Hello, fellow neuroscientist here! (End of my 3rd year in a PhD program. I do developmental research)
No notes, this was a solid break down!
It is extremely comical to me when media misunderstands the absurd intricacies of what they're describing especially when it comes to the brain. Johnny Depp's Transcendence movie is coming to mind.
I look forward to seeing more of your stuff and interacting with you in the future!
The other thing stories usually miss is the amount of data involved and the time it takes to transfer that data even at very high speeds. The brain contains at least petabytes of data regarding neurons, synapses, the interconnection map, the strength of the interconnections, and probably a lot more (glial cells, for instance, may be important to cogitation).
As an example, suppose we're talking 30 petabytes and a blazing speed of one terabyte per second. That's 30,000 seconds, which is over eight hours. And that's at one terabyte per second.
Yeah absolutely, it's a bonkers amount of data to be transferred, and an even crazier amount for a computer to hold in memory to simulate. The computational advances that would be needed would be enormous
I wouldn't bet against computer technology reaching the memory capacity (I'm old enough to be amazed that I have a 2 Terabyte hard drive -- I remember my first 5 Megabyte drive). But it is a formidable amount of data, especially if you try to transfer it.
So, I quite agree with your premise that mind uploading is far in our future (if it happens at all). I'm a big fan of Sawyer and loved Mindscan. And Upload is a fun show. But I think mind uploading is the last thing we'll solve when it comes to hardware minds. IF we ever solve it, it won't be cheap or easy.
FWIW, I'm skeptical conventional computers can simulate a *mind* -- the organ of the brain, yes, just as we can simulate the biological function of a heart or kidney. But I think it's an open question as to whether that digital simulation will think or have a mind. I agree with Penrose that consciousness doesn't seem algorithmic -- doesn't seem *computable* (some things aren't).
I agree the computer hardware is the easy part in all this. I also think it's far from trivial and the computational power required for it is far beyond present day capabilities.
I'm not such a big fan of Penrose on the mind (he's pretty far outside the academic mainstream in Phil of mind or neuroscience on this stuff). Personally I'm a functionalist when it comes to consciousness, under which simulation is definitely a possibility. I'll probably write posts exploring at some point.
I'll certainly read them with interest. I have a WordPress blog going back to 2011 and have written a lot of posts about computationalism over the years -- mostly expressing why I'm skeptical. I spent years debating a fellow blogger who was also a functionalist, so I'm fairly aware of the arguments. I don't know to what extent you're interested in debating the topic. Happy to let it go or have at it.
I'll say that, from a functionalist point of view, it's hard to say why a simulation wouldn't work. A key question is the granularity necessary. How much detail about synapse function must be included? Do glial cells need to be factored in? A cellular level sim seems inadequate, so how far down must it go? Molecular? Atomic? Quantum?
I think it boils down to whether consciousness is in the process or the result (the journey or the destination). A digital sim can capture the result, but I'm unsure it can capture the process.
FWIW, I don't see why a *physical* sim wouldn't work. Asimov's positronic brains replicated the structure of the brain. (In contrast to functionalism, I'm a structuralist. I think the physical structure is crucial.)
I'd definitely be happy to have the debate in the context of anything I post about it. If you have any posts of your own that you think make your case well, I would be interested in reading them -- they could become good fodder for my own posts 🙂
Okay, well, I’ll stay tuned for future posts. Been a while since I discussed it with someone, especially someone with training in a key related field. Looking forward to it.
Of the posts I’ve written about computationalism, this one from 2020 is possibly the most representative:
I've thought about writing about the future of specific technologies (AI, Brain machine interfaces), but not a general futurology post. That might be an interesting way of framing things for a future post. I'll have to see where things take me as I mull over possible topics and ideas.
Don't expect anything on that in the next month or so, I've got a backlog of written or in progress posts on other topics, but I definitely foresee myself coming back to "futurology" in some form or other.
Isn't the concept of a mind upload the same thing as a soul? I don't believe in souls and so if you were dying and had your brain uploaded into a robot, "you" would continue on in your old body and the robot would be a new person. This reflects my belief that "you" are a construct of your brain, and do not exist independently of your brain. So to use the soul analogy, you don't transfer your soul from one body to another, but rather a second soul is created that goes with the second body. And when your first brain dies, you end.
Usually souls are conceived of as a different (non-physical) "substance" that is the mind/consciousness.
Personally I'm inclined towards some sort of functionalism (https://iep.utm.edu/functism/)--the view that the mind is defined by the information processing functions it does. Therefore it doesn't matter how those functions are being performed (whether by a computer simulation of a brain or a brain), the mind is these functions being performed. Just like software can be abstracted away from the underlying hardware, so can mind be abstracted away from the underlying brain. It also means then minds are "multiply realizable"--in other words, they can be copied.
