3 ms·

We should clarify the definition of "uploading". I see two cases: 1. We gradually replace each neuron in-place - without any external connections. This means a

We should clarify the definition of "uploading". I see two cases: 1. We gradually replace each neuron in-place - without any external connections. This means a biological brain slowly turns into a silicon brain (or whatever material used for new neurons) 2. We gradually replace each neuron with a connection to an external computational system which provides the signals that the original neuron would output. This is what I mean when I say "brain uploading". The first case is more straightforward, and I'm not discussing it here - I expect it to be possible when the technology is ready. The second case might have fundamental issues. The main issue is about preservation of consciousness - let's try to define it first. Consciousness is a property of a brain, which most likely depends on the signals between neurons, and which might or might not additionally depend on each neuron internal electrochemical state. When we change a brain we change its consciousness. If we change it too much, we might alter consciousness to the degree where it is not longer the same identity. This might be happening to a degree as a person grows, though I'd argue that the person would still be the same, in a sense that the person would consider himself to remain the same person over the years. For example, I remember myself 20 years ago, and even though I was somewhat different in terms of views, habits, drives, etc, I'm sure I was still me - I was not my neighbor Mike - I'm still the same captain of mostly the same ship. If we replace neurons as in case 1 above, I believe the continuity of consciousness can be preserved. But in case 2, I'm not so sure, because we are fundamentally changing the structure of the brain, not just individual components. Going back to my original example of just 3 neurons, replacing one neuron with a silicon based device in-place does not fundamentally change the overall structure of the 3-neuron brain. It's still just 3 neurons operating together as a whole. However, if we replace one neuron with a connection to an external computer, we no longer have the 3-neuron brain. We have 2 neurons which operate as before, because they still receive the same inputs from the third neuron as before, but the third neuron has fundamentally changed - it is now receiving external signals that the original third neuron did not receive, and therefore, the structure of the original brain is now different. I don't know if we can claim the consciousness of the original 3-neuron brain is preserved. The problem is that the other two neurons cannot adapt to this change, because from their point of view nothing has changed. Now if we consider that the brain has 100B neurons, and we replaced 50B neurons per case 2 - now I'm not sure what is happening to the consciousness because half of the brain is receiving external signals from some complex computational system. Does new consciousness extend to that external system? Why, or why not? If yes, how is it impacted? Re: "emulating a mechanical watch in software". Can we upload the mechanical watch to the cloud? No, we can only replicate it. But that is not relevant to our discussion, because a mechanical watch has no consciousness which could change or be transferred.

542354234235 2y ago
That is an interesting perspective. I actually think that a gradual shift to an emulated brain would keep you more “you” than replacing neurons in place with non-biological neurons. The reason is the emulation and is why I used the watch analogy, which was not about duplication vs transferring, but was about how emulation of physical objects and the connections between them is the same, for all intents and purposes, as the “real” version. In your 3 neuron brain analogy, the emulated neuron would not be receiving external signals. It’s physical properties would be emulated in software so the neuron would function i.e. behave the same way, as the original. Much the same way that an emulated mechanical watch mainspring will tension when you “wind” it and release that energy to “power” the watch the same as the watch it was emulating. Of course the winding, tension, etc. aren’t really happening. They are software running in a computer, but the key is they are behaving in the same way as a real watch component. Back to our emulated third neuron, it would not really be going through the electrochemical process of changing its action potential based on the chemical signal received from one of the other two neurons. What it would be doing is emulating that process in software in the same way as the biological one it replaced. It is not “receiving external signals” in that it is not a computer serving as a neuron, or like plugging your brain into a computer so you can directly interface with it. It is a computer running software and that software is running a physics engine of a neuron and the only external signals transmitted to that neuron running in software is an emulated duplicate of the signal sent from one of the other two neurons. However the neuron responds based on the rules of the physics engine, which again is built so that the neuron will behave the same as its biological model, would be converted to a real signal sent back to the other two neurons. So, nothing would fundamentally change for the other neurons to adapt to. Now, lets look at the non-biological silicon (or whatever material) brain. A silicon neuron would not behave the same as a biological one because it is made of totally different materials, likely chosen to act faster, more efficiently, more accurately, etc. In that case, the new neuron would not respond in the same as the biological one it replaced. I would say that would alter the functioning of the brain much more drastically.
- kadushka 2y ago
  I'm having trouble understanding, or perhaps, accepting, your point of view. When you say "It’s physical properties would be emulated in software", what really happens is still electrical signals flowing between transistors. "Software" is a concept - it does not exist in a physical world - it's our interpretation of the physical processes that happen when we apply voltage to transistors. I'm not 100% sure, but I think that consciousness might similarly be a concept - "our" interpretation of what is going on between our biological neurons. Note I put "our" in quotes, because it's self-referential here, and that's why I'm not 100% sure. But if we get back to my 3 neuron scenario, the third neuron which is emulated by the external system has not disappeared into the "software". It's still a physical device, which functions in a physical world, and is part of the original brain, despite the fact that it is now physically different from the original biological neuron it has replaced. It still produces (relays) the same signals towards the original two neurons, but it does not compute those signals. Both the inputs and outputs of this new proxy neuron are different, because it has to communicate with the external system. The two original neurons still communicate with a physical device, not with some abstract software entity. However, if we look at consciousness differently - as something that is purely in the signals and not in the physical devices carrying the signals, then it changes the perspective. In that case it's harder for me to imagine the effect of extending computational signals into the cloud, because I'm not sure if the external signals become an extension of the original consciousness, or if they become more like sensory inputs to the local brain - but then the local brain becomes smaller? It's hard for me to comprehend this perspective.

