You can measure your reaction time on a phone in ten seconds. Getting a number is easy. Knowing what the number means on a day when you slept five hours is the harder part, and it is where most people misread themselves badly.
The short version: sleep loss does not slow everything down slightly. It makes you mostly normal with occasional holes. That distinction matters more than the average, because the holes are what cause crashes, dropped catches and missed cues.
What Researchers Actually Measure
Nearly all of the sleep and reaction research runs on one test: the psychomotor vigilance task, or PVT. It is deliberately boring. A counter appears on screen at random intervals over ten minutes, and you press a button as fast as you can each time. No strategy, no learning curve, nothing to get good at.
That dullness is the design. The PVT has almost no practice effect, so you can run it on the same person daily for two weeks without their score improving simply from familiarity. It exposes sustained attention rather than skill.
The important output is not the mean. It is the count of **lapses**: responses slower than 500 milliseconds. A rested adult produces almost none. A sleep restricted adult produces them in clusters, and each one is a moment where the brain briefly stopped processing the signal at all.
Sleep scientists call these microsleeps when they get long enough. From the inside, a lapse feels like nothing. You do not experience a gap. You experience having been present the whole time, which is precisely the problem.
The Numbers From One Short Night
Two studies define most of what we know.
Dawson and Reid, 1997. Published in Nature, this is the study behind every road safety poster comparing tiredness to alcohol. Participants performed a hand eye coordination task after increasing hours awake, and separately after controlled doses of alcohol. Around 17 to 19 hours without sleep, performance dropped to the level seen at a blood alcohol concentration of roughly 0.05 percent. By about 24 hours awake, it was near 0.10 percent, which is over the legal driving limit almost everywhere.
Van Dongen and colleagues, 2003. This one is more relevant to normal life, because almost nobody pulls all nighters regularly but plenty of people run on six hours. Participants spent two weeks limited to four, six or eight hours in bed. The six hour group accumulated attention deficits steadily across the fortnight until, by the end, their lapse counts resembled people who had gone one to two full nights without sleep.
The result that made the study famous was subjective. Sleepiness ratings in the restricted groups rose for the first few days and then flattened. People stopped feeling worse while continuing to perform worse. There was no internal signal tracking the actual decline.
Belenky and colleagues published a similar dose response study the same year and added an uncomfortable detail about recovery. After a week of restriction, three nights of eight hours in bed did not return the most restricted group to baseline. Sleep debt is not settled by one long Saturday.
Why the Slowdown Is Uneven
If you plot every response from a tired person, the shape is not the rested curve shifted right. It is the rested curve with a long tail attached.
That reflects what is happening physiologically. Sleep pressure builds through adenosine accumulation and does not shut the brain off uniformly. Instead, local groups of neurons drift into sleep like activity while the rest of the cortex stays awake, an effect documented in animal recordings and consistent with human imaging. You remain awake by any external measure and still miss signals in short bursts.
This is why averages mislead. If your last ten taps averaged 260 milliseconds, that number looks fine even if two of them were 700 and the other eight were 230. In a reflex game, the two bad ones end the run. On a motorway, the two bad ones are the whole story.
If you want the baseline figures a rested reaction time should land near, average reaction time by age has the distributions.
Time of Day Matters as Much as Total Sleep
Two other clocks are running underneath the sleep debt.
Sleep inertia. The grogginess right after waking is not a metaphor. Cognitive performance in the first fifteen to thirty minutes after waking is measurably impaired, sometimes for up to an hour, and it is worse when you wake out of deep sleep. Testing your reflexes ten minutes after the alarm produces a number that describes almost nothing about your day.
The circadian trough. Independently of how long you have been awake, alertness dips hardest in the early hours of the morning, roughly between 3 am and 6 am, and again in a milder afternoon dip around 2 pm to 4 pm. This is why a night shift worker who slept eight hours during the day still tests badly at 4 am. The debt is not the only variable.
Practical consequence: comparing your reflex score at 11 pm against yesterday's score at 10 am tells you about the clock, not about progress. Test at a fixed time or the comparison is noise.
What Recovers It, and How Much
Ranked by how much evidence supports them:
- Sleep. Obviously, and there is no substitute. Recovery from short term loss is fast but incomplete after chronic restriction, which is the Belenky finding above.
- A short nap. Ten to twenty minutes improves alertness and reduces lapses, with benefits appearing after the inertia clears. Longer naps risk waking out of deep sleep and dropping you into worse grogginess than you started with.
- Caffeine. Genuinely effective at reducing PVT lapses in sleep restricted people. It masks rather than repairs, and taken late it damages the sleep that would actually fix things. How long caffeine stays in your system covers where the cutoff should sit.
- Bright light. Helps alertness in the moment, particularly during a circadian trough. Useful for night work, not a repair mechanism.
- Movement and cold water. Small, short lived. Fine for a specific twenty minute stretch, useless as a strategy.
Nothing on that list except sleep changes the underlying deficit. Everything else buys a window.
Testing Yourself Honestly
If you are going to track reaction time at all, the protocol matters more than the app:
1. **Fix the time of day.** Same hour, every session. Ideally two to three hours after waking, once inertia has cleared.
2. **Run enough trials.** Five taps tell you nothing. Twenty or more start to show the distribution.
3. **Record the worst trials, not just the best.** Your slowest responses carry the sleep signal. Your fastest one is mostly luck.
4. **Log hours slept next to the score.** The correlation only appears across a couple of weeks of data. One tired Tuesday proves nothing.
5. **Keep the device constant.** Screen refresh rate and touch sampling add tens of milliseconds and differ between phones, which is why phone results are good for tracking yourself and bad for comparing against anyone else. The ruler drop test avoids that hardware layer entirely if you want a check.
Where a Reflex App Fits
Zync is a reflex test and tap to hit game, and the useful thing about a test that takes under a minute is that you will actually run it daily. A tracked series against your own sleep log will show you the pattern the research describes: stable averages, and bad nights turning up as scattered slow trials rather than a uniform slide.
Two limits worth stating plainly. A phone test measures simple visual reaction on a touchscreen, which is not the same as the choice reactions and depth judgment that driving or sport demand, and it will not train those either. And no reflex app can tell you that you are impaired in the moment, because the whole finding here is that the impairment does not announce itself. Your hours slept is the more honest number. If you want the trainable side of the picture, how to improve reaction time covers what practice does and does not change.
The Bottom Line
One short night costs you a few tens of milliseconds on the average tap and adds lapses that were not there before. Two weeks of six hour nights compound into a deficit similar to skipping a full night of sleep, and your own sense of alertness stops tracking it after about three days.
That last part is the reason to care. Every other kind of impairment gives you a warning signal. This one takes the warning signal away first and the performance second.