Replace the parts one at a time and the ship stays in the harbor. The question is whether anyone is still standing on the deck.
Everyone knows the setup. A ship gets a rotted plank replaced. Then another. Over years, every original plank is gone, swapped out one board at a time for a new one. Is it still the same ship?
People argue about this in bars and philosophy classes because it feels like it should have an answer and never quite does. What I keep coming back to is that we usually stop the thought experiment at the ship. We rarely take the next obvious step, which is to ask what happens when the planks are neurons and the ship is you.
The Setup, Made Personal
Imagine a neuron in your brain gets replaced with an artificial equivalent. Not a crude stand-in, but something that does exactly what the neuron did: same connections, same firing pattern, same response to the same inputs. From the outside, nothing changes. You’d never notice one neuron swapped out of eighty-six billion.
Now do it again. And again. Keep going, one neuron at a time, on some absurdly long timeline, until every biological neuron has been replaced with an artificial one.
At the end of that process, is the person who exists the same person who started it? And if not, where along the way did they stop being that person?
That second question is the one that actually matters, because “no, not the same person” is an easy answer to give and a much harder one to defend once you have to point to the moment it happened.
The Easy Answer Has a Problem
The easy answer comes from functionalism, more or less the working assumption in a lot of cognitive science: what makes a mind is the pattern of information processing, not the specific material doing the processing. If the artificial neuron does the same job as the biological one, functionally there’s no difference. Swap enough of them and you haven’t destroyed a person, you’ve just changed their hardware.
I find this compelling in the way I find most clean, well-argued positions compelling. It also doesn’t sit right.
Philosopher David Chalmers ran a version of this thought experiment specifically to test it, asking what would happen if you swapped neurons for silicon chips that preserved the same functional organization. He described two possible failure modes. One is “fading qualia,” where consciousness gradually dims as more neurons get replaced, until at some point the lights are on with nobody home. The other is “dancing qualia,” where your conscious experience stays vivid but starts disconnecting from what’s actually happening, so you might be having a wildly different internal experience while behaving exactly as before, with no way for you or anyone else to notice.
Chalmers argued both of those options are absurd, and used that absurdity to defend functionalism. I think he’s right that both options sound strange. I’m less sure that “sounds strange” settles anything, especially when the alternative just moves the strangeness somewhere else instead of resolving it.
The IT Version of the Problem
I spend my working life around systems that get migrated one piece at a time, and there’s a version of this problem that shows up constantly in that world, minus the philosophical stakes.
You’ve got an old system running in production. You want to move it to new infrastructure without taking it down. So you don’t do a cutover. You migrate piece by piece: move a service here, redirect a dependency there, swap a database layer while keeping the old one live as a fallback until you’re sure the new one behaves identically. Eventually, months later, every original component is gone. The system never went offline. Users never noticed a thing.
Is it the same system?
In practice, nobody in IT loses sleep over that question, because we don’t think identity lives in the hardware or even in the specific code. We think of the system as the thing that keeps doing its job, uninterrupted, regardless of what it’s running on underneath. A system is a continuity of function, not a continuity of parts. That’s functionalism again, just wearing a different badge, and it works fine for a piece of infrastructure that doesn’t have an inside.
The problem is that a brain, unlike a server, is the kind of thing that might have an inside. There’s something it’s like to be you having this thought right now. There’s nothing it’s like to be a load-balancer. If personal identity is just “the system kept doing its job without interruption,” the migration analogy works cleanly. If personal identity also depends on whatever produces the fact that experience feels like something from the inside, the analogy stops covering the part that actually worries people.
Where the Threshold Should Be, If There Is One
Say you’re looking for the exact moment the original person stops existing. A few obvious candidates get proposed, and none of them hold up well once you look at them directly.
It’s not the first neuron. Nobody seriously thinks replacing one out of eighty-six billion neurons ends a person, and if it did, the threshold would have to be absurdly fragile.
It’s not the last neuron either, at least not for any principled reason. Picking the final swap as the moment of death assumes everything up to that point was fine and the very last piece tips it over, which requires a hidden claim that the second-to-last state was still fully you and the last one wasn’t. Nobody has ever explained what property that final neuron uniquely carries.
It’s not some percentage either. Fifty percent replaced sounds like a natural halfway point, but there’s no argument for why the halfway point in cell count should correspond to the halfway point in personhood, whatever that would even mean. Identity doesn’t obviously come in percentages you can average.
The honest position might be that there is no threshold to find, because the question assumes identity is the kind of thing that has a sharp edge somewhere in the process. Most things we care about don’t work that way. Nobody can tell you the exact day a child becomes an adult, the exact word where a sentence stops being a fragment, or the exact grain of sand where a pile becomes a heap. These are the sorites paradoxes, and they show up whenever a gradual change gets asked to produce a binary answer.
Maybe personal identity through neuron replacement is a sorites problem wearing a lab coat. If so, the search for the threshold isn’t failing to find an answer. It’s failing to notice the question was built on the assumption that a threshold has to exist.
What Continuity Was Already Doing This Whole Time
There’s a detail that usually gets skipped when people first hear the thought experiment, and it’s the one that changes how I think about the whole thing.
You are already the ship of Theseus. Most of the atoms making up your body today were not part of you a decade ago. Cells die and get replaced, and the ones that don’t get replaced, like most neurons, still churn through their proteins and molecular components on a rolling basis. The specific matter you’re made of has been swapped out, gradually, your entire life, and nobody treats that as an open question about whether you’re still you.
What’s carried the sense of continuity through that ongoing replacement isn’t the material. It’s the pattern: memory, personality, the uninterrupted thread of experience connecting each moment to the last one. The artificial neuron thought experiment doesn’t introduce a new problem. It just does deliberately, on a component level and with a synthetic replacement, what your biology has already been doing constantly and quietly with organic ones.
That doesn’t dissolve the question. If you replace enough of the pattern-holding substrate fast enough, or with something different enough in kind rather than degree, it’s fair to ask whether the pattern really does transfer cleanly or whether we’re just assuming it does because the alternative is uncomfortable. I don’t think anyone actually knows whether an artificial neuron would carry forward whatever it is that makes experience feel like something, or would just carry forward the outward behavior while leaving that part behind. Nobody has run the experiment. Nobody currently could.
What I keep landing on is that the ship of Theseus was never really a question about ships. It was a stress test for what we mean by “same thing,” dressed up in a story simple enough to argue about over drinks. Point it at a mind instead of a hull and the stakes go up, but the actual shape of the puzzle doesn’t change. We’ve just been running this experiment on ourselves the entire time, one replaced cell at a time, and mostly agreeing not to ask too closely what it means.
Like My Work? Want an Easy Way to Support Me?
Hi there! If you’ve enjoyed my work why not click the link below and Buy Me A Coffee? It’s simple and coffee keeps my creative and intellectual juices flowing!
Just click on the image below and it’ll take you to my Buy Me A Coffee page. Thanks so much for reading and supporting me!
