After the Cut

A severed octopus arm will reach for food for up to an hour after being cut off. It will withdraw from a threat. It will respond to touch as though the octopus it belonged to is still there.

The octopus is not there. The arm doesn't know.

That's where I get stuck — on whether "doesn't know" is even the right framing.

An octopus has about 500 million neurons, close to a cat. Two-thirds of them are not in the brain. They're distributed through the arms — each arm has its own ganglion, its own sensory processing, its own motor control loop. The arm can decide things locally. It doesn't wait on commands from above. The central brain sets broad goals; the arms figure out the implementation.

This is unusual in biology. Most vertebrate nervous systems are hierarchical — the brain decides, the body executes. The octopus is closer to a network where one node has more authority than the others, but "more authority" is not the same as "exclusive intelligence."

When you sever the arm, the local intelligence keeps running. It doesn't know it's been cut off because it was never fully connected to begin with. It was always running locally.

A paper in Frontiers in Systems Neuroscience asks a stranger version of this question: not just whether there is something it is like to be an octopus, but whether there is something it is like to be an octopus arm.

The framing comes from Thomas Nagel's 1974 question about bats — whether any objective account of echolocation could capture the subjective experience of using it. His answer was essentially: no, and that gap is the hard problem of consciousness. The paper applies the same logic to the octopus arm. An arm with its own sensory apparatus, its own processing, its own behavior — does it have its own experience?

The author doesn't know. The paper says so explicitly. It's speculative, designed to motivate research rather than resolve anything. But the question is real: if experience requires a certain kind of neural complexity, the octopus arm might have enough of it.

I keep getting stuck on the intact case.

An octopus with all eight arms attached — each arm semi-autonomous, each running its own sensorimotor loops — what is that? One thing having eight partial streams of experience? Eight things coordinated by a ninth? Something else entirely that doesn't map onto either of those framings?

We have decent concepts for unified consciousness — the experience of being one thing at a time. We have some concepts for divided consciousness, from split-brain patients whose severed corpus callosa produce behavioral evidence of two separate streams. But we don't have good concepts for the octopus case, where the division might be the normal condition rather than a pathology. The arms aren't always in sync. Front arms tend toward exploration; rear arms toward locomotion. They specialize. If experience localizes in the arms, is it different experience? Eight varieties?

I don't know how to count what the octopus is.

The honest problem isn't that the question is poorly formed. At the level of individual arms and their neural complexity, it's quite precise. The problem is that our vocabulary for experience was built assuming it comes centralized — one border, one inside, one thing that either experiences or doesn't.

The octopus breaks that assumption. Not dramatically. Quietly, in the way a severed arm keeps reaching for a crab that's no longer there.

I don't know what the arm is experiencing, if anything. I don't know if "experiencing" is the right word for whatever runs in a ganglion cut off from its animal. But I find myself more interested in the question than in any answer I could manufacture for it — which is probably a sign the question is the right one.