{{ example ? 'Example' : 'Test' }}{{ snapshot ? snapshot.trials : trials }} trials ComparisonCongruent / incongruent

Condition	Mean (ms)	Median (ms)	Correct	Wrong	Timeout	Anticipation	Interrupted	Accuracy (%)	Copy
{{ row.condition }}{{ example ? ' · example' : '' }}	{{ row.correct ? ms(row.mean_ms) : '—' }}	{{ row.correct ? ms(row.median_ms) : '—' }}	{{ row.correct }}	{{ row.wrong }}	{{ row.timeout }}	{{ row.anticipation }}	{{ row.interrupted }}	{{ pct(row) }}

{{ chartState === 'empty' ? 'No included correct times to plot.' : 'The chart could not load. The trial ledger retains the response times.' }}

Trial	Block	Condition	Word → ink	Response	Outcome	Time (ms)	Input	Copy
{{ row.trial }}	{{ row.phase }}	{{ row.condition }}	{{ row.word }} → {{ row.ink }}	{{ row.response }}	{{ row.status }}{{ row.phase === 'practice' ? ' · unscored' : '' }}	{{ row.rt_ms === null ? '—' : ms(row.rt_ms) }}	{{ row.method }}

Similar tools

Reaction Time Test Test your visual reaction time with simple or choice responses and compare repeated sessions with clear reasons for excluded attempts.
Sequence Memory Test Practise visual sequence memory with forward or backward tile recall and compare your own results across sessions saved in your browser.
CPS Test Measure clicks per second in a timed browser trial, compare pace and consistency, and filter near-duplicate presses when checking switch bounce.
Typing Speed Test Test your typing speed and accuracy against an employer's requirements, with word or character error counts calculated in your browser.

When RED appears in blue ink, reading the word and identifying its ink lead to different answers. A colour-word task asks for the ink colour, so the answer is blue. This trial is incongruent because word and ink disagree; BLUE in blue ink is congruent. The Stroop effect is the interference that can arise when the word's meaning competes with the required colour response.

Congruent BLUE in blue ink and incongruent RED in blue ink both require the answer blue.

For a fluent reader, a familiar word can activate its meaning even when reading is irrelevant. Selecting the ink colour therefore involves keeping the intended response in focus while the word suggests an answer of its own. Psychology demonstrations use this conflict to explore selective attention, the ability to respond to relevant information while ignoring a distraction within the same item.

The useful comparison concerns both speed and accuracy. Longer responses to conflicting words are consistent with interference, but fast errors are not successful colour identification. Slowing down can preserve accuracy, while rushing can increase mistakes. When averages include only correct answers, errors also change which attempts contribute to the comparison.

A response time includes seeing the colour, choosing an answer and making a response. Familiarity with the words and response method, practice, and the ability to distinguish the inks all affect that process. Keeping those conditions similar makes repeated attempts easier to interpret, although a short session can still vary substantially.

Stroop's 1935 ink-naming experiment used spoken responses to printed lists and compared conflicting colour words with solid colour patches. Modern manual variants can compare matching and conflicting words using keys or buttons. These variants explore the same interference principle, but their different responses and comparison conditions produce different measurements.

A short browser demonstration supports exploration of colour-word interference. It does not provide a diagnosis, a clinical attention score or an intelligence ranking.

How to Use This Tool:

Use one response method throughout the session. The default setup gives eight practice attempts before a separately started test of 48 scored trials.

Choose 48 or 96 in Scored trials before starting.
In Advanced, leave Response timeout override blank for 3000 ms, or enter a value from 1500 to 5000 ms. If Check setup appears, select a supported trial count and clear or correct the timeout.
Select Start practice. Wait for the fixation cross to become a word, then answer its ink colour using R for red, G for green, B for blue or Y for yellow. The labelled buttons also accept mouse or touch responses. Practice feedback helps you check the mapping.
Select Start test when Practice complete appears. Practice never enters the scored comparison, and the test waits for this deliberate start. The chosen trial count and timeout stay fixed for the attempt.
Choose Resume after an interruption. Hiding the tab or losing window focus pauses a running session and voids any unfinished active trial; resuming continues with the next planned trial. Choose Stop to end the attempt and retain its partial results.
Inspect Condition comparison when the test finishes. Check the correct, error and exclusion counts before interpreting the time difference. Repeat with practice begins a fresh attempt and replaces the previous observations.

