Your mother-in-law texts you a photo. Your baby, the one who screamed through forty minutes of rocking last Tuesday, is asleep on her shoulder. Asleep in eleven minutes. She adds a little note — he went right down! — and you look at it standing in a kitchen that still smells like the burnt toast from the morning, and something ugly and quiet moves through you.

You are the person who knows the exact weight of this child. You know which shoulder he prefers, the pitch that settles him, the seam on the swaddle that catches his chin. And he will not sleep for you.

Here is the part nobody says out loud at the pediatrician's office: your baby sleeps worse for you because of who you are to him, not in spite of it. This is not a failure of your parenting. It is one of the more elegant pieces of evidence that the attachment is working.

The infant brain is running a matching algorithm

Babies are, from very early on, extraordinary pattern detectors. Within days of birth, newborns orient preferentially toward the smell of their own mother's breast pad over a stranger's — the classic olfactory recognition work that Aidan Macfarlane and others established decades ago. They discriminate their mother's voice from another woman's voice. They track the specific rhythm of the person who holds them most.

This matters enormously for sleep, because sleep in infancy is not a switch. It is a transition, and transitions are governed by expectations. Your baby has learned — through hundreds of repetitions, most of them at 2 a.m. — that a particular constellation of smell, voice, heartbeat, and arm position predicts a particular outcome. Milk. Bouncing. Being picked up when the crying escalates. That constellation is you.

In learning terms, you have become what behavioral scientists call a discriminative stimulus: a cue that signals which behaviors will pay off. Not a comfort object. A signal. And when the signal is present, the behavior it predicts becomes far more likely to appear.

Grandma is not that signal. Grandma's shoulder carries no history. There is no learned contingency between Grandma's presence and the arrival of a feed, so the baby's nervous system does something remarkably simple: it stops running the routine and goes to sleep.

Conditioned arousal, or: why your presence wakes him up

There is a term in adult insomnia research — conditioned arousal — that describes what happens when a cue that should mean sleep has instead been paired, over and over, with effortful wakefulness. The insomniac's bed becomes a place their body braces at. The cue itself starts producing alertness.

Something structurally similar happens between a baby and their primary caregiver, particularly a breastfeeding one. If your arrival at the crib has, for months, reliably preceded a feed, then your smell drifting into the room does not soothe. It announces. It cues the anticipatory arousal that precedes eating — a real, measurable physiological readying, not a mood.

This is why the classic scene happens: you tiptoe past the door, and a baby who was breathing evenly begins to stir. You did not make a sound. You made a smell. Lactating parents in particular are broadcasting a chemical signal that says food is nearby, and the infant's arousal system — which is doing exactly what it evolved to do — responds by nudging him up out of the light stage of sleep he was already cycling through.

You are not disturbing him. You are advertising.

The other half: he saves his hardest feelings for you

There is a second mechanism, and it is the one that hurts less once you see it plainly.

Attachment research — Bowlby's framework, Ainsworth's observational work — describes the primary caregiver as a secure base. The behavioral signature of a secure base is not that the child is calm around you. It is that the child brings you their dysregulation. A toddler who holds it together at daycare all day and detonates in the car with you has not been ruined by you. He has been waiting for you, because you are the person with whom it is safe to fall apart.

Bedtime is dysregulation's prime hour. Sleep pressure is high, cortisol is doing its evening thing, and the baby's capacity to self-manage is at its lowest ebb of the twenty-four. Of course he protests hardest at the person he trusts most. Protest is a bid, and bids are only made to someone you believe will answer.

The babysitter gets compliance. You get the truth.

Why this makes you feel like a fraud

I want to name the thing underneath the logistics, because the logistics are not why you googled this at 11:40 p.m.

When someone else settles your baby easily, the story that arrives — uninvited, fully formed — is I am doing something wrong, and they can see it. It lands somewhere close to worth. You have organized your whole life around this child's wellbeing and the child appears to be voting against you.

But the evidence points the other way. Ease of settling is not a measure of bond quality. It is a measure of cue history. The relative stranger settles the baby easily for the same reason a hotel bed sometimes gives you the best sleep of the year: nothing in the room has been conditioned to mean anything yet.

The person with no history has no history working against them. That is the whole trick. It is not a superior technique.

Your next moves

Cue histories are learnable, which means they are also re-learnable. Here is what to actually do, starting tonight.

  • Break the smell-equals-milk contingency at least once a day. For one nap, feed the baby fully at the start of the wake window — after waking, not before sleeping — so that your presence at the crib is no longer the reliable herald of a feed. This is the single highest-leverage change for breastfeeding parents, and it usually takes seven to ten days for the association to loosen.
  • Build a new pre-sleep cue that isn't your body. Same three-step sequence, same order, every single time: dim the room, one specific song, one specific phrase. Boring is the point. You are constructing a discriminative stimulus that predicts sleep rather than feeding, and it needs enough repetitions to compete with the one you already built.
  • Hand off exactly one settle per day to your partner — the same one. Not whichever one you're most desperate about. Consistency is what makes the new cue legible to the baby. The bedtime settle is the hardest; start with the second nap instead.
  • Put the baby down at the right moment, not the calm moment. Almost every "he won't sleep for me" story I have heard, when you pull the thread, is a timing story wearing a rejection costume. A baby put down twenty minutes past the edge of his window will fight anyone. It only looks personal because you were the one holding him.
  • Leave the room after the transfer instead of hovering. Your continued presence keeps the arousal cue live. Ninety seconds of standing frozen by the crib is often the exact ninety seconds in which he re-alerts.

The part that timing can't fix, and the part it can

Notice how much of the above collapses into a single variable: when. A baby placed down inside his window settles for almost anyone. A baby placed down outside it fights almost everyone, and the parent holding him at that moment absorbs the whole meaning of the failure.

Most of us are guessing at that window — reading yawns that arrive too late, running a schedule that fit three weeks ago, calibrating against a friend's baby who sleeps differently. Drowsy exists for that one narrow, load-bearing question: given how this baby slept last night and how long he has been awake, when is the next window actually open? It watches the pattern so you can stop doing arithmetic at the edge of exhaustion and just pick him up at the right moment — which, it turns out, is most of what the babysitter was doing right.

You can see your baby's next window at drowsy.lumenlabs.works. And then, if it helps: the photo your mother-in-law sent is not evidence against you. Your baby fell asleep on a shoulder that meant nothing to him. Tonight he will fight the one that means everything.