Mental Rotation Test — Free Spatial Reasoning Check

On every trial, Soundary's mental rotation task builds a new flat shape from 7 to 9 joined squares. Only asymmetric shapes are used — their mirror image can never be rotated back onto the original — so every trial has one right answer. The right-hand figure is turned by 0°, 60°, 120° or 180°, and on half the trials also flipped. You start with 8 practice trials with feedback (one same and one mirror pair per angle), then 48 test trials in random order, 6 for each angle in each condition. Each trial allows 7.5 seconds; no answer counts as an error, and a 0.5-second blank separates trials.

The headline score is your accuracy on the 48 test trials, plus mean reaction time on correct answers. The detailed view separates same-shape and mirror accuracy, lists mean times per angle, and estimates your rotation speed (ms/°) and baseline decision time at 0°. Rotation speed is the slope of a straight line fitted to reaction time against angle, using only correct 'same' trials. This follows Shepard and Metzler (1971), whose linear function described the time to recognize two figures as the same; they also found picture-plane rotation no faster than rotation in depth. With fewer than 4 correct 'same' trials, or all at one angle, no slope is computed.

Baseline time is the straight-line estimate of your reaction time at 0°, where no rotation is needed — the time to see the figures, compare them and press a button. Device delays, such as touchscreen lag or finger versus mouse, add roughly the same amount to every trial, so they raise your mean reaction time and baseline but barely touch rotation speed, which is computed from differences between angles. When comparing a phone result with a computer result, rotation speed is the fairer number. Conversely, if only the baseline rises on the same device, general slowing such as fatigue or hesitation is a likelier cause than rotation itself.

The commentary in the detailed view uses Soundary's own guide bands: a slope of 0.3 ms/° or less is nearly flat, which may mean you compared features rather than rotating; under 2 is fast; 2 to under 5 is the typical range; 5 or more means large angles took relatively long. These bands are reading aids, not population norms. Also compare same-shape and mirror accuracy: if only one is much lower, a habit of leaning on one button when unsure is a likelier explanation than rotation ability. Each angle has 12 trials (6 same, 6 mirror), so the per-angle counts show where you wavered.

Shepard and Metzler (1971) introduced the task with asymmetric 3D figures built from cubes, and Cooper and Shepard (1973) extended it to the picture plane, showing normal and mirror-reversed letters at various angles. Because it uses flat shapes, Soundary's version belongs to this picture-plane family. Vandenberg and Kuse (1978) turned the idea into a paper-and-pencil group test. A meta-analysis of neuroimaging studies by Zacks (2008) found that mental rotation consistently increases activity in the intraparietal sulcus and nearby regions, and that this activity changes as the task is made harder or easier.

Age and practice both move scores. A meta-analysis of 80 samples (Techentin, Voyer & Voyer, 2014) found that older adults performed worse than young adults on spatial tasks, with a larger gap in response time than in accuracy — so comparing your own results, taken on the same device at a similar time of day, means more than comparing with other people. Practice matters too: in Kail and Park (1990), participants who made 3,360 rotation judgments became much faster, but the gain did not generalize to other domains. A higher retest score may reflect familiarity with this task rather than a broader gain in everyday spatial skill.