Slowly replacing neurons and synapses with artificial components?
This doesnât seem a problem in principle, or even in practice. See the monkey with the neuroprosthetic arm
here â a robotic arm wired right to the monkeyâs brain. The brain is an electrical device, and the electronics are well within practical ranges and tolerances for our non-biological electronics.
Competing efforts in AI look to virtualize the human neuron mesh completely; all the neurons are âsoftware neuronsâ, and what is emulated is
topology and architecture of the neuron mesh. Thatâs got some good advantages (super flexible, easy to maintain and improve, cheap), but itâs terribly slow with current hardware/software.
Thereâs nothing about a âprosthetic neuronâ that is problematic in terms of electronics or signaling in the human brain, so far as is known (and thatâs a good bit, now).
OK -
I´ll stick my neck out and say it cant be done - that is you cannot perfectly replicate the brain, and my feeling is that is due to inherent non-linearity of each component that makes up teh whole.
If you could manufacture prosthetic neurons, you would simply clone (or properly replace) the very same non-linearity and parallelism of the human brain. Maybe you donât understand whatâs being suggested here: take a box with 100 billion prosthetic neurons (plastic and metal, etc. finely manufacture by IBM or Raytheon or some such), and replace bio-neurons one at a time. Now, youâd have to do that quick in practice, or the patientâs anesthesia will run out, and 100 billion(or more) prosthetic neurons is a lot to switch out, to be sure. But in principle, the replacement happens on a 1:1 basis, and when complete, the âprosthetic brainâ, all copper, glass and plastic now, has EXACTLY THE SAME NETWORK ARCHITECTURE as the bio-brain it replaced.
Itâs precisely non-linear and parallelized in the exact same way the bio-brain was.
Once thatâs understood, it should make an important point apparent: the challenge of the brain is not tied up in its âorganic-nessâ, but in its complexity and network configuration. Itâs the
pattern and structure of the connections that is essential, not the medium of the neuron.
If thatâs the case, then a âprosthetic brainâ is really a challenge of producing sufficient patterns and structures, and is not contingent on being biological at all.
Each nueron is unique in that it might have variation of density, fluctuations of forces. Perhaps even at the aub-atomic level there are variations.
Yes, but that makes artificial intelligence all the more practical. Neurons vary in their tolerances and performance, from neuron to neuron. They arenât perfect, or perfectly consistent from one to another. This is fortunate, as any prosthetic neuron would not be âperfectâ either.
But it wouldnât need to be. It would just need to operate in the wiggly tolerances of the biological neurons.
Then there is chaos theory. You know the famous butterfly effect. Well, it means that even allowing the tiniest of inacuracies in the initial conditions and you might get wildly different results.
Yes, but statistics! Probabilistically, these stabilize and become predictable at scale, just like everywhere else in physics. Bio-neurons and prosthetic neurons are at parity here, with the possible exception that prosthetic neurons are likely to superior in performance in a way something like a titanium prosthetic arm is stronger and lighter than the real thing. Perhaps prosthetic neurons would have be degraded a bit in their manufacturing to effect more realistic emulation of the bio-neurons.
Now if there are variations at the sub-atomic level, and if we need to iron those out two, then we are alread yundone, because we are already entering the Heizenberg Uncertainty Principal area.
Why? The Heisenberg Principle seems hard to place as relevant here, and even if that should be done, it applies to the prosthetic neuron as much as the bio-neuron.
So no I cant see it working at a materialist human-scientific level.
But for God nothing is impossible of course, and He will recreate our bodies.
I canât see the problems, here. I donât see that youâve identified any. There may be problems, but they arenât mentioned here.
-TS