The more likely route to uploading is probably destructive brain scanning and imaging, which requires improvements that are more incremental and economical than fundamental – ie we can already do it in some smaller scale form today.

The gradual nanobot replacement scenario is more philosophically pleasing, but it is essentially equivalent in end result.

My simplistic argument was showing that *if* physics permits this gradual neural replacement uploading – such that it preserves full functional brain equivalence, ie there is no essential change to the person’s thinking – then we must accept that physics trumps philosophy and that if this is possible than there is no pure *philosophical* argument against uploading.

A philosopher might say that this can be reduced down to a tautological acceptance of functionalism, but so be it – thats not worth arguing about. Uploading seems to be possible from physics, so its an engineering issue. But many people seem to have some lingering suspicion that it isn’t even possible for some deeper philosophical reasons.

This of course just all comes back to the issue of AI in general. Physically we have the brain as an example, we know these particular information processing systems can learn language and develop into minds, and the challenge is largely one of reverse engineering the brain. Philosophical doubts are reasonable only before this technological hurdle is reached. Afterwards people will just accept that machines can be conscious for the same simple functional reasons that they accept other human brains are.

In the same fashion, people will eventually accept that uploading is possible – that it really will be a continuation of your individual conscious experience – once they meet and talk to successful uploads and identify them as the same people they knew in the flesh – a personal turing test if you will. Perhaps some sects of people will insist that these uploads are not the same individuals if they don’t have the same physical neurons – that they are some sort of philosophical zombie – but I suspect this will be a minority position, like the belief that only certain people truly have souls.

No, of course not. Personality changes over time and as we learn new things, or get exposed to new ideas, we change our perceptions and outlook. This is why arguing that we can just replace the whole brain and come out with the “same exact” person is not a valid statement to make.

“You seem to have replaced one person’s pie in the sky optimism with your own pie in the sky pessimism.”

How so? Did I say that somehow enhancing the brain with cybernetic implants would be totally useless and impossible by any means? Not at all. I simply made a note that if you haven’t done the scientific research, you don’t get to claim what will work and how on the basis of your imagination, defending the idea of cyber neurons in the comments. Why argue with something I never wrote in the first place?

Now on my side, I have to take issue with something you said…

“… exploration of the frontier of knowledge requires sacrifice and I can assure you there are those that will gladly make that sacrifice.”

As a tech designer, I would not let someone sacrifice their life to QC my software. For me to kill another human being with a mistake solely to see if I had it right is simply too great of a risk. It doesn’t matter if they want to give their lives willingly for the sake of progress. I simply refuse to kill for a breakthrough and would much rather find a far less dangerous way to test my designs.

I think the idea of developing a functionally equivalent, the ‘exact’ copy, of the state of the human brain is dependent upon the existence of a static process in which to emulate. However the brain depends on literally trillions of stochastic processes from which consciousness emerges, and none of these processes are overly important to the final expression of being that which you and I describe as consciousness. You make the false assumption that the current state is the only possible description of that being, and ignore the temporal aspects of that consciousness. Our thoughts, feelings, memory and learning all occur within both time and space. As such both the wiring of the cortical neurons as well as their axonal synaptic conduction delay is of prime importance to this learning and memory. And that delay is actually a series of electrochemical pathways through which information processing emerges, given enough complexity in the network. So while there will likely be errors at any one of these levels in a transfer, the brain has already evolved to maintain itself when these sort of errors occur naturally.

Humans are essentially a complex network, in both our genetic and neuronal structure. This both makes understanding how we function more difficult, but more importantly gives our bodies and brains a robust structure. This is bad when trying to fight cancer, but good when someone has a stroke. For this reason I believe your complaint about bugs and errors is relatively misplaced, especially when you consider the uploading process occurs over time. Since the human will certainly still be alive, the neuronal processes will continue unabated. As the individual neurons are replaced by their mechanical doppelganger, the existing network will adapt in the needed ways to accept that new process within the whole. If ‘bugs’ were introduced in the wiring of the network, the brain’s own robustness would simply overcome them and rewire itself as necessary.

