Sustained Attention
to Response Task
A browser implementation of Robertson and colleagues’ 1997 paradigm. Digits arrive faster than you can deliberate about them, so responding becomes automatic — and the moment your attention drifts, the task catches it, because the one thing you must not do is the thing your hand is already doing.
The task
Every ninth digit is a 3, and the digits change size from trial to trial so that each one has to be read rather than recognised by shape. Expect to press on some of them: healthy adults typically do on around two in five, and a task nobody failed would have nothing to measure. On a phone, tap anywhere instead of SPACE.
Protocol & options
- No-go digit
- 3
- Digit on screen
- 250 ms
- Mask
- 900 ms
- Trial length
- 1150 ms
- No-go rate
- 1 in 9
These are fixed rather than adjustable, which is the one place this task differs from the others here. The SART’s timing is not a comfort setting — it is what makes responding automatic, and the published error rates only describe a run that used it. A session at a slower pace would be a different instrument wearing the same name, and could not be compared with your own earlier runs or with anyone else’s.
This is a research-style measurement instrument and an educational demonstration — not a medical or diagnostic device. A single session is noisy, SART error rates overlap heavily between people with and without an attention diagnosis, and the measure moves with sleep, caffeine, motivation and how cautious you decide to be. Nothing here can confirm or rule out a condition or guide medication decisions; that's a clinician's job.
SART (Sustained Attention)
What this test measures
The Sustained Attention to Response Task (Robertson, Manly, Andrade, Baddeley & Yiend, 1997) is a go/no-go task built back to front. Digits from 1 to 9 arrive every 1.15 seconds and you respond to all of them except the 3. Because eight trials in nine call for a press, responding quickly stops being a decision and becomes a rhythm — and the rare digit you must withhold on is precisely where a wandering mind gives itself away.
That inversion is the whole design. In a conventional vigilance task the rare event is the one you respond to, so a lapse of attention produces nothing — a miss that looks the same as sitting still. Here a lapse produces an action: your hand presses on the 3 because it was already pressing on everything. The failure becomes a positive event in the data, which is why the SART became the standard behavioural probe for mind-wandering rather than just another inhibition test.
Errors on this task are common and are supposed to be. Healthy adults press on roughly two in five no-go digits; a version nobody failed would have no room to vary with attention. What distinguishes a good session from a poor one is less the raw count than the pattern around it — how consistent the responding was, and whether it had already sped up in the seconds before each error.
What the report shows you
Alongside the error rate, the report contrasts your response times in the four trials leading into an error with those leading into a correct hold. Across the literature the run-up to an error is measurably faster, and that gap is the closest thing this task has to a direct read on attention drifting off the job: the error does not begin at the 3, it begins earlier, when responding slips into automatic pacing. It also reports response-time variability, post-error slowing, and whether you slowed across the session.
One honest limitation is worth knowing before you start. The error rate can be driven almost to zero simply by responding slowly and deliberately to every digit, which measures how cautious you decided to be rather than how well your attention held. The report checks your response times against your error rate and says so when it sees that pattern, and the session score weighs consistency alongside errors for the same reason.
How to use it
Respond as fast as you comfortably can and accept that some 3s will get past you — that is the task working, not you failing it. Take it under consistent conditions and the trend is what carries the information: attention state moves with sleep, time of day, and stimulation, which is what makes this worth repeating. Signed-in sessions feed the sustained-attention trend on the Progress page.
Sapiens Bench tasks are educational instruments for self-tracking, not medical or diagnostic devices.