Social behaviour and migration

A frozen pond seen in cross-section: under the ice, swallows huddled together among the reeds; in the pale sky, above snow-covered hills, a line of swallows moves away along a dotted line. — An old belief: swallows supposedly spending the winter under water, among the reeds of a pond; in the sky, the real answer, a line of swallows heading off southwards. Erosophia illustration, réalisée avec l'aide d'une IA

For a very long time, nobody knew where birds went in winter.

They left. They came back. In between, nothing.

Migrating, and living in groups

A migration is a regular, seasonal movement between two areas used for different purposes — most often an area where animals breed and an area where they spend the harsh season. It is neither wandering nor flight: the journey is repeated, year after year, often on the same dates and along the same routes.

The reason is almost always food. At high latitudes, summer brings great abundance and very long days in which to raise young; winter offers nothing. Leaving is costly, but staying would cost more.

The distances covered are greater than one might imagine. Every year the Arctic tern travels from the Arctic to the Antarctic and back. Tags fitted to individual birds record circuits of 50,000 to 90,000 kilometres a year: this is the longest regular journey known in any animal, and it is made by a bird weighing around a hundred grams. The East African wildebeest follows the rains in a loop. The humpback whale feeds near the poles and gives birth in the tropics, where it fasts. The monarch, for its part, takes several generations to complete its cycle: the individual that sets off again in spring is not the one that arrived in autumn.¹

To find their way, these animals do not rely on a single sense but on a whole array of cues: the position of the Sun, that of the stars, the Earth’s magnetic field, smells, the shape of coastlines and of the terrain. When one cue is missing, the others compensate — which explains why it was so difficult to establish which one mattered.²

Living together

Many of these movements are made in groups, and a group is more than the sum of its individuals. It brings three measurable advantages: more eyes to spot a predator, less individual risk of being the one that gets caught, and sometimes less effort — birds flying in formation save energy by making use of the wake of the birds ahead.³

It also comes at a cost: food has to be shared, and diseases spread faster.

One point needs to be made straight away, because everything else on this page depends on it. A group that moves in a coordinated way does not necessarily have a leader, nor does it necessarily follow a plan. A flock of starlings rippling across the sky obeys no one: each bird adjusts its position to its six or seven immediate neighbours, and the overall shape arises from these local adjustments. What, seen from outside, looks like a collective decision is merely a simple rule, applied by each individual.

This is precisely the trap that this page will examine: we have words — instinct, leader, decision — that give the impression of explaining, when all they do is name.

What could not be known

The problem was not a lack of curiosity: the question has been documented since Antiquity. The problem was the complete absence of instruments for observing over long distances.

For want of any means of measurement, the gap was filled with explanations: Aristotle believed that certain swallows, instead of migrating, hibernated in holes, and that certain birds changed into other species (redstarts becoming robins in winter); in the sixteenth century, people still imagined them spending the winter under water, among the reeds. It took the invention of ringing — the first numbered rings were fitted in Denmark in 1899 — and then the emergence of Argos tags and satellite tracking, to map their real routes.⁴

The fundamental lesson remains the same:

The question was not too difficult. It lacked an instrument.

As long as something cannot be measured, a question remains open — and speculation takes the place of observation.

The word “instinct”

Here is a word that needs examining, because it is used so widely:

“The bird migrates by instinct.” “It’s maternal instinct.” “Salmon swim up the river by instinct.”

The question is what these sentences actually contribute.

They name. They do not explain.

To the question how does a bird find its way?, the answer by instinct teaches nothing. It supplies a tautological label in place of a mechanism.

Yet the real mechanisms exist and are still being brought to light: magnetoreception, celestial cues, circannual clocks and social learning.⁵

This site has listed some thirty words that produce a false belief. This one is of a particular kind: it does not mislead about the nature of a fact, it gives the illusion of having explained it.

Note on rigour: In modern ethology, the term “instinct” retains a restricted technical use to refer to fixed motor patterns (Fixed Action Patterns). It should not, however, be confused with a causal explanation of behaviour.

The word “instinct” was disowned by those who used it

Here is a fundamental chapter in the history of the behavioural sciences.

In the 1950s, the study of animal behaviour rested largely on a strict dichotomy between “innate” and “learned” behaviours.

In 1953, the psychologist Daniel Lehrman published a thorough critique. His central argument is decisive:

Classifying a behaviour as “innate” says nothing about how it developed. The word does not describe a mechanism: it implicitly declares that there is nothing left to analyse.⁶

The reversal

In 1963, one of the founders of modern ethology — Niko Tinbergen, a future Nobel laureate — wrote a sentence that deserves to be read carefully.

