Mental Rotation test online: same figure or mirror?

Two figures made of 10 cubes appear side by side. The right one is turned in the picture. Decide whether it is the same figure, or its mirror image that no turn could ever match.

Start the Mental Rotation test

48 pairs · about 5 min · phone or computer

After the last pair you see how many pairs you told apart and how much each degree of turn cost you, and a free account keeps the result.

Same figure, turned 100 degreesLeft: the figure · right: turned or mirrored

How it works

4 practice pairs with feedback, then 48 timed pairs with none. Each pair stays up for up to 7.5 seconds, so the whole test takes about 5 minutes.

  1. 1 · First

    4 practice pairs

    After each answer the screen says whether you were right, and which kind of pair it was. The practice uses 2 figures the test never shows, and it is not scored.

  2. 2 · Each pair

    Same or mirror

    The left figure is always shown the same way. The right one is either that figure or its mirror image, like a left hand next to a right hand. Half of the pairs are each kind.

  3. 3 · The turn

    0 to 150 degrees

    The right figure is turned in the picture by 0, 50, 100 or 150 degrees, 12 pairs at each angle. A small fixed turn in depth means even the 0 degree pairs are not copies.

  4. 4 · Your answer

    Tap or press a key

    Tap Same or Mirror, or press the left arrow for same and the right arrow for mirror. With no answer after 7.5 seconds the next pair comes up, and that pair counts as neither right nor wrong.

The figures follow the design Roger Shepard and Jacqueline Metzler introduced in 1971: 10 cubes joined face to face into 4 straight arms, with a right-angle turn between arms. This test builds 8 such figures from a fixed seed, checks that each one differs from its own mirror image and that every cube shows, and draws them in 3 shades of teal. Every attempt sees the same 48 pairs in the same order.

A plus sign holds the centre for half a second between pairs. A small counter above the figures shows which pair you are on, and a thin bar at the top tracks your progress.

What it measures

Mental rotation is the ability to turn a picture of an object in your mind and compare it with another view. It is one of the most studied skills in visuospatial reasoning.

Turning the figure in your mind

  • Building a 3D picture of the figure from a flat drawing.
  • Turning that picture until it lines up with the other figure.
  • Checking the arms one by one to see whether they match or point the opposite way.

Also involved

  • Visual detail: seeing which arm is in front and which runs behind.
  • A decision under time pressure, since a fast guess can be wrong.
  • Response speed, which adds the same amount to every pair and so does not change the rotation cost.

The classic finding is that the time to answer grows in a straight line with the angle between the two figures. Shepard and Metzler (1971) found this for turns in the picture plane and for turns in depth alike, as if people rotated one figure in their mind at a steady speed. The slope of that line, in milliseconds per degree, is the speed of the turn; the starting point holds everything else, such as seeing the figures and pressing the button.

Mirror pairs are left out of the slope. No turn lines up a figure with its mirror image, so a mirror pair has no angle that the mind has to cover.

Spatial skills like this one are often linked with mathematics, and researchers still debate why (Hawes and Ansari, 2020). That is a reason some people with questions about number skills or dyscalculia look at spatial tasks too. A low score here says nothing about dyscalculia on its own.

A test, not a diagnosis.

48 pairs describe how this attempt went. This is not an IQ score, and it cannot diagnose or rule out any condition.

What you get

Your result appears as soon as the last pair is over.

On the result screen

  • How many of the 48 pairs you told apart, and your score: right answers minus wrong ones.
  • Your median time to a correct answer on same pairs at each of the 4 angles, as 4 bars.
  • Your rotation cost in plain words, for example: each extra degree of rotation cost you about 9 ms.
  • A clear note that there is no comparison with other people yet.
  • For clinicians: accuracy on answered pairs, same and mirror accuracy, median and spread of correct times, slope and intercept, plus a CSV file with every pair.

When you keep it

  • A free account saves your numbers, ready to open on any device.
  • Listed under fluid reasoning in your results, next to Matrix Reasoning and Pattern Completion if you take them.
  • Continuing from the hub brings up a suggested next test.

How this version is scored

Each pair is stored with its figure, its angle, whether it was same or mirror, your answer and your time in milliseconds.

Told apart
A pair answered correctly: Same for the turned figure, Mirror for the mirror image.
Score
Right answers minus wrong answers, from -48 to 48. Guessing on every pair lands near 0, so a wrong answer costs as much as a right one gains.
Timed out
A pair with no answer after 7.5 seconds. It counts as neither right nor wrong, and it is left out of accuracy and times.
Accuracy
Right answers as a share of the pairs you answered.
Response time
Measured by the browser clock from the moment the figures appear. The summary uses the median of your correct answers, so one slow pair cannot pull it far.
Rotation cost
A straight line fitted through your median correct time on same pairs at each angle. Its slope is the extra time per degree. It needs at least 2 correct same pairs at 3 of the 4 angles; otherwise the screen says it cannot be measured.

Read speed and accuracy together. A tiny or negative slope with many wrong answers usually means quick guesses, not fast rotation. A steep slope with almost every pair right is a careful, accurate attempt.

Why there is no percentile yet. Published norms come from other versions: the paper test of Vandenberg and Kuse (1978) and its redrawn form (Peters et al., 1995) show a target with 4 options, use turns in depth and score a timed sheet. Online versions have reached very large samples (Peters, Manning and Reimers, 2007), but with their own items. None of those tables fits 48 single pairs of these figures, so this test reports raw numbers until it has data of its own.

Before you start

A steady few minutes gives the cleanest numbers.

For a cleaner attempt

  • Any screen works: on a phone the 2 figures sit side by side at about 150 pixels each.
  • Pick a quiet moment with about 5 minutes to spare.
  • Answer as soon as you are sure; there is no prize for waiting until the time runs out.
  • Stay on the tab until the result appears: leaving loses your progress.

Interpret cautiously when

  • You are tired, unwell or in a hurry.
  • Low vision or a very small screen makes the cubes hard to see.
  • You have done mental rotation tasks recently: practice effects on these tasks are large (Peters et al., 1995).
  • You have taken this test before: the 48 pairs stay the same between attempts.

Questions

What does a mental rotation test measure?

It measures how well and how fast you can turn a picture of an object in your mind to compare it with another view. Researchers count it as a visuospatial reasoning skill. The number of pairs you tell apart shows accuracy, and the rotation cost shows the speed of the turn.

I am slow but accurate. Is that bad?

No. Many people take their time and get almost every pair right, and accuracy matters as much as speed here. A steep rotation cost with few errors simply means you turned the figures carefully. Fast answers with many errors are the pattern to read with caution.

Is this an IQ test?

No. It is one narrow task about spatial reasoning, and it gives no IQ score. Even full intelligence batteries measure many abilities, and a single short test cannot stand in for them.

Can I take the Mental Rotation test on a phone?

Yes. The figures and the 2 answer buttons are sized for a phone held upright. On a computer you can also answer with the left and right arrow keys.

Can I get better at mental rotation?

Spatial skills respond to practice. A review of 217 training studies found lasting gains that carried over to untrained spatial tasks (Uttal et al., 2013). A retake of this test will also improve partly because the pairs are the same, so compare a retake with care.

Sources

By Cristian Dominguez Rein-Loring, Creative Technology Lead · Updated October 12, 2026