The Algorithm Ate My Childhood: Adolescent Brains, Behavioural Design and the Under-16s Ban
The UK's coming under-16s ban is mostly about access, but the harm sits in behavioural design aimed at developing brains, and new defaults won't redesign it.

📌 AI AND AI INTEGRATION
A fourteen-year-old posts a photo, puts the phone down, and picks it up again ninety seconds later to see how many likes it has collected. Then they do it again, and again, because the feed has taught them that the next refresh might be the one that pays out.
In June 2026 the UK government announced that under-16s will be banned from social media, with the rules due before Parliament by the end of the year and the ban expected to take effect in 2027, and the reaction has been predictable. Tech companies are lobbying, parents are cheering (the government says nine in ten parents backed a ban), and politicians are doing that thing where they sound decisive without quite understanding what they’re being decisive about.
The problem is that almost everyone is arguing about access, about whether kids should be on platforms or not, and I think that’s the wrong question. The better question is what the behavioural architecture of these platforms is doing to developing minds, and whether a ban would even address it.
I’ll save you the suspense, because I don’t think it would.
That isn’t an argument for doing nothing. The answer is to understand what’s happening at the level of cognitive architecture, developmental psychology and behavioural design, and then to build something better. This essay is an attempt to do that, using the STAR Framework as one lens for why a one-size-fits-all policy for young people misreads the problem it is trying to solve.
The Optimisation Engine in Your Pocket
Social media platforms are not neutral tools. They are optimisation engines designed to maximise engagement, and much of that engagement comes from working with the grain of our cognitive biases. You don’t need a conspiracy theory to see it, because it’s the business model. It shows in the design patterns, and it shows in the testimony of people who built these systems and later spoke out about them. Sean Parker, Facebook’s founding president, said in 2017 that the company knew from the outset it was exploiting “a vulnerability in human psychology”.
Every notification, every infinite scroll and every autoplay video is a behavioural intervention. The feed works on a similar principle to the variable rewards that make slot machines so compelling, because you never know which refresh will pay out, and researchers modelling more than a million social media posts found patterns consistent with reward learning. The social proof that makes you check what everyone else is doing is the same pull that sends that fourteen-year-old back to count the likes. And the first numbers a teenager sees, follower counts and likes, become the yardstick for everything after.
These platforms have, in effect, industrialised what we know about cognitive biases. They’ve taken the shortcuts our brains use to process information and turned them into engagement levers. For adults this is hard enough to manage, and the evidence suggests we’re not managing it particularly well either. For adolescents it’s a different conversation entirely.
It helps to look at the specific biases at play one by one. Confirmation bias, our pull towards what we already believe, meets an algorithm that learns to serve more of it, which can narrow a young person’s view before they’ve finished forming it. Availability heuristic means that whatever the algorithm surfaces most often comes to feel most common, most important and most true. Bandwagon effect means that content with high engagement signals gets treated as credible simply because others have engaged with it. Negativity bias means that outrage and conflict tend to travel further than nuance, and one large study found that posts about the political out-group were among the strongest drivers of shares.
Each of these biases is a feature of human cognition, and each is being exploited at industrial scale. Adolescents may be more exposed to these effects, because they are still developing the habit of noticing when their thinking is being shaped.
The Dual Process Problem
If you’re familiar with Dual Process Theory, you know the basics. System 1 is fast, intuitive and emotional, while System 2 is slow, deliberate and effortful. Healthy decision-making needs both systems working in concert, with System 1 generating the impulse and System 2 evaluating it.
Think of System 1 as the accelerator and System 2 as the brakes. Both are necessary, because the accelerator gets you moving and the brakes keep you alive.
Social media platforms are designed to keep you in System 1 as much as possible. There’s the small reward of a notification, the emotional pull of a viral video, the urgency of a disappearing story and the autoplay that removes the need to make an active choice at all. Each of those design choices nudges you past deliberation and towards an immediate response, because the platform would much rather you scrolled than stopped to think.
