Train the right thing, and your brain might thank you.
For decades, the promise that mental exercises could stave off cognitive decline was more marketing than medicine. Now, a 20-year American study involving nearly 2,800 older adults has produced something rarer: a rigorously tested signal suggesting that one specific form of cognitive training — speed-of-processing exercises, paired with booster sessions — may reduce the risk of dementia by 25 percent. The finding does not overturn what we know about the brain-training industry's long history of overreach, but it quietly insists that the question of how we age mentally is still worth asking with care.
- A landmark 20-year trial has found that speed-of-processing training with booster sessions produced a statistically significant 25% reduction in dementia risk — the only intervention among four tested to achieve this threshold.
- The result arrives against a backdrop of industry-wide credibility damage, from Lumosity's $2.9 million FTC settlement to decades of neurobics fads that delivered little beyond task-specific improvement.
- Researchers cannot yet fully explain the mechanism, and dementia diagnoses were drawn from Medicare claims rather than specialist evaluation, leaving a significant asterisk on an otherwise gold-standard study.
- The leading explanation — cognitive reserve, the brain's buffer between silent damage and visible symptoms — suggests training may not erase disease so much as delay the moment it breaks through.
- With AI increasingly absorbing the cognitive labour people once performed themselves, the deliberate effort to keep minds actively engaged may be growing more consequential, not less.
For years, the brain-training industry sold certainty through apps and clever promises — Sudoku, memory games, even brushing teeth with the wrong hand. The research consistently showed people improved at the tasks themselves, and little else. But a genuinely different result has now emerged from one of the longest and most rigorous cognitive-training experiments ever conducted.
Published in Alzheimer's & Dementia, the ACTIVE trial followed nearly 2,800 healthy older adults across 20 years, randomly assigning them to memory training, reasoning training, a control group, or a speed-of-processing game in which objects flashed on screen and difficulty climbed with performance. Two decades later, only one group stood out: those who completed at least eight hours of speed training followed by at least one booster session showed a 25 percent lower risk of developing Alzheimer's or a related dementia. Speed training without the booster produced no measurable effect. Johns Hopkins neurologist Marilyn Albert, a co-author, noted the commitment involved was modest.
Caution remains warranted. Diagnoses came from Medicare claims data rather than specialist clinical assessment, and no one can yet fully explain why speed training succeeded where memory and reasoning training did not. The leading theory centres on cognitive reserve — the concept, developed by Columbia's Yaakov Stern, that the brain maintains a buffer between accumulating damage and the appearance of symptoms. Education, demanding work, and sustained mental engagement all correlate with a larger buffer. Speed training may not erase disease so much as thicken the wall that holds it back.
The distinction between this finding and the broader brain-training marketplace matters enormously. Lumosity was fined $2.9 million by the FTC in 2016 for claiming its games could prevent dementia without the science to support it. One properly tested, adaptive speed game showing a real long-term signal is not the same as a $99 Instagram course from a self-proclaimed memory champion.
A further question hovers at the edges of the research. Early findings suggest that outsourcing thinking to AI may reduce people's sense of ownership over their own ideas and impair recall. No one has demonstrated long-term harm, but as neuroscientist Lila Landowski observed, relying on AI for cognitive work is like hiring a personal trainer to lift the weights for you. If deliberate mental effort matters for brain health, the temptation to delegate it may carry costs not yet fully understood.
For years, the brain-training industry has sold us certainty wrapped in shiny apps and clever promises. Sudoku would sharpen your mind. Memory games would ward off decline. Even brushing your teeth with the wrong hand might rewire your neural pathways. Most of it didn't work. The research kept showing that people got better at Sudoku, better at the apps themselves, and that was where the benefit ended.
But now something genuinely interesting has emerged from a place where such claims are usually tested most rigorously. The journal Alzheimer's & Dementia published results from the ACTIVE trial, a 20-year study that stands as one of the longest and largest cognitive-training experiments ever conducted in the United States. Nearly 2,800 healthy older adults were randomly divided into four groups. Some trained their memory. Others worked on reasoning. A control group did nothing. And one group played a speed-of-processing game—objects flashing on a screen, the player identifying where they appeared, the difficulty climbing as performance improved.
