Brain Training Apps: What the Evidence Actually Shows

Brain training apps reliably make you better at brain training apps. Twenty years of research has produced a consistent pattern: practise a speed or memory game and you improve at that game, and often at tasks that closely resemble it, but the gain rarely shows up as better memory at work, sharper thinking in general, or protection against cognitive decline. That is a narrower claim than the marketing makes, and a more useful one than the flat dismissal you sometimes hear.

This piece sets out what the research actually found, what the main apps train, what has better evidence behind it, and how to judge a brain training subscription if you still want one. The short version: the exercises do something, the something is smaller and more specific than advertised, and several unglamorous habits have stronger support than any app.

What the evidence says

The 2014 consensus statement

In October 2014 the Stanford Center on Longevity and the Max Planck Institute for Human Development published a consensus statement on the brain training industry, signed by around 70 cognitive scientists and neuroscientists. Their conclusion was blunt: the scientific literature did not support the claim that brain games reduce or reverse cognitive decline, and the promise of broad cognitive benefit was being sold well ahead of the evidence. They also raised the opportunity cost — hours spent on a game are hours not spent on exercise, learning, or seeing other people.

The statement did not go unchallenged. A rebuttal letter, organised through a site set up by figures connected to the training industry, was signed by a comparable number of researchers who argued that the trial evidence was stronger than the consensus allowed. That disagreement is real and has not been resolved. It is worth knowing that some of the loudest voices on the optimistic side have commercial interests, and that this does not automatically make them wrong.

The Lumosity settlement

In January 2016 Lumos Labs, the maker of Lumosity, settled with the US Federal Trade Commission over deceptive advertising. The FTC's complaint was not that the games failed to exist or failed to be games — it was that the company had advertised real-world benefits it could not support: better performance at school and work, and reduced cognitive impairment associated with ageing and serious conditions. The judgment was for $50 million, suspended on payment of $2 million given the company's finances.

The settlement is the clearest official statement anyone has made about this category. It concerned advertising rather than the underlying science, but it established that "play our games and your life gets better" is a claim requiring evidence that, at the time, did not exist.

The ACTIVE trial

The strongest evidence in the other direction comes from ACTIVE (Advanced Cognitive Training for Independent and Vital Elderly), a large randomised trial that enrolled roughly 2,800 older adults and assigned them to memory strategy training, reasoning training, speed-of-processing training, or no training. Each group improved at the ability it trained, and for reasoning and speed those gains were still detectable years later. This is a genuine, durable effect, and it is more than critics sometimes acknowledge.

The complications start at the edges. Improvements in self-reported everyday functioning were modest and came mainly from participants' own reports rather than objective measures, and transfer to daily life remains debated. A later secondary analysis reported a lower rate of dementia diagnoses in the speed-training group over ten years. That analysis was exploratory rather than a prespecified outcome, the trial was not designed to detect dementia, and it has not been replicated — so it cannot support prevention claims, even though it is frequently quoted as though it does.

The transfer problem

Underneath all of this sits one question: does training a narrow skill generalise? Researchers call improvement on similar tasks "near transfer" and improvement on unrelated ones "far transfer". Near transfer is easy to produce and shows up in study after study. Far transfer is the thing people are actually buying, and it is the thing that keeps failing to appear.

A large 2016 review in Psychological Science in the Public Interest examined the literature behind commercial brain training and found extensive evidence that people improve at trained tasks, less for closely related tasks, and little that the training improves everyday cognitive performance. Meta-analyses of working-memory training have reached similar conclusions. Some trials also lack an active control group, which matters: if the comparison group does nothing, improvements may reflect motivation, expectation, and familiarity with being tested rather than the training itself.

The apps people mean

"Brain training app" covers several different products with different rationales.

Lumosity is the category's most recognisable name: a large library of short games built around memory, attention, processing speed, flexibility, and problem solving, wrapped in daily sessions and progress scores. It runs on a free tier with a paid upgrade. Post-settlement, its marketing is noticeably more careful than it once was.

Peak follows a similar model of short daily game sessions across memory, attention, language, and mental agility, with some exercises developed in collaboration with academic researchers. Free tier with paid upgrade. Its strength is presentation and variety; the underlying transfer question is the same one every game-based trainer faces.

Elevate is the outlier in the group, and the more defensible design. Rather than abstract puzzles, it drills skills with obvious real-world referents: reading comprehension, writing precision, listening, mental arithmetic, vocabulary. Practising comprehension to get better at comprehension is a far shorter inferential leap than practising a spatial game to get better at thinking. Free tier with paid upgrade.

BrainHQ, from Posit Science, is the app with the closest link to the research literature. Several of its exercises descend from laboratory paradigms, including the useful-field-of-view speed task used in ACTIVE. That lineage is real and worth something. It also means the honest claim is specific — trained visual processing speed — rather than general mental sharpness. Subscription based.

CogniFit leads with assessment: a battery that profiles cognitive domains and then assigns training, positioned towards clinical and research use as well as consumers. Subscription based. Detailed scoring can create an impression of clinical precision that consumer-grade measurement does not really support.

A fair summary of the category: these are competently made products that give you structured, measurable practice at specific tasks. What none of them can honestly promise is that the practice becomes general cognitive ability.

What works better for a healthy brain

The interventions with better evidence are less interesting to sell, which is part of why they get less attention.

