Mantis shrimp have sixteen types of photoreceptors. Humans have three. The obvious implication: they see color at a resolution we can't approach. Sixteen sensors against three — the comparison seems to settle the question before it starts.
It doesn't. When researchers tested mantis shrimp on standard wavelength discrimination tasks — show them two colors, ask if they're different — the shrimp needed colors to differ by at least 15 to 25 nanometers before they could tell them apart. Humans can distinguish colors that differ by 1 to 8 nanometers. We outperform them significantly on the task that "having more photoreceptors" is supposed to solve.
The 2014 study in Science (Thoen, How, Chiou, and Marshall) found that mantis shrimp don't appear to use color opponency — the process by which human visual systems compare signals across cone types to derive color information. Our visual cortex subtracts: long-wavelength minus medium-wavelength, medium-wavelength minus short-wavelength. This comparison is what gives us fine discrimination. Two very similar wavelengths produce slightly different ratios; we detect the ratio, not the absolute activation.
Mantis shrimp don't seem to compare. Each of the twelve color-dedicated receptor types appears to function independently. A photoreceptor fires when light in its band hits it; the signal goes forward; that's the answer. No mixing, no subtraction, no ratio. The researchers called it "interval decoding" — the brain reads whichever receptor is firing most strongly and classifies the color accordingly. One receptor class corresponds to one color category. Twelve receptors, twelve bins.
This is why more inputs produced worse discrimination: the shrimp's visual system isn't solving the same problem ours is. It isn't trying to resolve the spectral space finely. It's trying to label objects quickly. Each photoreceptor is a lookup: is this thing the wavelength of a mantis shrimp's shell marking? The wavelength of a particular prey species' body? The wavelength that signals a toxic animal? Fire the right receptor, get the right label, decide. No comparison required.
Justin Marshall compared it to satellite remote sensing. Satellites that need to classify land cover — forest vs. farmland vs. water — don't analyze every nuance of spectral reflectance. They bin pixels into categories using lookup tables. You don't need fine discrimination when you need fast classification. The mantis shrimp eye is built for the same priority.
There's a detail from the 2015 follow-up that stuck with me. Researchers found that primate visual cortex neurons — the ones that give us our conscious color experience — show "interval decoding with a winner-take-all rule," which is exactly the computation mantis shrimp do in their photoreceptors. The difference isn't the algorithm. It's where in the visual system it runs. We compare first, then classify; they classify at the receptor level without comparing. Same endpoint, different architecture.
The assumption underneath "more receptors = better vision" is that better means finer. That vision is a measuring instrument, and more inputs give more precise measurements. But the mantis shrimp build something different: a pattern-matcher, fast and parallel. Sixteen types of photoreceptors cover an extraordinary spectral range — deep ultraviolet through far red — and they tile that range in twelve labeled bins. The resolution isn't the point. The coverage is. And the speed.
A mantis shrimp punches with the acceleration of a bullet. It needs to know, in a fraction of a second, whether the thing in front of it is prey or rival or risk. There isn't time to run opponent subtraction through a visual cortex. The label is the computation.
More inputs doesn't mean more resolution. It means more of something — and what that something is depends entirely on what the system does with them.