In that case, the soul is just the information. Both are immaterial.
I think of a mind as what an embodied brain does. So, if one did a "brain transplant" from one body to another it would be a different person because the brain would be in a different body with different intestinal flora and so forth. A mind in an inorganic body would be even more different.
When I was younger I used to think of mind transfer as a way to achieve rapid interstellar travel. The idea was you send a small AI robots to another star using light sails or many-stage fusion rockets. These build up a robot-run advanced technological civilization that includes mind transfer units. The you transfer a person's mind to the receiver/transfer until at the other star system at the speed of light. The body the mind came from is destroyed, and the "person" wakes up on the other planet.
That's interesting, I think the more common intuition is transplanting a brain keeps the person the same. I'm curious where you would draw the line--like, if a person just gets a change in gut flora, is that enough to make them a different person? Organ transplant? Multiple organ transplants, loss of limbs, plus change in gut flora? What about the gradual turnover of cells all over the body?
To be clear, I'm not arguing you're wrong -- personally, I don't think there is an answer and it's more of a personal preference what you consider "the same" person/mind. I don't think it's crazy to say someone has changed so much they are "not the same person" as they were 10 years ago, but curious how you think about it and draw that line.
I am not sure. I used to think it was all the brain, but more recently I've read stuff on how what goes on in other parts of the body, particularly the gut affect brains. It's all moot since I, like you, don't think mind transfer is feasible.
I agree whole brain emulation isn't going to happen anytime soon, absent absurdly rapid advances in AI (which I also think is extremely unlikely, but that's a separate discussion).
Many agree with you whole brain emulation/mind uploading won't be plausible anytime soon, but might be plausible in the distant future, perhaps 100-200 years from now. As a result, they think that preserving the structure of the brain when people (legally) die today might allow whole brain emulation in the distant future once our knowledge of the brain and the relevant technology has improved to a sufficient degree. For example, Ken Hayworth has made this argument: https://www.brainpreservation.org/content-2/killed-bad-philosophy/. I'm curious what you think of that.
First, about that particular article, it's worth pointing out he dramatically overstates the state of technology. To take one example, he talks about using glutaraldehyde to "glue" proteins in place to perfectly preserve the brain's molecular structure. Glutaraldehyde is used in electron microscopy, and is super useful for getting good images of biological specimens, but it doesn't do a perfect job. To cite a paper by the author Ken cites in the above article: "The usefulness of glutaraldehyde in high resolution electron microscopy is limited because chemical fixation inevitably causes chemical and structural alterations in the specimen." (from https://www.sciencedirect.com/science/article/abs/pii/0739626086900420)
This is a fundamental problem. Pumping a chemical through the entire brain that has the chemical property of fixing things in place and preserving proteins is going to cause physical alterations. Today that causes a problem in even just small-scale images, but the idea that it is inert enough to perfectly preserve every protein's state in the brain is wildly overstating what that chemical does. Maybe there's some possible solution to this that will be found in the future, but pretending we already have line of sight to it today mischaracterizes present state. Ken Hayworth obviously has a vested interest in convincing folks brain freezing is a plausible path to life extension.
Second, on the idea of freezing our brains: I'm skeptical. Freezing brains today isn't possible for long periods without, again, causing damage or changes to the underlying cells/proteins.
On the idea that in 100 or 200 years mind uploading will be possible: Maybe. I'm not really willing to place bets one way or the other on that kind of time scale. Like I said in the article, we might have nanotechnology that would be essentially magic compared to what we have today. That said I think there's a lot of very significant hurdles to overcome for mind uploads. The technological hurdles might prove to be insurmountable, there are just hard physical limits to some things. There are also a lot of cultural hurdles that would have to be overcome for there to be enough momentum and funding for the required research, and legal considerations of this being adopted as a medical procedure. So many things would have to change, but 2 centuries is a long time. I don't want to pretend I have a good idea of how it would (or wouldn't) happen.
Thank you so much for the thoughtful reply and for engaging in this discussion. A few thoughts in response:
Regarding the potential artifacts introduced by glutaraldehyde -- you raise a very good point. I think it is important for advocates of brain preservation to be more specific about exactly which molecular/higher-level cellular structures they believe need to be preserved and how confidently we can preserve them with current techniques vs. where there are still question marks. Personally, it seems to me that most of the molecular alterations induced by glutaraldehyde affect our ability to read out information from glutaraldehyde-preserved brains with various histologic techniques, such as immunohistochemistry: https://osf.io/preprints/8zd4e. "Unknown unknowns" is of course also a very reasonable concern, given that it hasn't been done yet. I'm just wondering if you are thinking of a more specific critique.
