create a composite image. (i.e., even rows in this new image are from the first image, and odd rows are from the second image). The result is something like this:

Click/open up the full-size image and zoom in on the image/inspect it closely (you may have to magnify a few times). You should see alternating gray and red/pinkish lines. Now, take a step or two back and look at the whole scene — do you see colors? I see red, yellow, and blue!

(For those curious if this only works because you have an expectation of what the colors of the Mondrian should be, check out some other pictures I’ve made.)

*Note: some computer browsers might do some funny interpolation which might mess up the results of the experiment. If you’re not getting good results, try downloading the image from the link above and then opening it in photoshop/preview/an image viewer.

**Final Note: This is a cross-post from my Quora answer, with the exception of having the ability to embed short clips of video in the article thanks to the good people at Reelsurfer (Disclaimer: I am friends with one of the founders of Reelsurfer).