What to Look for in a Brain Game

Walk through any app store and search "brain training." You'll find hundreds of products, most wrapped in the same visual language: glowing neurons, upward-trending graphs, a percentile score implying you're being measured against humanity. The category earned its skepticism honestly. In 2016 the Federal Trade Commission fined the biggest name in the business, Lumosity, $2 million for advertising benefits its science couldn't support, and in 2014 some seventy researchers signed a public statement objecting to the gap between brain-game marketing and brain-game evidence.

And yet the underlying research isn't empty. Cognitive training reliably improves the skills it trains, sometimes durably; the large NIH-funded ACTIVE trial found gains that persisted for years after a handful of sessions. The category's problem was never that training does nothing. It's that products claimed far more than training does, and consumers had no tools for telling the difference.

This is an attempt to hand you those tools. We build a brain-training product, so read this the way you'd read a wax company's guide to wax: useful information from a source with an interest, laid out plainly enough that you can check our work. Every test below applies to us. Some of them, frankly, are hard to pass, which is rather the point.

Test one: does it adapt?

The founding principle of all training, physical or mental, is progressive overload. A task you can do easily produces no adaptation; a task hopelessly beyond you produces frustration and quitting. Learning happens at the edge, and the edge moves.

So the first thing to examine is whether difficulty adjusts to you, continuously. Good training software behaves like a good coach: when you're succeeding consistently, it adds load, more items, less time, more interference; when you're failing, it backs off until you're succeeding again, keeping you at a challenge level where success is frequent but never guaranteed. If an app serves fixed puzzles that a week of practice renders comfortable, you're not training anymore. You're rehearsing, pleasantly. Rehearsal has its place, but nobody should sell it to you as a workout.

A related tell: does your score keep climbing indefinitely on the same task? Early gains on any new task are largely you learning the task itself, its controls, its rhythm, its tricks. That's real learning, but it's the shallowest kind. A serious product expects that curve, keeps moving the target after it flattens, and is candid that "your score doubled" partly means "you got familiar."

Test two: is there real variety?

Memory is not one thing. Recalling a name, holding a phone number, remembering where you parked, and noticing that you've already told someone this story are different skills with different machinery. The research is blunt about the consequence: training transfers narrowly. Improvement on one task extends, at best, to closely similar tasks.

That makes task variety a structural honesty test. A product built around one endlessly repeated mini-game is, whatever its charts imply, training one narrow skill. A credible program trains a portfolio, across memory, attention, and speed, and rotates it, for the same reason your physical training has intervals and long days and strength work. Variety isn't garnish. Given narrow transfer, it's the only way a training program can be broad at all.

Test three: does it point toward life?

Since transfer is narrow, ask the sharpest question in the whole evaluation: transfer to what? What, specifically, would this task make me better at outside the app?

Some training tasks wear their answer openly. Practice attaching names to faces under time pressure and recalling them after a delay, and you are practicing something you do at every meeting and reunion; the skill and the life situation are nearly the same shape. Speed-of-processing tasks descend from a paradigm with a track record in research settings. But some tasks are simply games with a cognitive costume, and the tell is that no one can finish the sentence "this makes you better at..." without hand-waving. You deserve the finished sentence. A product confident in its design will tell you, per exercise, what's being trained and why that target was chosen, and will admit the honest residue: better at the task for sure, better at close cousins probably, and beyond that, unproven.

Test four: read the claims like a referee

Here is the fastest sorting mechanism in the category, and it requires no scientific training at all. Find the product's biggest promise and check which side of one line it falls on.

Claims about skills are potentially legitimate: train your memory, sharpen your attention, get faster at processing. These can be backed by evidence, and you can even verify them yourself, since you'll feel a trained task getting easier. Claims about disease are the red flag: anything promising to prevent, delay, or protect against dementia, Alzheimer's, memory loss, or cognitive decline. No brain-training product has earned that claim. The research that gets cited for it, most often an exploratory ten-year analysis from the ACTIVE trial linking speed training to lower dementia rates, is a genuinely interesting scientific lead that its own context demands be held loosely: a secondary analysis, not the trial's primary endpoint, still awaiting replication. When a company converts that into a promise of protection, it is spending the credibility of researchers who were careful with money they never authorized. The FTC's Lumosity complaint was, at bottom, about exactly this conversion.

While you're reading, apply two smaller filters. Beware the phrase "clinically proven" unattached to any named, published study; real evidence has citations. And discount testimonials entirely, since every product in every category has delighted customers, including products that do nothing.

Test five: does it respect you?

Finally, some softer tells that predict whether a product deserves a place in your daily life.

Watch the emotional register of the marketing. There's a difference between an invitation to train and an appeal to dread. Products that lead with fear of decline are telling you their theory of you, and fear-first companies tend to be claim-inflators, because fear is what inflated claims are for. Watch the engagement machinery: streaks and reminders used gently serve your goals, while guilt-tripping notifications and manipulative urgency serve retention metrics. And watch whether the product ever tells you anything against its own interest, such as the limits of transfer, or the fact that the best-evidenced brain intervention for older adults is aerobic exercise, which costs nothing and belongs to no app.

That last one is the test we'd most like to be held to. A brain-training product is a tool for practicing specific, useful mental skills a few focused minutes a day, inside a life that also includes movement, sleep, and people. Sold as that, it can be honestly excellent. Sold as armor against what you fear, it's Lumosity in 2015. Now you know how to tell them apart, whoever's logo is on the screen.