A very thorough and well written explanation of the problem and the serious doubts involved in this idea of recording/playback of brain states, it's complexity, and the likelihood that it at all represents "consciousness". Great read, thank you for sharing.
I think there are orders of magnitude more complexity missing in the 64 x 10^22 bits of information per state/sec that need to be considered if we want to successfully represent brain function in a computational model. Looking at this from a computer science perspective, were we to hope to make a single scan of the brain at a particular point in time to be able to later run that configuration in some brain simulation model (a la "mind upload" as explored in a lot of sci-fi), I think we should consider _at least_ the following factors:
* Neuronal connectivity map (many-to-many graph of the dendritic/axonic connections between the neurons.
* Action potentials of each neuron or in other words the present state of ions in the cell (though I believe this information could be lost at the cost of subjective short term memory loss)
* Current cellular regulatory state of ion channels (various types) in the cell
* Current cellular regulatory state of the production of each of various neurotransmitters
* Current cellular regulatory state of the receptors of each of the various neurotransmitters and other related channels
This information would need to be gathered in all structures of the brain including the accessory cells (non-neurons) that don't participate in the overall map but do have a biochemical relationship to brain state. The information relating to the real-time neurochemical state of the cells could possibly be read from the epigenetic expression of the DNA in the nucleus, which may or may not be a simpler way to represent or read the information.
In addition to considering the computational requirements for storing and processing this information to run the model, we also must take into account the feasibly of capturing this information without disturbing or destroying the state. This is an idea I've been brewing for a while, hopefully I can try to take a stab at the calculations some day.
I think though the exercise would lead to similar conclusions as yours regarding the infeasibility of the proposal. This not even taking into account the role the senses and the body play on emotion and as such personality and behavior.
Love these kinds of analysis, these were the kinds of ideas running through my head while reading Reality+ and it's great to see them laid out in such great detail.
I agree with what you're saying about requirements. I did take a stab at estimating the static data for representing the mind. And that doesn't include the code/data necessary to run the model. You make some good points about the cellular machinery.
Exactly as you say, a connectome is necessary, and it turns out dominate the size. I went with 64-bit pointers and two-way many-to-many links. That's 500 trillion times 128 bits, and I guessed 64 bits would be the minimum to represent synapse type, state, etc. Possibly auxiliary tables for common data to reduce duplication. The hundred million neurons turn out to be a drop in the bucket in comparison. My model, a minimal one, was a hair over 12 petabytes (the hair because of the neuron states).
All of which makes me very dubious we live in a simulated reality. The computation requirements are just too formidable for such to be common.
>All of which makes me very dubious we live in a simulated reality. The computation requirements are just too formidable for such to be common.
From my perspective the simulation hypothesis should receive about the same skepticism as Solipsism. I think there is a huge lack of understanding of physics and computer science in the philosophy of mind community and as such it is difficult to dispel the myth. Hopefully efforts like yours can break down the edifice. I believe there is a way to prove it absolutely, especially when you consider the level of precision and consistency we can measure in physics from the LHC to the JWST.
The same skepticism as solipsism, yes, exactly. Good way to put it.
And I very much agree a lot of the philosophy community is behind the times when it comes to the physical sciences in general, let alone computing. They're steeped in outstanding thinkers from centuries ago, and those thinkers, in terms of physical reality, were massively ignorant. They don't deserve the unqualified reverence they're often given. (I think Berkeley and Hume were just plain wrong on some points.)
When I wrote my original comment I had this skeptical voice in the back of my head making the drop-in-the-bucket argument about the comparison between neuronal configurations and the connectome but I continue to think the informational complexity is being underestimated.
If we assume there are ~170 billion cells in the brain we are interested in (neurons + glial cells) and that we need to measure the expression of ~500 genes (rough estimate taking into account the genes for the multitude of neurotransmitters, their receptors/channels, and the associated regulatory genes) using the same 64 bits for each data point, we get an answer on the order of 10^15.
So I was almost certainly exaggerating by saying "orders of magnitude more..." but it would seem that at first glance the neuronal state _could be_ just as complex as the connectome. I tried searching to find any literature on this subject but was unable to, though my lack of familiarity with the correct technical terminology is often a hindrance.
