Carlos Basabe

I drew oodles of psychedelic drugs for an article in the current issue of Men’s Health, about how scientists are testing out the use of these substances as possible treatments for anxiety and depression.

We have three cones: long (blue), medium (green), and short (red), that change their membrane voltage based on the wavelengths that reaches them. This spectral sensitivity is arranged so that there are a few wavelengths that produce a peak response, but there’s a lot of overlap of in responses. That is, many wavelengths stimulate more than one cone. (Test it out! Draw a few vertical lines on the image below and see how many curves it can pass through.) 

The colored areas above highlight the perceptive regions where your eyes and brain can to easily differentiate color. The gray areas below the curves are areas that have to be disambiguated somehow. Let’s call them “ambiguous zones”.

Notice how big the ambiguous zone is under the red and green curves. This means that there are LOTS of wavelengths that cause red and green cones to respond. So how do we tell the difference in colors if the response between red and green is the same? One way is to compare the amount of response  across the three cone types. For example, if the color is a bluish green, (say 550 nm on the graph above), the green cone will respond more (that is, its membrane voltage is higher) more than the blue cone. Another way is to use local spatial information and compare the ambiguous color to the colors surrounding it.

The posted photo has lots of color values in the red-green ambiguous zone, and your brain is having a hard time telling them apart.  In the case of overlapping values of color where response is similar across cones, our nervous systems need to do a little extra somethin somethin. Enter the Retinal Ganglion Cells.

Many photoreceptors, such as cones, send their signals to one or a few retinal ganglion cells nearby. Retinal ganglion cells do a lot of spatial processing in vision. If you think of your retina as an LED screen, photoreceptors (in this case, cones) would be the individual LEDs, and ganglion cells would break up the screen a grid-like fashion, so your brain can process what’s happening in one screen position relative to another. The subsections of screen are called receptive fields. (You can also think of RGC’s as that one person who always has the gossip on people physically near them. They always know what’s up, and they’ll always tell you what’s going on where they’re at.)

Retinal ganglion cells are super important in perceiving contrast and edges. They respond to opposite responses that happen in that little region of the retina. Retinal ganglion cells circular receptive fields made up in what’s called center-surround orientation. Usually, for a cell to send a signal, the center of the circular region MUST be light while the outside of the circular region MUST be dark, or vice versa. This works for color, too. For a ganglion cell on the left to fire (called a green-center ganglion), the retinal region must see mostly green, and NO red. For the cell on the right (called a red-center), the retinal region must see mostly red and NO green. (Remember, white light has all wavelengths, so it’s gonna set everything off.)

In that photo, we’re seeing various shades of color that fall in that ambiguous zone under red and green, and these green-center retinal ganglion cells get a lot of green, but there are a few red-center retinal ganglion cells that are stimulated too, since we’re in this ambiguous region between red and green perception. SO! When you look at these shades of green next to one another, your red-green opposing retinal ganglion cells will fire, telling your brain that the ambiguous color is red, and not, in fact green. If you take a slice of the image, you don’t perceive this effect as strongly because you can no longer stimulate all the cones in the circular receptive fields in entirety.

Last step: let’s test a hypothesis.

If this illusion happens because the color stimulates two cones almost equally, and relies on retinal ganglion cells to tell the difference locally, we should be able to do this with the blue/yellow color boundary as well. (Yellow is detected largely by the green cone.) We can do this by taking a picture of lemons…

and tinting it so that all the “yellow” color values end up in an ambiguous color zone… et voila! The illusion occurs again!

To see if the perceived yellow colors are in the ambiguous zone, I approximated the wavelengths by matching the color value (without shading) to wavelength using this converter. (Since I’m matching it by eye, I have low confidence in my estimation, so I made the representative rectangles wider than they actually should be.) Look at where the “yellow” values fall in relation to the responses of the cones. The “yellow” values are closer to green cone’s response curve. Our “test” is successful, the “yellow” colors actually do fall into the ambiguous zone! Huzzah!

If you’re handy with Illustrator, Photoshop, Affinity, and the ilk, try this out yourself! You should be able do it with any image that includes opposing colors (red/green, yellow/blue).

In short, there are many wavelengths to which many cones respond almost equally. Your retina has some tools that detect contrast in both brightness and color value. When a color value that falls between to opposing colors (blue/yellow, red/green), your brain relies on outputs from retinal ganglion cells to locally compare values in a spatial subregion on your retina, and tell them apart. This strategy works pretty well with natural scenes, but digital manipulation exposes where it fails. Cool!

People with color blindness basically see this way all the time. Depending on the type of color blindness, folks may have much bigger ambiguous zones. So their color differentiation relies more heavily on their retinal ganglion cells!

Here’s the technical literature (pdf) that clarified all of this for me.