Why Is It Hard to Say the Colour of a Word? Reading Wins

The demonstration is easy to run and hard to resist. A participant is shown colour words printed in coloured ink and asked to name the ink rather than read the word. When the two match, naming is quick. When they conflict, so that the word red appears in blue ink, naming the ink takes noticeably longer and errors rise, and the delay persists no matter how hard the participant tries to ignore the word. The finding was published by John Ridley Stroop in 1935 and has been replicated thousands of times across languages, ages and circumstances, which makes it one of the most robust results in the subject and a standard tool rather than a curiosity.

Several explanations have been offered and they are not mutually exclusive:

• •Reading is far more practised than colour naming and runs faster
• •Word recognition proceeds without being asked, so both responses become available
• •The two responses compete because both are words to be spoken
• •Selecting the weaker response requires suppressing the stronger one, which costs time
• •The conflict must be detected before it can be resolved, adding a further delay
• •Reversing the task and asking for the word rather than the ink produces almost no interference

The task became an instrument because it measures something clinically and experimentally useful. It provides a quantitative index of the ability to select a response against competition, which is treated as a component of executive function and which changes across the lifespan, developing through childhood and declining in later life. It is used in assessing conditions affecting frontal brain regions, and in research on attention disorders, schizophrenia and the effects of fatigue, alcohol and sleep loss. Emotional variants replace colour words with words relevant to a participant's concerns, and the resulting delay is used to measure preoccupation in anxiety, addiction and eating disorders. Brain imaging studies use it constantly because it produces reliable conflict on demand.

The basic design has been adapted extensively and the adaptations show what the original was really measuring. Replacing the words with numbers, so that a participant counts how many digits are shown while the digits themselves are numerals, produces the same interference without any colour at all. Replacing them with pictures and labels does the same. Arrows pointing one way while positioned on the opposite side produce a closely related conflict. A spatial version asks for the location of a word that names a different location. Emotional versions substitute words relevant to a worry or a craving. What all of these share is a task requiring one response while an irrelevant and better-practised route to a different response is available, which is the general structure rather than anything about colour.

The finding is frequently overstated and a few clarifications matter. It does not show that reading is literally automatic in the sense of being uncontrollable, since interference is reduced by practice, by manipulating where attention is directed, and by making the colour easier to name, so the effect is a matter of relative speed rather than an absolute compulsion. It does not measure a single ability, since the versions used in different studies differ enough that they correlate only moderately with each other. It does not appear in people who cannot read, including young children, which is the clearest demonstration that the interference is learned. And individual scores are noisy, which limits how much can be concluded about any one person from a single administration.

Naming the ink of a conflicting colour word is reliably slower than naming matching ink, because reading proceeds without being requested and produces a competing response. The result has been replicated for ninety years and is used to index the ability to select against competition. It does not appear in people who cannot read, which shows the interference is learned rather than built in.