I was at least able to find this article supporting the idea that this information is relevant to capturing aspects of human brain function and cognition:
I have no doubt my estimates are underestimates. The goal is only to find a minimal requirement — which turns out to be huge enough — with the acknowledgement that the reality would be much larger. As you mention, the glia cells are important. The myelin sheathing also matters in affecting the signal speed along axons. Neurons have an "on" and "off" state, but the on state consists of pulses, and the duty cycle of those pulses carries analog information. Some studies suggest the rise and fall times of the individual pulses carry information.
There's also the EM environment of the brain. It's strong enough that we can pick up brainwaves from outside the skull. I have to believe they play some role inside the skull. General blood chemistry probably also matters. Certainly, a few molecules of LSD or similar change cognition considerably.
Not an area I know much about, but my general sense is that genetics may not be the most fruitful path. The brain has a genetic template, but I would think the physical environment it grows in matters, and what a brain learns during a lifetime certainly changes it. Getting down to the genetic level is interesting, though. All cells would share the same genes.
I'm talking about measuring the epigeneitc markers (histone configuration, methylation etc) to understand the current regulatory state of the cell. Genes are literally physically stretched out or tagged in different ways that affect their expression and these markers or structures can change over time. When a neuron up regulates it's production of a particular neurotransmitter or it's associated receptors, there is an epigenetic change that occurs in the expression of the relevant genes. There are also other genes that control/regulate these epigenetic expressions. Much of this is still poorly understood and we have not the faintest idea I think on how to measure this in a living organism.
Some good reads:
The Gene by Siddhartha Mukherjee
Transformer by Nick Lane
Transformer is mostly about the Krebs cycle but it also talks a good deal about epigenetic gene expression and how it's dynamics effect cellular function.
Lotta interesting stuff. Ultimately, I think it can boil down to what can be functionally a blackbox and what details a simulation or recording has to have to produce results identical enough to qualify as a match. At the very least, we need good models of neurons and synapses and understanding details like you get into has to be part of that. Can’t make a black box without knowing what’s supposed to be inside. The question might be what details are necessary for understanding versus for implementation.
I’m also reminded of Brain Greene’s discussion of spacetime slicing in “The Fabric of the Cosmos” (and I’m sure many other books of his). Essentially Brian asks that we picture all of 4d spacetime as a loaf of bread that can be sliced into 3d chunks. Special Relativity tells us that there is more than one way to slice this loaf and depending on how we slice it, it will seem to tell different stories with respect to the order of events. While this doesn’t change the overall structure of space time, it does mean that things will appear to have a different ordering if we look at it from different perspectives.
I was drawn to this analogy in your discussion on the difficulty of creating distinct states in time with an asynchronous analog system like the brain. I believe the nature of these states doesn’t just relate to the frequency at which we are taking the snapshots but also the perspective from which we are doing it.
Taking the analogy into a hypothetical apparatus we can think of some scanning device that, let’s say, using an electromagnetic field measures in some sense the structural state of the brain at given moments via pulses it emits.
If this analogy is more than just a pretty metaphor, the states we measure would not look exactly the same depending on how we position the device and surely other particular implementation details. We might seem to see the order of firing taking place in the brain change depending on how we record events. This doesn’t change the overall fact that, given the arbitrarily “perfect” precision that we would supposedly need, in any case we will capture all of the relevant information regarding the brain’s activity.
This seems to imply that the supposed subjective experience of these sequential brain states could have various interpretations depending on the exact method we use to read the data which to me seems to be another logical contradiction leading us to refute the idea that conscious experience can be captured in frames of brain states. But maybe I am stretching the metaphor too much.
I think there would be no worries about simultaneity so long as both the brain and scanner were in the same inertial frame. There is also that causality is always respected regardless of relative velocity, so neuron A affecting neuron B is the same from all inertial frames. But the specific timestamps would depend on the frame.