Interpreting Results:

The headline difference is the incongruent mean minus the congruent mean, using included correct responses. A positive value means conflicting words took longer on average. A negative value means they were faster in that session. A value close to zero means the two averages were similar. None of these outcomes is a pass, a failure or a clinical category.

More correct trials needed means at least one condition has fewer than four included correct responses. Individual condition means and medians can appear before that requirement is met. A dash means the relevant statistic is unavailable, not that a response took zero milliseconds or accuracy was zero.

Read Correct, Wrong, Timeout, Anticipation and Interrupted together. Even 100% accuracy can coexist with excluded attempts. The four-response minimum permits a comparison to be calculated; it does not establish that the result is stable. If few correct responses remain, repeat under steadier conditions before drawing conclusions from the gap.

Response spread shows individual included correct times in trial order. Compare it with the mean and median to see whether a few slow responses influence the average. Practice and excluded attempts leave gaps in the plotted trial numbers and remain visible in Trial ledger. If the chart cannot load, the ledger still retains the recorded times.

Results shown while paused, or under Partial session after stopping, describe only the attempts recorded so far. A numerical difference can appear once its minimum counts are met even though the planned session is unfinished.

Technical Details:

Congruence describes the relationship between a word and its ink, independently of whether the response is correct. An incongruent word answered with the right ink colour is a correct response. Timing statistics compare those correct responses across conditions, while accuracy records unsuccessful answers and missed deadlines separately.

Formula Core

For each condition, the mean is the sum of its included correct response times divided by their count.

\bar{t} = \frac{\sum_{i = 1}^{n} t_{i}}{n}

Here, t_i is one included correct time in milliseconds and n is the included count for that condition. The median uses the same times, sorted from fastest to slowest. It is the middle value for an odd count or the average of the two middle values for an even count. A slow included response affects the mean more directly than the median.

The comparison subtracts the congruent mean from the incongruent mean and retains its sign.

D = {\bar{t}}_{I} - {\bar{t}}_{C}

I denotes incongruent trials and C congruent trials. Both means and the difference D are in milliseconds. The difference is calculated from the means, not from the medians or paired trial-by-trial differences.

Accuracy is calculated separately for each condition. Correct responses form the numerator; correct responses, wrong responses and timeouts form the denominator.

Accuracy (%) = 100 \times \frac{Correct}{Correct + Wrong + Timeout}

For example, two correct responses, one wrong response and two timeouts give 40% accuracy. Two additional anticipations, one interruption and a practice attempt do not enter that denominator. Timing and accuracy therefore describe different subsets of the attempted trials.

Which attempts count

Apply the following rules in order to each scored attempt. The first matching rule determines its outcome. All practice attempts stay outside the scored timing, accuracy and response spread, regardless of their outcome.

Scored-trial outcome rules in evaluation order and their contribution to timing and accuracy.
Outcome	Rule	Contribution
Interrupted	An active trial is interrupted by loss of focus, a hidden tab or stopping.	Excluded from timing and accuracy.
Anticipation	A response arrives before timing starts or in less than 150 ms.	Excluded from timing and accuracy, even if the ink choice is correct.
Timeout	No timed response is recorded before the deadline, or the recorded time reaches or exceeds it.	Excluded from timing; counted in the accuracy denominator.
Correct	The response matches the ink, with a time of at least 150 ms and below the deadline.	Included in timing, plotted points and both parts of the accuracy fraction.
Wrong	The response names a different ink within the same accepted time interval.	Excluded from timing; counted in the accuracy denominator.

At the default 3000 ms deadline, a correct response at exactly 150 ms qualifies, while one at exactly 3000 ms is a timeout. The 150 ms anticipation rule, the timeout range and the minimum of four correct times per condition are choices for this educational protocol. They are not validated clinical cutoffs.

Every recorded outcome consumes a planned trial. Excluded attempts are not replaced, and resuming does not replay an interrupted trial or add the paused duration to a response time.

Balanced presentation and the comparison baseline

The planned test contains equal numbers of congruent and incongruent trials. With four colours, there are four matching word-and-ink pairs and twelve ordered unequal pairs. Each unequal pair appears equally often.

