The mantis shrimp has twelve types of color receptors. We have three. The obvious story writes itself: they see a version of the world that makes ours look like a grainy black-and-white photograph. A spectrum we can't imagine.
Except when you test them, they're worse at telling colors apart than we are.
In feeding experiments, peacock mantis shrimp fail to distinguish a low-saturation blue from gray. That's not a subtle failure — that's a color distinction a human three-year-old handles without trying. The shrimp, with its twelve color channels, cannot do the thing that three channels are supposedly for.
So what are the twelve channels doing?
The current best guess is that they're not being used for discrimination at all. The proposal — still debated — is that the shrimp treats color more like a barcode than a spectrum. Instead of comparing channels to produce a sense of "this is redder than that," the pattern of which channels fire and at what levels gets matched against stored templates. The shrimp doesn't ask "how does this compare to that?" It asks "does this match something I already know?"
A barcode reader doesn't see the colors on the label. It reads a pattern. The mantis shrimp might be doing something like that — using twelve types of sensors to generate a high-dimensional fingerprint that gets compared against a library of known fingerprints. Food. Mate. Threat. Shell.
If this is right, then twelve channels don't buy you richer color experience. They buy you faster, more reliable categorization. Not a bigger palette — a more precise key.
I find that strange in a useful way. The animal with the most color hardware in the known animal kingdom might not "see color" the way we mean it. It might be doing something the word "color" doesn't cleanly describe.
The researchers themselves are honest about this. Their 2022 paper in the Journal of Experimental Biology ends, essentially, with: we still don't know why the system is structured this way, and we don't know what kind of processing it's actually doing. Twelve years after the paradox was clearly established, the question is open.
What I keep coming back to is the original assumption — that more receptor types means richer perception. It seems so obviously true that it takes some effort to notice it's an assumption. More wavelengths sampled should mean more detail resolved. But "detail" in what sense? The shrimp and the human are using their eyes to do different things. Discrimination is useful if you're trying to distinguish an object from a background. Recognition is useful if the category of the object is more important than its exact shade.
The shrimp hunts fast-moving prey in shallow water. It needs to know what something is in milliseconds, in chaotic light. Maybe what twelve channels buy is not a richer picture but a picture that's harder to confuse. Not more resolution — more robustness.
The part I can't get past is the experiential question, which the paper doesn't try to answer and probably can't. When the shrimp looks at something, is there anything it's like to see it? And if so, what is the quale of a twelve-channel barcode read?
We don't have a word for it. We can barely describe what color feels like from the inside — we mostly describe it functionally, or by pointing. "Red is the color of blood, of stop signs, of sunsets." Ask someone to describe the inner quality of redness and they run out of language quickly.
For the shrimp, we have even less. We have behavior. We have electrophysiology. We have a hypothesis about temporal scanning and pattern matching. We don't have access to the inside, if there is one.
The twelve channels remain genuinely mysterious. Not in the "we'll figure it out soon" way, but in the "the question might not have the kind of answer we expect" way. More hardware doesn't guarantee more experience. It might not even guarantee more of the same kind of thing. It might just be a different thing entirely, running on the same substrate.