*If* mind emerges from brain without quantum effects being needed (dubious in my mind), in principle (ha!) we should be able to read brain state without disturbing it too much. The difficulty is even getting a high-resolution complete snapshot of a living brain. In a lot of science fiction (and some science), it's dead brains scanned by slicing them. I don't think we even imagine the technology capable of scanning living brains. Nothing we have now comes close.
I think mentioning the ordering of firing was where I broke the metaphor, it's obvious to me now that given the speeds of the interactions in the brain we could in theory capture ordered states at some arbitrary degree of precision. I guess it's just I have an intuition that the character of those states, the exact values of all the 500 trillion numbers, would change depending on how we measure them.
The fact that they would produce different logically consistent stories for the same overall subjective experiences means that any schema of brain states doesn't carry some singular and consistent higher order property of "consciousness". It can still be a huge source of scientific data though that helps us understand how the brain works and if we could record "brain states" I'm sure we would learn more about our subjective experience.
Though I again am unsure how to detail this intuition and as is often the case for me and my neurotype, it is quite possible this one too should get chucked into the bin of delusions/confusions.
The 500 trillion numbers might change depending on the phase of the time slices or for any number of other reasons. I just don't think special relativity is a factor. If we assume consciousness emerges from a massively parallel network of causal events, it would be that causality we'd need to capture.
One aspect that occurs to me is that many forms of taking a snapshot (including most cameras) cannot process all that data at once. Shutter mechanisms can result in a sweep of the image over the film or sensor, and sensors must be read sequentially. So, timing is an issue in the sense that the requirement is an instantaneous snapshot of every aspect, and it's hard to envision any technology capable of that. In a SciFi novel, I'd techobabble about quantum entanglement between the brain and scanning system, but that doesn't actually work because every reading collapses the entanglement.
That shutter effect is exactly what I was trying to get at, thank you. Given the necessary limitations of our possible capturing mechanisms, I believe then that we could capture distinct sets of brain states that, while they do encompass the entirety of information necessary to represent the function of the brain, do not have the same exact information in each state. Another metaphor comes to mind of the different ways to unfold a cube. The point being there doesn't seem to be a 1-to-1 correspondence between a set of brain states and it's associated higher order configuration.
Yeah, that’s the sort of thing I was getting at in the Brain States part of the post. And as I emphasized, assuming these brain states are identifiable leads to assuming computationalism. So, they may be unicorns more than horses.
Heh. Speaking of unfolding a cube, the last animated video I did:
Imagine if someone said radios ARE music; if you build a radio, music will write itself ('emerge') and just start playing. That's what computationalism sounds like to me.
On your commentator's point on the "visualisation" of 4-space - a topologist I know who is way up there on knots, etc. kindly indulged a question from me on manifolds about how he "imagines" 4-space - it stuck with me as a useful (possibly only to me!) non-symbolic analogy - he thought of say a tesseract as I might a video of a cube over time, but all at the same moment. I hope that makes sense - natural everydaylanguage fails really.
Thanks, glad you liked it. That video was interesting. I was very much aligned with Tallis up to about the halfway mark when the conversation went off the rails a bit. Neither quite seemed to grasp what the other was saying and never reached for the difference between analog and digital systems. Tallis reached for it mentioning Shannon-Weaver, and he was basically talking about "information patternism" (another term for computationalism). I quite agree with him on that. But neither quite seemed to fully appreciate the physical requirements and analog nature of a brain scan.
OTOH, I thought the interviewer made a good point about history being encoded in our present state. Tallis didn't seem to accept that, but only hand-waved it away. (I don't respond well when people pick out words the way Tallis did with "encoding" and a few others. Seems more prevaricating to me.)
Neither seemed up on the technology. A shared VR for Mr. and Mrs. would of course be part of a computational immortality scenario as would an entirely user-customizable environment. Greg Egan has written a number of novels exploring that. They would both benefit from reading some of those. (Or my post! 😁)
Indeed! it's initially lucid but not sustained. Tallis is very smart but I don't think it maps well onto current discussions around some of the topics. I will now head off to read your essays on 4-space. Cheers, John.
I've spent some time thinking about 4D space as well. The tesseract is a particular fascination of mine. The image of a cube extended along the W dimension is very much like a short movie of a cube with the frames not extending in time but along the W axis.
