Three short levels, three or four minutes, one brain age — and every single year of it is shown as an equation you can recompute yourself
Find a few minutes where nobody interrupts you — a careless run tells you nothing
Your scores stay in this browser, on this device
An unfinished run saves nothing — reloading mid-run starts you over
The shape of a run: three short levels each produce one explainable raw metric, each metric maps through a published table to one age, and the three ages are averaged with equal weight into the brain age. The order is fixed — speed, then memory, then focus — because changing it changes how fatigue and practice are spread, which would make runs incomparable. A first run takes about 3–4 minutes; from the second run on the tutorials collapse and the practice go can be skipped, roughly 2.5 minutes.
Level 1, the tap level (speed): four plain baseline taps, then 12 judged trials — 8 blue discs that want a tap and 4 amber squares that do not, reshuffled every run, with trial 1 always a disc. Each stimulus is preceded by a random 800–2200ms blank. A disc waits at most 1.5s before it counts as a miss; a square clears itself after 1.2s. A tap inside 120ms of the stimulus painting is a false start: that trial is voided and re-served, scoring nothing and penalising nothing, so jumping the gun can never mint a fake fast reading. The metric is net decision time = median of the 8 correct disc taps minus the median of the 4 baseline taps.
Level 2, the memory level (working memory): tiles on a 3×3 grid light for 500ms with a 250ms gap, then you tap the sequence back. It starts at length 3, grows by one on every clean sequence, and caps at 9. One wrong tile ends the sequence immediately and the correct order is replayed slowly; the first failure repeats the same length, and two failures end the level. Five seconds with no input counts as a failure. The metric is the longest sequence you actually completed.
Level 3, the switching level (attention and task switching): digits 1–8 and letters A–H are scattered on a 4×4 grid and you alternate 1 → A → 2 → B … 8 → H. One completed round of 16 steps triggers an instant reshuffle in which at least 8 cells change place. The clock is a fixed 40 seconds and starts on your first tap. A correct tap dims the cell and names the next target; a wrong tap flashes, advances nothing and costs no time. The metrics are total correct steps and wrong taps.
The conversion tables (an entertainment scale we set ourselves, printed in full on the page so you can recompute any result): speed uses speedRaw = net decision time + 15×false taps + 25×misses, then age = 18 + 62 × (speedRaw − 60) / 360, with speedRaw ≤60ms reading 18 and ≥420ms reading 80. Memory uses age = 80 − (span − 3) × 62 / 6, so a span of 3 reads 80 and a span of 9 reads 18. Focus uses focusRaw = correct steps − 2 × wrong taps, then age = 80 − (focusRaw − 8) × 62 / 34, with focusRaw ≤8 reading 80 and ≥42 reading 18. The three ages are averaged with equal weight and clamped to 18–80. There is no randomness anywhere in that chain: identical raw scores always produce an identical brain age.
When a part cannot be measured, and the tier names: the speed part needs at least 3 baseline samples and 5 correct disc taps, the memory part needs at least one completed sequence, and the focus part needs at least 4 correct steps. If any part falls short, no brain age is assembled — the result page shows only the parts that were measured and says why. The five tiers are 18–29 razor sharp, 30–39 firing well, 40–49 right in the middle, 50–64 a step behind, 65–80 not your day; those thresholds are our own entertainment scale and do not represent any real population distribution.
Relaxed mode and passing the phone: relaxed mode enlarges the type, stretches the disc dwell by half, slows the memory playback to 800/400ms while allowing three failures, and gives the switching level 60 seconds. Because none of that is comparable with a standard run, relaxed mode explicitly produces no brain age and writes nothing to your history. Pass-the-phone supports 2–6 players on one identical board; each handover walks through hand it over, confirm who you are, then say you are ready, and no stimulus appears until that third tap. Once everybody has played, the board ranks by lowest brain age, then longest span, then most focus steps, then fastest net decision time.
Records, privacy and the usual misreadings: scores live only in this browser on this device, the newest 20 runs are kept, and settings has a one-tap wipe. There is no sign-up, no email, no daily cap and no network request of any kind while you play. Reloading mid-run voids that run — a game that measures milliseconds and runs a clock cannot be honestly restored. Repeat runs get easier and the number naturally drops, which is why the result page keeps your first score alongside every later one: the first run is the meaningful one, and later runs are better read as "how am I doing today".
⚠️ For entertainment only — not a screening or diagnostic tool.
Last updated: 2026-08-30 · Saved on this device only
