Sustained Attention Test

Robertson and colleagues published this task in 1997 under the title 'Oops!', which is the right name for it. Eight of every nine digits need a press. Within about twenty seconds, pressing turns into a rhythm. The digit you must not press then has to break a rhythm rather than a decision.

That is the whole trick. Older watchkeeping tests ask you to spot a rare event, so failing looks like doing nothing at all. This one flips that round. Failing looks like doing something, which turns a lapse of attention into a button press you can count.

It was tested in people with brain injuries. Their error counts matched the everyday slips their relatives reported.

The result card shows three numbers, because an error count on its own is easy to fake by going slowly. Your average press time says whether you were fast enough for the task to be hard at all.

The third number is your variability. That is how much your press times jumped about, and it is the one attention researchers care about most.

Variability picks up mind wandering better than errors do. Seli and colleagues showed that press times turn erratic in the seconds before a lapse. So a run with few errors and a wide spread usually means you drifted often and got lucky about when the 3 appeared.

The original runs 225 digits over roughly four and a half minutes. It shows them in five randomly mixed sizes, so you cannot watch for a shape and stop reading. This version runs 45 digits by default, a little over a minute, because nobody finishes a four-and-a-half minute task on a web page.

The pace is slower too. The published task shows each digit for 250 milliseconds and then blanks the screen for 900, so a trial lasts 1.15 seconds. Here the digit is shown for the same 250, and the blank runs 1,250. A trial lasts 1.5 seconds.

That extra third of a second is there because you answer with a button rather than a key already under your finger. On a phone it is a thumb crossing the screen. It also makes the task slightly easier, so expect a few fewer wrong presses here than the published figures would suggest.

Shorter runs mean fewer chances to slip, which makes the error count jumpier. Five errors out of five chances and five out of twenty-five are very different results, and a short run cannot tell them apart cleanly.

That is what the 90-digit option is for. Use its numbers if you plan to track yourself over time. The mixed sizes are kept in both.

This is not attention span in the everyday sense. Nothing here estimates how long you can concentrate on a book or a lecture. It measures how well one specific habit can be interrupted over a minute or two, under conditions built to make interrupting hard.

A high error count often means you were tired, bored, interrupted or holding a phone. Repeat the task under the same conditions before comparing runs.

For a measure that tracks tiredness instead, use the psychomotor vigilance task. For the self-control side of the same coin, use the go/no-go test.

Two further readings if the mechanics interest you. What the Stroop effect actually shows covers why subtracting one score from another beats a single total. Average reaction time by age shows how much any timing measure moves for reasons that have nothing to do with attention.