Appreciate the raindrop analogy. Just as we can't truly capture the continuous, dynamic nature of rainfall by taking discrete snapshots, we can't reduce consciousness to a series of recorded states. It's not just a technical limitation I feel - it's a fundamental misunderstanding of what consciousness is.
What I found most interesting is the distinction between playback and computation. The Mandelbrot set for example makes this clear - displaying a pre-computed image is different from doing the actual computation. Similarly, playing back recorded brain states (even if we could somehow capture them) wouldn't recreate the actual experience of consciousness.
I think this gets at a deeper issue in AI and consciousness debates: we often mistake simulation for replication. Just because something can mimic the outputs of consciousness doesn't mean it has consciousness, just as a video of rain isn't actually making it rain. Just some thoughts.
Thank you, Neela, I'm glad you got something out of it. It sounds as if you not only have similar thoughts but have given this some thought. I think you're right on about simulation versus replication. I even see three possibilities: Emulation, Simulation, and Replication.
Emulation is a functional "blackbox" approach that goes after only the functionality of the mind irrespective of the physics of the brain. Simulation models the physics of the brain on the physicalist assumption that mind=physics. Replication models the physical structure of the brain, the physical neural network. I think the last might work, but I'm skeptical of the first two.
For reasons this whole series of posts tries to explain. It begins with this post:
Which lays the groundwork for the ones that follow (in "My Best Guess" newsletter) and explains more about what I see as the three options for machine minds.
A very thorough and well written explanation of the problem and the serious doubts involved in this idea of recording/playback of brain states, it's complexity, and the likelihood that it at all represents "consciousness". Great read, thank you for sharing.
I think there are orders of magnitude more complexity missing in the 64 x 10^22 bits of information per state/sec that need to be considered if we want to successfully represent brain function in a computational model. Looking at this from a computer science perspective, were we to hope to make a single scan of the brain at a particular point in time to be able to later run that configuration in some brain simulation model (a la "mind upload" as explored in a lot of sci-fi), I think we should consider _at least_ the following factors:
* Neuronal connectivity map (many-to-many graph of the dendritic/axonic connections between the neurons.
* Action potentials of each neuron or in other words the present state of ions in the cell (though I believe this information could be lost at the cost of subjective short term memory loss)
* Current cellular regulatory state of ion channels (various types) in the cell
* Current cellular regulatory state of the production of each of various neurotransmitters
* Current cellular regulatory state of the receptors of each of the various neurotransmitters and other related channels
This information would need to be gathered in all structures of the brain including the accessory cells (non-neurons) that don't participate in the overall map but do have a biochemical relationship to brain state. The information relating to the real-time neurochemical state of the cells could possibly be read from the epigenetic expression of the DNA in the nucleus, which may or may not be a simpler way to represent or read the information.
In addition to considering the computational requirements for storing and processing this information to run the model, we also must take into account the feasibly of capturing this information without disturbing or destroying the state. This is an idea I've been brewing for a while, hopefully I can try to take a stab at the calculations some day.
I think though the exercise would lead to similar conclusions as yours regarding the infeasibility of the proposal. This not even taking into account the role the senses and the body play on emotion and as such personality and behavior.
Love these kinds of analysis, these were the kinds of ideas running through my head while reading Reality+ and it's great to see them laid out in such great detail.
Thank you, glad you got something from it.
I agree with what you're saying about requirements. I did take a stab at estimating the static data for representing the mind. And that doesn't include the code/data necessary to run the model. You make some good points about the cellular machinery.
FWIW: https://logosconcarne.substack.com/p/digital-simulation
Exactly as you say, a connectome is necessary, and it turns out dominate the size. I went with 64-bit pointers and two-way many-to-many links. That's 500 trillion times 128 bits, and I guessed 64 bits would be the minimum to represent synapse type, state, etc. Possibly auxiliary tables for common data to reduce duplication. The hundred million neurons turn out to be a drop in the bucket in comparison. My model, a minimal one, was a hair over 12 petabytes (the hair because of the neuron states).