On Ken Hayworth "obviously" having a "vested interest" -- I may have misunderstood, but is there a financial incentive on Ken's part that I'm missing?
Lastly, I am with you on the long timelines and huge uncertainties involved. I agree wholeheartedly that placing confident bets on the state of neuroscience and computing 100-200 years from now is a fool's errand. Still, given the stakes involved, I believe we have an ethical obligation to offer people the opportunity to preserve the information in their brain at the time of their death, if they desire this, on the highly uncertain chance that it does happen to be possible in the future. Unfortunately, people can't currently consistently access this procedure in a high quality manner after their death, for a variety of reasons, including medicolegal barriers such as involuntary autopsy that cannot be opted out of, and the potential for familial interference. I personally think that's problematic, because I think people should be allowed to exercise their body autonomy in this way. I'm curious to hear if you agree.
Thanks again for the discussion and for sharing your expertise.
I think you hit the nail on the head with pointing out it all depends on what structures need to be preserved. Maybe there's a good argument for why the type of damage the chemicals he's advocating for doesn't matter. But it's not true we can perfectly preserve a brain today. Reading other things Hayworth has written, I think the idea is we could reproduce a connectome in the future, but given we can't currently go from connectome to simulation in much simpler organisms now, there's a lot more that needs to be fleshed out there. I think they're betting that improved neuroscience understanding will let us paper over a lot of this and build a simulation with imperfect information. Maybe that's right! I just don't think it's a given.
As far as the vested interest comment, Hayworth is president of the Brain Preservation Foundation. I don't know if he's paid for that, but certainly it's in his interests to present a rosey picture to donors. I don't think he's lying or being dishonest, I think he's a true believer, but he is incentivized to present a rosier and simpler view of the matter IMO. To his credit, it's not like he's a weird fringe element, he does real science and his foundation has been involved in real improvements in preservation techniques. But he's also not exactly espousing the scientific consensus.
In terms of the ethics: I don't know enough about the circumstances in which your wishes would be overridden and I can't say it's an issue I've thought too deeply about
This was the very first trope I took on when I started writing for IGMS. Almost ten years later, and we're still super-impressed by 1mm^3 of just structural data, no activity at all.
I'm still very interested in Hofstadter's biological upload, the copy / model of his wife he carries around in his own brain. My mom just died this past winter, and while I lived at home my model might have been that detailed, it has degraded / abstracted over the past 30+ years.
I'm a fan of Hofstadter's view as well, the idea that our model of loved ones is (from a functionalist perspective) a version of their mind. I'm not confident these models are high enough fidelity to really be thought of as a mind even when those models are more detailed, but it's an interesting idea.
I'm sorry to hear about your mom's passing. I hope you're doing well.
I'm sure most of them are not, either. That was actually my point, that the constant and intense engagement is probably the key, like the way novelists talk about their characters "making their own decisions."
Great article! It raises some fascinating questions about consciousness. Mind uploading might be theoretically possible (although not practically achievable with our current technology and understanding), but only under certain theories of consciousness, like functionalism. Functionalism is currently the predominant view among neuroscientists and computer scientists. It's intriguing to consider how science and art tend to converge on similar themes. I wonder how our science fiction might differ if another theory of consciousness, such as panpsychism or identity theory, were the standard view instead.
I'm very much looking forward to reading more of your work.
The 100 shows mind uploads. Though acting isn't all that, the script is.
”Playing with these ideas is a fun and challenging way to pose questions about what it is we value—if people are ends in themselves, what happens when we can copy them (or change them) at will?"
Who would be having access to these eluded copies I wonder? Considering human desire for power and gain throughout the history, these "fun" ways
of thinking would simply be dangerous. If I need security in life to protect my body, now I have to worry about my copy laying into some lair?! The show I mention above deals with snatching bodies, murdering others to extend your physical life. Talk about value... . Hopefully we will satisfy the wonder to know while in the flesh. It isn't all that complicated, rather much simplier. Maybe this is where it hides!
Transfer, a french movie deals with uploads and yep, you guessed it, aquired by violence, consent override. Travelers, all four seasons too. For those who have watched it, Vincent resident one had to jump into a woman's body to hide from the Director, AI. Loved what The Director was capable of doing.
I find no correlation between shows, movies, fiction in general and these concepts becoming reality. But I do know human nature, and I draw from there. Nothing good would happen in the what if scenario. Keep wondering.
I was debating my brother on this same topic. He thinks that with Neuralink we are close to uploading knowledge to the brain like in the matrix. I told him at best we have a better video game controller and no one will ever be able to read or write intelligible information to the brain directly. We are chaotically made and individually experienced.
