Infant night waking

Why do human infants keep demanding night feeds long after their nutritional needs would require them?

By nine or ten months a healthy infant is eating solid food, growing steadily, and can plainly go ten hours in the daytime without a feed. The same infant wakes three times a night wanting one. Night-weaning a baby of that age does not, as a rule, stall its growth — which tells us the feeds are not required. So the behaviour has outlived the need that supposedly produced it.

That is an odd thing for a costly behaviour to do. It exhausts the person the infant most depends on. If we assume, as it is natural to, that the infant's interests and the parent's interests point the same way, the pattern looks like a malfunction. Perhaps the assumption is the thing to question.

First, correct the observation, because it is wrong in a way that matters. Infants do not wake three times a night. They wake far more often than that — and so do adults. Sleep runs in cycles, roughly fifty to sixty minutes in an infant against something nearer ninety in an adult, and every cycle boundary brings a brief arousal. That is normal architecture, not a defect.

So what varies between the baby who "sleeps through" and the one who does not? Video studies of infants left alone with a camera found that trained sleepers still surfaced at those boundaries. What they did differently was resettle without calling out. The parent's experience of "waking" is really the experience of being signalled. That is a different thing to explain, and a much more interesting one, because a signal is directed at someone.

Now ask the question selection would ask. Whose interests does the signal serve? A parent's fitness is spread across all the children she will have. An infant shares half its genes with each future sibling and all of them with itself, so it values its own share of care at roughly twice the weight the parent's optimum assigns it. Wherever a resource can be pushed past the parental optimum, we should expect offspring that push — and parents that resist. This is Trivers's parent-offspring conflict, and it predicts that weaning, in every mammal examined, is a fight rather than a handover.

Night feeding has a further twist that makes it a particularly sharp instrument. Frequent suckling suppresses ovulation, and nocturnal suckling appears especially effective at doing so. An infant that keeps its mother feeding through the night is, incidentally, postponing the arrival of a competitor for her care. David Haig proposed this explicitly as the function of persistent night waking — and I want to be plain that it is a hypothesis with suggestive support rather than a demonstrated mechanism. It is hard to test directly, and I do not know that it is true.

What the weaker, safer version of the argument gives us is still enough: the infant's optimum and the parent's optimum are not the same, so there is no reason to expect the infant to stop asking at the point the parent stops needing to give. The question assumed a shared ledger. There isn't one.

Three refinements hold this together.

The first is that "signalled waking" and "waking" come apart, and almost every argument about infant sleep is really an argument about the first while sounding like it is about the second. Sleep training does not stop arousals. It changes what an infant does at one, by changing what it expects to follow. That is why it works, and also why it does not make the infant sleep more.

The second is that this is not a rival to attachment. A responsive parent is not making a mistake that a firmer one avoids; responsiveness is one side of a negotiation in which both parties have real stakes. What the conflict view adds is that the infant's demand is not a readout of the infant's need, so a parent cannot infer the second from the first — which is exactly the inference that makes night waking feel like a reproach.

The third is a caution about cost. Cross-cultural work finds that where mother and infant sleep together and feed on cue, the frequency of night feeding stays high but is not reported as a problem. That matters two ways: it is evidence against the idea that Western parents have created the waking by mishandling it, and it is a reminder that "the problem" is partly the arrangement of the household rather than the behaviour of the child.

Here is what would falsify the account. It claims the persistence is about the divergence of interests, not about need. So night waking should not track nutritional demand — and it does not, since it persists after solids and often intensifies around the end of the first year rather than fading with it. It also claims the observable behaviour is signalling, so an intervention that changes only the response should change the signalling while leaving the arousals intact. If the video studies had found trained infants sleeping unbroken, the whole reframing would collapse.

• How weaning conflict resolves in other mammals, and what sets its timing.
• Why genomic imprinting predicts opposite pressures from an infant's maternal and paternal genes.

Every term the collection defines is gathered in the glossary.