Attention Span Test

What the test asks

The psychomotor vigilance test, PVT for short, asks one thing: watch a box and press as soon as a counter starts running in it. The counter comes after a random wait, so there is no rhythm to settle into, and the run goes on for minutes. Your fastest press says little. The slow presses, the misses and the presses that come too early show how steady your attention stayed.

This page gives you the two versions whose rules are published in full: the standard 10-minute test and a 3-minute version made for places where 10 minutes is too long. Each run is saved with the local time and, if you enter it, how long you slept, so you can see whether your own results change with the time of day or after a short night.

The rules used here

• Stimulus: a millisecond counter in a box. It starts at 0 and runs until you press. Your reaction time then stays on screen for 1 second.
• Waits: the time from your press to the next counter is random, 2 to 10 seconds in the 10-minute test and 1 to 4 seconds in the 3-minute test. The 1 second with your time is part of that wait.
• False start: a press with no counter on screen, or less than 100 ms after it appeared. The box shows “FS” for 1 second and a new wait begins.
• Lapse: a reaction time of 500 ms or more in the 10-minute test, 355 ms or more in the 3-minute test. A counter left for 30 seconds counts as a lapse of 30,000 ms.
• End: the test ends with your first answer to a counter after the 3 or 10 minutes are up, so a run is a little longer than its nominal length.
• Practice: five counters with the same feedback come first. They are required before your first run of each length and are never counted. The number five is this page's choice.

How the result is computed

Only answers to a counter of 100 ms or more are valid; lapses and the 30-second misses are among them. From these the page computes:

• Response speed: each reaction time in milliseconds is turned into presses per second, 1000 / RT, and these values are averaged. A 250 ms answer gives 4.0, a 500 ms answer 2.0. The average of speeds is less dominated by one very long lapse than the average of times.
• Lapses and false starts: counted as defined above. The 3-minute test also shows how many answers took 500 ms or more, to set beside the 10-minute rule.
• Median reaction time with a 95% interval: the page resamples your valid times 2,000 times and takes the middle 95% of the resampled medians.
• Slowest 10% as response speed and fastest 10% as a time.
• Performance score:

score = 100% − (lapses + false starts) / (valid answers + false starts)

A score of 100% means no lapses and no false starts. For the 3-minute test the score uses the 355 ms lapses, as in the study that introduced that version. The chart under the result shows every answer through the run with the median of each minute, which is where slowing over time shows up.

Things to try

• Do a 3-minute run in the morning and one in the evening for a week, then compare the time-of-day rows of the history.
• Enter your hours of sleep before each run and look at the sleep rows after a few short and a few long nights.
• Run the 10-minute test once and watch the median line in the chart: does it climb towards the end?
• Try one run with a keyboard and one with a touch screen. The device label keeps them apart, and the numbers differ.
• In practice, press a little before the counter on purpose, to see how a false start looks.

What to look for

Basner and Dinges compared the usual PVT outcomes in sleep loss experiments and recommended two of them as the main ones: response speed and the number of lapses. The median and the mean reaction time were less sensitive. So start with those two numbers, then the per-minute chart. Pressing early to look fast does not help: every false start lowers the score just as a lapse does.

One run is a single measurement. A late night, coffee, a noisy room or a phone call before the run can each move it. A pattern over several runs at similar times says more than any one result.

Where the rules come from

The 10-minute rules on this page follow the criteria that Mathias Basner and David Dinges (2011) proposed so that studies would run and score the PVT in the same way. They note that the 10-minute test has proved highly reliable, with test-retest reliability above 0.8 for key outcomes such as lapses. The 3-minute version, PVT-B, is from Basner, Mollicone and Dinges (2011), who checked it against the 10-minute test in 74 healthy adults during total and partial sleep deprivation. They shortened the waits to 1 to 4 seconds and chose the 355 ms lapse threshold because, in comparable conditions, it brought the number of lapses in 3 minutes up to the level of the 10-minute test with 500 ms. The same paper notes that PVT performance does not improve when the test is repeated, which is why comparing your own runs is a fair use of it.

Limits

• Your device adds delay. Screen, input and browser add time to every answer. In a study of online test software, the added reaction-time delay was 71 to 87 ms on average depending on the software, with a spread (SD) of 8 to 28 ms (Anwyl-Irvine et al., 2021). The page cannot remove it, which is why runs are compared only with runs on the same input and screen refresh rate.
• The original is a dedicated device. The published values were measured on purpose-built hardware with a button, not on a phone or a laptop, so no published average is shown next to your result.
• The 355 ms threshold was set for sleep-deprived adults in a laboratory. It is a rule for counting, not a line between normal and abnormal.
• Time of day and sleep travel with other things: meals, caffeine, mood, light and how you sat. The history groups show patterns in your own data, not their causes.
• A hidden page cancels the run. Switching to another tab or app stops it, and nothing is saved, because a vigilance run only counts without a break.
• This is not a medical test. It cannot detect ADHD or a sleep disorder, and it cannot tell you if you are fit to drive or work.

Controls

• Keyboard: Space or Enter answers a counter while a run is on, and starts a run when the box has focus. Esc stops a run. Tab moves between the box, the buttons and the settings.
• Touch: tap the box to answer. While a run is on, the box does not scroll the page.
• Mouse: click the box to answer.
• Buttons: Start the 3- or 10-minute test, Stop, Download CSV (one row per saved run), Clear history (asks first, keeps the settings) and the language link under the history.
• Settings: length, hours of sleep and the practice counters. They are locked while a run is on.

Frequently asked questions

Why is a lapse 355 ms in the short test but 500 ms in the long one?

With waits of only 1 to 4 seconds and 3 minutes on the task, people answer faster and lapse less often than in the 10-minute test. The authors of the short version lowered the threshold so that the number of lapses matched the standard test, and the page keeps both rules apart.

Why are my times slower on my phone?

A touch screen and its browser usually add more delay than a keyboard on a computer, and the screen may refresh at a different rate. The device label of each run (for example “Touch · 60 Hz”) keeps such runs in their own group in the history.

Sources

• Basner M, Dinges DF (2011). Maximizing sensitivity of the psychomotor vigilance test (PVT) to sleep loss. Sleep 34(5): 581–591. The rules of the 10-minute test and the recommended outcomes.
• Basner M, Mollicone D, Dinges DF (2011). Validity and sensitivity of a brief psychomotor vigilance test (PVT-B) to total and partial sleep deprivation. Acta Astronautica 69: 949–959. The 3-minute version and its 355 ms threshold.
• Anwyl-Irvine A, Dalmaijer ES, Hodges N, Evershed JK (2021). Realistic precision and accuracy of online experiment platforms, web browsers, and devices. Behavior Research Methods 53(4): 1407–1425. How much a browser and a device add to a measured reaction time.