What he considers “heuristically harmful” is not the distinction itself: it is calling a behaviour “innate” at the end of a series of eliminations — one learning process has been ruled out, then another, so whatever remains must be innate.⁷

In other words: he does not condemn the word, he condemns the way it is arrived at. Declaring that something is innate because nothing has been found that could have taught it amounts to drawing a conclusion from an absence.

Put another way: framing the question in these terms blocks research.

Tinbergen came round to Lehrman’s critique. Konrad Lorenz maintained his position. This debate profoundly restructured the discipline.

What can legitimately be said

The methodological consequence is precise:

Only conclusions formulated negatively are sound. One can establish that a given factor plays no part — for example, that a bird reared in acoustic isolation still produces, in some species, the basic structure of its species’ song.⁸

But one cannot conclude that a behaviour is “free of ontogeny”. All behaviour arises in the course of development: the environment in the womb or in the egg, maternal steroids, early sensory stimulation.

A classic example: in many bird species, the overall coordination of a courtship display is functional without prior learning — but the detail and fine-tuning of its performance are shaped by social experience.

Neither entirely innate, nor entirely learned. The question itself was badly posed.

How many meanings in a single word?

In their critical review of ethological concepts, the researchers Matteo Mameli and Patrick Bateson (2006) listed twenty-six different definitions of “innateness” in the scientific literature, before singling out eight with distinct properties.⁹

Their argument is unanswerable: as long as it has not been proved that these eight properties are correlated with one another — which the data contradict — the word is a source of ambiguity.

Two people can use the word “innate” while thinking of completely different properties without realising it.

The six meanings and the writing rule

As early as 1991, Patrick Bateson pointed out that a behaviour described as “innate” could mean:

Present at birth;
Caused by a genetic difference;
Shaped by natural selection;
Invariant during development;
Shared by all members of a species;
Not learned.

These six criteria do not necessarily coincide. A behaviour can be present at birth without being shared by the whole species, or be shaped by natural selection while still requiring learning.

Bateson draws a simple conclusion from this: classifying behaviour into “innate” and “not innate” merely causes confusion.¹⁰

Comparison with a functional term

The word “reflex” rests on an identified neural arc: receptor, afferent nerve, spinal centre, motor neuron, effector. It can be observed, measured and timed.

“Instinct” does not refer to a circuit. It is a label stuck on something not yet explained.

A warning about anthropomorphism

The analysis of collective behaviour frequently suffers from anthropomorphic projections: leader, hierarchy, discipline, society.

The history of the concept of the “alpha wolf” illustrates this bias: based on observations of unrelated captive animals, this invalid model took more than fifty years to be corrected in popular culture (the researcher L. David Mech having invalidated his own initial theory himself).¹¹

Visible order does not prove there is a leader

Faced with a flock of starlings, a shoal of fish or a column of ants, the mind immediately looks for someone in command. Yet no individual is in charge. These patterns emerge from very simple local rules — keeping a distance from one’s immediate neighbours, matching one’s speed and direction to theirs, dodging an approaching predator.¹²

How it was measured

This was known from computer models, but it had never been checked on real birds. A team did so by photographing, from a rooftop in Rome, flocks of several thousand starlings — with two cameras set some distance apart, which makes it possible to reconstruct the position of each bird in three dimensions.¹³

Result: each starling adjusts its trajectory to six or seven neighbours.¹⁴

Then comes the detail that changes everything. These six or seven are not the birds located within so many metres. They are the six or seven nearest, whatever the distance between them.¹⁵

The nuance seems slight. It is decisive. In a tight flock, six neighbours fit within a metre; in a spread-out flock, they have to be sought much further away. The bird counts neighbours, it does not measure a distance.

The authors explain why this matters: when a falcon attacks, the flock stretches and scatters. A rule based on distance would then leave some birds alone, out of reach of any neighbour — the group would tear apart. A rule based on number holds firm. Counting one’s neighbours rather than measuring them is what keeps the flock from breaking up at the very moment it is attacked. It is the same mechanism as the one described in our page on chance and data (9.6), where individual behaviours added together produce a statistical regularity.¹⁶

The methodological rule remains strict: settle a question only where empirical research settles it, and document the state of uncertainty where it does not.

Note 1

The Baltic terns completed a 50,000 km annual migration circuit, exploiting ocean regions of high productivity

Alerstam, Thomas et al., “Hypotheses and Tracking Results about the Longest Migration: The Case of the Arctic Tern”, Ecology and Evolution, vol. 9, no. 17, 2019, p. 9511-9531

The total travel distance in the course of the non-breeding period was 90,000 ± 2000 km, which substantially exceeds previous estimates for this species.

Fijn, R. C., Hiemstra, D., Phillips, R. A. et van der Winden, J., “Arctic Terns Sterna paradisaea from the Netherlands Migrate Record Distances across Three Oceans to Wilkes Land, East Antarctica”, Ardea, vol. 101, no. 1, 2013, p. 3-12