Most brain-training hacks don’t work, but a 20-year study found one that does

Since the dawn of using our brains, we’ve wondered how to strengthen them, and I’ve been as guilty as anyone of craving a quick life hack that will galvanise my grey matter.

I dived into Sudoku and shiny apps that promised to keep my mind young. But the research kept finding they mostly made me better at Sudoku and playing specific shiny apps. I even brushed my teeth with my left hand for a while to help neuroplasticity kick in.

But now a respectable journal, Alzheimer’s & Dementia, has linked a computer game to lower dementia risk, and some experts are cautiously excited. The 20-year ACTIVE trial is the largest, and one of the longest-running, tests of cognitive training ever conducted in the United States. It randomly assigned just under 2800 healthy older adults across four groups. One engaged in memory training, one in reasoning training, and of course there was the do-nothing control group (often my favourite group!).

And one group tackled a “speed of processing” game, involving objects flashing on screen while the subject identified their locations. The test became harder and faster as they improved.

Twenty years on, those who did at least eight hours’ speed training over five to six weeks, plus at least one 75-minute booster session, recorded a 25 per cent lower risk of Alzheimer’s or a related dementia. For neuroscience, this is a seriously high effect.

The “speed-trainers plus booster” group were the only cohort to show a statistically significant reduction. Even basic speed training without a booster session didn’t move the neurological needle.

“It’s really a very modest amount of training,” notes Johns Hopkins neurologist Marilyn Albert, a co-author on the research paper.

But before you take a mid-year’s resolution to speed-train your way to life-long brain health, take a breath. The dementia stats were derived from US Medicare claims, not specialist clinical diagnosis. And no one yet has a rock-solid reason why speed worked when memory and reasoning did not.

So yes, this is a genuinely interesting and promising signal. And it comes from a gold-standard trial. But it is certainly not a cure for dementia. I don’t want to undersell it, but it is at best intriguing initial evidence with a large asterisk.

The leading candidate for why speed-based gameplay worked is cognitive reserve, a concept pioneered by Columbia University neuropsychologist Yaakov Stern. Reserve is the brain’s buffer: the gap between the damage quietly accumulating inside your skull and the moment degenerative symptoms actually surface. Two people can carry near-identical Alzheimer’s pathology, yet the one with more reserve keeps functioning for years longer before anything slips. Education, a demanding job and a lifetime of mentally stimulating activity all tend to go with a bigger buffer, or cognitive reserve.

So perhaps while training the brain may not erase the disease, it might pad the buffer that holds it at bay. This is where a lot of the smart research money is now heading.

And let me, someone who could not be more manifestly unqualified to opine on neurobiology and brain health, throw a curve ball in here.

Early work, including a much-hyped Massachusetts Institute of Technology preprint, hints that people who lean on AI to write display less ownership of the result and, for example, struggle to quote it back even minutes later.

Nobody has shown that this dents long-term brain health, and the authors themselves stress that these are preliminary results. But if a generation quietly offloads more of its hard thinking to a chatbot, does the deliberate work of keeping a mind sharp start to matter more, not less?

As the University of Tasmania neuroscientist Lila Landowski told me, “relying on AI excessively is like having a personal trainer lift the weights for you”.

The brain-training industry has a long history of selling certainty that stands on shaky ground. One of the most noted examples dates to 2016, when the US Federal Trade Commission took on Lumosity, one of the world’s most high-profile brain-training empires. I can remember my mum proudly telling me every time she mastered a new level of a Lumosity game – and good on you, Mum, for doing that rather than doom-scrolling TikTok!

Well, Lumos Labs settled for $US2 million [$2.9 million] over claims its games could stave off memory loss, dementia and even Alzheimer’s. Jessica Rich of the FTC ruled: “Lumosity simply did not have the science to back up its ads.”

Remember the life-hacker telling us to brush our teeth with the wrong hand. That one comes from “neurobics”, a 1990s notion that using the non-dominant hand rewires your brain. Researchers who study motor learning and brain asymmetry agree that brushing left-handed for a week makes a right-handed person very good at brushing left-handed. And that’s all. The evidence that this spills over into sharper general thinking is at best disputed. And you could well miss a bit of plaque while you are at it.

Yes, one specific, adaptive, properly trialled speed game seems to show a real long-term signal in this study. That is worth getting excited about. But a “two-time former European memory champion” flogging a $99 course on Instagram (trust me, there are plenty of them) is not the same thing. Nor is playing chess while smashing it out on your Peloton.

Train the right thing and your brain might say thank you. Train the wrong thing, and it’s a “two-time former European memory champion” who probably will!

Adam Spencer was the University of Sydney’s inaugural ambassador for mathematics and science and writes about AI, mathematics and general geekery at adambspencer.substack.com.