Skip to content
Kura

8 claims, plus 4 to avoid

Mixed, modest, or oversold

Real signals that the popular version has stretched, and pleasant things that do a bit less than promised. Each card leads with what is actually true and then names the claim you will hear, in that order and on purpose.

  1. Mixed Screens · Early learning

    Screens mostly matter because of what they replace.

    More screen time goes with slower language, but the evidence is observational and the mechanism looks like displacement. Video calls with a live person are fine.

    The claim you will hear
    Screens damage babies’ brains.

    The association between more screen time and lower language skills is consistent across 42 studies, but it is observational, and the likeliest mechanism is that screens replace conversation rather than damage anything directly. Two details support that reading: (1) watching together and higher-quality content go with better language, and (2) toddlers learn from a live video-chat partner who responds to them but not from the same partner pre-recorded. The AAP advice (none under 18 months except video calls, then small amounts watched together) strikes me as reasonable rather than sacred.

    Read on: why it works, the evidence, the caveatsShow less

    Why it works

    A baby learns language from turns, and a screen does not take turns, so every hour in front of one is an hour not spent in the loop. The exception proves the rule: a grandparent on a video call responds to the baby’s sounds, and toddlers learn from that.

    The evidence

    Madigan’s meta-analysis of 42 studies (18,905 children) found greater screen use associated with lower language skills, and later onset, higher-quality programming and co-viewing associated with better language. All of it observational.

    Myers and colleagues showed that toddlers aged 12–25 months learned new words and patterns from a live video-chat partner but not from a pre-recorded version of the same interaction.

    The AAP’s 2016 policy: avoid screens other than video chat under 18 months; 18–24 months high-quality programmes watched together; ages two to five at most an hour a day, watched together.

    Caveats

    No trial has randomised babies to screen time, and none will. The association is consistent, the causal story plausible, and the effect sizes small.

  2. Good Sleep · Responsive caregiving

    Sleep training is contested in public and fairly settled in the trials.

    Trials of graduated waiting found no effect on attachment or stress hormones, better parental mood, and no harm at a five-year follow-up.

    The claim you will hear
    Letting a baby cry damages attachment.

    Randomised trials of graduated waiting (graduated extinction, waiting a little longer each time before going in) found no differences in attachment security, stress-hormone levels or behaviour, and less parental stress, and a five-year follow-up of a large Australian trial found no differences at age six in emotional or behavioural problems, sleep, stress regulation or the child-parent relationship. The benefit is mostly parental sleep and mental health, which loops straight back to the responsive-caregiving card. I will admit this one surprised me.

    Read on: why it works, the evidence, the caveatsShow less

    Why it works

    Behavioural sleep methods teach a baby who is developmentally ready (usually past six months) to settle without a specific parental action. The fear is that unanswered crying at night undoes daytime responsiveness. The trials measured exactly that fear, with cortisol samples and attachment assessments, and did not find it.

    The evidence

    Gradisar’s trial (43 infants aged 6–16 months) compared graduated extinction, bedtime fading and a control. Both methods reduced time to fall asleep and night waking; there were no differences in afternoon cortisol, attachment security at 12 months, or child behaviour, and parents in the intervention groups reported less stress.

    Price’s five-year follow-up of 326 children from a randomised trial found no difference between intervention and control at age six on emotional or behavioural problems, sleep, cortisol regulation, or the child-parent relationship, and no lasting effect on maternal depression either way.

    Caveats

    The trials are small and follow-up covers years rather than decades. They rule out the large harms critics feared; they cannot rule out tiny ones.

    Methods vary. “Cry it out” with no checking is not what was tested; graduated checking and bedtime fading were.

  3. Mixed Childcare · Early learning

    The famous preschool wins were for very disadvantaged children.

    For well-resourced families, quality of care matters a little, hours matter a little, and family factors dwarf both.

    The claim you will hear
    Early preschool returns 7–10% a year (or the opposite, that daycare harms babies).

    The long-term wins everyone cites, Abecedarian and Perry, were intensive programmes for very disadvantaged children in the 1960s and 70s, and they did produce remarkable adult outcomes. For ordinary families, the large US childcare study that followed 1,300 children to age fifteen found that higher-quality care predicted slightly better academic scores, that more hours predicted slightly more impulsivity, and that both effects were small next to family factors. My read is that the setting matters less than whether the adults in it are responsive.

    Read on: why it works, the evidence, the caveatsShow less

    Why it works

    A well-resourced home already supplies most of what a good programme supplies: talk, books, routines, responsive adults. For children without those, a programme can be the difference, which is why the same intervention shows very large effects in one population and close to none in another.

    The evidence

    Abecedarian randomised disadvantaged infants to full-day, high-quality care from birth to five. In their mid-30s, treated participants had markedly lower cardiovascular and metabolic risk factors. Perry Preschool, two years of preschool for very disadvantaged three- and four-year-olds, is the source of Heckman’s 7–10% annual return estimate.

