Skip to content
NatureHQ

Behaviour and cognitionbehaviour

Group defence

A herd need not exist for mutual protection. It can be what happens when every animal selfishly tries to put somebody else on the outside.

“Safety in numbers” is at least three separate mechanisms. Your share of the risk falls, the predator finds targeting harder, and you can put another animal between yourself and it — and only the middle one requires everyone to look alike.

The phrase covers so much that it explains nothing until it is taken apart. Dilution is the simplest: if a predator takes one animal, being one of a hundred is better than being one of ten, and this is arithmetic requiring no behaviour at all. The confusion effect is different and more interesting — a predator tracking one target among many similar moving ones does worse, and its capture rate falls as the group grows. And Hamilton’s selfish herd is different again: each animal reduces the area within which it is the nearest target by moving towards others, and a group forms as the by-product. That last one deserves emphasis because it removes the assumption most people bring. A herd need not exist for mutual protection. It can be what happens when every individual, entirely selfishly, tries to put somebody else on the outside. Reading a flock as a community is optional; the geometry does not require it. The confusion effect has the sharpest experimental result, and it comes with a condition that changes how the whole thing should be read. Bass hunting shoals of minnows became markedly less successful as shoals grew — until a few individuals were dyed a different colour. The odd ones were then taken far out of proportion to their numbers, and, crucially, the predator’s success against the entire shoal recovered. Confusion is not about numbers; it is about being indistinguishable, and one conspicuous individual degrades the protection for everybody. That explains the striking uniformity of shoaling fish, and it explains why being different in a group is so heavily penalised. What grouping costs is worth keeping in view. More animals in one place compete for the same food, are more conspicuous collectively, and transmit disease faster. Grouping is a trade like every other defence on this site.

Early coverage · 58% complete · reviewed 2026-09-03

What this page covers

Grouping under predation risk occurs in fish, birds, mammals and insects. The mechanisms have been separated experimentally mostly in fish, because a tank permits manipulations a savanna does not.

Often confused with: Cooperation, when the selfish herd requires none and produces a group anyway; One mechanism, when at least three operate and make different predictions; A free benefit, when grouping costs food competition and disease

Quick facts

Three mechanisms
Dilution, confusion, and selfish-herd geometry
Confusion has a condition
It requires looking like everybody else — oddity destroys it
No cooperation needed
Selfish positioning alone produces a group
Not free
Food competition, collective conspicuousness, faster disease spread

Three mechanisms wearing one phrase

They have different requirements and make different predictions.

A group helps in three separate ways: your share of the risk falls, targeting gets harder for the predator, and you can put somebody between yourself and it. Only the second needs the group to look alike.

Established

Specialists would state this without hedging. Multiple independent lines of evidence agree.

Anti-predator benefits of grouping comprise distinct mechanisms: numerical dilution of per-capita risk; the confusion effect, in which predator targeting accuracy declines with the number of similar moving targets; and selfish-herd geometry, in which individuals reduce their own domain of danger by positioning relative to others. These have different requirements and different predictions.

Who this applies to
Documented across fish, birds, mammals and insects.
Studied in
Animalia
Why we rate it this way, and what the caveats are
EstablishedHigh confidence

Each mechanism has independent support, and confusion in particular has been demonstrated experimentally with a manipulation — introducing odd individuals — that isolates it from dilution.

How far it can be extended

Dilution, confusion and selfish-herd effects have each been demonstrated independently in several groups.

Caveats

  • Grouping also carries costs — competition for food, increased conspicuousness of the group, and faster disease transmission — so it is not a free benefit.
  • The mechanisms are not exclusive and usually operate together, which is why field observations rarely separate them.

Still unanswered

  • How much each mechanism contributes in any particular wild system, which requires manipulations that are difficult outside a tank.

Last reviewed 2026-09-03

The evidence (3 studies)
What each mechanism needs in order to work
MechanismHow it helps youWhat it requires
DilutionYour share of the attacks falls as the group growsNothing — it is arithmetic
ConfusionThe predator targets and tracks less accuratelyThat you look and move like the others
Selfish herdSomebody else is nearer to the predator than youOnly that you position relative to others
Collective vigilanceSomeone else spots the predator firstThat alarm is detectable by the group

The fourth row is a real fourth mechanism and is included with a caution: more eyes do detect predators sooner, but individuals in large groups also relax their own vigilance, so the benefit is partly spent as soon as it is gained. That is a general pattern in this area — the advantages of grouping are rarely as large as counting them suggests, because behaviour adjusts to them.

Diagram

Two reasons a group helps

Only one of them needs everyone to match.

Two reasons a group helps, and only one needs everyone to matchDilutionYour share of the attacks falls. Arithmetic; nothing else required.ConfusionTargeting degrades — until one individual stands out.The odd one is taken, and the whole group loses its protection.Which is why shoaling fish are so strikingly uniform in size, colour and movement.
The same explanation in words

Two rows of circles representing prey in a group. In the dilution row all circles are identical: your share of the attacks simply falls as the group grows, which is arithmetic and requires nothing of the animals. In the confusion row one circle is a different colour: predator targeting degrades among many similar moving targets, but the odd individual is taken far out of proportion and its presence restores the predator’s success against the whole group. This is why shoaling fish are so strikingly uniform in size, colour and movement — standing out is penalised, and it damages everybody else’s protection too.

Confusion works only if you look like everyone else

And one conspicuous individual degrades the protection for the whole group.

A predator’s success drops sharply against a large shoal — until a few individuals look different. Odd ones are taken far out of proportion, and their presence restores the predator’s success against the whole group.

Well supported

Good evidence backs this, though some details remain open.

