Choice Reaction Time Test
A simple reaction test asks one question: did it change? This one asks a harder one: which one changed? Four pads, one lights up at random, and you have to hit the correct one. That extra decision costs you time. That’s Hick’s law, and you can feel it here.
- 300–500 ms
- Typical four-choice reaction
- 4 pads
- One lights up at random
- Wrong pad
- Counts as an error, not a hit
Hit the pad that lights up
Four pads. One turns blue at random, so tap it fast. Ten rounds.
The cost of choosing
Here is the thing most reaction tests hide: in real life you almost never just react. You react and choose. A keeper guesses a corner. A driver picks the brake over the horn. A player picks which enemy to track. That choosing happens in the slowest, most flexible part of the reaction (the decision step in the middle), and it is exactly what this test measures.
Watch what happens to your number. A simple reaction might be 250 ms; with four pads to pick from, you will likely land north of 400. That gap is not you getting worse at reacting. It is the price of the decision, and Hick's law says it grows predictably: every time you double the number of options, you pay roughly the same extra slice of time. Two pads, four pads, eight pads: each step costs about the same, which is why a cluttered choice punishes you more than a simple one.
Working the numbers out yourself
Hick's law is usually written as a sum you can actually do: RT = a + b × log2(n + 1), where n is how many options you are choosing between. The a is everything that is not the decision (seeing the stimulus, moving your hand). The b is what one extra bit of decision costs you.
The constants below, a = 200 ms and b = 150 ms per bit, come from Card, Moran and Newell's Model Human Processor (1983). They are round teaching numbers rather than measurements of you, so read the shape of the column, not the exact values.
One thing worth knowing if you compare sources: Hick's original 1952 form uses log2(n + 1), while plenty of textbooks simplify it to log2(n). The extra one stands for the case where no signal arrives at all. The two give slightly different absolute numbers, 548 ms against 500 ms for four options here, but identical behaviour: our simple vs choice explainer works through the simplified version if you want to see both.
| Options (n) | log2(n + 1) | The sum | Predicted RT |
|---|---|---|---|
| 1 (simple) | 1.00 bits | 200 + 150 × 1.00 | 350 ms |
| 2 | 1.58 bits | 200 + 150 × 1.58 | 438 ms |
| 4 (this test) | 2.32 bits | 200 + 150 × 2.32 | 548 ms |
| 8 | 3.17 bits | 200 + 150 × 3.17 | 675 ms |
Look at the jumps: 1 to 2 costs 88 ms, 2 to 4 costs 110 ms, 4 to 8 costs 127 ms. Doubling the options adds a similar slice each time rather than doubling your time, which is the whole point of the logarithm. Going from four choices to eight is nowhere near twice as slow.
Two honest caveats. Your own a and b will not be these numbers, so do not treat 548 ms as a target. And the law describes unpractised choices best: with heavy practice, or when one option is far more likely than the others, the slope flattens and can nearly vanish. Proctor and Schneider's 2018 review in the Quarterly Journal of Experimental Psychology walks through where it holds and where it breaks.
How to read your result
- Median, not best. We take the median of your correct hits so one lucky pad does not flatter you.
- Accuracy counts. Hitting the wrong pad fast is not skill. The test tracks your errors separately.
- Compare it. Run the simple test and subtract: the difference is your decision time.