Free N-Back Test

What the n-back task actually measures

The n-back task is the closest thing cognitive psychology has to a stress test for working memory. Nothing about it is hard to understand: a square appears in one of nine cells, then another, then another, and your only job is to say whether the current position matches the one from two steps ago. The difficulty is not in the rule. It is in the fact that the rule forces you to hold a short list in mind, update it on every single trial, and discard the item that has just fallen out of range — all while a new item is already arriving. That constant add-and-drop cycle is what psychologists call updating, and it is one of the three classic executive functions alongside inhibition and task switching.

This is very different from a span test. When you memorise a phone number you are storing information and then retrieving it once. Here you never get to rest on a stable memory: the contents of the buffer change every trial, and the moment you let one slip you lose the thread for the two trials that follow as well. That is why the n-back has a reputation for feeling far harder than it looks written down.

How this version is built

You get twenty-five trials on a three by three grid. Each trial lights one cell for eight tenths of a second and then leaves the grid blank until the next one appears, giving you a window of about a second and a half to respond. Roughly a third of the trials are planted matches, which is the usual proportion in laboratory versions — enough that matches feel frequent, not so many that simply pressing constantly would score well. The remaining trials include deliberate lures: positions that repeat from three or four steps back rather than two. Those lures are the whole point. Without them anyone could pass by noticing "that looked familiar" rather than actually tracking distance.

The first two trials cannot be matches, since there is nothing two steps behind them yet. On those the correct response is to do nothing at all. Your score is the percentage of the twenty-five trials you handled correctly, counting both the matches you caught and the non-matches you correctly left alone.

Reading your score

One caution about the low end: a score under sixty percent very often means the rule was misread rather than that memory failed. Plenty of first-time takers press on every repeat regardless of distance, which produces a flood of false alarms. Read the rule once more and retake it before drawing any conclusion.

Why false alarms dominate the errors

If you look at where points are actually lost, misses are the minority. The bigger leak is pressing MATCH on something that merely felt familiar. Human memory tags recently seen items with a kind of warmth, and that warmth does not carry a timestamp. A cell you saw three trials ago feels almost exactly as recent as one you saw two trials ago, so the feeling of recognition is useless for this task. Good performers stop relying on it entirely and instead maintain an explicit two-item list — "top-left, then centre" — rewriting it out loud in their head on every trial. It feels laborious, and it is, but it is the only strategy that survives the lures.

Getting a cleaner result

Sit somewhere quiet and give the run your full attention for the forty seconds it lasts; n-back is unusually sensitive to divided attention, and a single glance away costs you two or three trials rather than one. Keep your hand resting on the button so responding costs no thought. Do not try to speed up — the pace is fixed, and rushing only means you commit before the position has registered. If you find yourself completely lost mid-run, the best recovery is to skip the next two trials deliberately, rebuild the list from scratch, and carry on rather than guessing your way through the rest.

How it differs from other memory tests

A sequence memory test asks you to reproduce a growing pattern in order, which rewards chunking and rehearsal. A digit span test measures how much you can hold at once. The n-back measures neither capacity nor recall but maintenance under interference — the ability to keep a small amount of information correct while new information keeps overwriting it. That is why n-back scores correlate only modestly with span scores despite both being called working memory tasks. If you want a broader ability estimate rather than one narrow executive function, take the culture fair IQ test instead, and treat this one as what it is: a focused look at how well you juggle.