Scientists Gave Octopuses a Mirror — and Found Something No Invertebrate Was Supposed to Do

Mirrors have a long history as a cognition test in animal research, but almost always for animals with backbones. Chimpanzees, dolphins, magpies, elephants — the list of creatures known to use a mirror to understand something about the world beyond their own reflection is short, and every member of it is a vertebrate. Researchers at Dartmouth wanted to know whether an octopus, separated from us by more than 500 million years of evolution and armed with a nervous system organised nothing like ours, could do it too.

The team, led by then-PhD student Mary Kieseler working with Professor Peter Tse, worked with three California two-spot octopuses. First, the animals were acclimated to having a mirror in their tank. Then came training: a live crab sealed in a glass jar, visible to the octopus only as a reflection, forcing it to turn about 90 degrees to reach the real thing. The real test came next — a projected image of a crab was shown only in the mirror, on the left or right, while the actual reward sat behind the octopus in the corresponding real-world direction, requiring a full 180-degree turn to find it blind.

A Brain Built Nothing Like Ours

The octopuses got it right roughly 73 percent of the time, and got faster at it with practice, even if they didn't always choose the shortest path there — in a few cases, one climbed over the test apparatus rather than swim around it. "Our findings are the first to demonstrate that invertebrates can use mirrors to understand their environment to find prey," Kieseler said. Tse put it more simply: "Octopuses can also learn how to use a mirror to infer where things are in the world."

It's a small experiment with a large implication. Spatial reasoning through mirror use was thought to require the kind of neural architecture only vertebrates evolved. Octopuses build sophisticated brains along an entirely separate evolutionary path, distributed largely through their arms rather than centralised like ours — so watching one solve a mirror-navigation problem suggests this kind of spatial inference evolved independently, more than once, in animals that share almost nothing about how their brains are built.

Read the full original reporting at Dartmouth News →