I watched a very long podcast episode with John Smart and Max Marty
https://www.youtube.com/watch?v=rBhjRzxj6eY
John was diving deep into some topics that most people would struggle with but that he clearly thinks about regularly. Max spent a lot of time trying to pin down how John viewed duplicate minds. This is a source of angst for many people and Max couldn't quite accept that a duplicate would really be him. So I'm going to rephrase it based on some of John's ideas mixed with how I regularly think about it.
First of all, if you preserve memories and then duplicate them on other hardware, it's much more than just recorded events from your past. When we talk about memories, we're talking about memories that are episodic, semantic, implicit, emotional, procedural, perceptual, habitual, motivational, interpretive, heuristic, cognitive, somatic, reflexive, social, cultural, identity, contextual, and many other kinds. Max actually accepted that the duplicate would think that it was him, which is a leap that many people can make if they think about it for long enough. Max estimated that 50% of people can accept that.
To finish the thought experiment, you just have to envision a world where we routinely make duplicates for various reasons. Physics says that it should be possible, so it will almost certainly happen. Duplicates will be created for revivals, backups, transportation, multitasking, etc. Now, in such a world where everyone does it and has been doing it for a long time, you will basically get used to the idea. It will be so useful and so enriching that your angst will fall away. This has happened many times in history. People quickly got used to being flung through the air in an aluminum can, not because anyone from the 1800s would have thought that was a good idea, but because it was useful and valuable. We don't cut people wide open and replace organs because we naturally enjoy it, but because it's extremely useful and valuable. It solves problems that cannot be solved any other way. As John was trying to explain, those few who choose not to use duplicates will quickly fall behind like the Amish.
The transition to using duplicates won't happen quickly. It will be a very gradual process that never really has an endpoint. So you could also imagine some intermediate stages. I'll pick a "halfway" point as a scenario. Imagine that we all have complex brain implants. We would do much of our thinking in external hardware which we would find very useful of course. The implants would have been gradually getting better for decades, but we would still die of old age. Many of our memories would be duplicated in this hardware because the memories would get gradually duplicated as we used the implants on a daily basis. Having our memories duplicated in hardware would be very useful because they would get retained permanently instead of fading. They would also be more accessible, better organized, and faster. We would also have lots of memories that would have been shared by loved ones or downloaded from others. It would get to the point where we would have far more memories in hardware than in our biological brains. The definition of memory in this context is extremely broad. In this world where we are halfway to being able to make a duplicate, we would already have most of our memories outside of our biological brains. People in a society like that would have much less resistance to the idea of making a duplicate.
Physics says that both the original and the duplicate are actually the same mind. Resistance to the idea of a duplicate is just another form of vitalism, believing that there is a "life force" distinct from physical laws. People will need to adapt to reality if they want to move forward with the rest of humanity. Many people currently have trouble with this, but as it becomes possible, duplicates will simply be a non-issue because we will all happily go along with it.
Memories
-
PCmorphy72
- Posts: 32
- Joined: Sun May 26, 2019 12:39 pm
Re: Memories
Moving from a related thread, “suspending” the well-addressed fibroblast/cloning argument (and even the durable digital archives argument too, though not yet addressed at all – by the way, Andy, did you take a look at the Wikipedia M-DISC page? At the moment I have only 45 MB on the SecurSafe free space…) I admit I didn’t read the recent paper “What are memories made of? A survey of neuroscientists on the structural basis of long-term memory” (proudly, at least one of its authors is here in this forum), since on the page https://www.sparksbrain.org/ourPreserva ... roach.html I had read only statements such as:
“While immersion is substantially slower than perfusion to preserve the brain, some studies have shown that it can still achieve good preservation quality, particularly in cortical regions, which are believed to be important areas for long-term memory storage.” (implicit optimism)
“While we believe our current protocols are the best available practices and that they may be sufficient to preserve the information encoding long-term memories and personal identity, we clearly cannot claim 100% certainty about this.” (explicit optimism)
What I read some days ago (from other sources) was that aldehyde fixation in the brain preserves micro architecture (neurites, synapses, spine morphology) with excellent fidelity and stability, which is valuable for future connectome tracing, but also that even optimistic scenarios face losses in epigenetic marks, long range phasing, structural variants, protein DNA interaction maps, and, more importantly, degradation of the “parameterization” needed to infer original “synaptic weights”: precisely the layers that help infer “cell type specific expression and synaptic protein repertoires.”
