Do Screens Make Minds Worse? What the Randomised Evidence Actually Shows

The question

Two claims circulate with almost identical confidence and no shared evidence. The first is defensive: screens are damaging children's brains, and we must cut them back. The second is optimistic: cutting screen time will make children sharper. Both trace to the same correlational literature, and both inherit the same measurement problem.

The underlying question is narrower and more answerable than either: when researchers actually took screen time away from people and measured what happened to their minds, what did they find?

This brief separates four evidence streams that headlines treat as one: correlational studies linking screen use to cognitive scores; randomised trials that reduced screen use and measured outcomes; acute laboratory experiments on distraction, which estimate the cost of a phone being present, not of a life being restructured; and natural experiments on school phone restrictions, which remove access during a fixed window of the school day.

I label claims the way this archive does elsewhere: an observation isolates no cause, a quasi-causal estimate leans on policy timing or comparison groups, a causal estimate comes from random assignment, and an extrapolation extends a result beyond what was tested.

Stream 3 deserves a warning label: it is the strongest-looking evidence and the least transferable. Böttger et al. (2023) re-analysed 22 studies and 43 effects on the brain-drain finding that a smartphone in reach degrades task performance, and report "a significant overall negative effect of smartphone use and presence", while adding that "not all cognitive domains are equally affected" and that "the heterogeneity of the effects reinforces this finding". That is a real causal effect — of presence during a task, over minutes, in adults. It says nothing about what happens to a child whose afternoon is restructured, and it is the basis on which the pessimistic claim is usually overstated.

What the randomised trials actually tested

Four randomised studies are the backbone of the causal record. Two are household trials of families, one a student trial, one an adolescent detox pilot, spanning roughly ages 16 months to 23 years.

Three things stand out.

First, the interventions are large and compliance is high. In the Danish trial, 97% of intervention children met the screen limit; the trial was registered as NCT04098913 and protocoled by Rasmussen et al. (2020) before its results appeared. Whatever is wrong with this literature, it is not that the treatments were too weak to bite.

Second, three of the four trials never measured cognition at all. Pedersen et al. (2022) measured physical activity and EEG sleep; their abstract reports that "no significant between-group mean differences were observed between intervention and control for the electroencephalography-based sleep outcomes", while the primary outcome moved hard — a between-group difference in leisure non-sedentary activity of 45.8 minutes per day (95% CI, 27.9-63.6; P < .001). The prespecified secondary analysis, Schmidt-Persson et al. (2024), found "a statistically significant between-group mean difference in the total difficulties score, favoring the screen media reduction intervention (-1.67; 95% CI, -2.68 to -0.67; Cohen d, 0.53)", with the largest gains in internalising symptoms (−1.03) and prosocial behaviour (0.84). Pieh et al. (2025) measured depression, stress, well-being and sleep quality: their chosen outcomes moved — a meaningful causal result about time use and symptoms, and nothing about cognition.

Third, where cognition was measured, the result was null or transient. Pickard et al. (2024) is the cleanest test: it removed pre-bed screen use, replaced it with an active bedtime-box comparator that isolates the screen effect from the warm-interaction effect, and found that "there was no observable effect of PASTI on objective measures of attention". Its parent-reported differences (effortful control d −0.40, inhibitory control d −0.48) run the other way — the authors attribute them to score increases in the bedtime-box group — and the trial's own verdict is that "a future full confirmatory trial is needed before PASTI's adoption by parents and pediatricians". Sullivan et al. (2026) is the honest counterpart: at 21 days "one or both detox groups showed tentative improvements in sleep duration, daytime sleepiness, depressive symptoms, perceived stress and working memory relative to the controls, however, these changes were not maintained at a 2-month follow-up", in a pilot with 13% attrition and small arms.

The claim that survives this evidence is narrow and worth stating plainly: in the small set of randomised studies that have reduced recreational screen time, cognition has not improved.

