Face symmetry test. See how even your two sides are, and whether anyone would notice.

Is there a normal amount of symmetry? There is no set number. Clinical studies measured the two sides of healthy faces (Farkas & Cheung 1981), including faces chosen as pleasing (Peck, Peck & Kataja 1991), and found differences between them.

Why 100% symmetrical faces look odd: Nobody has a 100% symmetrical face. When researchers mirror one half of a photo onto the other to make it perfectly even, people rated the mirrored faces as less natural (Zaidel & Cohen 2005).

How this test works: We measure along the line through your pupils, so a head tipped to one side doesn't count. We draw a centre line through the middle of your face and compare how far each cheekbone and jaw corner sits from it, as a share of your face width. This test compares your two sides as a share of your face width, so it says which side is wider, not by how many millimetres. It also can't undo a head turned to one side, which makes the nearer side look wider, so use a photo where you look straight at the camera.

Where the answers change: the test looks at your cheekbones and your jaw corners separately, each as a share of the width between your cheekbones. A tapped dot can easily land a few pixels off, and that alone can move a pair by several percent, so the test only names a wider side when one pair differs by 11% or more. Under that, it reads as no difference anyone would notice. A pair more than 30% apart, or a dot on the wrong side of the centre line, asks you to check the dots. These lines are this test's own, not a clinical standard.

How steady the answer is: we moved all nine dots at random on the four example faces, 10,000 times each, by about 6 and then about 10 pixels of a 720-pixel-wide photo (roughly 3 and 5 pixels on a phone screen). Measured from their own dots, the example faces differ by 8.3% at the cheekbones and 9.3% at the jaw (face 1), 7.4% and 3.1% (face 2), 10.8% and 9.0% (face 3), and 0.3% and 0.4% (face 4), so all four read “No difference anyone would notice”. With the smaller error, that answer held in 90.8% and 99.8% of runs for faces 2 and 4; with the larger one, in 71.2% and 94.1%. Faces 1 and 3 sit closer to the 11% line, so a dot a few pixels off flips the answer to “One side is a little wider” in 40.5% and 65.3% of runs (56.1% and 71.0% with the larger error). On face 3, hair covers her cheek edge on the right of the photo and her head is turned a touch, so the left of the photo measures wider at the cheekbones. So the photo matters most: look straight at the camera, with both sides of your face in view.