Long term memory relies on synaptic plasticity distributed across specific circuits: those degradations suggest that — even under optimistic assumptions — the molecular substrates of synaptic plasticity are not faithfully preserved. The structural map may remain highly legible, but the dynamic functional code of long-term memory embedded in molecular details (engrams, synaptic weights, plasticity histories) risks being scrambled beyond reliable recovery. In practical terms, this means that most of the fine-grained information encoding personal memories could be lost, leaving only probabilistic reconstructions (with lower confidence in original synaptic weights) rather than authentic retrieval.
Trying to be more technical, I read that long term memory depends on molecular parameters which are relevant to synaptic weights, such as receptor composition, scaffold configuration, PTMs, and local RNA/proteostasis, which fixation degrades or disrupts. Specifically:
In any case there is a common boundary condition: it’s reasonable to expect future methods to extract more from fixed brains than architecture alone (multi omics, modeling), but their success probabilities still hinge on the fidelity of preserved molecular parameters; without reliable molecular states, future decoding reduces to probabilistic inference of synaptic weights, with lower confidence per engram.
Scientific Papers:
“While immersion is substantially slower than perfusion to preserve the brain, some studies have shown that it can still achieve good preservation quality, particularly in cortical regions, which are believed to be important areas for long-term memory storage.” (implicit optimism)
“While we believe our current protocols are the best available practices and that they may be sufficient to preserve the information encoding long-term memories and personal identity, we clearly cannot claim 100% certainty about this.” (explicit optimism)
What I read some days ago (from other sources) was that aldehyde fixation in the brain preserves micro architecture (neurites, synapses, spine morphology) with excellent fidelity and stability, which is valuable for future connectome tracing, but also that even optimistic scenarios face losses in epigenetic marks, long range phasing, structural variants, protein DNA interaction maps, and, more importantly, degradation of the “parameterization” needed to infer original “synaptic weights”: precisely the layers that help infer “cell type specific expression and synaptic protein repertoires.”
Long term memory relies on synaptic plasticity distributed across specific circuits: those degradations suggest that — even under optimistic assumptions — the molecular substrates of synaptic plasticity are not faithfully preserved. The structural map may remain highly legible, but the dynamic functional code of long-term memory embedded in molecular details (engrams, synaptic weights, plasticity histories) risks being scrambled beyond reliable recovery. In practical terms, this means that most of the fine-grained information encoding personal memories could be lost, leaving only probabilistic reconstructions (with lower confidence in original synaptic weights) rather than authentic retrieval.
Trying to be more technical, I read that long term memory depends on molecular parameters which are relevant to synaptic weights, such as receptor composition, scaffold configuration, PTMs, and local RNA/proteostasis, which fixation degrades or disrupts. Specifically:
- Crosslinking effects: Protein-protein and protein-DNA crosslinks occlude reactive sites, distort conformations, and mask epitopes; enzymatic access is hindered (though we already noted this type of defect could be theoretically repaired).
- Phosphorylation states: Labile post translational modifications (e.g., phosphorylation) are not reliably preserved, degrading signaling histories tied to plasticity.
- Receptor stoichiometry and scaffolds: Stoichiometric balances (AMPA/NMDA, auxiliary subunits) and scaffold organization (PSD95, SAP97, Homer) are chemically perturbed, altering the “parameterization” of synaptic weight.
- RNA and local translation: mRNA integrity (spine-localized transcripts) and translation markers degrade; ribosomal and proteostasis signals that set recent plastic changes are lost.
- Chromatin and regulatory context: Crosslinking and ischemia distort chromatin topology and accessibility, weakening inference on activity dependent regulation.
In any case there is a common boundary condition: it’s reasonable to expect future methods to extract more from fixed brains than architecture alone (multi omics, modeling), but their success probabilities still hinge on the fidelity of preserved molecular parameters; without reliable molecular states, future decoding reduces to probabilistic inference of synaptic weights, with lower confidence per engram.