Where the effects do appear

The randomised record is not empty. It is empty of cognitive gains and populated with three other kinds of effect.

Time substitution is large and reliable. Cutting recreational screen media for two weeks moved children's leisure non-sedentary activity by 45.8 minutes per day in the Danish trial. Roughly three quarters of an hour of sedentary leisure converted to movement: the strongest causal finding in this literature.

Sleep effects are real but conditional. Pickard et al. (2024) produced small-to-moderate gains in objectively measured sleep efficiency — "modest preliminary beneficial effects on sleep". Pieh et al. (2025) found a small improvement in sleep quality (η² = .048). But Pedersen et al. (2022), whose behavioural shift was the largest measured anywhere here, found no EEG sleep difference. Sleep is not an automatic conduit: whether cutting screens buys sleep depends on what replaces them and when the screens were used.

Symptom effects are the best-supported outcome in the whole field. The Danish secondary analysis found improved behavioural and emotional symptoms over two weeks, with the caveat that "more research is needed to confirm whether these effects are sustainable in the long term". Pieh et al. (2025) found reduced depressive symptoms and stress at three weeks and then "screen time increased rapidly after the intervention and at follow-up the values were once again approaching the initial level"; the authors nonetheless conclude that their data "suggest a causal relationship, rather than a merely correlative one, between daily smartphone screen time and mental health". Sullivan et al. (2026) found tentative gains in depressive symptoms and perceived stress. Modest effects that decay quickly — the same shape the wider literature reports for adolescent mental health.

That pattern — effects on sleep and symptoms, none on cognition, all short-lived — is this brief's central observation. What screen reduction actually does is rearrange the near term: more activity, sometimes more sleep, slightly better mood. It does not obviously rebuild anything structural in attention or executive function.

Why cognition is the least-measured outcome

The absence of cognitive outcomes in the randomised trials is not an accident of design; it reflects where the field's incentives point. Screen research grew out of paediatric obesity and mental health, where outcomes are cheap. Cognitive testing is expensive and produces numbers a general audience reads as a verdict on a child's intellect. There is also an asymmetry the other way: if a trial did find a cognitive improvement, a family's two weeks of less screen use changes sleep, activity, mood and parental behaviour at once, so any gain would be a composite a single trial cannot separate.

The result is an evidence gap with a clear shape: no randomised trial of recreational screen-time reduction has used a validated cognitive battery as a pre-registered primary outcome with device-logged exposure and follow-up beyond a few weeks. The work now under way does not close it: the PeerScreen protocol (Schmidt-Persson et al., 2026) is a four-week cluster-randomised trial in peer groups of 13- to 24-year-olds — smartphones to one hour a day, other screens to 14 hours a week — measuring mental health, sleep, physical activity, heart-rate variability and peer-group behaviour, but not cognition. Its framing of the field is telling: "causal evidence remains limited due to confounding and reliance on self-reported measures".

The observational literature: small, contested, and easy to over-read

Most of the supporting evidence for the pessimistic claim is correlational, and three recent syntheses converge on "small and complicated".

Liu et al. (2026), a systematic review and meta-analysis of 87 studies (154 datasets), found a weak negative correlation, with two findings that matter more than the average: absolute screen time had no predictive effect, while exceeding recommended thresholds impaired executive function; and passive viewing was negative while active, interactive use was not. Gender moderated the relationship, with weaker effects in girls.

Özcan and Kisbu-Sakarya (2025), a meta-analysis of 30 studies, 121 effect sizes and 7,740 children aged 0–6, found the overall relation between media exposure and executive function non-significant, with moderators setting the pattern: low-duration interactive media associated with better executive function, high-duration receptive media with worse.

Padmapriya et al. (2026) provide the broadest map yet: 121 systematic reviews (52 with meta-analyses), 2,790 unique primary studies and more than nine million participants. The quality assessment is the punchline — 39% of reviews were rated critically low quality, 32% low, and by GRADE 92% reached only low or very low certainty. The base was predominantly cross-sectional (94 reviews), with 83 including longitudinal studies and only 38 intervention or experimental. The findings should not be interpreted as causal.

