Three weeks ago this newsletter ran a piece on olo, the color a Berkeley laser produces by firing at one class of cone cell and nothing else. It hung on Rivka Galchen's New Yorker feature, and it treated the tetrachromat women in the back half of that article as supporting material. That was the wrong call.

This week Niko McCarty pulled the tetrachromacy section out of the same feature, wrote it up on X, and collected roughly 177,000 views in two days. Nobody went viral for the laser. They went viral for the women.

So I went and read the papers underneath, which turned out to be a longer walk than expected, and the story that came back is not the one in either the feature or the post. It is not that a rare woman has a fourth color channel and the rest of us do not. It is that fourth channels are lying around everywhere, in carriers and in ordinary retinas and in the output of a laser, and the visual system almost always declines to do anything with them.

The idea is older than the equipment, and the two men who had it first both died badly.

Hessel de Vries, a Groningen physicist better known for the tree-ring correction that made radiocarbon dating trustworthy, published the speculation in Physica in 1948. If anomalous trichromacy is X-linked, the daughters of anomalous men should carry four pigments, and somebody should go look. In December 1959 de Vries, obsessed with his lab assistant Anneke Hoogeveen and having left his family, killed her with a chisel at her parents' house and then killed himself with cyanide. He had been in line for a Nobel. The carrier question sat mostly untouched until Gabriele Jordan picked it up at Cambridge in the nineteen-eighties, and the first careful data did not arrive until 1993.

Meanwhile, and unknown to any of them, the question had already been answered in a different form and in a different language.

In Moscow in 1956 the physicist M. M. Bongard and his colleague M. S. Smirnov reported that human vision outside the fovea is tetrachromatic. Their method, which they called replacement colorimetry, is what we would now call silent substitution. In the periphery you can set up a match using three primaries and it will hold in the steady state, but swap one such match for another and a transient flashes. Match with four primaries and the substitution is silent, and the match stays stable across changes in light level and in chromatic adaptation. They believed the fourth receptor was the rods, a reasonable guess in 1956 and decades before anyone knew about intrinsically photosensitive retinal ganglion cells.

Bongard died in the Pamirs in 1971. Marina Danilova and John Mollon published the first English translation of the 1956 paper in Vision Research in 2021, timed to the fiftieth anniversary of his death. Sixty-five years in Russian.

Mollon's commentary makes the connection that reorganises this whole subject. The tetrachromacy of the parafovea, he suggests, is analogous to the weak tetrachromacy of carriers. In both cases you get trichromatic matches that refuse to stay stable under adaptation, because there is a fourth signal in there that will not be summarised by three numbers.

Which means the extra channel is not the rare thing. You have one. It is sitting in your peripheral vision, it has been there your whole life, and you have never once noticed it.

One more note on Bongard, because it is too good to leave out. He is not remembered for any of this. He is remembered for the hundred visual puzzles in the appendix of his 1967 book, which cognitive science adopted as Bongard problems and which are still hauled out to argue about whether machines can perceive categories at all. The man who gave artificial intelligence its favourite test of visual abstraction also found a dimension of human color vision that nobody claimed, and the two fields went seventy years without noticing they shared him.

Here is what the feature has and does not quite say. Every strand in it tests the same proposition. Put a signal in front of an adult visual system that evolution never arranged for it to receive. Does the brain build a channel for it, or does the signal go nowhere?

The two engineered interventions work. The two that occur naturally, in millions of people, across entire lifetimes of exposure, mostly do not. That inversion is in neither the feature nor the thread, and it is the thing worth carrying out of all of this.

Jordan and Deeb's paper asks the question outright in its second paragraph. Does her visual system have enough plasticity to take advantage of the input from an extra class of cone? Their answer, assembled over thirty years, is that the retina was never the hard part.

Four things have to go right downstream. The pigments need enough spectral separation to produce a distinguishable signal. The cone ratios have to cooperate, since a mosaic dominated by long-wave cones gives the extra class a cleaner surround to be compared against. Optical density matters, because a pigment at low density has a narrower absorption spectrum and behaves differently than its peak sensitivity predicts. And the anomalous signal has to be strong enough during development to recruit its own population of post-receptoral channels, rather than being drowned out by the ordinary L and M comparison running alongside it.

That last one is the whole story. The extra signal arrives, and the cortex has no particular reason to give it a lane. Olo does not have this problem, because the laser removes the competing signal entirely. There is nothing for the novel input to be masked by. The monkeys did not have it either, because a genuinely new pigment in a two-pigment retina is the loudest thing in the room.

None of which means four channels are hard to use in principle. Birds, reptiles, many shallow-water fish and at least one butterfly are tetrachromatic and manage perfectly well. Mary Caswell Stoddard's group trained wild hummingbirds at feeders in Colorado and showed they discriminate nonspectral colors, the ones no single wavelength can evoke. For us purple is the only one of those. For a hummingbird there is also ultraviolet plus red and ultraviolet plus green, and the team's survey of plumage and plant spectra found plenty of natural colors falling in exactly those categories. The birds are not merely seeing further into the ultraviolet. They are using the fourth cone to carve up a color space we do not have.

There is also a piece of intellectual honesty in the Cambridge literature that deserves preserving. Mollon and Jordan had originally floated an adaptive story. Perhaps tetrachromacy persists because a woman who could read subtle signals in a face, the health of a child or the fitness of a partner, would have an edge. Their own data killed it. Behavioural tetrachromacy is too rare to be maintained by any advantage to the carriers, so if there is one, they concluded, it probably belongs to the sons. The trait survives, on their reading, for the benefit of the people who cannot see straight.

A footnote, offered with the caveat that sixteen months is early for citations to accumulate.

As of this week the olo paper has nine citations and none that Semantic Scholar flags as influential. Deduplicate the preprint and journal versions of one of them and it is closer to seven distinct works. One is the Oz team citing itself. Two more come from the adjacent adaptive optics groups at Berkeley and Washington. One is a news article in Nature, written by a journalist. One is a framework paper about blue light and digital therapeutics. One is a position paper on olfaction standardisation for embodied AI, which I have not been able to make sense of.

The remaining two are philosophy. One in Synthese, on geometrical arguments against similarity structuralism. One in Philosophy and the Mind Sciences, on linking the structure of neuronal mechanisms to the structure of qualia.

The most talked-about result in color vision in a decade has so far been taken up mainly by people arguing about qualia. Which, on reflection, may be the correct audience.