N-Back Test Online

How to play the n-back test

1. Tap Start. After the countdown you see nine holes in a 3×3 grid, your level in big type above them (for example “2-back”) and a trial counter such as “Trial 5/16”. Your first run is 1-back.
2. Coach Mochi pops out of one of the nine holes for half a second. A new turn starts every 2.5 seconds, or every 3 seconds with Large text & slower pace on.
3. If Mochi is in the same hole as n turns ago, tap the big Match! button. At 1-back you compare with the turn just before. At 2-back you compare with the one before that. The line above the grid reminds you: “Same hole as 2 back? Tap Match!”
4. If it isn’t a match, do nothing. Not tapping is your “no match” answer. A short note under the button tells you how it went: Nice match!, Not a match or That one was a match.
5. A run has 16 turns at every level, so it takes about 40 seconds (about 48 seconds with Large text & slower pace). The first n turns are a warm-up (“Warm-up: just remember the holes.”) and can never be matches, because there is nothing to compare them with yet. The 16 − n turns after them are scored, and about 30% of those are matches. The same hole never lights up more than twice in a row.
6. N-Back only has Training mode, so there is no fixed test. Each run sets your next level: 85% or better moves you up one n, below 60% moves you down one, and anything in between keeps you where you are. Levels go up to 9-back.

What the n-back measures

The skill here is working memory updating: keeping a short list in mind while it keeps changing. At 2-back you hold the last two positions. Mochi pops up somewhere new, you compare that hole with the older one, drop that one, and shift everything along. Then you do it again, every 2.5 seconds. Each decision is only “yes or no,” but the hold, compare and update loop gets heavy fast as n grows.

Wayne Kirchner described the task in 1958 in a study of age differences in short-term memory. It remains a common working memory task in research labs. A 2018 study of 533 adults aged 20 to 80 found that reaction times rose and the share of detected targets fell from the youngest group to the oldest (Gajewski et al., 2018).

Honest limits:

• This is a game and a training habit. It is not a clinical test and it can’t diagnose anything.
• This version (positions only, 16 turns per run, played on a phone) has no published norm.
• Practice makes you better at n-back, and that is all we claim. A meta-analysis of 87 publications and 145 comparisons found no convincing improvement in intelligence, reading or arithmetic when working memory training was compared with an active control group (Melby-Lervåg et al., 2016). A broad review of brain-training studies reached a similar conclusion: people improve on the trained task, with little evidence of transfer to everyday performance (Simons et al., 2016).

N-back averages and how the score works

There is no peer-reviewed average for this exact version, so we won’t make one up. Peer comparison may come later. For now, the person to beat is you.

Your score is your accuracy: (matches you caught + non-matches you correctly left alone) ÷ scored turns × 100. At 2-back, for example, 14 of the 16 turns are scored and 4 of them are matches, so every mistake costs about 7 points. A missed match and a false tap cost the same.

One thing to know: tapping nothing at all still scores about 70%, because about 70% of the scored turns are non-matches. At 2-back you’d get all 10 non-matches right and miss all 4 matches: 71.4%. So read 70% as the floor, not as a decent result, and treat your level (n) as your main progress marker. The result screen tells you when it changes (“Level up! Next time: level 3”), and your records page shows your current training level.