So, I'm going to point out some quibbles in the OP, and then get into it.
First, brain in a jar isn't 100 or even 50 years off. It's more 15-30 years off.
I think the worst part, in fact, would be that the whole system would likely overlook how to get your rocks off.
So, I'm going to point out some quibbles in the OP, and then get into it.
First, brain in a jar isn't 100 or even 50 years off. It's more 15-30 years off.
The computer tech is, the biological side of it we simply don't know enough to make such a prediction.
Second, you would have to at least get the whole spinal column too for any useful results, and possibly any neural clusters particularly around the larynx, tongue, eyes, arms, hands, legs, and feet. This is because muscle memory and signal preprocessing is very important.
I think the worst part, in fact, would be that the whole system would likely overlook how to get your rocks off.
The simulation needs to include the hormonal aspects.
The computer tech is, the biological side of it we simply don't know enough to make such a prediction.
Second, you would have to at least get the whole spinal column too for any useful results, and possibly any neural clusters particularly around the larynx, tongue, eyes, arms, hands, legs, and feet. This is because muscle memory and signal preprocessing is very important.
The simulation needs to include the hormonal aspects.
Oh, the biological side is easy enough. It comes down to mapping which neurons express surface receptors for the hormone in question, and what effect that exposure has as to antagonizing or agonizing the signal.propegation. I suspect it's just a matter of breaking out the bias on the network component+wise. The other major difficulty is in mapping how actual reinforcement of connectivities and other "plasticity events" happen.
The computer tech is, the biological side of it we simply don't know enough to make such a prediction.
Second, you would have to at least get the whole spinal column too for any useful results, and possibly any neural clusters particularly around the larynx, tongue, eyes, arms, hands, legs, and feet. This is because muscle memory and signal preprocessing is very important.
The simulation needs to include the hormonal aspects.
Oh, the biological side is easy enough. It comes down to mapping which neurons express surface receptors for the hormone in question, and what effect that exposure has as to antagonizing or agonizing the signal.propegation. I suspect it's just a matter of breaking out the bias on the network component+wise. The other major difficulty is in mapping how actual reinforcement of connectivities and other "plasticity events" happen.
I don't think it's nearly that simple. We need to know what matters (simulating everything is impractical) and we need to be able to do it without losing the information in the process--you could record the surface easily enough but how do you get the information beneath without the brain dying and decaying while you're doing it?
The bigger issue I think will come with regards to atypical people. As it is, we're right around where I expected us to be the last time this came up here.
I don't think it's nearly that simple. We need to know what matters (simulating everything is impractical) and we need to be able to do it without losing the information in the process--you could record the surface easily enough but how do you get the information beneath without the brain dying and decaying while you're doing it?
So, for the most part it is a function of making an identification of the qualities of a given neuron. At it's basic level, the neuron is just a machine with various levers that bias it one way or another. If you look at the mechanics of neurons in general, and the mechanical elements of any given neuron specifically (by it's chemistry, methylation, and so on), it is just a matter of properly tagging those qualities. It doesn't need to be seen in real time as long as you can figure out sufficient minutae to correlate what is seen to how whatever is being seen behaves in general.
At any rate, I'm talking full destructive scans: turn me into a meat popsicle and shave me down layer by layer with a diamond razor and feed each one through an electron microscope, makes no nevermind to me. As long as the model for reconstructing and simulating the resultant data is sufficient, I'm not too bothered. If they get it wrong the first few times? Well, I hope they save the data at least. That's actually one of my contracts with myself, to not shy away from at least trying to evaluate whether I can handle and live with particular experiences.
The bigger issue I think will come with regards to atypical people. As it is, we're right around where I expected us to be the last time this came up here.
Why would atypical people be a problem? If the process doesn't cover whatever it is that makes a person "atypical", it implies that we're using some sort of statistical approximation for some features. Which means that although the simulated brains, even if they may mostly behave like the originals, are fundamentally different persons. Or maybe I misunderstood what you meant by "atypical".
That would be more like the AIs in Caprica or Transcendence: a traditional AI that has enough information about a person (social media record in the former case, a non-invasive brain scan of Johnny Depp in the latter) that it behaves like one, but is not.
The bigger issue I think will come with regards to atypical people. As it is, we're right around where I expected us to be the last time this came up here.
Why would atypical people be a problem? If the process doesn't cover whatever it is that makes a person "atypical", it implies that we're using some sort of statistical approximation for some features. Which means that although the simulated brains, even if they may mostly behave like the originals, are fundamentally different persons. Or maybe I misunderstood what you meant by "atypical".
That would be more like the AIs in Caprica or Transcendence: a traditional AI that has enough information about a person (social media record in the former case, a non-invasive brain scan of Johnny Depp in the latter) that it behaves like one, but is not.
You can't get away from statistical/ML/AI hinting in the process of reconstruction for the reasons Loren states, mostly accounting for the distortion created in the process of actually doing the destructive scan. You have to have some.model that corrects and makes assumptions. To that end, you would have to slice up a lot of brains near time of death at different cadences. Connectivities can be inferred, but only within the bounds of the analytical model.
