Ask a room whether social media is bad for us and you will get a fight about mood. Depression rates. Anxiety scales. Suicide trends. Whether the effect size is enormous or laughable.
That fight has run for years, and the evidence under it is thinner than either side admits. One well-known analysis by Amy Orben and Andrew Przybylski found the effect of digital technology on well-being roughly equivalent to that of eating potatoes — smaller than the effect of wearing eyeglasses or eating breakfast.
Correlational findings are also miserable to interpret. Depressed girls may use more social media; or some third trait may drive both. As Paul Bloom summarizes the literature, many investigators believe the relationship is either absent or negligible.
The mood fight is a distraction; the deeper risk is that platforms shift values and train abandonment rather than persistence.
Bloom adds a further caution, borrowing Eric Turkheimer's "gloomy prospect": individual behavior is caused by a vast network of interacting causes that differ person to person and loop back on each other. It sucks to be a social scientist.
But the mood fight is a distraction. The more consequential finding is that social media may be doing something to us that no depression scale would catch.
New research suggests these platforms are shifting our actual values — tilting them toward achievement and conformity, according to work from Ertugrul Uysal at ETH Zurich, circulated by the Society for Personality and Social Psychology. A bad mood is a state you recover from. A shifted value is a preference you carry into everything else.
Now set that beside a second paper: an effort-recalibration framework published in Nature Human Behaviour. Its premise is that the brain runs as a continuous cost-benefit calculator, constantly weighing expected reward against subjective effort cost.
Digital platforms skew that market. Infinite scroll, notifications, algorithmic recommendations, likes, short video — all of it lowers entry friction while maximizing immediate reward. Exploration becomes cheap. Exploitation — staying with one demanding thing long enough to extract its delayed payoff — becomes expensive by comparison.
This resolves a stubborn puzzle. In labs, people perform fine on attention tests. Give them structure and clear stakes and the engine works perfectly. The authors' read: nothing is broken. What changes is what people choose when left to their own devices.
Through repetition, the subjective weight of mental exertion inflates. Slow, awkward beginnings — a dense first page, a new instrument, a hard problem — get flagged by the decision system as bad deals. The threshold for switching drops.
The book does not get harder. It just starts to feel like a worse bargain than the phone.
That alone would be enough to worry about. But there is a darker piece, and it kills the assumption most of us carry about why people consume what they consume.
The standard model of media selection is self-serving. You feel bad, you reach for something that makes you feel better. That model is incomplete.
An expanding body of evidence suggests people — including those with psychological vulnerabilities — actively seek out content that aggravates distress rather than relieves it. When negative self-concepts are salient, users gravitate toward content that validates those self-views, even at the expense of their well-being, according to a preprint that extends the Selective Exposure Self- and Affect-Management model to include self-defeating motivations.
Mental health content is the clearest case, because the dynamics are visible and the harm is concrete. The authors argue the same process operates in other domains.
Put the three findings side by side and a coherent picture forms. Platforms reprice effort downward. That repricing reshapes what people value. And the content people select in their worst moments is often the content that confirms the worst things they believe about themselves.
None of this requires smartphones to be uniform psychological poison. The effort framework explicitly resists that framing. A phone can host a long-form essay or a language app as easily as a scrolling feed. The critical variable is the effort-and-reward architecture of the specific application: does it reward deliberate goal pursuit, or goal-free rapid switching?
That distinction is also where the policy conversation has gone wrong. The American Psychological Association, responding to the recent Meta settlement, argues that age limits are one piece alongside developmentally appropriate design, ongoing adult guidance, and social media literacy — a shared responsibility among platforms, policymakers, families, and young people.
Critics of that position are not wrong to be suspicious. Read the replies and you find the argument that age limits and design tweaks are a public relations move, or a cop-out that dumps safety onto families while multi-billion-dollar companies keep engineering the loop. That objection has force.
But it tends to collapse into "ban the screens," which attacks the wrong variable. If the mechanism is effort valuation and self-defeating selection, removing access is a blunt instrument aimed at a design problem. The framework offers a sharper question: does this environment reward persistence, or train abandonment?
Bloom's own position is instructive. He is skeptical of the causal case, and he resists "better safe than sorry" reasoning on