Cognitive Psychology
A word appears in coloured ink. Name the ink, not the word. Almost everybody can do it, almost nobody can do it as fast when the word disagrees — and the interesting question is what the neutral condition adds to that sentence.
Illustrative browser timing — a demonstration, not a measurement of anybody
By the end you should be able to split a Stroop effect into interference and facilitation against a neutral baseline, and say what moves each of them.
About 25 minutes. Nothing you do here is saved or sent anywhere.
The task requires telling four ink colours apart. If colour discrimination is difficult for you, a two-colour version (blue and yellow, which differ in brightness as well as hue) is available in the settings. The worked example reaches every conclusion on this page without responding to a single colour.
Nothing flashes. One word is on screen at a time and it stays until you answer — no trial is timed out. Nothing animates, so there is nothing for a reduced-motion setting to switch off.
Three kinds of trial. On a congruent trial the word names the ink it is printed in. On an incongruent trial it names a different colour. On a neutral trial the word is not a colour word at all — an ordinary word such as chair — printed in one of the same inks. Your job is always the same: name the ink.
Press the button for the ink colour, or use the number keys shown on the buttons. The word stays on screen until you answer, so nothing is lost by thinking — but the measure is how long you took, so answer as soon as you are sure.
Six untimed practice trials with feedback, then a scored block with no feedback until it ends. Results are withheld during the block on purpose: a running mean would change how you answer the next trial.
Status Not started.
Answer the question above to unlock the experiment.
| Condition | Trials | Correct | Mean RT | Median RT | Spread (SD) | Versus neutral |
|---|
| Block | Source | Colours | Mix | Scored trials | Interference | Facilitation |
|---|
The Stroop effect is not a fixed quantity that a person carries around. It depends on what the task asks for and on what the rest of the trials look like. For each change below, predict what happens to the difference between incongruent and congruent trials.
Compare incongruent with congruent and you get one number. That number hides a question: did the mismatching word slow you down, or did the matching word speed you up, or both? Only a baseline can tell you. Interference is incongruent minus neutral; facilitation is neutral minus congruent. In most published work interference is much the larger of the two, and that asymmetry is itself something an account has to explain.
"Reading is automatic" is the usual one-line summary, and it is doing less work than it looks. It has to mean something like: a skilled reader's response to the word is produced quickly and without being asked for, and it competes with the response the task actually wants. Notice that the explanation is about competition between responses, not about the word being impossible to ignore — the ink is named correctly on almost every trial. What is costly is resolving the conflict, not preventing the reading.
If most trials are incongruent, the effect shrinks: the system adjusts how much weight it gives to the word. If the instruction is reversed — read the word, ignore the ink — the interference from the ink is small or absent, which is the asymmetry the automaticity account was invented to handle. An effect that moves this much with context is not a property of a person.
The modern reading is in terms of conflict monitoring and control: a conflict is detected, control is adjusted, and the adjustment shows up on the next trial and across the block. That account is productive, and it is not the only one. Accounts based on the relative speed of the two processes, on translating a colour into a word, and on learning the statistics of the stimulus set all survive the basic finding. The demonstration on this page cannot choose between them.