Children's play

Why do children spend so much of their time on activities with no obvious purpose?

Give a child a free hour and no instructions and she will not rest. She will build a tower in order to knock it down, chase a friend who is not running anywhere in particular, or spend twenty minutes being a dog. None of it feeds her, warms her, or wins her anything.

That is easy to file under "children are children". But the same behaviour turns up in kittens, ravens, otters, young horses and juvenile rats, and evolution is not sentimental about time. A young animal at play is burning fuel it could be growing on, risking a sprain, and making itself conspicuous while it does. So the interesting question is not why play is enjoyable. It is why something so plainly costly is so hard to find a mammal that does not do it.

Start by taking the cost seriously. Play consumes energy, produces injuries, and pulls attention away from watching for danger — young animals have been recorded being taken by predators mid-game. And notice how the body itself rates it: play is among the first behaviours to disappear when a young animal is hungry, cold, or ill. That is the accounting of a luxury item. What should we make of a luxury item that persisted across nearly every mammal lineage?

Before answering, it helps to say what counts, so we are not just gesturing at "fun". Researchers converge on a cluster of criteria: the behaviour is incompletely functional in the context it appears in, it is done for its own sake, it differs from the serious version by being exaggerated, repeated, or left unfinished, and it happens only when the animal is fed, safe and unstressed. That last condition is the informative one. Play is what an animal does with surplus — which is precisely why it can afford to be expensive, and why it needs a payoff to have survived.

So what might it buy? The obvious candidate is motor skill: practise now, be competent later. But look closely at what young animals actually rehearse. They leap past their landing, spin until they topple, throw themselves off balance on purpose. If straightforward practice were the point, you would expect the successful movement drilled over and over. Instead you see the loss of control drilled over and over. One proposal reads play as training for the unexpected — deliberately putting yourself into awkward positions while the stakes are low, so that recovery is practised, not just performance. That fits the exaggeration rather than explaining it away.

Now the social side, where the evidence is sharpest. Watch two young animals play-fight and you find something a real fight never contains: the stronger one pulls its bite, hands over the top position, and lets itself be pinned. Dogs bow, chimpanzees wear a relaxed open-mouthed face, children announce the frame outright — this is pretend. Why would a rehearsal for combat include letting the other side win? It would not. But it makes complete sense if the point is keeping a partner willing to continue, because a partner who is hurt or humiliated simply leaves. Self-handicapping is the giveaway that play fighting is not practice fighting. It is practice at negotiating.

And when it is removed? Rats raised with companions who are drugged out of playing — socially housed in every other respect — grow up clumsy at reading and answering another rat's signals. Something was being calibrated, and it was not muscle.

Two refinements, and then an honest limit.

First, "play" is not one behaviour. Tumbling, poking at objects, chasing, and pretending to be a dragon are different activities with different likely payoffs, and evidence for one is not evidence for another. Second, the argument above is evolutionary and comparative: it explains why the capacity persisted across species. That is not the same as knowing what any particular hour of any particular child's play is doing for her.

And that gap is where the honesty is required. The general case — play is costly, near-universal, carefully regulated, and its absence leaves marks — is strong. The specific case is much weaker. Reviews of pretend play in children have found the evidence that it causes measurable gains in language, reasoning or self-control to be thin: small studies, weak controls, and experimenters who knew which group they were rating. "Play-based learning" advocacy routinely outruns that. Human deprivation evidence is worse still, because children who lack play usually lack a great deal else, so the missing ingredient cannot be isolated.

There is also a quiet trap in the functional argument itself. If play is defined partly by being spontaneous and done for its own sake, then an activity engineered to deliver an outcome is drifting out of the category while claiming its benefits.

• Why play declines sharply after adolescence in most species, and what that suggests about its function.
• Whether the recorded fall in unsupervised childhood play has measurable consequences, and how one would ever tell.

Every term the collection defines is gathered in the glossary.