Stroop Test Online — Free Color-Word Interference Check

Soundary's Stroop test runs 20 trials with 4 color words: red, green, blue and yellow. Word and ink color are drawn independently at random, so on average about 1 in 4 trials is congruent (word and ink match) and the rest are incongruent, with the exact count varying between runs. Each trial lasts exactly 1 second, and tapping the button that matches the ink color within it counts as correct. A wrong tap doesn't end the trial, so you can still fix it, but no correct tap within 1 second is a miss. Answering early doesn't skip ahead, so the trials take about 20 seconds.

The top of the result screen shows your correct answers out of 20 and your average response time. Missed trials enter that average as the full 1 second (1000 ms), so more misses also mean a longer average. The detailed report, unlocked by sharing or signing in, splits congruent and incongruent trials, showing accuracy for each and a mean response time based on correct answers only, and turns the gap between them into the interference effect. The screen describes 0 ms or less as no slowdown, under 100 ms as small, under 250 ms as typical and anything above as large; these are Soundary's own guide bands, not age norms.

The Stroop effect normally shows up as slowing, but with a 1-second limit the responses that suffer the most interference may not finish in time and turn into misses instead. So if your interference effect looks small while your incongruent accuracy is clearly lower than your congruent accuracy, the interference may have shown up as errors rather than milliseconds. Also, a run contains only about 5 congruent trials on average, so the congruent mean rests on a handful of responses. A single interference value can therefore swing a lot, and if either condition has no correct response at all, the screen marks the effect as not measurable.

Stroop's (1935) original experiments had people name the ink colors aloud. Augustinova et al. (2019) reported that Stroop interference is substantially larger when people answer by voice than when they press keys, so a tap-based score like this one can't be compared directly with spoken card versions or other apps. The color words also appear in the language of the page you're on, and words in a less familiar language are read less automatically, which can change the size of the interference. When you compare your runs, keep the page language the same as well.

In the second experiment of Stroop (1935), 100 students took an average of 110 seconds to name the ink colors of 100 color words, 74% longer than the 63 seconds they needed to name the colors of 100 colored squares. In the first experiment, reading the words while ignoring the ink barely slowed anyone down. That asymmetry, where the word disrupts color naming but the color hardly disrupts reading, is the heart of the Stroop effect. MacLeod (1991) later reviewed some 400 studies spanning more than 50 years, distilled 18 reliable findings any theory must explain, and concluded that reading automaticity or speed differences alone fall short.

Responses slow down overall with age, but whether Stroop interference itself grows is still debated. Verhaeghen and De Meersman (1998) meta-analysed 20 studies and concluded that the apparent age increase is largely explained by general slowing, whereas Nicosia et al. (2021) reanalysed trial-level data from 33 studies and reported an age difference beyond that slowing. Either way, this number isn't a score to rank against peers. It is most useful next to your own earlier runs in the same language on the same device, read with a note of whether you were tired or short on sleep that day.