542354234235 2y ago

That is an interesting perspective. I actually think that a gradual shift to an emulated brain would keep you more “you” than replacing neurons in place with non-biological neurons. The reason is the emulation and is why I used the watch analogy, which was not about duplication vs transferring, but was about how emulation of physical objects and the connections between them is the same, for all intents and purposes, as the “real” version. In your 3 neuron brain analogy, the emulated neuron would not be receiving external signals. It’s physical properties would be emulated in software so the neuron would function i.e. behave the same way, as the original. Much the same way that an emulated mechanical watch mainspring will tension when you “wind” it and release that energy to “power” the watch the same as the watch it was emulating. Of course the winding, tension, etc. aren’t really happening. They are software running in a computer, but the key is they are behaving in the same way as a real watch component. Back to our emulated third neuron, it would not really be going through the electrochemical process of changing its action potential based on the chemical signal received from one of the other two neurons. What it would be doing is emulating that process in software in the same way as the biological one it replaced. It is not “receiving external signals” in that it is not a computer serving as a neuron, or like plugging your brain into a computer so you can directly interface with it. It is a computer running software and that software is running a physics engine of a neuron and the only external signals transmitted to that neuron running in software is an emulated duplicate of the signal sent from one of the other two neurons. However the neuron responds based on the rules of the physics engine, which again is built so that the neuron will behave the same as its biological model, would be converted to a real signal sent back to the other two neurons. So, nothing would fundamentally change for the other neurons to adapt to. Now, lets look at the non-biological silicon (or whatever material) brain. A silicon neuron would not behave the same as a biological one because it is made of totally different materials, likely chosen to act faster, more efficiently, more accurately, etc. In that case, the new neuron would not respond in the same as the biological one it replaced. I would say that would alter the functioning of the brain much more drastically.

kadushka 2y ago
I'm having trouble understanding, or perhaps, accepting, your point of view. When you say "It’s physical properties would be emulated in software", what really happens is still electrical signals flowing between transistors. "Software" is a concept - it does not exist in a physical world - it's our interpretation of the physical processes that happen when we apply voltage to transistors. I'm not 100% sure, but I think that consciousness might similarly be a concept - "our" interpretation of what is going on between our biological neurons. Note I put "our" in quotes, because it's self-referential here, and that's why I'm not 100% sure. But if we get back to my 3 neuron scenario, the third neuron which is emulated by the external system has not disappeared into the "software". It's still a physical device, which functions in a physical world, and is part of the original brain, despite the fact that it is now physically different from the original biological neuron it has replaced. It still produces (relays) the same signals towards the original two neurons, but it does not compute those signals. Both the inputs and outputs of this new proxy neuron are different, because it has to communicate with the external system. The two original neurons still communicate with a physical device, not with some abstract software entity. However, if we look at consciousness differently - as something that is purely in the signals and not in the physical devices carrying the signals, then it changes the perspective. In that case it's harder for me to imagine the effect of extending computational signals into the cloud, because I'm not sure if the external signals become an extension of the original consciousness, or if they become more like sensory inputs to the local brain - but then the local brain becomes smaller? It's hard for me to comprehend this perspective.