Reaction time and the Stroop test: what these tasks actually measure

Two tasks that feel alike and are not

Reaction time and Stroop are often grouped together as quick brain games, and they do share a surface: both are short, both give you a number in milliseconds, and both feel mentally effortful. Underneath they measure different things. One measures how fast you can respond at all. The other measures what it costs you to produce the right answer when an easier wrong answer is already available. Confusing them is the most common reason people misread their own results.

A reaction task starts with a signal

In a simple reaction-time task, you wait for a signal and respond as fast as you can. There is nothing to decide and nothing to remember. What you are measuring is the whole chain from the signal reaching your eye to your finger moving, which includes perception, the decision that something happened, and the motor response itself. It is the cleanest measure in this family precisely because it removes choice.

Choice reaction time adds a decision

Add options and the task changes. If two signals can appear and each has its own response, you are now measuring decision time on top of everything else, and your times lengthen. The more alternatives there are, the longer the average response, which is one of the oldest regularities in experimental psychology. This is why a choice task and a simple task are not comparable, even on the same device on the same day.

What reaction time is not

It is not a measure of intelligence, of skill, or of how sharp you are as a person. It is a measure of a particular chain of events on a particular device at a particular moment. A large part of any score is the hardware and the software: touch screens, refresh rates and input polling all add delay that has nothing to do with you, and that delay differs between phones.

Why reaction times vary

Sleep, attention, interruptions, practice and device timing can influence a result. A single slow response does not identify which factor mattered. Keep the device and conditions comparable, inspect several trials and do not use a phone game to decide whether you are medically fit, safe to drive or ready for a hazardous activity.

The Stroop task, and where it comes from

In a Stroop task you name the colour of the ink a word is printed in, while the word itself spells a different colour. The word RED printed in blue should be answered blue. The task is named after John Ridley Stroop, whose 1935 paper reported that this interference slows people down reliably. It is one of the most reproduced findings in psychology and has been studied for the better part of a century.

Why reading gets in the way

For a literate adult, reading is close to automatic: you do not decide to read a word, it simply happens. Naming a colour is not automatic in the same way. So on an incongruent trial two answers arrive, the automatic one and the correct one, and you must suppress the first to produce the second. That suppression takes measurable time, and the amount of time it takes is what the task is designed to capture.

The Stroop effect, measured

The difference between your times on congruent trials, where the word and the ink match, and incongruent trials, where they conflict, is the Stroop effect. That difference is the number that matters, not your raw speed. MacLeod reviewed half a century of research on this effect in 1991, and the review remains the standard reference for how robust the phenomenon is and how many variations of it exist.

What a Stroop task involves

A Stroop task requires following the colour-response rule while dealing with potentially conflicting word information. Performance depends on several processes, including attention, reading and response selection. It is not a pure measure of one mental faculty and does not diagnose an attention or memory disorder.

How both differ from dual n-back

Dual n-back loads maintenance and updating: you hold a moving window and rewrite it constantly. Reaction time loads speed. Stroop loads inhibition. These are related but separate abilities, and being strong on one predicts surprisingly little about the others. Jaeggi and colleagues examined how n-back performance relates to reasoning measures in 2010, and even there the relationship is more modest and more complicated than popular accounts suggest.

Compare similar conditions

Different devices and apps can use different timing and scoring methods. Compare your own results under similar conditions, record errors as well as response time, and keep results from different tasks separate. More observations can clarify a pattern, but they do not remove device limitations or establish its cause.

Reaction time and intelligence

There is a real, well-documented correlation between reaction-time measures and cognitive test scores at population level, examined in a large study by Deary, Der and Ford in 2001. That correlation says something about groups and almost nothing useful about you. A single person's score on a phone, taken once, tells you about that morning rather than about their mind, and no responsible app should present it otherwise.

What to do about errors

Speed alone rewards guessing, because you can always be faster if you stop caring whether you are right. Any honest use of these tasks reads speed and accuracy together, and treats a fast run with several errors as worse than a slightly slower run that was clean. If a task reports only your time, be sceptical of what the number is worth.

Use results as task feedback

Read these scores as feedback about the task you completed. You can note an unusually slow run and check for interruptions or mistakes, but it is not a readiness test for work, driving or another demanding activity. Use the same device and protocol if you want a personal record of practice.

Where they fit in Dual N-Back

Dual N-Back includes short cognitive games alongside the main exercise, including reaction time, Stroop and visual memory, and reports your results for each. They are there as practice and as a way to see your own performance, not as an assessment. The app states plainly that scores describe performance on these tasks and that it does not measure IQ or promise an increase in intelligence.

Further reading

• Stroop (1935), Studies of interference in serial verbal reactions, Journal of Experimental Psychology
• MacLeod (1991), Half a century of research on the Stroop effect: an integrative review, Psychological Bulletin
• Deary, Der & Ford (2001), Reaction times and intelligence differences: a population-based cohort study, Intelligence
• Jaeggi et al. (2010), The relationship between n-back performance and matrix reasoning, Intelligence