Stroop Test Online
Practice color-word interference with balanced deterministic sessions, transparent timing and no invented percentile.
How the task works
The Stroop color-word task creates a deliberate conflict between a word you can read automatically and a color response you are asked to make. In the simplest case, the word Blue can appear in red ink. The correct response is Red because Classic mode asks for the ink color. A congruent trial is easier to describe: the word Blue appears in blue ink, so word meaning and ink color point to the same answer. The interesting comparison is not whether one isolated trial felt hard, but how a balanced set of congruent and incongruent trials changes accuracy and response time.
This implementation uses six answer colors and deterministic seeded trial orders. A Practice URL therefore recreates the same sequence rather than silently giving a different course every time you reload it. That matters when you want to compare input methods, repeat an interrupted explanation, or verify that a surprising result came from your responses rather than a changed trial bank. Fresh Practice deliberately makes a new seed when you want a different order.
Classic, Mixed, Rule Switch and Endurance
Classic 30 contains exactly fifteen congruent and fifteen incongruent trials. You always answer the ink color. Mixed 40 adds neutral non-color words so not every trial contains a competing color meaning. Rule Switch 36 changes the active rule: an unboxed word asks for ink color, while a boxed word asks for the color-word meaning. That mode is not simply a faster Classic test; it asks you to keep the current rule in mind while the stimulus changes. Endurance 60 is a longer mixed set intended for consistency rather than a one-click personal record.
The modes deliberately remain separate in local bests. A short Classic result is not compared directly with a sixty-trial Endurance run, and Rule Switch is not treated as if it measured the same response rule as Classic. This keeps Progress useful without inventing a universal cognitive score.
Reading the result
Accuracy is the literal proportion of trials answered correctly. Mean reaction time uses correct trials only so a random wrong click does not look like an extremely fast valid response. Congruent mean and incongruent mean are also calculated from correct trials in their own categories. The displayed interference cost is your incongruent mean minus your congruent mean for that session. A positive number means your correct incongruent responses were slower on average during that run. A negative value can happen, especially in a small sample, and should not be treated as a diagnosis or a permanent trait.
Browser timing is useful for repeatable personal practice, but it is not a laboratory instrument. Display refresh, touch versus mouse input, keyboard travel, operating-system scheduling, browser load and even how familiar you are with the six response positions can move the numbers. Compare your own runs under similar conditions instead of treating a single millisecond value as an absolute ranking.
Accuracy before speed
Trying to force the lowest possible reaction time can encourage guessing. A better Practice routine is to keep accuracy high enough that the reaction-time averages still represent genuine rule-following responses. If errors rise sharply, slow down and re-establish the instruction. In Rule Switch, read the box cue before responding. In Classic and Mixed, ignore the word meaning and name the ink color. The local score gives accuracy most of the weight for this reason.
Number keys 1 through 6 mirror the six visible color buttons. Touch and pointer controls use the same answer mapping. If you change from touch to keyboard, treat that as a new condition rather than assuming the reaction-time difference represents a change in attention.
What this browser task does not establish
A Stroop-style task is often discussed in psychology because conflict between word reading and color naming is easy to create and measure. That does not make this website a clinical assessment. The site does not diagnose ADHD, attention disorders, dementia, neurological disease or any other condition. It does not estimate IQ, executive-function age or a population percentile. It also does not claim that repeatedly playing this game causes broad intelligence gains. The honest use is narrower: practice a defined color-word rule, measure your own accuracy and response timing, and observe how your own sessions change under comparable conditions.
Language matters too. A person who reads the displayed color words differently, is still learning English, or has difficulty distinguishing one of the chosen colors may experience the task differently. The result should therefore be interpreted in the context of the task and device rather than as a universal label.
Local-first and deterministic by design
No account, external puzzle corpus or paid API is required. The browser creates each seeded trial stream locally, stores completed personal records locally, and keeps specialist Daily history separate from the main Today’s 6 schedule. The generator controls the number of congruent, incongruent and neutral trials instead of relying on chance to produce a useful balance. Independent release verification regenerates thousands of sessions and checks the answer rule for every trial.
If the page loses visibility during an active run, the session is invalidated rather than recorded. That prevents a background-tab pause from producing a misleading reaction-time result. An interrupted run can be restarted with the same seed if you want the same deterministic course again.
A practical Practice routine
Start with Classic 30 until the instruction feels automatic. Then use Mixed 40 if you want neutral trials among the conflicts. Rule Switch is useful only after you understand both response rules; the box is the rule cue, not decoration. Endurance 60 is better saved for a time when you can complete the full set without leaving the tab. Review recent local sessions by accuracy first and reaction time second. If your average becomes faster while errors also climb, the change is not a clean improvement.
Because every Practice course can be reproduced from its URL seed, you can repeat the exact same sequence to compare two input methods. For normal training, use Fresh Practice so the order changes and you are not simply memorizing a specific sequence.