1) I'm surprised that you as a neuroscientist make a definitive claim that we need to simulate every atom of our body to ensure that "you" is still "you". Clearly, cutting my nails destructs my persona ;) (Less flippant, it's kind of hard to say which parts of us make us ourselves. I certainly don't see much of a consensus on that)
2) What about the old idea of replacing one neuron at a time with a simulation? (The "Moravec Upload") Caveat: Provided our identity is just defined by our neurons, which is certainly more than up for debate.
3) I'm not aware that there's a claim of exponentially increasing human intelligence for the singularity. The idea is for technology to become self-improving. We'll see about that, but it's certainly a different claim than what you're refuting.
Preface: Obviously I'm not making any 100% claims. Nothing is impossible, someday, etc etc
1) Ok, I was generalizing for the sake of brevity. Agree that there isn't a consensus, but it's pretty safe to say that it is somewhere between "your brain" and "everything".
2) To do this, you would also have to effectively freeze all the other neurons in your brain. Neurons don't operate in isolation. If you remove one from the network, the rest start (ever so slightly) re-adjusting weights, synapses, dendritic arbors, etc. So I suppose if you could put someone into a hibernation where their neurons were effectively frozen, this may be possible. Ignoring the invasiveness aspect too.
3) I suppose this boils into a big argument about what intelligence actually is, and I don't want to start an internet argument about that. Suffice to say, no one really has any idea how to define or quantify intelligence. How do you improve intelligence? It certainly isn't the addition of more knowledge (although that helps you become effectively more intelligent, since you can stand on the shoulders of those before you). If you take my brain and speed it up so I can do the same thing but faster, am I more intelligent? I don't have a good answer to that.
Perhaps hyper intelligence is nothing more than being able to think intelligently really really fast.
> "If you remove one from the network, the rest start (ever so slightly) re-adjusting weights, synapses, dendritic arbors, etc."
I am not terribly familiar with neuroscience, so you will have to forgive me for asking, but why would this be particularly troubling? Is the concern that surrounding neurons would neglect to "hook up" with the synthetic one, once they got used to nothing being there?
Yep, pretty much. And even if they do reconnect, it is highly likely that their relationship will be different. Change in number of synapses, strength, even physical locations make a difference. Synapses at the end of a dendritic arbor affect the neuron differently than those that are close to the root. Even inter-synapse spacing is important, as adjacent synapses have modulatory effects on the reactivity of it's neighbors.
None of this is overly important per-se. Your brain doesn't have a specific pattern of arrangements that must be just so to make a human. But as you start changing these little details, the changes will eventually bubble up to create a different "you".
How much different is hard to say, maybe not even enough to be noticeable. This is firmly in "theoretical" territory. But considering how drastically people can change from relatively small strokes, I would hazard that there isn't a huge amount of deviance allowed.
Perhaps it's possible if you can do it finely enough that only single neurons are changing, allowing time to resynapse (or artificially recreating those connections?). Not sure.
Trouble getting real neurons and synthetic neurons to interact would certainly throw a wrench in everything, but I am not so sure that concerns about changes to the individual are much of a show stopper. If you are digitizing a "person", they are going to be dramatically altered one way or the other.
Either way though you are going to create an intelligence that will likely be unable to relate to biological people, and biological people will probably have just as much difficulty relating to it. I wouldn't be surprised if an uploaded person was viewed as some sort of psychopath after the procedure, no matter how perfect the procedure was. (And considering how biological people will probably treat it, it would probably be very justified in thinking the same of them)
As long as the individual is happy with the results, no matter how changed they become, that is probably the best we could hope to shoot for.
2) What about the old idea of replacing one neuron at a time with a simulation? (The "Moravec Upload") Caveat: Provided our identity is just defined by our neurons, which is certainly more than up for debate.
3) I'm not aware that there's a claim of exponentially increasing human intelligence for the singularity. The idea is for technology to become self-improving. We'll see about that, but it's certainly a different claim than what you're refuting.