Two further results sharpen the picture. On academic attainment, Paterna et al. (2024) found a correlation of r = −0.110 across 29 studies and 48,490 students — real, small, and about the size of a rounding error in most decision contexts. And when adolescents with problematic smartphone use were compared with others on executive function, Tauste-Garcia et al. (2025) reported a difference on parent-reported measures (d ≈ 0.53) but not on performance-based laboratory tasks — the signature of a reporting difference rather than a cognitive deficit.

Direction of causation is the other open question, and the longitudinal evidence refuses to settle it. Reid et al. (2026), a scoping review of 58 longitudinal studies, found at least one significant negative screen–executive-function relation in 81% of them (47 studies); 34 studies found earlier screen use predicting later executive function, but 9 found the reverse — earlier executive-function difficulties predicting later heavier screen use — and 4 found both. All eight neuroimaging studies found heavier screen use predicting change in brain regions implicated in executive function. Bidirectionality is the finding: children who find regulating hard may be the same children who scroll.

The methodological precedent is worth naming. Orben and Przybylski (2019) applied specification-curve analysis to three large datasets (n = 355,358) and found the association between digital technology use and adolescent well-being "negative but small, explaining at most 0.4% of the variation in well-being", with effects "too small to warrant policy change". That was well-being, not cognition, but the caution about effect sizes in large correlational datasets applies with equal force here.

Measurement: the exposure variable is broken

Before asking what screen time does to the mind, it is worth asking whether "screen time" is a real variable. It mostly is not.

Parry et al. (2021), a pre-registered meta-analysis of 106 effect sizes, found that "self-reported media use correlates only moderately with logged measurements, that self-reports were rarely an accurate reflection of logged media use and that measures of problematic media use show an even weaker association with usage logs". In 137 Czech adolescents comparing self-reports with device traces, Tkaczyk et al. (2024) found end-of-day reports were 32.2 minutes (14%) shorter than the trace and single-time estimates 55.2 minutes (18%) shorter; two-thirds of participants under-reported at least once, and within-person agreement for daily screen-time reports was r = 0.29, collapsing to r = 0.11 for phone-checking frequency. In 495 ABCD adolescents, Zhao et al. (2025) compared passive tracking with self-report: recorded social-media use was higher than past-year self-report (1.64 ± 1.93 h vs 1.44 ± 1.97 h) but close to post-sensing self-report (1.63 ± 1.93 h), and "higher picture vocabulary scores were associated with lower odds of under-reporting social media use (OR = 0.96, 95% CI: 0.93-0.99)". The measurement error is not random with respect to the outcome of interest.

The implication for causality is severe. If heavy users under-report their use, then any analysis of screen time and cognitive scores is partly regressing noise, and the noisier the exposure, the more likely the "effect" is an artefact of shared method variance between self-reported screen time and self-reported or school-assessed difficulties. The studies with the cleanest exposure — Pedersen et al. (2022), the ABCD tracking subsample — are also the ones that cannot deliver large samples.

Mechanisms: what would have to be true

For the pessimistic claim to be causally right, a specific chain must hold: screens displace sleep; displaced sleep degrades attention and executive function; the degradation is large enough to show up in test scores; and the effect survives the fact that most heavy screen use happens in the evening.

