The science of baby sleep

Newborn sleep is a different thing entirely

Babies aren't small adults at night. Newborns sleep a lot in total but in short bouts, spread around the clock — and the structure of that sleep is unlike ours.

Two practical consequences. First, brief night wakings are normal biology, not a problem you caused. Second, because babies pass through light sleep so often, the sound environment at those moments matters: a sudden noise during light sleep — a door, a sibling, a truck — can tip a surfacing baby into full waking.

The body clock comes online slowly

Newborns aren't born knowing night from day. Circadian rhythms — including the baby's own melatonin production — emerge over roughly the first 2–4 months, which is why day/night confusion is universal early on and why sleep gradually consolidates into longer night stretches across the first year. Light, feeding times, and consistent evening routines are the cues that help the clock set itself.

What sleep analysis actually shows

Researchers don't rely on what parents remember at 3 a.m. Infant sleep is measured directly — with polysomnography (brain waves, breathing, and muscle tone scored in the lab), actigraphy (wearable movement sensors at home), and videosomnography (infrared overnight video scored minute by minute). Three findings from that work change how you read your own baby's nights.

1. Babies fall asleep "backwards" — and stay light at first

Adults descend through deeper non-REM stages and reach REM later. Newborns typically do the opposite: they enter sleep through active sleep — the REM-like state full of twitches, flutters, grunts, and irregular breathing — and spend about half their sleep there, cycling roughly every 50 minutes (Grigg-Damberger, 2016). The practical upshot: the first stretch after falling asleep is the most fragile, which is why a baby who was deeply asleep in your arms can wake the instant they touch the mattress.

2. "Sleeping through the night" is really "resettling quietly"

Overnight video studies found that almost all infants — including the celebrated "good sleepers" — still wake briefly during the night. What separates them is what happens next: self-soothers resettle on their own, while signalers cry out for help (Anders et al., 1992). A quiet night usually means your baby woke and drifted back on their own — not that they never surfaced.

3. Sleep consolidates early, steeply, and very unevenly

Night sleep knits itself into longer stretches across the first year, with the steepest gains in the first four months; by around five months, roughly half of infants regularly manage a long overnight stretch — and perfectly healthy babies land on both sides of that line (Henderson et al., 2010). Wide variation is the norm, not a verdict on your parenting. This reorganisation is also the stretch many parents call the "4-month sleep regression": a baby who had been sleeping longer starts waking more for a few weeks while the body clock and sleep structure mature. It is a common phase in a steep, uneven process — not a step backwards.

What steady noise actually does

The womb is not quiet. For nine months your baby lived with a constant low-frequency rumble — blood flow, digestion, a heartbeat — so to a newborn, silence is the unfamiliar condition. Steady, featureless noise does two evidence-relevant things:

1. It helps settling. In the classic trial, 80% of newborns fell asleep within five minutes with white noise playing, versus 25% without (Spencer et al., 1990).

2. It masks the sounds that wake babies. A steady sound floor smooths over the door-slams and dog-barks that would otherwise stand out against silence — exactly when your baby is cycling through light sleep.

Why we lean "brown" rather than bright white

White noise spreads energy evenly across all frequencies, which gives it a hissy edge. Brown noise shifts the energy toward low frequencies — closer to the womb's acoustic world, which recordings show is dominated by low-frequency sound (Gerhardt & Abrams, 2000) — and softer on ears at the same loudness. That's the bed under every TwinkleHush sleep video.

How loud is too loud?

The one real risk with sound machines is volume. When researchers tested 14 infant sleep machines at maximum volume, several could exceed levels safe for adult workplaces, let alone infant ears (Hugh et al., 2014, Pediatrics). The fix is easy and doesn't reduce the benefit:

Why lullabies, and why so slow

Lullabies are one of the few genuinely universal human behaviors, and the research backs the instinct: in a multi-hospital NICU study, lullabies slowed infants' heart rates and improved feeding and sleep patterns (Loewy et al., 2013, Pediatrics). Decades of work on infant-directed singing (Trehub and colleagues) shows babies attend to slow, simple, repetitive melodies — and that caregiver singing regulates arousal.

That's why our lullabies are music-box simple: one melody, slow tempo, no drums, no dynamic surprises — layered over the noise bed rather than replacing it. And nothing beats the live version: your own singing voice is still the best lullaby technology ever tested. Ours is for the 3 a.m. moments when yours needs a rest.

Screens at bedtime — the honest part

The American Academy of Pediatrics recommends avoiding screen media for children under about 18–24 months (video chat excepted), and screen use near bedtime is associated with worse infant and toddler sleep, not better. No sleep video — including ours — is "good for" a baby to watch.

So why make videos at all? Because YouTube is where parents already are, and it's a convenient way to play long, seamless, gentle audio. We just design for the truth of it:

Safe sleep, always

Sound and songs are comfort tools — the non-negotiables of safe sleep are the AAP basics: baby on their back, on a firm flat surface, in a clear crib (no blankets, bumpers, or pillows), in the parents' room for the early months, smoke-free. If a question ever feels medical, it's pediatrician territory, not YouTube territory.