For adolescents, this creates a developmental problem that goes well beyond distraction, because the capacity for slower, deliberate thinking is still developing. Brain regions involved in impulse control and planning, including the prefrontal cortex, keep developing into the twenties, although different systems mature at different times and there is no single age at which a brain is “done”. That is normal development rather than a deficiency, and it’s part of the reason UK law stops under-18s buying alcohol or sitting on a jury, and in England and Wales shields them from most binding contracts. We recognise in other areas of life that adolescent decision-making is a work in progress, and yet adolescents are being placed inside behavioural environments designed to exploit the very cognitive architecture they haven’t finished building.
We’ve handed them a remarkably sophisticated piece of behavioural technology and said, “Here, figure it out.”
Heavier social media use in adolescence has been linked with poorer sleep and, less consistently, with anxiety and attention problems. Most of that research is correlational and many of the effects are small, and across more than 350,000 adolescents one analysis found that technology use explained at most 0.4% of the variation in wellbeing. So it can’t show that the platforms cause these problems, but it fits what you would expect from an environment built to keep people in System 1.
Imagine handing a sixteen-year-old the keys to a car and then removing the brakes. In a sense that’s what we’ve done, except the car is an algorithm and the destination is “maximum time on platform.”
What the Adolescent Brain Is Building
Some biological context helps here, because the policy conversation tends to treat adolescence as a social construct rather than a developmental reality.
The adolescent brain goes through a process called synaptic pruning, in which unused neural connections are eliminated and frequently used ones are strengthened. Pruning is especially pronounced in the prefrontal cortex and is shaped by experience, which is how the brain becomes more efficient, and it also means the experiences a teenager has play a part in shaping the brain they take into adulthood.
US teenagers averaged 4.8 hours a day on social media in a 2023 Gallup survey. If use helps tell the brain what to keep, it’s worth asking which habits hours of fast, reactive scrolling are rehearsing. We don’t yet know how much this shapes the adult brain, but it’s a fair question to ask of any environment that takes up that much of a developing life.
The hippocampus, which is critical for memory formation and contextual understanding, is also developing during adolescence. Whether a fragmented stream of fifteen-second clips, memes and headlines without articles affects how teenagers form memories is an open question, and the long-term research doesn’t exist yet.
The amygdala, which is central to how the brain responds to threat and emotion, stabilises earlier than the prefrontal cortex, and adolescents are especially sensitive to social evaluation such as rejection, exclusion and criticism. Social media, full of comparison, status signals and subtle exclusion cues, keeps putting teenagers in exactly the situations they are most sensitive to. One US study found that teenagers who habitually checked social media showed changes over time in how their brains responded to social feedback, though it couldn’t show that the checking caused them.
That is the biological context for the behavioural design problem. We’re talking about more than “screen time”, because this is the environment in which a brain is developing.
Young People Are Not a Monolith
This is where most of the policy conversation falls apart. The debate treats young people as a single category, “kids on phones”, in the same way marketing flattens “the consumer” as though millions of people share one brain. They don’t, and human beings never have been a monolith.
The STAR Framework is built on seven psychological pillars. In STAR’s model, which is my own framework and an untested interpretation rather than established science, people lean towards one of four broad motivational types, each organised around a different core need. Three of those needs, relatedness, competence and autonomy, come from Self-Determination Theory, which names those three. The fourth, security, is STAR’s own addition rather than part of SDT. I use the four types here as lenses on needs, describing decision-making styles rather than kinds of teenager, and I think looking through each in turn helps explain why the same technology can mean very different things to different young people.
Through the Socialiser lens, which centres on relatedness, connection and belonging, social media is social infrastructure rather than entertainment. It’s where friendships are kept going and where you stay in the loop and feel included. Seen that way, taking it away does more than reduce screen time, because it removes the village square for a young person whose social world runs through it. The need for belonging doesn’t go away with a ban, and the ban just removes the most accessible way to meet it.