Your article reads almost exactly like my arguments. A pleasure to read from someone who knows the difference between what can be and what is written by some English majors in Southern California.
I found your Substack recently and really enjoy the content! Similar to you, mind-uploading/brain emulation is among my favourite Sci-fi and philosophy topics.
I think you did a great job highlighting the pragmatic scientific implausibility of mind-uploading any time soon. However, I also think the 4E cognitive science literature (embodied/embedded/enactive/extended cognition) raises critical objections to the possibility of brain emulation as well. I’m short, we cannot just simulate brains in and of themselves because mind arises through dynamic interactions with body and both physical and social environments as well. Hence, the idea would be that you cannot emulate a brain in abstraction from the rest of the world.
Would be very interested to read a future post on your thoughts on 4E cognition and/or the related topics of active inference/predictive processing.
To me, it’s obvious that, however it works, consciousness is an emergent property of our material beings. I don’t care what kind of scanner or teleporter you use, MY experience of consciousness is not magically going to teleport with the (re)creation of my material self - that will just be a copy, understandably feeling like it is totally the old me.
How the thought experiment about forgetting to delete the old copy doesn’t make this logically obvious is a mystery to me. What am I not understanding?
Is it just the matter that seems important to you? Why would the atoms that make you up be special? They're all slowly replaced over the course of years anyways but you still think your self from a few years ago is the same as you. Atoms are indistinguishable in terms of their properties from other atoms, so it's not clear what the mechanism would be for the matter you're made of to matter.
That's a different question. One must be pedantic in these thought experiments.
Regarding having one's atoms replaced slowly, I'm not sure my experience of consciousness would alter depending on how many of them were replaced exactly how quickly and exactly where. I'd rather they weren't.
I'd say it's totally possible we wake up every day, ever so slightly altered, with the conviction that we are who we always were. But maybe the old self died - how would one tell the difference? They're gone. But I think we can agree that a centenarian feels quite different than his much younger self did.
Regarding being copied, uploaded or teleported, well - as the thought experiment IMO proves, things get a bit hairy if you forget to delete the original, don't they? You can't be in two places at once.
I'm trying to get at the implications of your proposed solution, like that you're a different person than you were previously (totally fine to accept it, just pointing out it's an implication many would find counter-intuitive), that it implies some unobserved property of atoms (extremely unlikely in my view). Based on what you've said here I'm also not clear on how it avoids the problems you're trying to avoid--what happens if a teleporter made two of you, but used half your matter for one and half for the other. If it's the matter that matters (ha) you're now in two places. That seems worse than the idea that the self can "branch" and create two people with equal claim to being the person who stepped into the teleporter.
Really cool article Tommy. It's a fascinating area, no doubt with a lot of philosophical discussion around it. I'm also fascinated by the brain imaging field and how it's progressing.
I did want to make one comment on your point about "simulating a rainstorm and expecting it to get wet". When people say this, the distinction is that on current computers they rely on a Von Neumann architecture and so everything you try to recreate ultimately gets boiled down to 1's and 0's. This is a big issue. Nature doesn't work like this, so that's why the comment about simulating v actual is valid - i.e. we need the AI system to be analogue based using real physics.
However, this doesn't mean it isn't 'theoretically' possible using neuromorphic chips. If we can design these chips in a way that uses actual physics via analogue mechamisms, rather than a 1's and 0's abstraction, then you open up the possibility of something being conscious. Otherwise there will always be a considerable layer of 'introduced information' inputted by humans in the design of the system, which constrains it's agency.
Great reflection and pulling together of how immense of a task it would be to map and download a brain. In some ways I am comforted by the idea that it is so difficult. And on the other side — I’m one of those people who think up whacky stories about uploading brains and interacting with singularity “machines”.
What do you think needs to happen before nanobots can actually interact and work with biological tissues?
Hey, I write stories like that too! I think the concepts are super interesting and rich, they're just implausible in practice.
In terms of nanobots working on humans -- I'm no expert on nanotech, so I'm not sure, but I don't think we're particularly close. More importantly for our purposes, I think there's a big leap from "preprogrammed nanobots capable of recognizing a particular kind of cell/tissue and injecting something" to "orchestrated swarm of nanobots capable of mapping out every cell and connection in a large organ and then record electrical activity from it all syncronously".
Hello, fellow neuroscientist here! (End of my 3rd year in a PhD program. I do developmental research)
No notes, this was a solid break down!
It is extremely comical to me when media misunderstands the absurd intricacies of what they're describing especially when it comes to the brain. Johnny Depp's Transcendence movie is coming to mind.
I look forward to seeing more of your stuff and interacting with you in the future!
Thank you! I look forward to seeing you around as well!
