Sleep Training Harms

Wendy Middlemiss and colleagues published a 2012 study in Early Human Development that has become the most-cited paper in the critique of extinction sleep training. The team measured cortisol (the primary stress hormone) in mother-infant pairs across a five-day inpatient sleep training program. On day one, when babies cried alone, their cortisol levels and their mothers' cortisol levels were synchronized: high together, low together. By day three, the babies had stopped crying. This was the success metric for the program. But their cortisol levels had not dropped.

The cortisol stayed elevated even though the crying had stopped. The mothers' cortisol, hearing no crying, had dropped to normal. The synchrony was broken. The babies were still in physiological distress. They had simply stopped signaling it. This is the central finding the field has been working with ever since.

Darcia Narvaez is a developmental psychologist at the University of Notre Dame whose work develops a framework called the "evolved developmental niche." The framework identifies the specific kinds of caregiving infants have evolved to expect across human history, including responsive caregiving to distress, breastfeeding, frequent physical contact, multiple consistent caregivers, free play, positive social support, and natural birth. Sleep training violates at least the first of these by design.

Her published work argues that extinction-based sleep training represents a failure of co-regulation, the early process by which an infant's nervous system learns to manage states of arousal by being met by a regulated adult nervous system. Co-regulation that does not happen, according to her framework, sets up adult patterns of dysregulation, anxiety, and impaired stress response. The framework is conceptual rather than experimental, but it pulls together a wide literature.

Bilgin and Wolke published a 2020 paper in the Journal of Child Psychology and Psychiatry examining whether parental use of cry-it-out affects infant attachment and behavioural development. The team followed 178 infants and their mothers, measured parental use of leaving the infant to cry at term, 3, 6, and 18 months, and assessed attachment at 18 months using the strange situation procedure plus parent-reported and observational behavioural measures. The headline finding was that contemporary parental use of cry-it-out in the first 6 months was not associated with adverse attachment or behavioural development at 18 months. The paper generated substantial controversy.

Commentaries by Davis and Kramer (2021) and others argued the study was underpowered for the kinds of subtle attachment effects predicted by theory and that the analytical choices may have overstated the null finding. Both the original paper and the critiques sit in the literature side by side.

Sarah Blunden and colleagues have published several papers, including in Sleep Medicine Reviews, critiquing the methodology behind the studies most often cited in support of sleep training. The critiques focus on several recurring issues.

Most of the studies are short-term, measuring outcomes at six months or earlier, well before any developmental effects would be expected to appear. Most of the studies measure the parent-reported outcome (parents say the baby sleeps better) rather than the infant outcome (the baby's actual sleep quality, stress markers, attachment).

Most studies are not blinded, so the parents and assessors know which intervention is being studied. Most use middle-class educated samples that generalize poorly. The Blunden work has been influential in the academic conversation but has not significantly changed the popular sleep training literature.

Allan Schore is a neuropsychologist whose decades of work on attachment neurobiology argue that the right brain develops largely through the dyadic relationship with the primary caregiver in the first two years of life. The mechanism, in his account, is moment-to-moment co-regulation: the infant cries, the caregiver responds, the infant's arousal regulates, the caregiver mirrors the regulated state, and over thousands of repetitions, the infant's brain develops the architecture for self-regulation.

When that loop is repeatedly broken, as it is by extinction sleep training, the developing brain has to make do with the alternative. The alternative, in Schore's framework, is a developmental pattern where the right brain wires for distress and dissociation rather than regulation. The framework draws on neuroimaging, animal studies of separation, and clinical data on adult patients with regulation disorders.

Harriet Hiscock and colleagues published a five-year follow-up study in Pediatrics in 2012 that has become the strongest counter-evidence in the sleep training debate. The team followed 326 families who had participated in a randomized trial of behavioral sleep interventions (including controlled crying) at infant age seven months.

The follow-up at age six tested behavioral problems, parent-child attachment, mental health outcomes, and cortisol patterns. The headline finding: no statistically significant differences between the sleep-trained group and the control group at age six. The Hiscock paper is regularly cited by sleep training advocates as evidence that the harm concerns have not held up at follow-up.

The methodological critiques include that the cortisol sampling was limited, attachment was measured through parent report rather than direct observation, and the sample was relatively small for the kinds of subtle long-term effects predicted by attachment theory. Both points have weight. We include this paper because the case has to engage with the strongest opposing evidence honestly.

Pamela Douglas and Peter Hill published a 2013 systematic review in the Journal of Developmental & Behavioral Pediatrics titled "Behavioral sleep interventions in the first six months of life do not improve outcomes for mothers or infants." The team examined the published evidence on behavioral sleep training delivered to infants under six months of age and concluded that the interventions do not reliably improve infant sleep, maternal mood, or maternal sleep, and that there is no evidence supporting their use in this age range. The review also noted that the trials needed to resolve the harm question (large, long-term, with rigorous outcome measures including infant stress markers and validated attachment measures) have not been conducted. The review is widely cited in the academic conversation about sleep training but has not significantly shifted the popular literature, which continues to recommend behavioral approaches in this age range.