X-Message-Number: 13222 From: "john grigg" <> Subject: the feasibility of cryonics... Date: Tue, 08 Feb 2000 00:22:07 PST Hello everyone, Cryonics has become a thread on the extropians digest once again and I want to share with all of you some of the best posts. They are definitely worth reading. Extropians are not the die-hard cryonicists I would have initially thought them all to be but they do bring up some excellent points. I look forward to reading the responses from the Cryonet. sincerely, John Grigg Date: Sun, 6 Feb 2000 12:36:59 -0800From: Subject: Re: CRYONICS: feasibility [was Re: Fox cryonics show] Robert J. Bradbury, <>, writes: >What I want to know is are there people who have read Ralph's >technical essay:> The Molecular Repair of the Brain > http://www.merkle.com/cryo/techFeas.html >and who *still* believe cryonics is infeasible?> >By infeasible, I mean do you have concrete reasons to believe >that something Ralph says is in gross error or missing in the >details of how one would molecularly reassemble the damage caused by >freezing? Another way of asking the same question is -- How >can you believe that trillions of nanobots with collective >ultra-human intelligence could *not* reassemble the 3D jig-saw >puzzle left by your frozen brain? I am signed up for cryonics, but I have never found this essay completely convincing. It seems to gloss over the hard part, which is the possibility of near-universal damage to the synaptic structures which we believe encode memories. Cryonicist Mike Darwin has written many times about the damage he sees to brain tissue which has been cryopreserved using current protocols. He sees devastating damage, ice crystals cutting everywhere, breaking and tearing tissues apart as they grow, making things look like a bulldozer has come through. Ralph writes of evidence that brain tissue can withstand conversion of up to 60% of water to ice and maintain some level of functionality, and says that current protocols have kept the ice level below 40%. But these experiments are not able to rule out the possibility of damage sufficient to eradicate memory. They are able to get some brain waves and other evidence that neurons are functioning to some degree, but there is obviously significantly damage. Ralph also suggests that because synaptic structures are large, composed of tens of thousands of atoms, that this will facilitate the task of reconstructing them. However this size is meaningless if the damage is occuring at this scale. If a large fraction of these 10000 atoms are badly disrupted it may not be possible to figure out exactly how they were arranged before freezing. Most of the latter part of the essay discusses the feasibility of measuring atoms and moving them around, and I am willing to stipulate to that. It is the middle step here, where we use our computer models to figure out where things were before the damage occured, that troubles me. The best analogy I can suggest is that the brain may be like a city. When it is frozen, it undergoes a terrible bombardment, and buildings are damaged or destroyed, rubble lying in the streets everywhere. In the aftermath, people can sweep up and get things going after a fashion, piles of rubble here and there. It can function to a limited extent, but it is not at all the city it was before. This is like the cat brain spontaneously recovering brain waves. To rebuild the brain is like a team of archeologists reconstructing the city, figuring out where each piece of rubble goes, getting the architecture back where it was. That's probably not going to be possible. We can put something together that may look nice, but it is an educated guess at best. Real cities have idiosyncracies and quirks, and when those are lost in the rubble they will never be restored. What we end up with is a ficticious city, one which has similarities to what was there before, but the original city is lost forever. I don't know that the brain necessarily works like this; maybe there will be enough information for a full restoration. But it seems to me that it is possible that this is the situation, and that this model is consistent with the limited studies which have been done. If so then Ralph's reasoning is incorrect, and from the available data we cannot conclude that cryonics is especially likely to work.Hal Date: Sun, 06 Feb 2000 14:59:16 -0600 From: "Eliezer S. Yudkowsky" <> Subject: Re: CRYONICS: feasibility [was Re: Fox cryonics show] Forget the problem of repairing the brain. Think of it in uploading terms. The theoretical question is: "If you know where every single atom in a frozen brain lies, could you extract an model of the brain's operation with such high resolution that the informational damage caused by freezing would not be significant (on a damage scale calibrated by the effects of neural death and quantum/thermal randomness in day-to-day operation)?" I think this sounds reasonable; damage to structure on the >10^6-atom-scale may appear to destroy the information contained on that level of abstraction, but the same information should still be obtainable from the internal structure of the mostly-untouched inner volumes of the 10^6-atom units being shoved around. If so, then the practical question is extracting the information without causing exactly the sort of atomic-scale disruption that would destroy that level of abstraction as well.- -- Eliezer S. Yudkowsky http://pobox.com/~sentience/beyond.html Typing in Dvorak Programming with Patterns Writing in Gender-neutral Voting for Libertarians Heading for Singularity There Is A Better Way Date: Sun, 6 Feb 2000 13:58:26 -0800 (PST) From: "Robert J. Bradbury" <> Subject: Re: CRYONICS: feasibility [was Re: Fox cryonics show] On Sun, 6 Feb 2000, Eliezer S. Yudkowsky wrote: >Forget the problem of repairing the brain. Think of it in uploading> >terms. Well Eliezer, not everyone may want to be reanimated as an upload. Some of us might want to putter around the galaxy in our interstellar star-hopper until we make some fatal mistake in navigation or alien diplomacy. Its so much more romantic that way compared with dying a thousand deaths in an SI VR. >The theoretical question is: "If you know where every single >atom in a frozen brain lies, could you extract an model of the brain's >operation with such high resolution that the informational damage caused >by freezing would not be significant (on a damage scale calibrated by >the effects of neural death and quantum/thermal randomness in day-to-day >operation)?" This is pretty much the question I'm trying to get at. Hal responded to some degree expressing his doubts. From my perspective, thequestions