Start with the game
Practice is the default. The main surface is the center-arrow task, not a wall of measurement settings.
V0.349 · GAME-FIRST SELECTIVE ATTENTION & INTERFERENCE
Flanker
Answer the direction of the center arrow while ignoring the surrounding arrows. The workspace keeps accuracy, correct-response speed and the observed incongruent-minus-congruent interference cost separate.
Practice
ProgressReady
Accuracy—
Median RT—
RunClean
Answer the center arrow only.
PausedStimulus timing and the response window are frozen until you resume.
Answer the center arrow. Ignore the crowd.
Congruent arrows agree with the center. Incongruent arrows point the other way.
Keyboard: ← / A = Left · → / L = Right
20 scored · congruent + incongruent · 1800 ms window
RUN COMPLETE
CleanPractice result
Accuracy—
Median correct RT—
Interference cost—
Accuracy cost—
Congruent—
Incongruent—
Neutral—
Practice run · no gate
Interference cost uses median RT from usable correct responses. Responses below 150 ms or above 1500 ms are retained as evidence but excluded from the RT medians. Read speed together with accuracy; this is not an IQ or clinical score.
Exact trial review
Finish a V0.349 run to inspect every scored trial.
Recent runs, best Benchmark & Daily streak
Measurement notes & fixed benchmark forms
Benchmark A–E preserve the inherited V0.250 deterministic 80-trial scored streams: 40 congruent and 40 incongruent, with left/right targets balanced inside each condition. V0.349 adds a five-trial guided comprehension gate before scoring.
The gate requires at least four correct responses across the five guided trials. Two or more misses end the run before scoring so obvious rule confusion does not become a benchmark result; this is a product comprehension check, not a diagnostic criterion.
Advanced V0.218 interference evidence & repeatability Lab
V0.218 · INTERFERENCE EVIDENCE & REPEATABILITY LAB
Train the conflict, inspect the evidence
Repeatable fixed blocks, validity flags, weak-condition practice and reproducible custom setups turn Flanker into a local training lab instead of a one-off score.
Separate speed and accuracy
The interference RT gap is shown beside accuracy cost and condition-specific evidence so a fast but error-heavy run is not misread as better control.
Repeat exactly
Daily and fixed A–E Benchmark streams are deterministic. Practice and Sprint can be shared or replayed with their exact seed and setup.
Local-first preview. Run history stays in this browser. Focus loss marks a run interrupted and automatically pauses active timing. Production/Hostinger and Today’s 6 are unchanged.