The sleep link is well supported in both directions. Chang et al. (2015) found that "participants reading an LE-eBook took longer to fall asleep and had reduced evening sleepiness, reduced melatonin secretion, later timing of their circadian clock, and reduced next-morning alertness than when reading a printed book". Perrault et al. (2019), tracking 12- to 19-year-olds with actigraphy for a month, found that "the reduction of screen time after 9 pm correlated with earlier sleep onset time and increased total sleep duration. The latter led to improved daytime vigilance" — though the design was a single-arm interventional phase, not randomised. In 43,755 US children, Twenge et al. (2019) found "associations with sleep duration were primarily due to portable electronic devices", and that "these results remained when demographic variables, diagnoses of anxiety or depression, physical activity, and BMI were included in the model". And the sleep-to-cognition link is quantitatively real: restricting healthy adults to four or six hours in bed for fourteen nights, Van Dongen et al. (2003) found, "resulted in significant cumulative, dose-dependent deficits in cognitive performance on all tasks", while "subjective sleepiness ratings showed an acute response to sleep restriction but only small further increases on subsequent days" — so subjective fatigue badly understates cognitive cost.

That chain is plausible but mostly short-range: sleep restriction produces measurable attention deficits within days, and recovery follows recovery. It does not obviously generate a durable generational decline in measured skill, and the field's own randomised evidence shows the sleep leg failing to appear even when the behavioural leg succeeded.

Two competing mechanisms deserve equal weight. Content and timing may matter more than duration: the meta-analytic evidence repeatedly finds passive viewing negative and interactive use neutral or positive, and Pickard et al. (2024) produced their null attention result by removing pre-bed use specifically, where the plausible harm channel is displacement rather than content. Hinten et al. (2025) make the same point in the laboratory, finding that "media pace does not affect children's cognition" (d = −0.123, 95% CI −0.331 to 0.086, p = 0.23) while fantastical content produced a small negative effect (d = −0.244), with heterogeneity so wide that the authors describe media exposure as impacting children "in some cases, negatively, in others positively, or not at all". And selection runs the other way: families who restrict screens may differ in ways that also predict child outcomes, which is why random assignment keeps mattering.

The strongest natural experiment: school phone bans

If randomised household trials are thin, the school-policy literature offers the largest experiments with objective outcome data. It also offers the most decisive answer to the cognitive question so far.

Allcott et al. (2026) evaluated lockable phone pouches using provider adoption records, independent GPS pings, teacher surveys, standardised test scores and administrative records, in a staggered difference-in-differences design. The first stage was strong: teacher reports put personal phone use in class falling from 61% to 13% of students, confirmed by GPS data. The downstream result was a precise null on achievement: a pooled test-score effect of −0.006 (SE 0.010) school-level standard deviations, about −0.003 student-level SD, precise enough to rule out average gains above roughly 0.007 student-level SD. High schools showed a modest positive effect (+0.026 school-level SD overall, +0.047 in mathematics), middle schools a small negative one (−0.025). Discipline rose in the adoption year and faded. Even scaling the estimate linearly to a policy eliminating in-school phone use entirely would imply gains of at most about 0.009 student-level SD.

This is a quasi-causal estimate of a school-hours restriction, not of household screen reduction, and the distinction is real: a school-day pouch removes the most disruptive window of phone use while leaving evening and weekend exposure untouched. But it is the best-identified evidence available for the proposition that removing screen access improves measured cognition in children, and it says: not detectably.

The rest of the school-policy literature is heterogeneous, and a companion brief in this archive covers it in detail. Two meta-analyses sharpen the pattern. Lai et al. (2025), reviewing 39 studies and 95,033 participants, found school interventions "modestly reduced screen time (SMD: -0.10, 95% CrI: -0.14, -0.06)" and "increased physical activity (SMD: 0.10)", with "no clear difference ... in BMI" and low certainty throughout (high risk of bias, I² of 85% and 88%). Celik et al. (2026), an arm-based meta-analysis of 17 studies, found school interventions effective at reducing screen time, inconclusive for problematic use, and advantageous for anxiety, depressive symptoms and life satisfaction. Population-level interventions move screen time a little and standardised achievement barely at all.