The other thing stories usually miss is the amount of data involved and the time it takes to transfer that data even at very high speeds. The brain contains at least petabytes of data regarding neurons, synapses, the interconnection map, the strength of the interconnections, and probably a lot more (glial cells, for instance, may be important to cogitation).
As an example, suppose we're talking 30 petabytes and a blazing speed of one terabyte per second. That's 30,000 seconds, which is over eight hours. And that's at one terabyte per second.
Yeah absolutely, it's a bonkers amount of data to be transferred, and an even crazier amount for a computer to hold in memory to simulate. The computational advances that would be needed would be enormous
I wouldn't bet against computer technology reaching the memory capacity (I'm old enough to be amazed that I have a 2 Terabyte hard drive -- I remember my first 5 Megabyte drive). But it is a formidable amount of data, especially if you try to transfer it.
So, I quite agree with your premise that mind uploading is far in our future (if it happens at all). I'm a big fan of Sawyer and loved Mindscan. And Upload is a fun show. But I think mind uploading is the last thing we'll solve when it comes to hardware minds. IF we ever solve it, it won't be cheap or easy.
FWIW, I'm skeptical conventional computers can simulate a *mind* -- the organ of the brain, yes, just as we can simulate the biological function of a heart or kidney. But I think it's an open question as to whether that digital simulation will think or have a mind. I agree with Penrose that consciousness doesn't seem algorithmic -- doesn't seem *computable* (some things aren't).
I agree the computer hardware is the easy part in all this. I also think it's far from trivial and the computational power required for it is far beyond present day capabilities.
I'm not such a big fan of Penrose on the mind (he's pretty far outside the academic mainstream in Phil of mind or neuroscience on this stuff). Personally I'm a functionalist when it comes to consciousness, under which simulation is definitely a possibility. I'll probably write posts exploring at some point.
I'll certainly read them with interest. I have a WordPress blog going back to 2011 and have written a lot of posts about computationalism over the years -- mostly expressing why I'm skeptical. I spent years debating a fellow blogger who was also a functionalist, so I'm fairly aware of the arguments. I don't know to what extent you're interested in debating the topic. Happy to let it go or have at it.
I'll say that, from a functionalist point of view, it's hard to say why a simulation wouldn't work. A key question is the granularity necessary. How much detail about synapse function must be included? Do glial cells need to be factored in? A cellular level sim seems inadequate, so how far down must it go? Molecular? Atomic? Quantum?
I think it boils down to whether consciousness is in the process or the result (the journey or the destination). A digital sim can capture the result, but I'm unsure it can capture the process.
FWIW, I don't see why a *physical* sim wouldn't work. Asimov's positronic brains replicated the structure of the brain. (In contrast to functionalism, I'm a structuralist. I think the physical structure is crucial.)
I'd definitely be happy to have the debate in the context of anything I post about it. If you have any posts of your own that you think make your case well, I would be interested in reading them -- they could become good fodder for my own posts 🙂
Okay, well, I’ll stay tuned for future posts. Been a while since I discussed it with someone, especially someone with training in a key related field. Looking forward to it.
Of the posts I’ve written about computationalism, this one from 2020 is possibly the most representative:
https://logosconcarne.com/2020/06/15/strong-computationalism/
Or you can use this link for a list you can choose from:
https://logosconcarne.com/tag/computationalism/
That's a brilliant essay. A little suggestion: would you consider writing another one on the topic of, this time, plausible futurology?
Thanks!
I would love to write about plausible futurology.
I've thought about writing about the future of specific technologies (AI, Brain machine interfaces), but not a general futurology post. That might be an interesting way of framing things for a future post. I'll have to see where things take me as I mull over possible topics and ideas.
Don't expect anything on that in the next month or so, I've got a backlog of written or in progress posts on other topics, but I definitely foresee myself coming back to "futurology" in some form or other.
Isn't the concept of a mind upload the same thing as a soul? I don't believe in souls and so if you were dying and had your brain uploaded into a robot, "you" would continue on in your old body and the robot would be a new person. This reflects my belief that "you" are a construct of your brain, and do not exist independently of your brain. So to use the soul analogy, you don't transfer your soul from one body to another, but rather a second soul is created that goes with the second body. And when your first brain dies, you end.
I don't think mind uploads are the same as souls!
Usually souls are conceived of as a different (non-physical) "substance" that is the mind/consciousness.
Personally I'm inclined towards some sort of functionalism (https://iep.utm.edu/functism/)--the view that the mind is defined by the information processing functions it does. Therefore it doesn't matter how those functions are being performed (whether by a computer simulation of a brain or a brain), the mind is these functions being performed. Just like software can be abstracted away from the underlying hardware, so can mind be abstracted away from the underlying brain. It also means then minds are "multiply realizable"--in other words, they can be copied.
