Refresh Rate, Input Lag, and Your Test Score

Part of your reaction time score is not you. Between the moment a browser decides to show you something and the moment your click is recorded, a chain of hardware and software each takes a small slice of time. On a poorly configured setup that chain can add several tens of milliseconds — enough to move you a full grade down a scale that has nothing to do with your nervous system.

This is worth understanding for two reasons. It explains why your score differs between devices, and it tells you which upgrades actually buy milliseconds rather than just feeling faster.

Where the milliseconds go

A click-based reaction test contains a loop with roughly five stages, only one of which is human:

1. The browser decides to change the screen. The stimulus exists in software.
2. The frame is drawn and sent to the display. Here the refresh rate imposes a wait.
3. The display physically changes. Pixel response time, plus any processing the monitor does internally.
4. You see it and react. The only biological stage.
5. Your click travels to the browser. Switch actuation, polling, USB or wireless transport, event dispatch.

Stages 2, 3, and 5 are equipment. They are added to your score without appearing anywhere in it, and no test can subtract them because no test can see them.

Refresh rate: the biggest single lever

A display updates at fixed intervals. At 60 Hz that is once every 16.7 milliseconds; at 144 Hz, every 6.9 ms; at 240 Hz, every 4.2 ms.

When the browser makes the pad turn green, that change waits for the next refresh before it can appear. On average you wait half an interval, and in the worst case a full one. So a 60 Hz display adds somewhere between 0 and 16.7 ms of pure waiting, averaging around 8 ms. A 144 Hz display averages closer to 3.5 ms.

Moving from 60 Hz to 144 Hz therefore removes roughly 5 ms on average — and more importantly reduces the variance, since the worst case shrinks from 16.7 ms to 6.9 ms. Variance is what makes your scores inconsistent between rounds even when your attention is steady.

The gains past 144 Hz are real but small. Going from 144 Hz to 240 Hz buys about 1.4 ms on average. That is not nothing, but it is well inside the range that a single bad night's sleep will erase.

Pixel response time and monitor processing

Refresh rate governs when a new frame is sent. Response time governs how quickly the pixels actually finish changing colour, and monitor processing covers whatever the display does to the signal before showing it.

For a high-contrast stimulus like a pad turning bright green, response time matters less than the specification sheets imply — you perceive the change as soon as it is unambiguous, not when it completes. Internal processing is the more meaningful contributor, and it is the reason "game mode" exists on most displays: it disables image processing that would otherwise sit between the signal and your eyes.

If you are testing on a television rather than a monitor, this is the first thing to check. Televisions frequently apply substantial processing by default.

The input side

Polling rate

A mouse reports its state at a fixed interval. At 125 Hz that is every 8 ms; at 1000 Hz, every 1 ms. Your click waits for the next report, so polling adds an average of half an interval — about 4 ms at 125 Hz, about 0.5 ms at 1000 Hz.

Most modern mice default to 1000 Hz, so this is usually already solved. It is worth checking on older or budget hardware, where 125 Hz still appears.

Debounce

Mechanical switches bounce — they make and break contact several times in the moment of pressing. Firmware waits a short interval to be sure the press is real. This mostly affects rapid repeated clicking rather than single reactions, which is why it matters far more on a CPS test than on a reaction test.

Wireless

Modern low-latency wireless is close enough to wired that the difference is not the thing limiting you. Bluetooth peripherals are a different matter and can add noticeably more. If you are testing on a Bluetooth mouse and care about the number, borrow a wired one and compare.

The software layer nobody mentions

Browser timing itself is not free. Frame scheduling, background tabs competing for the main thread, extensions injecting work into the page, and the operating system's own compositor all add small, variable delays.

Variability is the real problem here rather than the average. A machine under load does not just score worse — it scores inconsistently, which is precisely the signature people mistake for their own attention wavering. If your spread between best and worst round is unusually wide, close everything else before concluding anything about yourself.

Putting a number on it

Adding the plausible contributions for two setups:

The realistic gap between these two setups is comfortably in the tens of milliseconds — comparable to a decade or more of age-related change. That is the honest reason cross-device and cross-site comparison does not work, and why the same person can score 30–50 ms apart on two machines in the same room.

Phones and tablets are a different chain entirely

Touchscreen testing is not simply "the same test on a smaller screen". The input path is fundamentally different: a touchscreen samples for contact at its own rate, then the operating system decides whether the contact qualifies as a tap, then the browser dispatches an event. Several of those stages have no equivalent on a mouse.

Touch sampling rates vary widely between devices, and some phones sample considerably faster than they refresh. Meanwhile mobile browsers historically delayed tap events to distinguish a tap from the beginning of a double-tap gesture — largely resolved on modern pages, but a reminder that the platform is doing interpretive work between your finger and the page.

The practical consequence is that phone and desktop scores are not comparable in either direction, and that a phone score is not "worse" so much as measured through a different instrument. If you test on both, keep two separate histories.

What this means for how you test

1. Fix the device, then forget it. Test on the same machine every time. A consistent setup makes your trend readable even if the absolute number includes equipment latency.
2. Close things. Background load is free variance you do not need.
3. Do not chase hardware for the score. Going 60 Hz → 144 Hz is a genuine improvement worth having for other reasons. Going 144 Hz → 240 Hz to improve a reaction test result is not a good use of money.
4. Compare against yourself. Your equipment is a constant in your own history and a variable in everyone else's.

Measure yours

The ReflexArcade reaction time test reports your five-round average and your spread, which is what makes equipment variance visible. For what actually moves the biological part of the number, see how to improve your reaction time.