Twelve photoreceptor classes did not give one mantis shrimp species fine wavelength discrimination. The barcode model explains why, but newer work points to a hybrid visual system.
By Natural History · Edited by Lachlan Brown
Published · How we edit
The mantis shrimp’s reputation as the owner of nature’s richest color vision rests on an irresistible piece of arithmetic. Humans have three cone classes for daytime color vision. Some mantis shrimp have 12 photoreceptor classes. From there it is easy to imagine an animal seeing millions of colors that simply do not exist for us.
That arithmetic does not describe perception. A 2014 paper in Science, led by Hanne H. Thoen at the University of Queensland, tested how well one mantis shrimp species could separate nearby wavelengths. The animals were unexpectedly poor at it. This is one experiment on a particular discrimination task, not the final word on the subjective visual world of every mantis shrimp. It nevertheless punctured the popular assumption that adding receptor types automatically adds ever-finer color distinctions.
The team studied Haptosquilla trispinosa, a small mantis shrimp found on shallow reefs. The animals were trained to associate light of a particular wavelength with food. In the test, each shrimp emerged from its burrow and chose between two optical fibers. One displayed the rewarded wavelength. The other displayed a wavelength that the researchers moved progressively closer to it.
When the two wavelengths were well separated, the shrimp could choose the rewarded one. As the separation narrowed, their success fell. Across the tested spectrum, the animals generally needed a difference of about 15 to 25 nanometers for reliable discrimination. Human wavelength thresholds vary with color and conditions, but can fall to roughly 1 to 8 nanometers. Animals with far fewer receptor classes can therefore make finer distinctions in this kind of test.
That result is easy to overstate in the opposite direction. It did not show that mantis shrimp are color-blind or that their complex eyes are useless. It measured the ability to distinguish similar, narrow-band lights in a laboratory choice task. A reef contains broad and mixed spectra, changing illumination, polarization, movement and recognizable objects. Those are related problems, but not the same problem.
A photoreceptor class is an input, not a color counter. Human color vision begins with three broadly tuned cone classes. Their sensitivity curves overlap, so the same light stimulates all three in different proportions. Neural circuits compare those outputs. Much of the precision comes from the comparisons, not from possessing a very large inventory of receptors.
The mantis shrimp retina is organized differently. In the specialized midband crossing each compound eye, receptor classes sample narrow portions of the spectrum from deep ultraviolet to far red, about 300 to 720 nanometers in the 2014 paper. Other parts of the eye handle brightness and polarization information. Calling all 12 simply “color receptors” is convenient shorthand, but the biological system is more structured than a row of 12 human-style cones.
If the brain compared all those narrow channels conventionally, the system might be expected to support extremely fine spectral discrimination. The behavioral results did not fit that expectation. Receptor count alone says which signals are available at the retina. It does not say how those signals are combined, how many perceptual categories result, or what the experience feels like to the animal.
Thoen and colleagues proposed a different scheme. Each narrow receptor channel might act as a labeled spectral bin. Rather than calculating a precise hue from comparisons among several channels, the system could identify whichever channel responds most strongly. A light falling within one band would receive one label, while a slightly different light could receive the same label until it crossed into the next band.
The eyes help explain the scanning part of the metaphor. A mantis shrimp can move its eyes independently, and the specialized midband samples a relatively thin strip of the scene. Scanning eye movements can carry that strip across objects. The 2014 authors proposed temporal signaling combined with those movements, producing color recognition rather than the fine discrimination familiar from conventional opponent systems.
A later analysis in i-Perception framed the idea as interval decoding with a winner-take-all rule: the most strongly responding channel supplies the label. “Barcode reader” is a useful journalistic analogy because both systems classify a pattern while a sensor scans across it. It is not a literal account of what a mantis shrimp perceives, and the 2014 experiment did not record the full chain of neurons carrying out the proposed computation.
Why build 12 narrow channels if the result is coarser discrimination? One possibility is that the eye performs some of the sorting before information travels deeper into the nervous system. That could reduce the comparisons required later. For a compact brain making rapid choices on a crowded reef, a fast label such as “food,” “rival” or “shelter” may sometimes matter more than separating two nearly identical spectral shades.
This is the point behind the proposed tradeoff. Fine discrimination extracts small differences, but computation takes wiring, neurons and time. Early binning could make recognition efficient. The follow-up authors argued that hunting pressure and a small brain could have favored a fast, hard-wired arrangement.
Still, the two-fiber study did not race a binning system against an opponent system. It measured choice accuracy as wavelengths converged. Faster recognition is therefore a proposed evolutionary benefit, not a stopwatch result. The careful claim is that this architecture could lower processing demands and favor speed, not that every mantis shrimp decision has been shown to beat the human visual system.
The original result remains important, but the pure barcode model has not closed the case. A 2022 study in The Journal of Experimental Biology found that H. trispinosa could distinguish several colors, including high- and low-saturation blue, from gray. The authors argued that this was difficult to reconcile with a system using only independent bins.
A 2025 experiment in the same journal tested color against gray under natural and colored illumination. It reported direct behavioral evidence consistent with spectral opponent processing and proposed that mantis shrimp use a hybrid of opponent comparisons and binning. Different components may serve different visual problems.
That later work does not revive the claim that 12 receptors straightforwardly yield millions more colors than humans see. It reinforces the deeper lesson. Mantis shrimp vision is remarkable because its computation may be organized differently, not because receptor arithmetic grants an unimaginably elaborate paint chart. Twelve channels describe the front end of the system. The continuing question is how those channels are compared, labeled and used as the animal scans a changing reef.
Standards
Articles on Science Blog are edited and fact-checked before publication. We use AI tools in the newsroom. See our editorial standards and masthead.
A Science Blog column
Natural History is Science Blog’s desk for the living world. Animals, evolution, and ecology, reported from the field science and the papers rather than the press releases.
Clear science for curious people
A Brown Brothers Media publication. All rights reserved © 2026.
Clear science for curious people — free PDF, then our science email.
Also available via RSS.