Sustained Attention Test

A Psychomotor Vigilance Task (PVT) — the same style of test sleep researchers use to measure vigilance and fatigue.

Green stimuli appear at random 2–5 second intervals. Click each one as fast as you can, but never click early. 1, 3 or 5 minutes, no sign-up needed.

Duration:
Space key also works · stimuli appear at random 2–5 s intervals

What Is a Sustained Attention Test?

A sustained attention test measures your ability to maintain focus on a simple, repetitive task over a period of time. Unlike a simple reaction time test where each response is a fresh, isolated event, a sustained attention test forces your brain to stay in a state of continuous readiness — watching for an unpredictable signal, moment after moment, with nothing else to do. This "tedious vigilance" is precisely what makes the test so revealing.

The scientific roots go back to World War II, when radar operators began missing rare signals after long watches. N. H. Mackworth measured this decline for the first time in 1948, showing that detection of a rarely appearing signal dropped sharply within the first half hour of a watch (Mackworth, 1948). That experiment — the famous Mackworth Clock — founded the entire field of vigilance research.

Modern sustained attention research shows that the ability to stay vigilant varies enormously between people and is surprisingly fragile: in a 10,000-person study, sustained attention ability and strategy dissociated sharply across the life span (Fortenbaugh et al., 2017), and performance can degrade within minutes when the task is monotonous and the stakes are low.

The Psychomotor Vigilance Task: the Science Behind This Test

This page implements a compact version of the Psychomotor Vigilance Task (PVT), introduced by Dinges and Powell (1985) as a simple portable reaction-time task for studies of fatigue. The PVT works by presenting a stimulus at random, unpredictable intervals — the key ingredient. When the delay between stimuli is random, you cannot rhythmically anticipate; you must hold attention continuously. Every click is then a precise sample of your alertness at that instant.

Clinical versions of the PVT typically run 10 minutes with inter-stimulus intervals of 2–10 seconds (Dorrian, Rogers, & Dinges, 2005). For a browser-based self-test we compress this to 1–5 minutes with 2–5 second intervals — short enough to stay practical, long enough to observe the classic vigilance decrement: lapses and slowing responses in the later minutes.

The PVT is widely considered the gold standard among fatigue-sensitive tests because it is simple, has no learning curve, and its outcome metrics — lapses and slow responses — are extraordinarily sensitive to sleepiness (Dorrian, Rogers, & Dinges, 2005).

Why Lapses Matter More Than Your Average

If you glance at reaction-time results, your eyes go straight to the average. That's a mistake on a sustained attention task. The metric researchers watch first is the lapse — a response slower than 500 ms — because lapses are what change earliest and most reliably when vigilance fails (Basner & Dinges, 2011).

The reason is mathematical. Average reaction time is pulled along by the mass of normal, fast responses, so a handful of very slow responses barely moves it. Lapses are a direct count of the failures themselves. Two results can share an identical mean (286 ms) yet differ dramatically in alertness — one with a clean distribution and zero lapses, another with repeated 600–900 ms lapses that the mean hides. On this page we therefore show median, percentiles, lapses, and a per-minute breakdown, not just the average.

Sleep research makes the same point. In landmark studies, restricting sleep to 4–5 hours per night for a week produced clear, cumulative deficits in PVT lapses — while mood and many other measures recovered or plateaued (Dinges et al., 1997). Total sleep loss studies found the same signature: lapses and slow responses increase sharply, and reaction-time variability grows (Van Dongen et al., 2003). Vigilant attention has been described as the cognitive function most reliably impaired by sleep deprivation (Lim & Dinges, 2008).

In practice, that means: if your average speed looks fine but you accumulate lapses in minutes 4–5, that pattern is worth more attention than any single fast click.

Why We Show the Median (and Why It Differs from the Mean)

Reaction times are famously right-skewed: most responses cluster around a fast peak, while a long tail of slow responses drags the arithmetic mean upward. This distribution shape is a well-documented property of response-time data (Luce, 1986). In any skewed distribution, the median — the midpoint of your results — is a far more stable summary than the mean, because it is immune to the tail.