Planned condition counts and repetitions of each word-and-ink pair.
Scored trials	Congruent	Incongruent	Each matching pair	Each unequal pair
48	24	24	6	2
96	48	48	12	4

Practice contains one congruent and one incongruent attempt per ink colour. Practice and test order are shuffled separately, with the entire selected test shuffled together. Errors and exclusions can leave unequal included counts despite the balanced plan. The eight practice attempts and the 48- or 96-trial choices are also local protocol decisions.

Matching words can help identify the ink, an effect called facilitation. A neutral condition, such as colour patches or unrelated text, is needed to distinguish that benefit from slowing caused by conflicting words. Because this comparison contains only congruent and incongruent colour words, its difference cannot separate those two contributions or isolate a single mental process.

Timing and numerical precision

A fixation cross is scheduled for 500 ms before each word. Responses during that interval are anticipations. Practice outcome feedback is scheduled for 650 ms, while scored trials show Next for a scheduled 250 ms.

Response timing begins around the browser's scheduled presentation of the word and ends when the response is handled. These timestamps do not establish the exact physical onset of light on the screen or the instant of key or touch contact.

Calculations use unrounded recorded times. Displayed times and percentages use one decimal place, and rounding happens after classification and calculation. For example, 2999.999 ms is below a 3000 ms deadline but displays as 3000.0 ms. Check the outcome alongside the displayed time near a boundary. Decimal precision is not a claim of equivalent physical measurement accuracy.

Limitations and Privacy:

The observations describe one session with one device and response setup. This adaptation supplies no clinical norms, percentiles, confidence intervals or calibration of display and input latency.

Keep the device, browser, input method, trial count and timeout the same when comparing attempts. Keyboard, mouse and touch responses involve different movements and delays; subtracting condition means does not guarantee that all timing errors cancel.
The task requires reading the English colour words and distinguishing the displayed inks. Difficulty with either can affect performance independently of colour-word interference.
Practice can change familiarity with the words and response mapping. A faster repeat attempt alone does not demonstrate a lasting change in attention or a training benefit.
A shorter deadline can move slow responses into the timeout count, leaving faster responses in the mean. Compare errors and exclusions as well as the time difference whenever the setup changes.

Responses and scores are processed locally and kept temporarily in the current tab. Reloading clears them. The URL carries only the trial-count and timeout setup, so sharing it does not share observations or reproduce the same shuffled sequence. The chart may require a network download to display, but recorded responses are not sent for scoring.

The trial ledger retains practice and test attempts with their outcomes, available times and input methods. Save results before reloading, starting again or loading the example. Condition and trial tables support CSV and DOCX exports, and individual rows can be copied. The response spread can be saved as an image or as CSV containing its included plotted times. Those saved or copied records remain outside the temporary session.

Worked Examples:

These illustrative values explain the calculation. They are not expected scores or measurements of a person's attention.

The labelled +200 ms example

Choose View worked example in Advanced to replace the current results with eight synthetic responses. The four congruent times are 400, 500, 600 and 700 ms, giving a mean and median of 550 ms. The incongruent times are 600, 700, 800 and 900 ms, giving a mean and median of 750 ms. Subtracting the means produces +200.0 ms, with four correct responses and 100% accuracy in each condition. No measurements are taken, and the results and exports identify the values as an example. This demonstrates the arithmetic, not the reliability or expected size of a human Stroop effect.

A negative difference with all answers correct

Consider congruent times of 600, 800, 1000 and 1200 ms, compared with incongruent times of 500, 700, 900 and 1100 ms. Their means are 900 and 800 ms, so the comparison is −100.0 ms. All eight answers can be correct while the difference is negative. With only four included times per condition, this describes the small sample and does not establish superior attention or the absence of interference. Inspect the response spread and repeat with the same setup before treating the pattern as consistent.

References:

Studies of Interference in Serial Verbal Reactions, J. Ridley Stroop, Journal of Experimental Psychology, 1935.
Stroop task, PsyToolkit, updated 16 July 2026.
The Loci of Stroop Interference and Facilitation Effects With Manual and Vocal Responses, Augustinova, Parris and Ferrand, Frontiers in Psychology, 2019.
The timing mega-study: comparing a range of experiment generators, both lab-based and online, Bridges and colleagues, PeerJ, 2020.

Stroop Effect Test

{{ summaryTitle }}