N-Back Test

Wayne Kirchner invented this task in 1958. He wanted to know how well younger and older adults kept up with information that would not sit still. His volunteers watched a row of lights and reported the one lit a few steps back. The result that made it famous was how sharply people failed as the gap grew.

The task stayed quiet for decades, then became one of the most used tools in memory research. The reason is what it asks of you. Most memory tests let you learn a list and then recall it. This one never lets you finish. Every new letter asks you a question and paints over what you were holding at the same time.

There are two ways to be right and two ways to be wrong. Press on a match and you are right. Stay quiet on a non-match and you are right too. Press on a non-match and you are wrong. Miss a match and you are also wrong.

Count only the matches you caught and you could score perfectly by pressing every single time. So we do not count it that way. We score every trial, presses and silences alike, and show your catches and your false alarms separately underneath.

If your accuracy looks high and your false alarms are high too, you are guessing freely and the headline number is flattering you. About one trial in four is a match, so pressing on everything scores near 25 per cent, not 50.

The famous training studies used dual n-back. It runs a picture stream and a sound stream at once, and you answer each one separately. It is much harder, much more irritating, and much harder to read, because the two streams interfere with each other.

Single n-back on a grid is the cleaner measurement, and the version most people came here for. Nine positions, one lit at a time. There are no words in it, which is the point. A letter stream can be held in the silent voice you use to repeat a phone number, so it partly measures how well you rehearse. A position has nothing to say.

About one square in four is a real match. We also plant near misses one step either side of the target. So a position that feels almost familiar is often exactly the trap it looks like.

In 2008 Jaeggi and colleagues reported that training on dual n-back raised scores on reasoning tests. That result launched an industry.

In 2013 Redick and colleagues ran a bigger study. Their comparison group did a real task rather than nothing, which is the harder and fairer test. They found no such benefit. Reviews since have been consistently unimpressive. Training makes you better at n-back, and at tasks that look like n-back. That is mostly where it stops.

We will not tell you this makes you smarter. It will push your n-back score up, quickly at first, and that is a real and satisfying thing to chase. For what working memory is, and why span tests and n-back measure different things, see how working memory works. For the storage side, take the digit span test and read what counts as a good digit span.