X-Message-Number: 9172 From: Thomas Donaldson <> Subject: Re: CryoNet #9168 Date: Tue, 17 Feb 1998 00:17:50 -0800 (PST) Hi again! Some additional points: you suggest that we can read off memories from the frozen structure. Unfortunately, if that structure is broken and distorted, as it probably is from current suspension methods, your readout will also be just as broken and distorted. So it is not nearly so simple as you say. Not only that, but brains are 3-dimensional and micrographs are 2 dimensional, so you must have many micrographs to have any hope at all of reconstruction. Moreover, you will almost certainly have to know a good deal about how brains normally connect: given that the one you're looking at ISN'T normal but damaged, just how are you proposing to reconnect it without such knowledge? And I am saying that if you have that knowledge you will be close to knowing how to reproduce those connections in a biological system. I will discuss that a bit more, right here. The kind of knowledge you will need is that of such things as the processes by which neurons create new synapses, and the neurons which are likely to acquire those new synapses. For instance, one theory posits that pyramidal neurons do not change their axons, but the change in learning comes by changes in inhibitory neurons using GABA (this is presently a hypothesis, as yet unproven, but it tells us something about what might connect with what else). There are several different kinds of neurons which connect to our pyramidal neurons and use GABA as their transmitter. Moreover both axons and dendrites seem to be changeable: they move about a bit, presumably as some connections form and others disappear. You would need to understand how that happens, in detail: both the growth and the loss of synaptic connections. For any repair, we will need first to work out just what connects with what (currently a problem --- remember we need to do that not on a gross scale but on the very fine microscopic scale on which our neurons form memories) and then if something has happened to that connection, get it to regrow. If we revive the neurons it should not be difficult to use systems developed from lymphocytes, bacteria, and other such semi-independent creatures to put the proper trophic substances in the proper places. For what it's worth, a 2-dimensional micrograph doesn't come near to telling us about the damage due to freezing: so long as we could find membranes (though pierced and damaged) connecting one part of a neuron with another, then the problem becomes not that of forming connections between neurons but of repairing a single neuron. If and when that happens, it makes the problem much easier: no need to work out connections which no longer exist by inference from what you find in the damaged system. And however you examine your brains, frozen or not, you will have to examine them on that scale AND in 3 dimensions. Not only that, but your examination, to be at all thorough, must involve not just the APPEARANCE of the damage but also its CHEMISTRY. We work not just electrically but also chemically, and a knowledge of just how that chemistry works will be important in any attempt to find out the original structure of a damaged brain. You say that you do not like nanorobots, yet how do you propose to make your pictures of damaged brain structure at the required resolution without such devices? If you can work at that scale, you are very close to being able to repair at that scale. And for that it won't really matter what materials your repair machines (or (or highly modified systems of lymphocytes and bacteria, or whatever) consist of: a nanorobot of nonorganic materials can just as well spread around a trophic molecule as directly connect two parts of a severed dendrite. Of course, if the current plans to be able to freeze undamaged brains work out, then no such readout is needed, nor will we need any uploading at all. We merely revive the brain --- from vitrification or whatever. Finally, existing neural nets in computerdom do NOT match those of brains. Sure, just as in many other things we can obtain machines which in some ways come close. But the growth of new connections is essential to the way we learn; no neural net to which I am acquainted does that. If you go off and build one that does, then what I say will cease to be true. It's not enough just to imitate SOME pertinent features of neurons. To make a brain you will need to imitate all of them. Spiking, as you probably know, is likely to play an important role encoding our thoughts, and in that respect I'm not surprized that we can make electronic models behaving the same way. But we don't just think with our brains. We also learn with them. And we feel with them too. Finally, I will make some other points to explain just what I think about these issues. First of all, my problem with uploading of any kind comes from my desire not to be uploaded into a computer (though someday I might be STORED in a computer, inactive). I do not believe that computers can act as brains. I do not claim that this means that we could not build a brain of very different materials. I even think it LIKELY that we can do that. But if I understand you, you seem to believe that doing so will somehow be NECESSARY. Given that the strongest version of nanotechnology right now is BIOtechnology, I doubt that very much, even for low temperatures. (Right now there is intensive work on getting biochemistry to work in solvents other than water -- like glycerol, for instance. And actually applied industrial work doing this already exists). For that matter, if we really want to be revived, we want not just a brain but all the rest, which adds to the difficulty if you want it to consist of nonbiological materials. You set yourself a very hard problem there, not just to create a totally new form of life but to create instances of it which are close to YOU. And I assume that you too want not just to be able to think but to learn and feel. So over to you, now. And best wishes, and long long life, Thomas Donaldson Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=9172