Every time you try drowsy-but-awake, the same thing happens. You wait for the window, you lower your baby carefully into the cot, and the moment their back touches the mattress they are wide awake — or crying. Or both.

This is not a failure of execution. It is not that you aren’t doing it quickly enough, or smoothly enough, or at the right level of drowsiness. The reason drowsy-but-awake fails for most babies is that it asks them to do something their brain cannot yet do. Understanding why makes the failure make sense. And once it makes sense, the guilt becomes harder to sustain.

For the complete picture — including what to do instead — see our guide to drowsy but awake. This article focuses specifically on the neuroscience: what is happening in your baby’s brain during sleep onset, what self-soothing actually requires, and why the drowsy window is such an elusive target.

How babies actually fall asleep

Sleep onset is a neurological process. It requires the brain to shift from active wakefulness, through a transition of decreasing arousal, into the first stage of light sleep. In adults, this transition is typically smooth and self-directed — the brain has the architecture to manage it. In infants, it almost always requires external support.

Sleep architecture in infancy

Newborns have a relatively simple sleep cycle: active sleep (an equivalent of REM) and quiet sleep (non-REM equivalent). They fall into deep sleep rapidly after onset — often within minutes — and spend a higher proportion of total sleep time in active sleep than adults do. This is part of why early newborn sleep often looks deceptively easy: they drop off quickly and deeply.

By three to four months, sleep architecture matures significantly. The adult pattern of sleep stages begins to emerge, with lighter N1 and N2 stages becoming more prominent in the cycle. This is the neurological basis of what is commonly called the “four-month sleep regression” — not a regression, but a developmental shift that makes falling asleep independently harder, not easier, as the baby must now navigate more complex sleep stage transitions. See our guide to sleep regressions for more on this.

The practical consequence: the older a baby gets, the more likely they are to move through a period of near-wakefulness at sleep onset — a moment when, if they are being transferred from arms to cot, they are likely to rouse fully.

Proximity regulation

Young babies do not maintain the physical conditions necessary for sleep using only their own bodies. They use their caregiver’s body. Research on skin-to-skin contact and infant physiology documents this clearly: close contact with a caregiver regulates the baby’s heart rate, breathing, temperature, and cortisol levels — a process called co-regulation.

When a baby is moved from warm arms to a flat, still surface, these regulatory inputs are removed simultaneously. The nervous system responds. For many babies, this shift is sufficient to interrupt the transition into sleep entirely.

The self-soothing myth

“Self-soothing” is the phrase at the centre of drowsy-but-awake’s claim. The idea is that babies who fall asleep in their cot learn to soothe themselves between sleep cycles, while babies who are helped to sleep develop a dependency. But self-soothing is not a behaviour that can be learned on a training schedule. It is a capacity that has to develop.

What self-soothing actually requires

Genuine self-regulation — the ability to manage internal arousal without external support — requires three brain systems to be functional and working together:

  • The prefrontal cortex, which governs planning, inhibition, and top-down regulation of emotional responses
  • The limbic system, particularly the amygdala, which processes threat and emotional salience
  • The autonomic nervous system, which manages the body’s stress response

In infants, all three are immature. The prefrontal cortex undergoes particularly protracted development — key connectivity and myelination continue throughout childhood, with some researchers arguing that full development of these regulatory pathways doesn’t complete until puberty [1].

This is not a minor caveat. It is the foundational problem with the drowsy-but-awake framework. The advice assumes that self-soothing can be trained into an infant through consistent practice. The developmental neuroscience says otherwise: it is a capacity that emerges as the brain matures, on its own schedule, regardless of what techniques the parent uses at bedtime.

What “self-soothing” in babies actually looks like

The Goodlin-Jones et al. (2001) videosomnography study found that babies classified as “self-soothers” woke just as often as “signalers” — babies who called for parental help. The difference was not sleep quality. It was whether they vocalised [2].

