Whole-pattern comparison
Compare deterministic vector patterns under exact, rotation, mirror, or delayed rules and review accuracy, response time, and separate error types.
Compare complete patterns under an exact, rotation, mirror, or delayed rule.
Only an exactly identical pattern counts as a match.
Use the current rule to decide whether the two complete patterns match.
The generator builds complete 3×3 or 4×4 patterns from geometric blocks, line directions, and dots. A seed reproduces the trial; a different trial always changes at least one defined detail.
Results stay in local browser history. They describe the selected rule, timing, display, and input conditions, not general pattern intelligence or a standardized visual-memory result.
Same or Different keeps both patterns visible, so the task is direct comparison. Memory Delay shows pattern A, removes it, waits, and then shows pattern B; the response must rely on the retained representation rather than side-by-side scanning.
Display duration and delay are stored because they materially alter delayed trials. A delayed result should not be mixed into the same trend as a simultaneous result.
Rotation Match treats a 90° or 180° transformation of the same base pattern as a match. Mirror Challenge treats a left–right reflection as a match. In either mode, an additional dot change makes the pair different.
The current rule is visible before every block. This matters because a mirrored pair is deliberately correct in one mode but would be different under exact matching.
A perfectly symmetric pattern can look unchanged after a rotation or reflection, making the transformation label ambiguous. The generator rejects candidates equal to their own rotated or mirrored version before building a trial.
Automated tests replay seeded trials and cover this rejection rule. The goal is a visible, defined comparison—not difficulty created by an accidental label conflict.
A false match means answering Match when the pair is different. A false mismatch means answering Different when the pair meets the current rule. The two errors are reported separately because they reflect opposite decisions.
Median correct response time includes correct, answered trials only. Accuracy and timing remain separate rather than being compressed into a black-box recognition score.
Visual Memory asks for previously highlighted positions, Symbol Search checks targets in a row, and Odd One Out infers an exception in one grid. Pattern Matching compares two complete generated structures under an explicit transformation rule.
Delayed mode adds a memory interval, but that does not turn a short browser task into a general memory assessment.
Screen size, contrast, refresh timing, browser workload, touch precision, and keyboard familiarity can all change a session. Keep grid size, rule, display duration, delay, and response deadline aligned for personal comparisons.
Buttons include visible text and keyboard shortcuts. The vector cells remain sharp at high pixel density and in both site themes, but small displays can still make details harder to inspect.
These questions stay tied to the visible content on this page.
Continue with tools that pair naturally with this page.
Test visual memory by recalling highlighted grid positions, then review accuracy, level, mistakes, and browser-only session history.
Start
Find the unique symbol among similar distractors, adjust grid difficulty, and review completion time, errors, streaks, and categories.
Start
Scan symbol rows, decide whether a target appears, and review correct answers, errors, pace, and response consistency.
Start
Compare alternating scenes, find the changed detail, and review detection time, accuracy, hints used, and puzzle difficulty.
Start