    The NICHD Study of Early Child Care followed about 1,300 typical US children to age 15. Higher-quality care predicted modestly higher academic achievement; more hours predicted slightly more risk-taking and impulsivity. Family income, parenting and maternal education predicted far more than any childcare variable did.

    Nurse home visiting for low-income first-time mothers (Nurse-Family Partnership) reduced child maltreatment and, at fifteen-year follow-up, arrests. Again, a high-risk population.

    Caveats

    “Quality” in these studies means responsive, language-rich adult interaction as measured by observers, and it correlates only loosely with price.

  4. Mixed Language · Early learning

    Two languages are worth having. The brain-upgrade claim has not held up.

    Bilingual babies do fine in both languages. The claimed executive-function boost did not survive a correction for publication bias.

    The claim you will hear
    Bilingual children have better self-control and attention.

    Raising a baby in two languages is harmless and worth doing if you have them; vocabulary in each language may run a bit smaller for a while, but the total is about the same and both catch up. The widely repeated claim that bilingualism boosts executive function (attention, switching, self-control) has not held up: a meta-analysis of 152 studies found a tiny advantage that vanished once publication bias was corrected. Do it because it connects her to family and to a place, rather than as a cognitive strategy.

    Read on: why it works, the evidence, the caveatsShow less

    Why it works

    Babies sort speech sounds by statistical regularities, and two input streams are two sets of regularities that they keep separate remarkably early. The “advantage” theory was that constantly managing two languages trains the brain’s control system; it turned out to be mostly a publication effect, with positive studies published and null ones left in the drawer.

    The evidence

    Paap and Greenberg tested bilingual and monolingual adults on 15 indicators of executive processing across three studies and found no advantage on any of them.

    Lehtonen’s meta-analysis of 152 studies found a very small overall advantage that disappeared after correcting for publication bias.

    Caveats

    The replication problem is specific to the executive-function claim. The social, cultural and later-learning benefits of speaking two languages are not in question.

  5. Mixed Music · Responsive caregiving

    Sing to her, and skip the music class if the goal is IQ.

    Singing soothes and bonds, with experimental evidence. Music classes for cognitive gain do not deliver it, and the Mozart effect did not survive the meta-analysis.

    The claim you will hear
    Music makes babies smarter.

    Singing to a baby is well supported for what it actually does: in an experiment, babies aged seven to ten months stayed calm about twice as long listening to singing as to speech, even in a language they had never heard, and singing carries the rhythm and repetition that language learning feeds on. The claim that music makes babies smarter is a different thing. Two randomised trials of preschool music classes found no transfer to spatial, number or language skills, and a meta-analysis of 39 studies found the “Mozart effect” tiny, not specific to Mozart, and inflated by publication bias.

    Read on: why it works, the evidence, the caveatsShow less

    Why it works

    Singing is regulated, rhythmic, emotionally marked speech delivered face to face, which is why it soothes: it gives the baby a predictable pattern to lock onto and a parent who is clearly attending. Recorded music in the background does none of that, and structured music training builds musical skills rather than general intelligence.

    The evidence

    Corbeil, Trehub and Peretz played infants recordings of singing or speech and measured time to distress. Singing roughly doubled the time babies stayed calm, including songs in Turkish for babies who had never heard it.

    Mehr and colleagues randomised preschoolers to music enrichment, visual-arts enrichment or nothing, twice. Music improved music-related skills only.

    Pietschnig’s meta-analysis of the Mozart effect found a small effect on one spatial task, no larger for Mozart than for other music, with signs of publication bias.

    Caveats

    Music lessons later in childhood have their own benefits (discipline, joy, musicianship). The claim being set aside here is the infant cognitive one.

  6. Weak Touch · Responsive caregiving

    Baby massage is lovely and, for healthy babies, unproven.

    It helps weight gain in preterm babies. For healthy full-term babies the Cochrane review found low-quality evidence and no clear benefit, and no harm either.

    The claim you will hear
    Massage improves sleep, digestion and bonding.

    Massage has decent evidence for weight gain in premature babies in neonatal units. For healthy full-term babies under six months, the Cochrane review of 34 studies found low-quality evidence with no convincing benefit on growth, sleep, crying or attachment, and nothing suggesting harm. If you enjoy it, it is touch and attention (both good things); it just is not a treatment.

    Read on: why it works, the evidence, the caveatsShow less

    Why it works

    Touch is a real developmental input, and skin-to-skin has strong evidence. Massage as a specific technique adds structure to touch, but it has not been shown to add outcomes for babies who already get held a lot.

    The evidence

    The Cochrane review (Bennett, Underdown and Barlow, 2013) pooled 34 studies and found low-quality evidence for most outcomes, with the few positive findings coming from small studies at high risk of bias.

    Caveats

    It is a different story for preterm babies, where massage in the unit has evidence for weight gain and shorter stays.

  7. Weak Language · Early learning

    Baby sign does not speed talking. It may make you more attentive.

    A randomised trial found no language benefit for the child, but parents who used it became more responsive to non-verbal cues.

    The claim you will hear
    Signing babies talk earlier and have fewer tantrums.