Scientific Papers:
- German, A., Akdaş, E. Y., Fluegel-Koch, C., Fejtova, A., Winkler, J., Alzheimer, C., & Zheng, F. (2025). Functional recovery of adult brain tissue arrested in time during cryopreservation by vitrification. bioRxiv. https://doi.org/10.1101/2025.01.22.634384
- Abraham, W. C., Jones, O. D., & Glanzman, D. L. (2019). Is plasticity of synapses the mechanism of long-term memory storage? npj Science of Learning, 4, 9. https://doi.org/10.1038/s41539-019-0048-y
- Dudai, Y. (2002). Molecular bases of long-term memories: a question of persistence. Current Opinion in Neurobiology, 12(2), 211–216. https://doi.org/10.1016/S0959-4388(02)00305-7
-
jordansparks
- Site Admin
- Posts: 283
- Joined: Thu Aug 27, 2015 3:59 pm
Re: Memories
I think you are misinterpreting current literature. I will use an analogy. Let's say we have a book that we want to preserve. We could cryopreserve the book by soaking the whole thing in cryoprotectant and cooling in liquid nitrogen. For this analogy, we will assume that the text on the pages slightly degrades during that process. We will also assume that we have trouble getting the cryoprotectant to penetrate fully, so certain areas will get further damaged during the subzero cooling. We could, in theory, take that cryopreserved book, thaw it, open it, and read the text. But the thawing process is even more damaging, so the whole thing would sort of fall apart if we tried to read it. We could use techniques similar to those used with the dead sea scrolls, like doing CT scans and reconstructing the words on a computer screen. Or we could use micro-robotics to start with the frozen book and carefully remove the cryoprotectant bit by bit. The robots would simultaneously need to repair the damaged paper fibers and move some of the blurred ink particles back to their original positions.
And now the chemical fixation version of our scenario: We first open as many pages as possible to apply some thin epoxy, but it's hard to get to all the pages, so after that step, we soak it in a container of monomer for a few months to make sure it fully penetrates. Then we cure it. We now have the book embedded in a solid block of cured plastic. We can't possibly read that book with current technology. But the text is better preserved than with cryo because the epoxy caused very little initial damage and there was no cooling step that caused further damage. Could we read that book? We could again turn to CT scans, just like the cryopreserved book. We could also use some high tech robotics to shave thin layers off the top and read the text that appears with each layer. Or we could use a different kind of micro-robotics to remove all the epoxy bit by bit until all that's left is the original paper book, still completely intact. The micro-robotics would need to carefully cut out the plastic from around each paper fiber.
The literature that you are citing only applies to the first scenario, not the second. It's analogous to saying that you can't open the pages once you apply epoxy, but we already know that. That's the whole point. All of those issues are completely moot when you are already starting with fixed tissue. Of course we don't have the technology to read the letters in the book, but they are all sitting there in pristine condition, just waiting for higher technology.
I would also emphasize that fixation does not damage any of those structures that you listed. As one example of many, chromatin topology is preserved beautifully. DNA is spooled around histones which are protein. The crosslinking is, therefore, very similar to the image I made showing aldehyde preservation. All those spools are preserved with spatial accuracy right where they sit. Similarly, synaptic weights are preserved as well as they possibly can be. A synapse is a dense mass of proteins, and it all gets locked in place. Fixation does not cause disruption or damage to those structures. Quite the opposite. It preserves the structures, and by extension, the synaptic weights.
There just isn't any better way to preserve them. This includes all molecular information. So when Andy said, "as long as memory is not encoded in smaller structures than is currently believed" he was including all that molecular information as potentially part of the memory encoding. I think he was being too generous. I can't even really imagine what a "smaller" structure might be that couldn't be preserved with aldehyde. Maybe the quantum state of some electrons? Regardless, aldehyde is the very best preservation that scientists have ever come up with. There just isn't anything that would preserve better.
Finally, structural preservation is fully equivalent to memory preservation. There is no distinction. They are the same thing.
And now the chemical fixation version of our scenario: We first open as many pages as possible to apply some thin epoxy, but it's hard to get to all the pages, so after that step, we soak it in a container of monomer for a few months to make sure it fully penetrates. Then we cure it. We now have the book embedded in a solid block of cured plastic. We can't possibly read that book with current technology. But the text is better preserved than with cryo because the epoxy caused very little initial damage and there was no cooling step that caused further damage. Could we read that book? We could again turn to CT scans, just like the cryopreserved book. We could also use some high tech robotics to shave thin layers off the top and read the text that appears with each layer. Or we could use a different kind of micro-robotics to remove all the epoxy bit by bit until all that's left is the original paper book, still completely intact. The micro-robotics would need to carefully cut out the plastic from around each paper fiber.
