Do Brain Training Games Work? What the Evidence Says

Lumosity charged me $79 for a year's subscription. I played it for six months — fifteen minutes every morning, right after coffee, right before sitting down to write. My scores climbed steadily. I got faster at identifying patterns. I genuinely felt sharper.

Then I sat down to write and found exactly the same wall I'd been hitting before.

I'm not alone. Lumosity alone says it has attracted more than 100 million members, and market-research firms put the broader brain training market in the billions of dollars. The pitch is simple and deeply appealing: spend a few minutes a day on cognitive puzzles, and watch those benefits bleed into everything else — your work, your memory, your focus in meetings. It's intellectual exercise. If you train your body to carry groceries, why can't you train your brain to think better?

The problem is that analogy breaks down exactly where it matters most. Going to the gym makes you better at lifting things in the real world. Brain training apps, in most large independent studies, make you better at brain training apps. Understanding why — and what the research actually does support — is the more useful conversation.

The concept at the center of this debate is called far transfer. Close transfer means getting better at the specific task you practice. You play Tetris for a month, you get dramatically better at Tetris. Far transfer means that improvement carries over into something unrelated — that spatial reasoning in Tetris makes you better at reading maps, or that pattern-matching speed on a screen makes you more perceptive in meetings.

Brain training apps stake their entire value proposition on far transfer. That's where the evidence has a real problem.

The Study That Changed the Conversation

In 2010, Adrian Owen and colleagues at Cambridge University ran one of the most comprehensive tests of this claim. Working with the BBC program Bang Goes the Theory, they recruited more than 11,000 participants and assigned them to one of three groups. Two groups trained with cognitive tasks — one focused on reasoning, planning and problem-solving, the other on tasks resembling commercial brain training products. A control group practiced searching the internet to answer obscure questions.

After six weeks, the training groups had improved on the tasks they'd practiced. But when tested on untrained benchmark tasks, the gains did not carry over, even to closely related skills. The study appeared in Nature in 2010, and the conclusion was direct: no evidence of generalized cognitive improvement in healthy adults.

The Working Memory Trap

One category deserves its own examination: working-memory training. Working memory is the mental scratchpad you use to hold and manipulate information in real time — tracking the thread of an argument mid-conversation, keeping several variables in mind while solving a problem, remembering the beginning of a sentence by the time you reach its end. It's closely linked to fluid intelligence, and several high-profile products were built specifically to target it.

The most studied program is n-back training — a task where you track a sequence and respond when an item matches what appeared n steps earlier. Early studies suggested that sustained n-back practice might improve fluid intelligence, and the excitement was genuine.

A 2016 meta-analysis by Melby-Lervåg, Redick, and Hulme pulled together 87 publications on working-memory training and arrived at a clear conclusion. Gains on working-memory tasks were real and reliable. But they found no convincing evidence of far transfer to outcomes such as nonverbal ability, reading, and arithmetic when training was compared with a treated (active) control group.

That phrase — active control groups — matters more than it sounds. Many early studies compared trained participants against people who did nothing at all. When you instead compare them against people who spent the same amount of time doing something equally mentally engaging, the advantage of working-memory training nearly disappears.

When Researchers Took Stock

In 2014, two competing open letters circulated through the scientific community. One, from the Stanford Center on Longevity and the Max Planck Institute for Human Development and signed by roughly 70 scientists, argued that the evidence didn't support the marketing of brain games as a way to reverse cognitive decline. Another, signed by more than 100 scientists, countered that the evidence was more promising than critics acknowledged.

Rather than leave the question unresolved, Simons and colleagues reviewed the literature, closely examining the 132 papers cited by the counter-letter's proponents, and published a comprehensive assessment in Psychological Science in the Public Interest in 2016. Their conclusion was careful but pointed. They found extensive evidence that brain training improves performance on the trained tasks, less evidence for improvement on closely related tasks, and little evidence of broader transfer. As Simons put it, "we found little evidence for broad transfer from brain-training tasks to other tasks."