Through the Thinker lens, which centres on competence, the drive to understand, learn and build capability, social media looks like a library that happens to have a slot machine bolted on. Tutorials, experts, long threads on niche interests and the chance to benchmark your understanding against others all live there, so removing the platform removes a significant learning resource. What this need calls for is better curation and fewer engagement traps rather than less access.
Through the Adventurer lens, which centres on autonomy, the drive to explore, create, express identity and test boundaries, social media is a stage. The emphasis falls on producing rather than consuming, on making videos, writing posts, curating feeds and pushing creative boundaries. A ban removes creation as well as consumption, and it tells a young person in the middle of discovering who they are that one of the most accessible creative outlets they have is now off-limits.
Through the Realist lens, which centres on security, stability and reliability, along with simplicity and ease, social media can work as a radar. It’s a way to check what’s happening in a peer group, understand expectations and reduce uncertainty, so removing it can leave that need for stability with less to go on.
Read this way, a blanket ban doesn’t land on these four needs equally. Through the Socialiser lens it looks like isolation, through the Adventurer lens like censorship and through the Thinker lens like resource removal. Through the Realist lens it’s the loss of the radar, which may raise anxiety unless something better replaces it, and the ban on its own doesn’t provide that.
You cannot design effective policy for “young people” as a single block. You can only design it for the different needs young people bring to the world, and until the policy conversation acknowledges that, every intervention risks being a blunt instrument applied to a problem that needs precision.
The Real Conversation: Behavioural Architecture
The question I’d rather we asked isn’t “should under-16s be on social media?” but “what would social media look like if it were designed for adolescent wellbeing rather than adult engagement metrics?”
This is a behavioural design problem, and it has solutions that don’t depend on a ban.
Regulatory Focus Theory tells us that people operate in two modes, promotion focus (seeking gains and pursuing ideals) and prevention focus (avoiding losses and meeting obligations). I’d argue social media is almost entirely promotion-focused, with its whole gamification structure built around accumulation and growth, more followers, more likes, more content and more engagement. For adults this can feel compulsive, and for adolescents still working out what matters to them it risks making the pursuit of “more” the organising principle of their social life.
What if the default mode for adolescent accounts were prevention-focused, so that the platform actively helped young people avoid harm rather than endlessly optimising for engagement? I don’t mean parental controls, which many adolescents resent and some route around, but design that makes the safe path the easy path.
Self-Determination Theory identifies three fundamental psychological needs: autonomy, competence and relatedness. Social media satisfies relatedness (sort of) and competence (through comparison, which is a toxic form of competence feedback), but in my view it undermines autonomy. The algorithm decides what you see, the recommendation engine decides what you want and the notification system decides when you engage, while autoplay removes even the minimal autonomy of choosing to watch the next video.
For adolescents who are still building their sense of autonomy, this is corrosive, because they’re learning that their choices are made by systems rather than by them. What if platforms were required to give young users genuine control over their experience, through agency built into the design rather than parental controls imposed from above? What if a teenager could actually choose what the algorithm shows them, and understand why it’s showing them what it shows?
The Big Five personality model reminds us that adolescence is a period when personality traits are still changing. High neuroticism is linked with a stronger habit of social comparison and with more problematic social media use, and social media is essentially a social comparison engine. High openness makes a teenager more likely to explore, which feeds the algorithm more varied signals, but it does not, on current evidence, make them more prone to extreme or conspiracy content. If anything, a systematic review of radicalisation research found lower openness went with more radical attitudes, and a meta-analysis found no link between openness and conspiracy belief. Agreeableness, along with conscientiousness and emotional stability, is associated with more conformity. High conscientiousness may actually be protective, since lower conscientiousness is linked with more problematic use, but platform design works against conscientious behaviour such as deliberate use, time management and intentional engagement.
So what if young people had real controls over the features that bear on those differences, such as the ones that invite comparison, and could set them to suit themselves? That would respect personality variation without asking a platform to guess a teenager’s traits and treat them as an engagement score.
