Rivka Galchen's New Yorker piece hangs on one experiment. A Berkeley team pointed a laser at the back of a living eye, lit up a patch of color-sensing cells in a pattern that never happens in nature, and five people reported seeing a blue-green so saturated that no screen, pigment, or rainbow can match it. They named it "olo." The natural companion to the glowing-mice story is not another hidden signal. It is the opposite idea: that even the colors you see every day were never sitting out there in the light.

That is the payoff worth carrying to dinner. Biophotons are a real signal we cannot detect without a cooled camera. Olo is a reminder that color is not a signal at all. It is a verdict the brain returns.

The tool is Austin Roorda's adaptive-optics platform, the kind of retinal imaging built to study eye disease one photoreceptor at a time. Ren Ng, a computer-graphics professor, asked a mischievous question: what if you stimulated only the M cones, a firing pattern nature never produces because ordinary green light always hits L cones too? The system, named Oz, first maps a person's exact mosaic of cones, then uses a 543 nm green laser to deliver microscopic pulses to a chosen set of them, tracking the constantly moving eye in real time to stay on target.

They held roughly a thousand M cones on, and only M cones, across a patch about the size of a thumbnail at arm's length. Subjects saw olo: a deep, glowing teal. Asked to match it against the purest natural green available (a single laser wavelength near 510 nm), they said the natural green looked washed out beside it. The control is the tidy part. When the team deliberately let the laser drift onto neighboring L and S cones, the color collapsed back to plain green. The novel percept tracked the novel stimulation. Five subjects, including Ng and Roorda themselves.

Galchen's better move is to widen out from the gadget. Two older lines of research make the same point without a laser.

Tetrachromats. Most of us run three cone types. A small number of women carry a fourth, a quirk of the same X-linked genetics that gives men color blindness. Gabriele Jordan and John Mollon spent years hunting for functional ones and found them. One subject, a hematologist, could reliably pull one color out of a set that looked identical to everyone else. Another, a midwife, could see meconium staining in amniotic fluid that her trained colleagues could not. Their extra channel does not feel like a superpower from the inside. It is just how the world has always looked to them, which is the unsettling part: you would never know you were missing a dimension.

The dress. The 2015 photograph that split the internet into white-and-gold versus blue-and-black is the same lesson with none of the equipment. The dress is blue and black, shot indoors. The disagreement comes from color constancy: your brain silently guesses the lighting and subtracts it, so it can call a wheelbarrow red at noon and at dusk. Guess warm light and you subtract orange and see blue. Guess cool light and you subtract blue and see white. Two people, one image, different priors, different color. Nobody is wrong, because the color was never fully in the photo.