All of which makes me very dubious we live in a simulated reality. The computation requirements are just too formidable for such to be common.
>All of which makes me very dubious we live in a simulated reality. The computation requirements are just too formidable for such to be common.
From my perspective the simulation hypothesis should receive about the same skepticism as Solipsism. I think there is a huge lack of understanding of physics and computer science in the philosophy of mind community and as such it is difficult to dispel the myth. Hopefully efforts like yours can break down the edifice. I believe there is a way to prove it absolutely, especially when you consider the level of precision and consistency we can measure in physics from the LHC to the JWST.
The same skepticism as solipsism, yes, exactly. Good way to put it.
And I very much agree a lot of the philosophy community is behind the times when it comes to the physical sciences in general, let alone computing. They're steeped in outstanding thinkers from centuries ago, and those thinkers, in terms of physical reality, were massively ignorant. They don't deserve the unqualified reverence they're often given. (I think Berkeley and Hume were just plain wrong on some points.)
When I wrote my original comment I had this skeptical voice in the back of my head making the drop-in-the-bucket argument about the comparison between neuronal configurations and the connectome but I continue to think the informational complexity is being underestimated.
If we assume there are ~170 billion cells in the brain we are interested in (neurons + glial cells) and that we need to measure the expression of ~500 genes (rough estimate taking into account the genes for the multitude of neurotransmitters, their receptors/channels, and the associated regulatory genes) using the same 64 bits for each data point, we get an answer on the order of 10^15.
So I was almost certainly exaggerating by saying "orders of magnitude more..." but it would seem that at first glance the neuronal state _could be_ just as complex as the connectome. I tried searching to find any literature on this subject but was unable to, though my lack of familiarity with the correct technical terminology is often a hindrance.
I was at least able to find this article supporting the idea that this information is relevant to capturing aspects of human brain function and cognition:
https://pmc.ncbi.nlm.nih.gov/articles/PMC4969624/
Thank you for your feedback on my comments, it is nice to be able to discuss these ideas openly and receive respectful and constructive criticism.
I have no doubt my estimates are underestimates. The goal is only to find a minimal requirement — which turns out to be huge enough — with the acknowledgement that the reality would be much larger. As you mention, the glia cells are important. The myelin sheathing also matters in affecting the signal speed along axons. Neurons have an "on" and "off" state, but the on state consists of pulses, and the duty cycle of those pulses carries analog information. Some studies suggest the rise and fall times of the individual pulses carry information.
There's also the EM environment of the brain. It's strong enough that we can pick up brainwaves from outside the skull. I have to believe they play some role inside the skull. General blood chemistry probably also matters. Certainly, a few molecules of LSD or similar change cognition considerably.
Not an area I know much about, but my general sense is that genetics may not be the most fruitful path. The brain has a genetic template, but I would think the physical environment it grows in matters, and what a brain learns during a lifetime certainly changes it. Getting down to the genetic level is interesting, though. All cells would share the same genes.
I'm talking about measuring the epigeneitc markers (histone configuration, methylation etc) to understand the current regulatory state of the cell. Genes are literally physically stretched out or tagged in different ways that affect their expression and these markers or structures can change over time. When a neuron up regulates it's production of a particular neurotransmitter or it's associated receptors, there is an epigenetic change that occurs in the expression of the relevant genes. There are also other genes that control/regulate these epigenetic expressions. Much of this is still poorly understood and we have not the faintest idea I think on how to measure this in a living organism.
Some good reads:
The Gene by Siddhartha Mukherjee
Transformer by Nick Lane
Transformer is mostly about the Krebs cycle but it also talks a good deal about epigenetic gene expression and how it's dynamics effect cellular function.
Lotta interesting stuff. Ultimately, I think it can boil down to what can be functionally a blackbox and what details a simulation or recording has to have to produce results identical enough to qualify as a match. At the very least, we need good models of neurons and synapses and understanding details like you get into has to be part of that. Can’t make a black box without knowing what’s supposed to be inside. The question might be what details are necessary for understanding versus for implementation.