Competing explanations and what would change the answer

Three explanations survive contact with the evidence and should be carried forward rather than dismissed:

• The claim is about a small subgroup. Heavy, problematic, late-night use may harm cognition while ordinary use does not. The evidence is thin too — the problem-use literature is almost entirely cross-sectional, and Tauste-Garcia et al. (2025) found no difference on performance-based executive function — but it is the most plausible route by which a true effect hides inside a null average.
• The effects are real but transient. Both randomised symptom studies decay within weeks to months. A study measuring cognition right after an intervention could capture a genuine effect that a two-year follow-up would miss. Nothing in the record rules this out, and no trial has looked.
• Measurement may be diluting the signal. Under-reporting is not random: heavier users under-report more, and lower cognitive scores predict more under-reporting (Zhao et al., 2025). A true effect could sit in the data, attenuated by a factor no one has quantified.

I would raise my confidence in a small average causal effect on cognition if a multi-year cluster-randomised trial with device-logged exposure and a pre-registered cognitive battery shows a consistent gain across independent sites, in performance-based tasks rather than parent reports, persisting beyond the six-week window where every symptom effect in this literature has decayed. I would lower it further if a long-horizon trial with objective exposure and independent assessment returns a precise null, or if within-person analyses on device-logged panels show changes in logged recreational use do not predict within-person changes in validated attention. I would also revise upward if content-level randomisation — interactive versus passive content at matched duration — produced differential cognitive effects, since that would relocate the causal question from time to mode.

Best current answer and calibrated confidence

Best current answer. No randomised study has shown that reducing recreational screen time improves cognitive performance. The trials that reduced screen use well and measured cognition found a null (objective attention in toddlers, Pickard et al., 2024) or a transient effect (working memory in adolescents, gone at two months, Sullivan et al., 2026), while the same trials reliably moved physical activity, sometimes sleep, and often short-term symptom scores — near-term behavioural change with no demonstrated cognitive payload. The correlational literature carrying the pessimistic claim reports weak, heterogeneous, frequently bidirectional associations, measured with an unreliable exposure variable. The largest natural experiment with objective test scores — school phone bans with lockable pouches (Allcott et al., 2026) — finds average achievement effects near zero, with a bound tight enough to exclude gains above about 0.01 SD. The mechanistic story that would make the pessimistic claim true (screens displace sleep, sleep loss degrades attention) is real but short-range, and the one trial with a large behavioural shift did not find the sleep leg.

The honest reading is not "screens are fine". It is that the burden of proof has moved: the claim that screens cause cognitive decline is unproven, and the symmetric claim that cutting screens improves minds is unproven. Families optimising for sleep, activity and mental health have independent reasons to reduce evening and recreational use, and the randomised evidence supports those reasons directly; families optimising for test scores have been given none.

Calibrated confidence.

• ~85% that the average causal effect of reducing recreational screen time on measured cognition in children and adolescents is very small (|d| < 0.10) — not that it is exactly zero, but that no policy-relevant benefit is currently demonstrated.
• ~80% that the randomised literature, taken as a whole, does not support a claim that screen reduction produces cognitive gains.
• ~70% that sleep displacement is the most plausible mechanism for any genuine effect, and ~60% that it accounts for most of it, given that the largest behavioural shift produced no measurable sleep change.
• ~85% that observed correlations between screen time and cognitive scores are substantially inflated by measurement error, self-report artefact and reverse causation.
• ~60% that a well-powered multi-year randomised trial with device-logged exposure and validated cognitive outcomes would return a precise null for the average population. I am less confident here than elsewhere: the mechanism is plausible and the current trials are too small to bound a small effect.

Coverage limits. This synthesis rests on searches of Crossref, Europe PMC and the open web for randomised trials, meta-analyses and natural experiments published through October 2026, plus manual verification of each cited reference. It is not a formal systematic review: no dual screening, no protocol registration, no registry sweep beyond the protocols named here, no re-analysis of primary data. Where a claim rests on a single study — the school pouch estimates, the detox pilot — I have said so. Three unverifiable sources in an earlier draft were dropped rather than repaired, and five bibliographic errors were corrected.