They also flagged design problems across many of the studies, including weak control groups that didn't match the training group in engagement.

What the Evidence Actually Supports

The dispute is narrower: whether training on a proxy task transfers to performance on a different target skill. And for a few specific approaches, the evidence is considerably more positive.

Practicing the actual skill produces consistent gains because it involves the exact cognitive process you want to improve — no transfer required because you're already doing the thing. If you want to think more analytically, work through real analytical problems in your field. If you want to retain more from books, apply spaced repetition to their actual ideas rather than to abstract memory sequences. The mechanism is straightforward: direct repetition of the target behavior improves that behavior.

Physical exercise has a substantially stronger evidence base for general cognitive benefit than any brain training program. A randomized controlled trial published in the Proceedings of the National Academy of Sciences (Erickson et al., 2011, 120 older adults) found that a year of aerobic exercise increased hippocampal volume by about 2%, with larger gains linked to better spatial memory — though the memory finding has been debated. Proposed mechanisms, such as increased BDNF and improved blood flow, operate at the physical substrate of memory and learning itself, not at the level of a game mechanic.

Sleep is performing cognitive work that no app can replicate or shortcut. Memory consolidation, the pruning of weak neural pathways, the restoration of prefrontal function — these happen during sleep, not during app sessions. Trading sleep time to fit in brain training would be, based on the current evidence, a poor exchange.

What About Puzzles and Crosswords?

A quieter question sits beneath the app debate: what about traditional puzzles? Crosswords, sudoku, chess, jigsaw puzzles — do these hold up any better?

The evidence on far transfer is similarly mixed. Crossword experts become better at crosswords. Chess players develop a remarkable perceptual memory for board positions, but that skill is largely chess-specific. Transfer to general memory or processing speed appears modest in most studies, and the effect sizes are small.

That said, there's an honest case for puzzles that has nothing to do with making your brain broadly sharper. They're engaging. Crosswords expand vocabulary and consolidate general knowledge. They occupy a mental state between passive entertainment and effortful work that many people find genuinely restorative. Enjoyment isn't a trivial outcome.

The concept behind the longer-term case for puzzles is sometimes called cognitive reserve — the idea that a lifetime of varied mental engagement may build resilience against decline, even if no single activity is a direct cause. Researchers like Yaakov Stern at Columbia have studied this framework for decades, and observational studies consistently show that more education, more occupational complexity, and more varied leisure mental activity all correlate with later onset of dementia symptoms. These are correlational findings, not randomized trials, so no causal arrow can be drawn with confidence. But the pattern suggests that accumulated mental variety and challenge matter in ways that exceed what any specific training app can capture or measure.

Where to Actually Put Your Effort

The practical takeaway from this literature isn't "brain training is worthless." It's "be specific about what you actually want to improve."

A useful test: can you draw a direct line from the game to the target skill? If yes, there's a reasonable case for it. If you need two or three inferential steps to connect faster pattern-recognition on a screen to remembering a colleague's name in a meeting, the evidence suggests the connection probably isn't there.

Three practical redirects worth considering:

1. Replace brain training time with deliberate practice on something that actually matters — writing, analysis, a language, whatever your real target is
2. Treat sleep as cognitive infrastructure, not optional recovery time
3. If you enjoy puzzles, keep enjoying them — just count them as restorative entertainment rather than cognitive conditioning

how to build better memory with spaced repetition

The problem is that "you'll get better at our game" was never what anyone actually wanted. What people are after is the confidence that their mind is growing, not just that they're getting faster at a task that exists only inside an app.

The evidence offers a more grounded path to that confidence: practice the real thing, protect your sleep, move your body, and pay attention to what's actually improving in the work that matters. That's not a magic bullet. The research, at this point, is reasonably clear that there isn't one.

Design Your Evolution means pointing your effort where it has the most leverage. On the question of cognitive training, the evidence has already done most of the pointing.

What's the one skill you've been trying to develop indirectly — through a proxy rather than the thing itself? It's worth naming it.

the science of deliberate practice and skill development