A view on cryonics (was Re: [extropy-chat] Bad Forecasts!)
Rafal Smigrodzki
rafal at smigrodzki.org
Wed Sep 15 22:19:08 UTC 2004
Brett Paatsch wrote:
>Rafal Smigrodzki wrote:
>
>
>
>> Thus, the concept of "identity" is
>>in fact a whole slew of related but distinct concepts, with
>>different properties and referents.
>>
>>
>
>I've no problem with this but it doesn't seem relevant here. As
>Slawomir Paliwoda says in his excellent post on Personal Identity,
>"personal identity depends on a mind powered by the brain which is
>a physical object, and like all other objects, it can and should be
>subject to scientific investigation."
>
>And .... "once we agree that, in science, there's only one correct
>explanation for something at the exclusion of other theories, we can
>say that there can only be one explanation/description of personal
>identity that is true and all other theories are false."
>
>
### Now, this is news to me: do you think that Euclid's parallel axiom
is true? False? Neither? Non-scientific?
Or more: since the taste of rocquefort depends on chemical interactions,
which are subject to scientific investigation, what is the correct taste
of rocquefort?
Generally, science doesn't necessarily produce exclusive either/or
results, and doesn't directly make normative statements. The statement:
"I like the idea of having a self-similar material structure in the
future" is normative, and its normative content *may not* be subject to
verification - just as the normative content of your statement "I don't
care about having self-similar material structures in the future, except
if produced by continuity (of sorts) with my present structure".
------------------------------------
>
>
>>I accept that your (Brett Paatch's) identity is what you say, since
>>this is what is produced by the processes in your brain (and my
>>meta-rule for such definitions is that they are produced by the
>>brains in a self-referential manner) . It's a fact, albeit a fact
>>pertaining only to Brett Paatsch. Now, my own, Rafal's identity
>>is defined differently. It is also a fact. Although the definitions
>>are different, they are not contradictory - they apply to different
>>objects, just like the varied definitions of an "original" in the art
>>world. Even though your definition of self is different from mine
>>(not even analogous), I do not think you are mistaken - merely
>>different. Do you see the point?
>>
>>
>
>Yes, I understand what your saying, but I think that you have
>gone off on a tangent. You are taking us away from a discussion
>on cryonics and of the obstacles cryonics would inevitably have
>to overcome to be of interest to people like you and I. That is
>people that currently have our sense of self located in our brains.
>
>
### No, I am not going off at a tangent - because I am not like you. My
sense of self is not "located" in my brain, it's located in my
personality-defining information-like structures.
----------------------------
> You are currently located in
>your brain just as I am. Your aspirations for cryonics will have
>to deal with that. A cryonics procedure that goes hunting
>about the room to pick up astral travelling selves is unlike to
>be of interest to either of us.
>
>
>
### Yes, my personality-defining information is inscribed in my brain,
currently. My aspirations for cryonics deal with this excellently.
---------------------------------------
>>In other words, you are a universalist, believing that a
>>single definition of self (your own) is the only correct one
>>("all members of the species homo sapiens have their sense
>>of self....."), while I am a pluralist, letting everyone decide
>>where their own (and only their own) identities lie.
>>
>>
>
>You've split my sentence in the middle when you quote me
>above. I'd said "all members of the species homo sapiens have
>their sense of self inherent in the structure of their cellular brain".
>That is not me defining a self concept. That is me pointing out
>that whatever our self concepts are they are *located* if we
>are homo sapiens in our brain structure.
>
>
### I don't know how you can claim that a concept is "located" anywhere.
A concept may be written down in a particular book, or be embodied in a
patented machine, but the word "located" should only be used for
material objects.
Let me quote your full paragraph to which I referred above: "I do think
all members of the species homo sapiens have their sense of self
inherent in the structure of their cellular brain. I think that it is a
mistake to think that someone who is now a homo sapiens can be
abstracted out of their cellular substrate and yet somehow continue to
exist as disembodied pattern and then to be re-instantiated again. I
think that the self is lost in the process."
In other words, you say that for all humans, sense of self is "inherent"
in the brain and is inevitably lost during transfer of information. The
use of the universal qualifier makes it sound like a definition,
insisting that for all humans, whatever is transferred
("re-instantiated") cannot be self.
-----------------------------------
>>If you mean "functioning without any material substrate", then no,
>>of course not.
