The Stroop Task: How the Classic 1935 Experiment Works

The Stroop task might be the shortest experiment in psychology. Show someone a color word printed in mismatched ink, ask them to name the ink color, and measure how long they take. That pause is the whole experiment. This article covers how the original 1935 study was run, how the modern computerized version works, and what the numbers mean. If you want the phenomenon itself explained, start with What Is the Stroop Effect?.

One caveat before the history. The task has been rebuilt hundreds of times since 1935 — screens instead of printed cards, one word at a time instead of whole sheets, buttons instead of spoken answers. The core never changed: two features of a stimulus compete for one response, and the competition costs time.

How Stroop ran it in 1935

John Ridley Stroop tested students with three kinds of large cards, each holding a grid of 100 items. On one kind of card, color words sat in plain black ink and the job was to read them out. On another, colored squares replaced the words and the job was to name the colors. The third kind is the famous one: color words printed in mismatched ink — GREEN stamped in red, say — and the job was to name the ink color of each item, ignoring what the words spelled.

A stopwatch timed each card from start to finish. No computer, no button box. The whole apparatus was a stack of cards, a stopwatch, and an undergraduate saying colors out loud as fast as they could.

The results were lopsided. Naming ink on the mismatched card took dramatically longer than naming colored squares — commonly reported as half again as long, sometimes more. Reading the words, though, barely slowed down when the ink conflicted. Whichever way you stacked the conflict, reading won.

Congruent, incongruent, and neutral trials

Modern versions kept the logic and shrank the scale: one word at a time on a screen, one response each. That gave researchers trial-level data and a cleaner vocabulary. A trial is congruent when word and ink agree (RED in red ink), incongruent when they conflict (RED in blue ink), and neutral when there is no word to fight with at all — colored squares, or letter strings like XXXX.

Why the three conditions matter

The setup is what makes the experiment honest. Suppose you only measured how fast someone names mismatched ink and stopped there. A slow speaker or a slow starter would look impaired without any interference at all. The neutral card measures plain color-naming speed, the reading card measures plain reading speed — subtract, and what is left is specifically the cost of conflict, not general slowness.

That subtraction is also why the task travels well. It compares each person against themselves, so it works across ages, languages, and clinical groups without recalibrating a thing.

What interference scores look like

In computerized versions with button responses, a typical healthy adult answers congruent trials in roughly 600–700 milliseconds and incongruent trials in roughly 750–900. The interference score — incongruent minus neutral — lands somewhere around 100 to 200 milliseconds, and errors climb at the same time. Mismatched ink doesn't just slow people down; it trips them up.

Your own score moves around more than you would expect. Age moves it, and so do sleep, caffeine, time of day, and how much you have practiced. That is why anyone measuring this seriously tracks a personal baseline instead of comparing against a fixed line.

Variants built on the same recipe

The recipe — two features compete for one response — generalizes surprisingly far. The emotional Stroop swaps color words for emotional ones and checks whether personally meaningful words hold attention longer; it has been used in studies of anxiety and addiction. The numerical Stroop pits a digit's value against how many digits you see: a 3 printed three times against a 5 printed twice. Every variant keeps the subtraction and swaps the ingredients.

Try it yourself

Stroop Clash is our free browser version of the task: 20 trials, one word at a time, four answer buttons. You get accuracy and average response time at the end, both saved locally so you can watch your own baseline move. Play Stroop Clash.

Same disclaimer as in the last article: this is practice and measurement, not a diagnosis. The clinical-grade version lives in a neuropsychologist's office, not a browser tab.

Frequently asked questions