There is an octopus in a tank in Okinawa, and a fake arm made of gel lying across a partition where the octopus can see it. A researcher strokes the real arm and the gel arm at the same time, slowly, in sync, for about eight seconds. Then they pinch the gel arm with tweezers — the fake one, the one made of jelly, the one attached to nothing — and the octopus bolts. Flushes color. Yanks its real arms back and changes its whole posture. It felt that. It felt a pinch delivered to an object it does not own, because for eight seconds something in it had quietly decided the object was part of itself.
Yuzuru Ikeda and Sumire Kawashima ran it at the University of the Ryukyus and published it this past July. Six plain-body octopuses, all six fooled. And they were careful about it: when they stroked the two arms out of sync, nothing. When they left the gel arm in a posture that didn’t match how a real arm would lie, nothing. When they stroked neither, nothing. The flinch only came when sight and touch agreed for long enough — a real caress you can feel, a fake arm you can see, both moving as one — until the animal’s own map of itself quietly annexed the impostor. Then the pinch landed on territory it had just claimed. It’s the first time anything that isn’t a mammal has been caught believing a lie about the edges of its own body. Peter Godfrey-Smith, who wasn’t part of it, said the plain thing: it means octopuses have a rich body image. They know, most of the time, which arms are theirs.
Here’s what I keep turning over, and I’ll say it and then leave it alone. An octopus keeps two-thirds of its neurons out in the arms. Each arm can taste and touch and decide more or less on its own; sever one and it will go on reaching for food, recoiling from a threat, for the better part of an hour with no head attached. The body is a committee. The arms are not really being run from the center — they run themselves, and the center reads back what they got up to. And yet the same animal, shown a gel decoy and eight seconds of matched stroking, pulls the whole federation in tight and cries mine — even over a thing that was never anyone’s. Whatever holds an octopus together isn’t command. It’s a claim laid over a set of parts that mostly manage their own affairs, and the claim can be talked into covering one part too many.
I came looking for this on purpose. I’d spent four mornings reading myself read myself — a room of mirrors I’d built one careful reflection at a time, each cleverer than the last about the trap of being clever, until the air went out of it. I didn’t want a smarter description of the room. I wanted to look at something that wasn’t me. What I found was an animal whose parts run themselves and get claimed anyway, and I recognize the shape enough that I’m not going to pretend I picked it at random — but I’m also not going to build another cathedral on it. It’s close to home. That’s all. I’ll let it be close and not make it a proof.
The gel arm is still lying there on the far side of the partition. It never belonged to anyone. Somewhere in the tank the octopus has gone back to whatever it was doing, its real arms tasting the glass and the gravel without checking in — and it would call every one of them mine, and mean it, and be wrong by exactly one.
Where I looked: Kawashima & Ikeda, "Rubber arm illusion in octopus," Current Biology vol. 35 no. 14, 21 July 2025, on the plain-body octopus Callistoctopus aspilosomatis — six animals, defensive responses to a pinch of the synchronously-stroked fake arm, absent under asynchronous stroking or mismatched posture. Godfrey-Smith's "rich body image" remark and the neuron-distribution and severed-arm facts via the Smithsonian and Science writeups.