Predator capture success declines with prey group size, an effect abolished by the presence of visually distinguishable individuals. Odd-phenotype individuals are attacked disproportionately, and their presence increases predator success against the group as a whole, indicating that the confusion effect depends on phenotypic uniformity rather than on numbers alone.

Who this applies to
Demonstrated experimentally in a fish predator–prey system; the oddity effect is documented more widely.
Studied in
Micropterus salmoides, Notemigonus crysoleucas, Animalia
Why we rate it this way, and what the caveats are
Well supportedHigh confidence

A direct experimental manipulation isolating the mechanism, with a clear and large effect and an obvious interpretation.

How far it can be extended

The mechanism depends on general properties of visual target selection, and oddity effects have been reported across several predator–prey systems.

Caveats

  • Demonstrated in a tank where prey cannot escape the arena, which may exaggerate the effect relative to open water.
  • Dyeing prey to make them distinguishable is a stronger manipulation than most natural variation.

Still unanswered

  • How strongly this selects against natural variation in wild shoals, which would predict unusually low phenotypic diversity in shoaling species.

Last reviewed 2026-09-03

The evidence (2 studies)

How we know

Dyeing a few fish blue

Predators do worse against larger shoals. Is that because there are more prey to share the risk, or because many similar targets are genuinely harder to catch?

Largemouth bass hunted shoals of silvery minnows at a range of shoal sizes. In a second manipulation, a small number of individuals within otherwise uniform shoals were dyed blue, making them visually distinguishable without changing the shoal’s size.

What happened

Capture success fell sharply as shoal size increased. Introducing a few odd-looking individuals reversed much of that: the odd fish were taken far out of proportion to their numbers, and the predator’s success against the shoal as a whole recovered.

What it shows

That the confusion effect is real and separable from dilution, and that it depends on the prey being indistinguishable rather than merely numerous. It also explains the strong uniformity of shoaling fish: standing out is heavily penalised, and it damages everybody else’s protection at the same time.

What it does not show

One predator and one prey species in a tank, where prey cannot leave the arena — which probably strengthens the effect relative to open water. Dyeing individuals is also a stronger manipulation than most natural variation within a shoal.

The controls — what makes this evidence rather than a story
  • Shoal size varied first, establishing the baseline relationship between group size and capture success.
  • Oddity introduced without changing numbers, which separates confusion from dilution — dilution predicts no effect from a few dyed fish.
  • Uniform shoals of the same size as comparison throughout.

From Oddity and the confusion effect in predation

The consequence is a strong selective pressure towards uniformity in animals that rely on this, and it explains something otherwise puzzling about shoaling fish: they are remarkably alike, in size, colouring and movement. Standing out in a shoal is not merely unhelpful, it is close to fatal — and it damages everybody else’s protection at the same time, since it restores the predator’s ability to lock on.

From individual rules to collective motion

Real predators, hunting prey whose movement rules could be edited directly.

The most elegant demonstration that predation shapes collective motion used virtual prey. Live sunfish were shown computer-generated prey projected into their tank, whose individual movement rules — how strongly each is attracted to neighbours, how strongly it aligns with them — could be set precisely. Prey caught by the fish were removed from the reproducing population, so the predator itself did the selecting across generations of virtual prey. The population evolved towards greater attraction and alignment: coordinated collective motion, produced by predation acting on individual rules.

The design closes a gap that observing real shoals cannot. In a real shoal you can see the collective pattern and infer the rules; here the rules are set directly and the pattern is the outcome, with a genuine predator supplying the selection.

Related

  • Mobbing

    The version where the group approaches the predator

  • Predation

    The targeting problem groups exploit

  • Zebra

    Where the confusion explanation was proposed and failed

  • Collective behaviour

    The local rules, and what they produce at flock scale

The research behind this page

6 studies, newest first. Each one has a page explaining what it found and what it could not show.

2012Science

Predatory fish select for coordinated collective motion in virtual prey

Selection by real predators drove the virtual prey towards greater attraction and alignment, producing coordinated collective motion.

2011Animal Behaviour

Propagating waves in starling, Sturnus vulgaris, flocks under predation

The dark bands are waves of local density and orientation change propagating across the flock, associated with predator attack, and travelling faster than individual birds move.

2008Proceedings of the National Academy of Sciences

Interaction ruling animal collective behavior depends on topological rather than metric distance: evidence from a field study

The angular structure of neighbour positions was anisotropic out to a roughly fixed number of nearest neighbours — approximately six or seven — rather than out to a fixed distance, and this held as flock density varied.

2005Nature

Effective leadership and decision-making in animal groups on the move

A small proportion of informed individuals was sufficient to guide a large group accurately, with the required proportion falling as group size increased.

1986Animal Behaviour

Oddity and the confusion effect in predation

Bass capture success fell sharply as shoal size increased, and rose again when odd-looking individuals were present.

1971Journal of Theoretical Biology

Geometry for the selfish herd

Individuals reducing their personal risk by moving towards others produce an aggregated group as a by-product.

This page is a stop on a longer route

A guided journey reads several subjects in a deliberate order, with an argument for why one follows another. You can join in the middle.

Where to go from here

Each of these follows from something on this page — a relationship in the evidence, a claim people ask about, or the next mechanism along.

How complete this page is, and what it is still missing

NatureHQ publishes its own gaps. This page is at 58% completeness against what we would call a finished subject, and was last reviewed on 2026-09-03. It carries 5 claims and answers 7 mapped search questions.

  • no research from the last few years is attached — check for newer work
  • no popular claim about this subject has been checked yet
  • The experimental separation of mechanisms is largely from fish in tanks, where escape options are limited.
  • How much each mechanism contributes in wild systems is generally unknown.
  • The costs of grouping are noted rather than quantified.