Does it actually work?

The short answer

Three things are certain. You will get better at these exercises — quickly, and by a lot. You’ll be able to see exactly how much, because the app writes the number down. And it’s genuinely interesting to do, which matters more than it sounds, because nothing works if you stop.

One thing is honestly unresolved: whether getting better at these carries over into everyday thinking — sharper focus, a memory that drops fewer things, following a complicated argument without losing the thread.

Plenty of people who’ve trained seriously say it does. Most controlled trials haven’t managed to show it. Both of those statements are true at once, and the rest of this page is why.

What people who train seriously report

There’s a community around this kind of training that’s been going for well over a decade — people running the protocol for months at a stretch and posting what happened, in detail, without anything to sell.

The best-known name is Dave Asprey, who has written publicly about dual n-back as one of the things he credits for a measured rise in his own IQ score. He’s not alone: there are people on YouTube who’ve documented running quad n-back daily for months, and forum threads going back years where the same handful of things come up again and again — reading gets easier, conversations in noisy rooms get easier, the feeling of holding several threads at once without one slipping.

Those are claims, not findings. But there are a lot of them, they’re consistent with each other, and they come from the group the research has barely looked at: people who actually wanted the result and kept going.

Here’s the catch, and we’d rather say it than have you notice it later. Someone who takes an IQ test, trains for six months, and takes it again will usually score higher — partly for reasons that have nothing to do with training. You’ve seen the format before. You know the tricks the questions use. You’re more relaxed. That’s called the retest effect, it’s large, and it makes almost every self-reported “my IQ went up” story impossible to read on its own.

This isn’t us being difficult about it. It’s the exact problem the app was built to solve, and it’s why there’s a test in there that we deliberately never train. More on that below.

What the research found, and why people still argue

In 2008 a study found that training on dual n-back improved performance on tests of fluid reasoning — the ability to work out a pattern you’ve never seen before — and that the more people trained, the more they gained. It set off everything that followed.

Since then, results have gone both ways, and the disagreement is real. Research teams have run reviews across overlapping sets of the same studies and come to opposite conclusions: one group finds a small but consistent benefit, another finds nothing once you account for how the comparison groups were set up. When careful people reading the same evidence disagree that sharply, “the science says it doesn’t work” is not an accurate summary. Neither is the opposite.

Five reasons the trials might have missed something

None of these prove the training works. They’re reasons the studies we have aren’t well shaped to detect it if it does.

They were short. Most ran between two and five weeks. Nobody who reports real change from this kind of training is talking about three weeks — they’re talking about six months. A trial that stops at week four is measuring the warm-up.

The participants weren’t trying. Most studies recruited undergraduates for course credit or a small payment. This training is genuinely hard and gets harder the better you do; how much you push is the whole variable. A student finishing a required session is not doing the same activity as someone training because they want their mind to work better.

Usually one test, taken once. Transfer was often judged on a single reasoning test on a single day, sometimes under time pressure. That’s a narrow window, and a real effect can slip through a narrow window.

Averages hide the people who improved. These studies report the group average. If a third of people gain a lot and the rest gain nothing, the average is small and the headline reads “no effect” — while the interesting question, who gained and why, goes unasked. Some of the original work found exactly that pattern: the people who improved most at the training improved most on everything else.

Almost nobody came back later. Very few trials followed anyone up months afterwards. Something that takes a while to show, or shows only once the training becomes a habit, would simply never appear.

And the fair objection to all five: those are also exactly the arguments you’d reach for if you’d decided in advance that it works. We know. A reason a study might have missed something is not evidence that there was something to miss. That’s why we didn’t stop at making the argument.

So we put the instruments in the app instead

You don’t have to believe us, a forum, or a paper. You can measure it on yourself.

Five tests come with the app. You never practise them, their difficulty never changes, and they never feed into your training, so a score from today and a score from next month mean the same thing. Four of them measure the kind of thinking the training is aimed at.

The fifth is the one that makes the other four worth reading. It measures 3D mental rotation, and we deliberately never train it. It’s a control. If it climbs just as much as the other four, the honest reading is that you got better at taking tests — the retest effect — rather than at thinking. You’ll see that yourself, in the app, without having to ask us.

We built a test whose whole job is to catch us out, and then put it on the results screen. Most apps in this category would much rather you never had a way to find that out.

What we will and won’t promise

We promise you’ll get better at the exercises, that you’ll be able to prove it from your own numbers, and that every bit of encouragement in the app is attached to something you actually did. The applause fires when the difficulty really rises, never for showing up. The streak counts days you trained, and a forgiven day is not one of them. Nothing here inflates a number to keep you opening the app.

We won’t promise a number of IQ points, a better memory, or sharper focus at work. Nobody can honestly promise you those, and anyone who does is telling you something they don’t know.

What we can say is that the mechanism has better evidence behind it than anything else in the category, that the people who’ve pushed it hardest are the ones who report the most, and that for the first time you’ll be holding the instruments to check it on yourself.

Take the tests before you start. Train for a few weeks. Take them again. Then read your own numbers, control and all.

What your first month looks like → · How it all works →