An Extra Hour of Screens at Age 1, Tied to 1.47 Points Lower at 9

A one-year-old on a lap, a phone angled outward to buy fifteen quiet minutes at the end of a long day. It is the least dramatic thing that happens in a house, and it is the one parents keep relitigating at two in the morning, because the honest answer has always been that nobody followed those particular babies long enough to know.

A team from Inserm and the National University of Singapore has now done something close to that. Working inside the Growing Up in Singapore Towards healthy Outcomes birth cohort, they took parent-reported screen viewing time from 502 children at six separate points — ages 1, 1.5, 2, 3, 6 and 8 — and then waited. At age 9 the children sat the Wechsler Individual Achievement Test-Third Edition. At 10.5 they completed the Letter-Number Sequencing task from the Wechsler Intelligence Scale for Children-Fifth Edition, a standard measure of working memory: hold a jumble of letters and digits in mind, reorder it, say it back.

The design is the point. Most of what exists on this question is a single snapshot of screen time set against a single later score, which cannot distinguish a sensitive window from a lifelong habit. Repeating the exposure measure six times lets the authors ask a sharper question — not whether screens matter, but when. The paper was received in September 2025, accepted in April 2026 and published online on 3 June in the World Journal of Pediatrics.

The results are given as standard test points per additional hour per day of screen viewing, in models adjusted for family and environmental factors. Age 1 produced the largest coefficients in the study: each extra daily hour was associated with 1.47 fewer points on the age-9 achievement test (95% CI -2.37 to -0.57) and 1.12 fewer working-memory points at 10.5 (95% CI -2.07 to -0.17). Age 1.5 carried a smaller academic association, -0.95 (95% CI -1.85 to -0.06). Then the signal disappeared. Ages 2 and 3 showed no significant association with either outcome. At age 6, roughly the point of formal school entry, it reappeared: -0.88 for academic performance (95% CI -1.55 to -0.21) and -1.01 for working memory (95% CI -1.71 to -0.31).

The authors' own framing is restrained. "The effect sizes we saw at age 1 were the largest among all time points we examined," one told the release announcing the work, adding that this "suggests early infancy may be a window of heightened sensitivity." Their summary verdict fits in three words — the findings, the release says, support the principle that "less is better." They also note, in the same breath, that "an additional hour per day may not produce an obvious effect in every child."

A second set of models averaged each child's screen time across the whole decade. Cumulative exposure was consistently associated with poorer academic performance — but not with working memory, which is a genuine split the coverage has mostly skipped past.

The framing that travelled fastest — two dangerous ages, the years between them untouched — is the part the study cannot carry. Ages 2 and 3 returned estimates that did not reach statistical significance. That is not the same finding as no effect. With 502 children contributing twelve separate estimates across six ages and two outcomes, the analysis has enough precision to detect a moderate association and not much more; an interval that includes zero at age 2 is compatible with a real effect the sample was too small to resolve. Treating a significant estimate at 1.5 and a non-significant one at 2 as evidence of a cliff between them requires a comparison the paper does not report.

The significant results deserve the same scrutiny in reverse. Two of the five have upper bounds that all but graze zero: -0.06 for age-1.5 academic performance, -0.17 for age-1 working memory. Those are findings the data are compatible with, not findings the data pin down. And there are twelve single-time-point estimates on offer across six ages and two outcomes; nothing in the accessible material indicates a correction for testing that many. Five hits out of twelve, two of them borderline, is a pattern worth holding loosely.

Then there is the exposure itself. Screen viewing at age 1 was reported by parents, without any breakdown by content, device, or whether an adult was watching alongside — the authors flag all three as work for someone else. And an hour of screens in a twelve-month-old's day is not a free-standing variable. It travels with caregiving arrangements, with how much adult speech fills the room, with how depleted the adults are, with everything that also predicts a test score eight years later. Statistical adjustment narrows that problem; it does not close it. The reverse pathway is live too, since an infant who is harder to settle tends to get handed the screen.

Something still holds after all that. The two ages that lit up are the two the developmental literature would have nominated in advance — the first year, when language and attention are being laid down through back-and-forth with a person, and the transition into formal schooling, when a child's day reorganizes around sustained instruction. A result that lands where theory expected it is weaker evidence than a randomized trial and stronger evidence than a coincidence.

Scale matters for how a parent should hear this. These are population-level shifts of roughly one point per daily hour on tests built to spread children across a wide range. That is a nudge in an average, not a diagnosis, and the authors say as much. It sits alongside existing guidance from the WHO and the American Academy of Pediatrics — avoid screens before 18 to 24 months, under an hour a day from 2 to 5 — which this study neither validates nor overturns, having never tested a threshold.

The other caveat is the calendar. Children whose working memory was measured at 10.5 had their infancy screen time recorded about a decade earlier, on the devices of that decade. Whatever a one-year-old is handed in 2026 was not the exposure being measured here, and that gap is unlikely to shrink; longitudinal work on childhood is always reporting on a world that has already moved.

There is a version of this study that fits in a headline and a version that fits the data, and they are not the same length. The data version: in one Singapore birth cohort, screen time at 12 and 18 months and again at age 6 tracked with modestly lower achievement and working-memory scores years on, at roughly a point per daily hour, with two of the five significant intervals nearly touching zero and the middle years showing nothing the sample could resolve. The headline version — screens damage the brain at two specific ages and leave the years between untouched — overreaches twice, once by converting association into damage and once by converting a non-significant estimate into a demonstrated absence. What a careful parent can take from it is narrower and more useful than either: infancy is the period where this cohort's numbers were largest, the authors' own summary is that less is better rather than that any amount is harmful, and nothing here supports the belief that a toddler crossing some hourly line has been marked. The study is a reason to be thoughtful about the first eighteen months. It is not a reason to be frightened about the fifteen minutes.