Contains a lot of guesstimates though, which it freely admits throughout the text (in the abstract, not so much). It's a bit like a very tentative Fermi estimate.
A Moravec transfer could be a middle option between scanning a living brain and scanning a dead one.
Hi, biomedical engineering undergrad here. I have to go review some materials before I can give you a more precise answer to the question you're interested in, but my first thought upon reading this is that it's not just a matter of structural makeup, but mainly one of patterns of electrical activity in the brain. Most studies I've seen that touched upon high-level concepts like emotions, attention, or interest used fMRI scans to correlate brain activity in certain areas with exposure to certain stimuli. I'm not sure how much this says about the problem of consciousness, but if it is relevant, then you have an additional parameter to consider other than spatial resolution -- namely, time.
Thanks for looking into it. FYI, it seems like the time dimension won't be necessary since people have recovered after their brain activity stops e.g. Profound Hypothermia and Circulatory Arrest
I don't think you can conclude that "the time dimension won't be necessary" from the paper you linked.
(A) Because the brain temp in fig 1 (and many of the other experiments reported) was about 20 deg C, and neurons still fire at that temperature. (It's routine in many experiments to record synaptic activity in neurons at or near these temperatures). So you can't infer from the metabolic measures that neural firing has ceased. Neural firing patterns are almost certainly preserved to some substantial extent under the conditions in this paper (albeit with a lower expected spike probability).
(B) The figures show (to quote the paper) that "...cerebral damage is evident in biochemical terms at 30 minutes, as judged by cells at 37 minutes and by functional impairment at 45 minutes." During that same time interval, ATP consumption declined only slightly and indices of mitochondrial stress increased only slightly (<20% in each case). That makes it seem to me like arresting normal metabolic function even slightly causes widespread dysfunction.
Why did this paper make you think that time-varying patterns of neural activity wouldn't be necessary to reconstruct/understand psychological phenomena?
Super-resolution microscopy is an interesting recent development that won the Nobel Prize in Chemistry last year. Here's another article on the subject. It has been used to image mouse brains, but only near the surface. It won't be able to view the interior of any brain, but still interesting.
Bostrom's paper on whole brain emulation has a section on microscopy that might be of interest. The part of the paper on scanning starts on p. 40, and there's a table which compares different scanning methods on p. 53.
Thanks for the pointer. Sadly, the techniques discussed all require slicing up the brain before imaging.
I don't know the answer to your question, but my guess is that non-intrusive periodic backups, if and when when implemented, will not use this particular technology. Storing about 10^30 bits of data per backup is unlikely to be the way to go, given that most of them will be redundant.
That's a good point, shminux. But don't you think that the entire organ needs to be scanned at high resolution first before storing a compressed copy?
I think it depends on what is the physical correlate of consciousness, memory and identity. At this point we know very little about it. Clearly single molecules do not matter, and single neurons can die without obvious ill effects. It's not clear how glial cells affect consciousness. We have to learn a lot more about the issue to figure out what can be compressed out from such a scan, and once we understand it, we can probably use a simpler scanning technique.
Was going to try and provide an answer but lukeprog's link is more informative than what I would have written :P
One factor though that to me seems worth considering:
Imagine your goal of people being able to back themselves up turns out to never actually happen - either because of technical infeasibility, or we go extinct before figuring out how, or because static structural information is simply insufficient to derive psychological phenomena. (That last option might be a minority position on LW, but I'd venture to say it is not a minority position among neuroscientists and psychologists.)
If that WERE the case, then money and time donated to medial imaging is likely to STILL have enormous positive benefits, in terms of advancing basic science and diagnosis/treatment of mental illness. By contrast, if your end goal turns out to be infeasible, money and time spent on cryonics and plastination will turn out to have been largely wasted.
Or at the very least, it is much easier to imagine the former having a large societal benefit for reasons other than life extension, while the "side-effect" discoveries of brain plastination would not have the same obvious public benefit (even if those benefits were non-zero).
So it seems like if you're not sure which is the best approach and want to be fairly certain you're not wasting your time/money, it's better to dump resources into medical imaging, which would be justified by the spillover effects alone.
Thanks for pointing this out. I'd seen their site before but could get no solid details on how they'd actually accomplish the 'backups'. How'd you come across those details?
Oops, didn't realize that those (big picture) details weren't on their website. I think if you're curious you should contact them.
I'll answer your question first before questioning the line of thinking you're on...
Yes it is possible, although probably not in the way you are imagining. It would be surprising to me if a mechanism was found for remote scanning a brain at that resolution, not to mention the practical difficulties of recording the image. But as a proof of concept(ual analysis) I'll point you at Freitas' work to show that it would be possible in principle to in-situ destructively scan and then recreate neuron cells, a small number of cells at a time.
However I hope that answer shows a third possibility to consider beyond imaging and cryonics. Perhaps the most promising path through the singularity is intelligence augmentation / neural upgrades. Scan the neuron just as described above, but rather than re-create it with biology, use a machine neuron instead. Once the process is finished and biology left firmly behind, you can overclock the brain and/or add redundancy.
The trick to living forever should be strengthening the body, not regular backups in my not-so-humble opinion. Your backup is not you, just someone who thinks it is you. If you want to live forever, then focus on yourself.
To all my physicists in the house, will it ever be possible for a device to scan the contents of a human head at the molecular level (say, 5 x 5 x 5nm) while the subject is still alive? I don't have a physics background, so if you could also just point me to the materials I need to read to be able to answer the question, that would be wonderful as well.
The background: I want to live to see the far future and so I'm researching the feasibility of alternatives to cryonics that'll let people "back up" themselves at regular intervals rather than at the point of death. If this is even theoretically possible then I can direct my time and donations towards medical imaging researchers. If not then I'll continue to support cryonics and plastination research.
I'm looking forward to your responses!