Concretely: one lapse of 900 ms in a 15-response session raises your mean by ~40 ms but leaves your median untouched. If your median is 271 ms and your mean is 286 ms, the 15 ms gap is a direct measurement of how much your slow responses are dragging you down — the bigger the gap between median and mean, the less stable your attention.

That's also why we report the fastest and slowest 10%: they bracket the whole distribution. The fastest 10% tells you your true ceiling when attention locks on; the slowest 10% tells you how badly it slips at your worst moments.

How to Read Your Results

  • Median RT. Your typical reaction speed when alert. Well-rested healthy adults usually land between ~250 and ~300 ms on PVT-style tasks; below ~240 ms is unusually fast.
  • Mean RT. Always higher than the median when lapses exist. Compare the two — a growing gap signals instability.
  • Fastest / Slowest 10%. Your peak and trough states. A fast 10% below 230 ms with a slow 10% above 350 ms means your attention swings widely.
  • Lapses. Responses over 500 ms — the PVT's primary fatigue marker (Basner & Dinges, 2011). Zero is the goal for a 1-minute session; for 3–5 minute sessions, one or two lapses in the last minute are common even when well rested.
  • False Starts. Clicks before the stimulus. A few are harmless (impatience, anticipation); many indicate you're clicking rhythmically instead of watching — which invalidates the measurement.
  • Response Stability. Consistency across your session, on a 0–100 scale. Below ~60% means your attention is fluctuating a lot.
  • Vigilance Score. A composite of lapses, speed, and consistency. Above 80 is the well-rested range; below 60 is the "go get some sleep" range.

This is a recreational self-assessment, not a clinical instrument. Clinical PVT administration uses calibrated hardware, standardized 10-minute protocols, and normative data (Dorrian, Rogers, & Dinges, 2005). Use your results to notice patterns — never to diagnose.

How to Improve Your Sustained Attention

Sustained attention is trainable, but the biggest lever is not a training app — it's your sleep and arousal state. The research is unusually clear on this:

  1. Sleep first. Chronic partial sleep restriction — even 4–5 hours per night — produces cumulative vigilance deficits across days (Dinges et al., 1997), and total sleep loss degrades vigilant attention faster than almost any other cognitive function (Lim & Dinges, 2008). Retest after a full night of sleep before drawing conclusions about your attention.
  2. Watch your caffeine timing. Caffeine temporarily restores PVT performance and is widely used in fatigue management — but it cannot replace sleep. Time it before your demanding periods, not as an evening band-aid.
  3. Respect your chronotype. Alertness fluctuates across the day. A 3-minute session in your afternoon slump will differ from the same session in the morning. Track your best and worst times of day.
  4. Short breaks beat long grinds. Vigilance decrements build up with time-on-task. In work settings, brief breaks — even 30–60 seconds — measurably restore performance. At a computer, every 20–30 minutes, look away and blink.
  5. Practice the skill. Repeated testing itself produces familiarity with the pacing and improves consistency. Take the 3-minute test daily for a week and watch your lapse count trend down.

