Symmetry: special but out of the ordinary

What is it about?

Symmetry has fascinated mankind for centuries. In nature, symmetry is ubiquitous in plants and governs mate selection in many animals. In our art, many celebrated works of art are appreciated for their symmetry, such as the Vitruvian Man by Da Vinci, or our own Taj Mahal. Surely this means symmetry is special in the brain too? To study symmetry, we created symmetric and asymmetric objects by combining the same set of parts in all possible ways, and recorded from single neurons in the monkey visual cortex in a region crucial for object perception. Since symmetry is a property of the whole object but not of its parts, we reasoned that symmetric objects might be special in how their parts interact, compared to asymmetric objects. To our surprise, we found no such thing: neural responses to symmetric and asymmetric objects were equally well explained as a sum of their parts. So if symmetry is not special in neurons, then how does it become special in perception? To answer this we asked whether symmetric objects are more distinctive compared to asymmetric objects. This was indeed the case: Just as mixing very different paints results in similar colors, mixing different parts to create an asymmetric object makes it less distinct. In contrast, mixing identical parts to create a symmetric object maintains its distinctiveness. If symmetry makes an object distinctive, then can this distinctiveness explain how we perceive symmetry? Indeed, in separate experiments, we show that the speed with which humans detect symmetry in an object can be explained by how much that object stands out from other objects. Thus our results show that symmetric objects are not special because their parts combine in some special way, but rather because their parts combine in an entirely ordinary way.

Featured Image

Why is it important?

Symmetry is an extremely salient property, but its neural basis is unknown. Here we show that symmetry become special in perception even though it need not have any special property at the level of single neurons.