Since the transfer of biological to mechanical neurons will happen over a period of time, one can assume the existing neuron network will adapt with the new neurons and any ‘errors’ introduced will be resolved once the neuron begins firing and communicating. This is because if the neuron’s response to the graded inputs is not accurately tuned, it can easily tune itself through the natural processes of LTP. But I would rather believe that if we had nano-bots with the capability of replicating the processes of the neuron it would be simple enough for the nano-bot to observe the behavior of the neuron due to it’s presynaptic inputs, and mimic the essentially boolean reaction with enough accuracy to maintain the polychronous subnetworks which are likely the basis for the neuronal networks extremely high memory and learning capacity.

You seem to have replaced one person’s pie in the sky optimism with your own pie in the sky pessimism. Both lack any basis in science or research. However until you have some underlying understanding of how the brain works you cannot simply piss away the entirety of neuroscience research. Sure there is a great deal left to learn, but we have some fundamental understandings of the processes within the brain. And with a small understanding your arguments are not really valid.

A heart can be marginally functionally emulated with an artificial heart, but the complexity of the original’s construction implies various knock-on effects (i.e., emergent effects) that will always create some kind of slippage between the original’s and the simulation’s functionality. Thus, an artificial heart imposes different constraints on the organisms behavior than would a real heart (and even a real heart transplant imposes behavioral constraints). Perhaps an artificial heart may day fully emulate the behavior of a real heart, but that possibility is highly unlikely.

Emulating the behaviors of a neuron, even to very high tolerances, would very likely introduce slippage as the complexity grows. Notwithstanding the engineering difficulties (how would the support functions of glial cells be replicated when dealing with artificial neurons?), if indeed such a proposition were tenable, the likelihood that it would cause unforeseeable changes to the consciousness itself would border on the unethical. At the very best, it would be consciousness similar to what we know as consciousness, but more likely a fragmented and constrained consciousness — and at worst, a warped consciousness akin to Frankeinstein’s monster. Best to tread lightly.

I totally agree which is why I don’t even try to rule on the actual possibility, only give an opinion on the quality of the thought experiment and the practical considerations as far as my experience with computer science is concerned.

“… one could surmise that it is only this memory/habit section that truly defines who we are.”

Yes, one certainly could but we need to remember that the division is conceptual. How the brain is actually wired and where the information we want to deal with lives, is a very complex matter.

But you do have a very good point, so please don’t mind my technical objections. I know they’re low level implementation concerns but they do matter to someone who wants to derive functional requirements for new software and hardware.

For example:

1. A non-nanoscale device could be constructed to function as a portion of the brain. Such a device could be connected to a much larger machine that performs its functions.

2a. Given that the brain can loosely be divided into two sections: the “automatic” or hardwired section that controls things like metabolism or eyesight, and the flexible, learning section that governs things like memory and habit (the sum of all that you have learned), one could surmise that it is only this memory/habit section that truly defines who we are. The rest is just stimulus from our instinctual side. Artificial neurons would *not* need to be wired exactly. The brain’s memory relies on a system whereby the loss of a single neuron does not significantly degrade function. If the process occurred over enough time, the artificial neurons could wire themselves to mesh with our existing memories. Anything one was reminded of over the course of the process, one would retain.

I’m actually going to jump to Connell’s defense on this one. Trying to plot the exact function of a brain into an external matrix would only work in Ghost in the Shell, so you would need to get the actual flow of information and signals in the right format before trying to plot them over.

Otherwise, how would you transfer the weak electrical hum of the brain over to a computer network?

That’s about as useful as ontological arguments, and just as flawed. And worse, it relies entirely on the idea that the brain works as imagined, and that nanotechnology could even come close to functioning as neurons. Hell, your warp drive exposition at least relies on known physics – the nanobot idea hasn’t even gotten that far. This has nothing to do with theory – it’s strictly philosophy, if not outright fantasy. They might as well spend time arguing about angels on the point of a pin.

Here’s the point I find most amusing, however: If we were to actually achieve the technology to understand the function of each and every neuron, why bother screwing around with nanobots? Plot that info directly into that external neural matrix. What possible point does replacing neurons with an exact (heh!) machine replica achieve? That’s like replacing sand grains on the beach with carbon-fiber Sandeeta(tm).

A suggestion, to those who think that brain replacement/infinite neural lifespans are a great idea: you might be better off getting human brains working up to par first. No one needs a perpetual idiot.