Skip to content

What an N-Back Task Actually Asks You to Do

What an n-back task asks you to do, why dual n-back feels unlike other memory games, and what the training studies actually found about transfer.

By Published Updated 6 min read

Reign Creative LLC makes IQ Test & Brain Training Games, which this article links to, so we have a commercial interest in it. How we write and check articles.

Key takeaways

  • An n-back task asks one question repeatedly: does the item on screen right now match the item from n steps back? You answer on every item, so there is no pause and no end-of-round recall.
  • The load does not grow as the run continues. At 2-back you are always holding two items, which is why the task feels like constant maintenance rather than like filling a bucket.
  • Dual n-back runs two streams at once — typically a position and a sound — with an independent match decision for each, which is what makes it so much harder than the single-stream version.
  • Thompson and colleagues ran 20 sessions of adaptive dual n-back against an active control and found substantial, lasting gains on the trained task with no transfer to untrained working-memory or fluid-intelligence measures.
  • A 2014 consensus statement from the Stanford Center on Longevity and the Max Planck Institute for Human Development, signed by dozens of scientists, holds that the literature does not support claims that brain games improve general cognitive performance in everyday life.
  • IQ Test & Brain Training Games includes a short N-Back Lite round in its memory catalogue. It is a game mode, not a training protocol, and we make no claim that it changes anything beyond performance on itself.

An n-back task asks you one question over and over: does the thing on screen right now match the thing that appeared n steps ago? Items arrive in a continuous stream, you answer on every item, and there is no recall phase at the end. That single design choice is what makes the task feel unlike every other memory game, and it is also why it became the standard laboratory tool for studying working-memory updating.

The task, step by step

Suppose letters appear one per second and you are running a 2-back. Take this sequence, which is short enough to check by hand:

K, M, K, P, M, P, P

You cannot answer on the first two items — there is nothing two steps back yet. From the third item onward, compare each letter with the one two places earlier:

  • Item 3 is K; item 1 was K. Match.
  • Item 4 is P; item 2 was M. No match.
  • Item 5 is M; item 3 was K. No match.
  • Item 6 is P; item 4 was P. Match.
  • Item 7 is P; item 5 was M. No match.

Two matches out of five decisions. Now run the same sequence as a 3-back and the answers change completely: item 5 (M) matches item 2 (M), item 7 (P) matches item 4 (P), and the two matches you found before are no longer matches at all. Same letters, different task. That is worth sitting with, because it shows that nothing about the items themselves is being tested. What is being tested is your ability to keep a moving window open at a particular width.

Why the load stays flat instead of growing

Most memory games get harder by giving you more. A digit-span trial adds a digit. A Simon-style sequence adds a step. The demand accumulates, you hit a wall, and the run ends.

N-back does not work that way. At 2-back you are holding two items at the start of the run and two items at the end of it. The quantity never changes. What changes on every single step is the contents: you add the new item, drop the item that just fell out of the window, and keep the order straight — all while making a yes/no decision that itself takes attention.

This is continuous updating, and it is a genuinely different demand from storage. It is also why people describe n-back as exhausting rather than as difficult. There is no moment of relief, no phase where you get to unload, and no natural place to rehearse. Working memory is a small resource to begin with — Nelson Cowan's review puts the central store at roughly three to five meaningful chunks in young adults — and n-back spends that resource on maintenance and replacement at the same time.

Single versus dual

The version that made the task famous is dual n-back, and it is worth being precise about what "dual" means. Two independent streams run at once: commonly a square moving around a grid and a letter spoken aloud. Each has its own match decision at the same value of n, and the two streams are uncorrelated, so knowing that the position matched tells you nothing about whether the sound did.

That is not twice the work; in practice it is considerably more, because the two queues compete for the same limited store and both need updating on every step. Most people who can hold a comfortable 3-back on one stream find dual 3-back close to unplayable at first.

What the training studies actually found

This is the part the marketing usually skips, and the picture is genuinely mixed rather than uniformly negative.

The clearest negative result comes from Thompson and colleagues, writing in PLoS One in 2013. Participants completed 20 sessions of adaptive dual n-back — about 40 minutes a day, five days a week, for four weeks — compared against an active control group who trained on a multiple-object-tracking task for the same schedule, plus a passive no-contact group. Both training groups showed substantial and specific improvements on their trained tasks that persisted for at least six months. Transfer to anything else did not appear: no improvement on complex span measures, and none on the fluid-intelligence tasks relative to either control group.

A different framing comes from Gathercole and colleagues in the Journal of Memory and Language, who argue that transfer happens only when training causes you to learn a new complex cognitive skill that can then be applied to an untrained activity. On that account transfer is real but narrow: it follows shared structure between tasks, not general strengthening. Where a task can already be handled by routines you have, no new routine gets built, and nothing transfers even between tasks that look similar.