Twenty years later, the numbers told a story worth paying attention to. Those who completed at least eight hours of speed training over five to six weeks, followed by at least one 75-minute booster session, showed a 25 percent lower risk of developing Alzheimer's disease or a related dementia. For neuroscience, this is a substantial effect. The speed-training group with boosters was the only cohort to achieve statistical significance. Speed training alone, without the booster, produced no measurable reduction. Marilyn Albert, a Johns Hopkins neurologist and co-author of the research, called it "a very modest amount of training"—which is to say, the commitment required was not onerous.
Yet caution is warranted. The dementia diagnoses came from Medicare claims data, not from specialist clinical evaluation. No one has a definitive explanation for why speed training succeeded where memory and reasoning training failed. This is promising initial evidence from a gold-standard trial, but it carries a large asterisk.
The leading theory points to cognitive reserve, a concept developed by Columbia University neuropsychologist Yaakov Stern. Reserve is the brain's buffer—the gap between the damage silently accumulating inside your skull and the moment when symptoms actually surface. Two people can carry nearly identical Alzheimer's pathology, yet the one with greater reserve continues functioning normally for years longer. Education, demanding work, and a lifetime of mental stimulation all correlate with a larger buffer. Perhaps speed training doesn't erase disease; perhaps it thickens the wall that holds it back.
This matters because the brain-training industry has a long history of overselling its wares. In 2016, the Federal Trade Commission settled with Lumosity, one of the world's largest brain-training companies, for $2.9 million over claims that its games could prevent memory loss and dementia. The FTC's Jessica Rich was blunt: "Lumosity simply did not have the science to back up its ads." The left-hand tooth-brushing craze of the 1990s—marketed as "neurobics"—promised to rewire the brain through novelty. What it actually did was make right-handed people better at brushing left-handed. The spillover into sharper general thinking remained disputed at best.
One other shadow looms. Early research, including preliminary work from MIT, suggests that people who rely on artificial intelligence to write lose a sense of ownership over the result and struggle to recall it minutes later. No one has shown this damages long-term brain health, and the researchers themselves stress these are preliminary findings. But if a generation quietly outsources its hard thinking to a chatbot, the deliberate work of keeping a mind sharp might become more important, not less. As Lila Landowski, a neuroscientist at the University of Tasmania, put it: relying excessively on AI is like having a personal trainer lift the weights for you.
So yes, one specific, adaptive, properly tested speed game shows a real long-term signal in this study. That deserves genuine excitement. But the person selling a $99 brain-training course on Instagram—even if they claim to be a two-time former European memory champion—is not the same thing. Train the right thing, and your brain might thank you. Train the wrong thing, and you're just making someone else rich.
Citações Notáveis
A very modest amount of training— Marilyn Albert, Johns Hopkins neurologist and ACTIVE trial co-author
Relying on AI excessively is like having a personal trainer lift the weights for you— Lila Landowski, University of Tasmania neuroscientist
A Conversa do Hearth Outra perspectiva sobre a história
Why did speed training work when memory and reasoning training didn't? That seems backwards.
Nobody knows for certain yet. But the leading theory is about cognitive reserve—the brain's buffer against disease. Speed training might strengthen that buffer in a way the other trainings don't. It's not that it erases Alzheimer's pathology; it's that it gives you more cushion before symptoms surface.
So it's not a cure.
Not remotely. It's a 25 percent risk reduction in a study that tracked Medicare claims, not clinical diagnoses. That's genuinely interesting, but it's not a cure. It's a signal worth investigating further.
The article mentions AI and offloading thinking. Why does that matter here?
Because if people start relying on AI to do their cognitive heavy lifting, the deliberate work of training your brain might become even more important. Right now we don't know if that's true long-term. But the worry is real.
What about all the other brain-training apps and games?
Most of them make you better at the specific game, and that's it. The brain-training industry has a long history of selling certainty it doesn't actually have. Lumosity got fined $2.9 million for making claims it couldn't back up. This speed-training study is different because it's a 20-year trial with a control group, not marketing.
So should people actually do this speed training?
If you're interested in it and have the time, eight hours over five or six weeks plus a booster session is a modest commitment. But don't expect it to be a guarantee. And don't confuse it with the Instagram courses and the memory champions selling certainty. This is one piece of evidence, not a life hack.