What Is a Choice Reaction Time Test?
A choice reaction time test adds a small decision to a familiar speed challenge. Instead of responding whenever a single screen changes, you must identify which target is highlighted and select its matching number. The task combines noticing a visual signal with choosing and making the correct response. ReflexBenchmark offers three versions with 2, 4 or 8 possible targets, so you can examine your performance as the number of alternatives changes.
Each version contains 16 scored trials. The result is an adjusted average in milliseconds, not simply the average of your fastest clicks. Wrong answers and missed targets receive stated time penalties. A lower adjusted average is better within the same mode, but accuracy and the individual trial chart help explain how that number was achieved. This is a browser practice test, not a medical or psychological diagnosis.
How to Play the 2-, 4- and 8-Choice Modes
- Choose 2 choices, 4 choices or 8 choices before starting. Only the selected number of response tiles appears.
- Press Start Test and wait. After a variable delay, one numbered tile is highlighted as the target.
- Tap or click that tile, or press its matching number key (1–8). The two-choice version also supports the left and right arrow keys.
- Read the immediate correct, wrong or missed feedback, then wait for the next target.
- Complete all 16 scored trials to unlock the adjusted score, accuracy, chart and share card.
The targets are balanced within each mode: the two-choice test shows each target eight times, the four-choice test shows each four times, and the eight-choice test shows each twice. Their order is shuffled for each attempt. The waiting delay changes, so trying to anticipate the highlight is not a reliable strategy.
A response before the cue counts as a false start and causes that same trial to be retried. The test also retries a response recorded under 80 ms after the cue rather than treating it as a genuine completed trial. A missed cue is different: if no valid response arrives within four seconds, the trial is scored as a miss.
How Your Adjusted Reaction Time Score Is Calculated
For every completed trial, the test records a time or assigns a penalty. A correct answer uses its recorded response time. A wrong answer receives whichever is larger: 2,000 ms or your measured response time plus 750 ms. A missed target counts as 4,000 ms. The game then adds the adjusted values for all 16 trials, divides by 16 and rounds to a whole millisecond.
Worked example of the 16-trial formula
Suppose an imaginary session contains 12 correct responses at 320 ms each, 2 wrong responses measured at 400 ms and 2 misses. The correct responses contribute 3,840 ms. Each wrong answer contributes 2,000 ms (rather than 1,150 ms), adding 4,000 ms. The two misses add 8,000 ms.
(3,840 + 4,000 + 8,000) ÷ 16 = 990 ms adjusted average. Accuracy is 12 ÷ 16 = 75%. These numbers demonstrate the game formula; they are not benchmark averages or a prediction of anyone's performance.
This explains why the displayed headline score can be much higher than your average time on correct trials. The two measures answer different questions: adjusted average includes the cost of mistakes, while Correct RT Avg describes response speed only on trials answered correctly. The system saves the adjusted average as the main score.
Reading Your Accuracy, Trial Chart and Statistics
After the sixteenth trial, the result panel shows an accuracy percentage, the average time for correct answers, and a per-trial chart. Green bars represent correct responses; orange bars show wrong answers with their adjusted penalties; purple bars mark misses at four seconds. Each bar corresponds to one of the 16 actual scored trials, not a decorative illustration.
The expanded analysis also includes median correct reaction time, standard deviation of correct times, the fastest correct response, wrong-choice count, missed-target count and false starts. Median is the middle correct time after sorting, while standard deviation describes the spread of the correct times. If there are no correct responses, the statistics that require correct times are unavailable rather than estimated.
A low adjusted time with high accuracy is a different pattern from a few quick correct responses mixed with penalties. Look at where errors happened, whether correct times were steady, and whether one input method felt more comfortable. The result does not tell you which specific mental step caused a slow response.
Comparing Two, Four and Eight Choices Fairly
Increasing the number of alternatives makes the matching task different. With two tiles, it is easier to locate the only other possible response. With eight tiles, you have more possible targets and potentially different eye and hand movements. These sessions still use 16 trials, but the mode matters when interpreting the score.
In psychology, the Hick–Hyman relationship describes a commonly observed tendency for choice-response time to increase with the amount of response uncertainty. It is not a promise that adding options will cost a fixed number of milliseconds in this browser game. Practice, target layout, stimulus-response mapping and input hardware can alter the pattern. ReflexBenchmark does not calculate a validated personal “decision-speed coefficient.”
For personal progress, compare your current two-choice run with previous two-choice runs, and do the same separately for four and eight choices. Even if your eight-choice score is slower, it would not be accurate to subtract the two-choice score and label the difference as pure brain decision time. The modes differ in display arrangement and movement requirements as well as selection demands.
Why Choice Reaction Time Differs from Simple Reaction Time
The Reaction Time Test asks you to notice a screen change and make one response. Choice reaction asks you to decide which response matches the signal. Because these tasks do not use identical targets and controls, their scores should be treated as results from related but distinct tests. A fast simple reaction score does not guarantee a particular choice reaction score.
The Audio Reaction Time Test changes the cue to a sound instead of a highlighted tile. The Flanker Test introduces distracting neighboring signals, while the Stroop Test asks you to resolve competing information. Each activity emphasizes a different challenge; do not compare their raw numbers as if they had the same scoring units or difficulty.
Practical Tips for More Consistent Results
- Learn the controls first. Know which number corresponds to each response tile. In two-choice mode, try either number keys or arrows, then stick with the more comfortable method when comparing attempts.
- Wait for a visible target. An early tap is recorded as a false start and cannot improve the scored average.
- Balance accuracy and speed. A wrong choice costs at least 2,000 ms, making random fast guessing a poor way to improve your adjusted score.
- Keep device conditions consistent. Avoid comparing a touchscreen session with a keyboard session as though the physical responses were identical.
- Take short breaks. Repeated attempts when distracted or uncomfortable can make your results less consistent.
- Track a few sessions in one mode. Review accuracy and spread as well as the headline adjusted average; a single attempt can be unusual.
Browsers, screens and input devices all introduce their own delays. Refresh rate, touch sampling, keyboard behavior, background activity and how close your finger is to a tile may influence a measured interval. These factors cannot be fully calibrated away by this free online test.
Rankings, Score Saving and Important Limitations
If the score API accepts your completed result, it can save your adjusted score and show an estimated rank when enough comparable recorded attempts exist. The site separates two-, four- and eight-choice modes for ranking. Until the comparison group reaches the minimum requirement, the ranking remains pending. A site-recorded percentile is not an age-normed population percentile and does not prove a particular level of intelligence, driving ability or neurological health.
The server can check that the submitted mode contains 16 trials, that target counts are balanced, and that correct, wrong and missed responses agree with the adjusted score. But a browser-side timestamp cannot prove precisely when you perceived the target or whether an attempt was unaided. The test is useful as informal personal practice, not as a clinical assessment or safety-critical measurement. If you want to understand a persistent health concern, a browser result cannot replace professional evaluation.