    The one randomised trial, with 40 mother-baby pairs followed from eight to twenty months, found that teaching babies to gesture did not accelerate language development. The mothers who used signs did become more responsive to their babies’ non-verbal cues, which is a real if indirect benefit. Fine to do, in other words, and not magic.

    Read on: why it works, the evidence, the caveatsShow less

    Why it works

    Babies gesture spontaneously (pointing, reaching, waving), and those gestures do predict language. Teaching extra signs adds symbols, but the trial suggests it does not change the underlying pace of language development. What it does change is how closely the parent watches the baby - which is the first card again.

    The evidence

    Kirk and colleagues randomised mothers to symbolic gesture training, British Sign Language training, verbal training or nothing. No group’s babies developed language faster, and mothers in the gesture groups scored higher on responsiveness to their babies’ non-verbal cues.

    Caveats

    One trial, and a small one. It rules out a large effect rather than a tiny one.

  8. Weak Devices · Good health

    Flat spots mostly round out on their own. Helmets did not beat waiting.

    A randomised trial of helmets for moderate-to-severe flattening found no difference in skull shape at two years, and frequent side effects.

    The claim you will hear
    A helmet will fix her head shape.

    Back-sleeping produces a lot of flattened heads (positional plagiocephaly), and a helmet industry grew up around it. The one randomised trial (84 babies aged five to six months with moderate-to-severe flattening) compared helmets with simply waiting: at two years there was no difference in skull shape, and helmet side effects (skin problems, pain, sweating, feeling hindered in cuddling) were common. Tummy time and varying her head position are the evidence-based responses.

    Read on: why it works, the evidence, the caveatsShow less

    Why it works

    The skull is soft and reshapes with pressure in both directions. As babies spend more time upright and rolling, the pressure that flattened the head is removed and growth rounds it out; helmets try to accelerate that, but the trial suggests they add little to what growth does anyway.

    The evidence

    Van Wijk’s trial in the BMJ found full recovery in 26% of the helmet group and 23% of the natural-course group at 24 months, a difference within chance, with all parents in the helmet group reporting at least one side effect.

    The tummy-time review lists prevention of skull flattening among the benefits of time on the front.

    Caveats

    The trial excluded the most severe cases and babies with medical causes of skull deformity, which are different conditions.

4 product categories

No evidence, or harmful

Four product categories with documented harm, or with no benefit to weigh against the risk.

  • AvoidProducts

    “Educational” baby videos and infant reading programmes.

    One study linked each daily hour of baby DVDs to about 17 fewer words understood, and the maker ended up refunding buyers. Infant reading kits have no evidence at all.

    In a study of 1,008 families, each daily hour of baby-targeted DVDs was associated with about 17 fewer words understood (out of 90 tested) in babies aged 8 to 16 months, while reading and storytelling went the other way. Disney later offered refunds on Baby Einstein. Flash-card and “your baby can read” programmes have no supporting evidence, and the US Federal Trade Commission acted against the best known of them for deceptive claims.

    If someone gifts one, then use the box for something else and the time for a book.

    Sources

  • AvoidProducts

    Amber teething necklaces and benzocaine teething gels.

    No evidence of pain relief, and documented strangulation and choking deaths. Benzocaine gels can cause a dangerous blood disorder in infants.

    Amber necklaces are sold on the claim that warmed amber releases a pain-relieving acid. It does not, and beads on a cord around a baby’s neck have caused strangulation and choking deaths, prompting warnings from the FDA and the AAP. Benzocaine teething gels carry an FDA warning for infants because of methemoglobinemia, a condition that stops blood carrying oxygen. A cold washcloth or a solid rubber teether is the recommendation.

    If she is teething, then a clean cold washcloth to chew, a solid (not liquid-filled) teether, and a finger rub on the gum.

    Sources

  • AvoidProducts

    Baby walkers.

    About 230,000 US emergency visits over 25 years, mostly head injuries from stairs. They do not help walking and may delay it. Banned in Canada.

    Wheeled walkers let a baby who cannot walk move fast, usually toward stairs. Roughly 230,000 US children under 15 months were treated in emergency departments for walker injuries between 1990 and 2014, most for head and neck injuries, and studies also suggest walker use delays independent walking rather than advancing it. The AAP has called for a ban; Canada banned them in 2004. A stationary activity centre gives the same entertainment without the wheels.

    If one is already in the house, then take the wheels off or give it away.

    Sources

  • AvoidProducts

    Inclined sleepers, crib bumpers, weighted sleep sacks, positioners.

    Each has been linked to infant deaths. Inclined sleepers were recalled after dozens of deaths, and the AAP advises against all of them.

    The sleep-product market invents a new soft, angled or weighted thing every year, and the 2022 AAP safe-sleep statement names them: inclined sleepers (recalled in the US after dozens of deaths), crib bumpers (now banned in the US), weighted blankets and sleep sacks, and positioning wedges. None has evidence of benefit and each has been linked to deaths. The safe surface is boring on purpose.

    If it tilts, pads, weighs or wedges, then it is not for sleep.

    Sources

Next: which early patterns keep paying off, and three caveats.

What carries forward