Gyroscope Benchmarks / Mind
N-back test: how good is your working memory?
This test measures how accurately you keep a short sequence active and update it as each new tile arrives.
Why it matters
Working memory is the small workspace you think in. It holds a phone number while you find a pen, the first half of a sentence while someone finishes the second, the turn you just took while you decide on the next one. Updating is the part a 2-back measures: keeping the last two items ready, dropping the oldest as the newest arrives, and not mixing them up. You use that every time you follow a conversation in a loud room, cook two dishes at once, or read a paragraph and still know how it started.
It is also among the first things to go on a bad day. Short sleep, a cold, a hard week at work, a phone that buzzed twice during the round: all of them borrow from the same small workspace. In one study, two weeks of six-hour nights produced cognitive deficits equal to two full nights without sleep, while the people rated themselves only mildly impaired. Feeling sharp and being sharp are different measurements.
Van Dongen et al., Sleep, 2003
Working memory also tends to decline slowly across adulthood, and the change is hard to notice from the inside because you have nothing to compare it with. A d-prime from the same 2-back, on the same phone, once a month, gives you that comparison. One run is a noisy reading. Twelve runs on one line show whether your good weeks and bad weeks match what the app already knows about your sleep, your training, and your days.
What this test measures
N-Back Grid is a spatial 2-back task. A tile appears, then another, and you tap when the current position matches the position from two turns earlier. The sequence never stops moving, so you must keep recent positions available while replacing older ones.
Three raw measures tell the story. Hit rate is the share of true matches you catch. False-alarm rate is how often you tap on a non-match. Sensitivity, called d-prime, combines hits and false alarms so that frequent tapping cannot masquerade as good detection.
A meta-analysis assembled 668 activation coordinates from 24 n-back imaging studies. Across verbal and nonverbal variants, frontal and parietal regions were consistently engaged, with additional differences by task content and process.
Owen et al., Human Brain Mapping, 2005
That does not make one n-back result a pure measure of working-memory capacity. Across three experiments, Jaeggi and colleagues found insufficient reliability for ranking individual differences, while retaining the task's value for experimental working-memory research.
Do the test right now
- 1Open N-Back Grid in Gyroscope. A round takes about 20–40 seconds.
- 2Watch each tile position. On every new tile, compare it with the position shown two turns ago.
- 3Tap only when the positions match. Let non-matches pass without tapping.
- 4Keep hit rate, false-alarm rate, and d-prime. Do not use hit rate alone, because tapping too often can raise hits and false alarms together.
Reference bands
N-back results depend on stimulus timing, target frequency, load, and scoring. Gyroscope therefore uses practical d-prime bands for its own spatial 2-back version rather than transferring a lab norm.
| D-prime | Practical band |
|---|---|
| Under 0.5 | Near chance |
| 1.0–1.5 | Typical first run |
| Over 2.0 | Strong |
Owen et al., Human Brain Mapping, 2005. Research context for Gyroscope's practical bands.
The gaps from 0.5 to 1.0 and from 1.5 to 2.0 are transition zones, not missing diagnoses. Treat a score as one run under one protocol. The more useful comparison is your own repeated result on the same version.
What moves your score
Working-memory performance is state-sensitive. In a meta-analysis of 70 short-term sleep-deprivation articles covering 147 tests, working-memory performance showed a meaningful negative effect while lapses in simple attention showed the largest effect, Hedges's g -0.776.
Lim and Dinges, Psychological Bulletin, 2010
- Sleep loss and fatigue can make the sequence harder to update.
- Practice improves familiarity with the 2-back rule.
- Stress and illness can pull attention away from the current tile.
- Time of day can change alertness, so compare similar test windows.
How to improve
What raises d-prime is a cleaner updating routine, not faster tapping. Most lost points come from false alarms and from losing the thread after one miss, so the gains are in how you hold the pair and recover.
- Hold two, not the whole run. Keep only the last two positions in mind. When a new tile appears, compare it with the older of the two, then let that one go. Carrying five or six positions back is what makes the middle of a round fall apart.
- Rehearse in a rhythm. Name the positions to yourself on a steady beat: top left, center, top left. A rhythm keeps the pair in order when tiles come quickly, and order is what the 2-back tests.
- Keep your eyes on the center. Watch the middle of the grid instead of chasing each tile. The whole grid stays in view and each position registers as a place, not a movement.
- Decide before you move. Tap only when you are sure. A lucky guess counts as a hit, but every guess that misses counts as a false alarm, and d-prime charges you for each one.
- Let a miss go. When you lose the thread, do not rebuild the sequence you dropped. Start a fresh pair with the next two tiles. One lost match costs less than the four you lose while reconstructing.
- Run it with nothing else on. Silence notifications on the phone you are testing on and put any other device out of reach. In one study a phone on the desk, face down and silent, measurably reduced working memory and attention.
Ward et al., Journal of the Association for Consumer Research, 2017
Retest once a month, on the same phone and mode, rested and at a similar hour, and log the round even when it went badly. The same protocol each time is what gives the line its meaning.
Track it
Log it in the app.
Download Gyroscope and run N-Back Grid in about a minute. See how you compare to your peers, to the studies, and to your own baseline over time.
Download the appGet Gyroscope One