The literature that you are citing only applies to the first scenario, not the second. It's analogous to saying that you can't open the pages once you apply epoxy, but we already know that. That's the whole point. All of those issues are completely moot when you are already starting with fixed tissue. Of course we don't have the technology to read the letters in the book, but they are all sitting there in pristine condition, just waiting for higher technology.
I would also emphasize that fixation does not damage any of those structures that you listed. As one example of many, chromatin topology is preserved beautifully. DNA is spooled around histones which are protein. The crosslinking is, therefore, very similar to the image I made showing aldehyde preservation. All those spools are preserved with spatial accuracy right where they sit. Similarly, synaptic weights are preserved as well as they possibly can be. A synapse is a dense mass of proteins, and it all gets locked in place. Fixation does not cause disruption or damage to those structures. Quite the opposite. It preserves the structures, and by extension, the synaptic weights.
There just isn't any better way to preserve them. This includes all molecular information. So when Andy said, "as long as memory is not encoded in smaller structures than is currently believed" he was including all that molecular information as potentially part of the memory encoding. I think he was being too generous. I can't even really imagine what a "smaller" structure might be that couldn't be preserved with aldehyde. Maybe the quantum state of some electrons? Regardless, aldehyde is the very best preservation that scientists have ever come up with. There just isn't anything that would preserve better.
Finally, structural preservation is fully equivalent to memory preservation. There is no distinction. They are the same thing.
-
PCmorphy72
- Posts: 32
- Joined: Sun May 26, 2019 12:39 pm
Re: Memories
There is 1 point I had already agreed on, about crosslinking, and I can also agree on the 2nd point that “chromatin topology is preserved beautifully”, but I would have preferred if you had explained technically the other 3 points you have not addressed yet (i.e. why it is not correct that “fixation degrades or disrupts” precisely these 3 things: “Phosphorylation states”, “Receptor stoichiometry and scaffolds”, and “RNA and local translation”), instead of explaining only through a few concrete words about why “synaptic weights are preserved as well as they possibly can be” (i.e. “A synapse is a dense mass of proteins, and it all gets locked in place. Fixation does not cause disruption or damage to those structures.”).
I do appreciate those few words, as well as the amusing agreeableness of the analogy with a book, though I think it would not have been necessary for me to understand what you mean. By the way, I often think of the analogy of death with a library fire: in most “cryonics / brain preservation” debates (not exactly our case), the analogy is that they discuss which type of fire extinguisher to use: a dry chemical one (which leaves residue that soils materials and often makes them unusable — an analogy with chemical fixation) or a CO2 extinguisher (which leaves no residue but is less effective — an analogy with cryopreservation), while the real point is to act together as quickly as possible to prevent the fire from spreading (the fire is the real cause of disruption — an analogy with cerebral ischemia).
I would also like to know whether what you mean by “structural preservation” is the same as what Andy means by “geometric aspects”, since the contrast in who is more “generous” between you and Andy seems to me even more evident when comparing these two statements:
Andy McKenzie: “Yes, I agree that biomolecular information will most likely be necessary (or at least, this should be our assumption), not just the geometric aspects of the connectome.”
Jordan Sparks: “Finally, structural preservation is fully equivalent to memory preservation. There is no distinction. They are the same thing.”
So I think I should say: “Yes, Andy agrees more with me than with you.”
I do appreciate those few words, as well as the amusing agreeableness of the analogy with a book, though I think it would not have been necessary for me to understand what you mean. By the way, I often think of the analogy of death with a library fire: in most “cryonics / brain preservation” debates (not exactly our case), the analogy is that they discuss which type of fire extinguisher to use: a dry chemical one (which leaves residue that soils materials and often makes them unusable — an analogy with chemical fixation) or a CO2 extinguisher (which leaves no residue but is less effective — an analogy with cryopreservation), while the real point is to act together as quickly as possible to prevent the fire from spreading (the fire is the real cause of disruption — an analogy with cerebral ischemia).