Above both sits the 2014 consensus statement issued jointly by the Stanford Center on Longevity and the Max Planck Institute for Human Development, signed by dozens of cognitive scientists and neuroscientists, which holds that the scientific literature does not support claims that software-based brain games improve general cognitive performance in everyday life or prevent cognitive decline. Our own cornerstone piece on whether brain training apps work sets out both sides of that argument, including the meta-analysis that reached a more positive conclusion about n-back specifically.

Playing an n-back round without the marketing

IQ Test & Brain Training Games is our own app, built by Reign Creative, and it includes N-Back Lite as one of the twelve challenge modes in its memory catalogue — a short round of the task alongside things like Flash Recall, Rapid Recognition and Dual Attribute. The name is deliberately modest. A brief round inside a memory game is not the multi-week adaptive protocol the studies above used, and we do not present it as one. It is there because the task is interesting and unusual, not because it does something to you.

If you want to play it well, three things help:

  • Do not try to visualise the whole run. Hold exactly n items and let everything older go. Trying to keep a longer history is the most common way people fail a run they could have passed.
  • Say the window out loud, or sub-vocally. Verbalising the two or three current items gives the queue something to hold onto during the gap between stimuli.
  • Expect the errors to cluster after a match. Registering a match takes a beat, and the next item usually arrives during that beat. Knowing this does not fix it, but it stops you reading the following mistake as a collapse.

More on structuring short sessions so they actually repeat is in our guide to building a daily practice habit, and the other span and updating tasks are covered across the memory and attention hub. The app itself is described on the IQ Test & Brain Training Games app page, and the rest of the education and brain training category covers the arithmetic side of the same app.

Is it worth doing?

If you find the task interesting, yes — as practice and as entertainment, which are honest categories. You will get better at n-back, probably quite quickly, and that improvement is real.

What you should not do is buy the older promise. Getting better at holding a two-item window while a stream moves is a skill that applies to holding a two-item window while a stream moves. The strongest well-controlled test of the training protocol found exactly that and nothing more, and the broad scientific consensus is against the general-improvement claim. An n-back round is a game mode with an unusual demand. That is enough of a reason to play it, and it is the only reason we offer.

Good to know

Frequently asked questions

How does an n-back task work?
Items appear one at a time in a continuous stream. For each item you decide whether it matches the item that appeared n positions earlier and respond immediately. At 2-back, the fourth item is compared with the second, the fifth with the third, and so on. You never get a recall phase at the end; the decision happens on every item while the stream keeps moving.
What is the difference between single and dual n-back?
Single n-back runs one stream of items and asks for one match decision per step. Dual n-back runs two streams simultaneously — commonly a visual position and an auditory letter — and asks for an independent decision about each. The streams are uncorrelated, so you maintain and update two separate queues at once. That is why dual n-back is substantially harder at the same value of n.
Does n-back training make you smarter?
The evidence does not support that claim and we do not make it. Thompson and colleagues trained participants for 20 sessions of adaptive dual n-back and found large, durable improvements on the trained task but no transfer to untrained working-memory tasks or fluid intelligence. Other reviews report small effects under different designs. The disagreement is real, but nothing in it licenses a promise.
Why does n-back feel so different from other memory games?
Because it never lets the load settle. A span task gives you a sequence, a pause, and a chance to dump everything you are holding. An n-back stream requires you to add the new item and drop the oldest one on every single step, while also making a decision. It is continuous updating rather than accumulation, and that is an unfamiliar demand for most people.

Sources

  1. Failure of Working Memory Training to Enhance Cognition or Intelligence (opens in a new tab)PLoS One, via PubMed Central (PMC3661602) — accessed
  2. Working memory training involves learning new skills (opens in a new tab)Journal of Memory and Language, via PubMed Central (PMC6591133) — accessed
  3. A Consensus on the Brain Training Industry from the Scientific Community (opens in a new tab)Stanford Center on Longevity and Max Planck Institute for Human Development — accessed
  4. The Magical Mystery Four: How is Working Memory Capacity Limited, and Why? (opens in a new tab)Current Directions in Psychological Science, via PubMed Central (PMC2864034) — accessed

About the author

— The independent app studio that builds the apps covered on this site and publishes every article here. On Google Play it publishes as Reign Collective Apps.

We use AI tools to help research, draft and edit our articles. Each article lists the sources its factual claims rely on, and details about our apps come from their live Google Play listings. No article on this site has been reviewed by an independent expert unless the reviewer is named on it.

Spotted an error? Tell us — how corrections are handled is set out in our editorial policy.

Keep reading