The precise remaining gap, and why this brief stops here. No randomised trial of recreational screen-time reduction has pre-registered a validated cognitive battery plus academic attainment as primary outcomes, with device-logged exposure, follow-up beyond six months, and independent assessment. The two trials that measured cognition measured it in the wrong population at the wrong duration; the most powerful natural experiment measures a different treatment; and the registered work now under way does not measure cognition at all. The next material advance is not another explanation of existing findings — it is a test that does not yet exist, and it is a field trial rather than an analysis.

References

1. Allcott, H., Baron, E. J., Dee, T., Duckworth, A. L., Gentzkow, M., & Jacob, B. (2026). The Effects of School Phone Bans: National Evidence from Lockable Pouches. NBER Working Paper 35132. https://doi.org/10.3386/w35132

2. Böttger, T., Poschik, K., & Zierer, K. (2023). Does the Brain Drain Effect Really Exist? A Meta-Analysis. Behavioral Sciences, 13(9), 751. https://doi.org/10.3390/bs13090751

3. Chang, A.-M., Aeschbach, D., Duffy, J. F., & Czeisler, C. A. (2015). Evening use of light-emitting eReaders negatively affects sleep, circadian timing, and next-morning alertness. PNAS, 112(4), 1232–1237. https://doi.org/10.1073/pnas.1418490112

4. Celik, S., et al. (2026). The Impact of School-Based Interventions on Adolescents' Digital Use and Mental Wellbeing: A Systematic Review and Arm-Based Meta-Analysis. Adolescent Research Review, 11, 529–559. https://doi.org/10.1007/s40894-026-00287-5

5. Hinten, A. E., et al. (2025). Meta-Analytic Review of the Short-Term Effects of Media Exposure on Children's Attention and Executive Functions. Developmental Science, 28. https://doi.org/10.1111/desc.70069

6. Lai, N. M., et al. (2025). Review Article: The effectiveness of school-based interventions for reducing screen time — a systematic review and meta-analysis. Child and Adolescent Mental Health, 30, 223–237. https://doi.org/10.1111/camh.70022

7. Liu, C., Su, J., Wei, Y., Lin, S., Wu, X., Tang, Y., & Kan, P. (2026). Interplay between children's screen exposure and executive function: A systematic review and meta-analysis. Journal of Children and Media, 1–19. https://doi.org/10.1080/17482798.2026.2697013

8. Padmapriya, N., Chandrasekaran, M., Khan, S. F., Nishika, R., Huiyun, G. L., Chong, D., Sanabria, A. S., Raeside, R., Holly, L., Chia, A., Fogel, A., & Müller-Riemenschneider, F. (2026). Associations of digital media use with health, behavioural and developmental outcomes in children and adolescents: an umbrella review and meta-analysis. eClinicalMedicine, 104, 104225. https://doi.org/10.1016/j.eclinm.2026.104225

9. Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. Nature Human Behaviour, 3, 173–182. https://doi.org/10.1038/s41562-018-0506-1

10. Parry, D. A., Davidson, B. I., Sewall, C. J. R., Fisher, J. T., Mieczkowski, H., et al. (2021). A systematic review and meta-analysis of discrepancies between logged and self-reported digital media use. Nature Human Behaviour, 5, 1535–1547. https://doi.org/10.1038/s41562-021-01117-5

11. Paterna, A., Alcaraz-Ibáñez, M., Aguilar-Parra, J. M., Salavera, C., & Demetrovics, Z. (2024). Problematic smartphone use and academic achievement: A systematic review and meta-analysis. Journal of Behavioral Addictions, 13(2), 313–326. https://doi.org/10.1556/2006.2024.00014

12. Pedersen, J., Rasmussen, M. G. B., Sørensen, S. O., Mortensen, S. R., Olesen, L. G., Brønd, J. C., Brage, S., Kristensen, P. L., & Grøntved, A. (2022). Effects of Limiting Recreational Screen Media Use on Physical Activity and Sleep in Families With Children: A Cluster Randomized Clinical Trial. JAMA Pediatrics, 176(8), 741–749. https://doi.org/10.1001/jamapediatrics.2022.1519

