Reaction time test average, and why your own time changes
· 7 min read · Tyamka team
The average on a reaction time test is about 273 ms on Human Benchmark and 231 ms in a large lab study. Why they differ and why your time moves.
The average on a reaction time test depends on where it was measured. Human Benchmark, a popular online test, reports a median of about 273 milliseconds on its statistics page, for everyone who has clicked its screen when it turns green. A lab study of 1,469 adults, who pressed a mouse button when a target appeared on a calibrated screen, found a mean of 231 ms. That is 42 ms apart, by our own subtraction, and neither number is a norm you have to meet. If you want your own time next to them, the first level of our reaction time test takes less than a minute.
Where the numbers come from
| Source | Setup | Number |
|---|---|---|
| Human Benchmark, statistics page | Anyone online, on their own computer or phone, clicks when the screen turns green | Median about 273 ms |
| Woods and colleagues, Frontiers in Human Neuroscience, 2015 | 1,469 adults aged 18 to 65 in a lab, a gaming mouse, the hardware delay measured | Mean 231 ms, or 213 ms with the 17.8 ms hardware delay taken off |
| Francis Galton, 19th century, as cited by Woods | Young people, his own apparatus | Below 190 ms |
The Human Benchmark figure rests on a huge number of clicks, but nobody knows what device each click came from, whether the person was paying attention, or how many tries they had already had. It is a median over people's own hardware, and that hardware adds time of its own.
The lab figure comes from Woods and colleagues, who set out to measure simple reaction time as precisely as they could. They first measured their own equipment. The screen showed the target 11 ms late on average and the mouse registered a press 6.8 ms late, 17.8 ms in total, and they reported the mean both with and without that delay. They also point out that Galton's old numbers may be low partly because Galton seems to have recorded each person's shortest of three attempts rather than a typical one.
So the average is not one number. It depends on who was tested, on what equipment, and whether the figure is a median, a mean or someone's best try.
Simple reaction and choice reaction
All three numbers above are for simple reaction. There is one kind of signal and one possible response. In the Woods study the target appeared either left or right of the centre, but the answer was always the same button. As soon as the answer depends on the signal, for example "press left if the light is on the left, right if it is on the right", responses get slower. The link between the number of choices and the extra time is known as Hick's law, after William Hick's work in 1952.
This matters when you compare yourself with a number from the internet. A test with one target and a test with four possible targets are different tasks, and their averages cannot stand in for each other.
Why your time changes from one attempt to the next
Ten tries give ten different times. That is normal, and the lab data show by how much. In the Woods study the trial-to-trial standard deviation within one person averaged 40 ms. As an illustration, with a spread like that, single tries 40 ms either side of your own average are ordinary and say nothing new about you.
What moves the number:
- How long you waited. The pause before the target in that study was between 1 and 1.8 seconds. After the shortest pause, responses were about 15% slower, 28 ms, than after the longest. When the signal comes sooner, you are less ready for it.
- How long the session has lasted. The mean went from 228.1 ms in the first block of 20 trials to 237.2 ms in the last one. The authors read that as fatigue, and the effect was small.
- Guessing. If the wait before the signal is predictable, you can time the press instead of reacting to it. That is why careful tests use a random wait and do not count presses that come too early.
- Attention. A message arriving, a noise in the room, a glance away. We know of no measurement of this for tests like ours, but most people who have done a few rounds know what a slow round with their mind elsewhere feels like.
- The device. Every screen, keyboard, mouse and touchscreen adds a delay between the real event and the time the test records. In one lab study the added time ranged from about 58 to 133 ms on average, depending on the device and browser. The article on reaction time on a phone and a computer goes through it.
No age norms and no health verdicts
We do not give norms by age, and we do not say what a given time means for your health. The Woods study did find differences between age groups, but a lab average for a group is not a yardstick for one person on a phone, and a game is not a medical test. If you are worried about your reactions, ask a doctor, not a website.
How to read your own time
Compare yourself with yourself, on the same device and in similar conditions. A few ways to keep it fair:
- Use the same device and the same browser every time.
- Do several attempts and look at the middle one, the median, not the best. The best try is the one most likely to contain a lucky guess.
- Leave out a try where you were clearly distracted, but decide that before you see the number.
- Compare days, not single taps. One fast result is not a new level.
Say your middle time on your phone stays around 250 ms. That is your number on that phone. It does not need to match 273, 231 or anyone's screenshot.
How our test counts time
In our reaction time test a circle appears after a random pause of one to four seconds, picked to the millisecond, so there is no rhythm to catch. After each correct tap the target shows your time in milliseconds. A tap before the circle, or less than 0.12 s after it appears, is a false start and costs one of your three lives. We treat a tap that fast as a guess. The 120 ms line is our rule, not a measured human limit, and the article on the fastest reaction time explains where lines like this come from.
The clock starts at the first animation frame after the circle is added to the page, just before the pixels change, and stops when the tap arrives. The screen's own delay and the time the touchscreen or mouse takes to report the tap are therefore inside your time. We have not measured how large that part is for our game. In the lab study mentioned above, web pages that started the clock in a similar way recorded times 58 to 133 ms longer than the real interval, depending on the device.
The results screen and your record are in points, not milliseconds. Each correct answer gives 10 points, and a fast tap adds up to 10 more. Levels 1 to 3 have one target, levels 4 to 9 add a choice of two and then four places, and from level 10 a square sometimes appears instead of the circle, and you must not tap it. A time from level 1 and a time from level 8 are different measurements.
The short answer
For a simple reaction to a light, a careful lab study found a mean of 231 ms including a small hardware delay, and a large online test reports a median of about 273 ms on people's own devices. Single tries in one session spread around your own average, in the Woods study by 40 ms on average. How a race start uses the same idea is in the article on the F1 reaction test, and why a tap on a screen is far from braking in traffic is in the article on reaction time and driving.