The short version
- A review of every peer-reviewed study cited by leading brain-training companies found extensive evidence of improvement on trained tasks, less on closely related tasks, and little evidence of improvement in everyday cognitive performance.
- None of the studies the brain-training industry cites met all of the methodological best practices the reviewers identified as necessary to draw conclusions about everyday benefit.
- A meta-analysis of 87 publications and 145 experimental comparisons found no convincing evidence that working memory training transfers to nonverbal ability, verbal ability, word decoding, reading comprehension or arithmetic when compared with treated controls.
- In that meta-analysis, how much a participant improved on working memory measures was unrelated to how much far transfer they showed, which undercuts the mechanism the training is sold on.
- The ACTIVE trial of 2,832 older adults found each of its three in-person training programmes improved its own targeted ability but detected no effect on everyday functioning at two years.
In 2014 more than seventy scientists signed an open letter saying brain games do not provide a scientifically grounded way to improve cognition or hold off decline. A few months later 133 scientists and practitioners signed a letter saying the literature is full of demonstrations that they do.
Same field, same studies, opposite conclusions. A team of seven researchers then did the obvious thing nobody had done: they went to the website hosting the second letter, took the list of peer-reviewed studies the brain-training companies cite as their own best evidence, and evaluated every one of them against a defined set of methodological standards.
What the companies' own evidence shows
Three findings, in descending order of strength.
There is extensive evidence that brain-training interventions improve performance on the trained tasks. There is less evidence that they improve performance on closely related tasks. There is little evidence that training improves performance on distantly related tasks, or that it improves everyday cognitive performance.
And the methodological verdict: many of the published studies had major shortcomings in design or analysis that preclude firm conclusions, and none of the cited studies met all of the best practices the reviewers identified as necessary to draw conclusions about everyday benefit.
That is not a hostile reading of a hostile sample. It is a review of the evidence the industry itself nominated.
Near transfer, far transfer, and why the distinction is the whole argument
Getting better at the thing you practise is called near transfer and it is completely real. You will improve at the n-back task, the visual search, the digit span. That improvement is measurable, durable, and worth nothing on its own — it is the same phenomenon as getting better at crosswords.
far transfer is the claim that matters: does training a narrow skill make you generally better at thinking? A meta-analysis of working memory training pooled 87 publications and 145 experimental comparisons. Immediately after training there were reliable improvements on intermediate measures — verbal and visuospatial working memory, the things closest to what was practised. On far-transfer measures — nonverbal ability, verbal ability, word decoding, reading comprehension, arithmetic — there was no convincing evidence of improvement when training was compared with a treated control group.
The mediation analysis is the detail that ends the argument. Across studies, how much someone improved on the working memory measures was not related to how much far transfer they showed. If the training worked the way the marketing says, those two numbers would move together. They do not.
The strongest exception, and what it actually was
There is one large trial worth taking seriously, and it is not an app. ACTIVE randomised 2,832 independently living adults aged 65 to 94 to ten sessions of in-person group training in memory, reasoning or speed of processing, or to no contact.
Each intervention improved its own targeted ability, durably, out to two years: 87% of the speed group, 74% of the reasoning group and 26% of the memory group showed reliable improvement immediately afterwards. And then the sentence that gets left out: no training effects on everyday functioning were detected at 2 years.
The ten-year follow-up is where the interesting result lives. At a mean age of 82, participants in each training group reported less difficulty with instrumental activities of daily living than controls — effect sizes 0.48 for memory, 0.38 for reasoning, 0.36 for speed — with roughly 60% of trained participants at or above their baseline self-reported function versus 50% of controls. The reasoning and speed groups still showed gains on their targeted abilities. Memory training effects on memory performance were no longer maintained.
So the best case for cognitive training is: ten sessions with a trainer, in a randomised trial of nearly three thousand people, producing a self-reported difference in daily function a decade later. That is a genuine, hard-won result. It is also not what a subscription app is selling you.
The pattern this shares with the supplement aisle
You have seen this shape before if you have read anything else on this blog. An intervention produces a real, narrow, measurable change; the change is then described using a much broader word; and the broader claim is the one that gets sold. It is the same move that turns a millisecond on a speed test into "clarity" and the same reason the word nootropic is doing so much unearned work.
It also has an exact parallel in the supplements and ADHD literature, where cognitive training looked effective when scored by people who knew what was going on and dropped to non-significance under blinded assessment. Expectation is a real effect. It is just not the effect you are paying for.
What we'd actually tell you
If you enjoy the games, play them. They are cheap, harmless entertainment, and enjoying something you do daily is not nothing. Just do not buy a subscription in the belief that it is protecting your future cognition, because the evidence the companies themselves put forward does not support that.
If the goal is genuinely to think better, the intervention with the largest measured effect in this entire subject is not a game and not a capsule. Chronic sleep debt of six hours a night for two weeks produces deficits equal to two nights of no sleep, and the people it happens to cannot feel it. Start there, then get the boring blood panel, and treat everything else — apps included — as optional.
Good questions
Do brain-training apps actually work?
They make you better at the app. A review of every study the brain-training companies cite as their best evidence found extensive support for improvement on trained tasks, less for closely related tasks, and little for everyday cognitive performance. No cited study met all the methodological standards needed to conclude otherwise.
What is far transfer?
It is the claim that practising a narrow cognitive task makes you generally better at thinking. It is the entire commercial premise of brain training, and it is the part the evidence does not support. A meta-analysis of 145 experimental comparisons found no convincing far transfer to reasoning, reading comprehension or arithmetic against treated controls.
Will brain training prevent cognitive decline?
No trial has shown that. The largest study, ACTIVE, trained 2,832 older adults in person and found no effect on everyday functioning at two years, though after ten years trained participants reported less difficulty with daily activities. That is a self-reported difference from ten in-person sessions, not a result from a subscription app.
Is it worth paying for a brain-training subscription?
Only if you enjoy the games for their own sake. They are harmless and inexpensive entertainment, and there is nothing wrong with liking them. Paying on the belief that you are protecting your future thinking is not supported by the evidence the companies themselves nominate as their strongest.
Are puzzles and crosswords any different?
Not in kind. Practising any cognitive task reliably improves performance at that task, which is why crosswords make you better at crosswords. The evidence gap is the same one: improvement on the trained activity is easy to demonstrate, and generalisation to everyday thinking is what nobody has convincingly shown.
What actually improves cognitive performance then?
Sleep is the largest measurable lever, by a wide margin. Two weeks at six hours a night produces deficits equal to two nights of no sleep, and people cannot feel it happening. After that, rule out medical causes of poor concentration with a blood panel. Both beat any app or capsule for effect size.
Sources
- Simons DJ, Boot WR, Charness N, et al. Do "Brain-Training" Programs Work? Psychol Sci Public Interest, 2016. View study
- Melby-Lervåg M, Redick TS, Hulme C. Working Memory Training Does Not Improve Performance on Measures of Intelligence or Other Measures of "Far Transfer": Evidence From a Meta-Analytic Review. Perspect Psychol Sci, 2016. View study
- Ball K, Berch DB, Helmers KF, et al. Effects of cognitive training interventions with older adults: a randomized controlled trial. JAMA, 2002. View study
- Rebok GW, Ball K, Guey LT, et al. Ten-year effects of the advanced cognitive training for independent and vital elderly cognitive training trial on cognition and everyday functioning in older adults. J Am Geriatr Soc, 2014. View study
These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.
