Three tests that all report a number in milliseconds can be measuring three completely different abilities. A simple reaction time task measures how fast you detect a signal and move. A choice task measures how fast you decide which move to make. A serial task measures how fast you learn a pattern you may not even notice you are learning. Mixing up their scores is the single most common reason people conclude their reflexes are bad when they are not.
This page explains what each family isolates, why choice tasks are slower in a mathematically predictable way, what the serial version is actually used for, and how the three map onto the rounds of our own reaction time test.
One signal, one response, and you know in advance what both will be. The screen goes green, you tap. Because there is nothing to decide, the score is close to a raw sum of the physical stages: light hits the retina, the signal travels to the visual cortex, a prepared motor command is released, the nerve impulse reaches the muscle, and the finger moves far enough to register.
Typical adult results cluster around 250-273ms on a screen, of which a meaningful chunk is the device rather than the person. This is the task behind nearly every published age norm and percentile table, including the ones on our average by age page - which is exactly why it is the score to use when comparing yourself with anyone else.
Its weakness is that it rewards anticipation. If the wait before the signal is predictable, a participant can start moving early and post a number that is not a reaction at all. Any honest simple task randomises the interval and treats an early response as a fault, not a fast trial.
Now there are two or more possible signals, each with its own correct response - green means tap, blue means hold; left arrow means press left, right arrow means press right. Between detection and movement, the brain must identify which signal appeared and select the matching action. That inserted stage costs most people 80-150ms.
Crucially, choice tasks can be got wrong, which means they produce two numbers instead of one: speed and accuracy. Anyone can be fast if they are allowed to guess. A proper choice test either penalises errors in the score or reports the error count alongside the time, because a 320ms average with four wrong responses is worse performance than a 380ms average with none. This trade-off has a name - the speed-accuracy trade-off - and it means a choice score is only interpretable if you know the error rate that came with it.
The relationship between the number of alternatives and the time to choose is logarithmic, not proportional. Hick's law expresses it as RT = a + b × log₂(n + 1), where n is the number of equally likely alternatives, a is the non-decision time (detection plus movement) and b is how much each extra bit of decision costs that person.
The practical consequence is that decision time grows by a roughly constant amount every time you double the options. Going from 2 choices to 4 costs about the same as going from 4 to 8. The table below illustrates the shape using a = 180ms and b = 100ms - plausible values, but illustrative only, since both constants vary widely between people and setups.
| Alternatives (n) | log₂(n+1) | Predicted RT | Cost vs simple |
|---|---|---|---|
| 1 (simple) | 1.00 | 280 ms | - |
| 2 | 1.58 | 338 ms | +58 ms |
| 4 | 2.32 | 412 ms | +132 ms |
| 8 | 3.17 | 497 ms | +217 ms |
| 16 | 4.09 | 589 ms | +309 ms |
Two caveats keep this honest. The law describes untrained responses to equally likely alternatives; heavy practice, or options that are not equally likely, flatten the curve badly. And it says nothing about accuracy - a participant who speeds up by guessing will beat the prediction while getting things wrong.
A serial reaction time task is a long, uninterrupted stream of choice trials - often several hundred - where the participant responds to whichever of four positions lights up. The trick is that the positions usually follow a repeating sequence hidden in the stream. Responses to the sequenced portion get steadily faster; when an unsequenced block is slipped in, times jump back up.
That jump is the measurement. It shows that the participant learned the pattern, frequently without being able to describe it afterwards, which makes the serial task a standard tool for studying implicit or procedural learning rather than reflex speed. If you came looking for a serial reaction time test to check how quick you are, it is the wrong instrument: it reports how well you absorb structure, and a single short run tells you almost nothing.
| Simple | Choice | Serial | |
|---|---|---|---|
| Signals | One | Two or more | Continuous stream |
| Decision stage | None | Yes | Yes, repeated |
| Typical adult score | 250-273 ms | 340-450 ms | Falls across the session |
| Errors possible | Only early starts | Yes - report them | Yes |
| What it isolates | Detection + movement | Decision speed | Implicit sequence learning |
| Good for comparing to norms | Yes | Only to your own history | No |
| Trial count needed | 5-10 | 10-20 | Hundreds |
Our test runs three rounds precisely so that the differences between them are visible. Round 1 is a pure simple task: a red screen turns green after a random 1.5 to 4 second wait, an early tap is scored as a fault, and five valid trials are averaged. Round 2 is a two-alternative choice task in the go/no-go style: green means tap, blue means hold, and a wrong tap on blue carries a time penalty so that guessing cannot pay. Round 3 adds a moving spatial target, layering visual search and aiming on top of the decision.
Read the gaps, not just the total. A round 1 to round 2 gap under about 60ms suggests your decision stage is efficient; a gap over 150ms suggests decisions, not reflexes, are what slow you down in games, sport or driving. A large round 3 penalty with a small round 2 penalty points at visual tracking and aim rather than either. The conversion table has a chart for translating between the three, and the how to test guide covers running the equivalent drills offline. For teenagers comparing themselves against age norms, use the simple-task figures on the teen age chart, and if you are designing a class study the experiment template shows how to hold task type constant.