Does brain training actually work? 🔬

The short answer

You get better at the game. That improvement mostly does not go anywhere else.

That sentence is not cynicism, it is roughly where twenty years of research landed. It is also worth writing plainly on a site that is, itself, a collection of brain games — so let us go through what the evidence actually supports.

Near transfer and far transfer

Two terms make the whole debate legible.

• Near transfer — improvement carries to a very similar task. Practise 2-back, get better at 3-back. This is well established and not controversial.
• Far transfer — improvement carries to something genuinely different: your work, your studies, general reasoning. This is what brain training products sell, and it is where the evidence is weak.

Almost every impressive-sounding claim you have read is a near-transfer result described as though it were far transfer.

What happened with N-back

The story is worth knowing because it is the field in miniature.

In 2008, Jaeggi and colleagues reported that training on dual N-back improved fluid intelligence — the ability to reason about novel problems. If true, that would be remarkable, since fluid intelligence had been considered largely fixed in adulthood. The result was widely covered and launched an industry.

Subsequent work has largely failed to reproduce it. Larger studies with better control conditions found the gains shrinking or vanishing, and meta-analyses concluded that working memory training produces reliable improvement on trained and closely related tasks and little to nothing on fluid intelligence.

The methodological problem identified repeatedly was the control group. If your trained group does a demanding game and your controls do nothing, any difference could be motivation, expectation, or simply having practised sitting a test. Studies with active control groups — doing something equally engaging but unrelated — generally show much smaller effects.

The 2010 study that settled the public argument

A large trial run with the BBC recruited over 11,000 participants, had them train on brain games for six weeks, and tested them on unrelated cognitive tasks before and after. Participants improved substantially on the games they trained. On the untrained tests, the trained groups did no better than controls.

Later, in 2016, the US Federal Trade Commission fined Lumosity $2 million over advertising claims about improving performance at work and school and reducing cognitive decline. The FTC's position was straightforward: the company had not substantiated them.

So what does help?

This is the useful part, because several things have far better evidence than games do — they are just less marketable.

• Sleep. The single largest and most reliable lever on day-to-day cognitive performance. A short night measurably slows reaction time, degrades inhibition and reduces working memory. You can see this yourself: run the reaction time test after a good night and a bad one.
• Aerobic exercise. The most consistently supported intervention for cognitive function across the lifespan, with the strongest results in older adults. Effects are modest but real and repeatedly replicated.
• Learning something genuinely difficult. A language, an instrument, a craft. This is not "brain training" but it is the closest thing to it that produces durable change — largely because you are actually acquiring a skill rather than practising a test.
• Managing stress and mood. Anxiety and depression impair attention and working memory substantially. Treating them improves cognition in a way no game does.
• Not multitasking. Task switching carries a real, measurable cost — you can feel it on our task switching test. Doing one thing at a time is a larger practical win than any training programme.

Then why play these games at all?

Three honest reasons.

• Familiarity is worth real points in a testing situation. If you will face a cognitive assessment — a hiring process, an exam — the first few items of an unfamiliar task are spent working out the rules. Having seen the format removes that loss. This is near transfer, which is the kind that works.
• They are a decent condition check. Reaction time and inhibition are so sensitive to sleep and fatigue that measuring yourself repeatedly tells you something true about your own state.
• They are enjoyable. Which is a sufficient reason, and does not require any claims about intelligence.

How to spot an overstated claim

• "Scientifically proven" with no study named. Ask which study, on how many people, with what control group.
• "Improves memory" without saying which memory. Working memory, episodic memory and procedural memory are different systems.
• Testimonials instead of trials. People who improved will always exist; the question is whether they improved more than a control group.
• Claims about preventing dementia. This is the claim regulators act on, and for good reason — the evidence does not support it.
• "Trains your brain like a muscle." A metaphor, not a mechanism.

The bottom line

Brain games are a reasonable way to spend fifteen minutes and a good way to get familiar with cognitive task formats. They are not a route to becoming smarter, and anyone telling you otherwise is selling something.

If your goal is to think better tomorrow than today, sleep is a better investment than any of this — including our site.