The Ban Is a Symptom of Design Failure
The government’s own case for the ban is that “tech giants had their chance and failed”. That’s not surprising, because self-regulation in engagement-maximising industries, from tobacco to alcohol to junk food, has a poor track record. The pattern I see repeated is to build the optimisation engine, extract maximum value, wait for regulation, argue that regulation is disproportionate, comply minimally and find workarounds.
Internal documents have shown what some companies knew. Meta’s own research, reported in the Facebook Files, found that Instagram made body-image issues worse for one in three teen girls who already had them, a reading Meta disputed. TikTok’s documents, revealed in a US state lawsuit, showed it knew compulsive use could harm teenagers. In March 2026 a Los Angeles jury found Meta and YouTube negligent in the first social media addiction case to reach trial, a verdict both companies said they would appeal. The research on YouTube’s recommendations is more mixed, with one audit finding users drifting towards more extreme channels, another finding a modest ideological drift and a third finding that relying on the recommender led to less partisan viewing on average, but the pattern in the companies’ own files is hard to dismiss.
The ban also shows a failure of imagination. The government has added some design rules: alongside the ban it plans to block livestreaming and contact from strangers for under-16s across more services, including gaming, and in July it announced a default midnight to 6am curfew for 16- and 17-year-olds, with features that keep people scrolling switched off by default. But the centrepiece is still an access ban for under-16s, and the design rules are defaults rather than a rethink of how these platforms work.
We don’t ban children from swimming pools. We design pools with shallow ends, lifeguards and depth markings, and build the safety architecture into the environment. We don’t hand a five-year-old the keys to a car, but we do let them ride a bicycle, and we give the bicycle training wheels. The principle is consistent: adapt the environment to the developmental stage, not the other way round.
Social media needs the same approach, the technology to do it exists, and what’s been missing is the will.
What Behavioural Architecture Would Actually Look Like
If we took the behavioural science seriously, adolescent social media would look very different. Here’s what that might mean in practice.
Time-aware design. Screen time limits imposed by parents tend to feel like surveillance and control and so undermine autonomy, whereas platforms could actively surface the passage of time. “You’ve been scrolling for 45 minutes” is paternalistic, but a design that naturally creates pause points, moments where the brain can shift from System 1 to System 2, is architecture. Think of the services on a motorway, because nobody bans driving and we simply build in places where people can stop, reflect and choose whether to carry on.
Bias-interruption layers. Platforms already estimate users’ ages from their behaviour, imperfectly, and where a platform has reason to think it’s serving a teenager it could introduce deliberate friction at the points where cognitive bias exploitation is strongest: before a social comparison moment, before a notification triggers FOMO and before autoplay loads the next video. The aim is pausing rather than blocking, creating the space for System 2 to show up, like an “Are you sure?” dialogue that is designed into the experience rather than bolted on as an afterthought.
Needs-responsive design. Rather than sorting teenagers into types, design for the needs STAR’s lens highlights, which is an untested way of looking at it, and let young people choose. That would mean real connection tools rather than engagement bait, depth and curated learning paths rather than endless rabbit holes, creative tools rather than consumption loops, and clarity rather than anxiety-inducing ambiguity. Personalisation engines already segment users by behaviour, so the question is whether they segment for wellbeing or for engagement.
Transparent algorithms. If the recommendation engine is going to shape what a developing mind sees, that engine should be visible. I don’t mean open source, which is a different conversation with different risks, but explainable. “We’re showing you this because you engaged with similar content” is a basic design principle that somehow doesn’t apply to one of the most influential information systems we have. For adolescents, who are still learning how information systems work, that transparency matters even more, because making the algorithm visible is itself educational.
