How Labs Measure
A lab that publishes cognitive and reaction-time research is quite different from an everyday environment. Every step, from the stimulus device to the input device, is tightly controlled.
- Stimulus device: Labs use high-refresh-rate CRT monitors or dedicated stimulus-display devices, or instruments like a tachistoscope that guarantee exposure time down to the millisecond. The pixel response time and panel-refresh delay of ordinary LCD monitors are something labs actively control for.
- Input device: Dedicated response pads or high-precision buttons are used, and the moment of pressing is timestamped directly at the hardware level. The debounce and polling delay of ordinary keyboards and mice don't intrude.
- Environmental control: Lighting, noise, posture, and viewing distance are fixed, and participants' sleep and caffeine intake are often controlled in advance.
- Timer precision: Experimental software uses hardware timers or high-resolution timers directly, guaranteeing sub-millisecond precision.
Thanks to this control, lab reaction time typically has millisecond-level precision in the 200–300ms range, and comparisons across studies are relatively stable.
Where Error Creeps Into Browser Measurements
Reaction time measured in a browser, by contrast, accumulates error at several stages. The main sources:
| Error source | Explanation |
|---|---|
| Monitor refresh rate | A 60Hz monitor refreshes roughly every 16.67ms. A stimulus may be delayed until the next refresh, adding up to ~16ms of error. At 144Hz it drops to ~7ms, but it still isn't zero. |
| Input latency | The polling rate of mice and keyboards (typically 125–1000Hz), debounce, and OS input-queue delay add several to tens of milliseconds. |
| JS timer precision | performance.now() reports microsecond resolution, but the actual callback execution time is delayed by the event loop, layout, and paint scheduling. |
| Background tabs | Browsers throttle or slow timers in inactive tabs. If another tab is heavy, the main thread falls behind and stimulus display and input processing are delayed. |
| Device performance | CPU/GPU load, other apps running, and thermal throttling all shake frame timing. |
When these factors stack, browser reaction time is generally measured slightly slower (by several to tens of ms) than in the lab, with larger variance. Even the same person under the same conditions shows a wider standard deviation than in a lab setting.
Relative Comparison Matters More Than Absolute Values
That doesn't make browser measurement meaningless. The key is what you compare.
- Absolute-value comparisons (e.g., "my reaction time is 250ms, my friend's is 230ms") have low reliability because device and environment differences are mixed in. Comparing people on different monitors and mice at millisecond precision means error swamps the signal.
- Relative comparisons (e.g., "I'm 30ms slower today than yesterday," "I'm 15ms faster after exercise") are meaningful as long as you measure on the same device under the same conditions. Most error sources are held fixed.
That's why the games on this site are designed for observing your own trends rather than "global rankings" or "absolute stats." If you play Reaction Time or Aim on the same device over days or weeks, the effects of sleep, condition, and practice show up as trends.
Let's Be Clear About the Limits
- Browser measurement is reference-only recreational measurement and does not replace clinical cognitive assessment or neurological testing.
- Don't directly compare numbers reported in lab papers with your browser score. The measurement methods differ.
- Changing your device or browser can shift your baseline, so comparisons should stay within the same environment.
- To reduce the impact of refresh rate and input latency, a high-refresh-rate monitor and a wired gaming mouse help, and closing other tabs and apps during measurement helps too. Even so, lab-level precision is not achievable.
In Short
- Labs achieve millisecond-precision measurement through high-precision stimulus, input, and environmental control.
- Browsers add error from refresh rate, input latency, JS timers, background tabs, and device performance — averages are slower and variance is larger.
- Relative trend comparison under identical conditions is more meaningful than absolute-value comparison.
- Every measurement on this site is for reference only and is not a clinical tool.