Attention span test: method

What it measures

Sustained (vigilant) attention in a simple visual reaction-time task with random foreperiods: attentional lapses and response speed.

Procedure

After Start, a fixed 3-minute run. A circle in the response area turns green after a random interval drawn uniformly from 1 to 4 s; this interval includes the 1 s feedback after the previous response. The visitor responds by tapping or clicking the response area, or pressing Space or Enter. Each reaction time (ms, measured with performance.now) is shown for 1 s. A response during the wait is a false start: 'too early' is shown and a new interval starts. With no response within 30 s of the stimulus, the trial ends as a time-out and is recorded as a 30,000 ms response. When 180 s have passed, the run ends at the next scheduling point. No practice trials. Leaving the tab or window during the run interrupts it, and it must be restarted.

Parameters

Scoring

Primary metric: number of lapses = valid responses with RT ≥ 355 ms (fewer is better). Valid responses are those with RT ≥ 100 ms; time-outs enter as 30,000 ms valid responses and count as lapses. Also reported: mean RT over valid responses (ms, lower is faster), fastest and slowest valid RT, and false starts (responses under 100 ms plus responses before the stimulus), which are excluded from the RT statistics. Each response is also labelled early, slow or ok for the result chart.

Valid-run rules

Runs interrupted by leaving the page are not scored. The lapse count is reported only when at least one valid response exists (so a mean RT can be computed). Abuse limits: at most 400 recorded responses and 1,000 false starts per run. Runs with many false starts are scored, not excluded.

Differences from the classic task

The core parameters follow the published PVT-B: 3 minutes, 1–4 s intervals including feedback, a 355 ms lapse threshold and a 100 ms false-start threshold (Basner, Mollicone & Dinges, 2011). Differences: it runs in a web browser on the visitor's own device instead of a dedicated test device; input is a touchscreen, mouse or keyboard with unknown latency; the stimulus is a circle turning green instead of a running millisecond counter; there is one unsupervised run with no control of sleep, time of day or surroundings.

Limitations

Device and browser latency adds a variable delay that also moves responses across the 355 ms lapse threshold, so lapse counts are not comparable across devices or with lab values. Time of day, sleep, caffeine, distraction and practice are not controlled. Because time-outs are recorded as 30,000 ms, a single time-out raises the mean RT strongly. One 3-minute run is noisy. Not a diagnostic test of sleepiness or of any attention disorder.

Languages and review status

Language-independent stimuli, instructions in all site languages.