Your reaction time is a relay race run by your nervous system, and most of the legs are fixed. When an enemy appears, light hits your retina, a signal crawls through several processing stages, a decision forms, and a motor command runs down your arm to your finger. The whole trip takes roughly 200 ms for a simple, expected stimulus — and no amount of grinding makes your neurons conduct faster. This guide walks the actual pipeline, shows where the milliseconds go, and explains the one stage that does train: prediction.
Here is where your ~200 ms actually goes when you react to something you did not predict:
| Stage | Rough time | What happens |
|---|---|---|
| Phototransduction (retina) | ~20–40 ms | Photons hit photoreceptors; a slow chemical cascade converts light to a neural signal. This is the single biggest reason visual reaction lags audio. |
| Retina → visual cortex | ~30–60 ms | Signal travels the optic nerve to V1 at the back of the brain. |
| Perception & decision | ~50–90 ms | The brain recognises "enemy, shoot" and selects a response. The most trainable slice — familiarity shrinks it. |
| Motor command → muscle | ~50–80 ms | Premotor and motor cortex fire; the signal runs down the spinal cord and arm to the finger flexors. |
| Mechanical click | ~5–15 ms | Finger presses, switch actuates, mouse reports the input. |
Add the midpoints and you land near 200 ms — which is exactly the wall most healthy adults hit on a clean reaction time test. The stages with biology baked in (transduction, conduction, motor delay) are essentially uncoachable. Only perception/decision has real give.
When a web test shows you 150 ms, you did not out-run physics. You anticipated. Your brain noticed the rhythm of the test, predicted when green would appear, and launched the motor program early so the click landed near the stimulus. That is not lying — it is the exact same mechanism that makes you "react" to a CS2 peek in what feels like 80 ms. You did not react to the model; you pre-fired the angle you already expected.
This is the core insight for FPS: the way to beat reaction time is to not need it. A player who holds the right angle, pre-aims the head-level pixel, and expects the enemy has collapsed the task from "detect→decide→move→click" down to "confirm→release." That is why crosshair placement and game sense beat raw reflex every time. We expand this in reflex vs reaction time and the practical drills in how to improve reflexes.
Sound becomes a neural signal in the cochlea in about a millisecond; light takes 20–40 ms longer in the retina. That gap is why your auditory reaction time is consistently 30–50 ms faster than visual. In practice it means footsteps, reloads, and ability cues reach your decision layer before a visual would. A player with good positional audio and headphones effectively gets a head start on every fight. If you are serious about reflexes, treat audio as a primary input, not background — see our note on positional audio impact.
Moves it: warming up (cold start is slower), alertness and sleep, caffeine, display latency (a 60Hz panel adds 20–40 ms versus 240Hz), and a relaxed hand position. These are slack — environmental and hardware delays sitting on top of the fixed pipeline. You can quantify your own slack with our reaction-time improvement calculator.
Doesn't move it (much): grinding a reaction-click trainer for hours, "eye exercises," or supplements with reaction claims. The fixed stages stay fixed. The honest gain from pure reaction training, once slack is removed, is a handful of milliseconds — real but tiny next to what anticipation buys you.
The perception-and-decision slice is where thousands of hours actually pay off. Experts do not perceive a chaotic scene and then decide; they pattern-match. A veteran CS2 player sees a shoulder pixel at a known angle and the response is already loaded. Research on expert perception (chess, sport, driving) shows the same thing repeatedly: experts encode situations into familiar chunks and retrieve responses faster, not because their neurons fire quicker but because the decision is half-made before the stimulus arrives. That is trainable, and it is most of your real-world "reaction" in a shooter.
With a finger on a button, the floor is ~180–200 ms; the fastest verified results sit near 120–150 ms. Web tests below ~120 ms are anticipation, not reaction. Auditory is 30–50 ms faster than visual.
Sound becomes a neural signal in the cochlea in ~1 ms; light needs a slower retinal cascade that takes 20–40 ms longer. That is why footstep cues beat seeing the model.
Conduction and motor stages are fixed, so simple reaction time barely improves. The decision/prediction layer trains well, which is why experienced players appear to react far faster.
Yes, ~2–6 ms per decade after the early twenties, accelerating later. Experience and anticipation easily offset this modest loss.