Developmental calibration. The platform experience should change as the user matures, because a thirteen-year-old and a seventeen-year-old are at different developmental stages and the platform should adapt. Age checks are often easy to get round in practice, so the experience could also respond to how someone is actually using it. If a user is stuck in a System 1-dominant pattern, the platform can introduce more friction, and if they are using it deliberately, it can ease off the paternalistic interventions.
The Unoptimised Adolescent
What makes me uneasy is that the UK’s social media ban for under-16s is a policy that, for all its consultation, still treats under-16s as one group, in response to a problem that grew while platforms operated without meaningful behavioural design standards.
Bans are easy to get round, and Australia shows how. Three months in, more than eight in ten under-16s in one study were still using the restricted platforms, mostly on their own accounts, and age checks were often a simple question about age. When the UK brought in age checks for adult sites in July 2025, VPN apps briefly topped the App Store charts, although in the Australian study only about 3% of teenagers used one. So access drifts back and the design patterns persist, and some of the teenagers who work around a ban may be the ones most hooked, which is exactly who it was meant to protect.
What would work is harder, slower and less politically satisfying, which is redesigning the behavioural environment itself. That means building platforms that treat adolescent psychology as a design constraint rather than an exploitation opportunity, and applying the science we already have, from Dual Process Theory to Self-Determination Theory to the Big Five, to create digital spaces that develop young minds rather than harvest them. The technology is ready and so is much of the science, and what’s been missing is the recognition that this is a design problem at all.
Maybe that’s changing, and maybe the UK’s ban, flawed as it is, will push the conversation toward the real question, which is not whether young people should be online but what kind of online world we’re willing to build for them.
Go back to that fourteen-year-old refreshing for likes every ninety seconds. Taking the phone away doesn’t remove the pull of the next refresh, because that pull was designed. The algorithm didn’t eat their childhood, bad behavioural design did, and the fix is better architecture rather than a ban on its own.
David Chadderton is a behavioural strategist, Chief Marketing Officer and STAR Dynamics. He writes The Unoptimised Human.
References
Allyn, B. (2024, 11 October). TikTok knows its app is harming kids, new internal documents show. NPR.
Alonzo, R., Hussain, J., Stranges, S., & Anderson, K. K. (2021). Interplay between social media use, sleep quality, and mental health in youth: A systematic review. Sleep Medicine Reviews, 56, 101414.
Arain, M., Haque, M., Johal, L., Mathur, P., Nel, W., Rais, A., Sandhu, R., & Sharma, S. (2013). Maturation of the adolescent brain. Neuropsychiatric Disease and Treatment, 9, 449-461.
Barnes, C., Hall, A., Mantach, S., Oldmeadow, C., Attia, J., Backholer, K., Williams, C., Tefera, Y., Kay, F., & Wolfenden, L. (2026). Assessing early effects of Australia’s Social Media Minimum Age Act on adolescents’ social media use: Observational study. BMJ, 393, e363695. See also the BMJ Group press release (24 June 2026).
Bawden, A. (2026, 24 June). Four in five under-16s in Australia using social media despite ban, study shows. The Guardian.
BBC News. (2025, 28 July). VPNs top App Store charts as UK age verification kicks in.
Department for Science, Innovation and Technology. (2026, 15 June). Social media to be banned for under-16s in landmark government move to give kids their childhood back. GOV.UK.
DeYoung, C. G., Peterson, J. B., & Higgins, D. M. (2002). Higher-order factors of the Big Five predict conformity: Are there neuroses of health? Personality and Individual Differences, 33(4), 533-552.
Dong, S., & Yang, A. (2026, 25 March). Jury finds Meta and YouTube negligent in landmark lawsuit on social media safety. NBC News.
Gibbons, F. X., & Buunk, B. P. (1999). Individual differences in social comparison: Development of a scale of social comparison orientation. Journal of Personality and Social Psychology, 76(1), 129-142.
Goreis, A., & Voracek, M. (2019). A systematic review and meta-analysis of psychological research on conspiracy beliefs: Field characteristics, measurement instruments, and associations with personality traits. Frontiers in Psychology, 10, 205.
