What Mantis Shrimp Teach Us About Selective Attention Online

The mantis shrimp is often cited as a marvel of biological engineering. It has sixteen types of photoreceptors — compared to three in the human eye — and can see wavelengths from deep ultraviolet to far infrared, including polarised light that no human optical system can detect. By the numbers, it should experience the richest visual world of any animal on the planet.

But a 2014 study by Hanna and Marshall found something counterintuitive: when tested on colour discrimination tasks, mantis shrimp performed worse than animals with far fewer photoreceptor types. Pigeons and bees, with only four or five types respectively, discriminated colours more finely in controlled experiments.

The reason, as best researchers understand it: the mantis shrimp visual system isn’t built for comparing and analysing. It’s built for categorising and deciding — fast. Each photoreceptor type acts less like a continuous sensor and more like a binary flag. Does this wavelength match the signature I’m looking for? Yes or no. The result is a system optimised for speed and pattern matching, not comprehensive evaluation.

The mantis shrimp doesn’t ponder. It identifies and acts.

The cost of comprehensive perception

There’s a reason most animals don’t evolve richer sensory systems than they need for survival. Processing takes energy. Comprehensive sensory input creates decision overhead. A predator that stops to fully analyse every visual signal it receives is slower than a predator that acts on the most decision-relevant signals alone.

The mantis shrimp’s solution is elegant: use the full input spectrum, but classify rather than measure. Build a fast lookup table, not a detailed model.

Human attention works on a similar constraint, even if the implementation is different. Working memory is limited — estimates cluster around four chunks of information held simultaneously. Comprehensive attention to everything in the visual field is not possible; the brain selects, filters, and discards constantly. The question is not whether to be selective, but what the selection criteria are.

In natural environments, those criteria evolved over millions of years: motion signals danger, colour signals ripeness, social cues signal threat or opportunity. The signals that grabbed attention were, on average, the ones worth attending to.

Digital environments are different in a specific way: the signals are engineered.

Engineered salience vs. environmental salience

Notification badges, red indicators, unread counts, algorithmic feeds — these are all designed to trigger the same attention-capture mechanisms that evolved to respond to genuine environmental signals. A red circle on an app icon activates the same neural pathway as a flash of movement in peripheral vision. The feeling of urgency is real. What produced it is not a predator or an opportunity; it’s a design decision made by someone who wants you to open the app.

The mantis shrimp’s photoreceptor system is fast and efficient because the signals it evolved to detect — the colour patterns of prey, rivals, and mates — were reliable indicators of things that mattered. Its lookup table matches reality.

A human attention system running the same fast-lookup strategy on engineered digital signals gets a different outcome: fast response to things that are designed to feel urgent rather than things that are actually urgent. The speed benefit remains; the accuracy drops.

What selective attention actually looks like in practice

The mantis shrimp doesn’t avoid seeing the full spectrum — it processes what it sees selectively, at speed, based on what’s decision-relevant. The sixteen photoreceptor types aren’t a liability; they’re a sensing array feeding a fast classifier.

The equivalent for managing digital attention isn’t to see less — it’s to change which signals reach the classifier in the first place, and to slow down the ones that do.

A few patterns that work with selective attention rather than against it:

Batch rather than interrupt. Email and messages don’t need real-time processing. Moving from a reactive mode (respond when the signal arrives) to a scheduled one (check at defined times) changes the signal type from an interrupt to a routine. The mantis shrimp isn’t checking the reef for threats while it’s focused on a specific prey item; it hunts, then scans.

Make the relevant signal the prominent one. If a notification badge on every app icon gets equal visual weight, every badge competes for attention equally. Reducing badges to the apps that carry genuinely time-sensitive information makes the remaining signals more reliable indicators — closer to the mantis shrimp’s evolved lookup table, further from the engineered noise floor.

Set environmental limits before the session, not during it. The time to decide how long you want to spend on a site is before you open it, not after you’re mid-scroll. A decision made in advance, enforced by the environment, removes the moment-to-moment willpower cost. The mantis shrimp doesn’t negotiate with itself about whether to respond to a prey signature; the classification happens automatically. An environment that automatically ends a session at the limit is doing the same thing.

The shrimp’s visual system is fast and efficient because it doesn’t ask more of itself than it needs to. Sixteen photoreceptor types in service of fast classification. Not comprehensive analysis — selective, calibrated response.

The goal isn’t to stop noticing what’s on screen. It’s to change what produces the signal that captures attention in the first place.