I’m also reminded of Brain Greene’s discussion of spacetime slicing in “The Fabric of the Cosmos” (and I’m sure many other books of his). Essentially Brian asks that we picture all of 4d spacetime as a loaf of bread that can be sliced into 3d chunks. Special Relativity tells us that there is more than one way to slice this loaf and depending on how we slice it, it will seem to tell different stories with respect to the order of events. While this doesn’t change the overall structure of space time, it does mean that things will appear to have a different ordering if we look at it from different perspectives.
I was drawn to this analogy in your discussion on the difficulty of creating distinct states in time with an asynchronous analog system like the brain. I believe the nature of these states doesn’t just relate to the frequency at which we are taking the snapshots but also the perspective from which we are doing it.
Taking the analogy into a hypothetical apparatus we can think of some scanning device that, let’s say, using an electromagnetic field measures in some sense the structural state of the brain at given moments via pulses it emits.
If this analogy is more than just a pretty metaphor, the states we measure would not look exactly the same depending on how we position the device and surely other particular implementation details. We might seem to see the order of firing taking place in the brain change depending on how we record events. This doesn’t change the overall fact that, given the arbitrarily “perfect” precision that we would supposedly need, in any case we will capture all of the relevant information regarding the brain’s activity.
This seems to imply that the supposed subjective experience of these sequential brain states could have various interpretations depending on the exact method we use to read the data which to me seems to be another logical contradiction leading us to refute the idea that conscious experience can be captured in frames of brain states. But maybe I am stretching the metaphor too much.
I think there would be no worries about simultaneity so long as both the brain and scanner were in the same inertial frame. There is also that causality is always respected regardless of relative velocity, so neuron A affecting neuron B is the same from all inertial frames. But the specific timestamps would depend on the frame.
*If* mind emerges from brain without quantum effects being needed (dubious in my mind), in principle (ha!) we should be able to read brain state without disturbing it too much. The difficulty is even getting a high-resolution complete snapshot of a living brain. In a lot of science fiction (and some science), it's dead brains scanned by slicing them. I don't think we even imagine the technology capable of scanning living brains. Nothing we have now comes close.
I think mentioning the ordering of firing was where I broke the metaphor, it's obvious to me now that given the speeds of the interactions in the brain we could in theory capture ordered states at some arbitrary degree of precision. I guess it's just I have an intuition that the character of those states, the exact values of all the 500 trillion numbers, would change depending on how we measure them.
The fact that they would produce different logically consistent stories for the same overall subjective experiences means that any schema of brain states doesn't carry some singular and consistent higher order property of "consciousness". It can still be a huge source of scientific data though that helps us understand how the brain works and if we could record "brain states" I'm sure we would learn more about our subjective experience.
Though I again am unsure how to detail this intuition and as is often the case for me and my neurotype, it is quite possible this one too should get chucked into the bin of delusions/confusions.
The 500 trillion numbers might change depending on the phase of the time slices or for any number of other reasons. I just don't think special relativity is a factor. If we assume consciousness emerges from a massively parallel network of causal events, it would be that causality we'd need to capture.
One aspect that occurs to me is that many forms of taking a snapshot (including most cameras) cannot process all that data at once. Shutter mechanisms can result in a sweep of the image over the film or sensor, and sensors must be read sequentially. So, timing is an issue in the sense that the requirement is an instantaneous snapshot of every aspect, and it's hard to envision any technology capable of that. In a SciFi novel, I'd techobabble about quantum entanglement between the brain and scanning system, but that doesn't actually work because every reading collapses the entanglement.
That shutter effect is exactly what I was trying to get at, thank you. Given the necessary limitations of our possible capturing mechanisms, I believe then that we could capture distinct sets of brain states that, while they do encompass the entirety of information necessary to represent the function of the brain, do not have the same exact information in each state. Another metaphor comes to mind of the different ways to unfold a cube. The point being there doesn't seem to be a 1-to-1 correspondence between a set of brain states and it's associated higher order configuration.
Yeah, that’s the sort of thing I was getting at in the Brain States part of the post. And as I emphasized, assuming these brain states are identifiable leads to assuming computationalism. So, they may be unicorns more than horses.
