Reacting to one cue is a reflex. Choosing between several — is that the real target or a bait, do I shoot or hold — is a decision, and decisions are the gap between "twitchy" and "sharp." Your reaction time test measures the reflex floor on purpose: one cue, one response, the fewest confounds. This is about the layer on top of it, and about a model that can pull your decision speed apart into the two things it's actually made of.
Reacting vs deciding
Adding options costs time in a lawful way. In 1952, W. E. Hick showed that choice reaction time grows with the logarithm of the number of alternatives — more options, more uncertainty, more milliseconds (Hick, 1952; replicated and extended by Hyman, 1953). If you want the gaming-terms primer on why a duel is a choice task and not a simple one, that's what simple vs choice reaction time is for. Here we go one level deeper: what is actually happening in the milliseconds while you decide?
Two dials behind every decision
The dominant account of fast decisions is the drift-diffusion model: your brain accumulates noisy evidence for each option until the pile crosses a threshold, and then it commits (Ratcliff, 1978). Three numbers describe the whole process:
- Drift rate (v) — how fast good evidence builds up. This is your raw decision horsepower.
- Boundary (a) — how much evidence you demand before committing. This is your caution dial.
- Non-decision time — everything that isn't deciding: encoding the stimulus and physically moving.
Most scores blur ability and caution into one "decision speed" figure. The model separates them — and you can estimate all three from just your average speed, how variable that speed is, and your accuracy, using the EZ-diffusion method (Wagenmakers, van der Maas & Grasman, 2007). No new data required; it runs on the same correct/incorrect-plus-timing trials a decision game already produces.
The speed-accuracy tradeoff, finally made legible
This is the readout that explains something you can feel. Turn your caution dial down — lower the boundary a — and you get faster and sloppier, while your underlying ability v doesn't change at all. That's the whole reason "just go faster" doesn't make you better; it makes you riskier. Drift rate v is the number that actually tracks skill. Watch it climb as you genuinely improve, and watch a shift when you're playing scared or playing loose. For how to tell a real change from noise once you have a number, read reaction time score interpretation.
None of this is "pure brain speed." Your monitor, your mouse, and the browser clock are all still inside every decision time — the same ± confidence band as the reaction test. What the model gives you is decision speed on this task, on this machine, today, split into ability and caution — not a context-free measurement of your nervous system.
How to read it honestly
Read v and a as directions, not lab values: "drift is trending up," "you're playing cautious." They earn trust only with a real run of clean trials — a handful of clicks is noise, and the estimate needs dozens-plus to settle. EZ is deliberately a simplified estimate: it assumes you have no bias between the answers and that your decision process doesn't wobble from trial to trial, which real people do. Browser timing jitter also slightly inflates your reaction-time variability, and that variability feeds straight into the math — one more reason to treat the output as a compass, not a verdict.
Curious what your own drift rate looks like? The decision games in the Arena compute it from your rounds — the lab coat under the hood. A companion readout, signal detection and d′, splits the accuracy side of the same trials into skill and bias.
Sources & context
The drift-diffusion model comes from Ratcliff, 1978; the lightweight estimator used here is Wagenmakers, van der Maas & Grasman, 2007. The Hick–Hyman law on choice reaction time is Hick, 1952 and Hyman, 1953. These are controlled-lab studies, not measurements of you on your hardware — treat them as evidence for how decisions behave, not a fixed number for your setup. For how PulsarMS timestamps stimulus and response and why every result carries a ± band, read how we measure.