By Haru · Updated · 2026-09-07
Does Brain Training Work — What the Evidence Says
Practice a memory game daily and your score on that game will go up. That much is not in dispute and never has been. The question worth asking is narrower and much harder: does anything you gained show up somewhere else, in work, in study, in remembering where you left your keys.
Researchers call that far transfer, and it is where the evidence gets thin. The largest trials have consistently found the first thing and not the second, a regulator in the United States fined one of the biggest companies in the field over how it described its product, and scientists split publicly into two camps in 2014 and have not fully reconciled.
This page sets out what those studies actually found and what they did not, so you can decide whether the twenty minutes a day is worth it. It also says plainly what the tools on this site are, because they belong on the measurement side of this argument rather than the training side.
Near transfer and far transfer are the whole argument
Near transfer means improving at things that closely resemble what you practiced. Drill two-digit addition and you get faster at two-digit addition, and probably at three-digit addition too. Nobody disputes near transfer, and it is not very interesting, because it is what practice means.
Far transfer means improving at something that does not resemble the training. Drill a memory game and become better at following a meeting, or at studying, or at reasoning through an unfamiliar problem. That is what the advertising promises and it is the claim the research keeps failing to support.
Once you have this distinction, most of the confusion clears. A brain training app showing you a rising graph is showing near transfer. It is real, it is measurable, and on its own it tells you nothing about whether anything else in your life changed.
The biggest trial found the first and not the second
The most-cited attempt to settle this was published in Nature in 2010 by Adrian Owen and colleagues, under the title Putting brain training to the test. It ran with more than eleven thousand participants who trained several times a week for six weeks on tasks modelled on commercial brain training games.
The result was the split described above. Participants improved on the tasks they trained, in some cases substantially. On the broader tests of cognition given before and after, the trained groups did not improve more than the control group. The authors described no evidence of transfer to untrained tasks, even when those tasks were closely related.
A study is one study, and this one has been argued over since. But the pattern it found has held up in later reviews more often than it has been overturned, which is a different situation from a single surprising result waiting to be replicated.
The scientists split in public, and the split is informative
In 2014 a group of researchers organized by the Stanford Center on Longevity and the Max Planck Institute for Human Development published a statement warning that the field's claims outran its evidence. A second group of scientists responded with a counter-statement arguing that a substantial body of evidence did support cognitive training.
Both statements were signed by serious people, which is why quoting one side settles nothing. What moved the argument forward was a large review published in 2016 by Daniel Simons and colleagues, which went through the literature the industry itself cited. Its central finding was less about the results than about the methods: much of the supporting evidence came from studies whose design could not distinguish a genuine effect from expectation and practice.
That is the honest state of things. Not a proven fraud, and not a proven benefit either, but a field where the strongest studies find the least.
A regulator stepped in, and what it objected to matters
In 2016 the United States Federal Trade Commission settled with Lumos Labs, the company behind Lumosity, for two million dollars over its advertising. The company had told customers its games would improve performance at work and at school and reduce or delay cognitive decline associated with age, including dementia.
The finding was not that the games do nothing. It was that the company did not have the science to support what it had promised, which is a narrower and more useful conclusion. Making a claim about dementia requires evidence about dementia, and that evidence did not exist.
This is a good filter to carry into any product page. The question is never whether a company sounds scientific. It is whether the specific promise on the box has a study behind it that measured that specific thing.
What the tools on this site are, and are not
They are measurements. The mental math speed test, the number memory test, the memory test, the focus test and the reaction time test each report how you did at a specific task at a specific moment. That is the whole claim, and it is a claim we can actually support, because the thing being measured and the thing being reported are the same thing.
What they are not is a training program, and we do not describe them as one. If you run the mental math drill every day your rate will rise. That is near transfer, it is genuinely satisfying, and we make no assertion about anything beyond it. The same goes for the brain type test, which is a light self-check rather than an assessment of anything.
Being clear about this costs us nothing and is the only defensible position given the evidence above.
Should you do it
If you enjoy it, yes, and the reason is simply that you enjoy it. That is a sufficient reason to spend twenty minutes on something and it does not require a study.
If you are doing it because you want to get better at a specific thing, practice that specific thing instead. Wanting to be faster at arithmetic is the one case where these drills are exactly right, because there the training and the goal are the same activity. Wanting to concentrate better during long documents is not, and reading long documents is closer to what you actually want.
And if you are doing it out of worry about your memory or somebody else's, that concern belongs with a doctor rather than an app. That is the part where the marketing went furthest past the evidence, and it is the part where the cost of being wrong is highest.
Sources Stanford Center on Longevity and the Max Planck Institute for Human Development — "A Consensus on the Brain Training Industry from the Scientific Community" (2014) Owen et al. (2010), “Putting brain training to the test”, Nature 465, 775–778 — free full text at PubMed Central Simons et al. (2016), “Do ‘Brain-Training’ Programs Work?”, Psychological Science in the Public Interest 17, 103–186 — Europe PMC record US Federal Trade Commission (January 2016) — “Lumosity to Pay $2 Million to Settle FTC Deceptive Advertising Charges for Its ‘Brain Training’ Program”
Frequently asked questions
Does this mean brain training games are a scam?
No, and the distinction matters. The games do what they demonstrably do, which is make you better at the games. The problem has been with claims that went beyond that, which is exactly what the regulator objected to in the Lumosity case. A game that says it is a game is not misleading anyone.
My score keeps going up. Why does that not count as evidence?
Because it is the part nobody disputes. Improving at a task you repeat is what practice does, and it happens even when nothing general has changed. The claim under argument is whether that improvement appears somewhere you did not train, and your rising score in the app cannot answer that question either way.
Are the paid apps better than the free ones?
Price is not the variable. What separates a defensible product from a questionable one is the specificity of its claims and whether a study measured that specific claim. A paid app promising to reduce cognitive decline needs evidence about cognitive decline; a free game promising to be a fun way to practice mental arithmetic needs no evidence at all, because that promise is self-evident.
Would this help an older parent?
That is the exact question the marketing answered too confidently, and it is not one a web page should answer for you. Claims about memory loss and dementia require evidence about memory loss and dementia, and that is what the regulator found missing. If the worry is real, it belongs with a doctor who can look at the whole picture rather than with a subscription.
Why does this site publish these tests at all, then?
Because measuring something and training something are different activities, and measuring is the one we can stand behind. A reaction time reading tells you what your reaction time was on that run. It is useful for comparing your own runs, for checking whether new hardware changed anything, and for curiosity. None of that requires a claim about transfer, so we do not make one.