are: (a) Does the process of ice crystal formation develop enough force to break interatomic bonds or will the crystals almost universally disrupt H-bonds and Van der Waals interactions? (b) What is the probability that within a some fixed (sub-cellular) volume that I will have identical surfaces that preclude putting things back together exactly the way they were originally? With regard to (b), I find it hard to believe using nanodis-assemblers that I could not remove all of the ice *entirely* (e.g. creating approximately a freeze-dryed brain), then map all of the surfaces and not discover *exactly* where the various synapses should be positioned and setup the required repair operations. I'd envision you would have nanobots carefully positioned around the broken pieces, preprogrammed to slowly heat the tissue and execute vector movements designed to properly reposition things. Only if the damage includes the breaking of covalent bonds [e.g. (a)] does the process of repair become more difficult (having to replace broken chromosomes, proteins, etc.).> I think this sounds reasonable; damage to structure on the > >10^6-atom-scale may appear to destroy the information contained on that >level of abstraction, but the same information should still be >obtainable from the internal structure of the mostly-untouched inner >volumes of the 10^6-atom units being shoved around. If so, then the >practical question is extracting the information without causing exactly >the sort of atomic-scale disruption that would destroy that level of >abstraction as well. Since we seem to be able to do this level of "atomic" surface sensing (molecular fingerprint recognition using AFMs) and you have things like laser driven decomposition of surface layers (currently used for things like DNA sequencing via mass-spec), this seems highly feasible to me. Furthermore, taking apart a small number of cells will give you the genetic blueprint (and consequently the 3D blueprint) of all of the molecules (excepting perhaps viruses or nutrients) that should be in the cells. I see this as no different from paleontology where you can recognize a dinosaur from a fragment of bone sticking out of a rock formation. My sense of the molecular biology of the brain at this point is that you could probably destroy the contents of every single cell so long as you did a highly accurate inventory. Then once you get all of the cell membranes and synapses in place, you fill the cell with water, repopulate the molecular contents andoff you go. The contents of the cell is probably highly flexible as well. I would guess that as long as you get things "balanced" sufficiently that the cells will "run", you will be ok. The only places where the contents is important are when you are dealing with internal messangers involved in increasing synaptic strengths. Since you "died" anyway, it will probably be like coming out of a coma or a very deep sleep, so I suspect any disruption of short term memory formation, because you don't get the molecule counts completely correct, is going to be unimportant. You might end up being a different person than you would have been had you not undergone the suspension and reanimation but it begs the question of whether or not some of us end up as different persons because we experience high or low blood sugar once or twice a day.Robert Date: Sun, 6 Feb 2000 19:04:49 -0800From: Subject: Re: CRYONICS: feasibility [was Re: Fox cryonics show] Robert J. Bradbury, <>, writes: >This is pretty much the question I'm trying to get at. Hal responded >to some degree expressing his doubts. From my perspective, the >questions are: > (a) Does the process of ice crystal formation develop enough force > to break interatomic bonds or will the crystals almost universally > disrupt H-bonds and Van der Waals interactions? I am concerned about synaptic structure. I don't know how rigidly the synapse is held together, but I suspect that a growing ice crystal could separate, perforate and even shred the membranes. I would suspect that cell membranes are not strong enough to withstand ice crystal penetration. I have the impression that synapses are relatively gently bound together. > (b) What is the probability that within a some fixed (sub-cellular) >volume > that I will have identical surfaces that preclude putting things back > together exactly the way they were originally? The problem is that the shapes may not match very well once the ice is removed. As the ice is forming, the material around it is at least a semi liquid. Water flows out of the cells and they would partially collapse due to dehydration. The cells themselves would be shoved around by the ice, tugged and torn, and their shapes will change.Hal Date: Mon, 7 Feb 2000 00:29:52 -0800From: "Zeb Haradon" <> Subject: Re: CRYONICS: feasibility, and AI vision If cryonics is physically/mathematically possible, then it is technologically possible, and barring a return to the dark ages, will be technologically feasible. What I mean by physically/mathematically possible is that there may not be enough information in a frozen brain to interpolate what the corresponding working brain looked like. Suppose my brain is frozen, it changes from (unfrozen) state Z to (frozen) state Z'. If for all (frozen) x', there is one-and-only-one corresponding (unfrozen) x, then a few billion nanocomputers running in parallel and given a 3D scan of my frozen brain (Z') should be able to figure out how to put everything back together into Z. However, it may be that too much information is lost, perhaps there are several different possible states that can result in a frozen Z'.I toyed with the idea of proving that it's possible, today, using AI computer vision techniques. The idea would be to come up with a computer vision program which can look at microscopic slides of groups of cells that had been frozen and are suffering ice damage, and return a picture of what the group of cells looked like before being frozen. Use a neural net, and train it on several (thousand) examples, where you have photos of the cell group both before and after freezing, so you can use the correct examples to train the net. After much learning, if it's possible, and if it's designed correctly and has enough computing power, it should be able to get it right. Unfortunately, I know very little about implementing the techniques of computer vision. - -------------------------------------------------------------------------- Zeb HaradonMy personal website:http://www.inconnect.com/~zharadon/ubunix/ A movie I'm directing:http://www.elevatormovie.com ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=13222