13. Perrault, A. A., Bayer, L., Peuvrier, M., Afyouni, A., Ghisletta, P., Brockmann, C., Spiridon, M., Haller, D. M., Pichon, S., Perrig, S., Schwartz, S., & Sterpenich, V. (2019). Reducing the use of screen electronic devices in the evening is associated with improved sleep and daytime vigilance in adolescents. Sleep, 42(9), zsz125. https://doi.org/10.1093/sleep/zsz125

14. Pickard, H., et al. (2024). Toddler Screen Use Before Bed and Its Effect on Sleep and Attention: A Randomized Clinical Trial. JAMA Pediatrics, 178(12), 1270. https://doi.org/10.1001/jamapediatrics.2024.3997

15. Pieh, C., et al. (2025). Smartphone screen time reduction improves mental health: a randomized controlled trial. BMC Medicine, 23. https://doi.org/10.1186/s12916-025-03944-z

16. Rasmussen, M. G. B., et al. (2020). Short-term efficacy of reducing screen media use on physical activity, sleep, and physiological stress in families with children aged 4–14: study protocol for the SCREENS randomized controlled trial. BMC Public Health, 20. https://doi.org/10.1186/s12889-020-8458-6

17. Reid, C. A., Dhamankar, M., Marsh, S., Peterson, E. R., Erb, C. D., & Waldie, K. E. (2026). Have we found directionality? Longitudinal associations between screen use and executive function in children. A scoping review. Developmental Review, 80, 101266. https://doi.org/10.1016/j.dr.2026.101266

18. Schmidt-Persson, J., et al. (2024). Screen Media Use and Mental Health of Children and Adolescents: A Secondary Analysis of a Randomized Clinical Trial. JAMA Network Open, 7(7), e2419881. https://doi.org/10.1001/jamanetworkopen.2024.19881

19. Schmidt-Persson, J., et al. (2026). Effects of reducing screen media use on mental health, sleep, physical activity, heart rate variability, and behavioral patterns in peer groups of young people: a study protocol for a cluster-randomized controlled trial. BMC Psychology. https://doi.org/10.1186/s40359-026-04827-5

20. Sullivan, E. C., et al. (2026). Innovations in Practice: A pilot randomised trial on smartphone and social media abstinence — effects on sleep quality and psychological wellbeing in adolescents. Child and Adolescent Mental Health. https://doi.org/10.1111/camh.70118

21. Tauste-Garcia, I., Nieto-Ruíz, A., Moreno Padilla, M., Martín Tamayo, I., & Fernández-Serrano, M. J. (2025). Executive function in adolescents with problematic smartphone use. International Journal of Adolescence and Youth, 30. https://doi.org/10.1080/02673843.2025.2562950

22. Twenge, J. M., Hisler, G. C., & Krizan, Z. (2019). Associations between screen time and sleep duration are primarily driven by portable electronic devices: evidence from a population-based study of U.S. children ages 0–17. Sleep Medicine, 56, 211–218. https://doi.org/10.1016/j.sleep.2018.11.009

23. Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, D. F. (2003). The cumulative cost of additional wakefulness: dose-response effects on neurobehavioral functions and sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep, 26(2), 117–126. https://doi.org/10.1093/sleep/26.2.117

24. Zhao, Y., Han, X., Bagot, K. S., Tapert, S. F., Potenza, M. N., & Paulus, M. P. (2025). Examining measurement discrepancies in adolescent screen media activity with insights from the ABCD study. npj Mental Health Research, 4. https://doi.org/10.1038/s44184-025-00131-z

25. Özcan, M. Ş., & Kisbu-Sakarya, Y. (2025). Exposure to Interactive Media, but at Low Levels, Is Associated with Better Executive Functioning in Children: A Meta-Analysis. Media Psychology, 29(2), 109–141. https://doi.org/10.1080/15213269.2025.2471583