This is a crucial distinction. A baby who doesn’t signal isn’t necessarily sleeping better. They may be a baby who has learned that signalling isn’t effective — which can look like independent sleep from the outside, but is functionally different. Or they may simply be a temperamentally easy baby with a lower arousal threshold from birth.

Burnham et al. (2002) found that the strongest predictor of self-soothing at 12 months was high levels of quiet sleep at birth — a constitutional, not trained, characteristic [3]. Some babies are built for early independent sleep. Most are not. And neither group is the result of what technique their parent used.

Why drowsy-but-awake specifically fails

Even under ideal conditions, drowsy-but-awake faces a structural problem: it asks the parent to hit a target that may not reliably exist.

The window problem

The “drowsy but awake” state is a specific neurological transition — the point where sleep pressure is high but consciousness hasn’t fully lapsed. In adults, this state might persist for several minutes. In infants, particularly in the early months, it can be vanishingly brief. Sleep professionals who work with families have described it as the “unicorn sleep state” [1]: a thing that is discussed as if readily observable and reliably reproducible, but which many parents simply never find in their baby.

The advice gives no guidance for parents whose babies don’t have this window, because it was built on the assumption that the window is universal. It isn’t.

The breastfeeding problem

For breastfeeding babies, the mechanism works actively against drowsy-but-awake. Evening breast milk contains cholecystokinin (CCK) — a digestive hormone released in the baby’s gut during feeding that produces simultaneous satiety and sedation. It also contains tryptophan and sleep-promoting nucleotides that are present in higher concentrations in evening milk than daytime milk. By the time an evening feed ends, the baby’s chemistry has been specifically calibrated for sleep. In many cases, the baby is either fully asleep or snapping back to alertness as the breast is removed — with no reliable drowsy-but-awake window in between. For more on the biology, see our guide to feeding to sleep.

The sensitive baby problem

For temperamentally sensitive babies — those with higher arousal thresholds, stronger startle responses, or a greater need for proximity — no refinement of technique makes the transfer smooth. Being placed on a flat, still, solo surface triggers alarm because alarm is the appropriate physiological response to separation from a caregiver. These babies aren’t failing at drowsy-but-awake. They are succeeding at attachment.

What’s actually helpful

If drowsy-but-awake isn’t the mechanism, what is?

Environment, not state

The evidence more consistently supports the sleep environment than the specific state at sleep onset. A dark room, white noise, a consistent pre-sleep routine, and appropriate wake windows for your baby’s age all create conditions that support sleep onset across a range of methods. These are replicable, low-stakes, and don’t require hitting a moving neurological target.

Multiple sleep cues

Rather than engineering a single sleep association — the cot, the drowsy window — consider building a set of contextual signals: a particular song, a specific phrase, white noise, a dark room. No single cue becomes essential because several exist together. This isn’t about manufacturing independence. It’s about giving your baby a reliable sleep context.

Parental support as a feature

For many babies, parental help at sleep onset is not a problem to be eliminated — it is the appropriate developmental response. Rocking, feeding, holding, and carrying work because they tap into biological systems designed to facilitate sleep. Using them is responsive parenting, not dependency creation. See our guides to contact sleeping and feeding to sleep.

Waiting for development

The research is consistent: self-soothing develops primarily through maturation. Babies who need significant help at four months often need considerably less at eight months, and less again at 12 months — across a wide range of parenting approaches. Time and development do most of the work.

If you want to encourage independent sleep more actively from six months onwards, graduated approaches that keep parental support present — gradual retreat, stay-and-support — are better matched to what we know about infant brain development than techniques that require the baby to self-regulate before the capacity exists.

The bottom line

The way your baby falls asleep right now is a reflection of where their brain is developmentally — not a character flaw, not a parenting failure, and not a permanent situation. Drowsy-but-awake demands that your baby perform a neurological feat before the infrastructure is in place. That’s not a problem with your baby or your technique. It’s a problem with the advice.

For the emotional side of this, read If ‘Drowsy But Awake’ Isn’t Working, You’re Not the Problem. For the specific claims that underpin the advice, see The Myth of Drowsy But Awake.

References below.