The brain training industry is about $6 billion a year. Most of it is not buying you what it says it's buying you. In 2016 the Federal Trade Commission charged Lumosity's parent company $50 million for deceptive advertising — specifically, for claiming their games would reduce cognitive decline when the evidence didn't support that.
This is one of those areas where the space between the marketing and the actual research is enormous, and where the confusion has had real consequences for how people spend their time and money.
But inside the noise, there is one specific type of training that the data does support. And the story of why it works when everything else mostly doesn't turns out to be more interesting than the products.
The core problem: transfer
The central question in cognitive training isn't *can you get better at this task?* Of course you can. Anyone gets better at anything they practice. The real question is *does getting better at this task make you better at anything else?*
That's called transfer. And transfer is mostly what the brain training industry has been selling.
The promise is usually something like: *play these games for 15 minutes a day and you'll improve your general memory, attention, and cognition.* The implicit claim is that training specific cognitive tasks will transfer to broader cognitive function, daily life performance, and long-term brain health.
The evidence for that kind of transfer is weak to nonexistent for most brain training products.
A 2016 systematic review of the research (the one that helped drive the FTC action) found that in most cases, training on a specific task improved performance on that specific task — and no other. You got better at the game. The improvement didn't travel to anything that mattered in your actual life.
A 2017 systematic review by Sprague and colleagues was slightly more nuanced. Process- and strategy-based cognitive training improved trained domains but had weak transfer to untrained ones. Aerobic exercise — which isn't brain training at all — more consistently improved executive function. Strength training and aerobic/resistance combinations most consistently improved cognitive inhibition and working memory.
In other words: the best "brain training" in the literature isn't brain training. It's physical exercise.
The exception that actually works: speed training
The most interesting finding in the entire cognitive training literature comes from the ACTIVE study — Advanced Cognitive Training in Vital Elderly. It's a large, well-designed, long-running trial. Started in 2002. Followed 2,800 older adults for a decade.
Participants were randomized to one of four conditions:
- Control (no training)
- Memory training (strategy practice for verbal episodic memory)
- Reasoning training (problem-solving and pattern recognition)
- Speed training (computerized exercises that required rapid visual information processing)
Ten years later, the results were striking.
The memory and reasoning groups showed essentially no reduction in dementia risk compared to control. The ten-year dementia incidence was 9.7% for memory training and 10.1% for reasoning training — no different from the 10.8% control rate.
The speed training group had a ten-year dementia incidence of 5.9%. Roughly a 45% reduction.
Six weeks of speed training, years earlier, appeared to produce a measurable dementia-risk reduction a decade later. This is not a typical finding in the brain training world. Most things that claim this kind of effect can't substantiate it. This one held up.
Why speed training might be different
The ACTIVE speed training involved computerized visual-perceptual exercises that demanded rapid information processing — identifying targets as they flashed briefly, tracking multiple objects simultaneously, making quick decisions under time pressure.
A few reasons this may have had broader effects than other training types:
**Processing speed is foundational.** Nearly every cognitive task depends on how fast information can be encoded and manipulated. A system-level improvement in processing speed has the potential to cascade into other functions.
**The tasks demanded novelty and complexity.** They kept changing, required adaptation, and didn't let the brain settle into a groove. This is different from games that let you master a pattern and then repeat it.
**They engaged attention and inhibition together.** You had to focus on the target *and* ignore distractors. That combination engages more brain networks than either alone.
**Time pressure added neurological stress.** Hormesis again — brief controlled stress, adaptive response.
There's a follow-up line of research pointing at something similar. A 2014 study had older adults play Super Mario 64 — the 3D platformer — for a half hour a day over six months. The control groups either took piano lessons or did nothing.
The Mario group showed increased gray matter in the right frontal eye fields, the hippocampus, and the cerebellum, and improved performance on antisaccade tasks (a measure of executive control). The piano group showed changes too, but more diffuse. The no-intervention group showed the expected age-related gray matter loss.
The probable reason Mario worked: it required rapid spatial attention, working memory, executive control, and motor coordination, all interleaved, all under time pressure. Same underlying principles as the ACTIVE speed training.
What this tells us about training in general
The specific findings on speed training point at a deeper principle: *for cognitive training to transfer, it needs to demand broad, integrated neural engagement — not isolated skill repetition.*
This is probably why most brain training apps don't work. They train one narrow skill at a time — memorize this sequence, click this target, identify this pattern — and the training stays narrow. The brain is a use-dependent system, but what gets used only as the thing it's being practiced for tends to *only* improve in that specific context.
Integrated, complex, novel engagement works differently. Learning a new language, playing a musical instrument, picking up a sport that demands quick decisions under unpredictable conditions — these engage cognition the way life does. Transfer is more likely.
The real finding: cognitive reserve
The broader framework most of this research is pointing at is the concept of cognitive reserve. The idea: a lifetime of complex, varied cognitive engagement builds biological reserve that protects against decline. Education, complex work, rich social lives, intellectual hobbies — these all compound over time.
People with higher cognitive reserve can accumulate significant brain pathology without showing the cognitive symptoms you'd expect from the pathology alone. Roughly half of people over 90 with clinical dementia show less neuropathology than you'd predict. Half of people without clinical dementia show more. Reserve is doing enormous work in the variance.
Brain training apps are trying to buy their way into a concept that you build over a lifetime of real engagement with the world.
What to actually do
**Don't spend money on most brain training apps.** They're optimizing for your subscription, not your cognition. If you want to train specifically, the ACTIVE-style speed training programs (some are free or near-free now) are the exception.
**Exercise, especially resistance and aerobic together.** This is the most reliable "brain training" in existence. It builds BDNF, improves cerebral blood flow, maintains hippocampal volume, reduces all-cause cognitive decline.
**Pick up something new and complex every few years.** A new language. An instrument. A sport that demands coordination and strategy. Gardening in a new climate. A craft with technical depth. The specific activity matters less than the novelty and complexity.
**Keep working on things that make you think.** Continuing education — formal or informal — is one of the strongest and most consistent predictors of cognitive preservation into late life. Read real books. Have hard conversations. Engage with people who think differently than you do.
**Play games that demand integrated thinking.** Not memory puzzles. Strategy games, complex video games with time pressure, chess, bridge. If it's easy after a few weeks, it stopped training anything meaningful.
The thing being sold as brain training mostly isn't brain training. The real brain training is a life with ongoing demand in it. Most of the people whose cognition holds up into their 90s aren't playing brain games. They're still working, still learning, still making something, still connecting with people who make them think.
That's a harder product to package and sell. It's also the one that actually works.