Reaction Time Spacebar Test

Start the set, wait through a random 1.5–5 second delay, then press Space only after the page shows the word GO and changes the cue shape. A press before the signal is a false start and is excluded. After five valid trials the page calculates a summary. Random waits restart after false starts so a fixed rhythm cannot reliably predict the next signal.

The logical signal state is not enough by itself: the user must have a chance to see it. The tool schedules the signal, waits for two requestAnimationFrame callbacks, records performance.now(), then exposes the active cue. The Space key time uses the same monotonic clock. This reduces a common implementation error in which a timer starts before the browser has painted the visible change.

Median is the primary result because sorting five trials and taking the middle value reduces the influence of one unusually slow or fast sample. Mean adds all valid times and divides by their count. Fastest and slowest show extremes, while standard deviation describes spread around the mean. None of these values should be converted into a medical, neurological, driving, hiring, or occupational qualification.

One response is sensitive to attention, anticipation, an accidental blink, browser work, and random timing. Five valid trials provide a small but more useful set without making the tool exhausting. For repeatable comparisons, use the same display, refresh mode, keyboard, connection type, browser, and quiet environment. Compare medians across sets rather than promoting one fastest trial as a definitive human reaction value.

A Space press while the page says WAIT is recorded only as a false start. It never enters timing statistics. If focus changes or the tab becomes hidden while a trial is armed, the pending timer is cancelled and the trial is marked interrupted. You can retry that trial, but the interruption count remains visible. Repeat keydown events are ignored so a held Space key cannot generate a valid reaction.

The result includes display refresh, browser scheduling, operating-system input dispatch, keyboard scan and debounce behavior, USB or wireless transport, and page rendering. It does not isolate pure neural response time. No raw key stream or typed text is stored, and the current implementation keeps the trial set only in page memory. Use the result as a local browser measurement under documented conditions, not as a health or safety decision.

If you compare two sets, avoid switching the display refresh rate, power mode, keyboard connection, or browser between them. Close heavy background work and keep the page visible. A fastest trial can be interesting, but the median and spread better describe the five-trial set. A large spread may justify another set under calmer conditions; it does not identify why variation occurred. False starts remain visible because anticipation is part of interpreting the session, even though those presses are never included in valid reaction statistics.