In that case, the soul is just the information. Both are immaterial.
I think of a mind as what an embodied brain does. So, if one did a "brain transplant" from one body to another it would be a different person because the brain would be in a different body with different intestinal flora and so forth. A mind in an inorganic body would be even more different.
When I was younger I used to think of mind transfer as a way to achieve rapid interstellar travel. The idea was you send a small AI robots to another star using light sails or many-stage fusion rockets. These build up a robot-run advanced technological civilization that includes mind transfer units. The you transfer a person's mind to the receiver/transfer until at the other star system at the speed of light. The body the mind came from is destroyed, and the "person" wakes up on the other planet.
That's interesting, I think the more common intuition is transplanting a brain keeps the person the same. I'm curious where you would draw the line--like, if a person just gets a change in gut flora, is that enough to make them a different person? Organ transplant? Multiple organ transplants, loss of limbs, plus change in gut flora? What about the gradual turnover of cells all over the body?
To be clear, I'm not arguing you're wrong -- personally, I don't think there is an answer and it's more of a personal preference what you consider "the same" person/mind. I don't think it's crazy to say someone has changed so much they are "not the same person" as they were 10 years ago, but curious how you think about it and draw that line.
I am not sure. I used to think it was all the brain, but more recently I've read stuff on how what goes on in other parts of the body, particularly the gut affect brains. It's all moot since I, like you, don't think mind transfer is feasible.
I agree whole brain emulation isn't going to happen anytime soon, absent absurdly rapid advances in AI (which I also think is extremely unlikely, but that's a separate discussion).
Many agree with you whole brain emulation/mind uploading won't be plausible anytime soon, but might be plausible in the distant future, perhaps 100-200 years from now. As a result, they think that preserving the structure of the brain when people (legally) die today might allow whole brain emulation in the distant future once our knowledge of the brain and the relevant technology has improved to a sufficient degree. For example, Ken Hayworth has made this argument: https://www.brainpreservation.org/content-2/killed-bad-philosophy/. I'm curious what you think of that.
I have a few thoughts.
First, about that particular article, it's worth pointing out he dramatically overstates the state of technology. To take one example, he talks about using glutaraldehyde to "glue" proteins in place to perfectly preserve the brain's molecular structure. Glutaraldehyde is used in electron microscopy, and is super useful for getting good images of biological specimens, but it doesn't do a perfect job. To cite a paper by the author Ken cites in the above article: "The usefulness of glutaraldehyde in high resolution electron microscopy is limited because chemical fixation inevitably causes chemical and structural alterations in the specimen." (from https://www.sciencedirect.com/science/article/abs/pii/0739626086900420)
This is a fundamental problem. Pumping a chemical through the entire brain that has the chemical property of fixing things in place and preserving proteins is going to cause physical alterations. Today that causes a problem in even just small-scale images, but the idea that it is inert enough to perfectly preserve every protein's state in the brain is wildly overstating what that chemical does. Maybe there's some possible solution to this that will be found in the future, but pretending we already have line of sight to it today mischaracterizes present state. Ken Hayworth obviously has a vested interest in convincing folks brain freezing is a plausible path to life extension.
Second, on the idea of freezing our brains: I'm skeptical. Freezing brains today isn't possible for long periods without, again, causing damage or changes to the underlying cells/proteins.
On the idea that in 100 or 200 years mind uploading will be possible: Maybe. I'm not really willing to place bets one way or the other on that kind of time scale. Like I said in the article, we might have nanotechnology that would be essentially magic compared to what we have today. That said I think there's a lot of very significant hurdles to overcome for mind uploads. The technological hurdles might prove to be insurmountable, there are just hard physical limits to some things. There are also a lot of cultural hurdles that would have to be overcome for there to be enough momentum and funding for the required research, and legal considerations of this being adopted as a medical procedure. So many things would have to change, but 2 centuries is a long time. I don't want to pretend I have a good idea of how it would (or wouldn't) happen.
Thank you so much for the thoughtful reply and for engaging in this discussion. A few thoughts in response:
Regarding the potential artifacts introduced by glutaraldehyde -- you raise a very good point. I think it is important for advocates of brain preservation to be more specific about exactly which molecular/higher-level cellular structures they believe need to be preserved and how confidently we can preserve them with current techniques vs. where there are still question marks. Personally, it seems to me that most of the molecular alterations induced by glutaraldehyde affect our ability to read out information from glutaraldehyde-preserved brains with various histologic techniques, such as immunohistochemistry: https://osf.io/preprints/8zd4e. "Unknown unknowns" is of course also a very reasonable concern, given that it hasn't been done yet. I'm just wondering if you are thinking of a more specific critique.