I would also like to know whether what you mean by “structural preservation” is the same as what Andy means by “geometric aspects”, since the contrast in who is more “generous” between you and Andy seems to me even more evident when comparing these two statements:
Andy McKenzie: “Yes, I agree that biomolecular information will most likely be necessary (or at least, this should be our assumption), not just the geometric aspects of the connectome.”
Jordan Sparks: “Finally, structural preservation is fully equivalent to memory preservation. There is no distinction. They are the same thing.”
So I think I should say: “Yes, Andy agrees more with me than with you.”
-
jordansparks
- Site Admin
- Posts: 283
- Joined: Thu Aug 27, 2015 3:59 pm
Re: Memories
Andy and I are saying the same thing. Connectomics is the study of the shapes of the membranes of the neurons. Connectomics does not include biomolecular information. That information is very well preserved, but our current technology has trouble deciphering it. So he's saying our current technology can only give us the geometric shapes of the neurons and that we will probably need finer detail than that. The detail is all there and has been preserved well. He's coming at it from the lingo that's in current research papers, where they lack the technology to see most of the finer detail. I'm using lingo that's more in laymen terms and is bigger picture. When I'm talking about structure, I'm talking about all molecules. So there is no discrepancy between what Andy and I are saying.
-
PCmorphy72
- Posts: 32
- Joined: Sun May 26, 2019 12:39 pm
Re: Memories
I understand: both of you see it differently from what I had misinterpreted. But why? Is it only because “the detail is all there and has been preserved well”? Perhaps it would be more accurate to say that fixation essentially immobilizes molecules, which explains the preservation of morphology, while at the same time certain dynamic states (like phosphorylation, receptor scaffolds, and RNA translation) may be lost. And since these dynamic molecular processes are precisely what underlie long‑term memory consolidation (or do they not?), I wonder how structural preservation alone can be equated with memory preservation.
-
jordansparks
- Site Admin
- Posts: 283
- Joined: Thu Aug 27, 2015 3:59 pm
Re: Memories
We do know that long term memories are stored in some sort of molecular structure that is durable enough to last more than a few hours. Aldehyde fixation is, therefore, preserving those memories. It doesn't really matter what the structures are. You've listed some things that you think are getting damaged or not captured by fixation, but that's really not possible. If it's some sort of "dynamic" state that is short lived, we don't care about it. If it's a longer lived state, then we already are capturing it. You're really getting into the weeds with some of your specific concerns. I'm actually trying to avoid talking about those specifics because it's truly irrelevant. But here we go. I'll do one more: RNA translation.
1. Learning triggers changes in the nucleus. (captured by fixation)
2. RNA gets transported to synapses (the RNA gets locked in place by fixation)
3. Proteins get synthesized that cause changes to the synapses (these changes are captured by fixation)
So I'm not sure what part of that you think is not getting captured by fixation. This same theme can be repeated for any of your other concerns, including phosphorylation and receptor scaffolds. If they are transient changes, they don't matter. If they are durable, we are already capturing them.
1. Learning triggers changes in the nucleus. (captured by fixation)
2. RNA gets transported to synapses (the RNA gets locked in place by fixation)
3. Proteins get synthesized that cause changes to the synapses (these changes are captured by fixation)
So I'm not sure what part of that you think is not getting captured by fixation. This same theme can be repeated for any of your other concerns, including phosphorylation and receptor scaffolds. If they are transient changes, they don't matter. If they are durable, we are already capturing them.
-
PCmorphy72
- Posts: 32
- Joined: Sun May 26, 2019 12:39 pm
Re: Memories
Thanks Jordan, I realize now that my wording blurred the line between “consolidating features” (the dynamic processes that stabilize long‑term memory, such as transcriptional activity, nuclear topology adjustments, and epigenetic regulation in progress) and “consolidated features” (the relatively static states that store already stabilized memory traces).
My actual intention was to refer to the nuclear features that preserve already consolidated long‑term memory, not the ones still consolidating. Your distinction helped me see where my wording was misleading, misinterpreting literature just because of this swapping of meaning. I’ll refine my wording accordingly in any future posts on the subject.
My actual intention was to refer to the nuclear features that preserve already consolidated long‑term memory, not the ones still consolidating. Your distinction helped me see where my wording was misleading, misinterpreting literature just because of this swapping of meaning. I’ll refine my wording accordingly in any future posts on the subject.