Physical exercise. Aerobic activity has the most consistent supporting evidence of any behavioural intervention for cognitive health, through effects on cardiovascular and metabolic health, sleep, and mood. The effect sizes in trials are moderate and vary considerably between studies, so this is not a guarantee either, but it is the best-supported lever available.

Sleep. Memory consolidation happens during sleep, and chronic short sleep degrades attention, working memory, and emotional regulation in ways that are easy to demonstrate experimentally. Nobody has to argue for far transfer here: sleep loss impairs everything, and repairing it improves everything.

Social contact. Regular social engagement is associated with better cognitive outcomes in ageing across many observational studies. Causation is hard to establish — withdrawing from social life can be an early symptom rather than a cause — but conversation is genuinely demanding cognitive work, unpredictable in a way no app can reproduce.

Learning genuinely new and difficult material. In one well-known trial, older adults who spent months learning demanding new skills such as digital photography or quilting showed memory improvements that groups doing familiar or purely social activities did not. The FINGER trial in Finland found modest cognitive benefits from a multidomain programme combining diet, exercise, cognitive training, and vascular risk management — worth noting that cognitive training was one ingredient in a package, not the isolated cause.

Why does novelty beat reaction-time drills? A brain training game becomes easier as you learn its structure, and the point at which it stops being hard is the point at which it stops doing much. Learning something genuinely unfamiliar keeps producing difficulty, because each new piece of understanding opens ground you did not previously know existed. Difficulty that stays difficult is the active ingredient, and games are engineered to feel rewarding rather than to stay hard.

Learning something real instead

This is the case for treating curiosity as the training programme. Reading or listening to material you do not already understand demands sustained attention over minutes rather than seconds, builds schemas that new information can attach to later, and leaves you with knowledge you can actually use — in conversation, in work, in understanding the news. Those are, almost word for word, the benefits brain training advertises. The difference is that you get them as a by-product of learning something, rather than hoping they spill over from a game.

The practical objection is time. A daily study session is a bigger commitment than a three-minute puzzle, which is precisely why the puzzle wins on most weekday evenings. That gap is where short-form learning sits. Chunks, the app we make, is one example: a short narrated factual story each day on history, science, or philosophy, one free daily, on iOS and Android. It is not a treatment and it makes no cognitive-health claims. It is simply new material in a format small enough to survive contact with a normal schedule — which is the quality that determines whether anything gets done at all.

How to choose, if you still want one

Wanting one is reasonable. Many people enjoy these apps, and enjoying a daily cognitive habit is a perfectly good reason to keep it.

Look for specific claims tied to specific exercises, published trials with active control groups rather than do-nothing controls, and independent replication. Check whether the studies cited on a marketing page were funded or conducted by the company selling the product, and whether they measured a real-world outcome or another version of the trained task.

Distrust any claim about preventing, delaying, treating, or reversing dementia or Alzheimer's disease; testimonials standing in for trial data; "scientifically proven" without a named study; a single small trial presented as settled science; and brain-age scores, which are a gamified metric rather than a clinical measurement.

Ask before subscribing: what exactly does this train, and what is the evidence it transfers anywhere else? Would I do this if it were framed as a game rather than as self-improvement? What would I otherwise be doing with these twenty minutes? Is there a free trial long enough to find out whether I still want it in three weeks? Most subscriptions in this category die of boredom rather than disappointment, so the honest test is whether you would keep playing with no self-improvement story attached.

Frequently Asked Questions

Do brain training apps actually work?

They work for the narrow thing they train and rarely beyond it. Practice reliably improves performance on the trained exercises and on closely similar tasks, but the evidence that this generalises to everyday memory, attention, or intelligence is weak, and the 2014 consensus statement from around 70 scientists said as much. Treat improvement in your scores as evidence you have learned the games, not evidence your cognition has broadly improved.

What is the best brain training app?

There is no app with enough evidence to claim the title outright. BrainHQ has the closest connection to published trials, including exercises descended from the ACTIVE speed-of-processing training, while Elevate trains skills such as reading comprehension and arithmetic where the link to real-world use is shorter. If your goal is enjoyment and a daily habit, pick whichever one you will actually open.

Are brain games good for dementia prevention?

The evidence does not support dementia prevention claims, and the FTC action against Lumosity in 2016 concerned advertising of exactly that kind. One secondary analysis from the ACTIVE trial reported fewer dementia diagnoses after speed-of-processing training, but it was exploratory, not what the trial was designed to measure, and has not been replicated. Anyone worried about cognitive decline should speak to a doctor about cardiovascular risk, hearing loss, sleep, and physical activity rather than buying a subscription.

What is better than brain training?

Physical exercise, adequate sleep, regular social contact, and learning genuinely new and difficult material all have better supporting evidence than commercial brain games. None of them are quick, and none can be bought in a subscription, which is roughly why the games market exists. The most reliable cognitive investment available to most people is sleeping properly and learning something hard.

The bottom line

Brain training apps are practice, and practice is specific. Play them because you enjoy them, track your scores if you find that motivating, and hold on to the fact that the improvement you see is improvement at the game. If what you actually want is the thing the marketing describes — sustained attention, sharper thinking, knowledge that shows up when you need it — the route there is learning material you do not yet understand, consistently, over months.

If you want to go deeper on how learning actually works, our guides to cognitive load theory and the science of spaced repetition cover why difficulty and timing matter more than the format of the exercise. And if you want a daily habit built on real material rather than puzzles, Chunks gives you one narrated story a day, free, on iOS and Android.