Read the full guide to improving reaction time →

Frequently Asked Questions

Why do you count a response over 500 ms as a "lapse"?▾
The 500 ms threshold is the standard definition used in psychomotor vigilance research. It was chosen because responses slower than 500 ms are essentially missing responses — they indicate brief episodes of sleepiness or attention failure, not just slow reflexes. Basner and Dinges validated that lapse counts computed with this threshold are among the most sensitive measures of sleep loss available (Basner & Dinges, 2011).
Why is my average reaction time on this test different from the simple test?▾
Two reasons. First, a sustained attention task contains genuine lapses — 600–900 ms responses — that pull the mean upward; the simple test has almost none. Second, waiting through long random intervals changes your strategy: your brain cannot pre-activate for the moment the stimulus appears, so responses tend to be slower and more variable. Both differences are expected and are part of what the test measures.
I clicked before the screen turned green a few times. Does that matter?▾
Every early click is counted as a false start and the full 2–5 second wait restarts. A few false starts are normal — they show anticipation rather than inattention. Many false starts suggest you are clicking rhythmically instead of watching the screen, which makes the whole session invalid. If your false starts exceed your lapses, retake the test and focus on waiting for the green signal.
I slept badly last night. Will it show in my results?▾
Very likely. Sleep loss reliably degrades vigilant attention, and the effect appears exactly in the metrics this test highlights: more lapses, slower responses, and greater variability — often before the average reaction time changes noticeably (Lim & Dinges, 2008; Van Dongen et al., 2003). If you want a fair baseline, take the test twice: once after a normal night of sleep, once after a poor one, and compare the lapse columns.
Is this the same test used in sleep clinics?▾
The same task family, not the same protocol. Clinical PVT assessment typically uses a 10-minute duration with 2–10 second intervals, calibrated hardware, and normative data for interpretation (Dorrian, Rogers, & Dinges, 2005). Our browser version is shorter (1–5 minutes) and is designed for recreational self-assessment. The core logic — random stimuli, lapse counting, sensitivity to sleepiness — is identical.
Why is there no 10-minute option?▾
Longer sessions detect the vigilance decrement more reliably, but 10 minutes of staring at a screen is a lot to ask of a visitor, and the research-grade protocol needs calibrated equipment to stay meaningful (Dorrian, Rogers, & Dinges, 2005). The 3- and 5-minute versions already show the classic fatigue curve — lapses clustering in the final minutes. If you want a closer approximation of the clinical PVT, take the 5-minute version at the same time of day over several days.
Can I improve my sustained attention with training?▾
Yes, within limits. Repeated practice improves your consistency and familiarity with the task — expect your lapse count to drop over the first several sessions. The larger, better-documented effects come from lifestyle factors: protecting your sleep, managing caffeine, and respecting your natural alertness rhythm. No app can compensate for chronically insufficient sleep; the vigilance literature is unambiguous on this point (Dinges et al., 1997).

References

  • Dinges, D. F., & Powell, J. W. (1985). Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behavior Research Methods, Instruments & Computers, 17, 652–655.
  • Basner, M., & Dinges, D. F. (2011). Maximizing sensitivity of the psychomotor vigilance test (PVT) to sleep loss. Sleep, 34(5), 581–591.
  • Lim, J., & Dinges, D. F. (2008). Sleep deprivation and vigilant attention. Annals of the New York Academy of Sciences, 1129, 305–322.
  • Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, D. F. (2003). The cumulative cost of additional wakefulness: dose-response effects on neurobehavioral functions and sleep physiology. Sleep, 26(2), 117–126.
  • Dinges, D. F., Pack, F., Williams, K., et al. (1997). Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4–5 hours per night. Sleep, 20(4), 267–277.
  • Dorrian, J., Rogers, N. L., & Dinges, D. F. (2005). Psychomotor vigilance performance: Neurocognitive assay sensitive to sleep loss. In Sleep Deprivation: Clinical Issues, Pharmacology, and Sleep Loss Effects (pp. 39–70). Marcel Dekker.
  • Fortenbaugh, F. C., et al. (2017). Sustained attention across the life span in a sample of 10,000: dissociating ability and strategy. Psychological Science, 28(9), 1257–1267.
  • Mackworth, N. H. (1948). The breakdown of vigilance during prolonged visual search. Quarterly Journal of Experimental Psychology, 1(1), 6–21.
  • Robertson, I. H., Manly, T., Andrade, J., Baddeley, B. T., & Yiend, J. (1997). "Oops!": performance correlates of everyday attentional failures in traumatic brain injured and normal subjects. Neuropsychologia, 35(6), 747–758.
  • Luce, R. D. (1986). Response Times: Their Role in Inferring Elementary Mental Organization. Oxford University Press.