On Ken Hayworth "obviously" having a "vested interest" -- I may have misunderstood, but is there a financial incentive on Ken's part that I'm missing?
Lastly, I am with you on the long timelines and huge uncertainties involved. I agree wholeheartedly that placing confident bets on the state of neuroscience and computing 100-200 years from now is a fool's errand. Still, given the stakes involved, I believe we have an ethical obligation to offer people the opportunity to preserve the information in their brain at the time of their death, if they desire this, on the highly uncertain chance that it does happen to be possible in the future. Unfortunately, people can't currently consistently access this procedure in a high quality manner after their death, for a variety of reasons, including medicolegal barriers such as involuntary autopsy that cannot be opted out of, and the potential for familial interference. I personally think that's problematic, because I think people should be allowed to exercise their body autonomy in this way. I'm curious to hear if you agree.
Thanks again for the discussion and for sharing your expertise.
I think you hit the nail on the head with pointing out it all depends on what structures need to be preserved. Maybe there's a good argument for why the type of damage the chemicals he's advocating for doesn't matter. But it's not true we can perfectly preserve a brain today. Reading other things Hayworth has written, I think the idea is we could reproduce a connectome in the future, but given we can't currently go from connectome to simulation in much simpler organisms now, there's a lot more that needs to be fleshed out there. I think they're betting that improved neuroscience understanding will let us paper over a lot of this and build a simulation with imperfect information. Maybe that's right! I just don't think it's a given.
As far as the vested interest comment, Hayworth is president of the Brain Preservation Foundation. I don't know if he's paid for that, but certainly it's in his interests to present a rosey picture to donors. I don't think he's lying or being dishonest, I think he's a true believer, but he is incentivized to present a rosier and simpler view of the matter IMO. To his credit, it's not like he's a weird fringe element, he does real science and his foundation has been involved in real improvements in preservation techniques. But he's also not exactly espousing the scientific consensus.
In terms of the ethics: I don't know enough about the circumstances in which your wishes would be overridden and I can't say it's an issue I've thought too deeply about
That is really cool. Thank you for sharing. I'll read the other posts soon. I've just found Cognitive Wonderland ! And I'm so happy about it :-)
This was the very first trope I took on when I started writing for IGMS. Almost ten years later, and we're still super-impressed by 1mm^3 of just structural data, no activity at all.
http://www.intergalacticmedicineshow.com/cgi-bin/mag.cgi?do=columns&vol=randall_hayes&article=001
I'm still very interested in Hofstadter's biological upload, the copy / model of his wife he carries around in his own brain. My mom just died this past winter, and while I lived at home my model might have been that detailed, it has degraded / abstracted over the past 30+ years.
I'm a fan of Hofstadter's view as well, the idea that our model of loved ones is (from a functionalist perspective) a version of their mind. I'm not confident these models are high enough fidelity to really be thought of as a mind even when those models are more detailed, but it's an interesting idea.
I'm sorry to hear about your mom's passing. I hope you're doing well.
I'm sure most of them are not, either. That was actually my point, that the constant and intense engagement is probably the key, like the way novelists talk about their characters "making their own decisions."
Great article! It raises some fascinating questions about consciousness. Mind uploading might be theoretically possible (although not practically achievable with our current technology and understanding), but only under certain theories of consciousness, like functionalism. Functionalism is currently the predominant view among neuroscientists and computer scientists. It's intriguing to consider how science and art tend to converge on similar themes. I wonder how our science fiction might differ if another theory of consciousness, such as panpsychism or identity theory, were the standard view instead.
I'm very much looking forward to reading more of your work.
Thanks for the piece Tommy. Looking forward to more.
Thank you!
The 100 shows mind uploads. Though acting isn't all that, the script is.
”Playing with these ideas is a fun and challenging way to pose questions about what it is we value—if people are ends in themselves, what happens when we can copy them (or change them) at will?"
Who would be having access to these eluded copies I wonder? Considering human desire for power and gain throughout the history, these "fun" ways
of thinking would simply be dangerous. If I need security in life to protect my body, now I have to worry about my copy laying into some lair?! The show I mention above deals with snatching bodies, murdering others to extend your physical life. Talk about value... . Hopefully we will satisfy the wonder to know while in the flesh. It isn't all that complicated, rather much simplier. Maybe this is where it hides!
Transfer, a french movie deals with uploads and yep, you guessed it, aquired by violence, consent override. Travelers, all four seasons too. For those who have watched it, Vincent resident one had to jump into a woman's body to hide from the Director, AI. Loved what The Director was capable of doing.
I find no correlation between shows, movies, fiction in general and these concepts becoming reality. But I do know human nature, and I draw from there. Nothing good would happen in the what if scenario. Keep wondering.
