Yannis Behrakis / Reuters

We think of color as an innate part of the world. An apple is red, a leaf is green, the sky is blue—all the evidence of our eyes suggests that we are seeing something fundamental about physical reality. And yet there are moments when color seems more like an illusion. If you sit in a space with green light for a while—say, a green plastic Porta-Potty—the world is suffused with red when you emerge. People looking at the same photograph of the same dress have radically different perceptions of its colors. At dawn and dusk, everything looks bluer, such that flowers that are red in sunlight are almost purple.

Your brain is up to something. But what?

In the case of the blue dawn effect, a pair of researchers, one at CalTech, the other now a post-doc at Harvard, recently published some experiments that may finally reveal what is happening.

Let's back up and start with some basic biology: When we say we “see” color, we mean that the light striking our retinas causes nerve cells to fire. Our brains compare the signals from three different types of cells to tell what to perceive. These cells each have a protein, called a cone, that will respond to short, medium, or long wavelengths of light; you can think of them as being a trio of musical instruments. The medium cone hoots loudest when medium-wavelength light strikes it. The others respond more quietly. From your perspective, each chord—each combination of responses—is a color.

The cone cells all send their messages to the brain via the same set of middlemen, and sometimes, when one kind of cone works overtime, it can cause the output of one of the others to be dampened. This seems to be the biology behind the red-green effect: Cells with the medium-wavelength cone, also called the green cone, fire away like crazy while you're exposed to green light. But when they're no longer being stimulated, the suppressed short-wavelength, or red, cone cell surges up, turning the world scarlet. And even in dim conditions, when cones are not being activated by bright light, they keep sending out a baseline signal, a sort of dial tone.

This is where Max Joesch, now of Harvard, and Markus Meister of CalTech come in. As a post-doc in Meister's lab, Joesch was running experiments on how mice see motion, and he was getting some very strange results. Sometimes he'd get what he'd been led to expect, sometimes he wouldn't. When he got to the bottom of the discrepancies, he realized they involved the eye’s rod cells, which take over vision from cone cells when it gets too dark for them to fire. It turned out that under the lighting conditions of his experiments, sometimes rods were active, and they were doing something quite unexpected.

What Joesch and Meister uncovered was that the rod cells were dampening the dial tone of the mouse cones. That created a difference between the cones' outputs that helped explain Joesch's results—and could, extrapolated to humans, explain the blue dawn and dusk effect.

Mouse and human visual systems are different. But the same kind of neurological connection exists between humans' rods and red and green cones. Meister and Joesch hypothesize that when the light is dim, the rods are active, and they dampen the output of the red and green cones. But the long-wavelength cone cell, also known as the blue cone cell, keeps going all on its lonesome. That gives you the impression that you're seeing blue.

If this is what's happening, it would also explain a mysterious little phenomenon experienced by Viagra users. Some men report that that there's a bluish tint to their vision while they're on the drug—and the medication is known to make the rods active at light levels at which they'd normally be dormant. It could be that the rods are muffling the red and green cone's signals in this situation as well.

So when you're outside looking at your neighborhood as the sun goes down, pay attention for the moment when everything starts to look blue. That may be the moment when the cells in your eyes, like those in mice, create an unexpected and beautiful chord.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.