Haroon, M., Wojcieszak, M., Chhabra, A., Liu, X., Mohapatra, P., & Shafiq, Z. (2023). Auditing YouTube’s recommendation system for ideologically congenial, extreme, and problematic recommendations. PNAS, 120(50), e2213020120. Journal version.
Horta Ribeiro, M., Ottoni, R., West, R., Almeida, V. A. F., & Meira, W. (2020). Auditing radicalization pathways on YouTube. Proceedings of FAT* 2020.
Hosseinmardi, H., Ghasemian, A., Rivera-Lanas, M., Horta Ribeiro, M., West, R., & Watts, D. J. (2024). Causally estimating the effect of YouTube’s recommender system using counterfactual bots. PNAS, 121(8), e2313377121. Journal version.
House of Commons Library. (2026, 3 November). Proposals to ban social media for children (CBP-10468).
Lindström, B., Bellander, M., Schultner, D. T., Chang, A., Tobler, P. N., & Amodio, D. M. (2021). A computational reward learning account of social media engagement. Nature Communications, 12, 1311.
Maza, M. T., Fox, K. A., Kwon, S.-J., Flannery, J. E., Lindquist, K. A., Prinstein, M. J., & Telzer, E. H. (2023). Association of habitual checking behaviors on social media with longitudinal functional brain development. JAMA Pediatrics, 177(2), 160-167.
Meta. (2021, 26 September). What our research really says about teen well-being and Instagram.
Meynadier, J., Malouff, J. M., Schutte, N. S., & Loi, N. M. (2024). Meta-analysis of associations between five-factor personality traits and problematic social media use. Current Psychology, 43(27), 23016-23035.
Mills, K. L., Goddings, A.-L., Clasen, L. S., Giedd, J. N., & Blakemore, S.-J. (2014). The developmental mismatch in structural brain maturation during adolescence. Developmental Neuroscience, 36(3-4), 147-160.
Moodie, R., Stuckler, D., Monteiro, C., Sheron, N., Neal, B., Thamarangsi, T., Lincoln, P., & Casswell, S. (2013). Profits and pandemics: Prevention of harmful effects of tobacco, alcohol, and ultra-processed food and drink industries. The Lancet, 381(9867), 670-679.
Ofcom. (2026, 29 October). How age checks at age 16 can work to underpin UK’s social media ban.
Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. Nature Human Behaviour, 3(2), 173-182.
Rathje, S., Van Bavel, J. J., & van der Linden, S. (2021). Out-group animosity drives engagement on social media. PNAS, 118(26), e2024292118.
Rothwell, J. (2023, 13 October). Teens spend average of 4.8 hours on social media per day. Gallup.
Selemon, L. D. (2013). A role for synaptic plasticity in the adolescent development of executive function. Translational Psychiatry, 3, e238.
Solon, O. (2017, 9 November). Ex-Facebook president Sean Parker: site made to exploit human ‘vulnerability’. The Guardian.
Somerville, L. H. (2016). Searching for signatures of brain maturity: What are we searching for? Neuron, 92(6), 1164-1167.
UK legislation: Juries Act 1974, s1; Minors’ Contracts Act 1987; Sale of Goods Act 1979, s3; GOV.UK, Alcohol and young people.
Vincent, J. (2021, 15 September). Instagram internal research: ‘We make body image issues worse for one in three teen girls’. The Verge, reporting the Wall Street Journal’s Facebook Files.
Wolfowicz, M., Litmanovitz, Y., Weisburd, D., & Hasisi, B. (2021). Cognitive and behavioral radicalization: A systematic review of the putative risk and protective factors. Campbell Systematic Reviews, 17(3), e1174.
Eight minutes. Thirty-three questions.
A precise picture of how you operate: motivations, decision style, and blind spots. Free to start.
- S
- T
- A
- R
The canonical source.
Essays, when there is something worth sending. Choose the subjects you want. I will ask you to confirm the address before I write.