Hacker News Reader

Let's assume one neuron is replaced every second. No problem, nice and gradual. What if it's one every millisecond, every picosecond? Should make no difference, because it's still one by one.

It should make a difference how long it takes, though.

Assume that consciousness is a physical process. If new neurons take part in this process after the old ones die (so that we have continuity as part of normal "healing"), there must be some time scale over which those neurons become an effective part of the consciousness process.

At minimum, if consciousness is a "network effect", this would be on the order of the latency of inter-neuronal communications. If the cells are doing some processing which forms part of consciousness, the time scale would increase.

To throw out some wild numbers: neuronal latency seems to vary, but I found a number for visual-cortical neurons of about 80 ms. (http://redwood.berkeley.edu/bruno/npb261b/whitney-reading/kr...) And you can't replace more than a few cells at a time (I wonder what fraction?) if you want to avoid impairing consciousness during the process.

(This also doesn't include whatever the time is to make an effective copy of a given neuron and it's state; we're assuming instant perfect copying.)

>If someone scans my brain, builds an artificial copy, instantly cuts out my brain and inserts the new one, I clearly die (because I got my brain/me cut out) and my copy lives on. From the outside, I'm the same, but the first me died.

If you look at the physics the right way, your brain is cut out and a copy inserted billions of times a second already.

You have an old car, and you want to renew it. If you buy it brand new and toss the old one, you are destroying the old one. If you keep changing pieces of it, testing them to see they work, and keep going until you have replaced the whole car, it's still the old car, just brand new.

There's not much difference at the end, except that you can consider that the second way of doing things will let you still use the old car, only it'll be new.

I largely agree that gradual replacement has the far better claim to preservation of identity. I would have hardly any desire to be uploaded any other way.

However, the car of Theseus example allows for a worrisome wrinkle. What if you kept all the old parts during the process you describe, and then after, you re-assembled them? Which car has the superior claim to being identical with the old car? I think it obvious that the reassembled old parts have the superior claim over the car produced by part-by-part substitution of new parts.

Not for me. For me that's a brand new old car.

Yes, it has all the parts of the old car, and only those pieces, but it was assembled anew from zero.

A more difficult dichotomy would be: what if we exchange parts between two cars until each one is the other one. That's hard to decide.

I find that first intuition odd. What if we alter to situation to not involve the new parts? Simply disassemble the old car parts and reassemble them, is that a brand new old-car?

Or how about we make all the new parts plastic so that no functioning "car" is produced by the gradual part-for-part substitution. We end up with two cars, one an all-plastic car model, and one consisting of all the old parts assembled and functioning. Which has the better claim to being the old car?

Yeah, part exchanges quickly leave me with even less salient intuitions about the proper use of old identity terms than these more "splitting" oriented cases. What should we say if a lion talked and all that.