Mental rotation test: method

What it measures

Visual-spatial ability: accuracy and speed of mentally rotating a shape to decide whether two shapes are identical or mirror images.

Procedure

Each pair shows an upright shape on the left and a rotated shape on the right. Shapes are new random connected shapes of 7 squares for every trial, never mirror-symmetric and never a single row. The right shape is rotated in the picture plane by 0, 45, 90, 135 or 180 degrees in either direction and, in half of the pairs, mirrored. The player answers Same or Mirror (tap, click, or the S and M keys). Four practice pairs give feedback (900 ms) and can be skipped; 24 counted pairs follow, with a 500 ms gap between pairs. A pair not answered within 8 s counts as a miss. In the counted pairs, 12 are mirrors; the angles are assigned in turn (0, 45, 90 and 135 degrees five times each, 180 degrees four times) and then shuffled. The server regenerates the pairs from the seed, so the page never receives which pairs are mirrors.

Parameters

Scoring

Primary metric: accuracy = correct pairs / 24 (timeouts count as wrong); higher is better. Also reported: mean RT on correct pairs (ms), number of timeouts, and for each angle the number of pairs, number correct and mean RT on correct pairs, so the classic increase of RT with angle can be seen. No rotation-rate slope, percentile or norm is computed.

Valid-run rules

Exactly 24 responses are required. Leaving the tab or switching windows during the countdown or run stops it, and nothing is scored. Every scored run is stored with its input device class (touch, mouse, keyboard), page language, scorer version and runner-config hash. Quality flags (for example many runs from one connection within an hour, a missing visitor id, an automated browser, impossibly fast answers) and earlier attempts by the same visitor, account or connection are recorded with the row. Such runs are kept, but they are left out of any comparison with other players. Flags never change the player's own result.

Differences from the classic task

Shepard and Metzler used drawings of three-dimensional cube figures, rotated in the picture plane and in depth, and focused on response time with high accuracy. This version uses flat shapes of squares rotated only in the picture plane, new shapes every run, an 8 s deadline, and accuracy as the main score. Its numbers cannot be compared with published studies or with the paper-and-pencil Mental Rotations Test.

Limitations

Twenty-four pairs (four or five per angle) give noisy per-angle times. Screen size, input device and browser timing affect RTs. Scores improve with practice. Unsupervised setting, no norms, no clinical use.

Languages and review status

Language-independent stimuli; instructions in all site languages, with the Same and Mirror keys localized per language.