Reflex Test

A true reflex never reaches your brain. Touch something hot and your spinal cord fires the withdrawal before the signal has finished travelling upward — that loop runs in roughly 20–40 ms and you have no say in it whatsoever. Nothing you can do with a mouse is a reflex in that sense.

What every “reflex test” on the internet actually measures, including this one, is voluntary reaction time: perceive, decide, respond. That is not a criticism of the term — it is what people mean by reflexes in practice, and it is the part that varies between people and between days. It is also the only part you can train.

The distinction matters because it explains the numbers. If reflexes were spinal you would score around 30 ms. Real scores land near 400 ms because you are doing three things: seeing which way the arrow points, choosing the matching key, and executing. Two of those three are cognition, and cognition is what fatigue eats first.

This is a two-choice task, so it is slower than a simple reaction time test by design. Comparing the two directly is the single most common mistake people make with these numbers.

That progression is not random — it is Hick’s law, one of the most durable findings in psychology: response time rises with the logarithm of the number of choices. Doubling your options costs a roughly constant increment, which is why a champion with four escape buttons is genuinely harder to play well than one with two, independent of mechanical difficulty.

Practiced gamers commonly land under 320 ms here. If your score sits well above the adult range on a machine you know, the useful question is not “are my reflexes bad” but “is this today’s number or my number” — and one run cannot answer that.

Choice reaction time is more state-sensitive than simple reaction time, because the decision step degrades faster than the motor step. That makes this test a sharper daily signal, and a noisier one-shot measurement.

• Sleep loss. Costs you on both halves — the response gets slower and the decision gets sloppier, so errors rise alongside times.
• Stress and frustration. Hit decision-making harder than raw speed. A tilted player’s hands are usually fine; their choices are not.
• Alcohol. Degrades choice reactions at doses where simple reactions still look almost normal.
• Time on task. Both speed and accuracy drift over a long session, and people compensate by trading one for the other without noticing.
• Speed–accuracy tradeoff. The most under-appreciated one. You can always go faster by accepting more wrong keys. A score is only meaningful alongside its error rate, which is why this test reports both.

If your time improved but your errors climbed, you did not get faster — you moved along the tradeoff curve. Real improvement moves both.

Same chain as any timed test: the display has to draw the arrow, and the keyboard has to report your press. A 60 Hz panel adds about 8 ms on average before the stimulus physically exists, versus roughly 2 ms at 240 Hz. Keyboard scan rates add a few more milliseconds, and wireless peripherals on old Bluetooth add more still.

The total equipment gap between a laptop on integrated graphics and a tuned desktop runs somewhere around 20–40 ms — a real chunk of the difference people attribute to talent. It is also a constant for any given setup, which is the useful part: it cancels out entirely when you compare your score to your own history on the same machine, and never cancels out when you compare it to a stranger’s.

You can train the decision, and barely touch the reflex. That is the honest summary of a large literature.

• Practice compresses the decision step. As a stimulus-response mapping becomes automatic, the choice penalty shrinks toward the simple-reaction floor. This is most of what “good reflexes” means in practice, and it is real and trainable.
• Familiarity is the biggest lever in game. A player who has seen a wind-up ten thousand times is not reacting faster — they started moving earlier, because they recognised it sooner. That is pattern recognition wearing reflexes’ clothing.
• Sleep protects what you have. No training regime recovers what a bad night costs.
• Hardware is a one-time free gain. Unglamorous and permanent.

What will not work: raising your raw neural conduction speed. That ceiling is largely fixed. The recoverable performance lives in the gap between your ceiling and today — see the honest guide to increasing your reaction time.

Almost every mechanical moment in the game is a choice reaction, not a simple one. A hook coming at you is not “press button” — it is flash, or sidestep, or walk through and accept it, decided in the same window you are reacting in. That is precisely the shape of this test, and precisely why it is a better mechanical proxy than a pure click-when-green test.

The practical read: a 100 ms swing in choice reaction time is worth roughly one dodge in the close ones, and the close ones are where games are decided. Your ceiling is not the interesting number — the gap between your ceiling and today is, which is what the daily check tracks across this and four other measurements.