Reaction Time Test Methodology

What this test measures

The test records elapsed browser time between the visual panel entering its green state and activation of the test button. It is a simple visual reaction task: one cue and one response. It is not a clinical psychomotor vigilance test, neurological examination, or certified timing system.

Five-round protocol

1. Activate the test button to begin a round.
2. Wait through a randomized delay between 1.5 and 4.5 seconds.
3. Activate the button after it turns green.
4. An activation during the waiting state is rejected as an early attempt and the round is retried.
5. After five valid attempts, the score card is calculated.

Calculations

Median: the middle value after sorting the five attempts. This is the headline score because it is less sensitive than the mean to one unusually fast or slow attempt.

Average: the arithmetic mean of all five attempts. Best and worst are the minimum and maximum. Spread is worst minus best.

Consistency: a session-only description based on spread: very consistent at 20 ms or less, consistent at 40 ms or less, variable at 70 ms or less, and highly variable above 70 ms. These labels are session-only and do not claim a medical meaning or a first-party percentile.

Reference percentiles and Reaction Age

Reference percentiles are modeled from the distribution reported in Woods et al. (2015), Experiment 1: a calibrated simple visual reaction-time study of 1,469 adults with a mean of 231 ms and a between-person standard deviation of 27 ms. We fit that published distribution as a log-normal curve, which is the standard shape for right-skewed reaction-time data, and derive the share of the reference adult population that would be slower than a given median.

Reaction Age maps a median to the observed average of the closest age group in the same study (18–24 at 217.9 ms through 59–65 at 239.1 ms). It is an index for context and sharing, not a measurement of biological age.

Both features are reference estimates, not first-party measurements: this site does not operate a mandatory server-side user benchmark, and the estimates are not medical labels, diagnoses, or claims about your biology.

Reference feedback

The result card uses the session median to provide a qualitative human reference band anchored to Woods et al. (2015), which observed a 231 ms mean in a calibrated simple visual reaction-time study. The bands are interpretation aids, not medical labels or age-adjusted norms; the reference percentiles shown beside them are modeled estimates described in the next section.

The animal speed lens is deliberately labeled as a fun numerical landmark. A star-nosed mole prey-handling cycle and a fly escape circuit use different senses, movements, anatomy, and experimental protocols from a deliberate browser click. They must not be interpreted as equivalent measurements or a species ranking.

Browser timing

The timer uses performance.now(), which the W3C High Resolution Time specification defines against a monotonic clock. Browsers may reduce exposed resolution for privacy or security. A precise clock does not remove display, input, operating-system, event-dispatch, or browser latency.

Device and protocol limitations

A frame lasts about 16.7 ms at 60 Hz and 6.9 ms at 144 Hz, but frame duration is not identical to guaranteed added latency. Displays, touchscreens, mice, keyboards, Bluetooth connections, power modes, background work, and browsers can all change a measured result. Compare sessions only on the same setup and protocol.

Local history and optional publishing

The most recent score cards are stored in your browser using local storage so you can compare sessions. They leave your device only when you deliberately publish a result anonymously to the public dataset.

Voluntary public results

Separate from the modeled references above, the home page can show a community-results section built only from results that visitors voluntarily publish. Publishing is opt-in: nothing is uploaded until the visitor presses the “Publish my result” button on the result card.

Each published record contains the median, best time, average, spread, protocol, a device label, a timestamp, and a random submission identifier — no name, account, email, or IP address. Rows are inserted through an insert-only, row-level-security policy; an automated job then validates, dedupes, and aggregates them into counts, medians, and distributions over a rolling 90-day window. Because the sample is self-selected and browser-measured, the community section is a transparent web sample and not a calibrated laboratory study; it never feeds the reference percentiles or Reaction Age shown on the result card.

Research sources

• W3C High Resolution Time Level 3
• Woods et al. (2015), Factors influencing the latency of simple reaction time
• Wilkinson and Allison (1989), Age and simple reaction time
• Hudson et al. (2020), Sleep deprivation and vigilant attention
• Catania (2011), The sense of touch in the star-nosed mole
• PLOS Biology study of the Drosophila giant-fiber escape circuit

Corrections

Measurement or citation errors are corrected in the build source and the visible updated date changes only after a substantive revision. Report an issue through the contact page.