Does brain training work? The honest state of the evidence

The short answer

Yes and no, and the split matters more than the headline. Practising a cognitive task reliably makes you better at that task and at tasks that closely resemble it. That part is not in dispute. What is disputed — and what the marketing usually implies — is whether the improvement spreads to general intelligence, to memory in everyday life, or to protection against decline in later years. The largest and best-controlled studies find little evidence that it does.

So a brain-training app can honestly promise a measurable, repeatable improvement on the specific skills it trains. It cannot honestly promise that you will become smarter, remember where you left your keys, or hold off dementia. Anyone who tells you otherwise is ahead of the evidence.

Near transfer: the part that works

Researchers describe the spread of a training effect as transfer. Near transfer is improvement on tasks that share most of their structure with the trained one: practise a working-memory span task and you get better at other span tasks. Far transfer is improvement on something different in kind — reasoning, reading comprehension, driving, remembering appointments.

Near transfer is found consistently. It is also, in a sense, unremarkable: it is the same thing that happens when you practise an instrument, a language or a sport. You get better at what you practise, and the gains fade if you stop. Some of the improvement is genuine skill, some is strategy (you learn how to approach the task), and some is plain familiarity with the format.

None of that makes near transfer worthless. If a task genuinely exercises holding several items in mind while updating them, then getting better at it is a real change in a narrow ability. It is just not the same as a change in you.

Far transfer: the contested part

The modern brain-training industry owes a great deal to a 2008 study in which people who practised a working-memory task called dual n-back improved on a test of fluid intelligence, and the gain grew with the number of training sessions. Fluid intelligence is the capacity to reason through novel problems, long thought to be more or less fixed in adulthood, so the result was startling and was reported everywhere.

The follow-ups were mixed. Some replications found a similar effect; others, often with tighter controls, found none. The recurring criticism was methodological: small samples, control groups who did nothing rather than an equally engaging alternative, and outcome tests that resembled the training more than was acknowledged. When later meta-analyses pooled the studies, the far-transfer effect shrank as study quality rose, and in the best-controlled comparisons it was close to zero.

That pattern — a striking early result, a wave of enthusiasm, then a slow retreat as controls improve — is common in psychology, and it is not a scandal. It is how a field corrects itself. But it does mean that the strongest claim you can defend today is that far transfer from computerised brain training is small, inconsistent and possibly absent. Not that it has been demonstrated.

The 2014 statements, and what followed

In 2014 the argument became public. A group of prominent cognitive scientists and neuroscientists published a consensus statement saying the scientific literature did not support claims that brain games improve general cognition or slow age-related decline, and that marketing was running well ahead of the evidence. A second group of researchers replied with a statement of their own, arguing that the first had understated a real and growing body of positive findings.

Both sides agreed on more than the coverage suggested. Neither disputed near transfer. Neither claimed that any app could prevent dementia. The disagreement was about how much weight to give the positive far-transfer studies given their weaknesses, and about how companies should be allowed to describe their products.

Two years later a US regulator settled with one of the best-known brain-training companies over advertising claims it could not substantiate, and a large independent review of the published studies concluded that most were too weakly designed to support the claims being made from them. Since then the more careful companies have pulled their language back, and the rest have not.

What “brain age” and percentiles actually measure

Many apps convert your scores into a “brain age” or a percentile rank. It is worth understanding what those numbers can and cannot be.

A percentile requires a population. To tell you that you are in the top 12% for your age, the app needs scores from a large, representative sample of people your age, taken under comparable conditions, on the same tasks. Most apps have something much weaker: whichever of their own users happened to play, on whatever devices, however distracted, whether or not they had played before. The arithmetic is real; the sample does not mean what it is implied to mean.

A brain age is worse, because it borrows the authority of a medical measurement it is not. It is typically a single composite score mapped onto an age curve fitted to that same unrepresentative sample. It moves with your sleep, your mood, your familiarity with the game and the size of your screen. It sounds like a diagnosis and it is not one. If you are worried about your memory or thinking, the person to ask is a doctor, not an app.

Some apps decline to show these numbers at all and compare you only with your own past results; Braintide is one of them, and says so on the screen where the percentile would otherwise be.

What brain training is reasonably good for

Once the inflated claims are set aside, a well-made app is genuinely useful for a few things.

• A repeatable benchmark of specific skills. If the same tasks are run at the same difficulty each time, you get a number you can compare over weeks and months. That is more than most people have for their own attention or working memory.
• Engagement. Short, varied, slightly difficult tasks are a pleasant way to spend ten minutes, and there are worse things to do on a phone.
• A habit with a shape. Doing something deliberate and effortful every day has value in itself, and a daily workout gives that habit a structure.
• Noticing your own variation. Playing the same tasks after a good night’s sleep and a bad one makes the effect of sleep vivid in a way no article can.

What it is not good for is replacing anything on the list below.

What protects cognition better

The interventions with the strongest evidence for long-term brain health are not apps.

• Sleep. Memory consolidation happens during sleep, and even one short night measurably impairs attention and working memory the next day. Nothing in an app compensates for it.
• Physical exercise. Regular aerobic activity is consistently associated with better cognitive function and lower risk of decline, and it is the intervention public-health bodies actually recommend.
• Social contact. Conversation is demanding cognitive work: tracking meaning, predicting, remembering, switching. Isolation is associated with faster decline.
• Learning something genuinely new. A language, an instrument, a craft. These demand broad skill by their nature, and they come with the social and motivational scaffolding an app lacks.
• Managing the big risks. Blood pressure, hearing loss, smoking, heavy drinking and untreated low mood matter more for the brain than any game.

How to use brain training without fooling yourself

Treat it as deliberate practice of a few narrow skills, keep sessions short and regular, and judge progress by a fixed-difficulty benchmark rather than an adaptive score that climbs with the difficulty. Building a routine that stays honest covers the how, and choosing a brain-training app covers the questions that separate the candid ones from the rest. If you want to see the argument at its sharpest, dual n-back is where it started.

Frequently asked questions

Do brain games improve IQ?

Not reliably. A well-known 2008 study reported gains in fluid intelligence after working-memory training, but later replications were mixed and the largest meta-analyses find that the effect shrinks towards zero as study quality improves. You will get better at the games; a change in general intelligence has not been convincingly shown.

Can brain training prevent dementia?

There is no good evidence that it can. Physical exercise, sleep, social contact, hearing care and managing blood pressure have far stronger support. If you are worried about your memory, talk to a doctor rather than relying on an app.

Is brain training a waste of time?

Not if you want what it can actually deliver: a measurable, repeatable practice of specific skills, a short daily habit, and a clear view of how sleep and stress affect your own attention. It is a waste only if you expect it to make you generally smarter.

Why do apps show a brain age or percentile if they are meaningless?

Because they are motivating and they look scientific. A percentile needs a representative population the app almost never has, and a brain age is a composite score dressed as a medical measurement. Neither tells you much beyond how you compare with other users of that app, on that day.

Not a medical tool: Braintide is a game, not a cognitive assessment or a treatment. These guides summarise published research for general interest and are not medical advice. If you are worried about your memory or thinking, talk to a doctor.