Sibling regression

Why does an older child start behaving like a baby again once a new baby arrives?

A three-year-old who has used the toilet reliably for a year starts wetting herself. A child with a full vocabulary begins asking for things in a babyish whine. She wants the bottle she gave up, the cot she outgrew, to be carried when she can walk perfectly well. And the timing is not subtle: it starts within days or weeks of a new baby coming home.

The instinctive reading is that something has broken — that the stress of displacement has knocked out a developmental gain. But notice what that reading assumes: that the skill is gone. Is it? She still uses the toilet at nursery, still speaks in sentences to her grandmother. What has changed is not what she can do, but what she does, and in front of whom — a different kind of fact, wanting a different kind of explanation.

Take the child's point of view, without assuming she is calculating anything consciously.

Before the baby, what did her world reward? Growing up. Dressing herself got praise, speaking clearly got answers, using the toilet got delight and often an audience. Every increment of competence was met with attention — the most valuable thing a small child can obtain from an adult.

Now something enters the household that is smaller, louder, and unable to do anything. Ask the awkward question: what does the household now visibly reward? Crying. Being unable to walk. Needing to be fed. Not because anyone decided this — an infant genuinely must be attended to — but the child does not observe intentions. She observes outcomes. And the outcome she can see is that helplessness summons a parent instantly, while competence increasingly summons nothing, because a competent child is precisely the one you can leave alone for ten minutes.

So the payoff structure has been rewritten without anyone announcing it. What would a sensible agent do when the behaviour that used to earn a reward stops earning it? Try something else. And what else is available? Here is the part that makes it click: the earlier repertoire has not been deleted. Babbling, clinging, wetting, refusing to walk — all still in her, a few months back down the road she just travelled. It is the nearest untried option, and it demonstrably works, because she can watch it working on someone else all day.

Notice that this predicts something the "broken skill" account does not: the behaviour should be targeted. It should appear where the audience is and not where it isn't, intensify when the parent is with the infant, and fade when she finds another reliable route to attention. That is roughly what observational studies report — Judy Dunn's work following families across a second birth found increases in demanding and clingy behaviour and in toileting problems, alongside, in some children, spurts of new independence elsewhere. Both directions at once make no sense as damage. They make sense as a child probing for what now works.

The developmental literature treats this as normal and, in most families, transient — weeks to a few months, not a permanent turn. That framing matters, because the alternative framing changes the parents' behaviour, usually for the worse.

Two refinements, because "she is doing it for attention" is a phrase that has done a lot of damage.

The first: for attention is a description of the contingency, not an accusation. It does not mean manipulation, and it does not mean the distress is fake. The loss is real — exclusive access to a parent has genuinely ended, routines have genuinely been disrupted, and a tired household is genuinely less patient. A child can be authentically upset and, at the same time, be expressing that upset through whichever channel the environment currently answers. Those are not competing explanations.

The second: the contingency account is a model, not the whole story. Sleep loss, a changed schedule, a mother recovering from birth, and the strangeness of a reordered home would each produce some wobble on their own. And I would be careful about reaching further — it is tempting to narrate this in attachment terms, as a rupture in the bond, but regression after a sibling's birth is a common, self-limiting behavioural pattern, and it is a considerable overstatement to read it as evidence about the security of a child's attachment. The honest claim is narrower: behaviour follows what the environment answers.

Which is why the practical implication falls out on its own. If the old behaviour stopped paying, the fix is to make it pay again — attention attached to the competent version, given predictably rather than only when extracted, and not as a reward negotiated after a wet bed. Punishing the babyish behaviour attacks the symptom while leaving the contingency exactly as it was.

• Why some children respond to a sibling's arrival with a burst of precocious independence instead.
• How the same contingency logic explains behaviour that reappears in adults under stress or illness.

Every term the collection defines is gathered in the glossary.