Brain-Training Apps, Audited

The pitch is seductive: ten minutes a day, and your memory, attention, and reasoning stay sharp. The science is less accommodating. Commercial brain-training games reliably make you better at themselves — and that is where the benefit ends. This page walks the transfer problem, the big trials, and what the honest verdict leaves you with.

The Promise the Ads Make

The brain-training industry grew into a multi-billion-dollar category on a specific claim: short daily sessions improve general cognitive function — memory, attention, processing speed — in ways that matter in real life. In 2016 the U.S. Federal Trade Commission fined Lumosity, the category's biggest name, $2 million for advertising exactly that without adequate evidence. The settlement was not about whether the games do anything; it was about the gap between what the studies showed and what the ads said. That gap has a technical name, and it is the entire story of this page: transfer. One note before the evidence: none of this says the games are harmful or that their users are gullible. It says the gap between the studies and the sales copy is wide, measured, and worth knowing before you spend either money or expectations.

The Transfer Problem, Defined

Cognitive-training research distinguishes near transfer — improvement on tasks that resemble the training — from far transfer — improvement on different, real-world demands. Everyone agrees training produces near transfer: practice a digit-span game and your digit span improves. The commercial claim is far transfer: that the game upgrades the underlying machinery, which then runs everything better. For that to be true, the trained task would have to exercise some general mental capacity the way squats build general leg strength. The evidence says cognition does not work that way: skills are specific, and the rewiring machinery reshapes the circuits you use, not a generic "brain" — the lesson of the bus-driver studies arriving in app form.

The 11,000-Person Test

The decisive experiment recruited 11,430 adults through a BBC science program, randomized them to six weeks of online training in reasoning, memory, planning, or visuospatial skills, and tested the whole battery before and after (Owen et al., Nature, 2010). Results, carefully stated: everyone improved on the trained tasks — and almost nobody improved on anything else, including untrained tasks in the same domain. The improvement curves were training-curves, not brain-upgrades. This was the largest study of its kind, and its design answers the two classic criticisms of smaller studies — tiny samples and unblinded endpoints — which is why the field treats it as the benchmark result.

The Meta-Analysis Era

Single studies disagree; meta-analyses pool them. Two are worth naming. Melby-Lervåg and colleagues (2016) pooled the working-memory-training literature and found no evidence that training working memory improves intelligence or other far-transfer outcomes — gains appeared only on trained and closely related tasks. Sala and Gobet (2019) reviewed the broader cognitive-training field and reached the same conclusion with different methods: no enhancement of general cognition. The pattern also holds in domains people find surprising — chess instruction, for example, does not reliably raise academic or cognitive skills beyond the chessboard (Sala & Gobet, Educational Research Review, 2016). The specificity is the rule, not the exception. Pooling matters here because the individual studies are small and noisy: single trials have claimed everything from large far transfer to none at all, and the pooled answer has stayed consistent — the gains stay close to the task.

The Claims Ledger

Translated into the ledger the marketing department never publishes, the category's claims line up against the evidence like this. One rule of thumb while you read it: any claim that a game prevents dementia, raises general intelligence, or sharpens everyday memory is asking you to accept far transfer — the thing the trials did not find.

The Honest Exception: Structured Training in the Lab

One careful program keeps the door open a crack. The ACTIVE trial randomized 2,832 older adults to ten sessions of structured, supervised training in memory, reasoning, or processing speed — with progressive difficulty and human coaching, not a download (Ball et al., JAMA, 2002). A decade later, the reasoning and speed groups still showed small, task-family gains, and some measures of everyday function looked better than controls (Rebok et al., Journal of the American Geriatrics Society, 2014). Read it honestly: the effects were modest, specific to the trained skill families, and delivered by a format — supervised, escalating, sustained — that commercial apps do not use. It is evidence that training can be engineered well, not that games transfer. The active ingredients were escalation — difficulty that rose with performance — supervision, and booster sessions spread across years: none of which a downloaded game provides.

What to Do Instead

• 🎻 Learn something hard and real. Language, instrument, dance, craft — skills that combine novelty, difficulty, and feedback over months, which is what the novelty principle documents.
• 🏃 Invest in the replicated levers. Exercise and sleep have larger, better-evidenced effects on cognition than any app — the subject of the next-but-one page.
• 🧠 Know what the game is for. If an app keeps you engaged daily, that is a habit win, not a cognitive one — value it accordingly.
• 🔎 Check claims against regulators. A marketing claim that a product sharpens the brain is the exact gap the FTC settlement addressed; skepticism here is warranted.

Questions, Answered Briefly

• 📈 Isn't some transfer better than none? Near transfer is real and the game delivers it — the error is paying for the far-transfer claim. Value the product at what it does, not what it promises.
• ⚠️ Do the games hurt? There is no evidence of harm. The honest cost is opportunity: minutes a day that exercise, sleep, or a real skill would have spent better.
• 🧘 What about meditation apps? A different category with a different evidence base — attention and stress training, not cognitive transfer. The meditation protocol owns that literature.
• 💬 What do I tell someone who loves theirs? Enjoy it as leisure, the way you would chess or crosswords — and if the goal is brain health, add the levers from the novelty principle and the helpers.

Related Topics