I was debating my brother on this same topic. He thinks that with Neuralink we are close to uploading knowledge to the brain like in the matrix. I told him at best we have a better video game controller and no one will ever be able to read or write intelligible information to the brain directly. We are chaotically made and individually experienced.
Your article reads almost exactly like my arguments. A pleasure to read from someone who knows the difference between what can be and what is written by some English majors in Southern California.
Thank you!
I found your Substack recently and really enjoy the content! Similar to you, mind-uploading/brain emulation is among my favourite Sci-fi and philosophy topics.
I think you did a great job highlighting the pragmatic scientific implausibility of mind-uploading any time soon. However, I also think the 4E cognitive science literature (embodied/embedded/enactive/extended cognition) raises critical objections to the possibility of brain emulation as well. I’m short, we cannot just simulate brains in and of themselves because mind arises through dynamic interactions with body and both physical and social environments as well. Hence, the idea would be that you cannot emulate a brain in abstraction from the rest of the world.
Would be very interested to read a future post on your thoughts on 4E cognition and/or the related topics of active inference/predictive processing.
To me, it’s obvious that, however it works, consciousness is an emergent property of our material beings. I don’t care what kind of scanner or teleporter you use, MY experience of consciousness is not magically going to teleport with the (re)creation of my material self - that will just be a copy, understandably feeling like it is totally the old me.
How the thought experiment about forgetting to delete the old copy doesn’t make this logically obvious is a mystery to me. What am I not understanding?
Is it just the matter that seems important to you? Why would the atoms that make you up be special? They're all slowly replaced over the course of years anyways but you still think your self from a few years ago is the same as you. Atoms are indistinguishable in terms of their properties from other atoms, so it's not clear what the mechanism would be for the matter you're made of to matter.
That's a different question. One must be pedantic in these thought experiments.
Regarding having one's atoms replaced slowly, I'm not sure my experience of consciousness would alter depending on how many of them were replaced exactly how quickly and exactly where. I'd rather they weren't.
I'd say it's totally possible we wake up every day, ever so slightly altered, with the conviction that we are who we always were. But maybe the old self died - how would one tell the difference? They're gone. But I think we can agree that a centenarian feels quite different than his much younger self did.
Regarding being copied, uploaded or teleported, well - as the thought experiment IMO proves, things get a bit hairy if you forget to delete the original, don't they? You can't be in two places at once.
I'm trying to get at the implications of your proposed solution, like that you're a different person than you were previously (totally fine to accept it, just pointing out it's an implication many would find counter-intuitive), that it implies some unobserved property of atoms (extremely unlikely in my view). Based on what you've said here I'm also not clear on how it avoids the problems you're trying to avoid--what happens if a teleporter made two of you, but used half your matter for one and half for the other. If it's the matter that matters (ha) you're now in two places. That seems worse than the idea that the self can "branch" and create two people with equal claim to being the person who stepped into the teleporter.
Really cool article Tommy. It's a fascinating area, no doubt with a lot of philosophical discussion around it. I'm also fascinated by the brain imaging field and how it's progressing.
I did want to make one comment on your point about "simulating a rainstorm and expecting it to get wet". When people say this, the distinction is that on current computers they rely on a Von Neumann architecture and so everything you try to recreate ultimately gets boiled down to 1's and 0's. This is a big issue. Nature doesn't work like this, so that's why the comment about simulating v actual is valid - i.e. we need the AI system to be analogue based using real physics.
However, this doesn't mean it isn't 'theoretically' possible using neuromorphic chips. If we can design these chips in a way that uses actual physics via analogue mechamisms, rather than a 1's and 0's abstraction, then you open up the possibility of something being conscious. Otherwise there will always be a considerable layer of 'introduced information' inputted by humans in the design of the system, which constrains it's agency.
Thanks again for the post. Keen to see more :)
Nice! I know too many people parroting, "it's AI it's AI !" Great article!
Great reflection and pulling together of how immense of a task it would be to map and download a brain. In some ways I am comforted by the idea that it is so difficult. And on the other side — I’m one of those people who think up whacky stories about uploading brains and interacting with singularity “machines”.
What do you think needs to happen before nanobots can actually interact and work with biological tissues?
Hey, I write stories like that too! I think the concepts are super interesting and rich, they're just implausible in practice.
In terms of nanobots working on humans -- I'm no expert on nanotech, so I'm not sure, but I don't think we're particularly close. More importantly for our purposes, I think there's a big leap from "preprogrammed nanobots capable of recognizing a particular kind of cell/tissue and injecting something" to "orchestrated swarm of nanobots capable of mapping out every cell and connection in a large organ and then record electrical activity from it all syncronously".