The reaction time test you can take on this site in 30 seconds has a 160-year history behind it. The basic idea — measure how long it takes a person to respond to a stimulus — hasn't changed much since the 1860s. What's changed is the precision of the measurement, the things we've learned from it, and the accessibility of the test itself.
The beginning: Donders and the chronoscope
Modern reaction time research starts with Franciscus Donders, a Dutch ophthalmologist who in 1865 published a paper describing what is now called the subtraction method. Donders realised that by comparing reaction times across different tasks, he could isolate the time taken by specific mental processes.
His setup was a chronoscope — a mechanical device that used a vibrating tuning fork and a gear train to measure intervals to about 1 millisecond precision. The stimulus was a light or sound. The response was a finger movement that completed an electrical circuit, stopping the clock. Ingenious for 1865, and the results were surprisingly close to modern measurements.
Donders found three key results that still hold up. Simple RT (one stimulus, one response) was the fastest. Choice RT (multiple stimuli, multiple responses) was slower. Go/no-go RT (one stimulus, but you respond only on some trials) was in between. The differences, he argued, reflected the time taken by detection, identification, and response selection — the building blocks of any reaction.
Hick's Law and the cognitive era
The next major advance came in 1952, when William Hick ran a series of experiments varying the number of response options from 2 to 10. He found that reaction time increases linearly with the logarithm of the number of choices. Double the choices, add a fixed amount of time. Double them again, add the same amount again.
This finding — now called Hick's Law — has held up across decades of replication and has practical applications everywhere from interface design (fewer choices = faster responses) to sports training (anticipation reduces effective choice load). It's one of the most robust findings in cognitive psychology.
The 1950s and 60s were the golden age of RT research, partly because the Cold War was funding lots of cognitive psychology for pilot and operator selection. The US Air Force and Navy funded hundreds of studies on RT, choice RT, and complex decision-making. Many of the reference numbers used today trace back to studies from this era.
Miller's magical number
In 1956, George Miller published a paper that became one of the most cited in the history of psychology. The title: 'The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information.' The paper argued that working memory has a capacity of about seven chunks, regardless of what those chunks contain.
The paper wasn't strictly about reaction time, but it shaped how psychologists thought about information processing for decades. It also explains why the number memory test on this site tends to cap most people out around 7 digits — that's the natural ceiling of human working memory, as Miller identified 70 years ago.
The computer era
By the 1970s and 80s, computers had replaced mechanical chronoscopes in research labs. The first computer-based RT tests ran on PDP-11 minicomputers and offered sub-millisecond timing for the first time. The methodology hasn't changed much since — a modern web test using the performance API is essentially the same as a 1985 lab test, just cheaper and more accessible.
What did change was the scale of data collection. A typical lab study in 1985 might test 30 undergraduates over a semester. A modern web test can collect 30,000 attempts in a week. This has allowed much finer-grained analysis of RT distributions across age, geography, and demographic factors.
The web era
Human Benchmark launched in 2007 and was the first major web-based RT test to gain widespread popularity. By the mid-2010s it had collected millions of attempts, and the aggregated data became a reference point for casual RT comparisons worldwide.
The methodology on those sites is essentially identical to what we use here: high-resolution timing via the performance API, simple visual stimulus (typically a colour change), false-start detection, and a five-round averaged format. The numbers should be directly comparable across major web tests, with the caveat that hardware varies between users.
Where RT research is now
Modern RT research has moved beyond simple reaction time into more nuanced territory. Current hot topics include: how RT variability (not just average RT) predicts cognitive decline in aging; how RT changes across the menstrual cycle; how RT is affected by long COVID and other post-viral conditions; and how RT relates to specific genetic variants.
The basic test — click when the screen changes colour — remains useful because it's simple, fast, and well-validated. Every modern study that needs a quick measure of processing speed uses some version of it. The 160-year-old idea is still the standard.
If you want to contribute to the science while you test, just take the test honestly. Use a wired mouse, take it a few times, and don't try to game it. Your data joins millions of other attempts in the collective understanding of how fast humans actually react. Donders would be pleased.