Heh. Speaking of unfolding a cube, the last animated video I did:
https://youtu.be/0-fx4skYu30
Imagine if someone said radios ARE music; if you build a radio, music will write itself ('emerge') and just start playing. That's what computationalism sounds like to me.
That’s a great way to put it! It’s almost cargo cult thinking.
Great article, thank you.
https://youtu.be/76P3hpDNQ6k?si=R6R0ObfOYVe1umuV is Raymond Tallis and tangentially relevant I hope, but a good watch too, in my view.
On your commentator's point on the "visualisation" of 4-space - a topologist I know who is way up there on knots, etc. kindly indulged a question from me on manifolds about how he "imagines" 4-space - it stuck with me as a useful (possibly only to me!) non-symbolic analogy - he thought of say a tesseract as I might a video of a cube over time, but all at the same moment. I hope that makes sense - natural everydaylanguage fails really.
Thanks, glad you liked it. That video was interesting. I was very much aligned with Tallis up to about the halfway mark when the conversation went off the rails a bit. Neither quite seemed to grasp what the other was saying and never reached for the difference between analog and digital systems. Tallis reached for it mentioning Shannon-Weaver, and he was basically talking about "information patternism" (another term for computationalism). I quite agree with him on that. But neither quite seemed to fully appreciate the physical requirements and analog nature of a brain scan.
OTOH, I thought the interviewer made a good point about history being encoded in our present state. Tallis didn't seem to accept that, but only hand-waved it away. (I don't respond well when people pick out words the way Tallis did with "encoding" and a few others. Seems more prevaricating to me.)
Neither seemed up on the technology. A shared VR for Mr. and Mrs. would of course be part of a computational immortality scenario as would an entirely user-customizable environment. Greg Egan has written a number of novels exploring that. They would both benefit from reading some of those. (Or my post! 😁)
Indeed! it's initially lucid but not sustained. Tallis is very smart but I don't think it maps well onto current discussions around some of the topics. I will now head off to read your essays on 4-space. Cheers, John.
I've spent some time thinking about 4D space as well. The tesseract is a particular fascination of mine. The image of a cube extended along the W dimension is very much like a short movie of a cube with the frames not extending in time but along the W axis.
FWIW:
https://logosconcarne.com/2016/08/17/hunting-tesseracti/
https://logosconcarne.com/2019/04/04/the-4th-dimension/
https://logosconcarne.com/2020/07/01/sideband-69-inside-a-tesseract/
Excellent article Wyrd.
I had to read the thing twice lol
Appreciate the raindrop analogy. Just as we can't truly capture the continuous, dynamic nature of rainfall by taking discrete snapshots, we can't reduce consciousness to a series of recorded states. It's not just a technical limitation I feel - it's a fundamental misunderstanding of what consciousness is.
What I found most interesting is the distinction between playback and computation. The Mandelbrot set for example makes this clear - displaying a pre-computed image is different from doing the actual computation. Similarly, playing back recorded brain states (even if we could somehow capture them) wouldn't recreate the actual experience of consciousness.
I think this gets at a deeper issue in AI and consciousness debates: we often mistake simulation for replication. Just because something can mimic the outputs of consciousness doesn't mean it has consciousness, just as a video of rain isn't actually making it rain. Just some thoughts.
Thank you, Neela, I'm glad you got something out of it. It sounds as if you not only have similar thoughts but have given this some thought. I think you're right on about simulation versus replication. I even see three possibilities: Emulation, Simulation, and Replication.
Emulation is a functional "blackbox" approach that goes after only the functionality of the mind irrespective of the physics of the brain. Simulation models the physics of the brain on the physicalist assumption that mind=physics. Replication models the physical structure of the brain, the physical neural network. I think the last might work, but I'm skeptical of the first two.
For reasons this whole series of posts tries to explain. It begins with this post:
https://logosconcarne.substack.com/p/brains-are-nothing-like-computers
Which lays the groundwork for the ones that follow (in "My Best Guess" newsletter) and explains more about what I see as the three options for machine minds.
And thank you for stopping by to comment!