At 2:14 in the morning, you open the nursery door and find your baby standing. Not crying yet — standing. Two fists on the crib rail, knees locked, swaying slightly, staring into the dark with the concentration of a person doing something enormously important. And they are. You just wish they'd picked a different hour to do it.

Here is the part nobody tells you: this is not your baby breaking. This is your baby working. The night waking that arrives with crawling, with pulling to stand, with those first drunken steps across the rug — that waking is not a malfunction in the sleep. It is the learning. And understanding why changes what you do at 2 a.m., because you stop trying to fix something that was never broken and start protecting something that's fragile.

The brain that just learned something can't stop touching it

When a baby acquires a new motor skill, they don't acquire it once. They acquire it thousands of times a day. Karen Adolph's lab at NYU put step counters and cameras on new walkers and found that a toddler in the thick of learning to walk takes something on the order of 2,400 steps and falls roughly 17 times per hour of free play. That's not a child who occasionally walks. That's a child running an obsessive, high-repetition training program, and nobody assigned it. The drive came from inside.

Now ask yourself what happens when that child is placed, at 7 p.m., into a small padded rectangle with a horizontal bar at exactly grabbing height.

The crib becomes a gym. This is not misbehavior and it isn't stalling. Motor learning in infancy runs on what researchers call massed practice with variability — the baby repeats the movement constantly, and each repetition is slightly different, and the nervous system slowly extracts the pattern from the noise. A brain in the middle of that process is not a brain that quietly powers down when the lights go off. The circuits are hot. The moment sleep gets shallow — and infant sleep cycles through shallow every 40 to 60 minutes — the hands find the rail.

Sleep isn't collateral damage. It's the second half of the lesson.

Here's where it gets genuinely beautiful, and where the parenting internet usually stops short.

The practice doesn't produce the skill. The practice produces raw material. Sleep is where the material gets built into something usable. In sleep-dependent motor consolidation — well documented in adults, and increasingly in infants — the brain replays the day's movement sequences during sleep, strengthens the useful pathways, prunes the useless ones, and hands you back a slightly better version of the skill in the morning. This is why you've had the uncanny experience of putting down a baby who was army-crawling and picking up a baby who is crawling. Nothing visible happened in between. Everything happened in between.

Sarah Berger and Anat Scher's research on new walkers found that infants who napped after a locomotor problem-solving task performed better afterward than infants who stayed awake — the nap wasn't recovery from the effort, it was part of the effort. Meanwhile, the same body of work has repeatedly linked the onset of independent locomotion — crawling, then walking — with a measurable uptick in night waking. Both things are true at once. The skill disrupts the sleep, and the sleep completes the skill. Your baby is caught in a loop that is, from an engineering standpoint, magnificent, and from a parenting standpoint, at 2:14 a.m., absolutely ruinous.

So the honest framing is this: your baby is not sleeping badly despite learning to crawl. Your baby is sleeping badly in order to learn to crawl. The waking is the cost of the acquisition, and the acquisition is what you've been waiting for since the day they were born.

Why it feels like the sleep training stopped working

Parents in this window describe the same thing with eerie consistency. The baby who went down easily now fights bedtime. The baby who self-settled now shrieks at 1 a.m. — and when you go in, they're not distressed, they're upright, and the crying only started because they got up there and couldn't figure out how to get back down.

That last detail is the whole thing. Pulling to stand comes online weeks before lowering to sit does. The brain builds the up before it builds the down. So a baby will haul themselves to standing in a dark room at one in the morning, achieve the summit, and then discover they are stranded — locked knees, no exit strategy, and a genuine flare of fear. The crying is not sleep resistance. It's the cry of someone stuck on a ledge.

Same with the sitting-up phase. Same, later, with cruising, and with climbing. There is a reliable gap between the skill that gets you into a position and the skill that gets you out of it, and that gap is where most of the night waking lives.

The other pressure is subtler. New motor skills consume enormous energy and attention, which means the baby's tolerance for being awake changes — sometimes shortening (they burn out faster), sometimes lengthening (they're too fascinated to stop). The wake windows you were running on last month were built for a baby who couldn't cross the room. That baby is gone. He crawled away.

What actually helps

The instinct is to rescue: pick them up, lie them down, tuck them in, leave. Ten minutes later, they're standing again. You've now taught a fascinating game with an excellent reward.

The better move is counterintuitive. You don't stop the practice. You move it — into the daylight, where it belongs, and you make sure the missing half of the skill gets built before it's needed in the dark.

Your next moves

  • Teach the descent, in daylight, on purpose. Three or four times today, let your baby pull to standing at the couch or crib rail, then physically guide their hands down the rail and bend their knees for them. Ten repetitions. You're not entertaining them — you're building the exit route so they don't need you at 1 a.m. to find it.
  • Give them the floor before you give them the crib. In the hour before the bedtime routine starts, put them on a clear, unrestricted floor and let them crawl, cruise, and fall as much as they want. Massed practice earlier means less residual drive later. Do not do this during the routine — end with the calm part.
  • Lower the crib mattress today, not next week. The night your baby first pulls up is the night the mattress height becomes a safety issue, and standing babies find the rail faster than parents find the Allen key.
  • When they stand at night, go in once, lay them down without eye contact or conversation, and leave. If they pop back up, wait — give them ninety seconds to try the descent you've been drilling. Most babies, once they own the skill, will solve it themselves. Rescuing on every rep teaches them to summon you instead.
  • Add fifteen to twenty minutes to the wake window before bedtime, for two weeks only. New locomotion tends to raise the sleep pressure needed to override the excitement. Then reassess — this is a temporary adjustment, not a new schedule.

And then give it two to four weeks. Milestone-driven waking has a shape: it arrives fast, it peaks, and it resolves on its own once the skill becomes automatic — once crawling stops being a project and starts being just how you get to the dog. The families who come out of it worst are usually the ones who rebuilt the entire sleep architecture around a three-week storm.

The window is still the window

The hard truth of this phase is that the underlying biology hasn't changed. Your baby still has a sleep pressure curve. There's still a window each afternoon when the pressure has built enough for a nap to take hold and not so much that cortisol has kicked in and slammed the door. What's changed is that the old landmarks — the eye rubs, the yawns, the reliable 2:30 nap — got buried under a new and very loud motor obsession. The signal is still there. The noise just got louder.

That's the specific problem Drowsy was built for: reading the underlying rhythm — age, sleep pressure, how the last nap actually went — and telling you the next window with a real number instead of a guess, especially in the weeks when your baby's cues have gone haywire and you've forgotten what normal looked like. It won't teach your baby to sit back down. But it will tell you when to try.

If you're deep in the practice nights, Drowsy is there when you're ready. And when you open that door at 2:14 and find them standing in the dark — take one second, before the exhaustion lands, to notice what you're actually looking at. Somebody very small is learning to move through the world. It was always going to cost you a few nights.