How it differs
Two comparisons, because people mean two different things by brain training. Both are about design: what each tool trains, and what it hands you at the end.
Versus the big subscription apps
They’re polished, pleasant and built to be played daily, and there’s nothing wrong with any of that. This one is built to be measured, and that runs through every other choice in it.
| Big subscription trainers | Tanren Sensei | |
|---|---|---|
| What it trains | Typically one paradigm, dressed up several ways | Memory, reasoning and speed — fourteen exercises on one engine |
| How you’re measured | By the games you practise | By five tests you never practise |
| An untrained control | None | One test we deliberately don’t train, so a general test-taking gain shows up as one |
| How difficulty moves | Usually a level you unlock | Twenty-three dials, held near the edge of what you can do |
| What goes up | A score, often unexplained | The difficulty you can hold at 80% correct, with the sum shown |
| Account | Email, usually a subscription | None. There’s no email in the product |
| Where your data lives | Their servers | Your phone. No network calls unless you turn on update checks |
| Getting your data out | Varies, often not possible | JSON export and encrypted backup, any time |
The first two rows are the ones that matter. Breadth, because training one thing makes you good at one thing. And measurement, because if a trainer scores you on the games you’ve been practising, that number will rise whether or not anything else did.
One matter of public record
In January 2016 the US Federal Trade Commission settled with Lumos Labs, makers of Lumosity, over advertising claims about real-world benefits. The order included a $2 million judgement. The FTC’s announcement is here.
We mention it once, because it’s on the public record and it’s a large part of why a measurement-first product seemed worth building. It says nothing about what that company does today.
Versus dual and quad n-back
A better-informed audience, and a fairer fight. If you’ve run the protocol in Brain Workshop or a dual n-back app, this is the section that matters.
Nothing here says n-back is broken. Pushing you to the edge of what you can do is the mechanism with the best evidence behind it, and Brain Workshop is a well-regarded free tool that implements it faithfully. We kept the mechanism. What we added is breadth, and instrumentation.
| Dual / quad n-back | Tanren Sensei | |
|---|---|---|
| What you match | Things — a position, a letter, a colour | The relationship between things |
| Kinds of training | One | Three — memory, reasoning and speed |
| How it gets harder | One knob: how far back | Twenty-three dials across the roster, each adapted separately |
| What “quad” adds | Four streams at once | Not our direction — we deepen the relationship instead |
| Reasoning practice | None | Seven families of puzzle, including “there isn’t enough to tell” |
| How you’re measured | Your n-level, a training number | Five tests you never practise, plus a control |
| Interruptions | Usually ignored | Detected, and they count for nothing either way |
| Near-miss traps | Rare, or fixed | An adapted difficulty dial |
The strongest thing we can say about the task itself, rather than about what it achieves: matching a relationship a few steps back means holding a structure and updating it, not holding a list. And where a dual n-back session gives you one thing to get better at, a week here gives you fourteen.
What we’re not claiming
That this app works better. No study shows that, and no brain training program has been shown to reliably raise general intelligence — this one included.
What the tables above describe is design: what you’re trained on, and whether the thing measuring you is the thing you’ve been practising. Those are things you can check for yourself by reading, which is why they’re worth a table. Outcomes aren’t, so they aren’t in one.
Claims on this page last checked against the app on .