>>
>>
>
>Ok. So you think you would cease to exist as "you" while you exist
>only as a dataset of information for building you at a later date?
>
>
### What do you mean by "exist"? I would exist as static information, I
would not exist as a conscious information processor.
--------------------------------
>Sort of cake - recipe -cake. While your a recipe your not a cake.
>
>
### Yes, sure, I have no problem with existing as a recipe. I am both a
recipe and a cake at this time (I can be used as source of information
for making copies and I keep processing information at the same time),
in the future I might temporarily exist as a recipe only, and then
resume normal functioning.
------------------------------------
>>> <>### Let me ask you something: do you think that it could be in
>>> principle
>>> possible to analyze the material structure of your brain, and use the
>>> information to make a brain which would behave in a way consistent
>>> (similar enough) with you (e.g. identify pictures of your mother as "My
>>> mother", or on seeing things you like say it likes them, etc.etc.)?
>>
>
>It might be. In principle. It might be possible to create what Slawomir
>Paliwoda calls a perfect clone. I wouldn't think that perfect clone was
>me. And I don't think that clone could be created because I don't think
>the information to create it could be gathered to the requisite resolution
>without destroying the original me in the process.
>
>
### No, I didn't ask about the "perfect clone" - I merely asked about a
copy which would behave consistently with you (have largely the same
responses to the same stimuli as you do). The exact level of similarity
here might be as low as that sufficient to fool all the persons you know
into believing the copy is actually you.
Whether making such a copy can be achieved without disassembly of you is
not relevant to the question. And yes, I already know you wouldn't see
the clone as "you".
------------------------------------------
>>But, if you agree that a sufficiently
>>advanced technology could produce such a brain, then you
>>cannot say we disagree about facts.
>>
>>
>
>I can conceed the possibility, but I don't find it worth giving
>a lot of thought too, given that I realise I would not be my
>"perfect clone". And there are plenty of other 'facts' yet for
>us to disagree over.
>
>
### Great! You agree that the disagreement is about value - you "don't
find it worth giving a lot of thought to"! This is the point - you are
not interested in having such a clone, but I am.
-----------------------------------------------
>
>
>>Both you and me would agree that it is physically possible
>>to make copies of our brains that would act similarly to the
>>originals. The only difference is that I am sufficiently satisfied
>>with/enthusiastic about the prospect of having such a copy
>>in the future that I am willing to pay for cryonics (and yes, I
>>even value this situation equally with "survival" by
>>spatiotemporal continuity of cells), while a copy of you
>>would not elicit sufficient interest from you.
>>
>>
>
>I get what you are saying. But your right I wouldn't be sufficently
>satisified or enthusiastic. To me you are excepting that you
>will die (be materially decomposed and cease as a process) but
>are taking consolation that something just like you will be
>brought back to life (recomposed and started as a process via
>means which you conceed you do not in detail understand).
>
>Your diverting limited resources from solving problems or
>overcoming limits in your current you-process in order to endow
>life to some future you-process. Even though the current and
>future you-processes will not overlap in time. Correct?
>
>
### Yes, exactly! This is indeed the case, whether you call
dying-and-recomposing, or and I call it "survival", this is indeed the case.
In the future (e.g. after my premature death in an explosion of a liquid
nitrogen tank :) my recomposed copy may contact you, and while
acknowledging that by your definition I am dead and he is somebody else,
he tells you he remembers from a first-person perspective our present
discussion, and that life is great, he's been boozin' an' humpin' like
in the old times. Ain't cryonics great, or what?
---------------------------------
>
>
>In fairness, why should anyone want to provide numbers or technical
>analysis? What is in it for them? To do that they'd have to take on a
>burden at some opportunity cost of time. If you had laid out numbers
>and technical analysis in detail yourself then they might do you the service
>of checking and perhaps correcting any errors etc.
>
>
### Opinions of scientists are valuable only insofar as they rely on
superior knowledge and technical analysis. Without these, scientist's
opinions about a subject (e.g. rocquefort or cryonics) are not any more
significant than the opinions of carpenters or business executives. The
technical analysis must be disclosed for verification, or else it's
reasonable to infer that the opinion is groundless (based on ignorance).
---------------------------------------
>
>
>> Therefore, I am justified in treating
>>scientific detractors of cryonics as ignorant, since if they knew any
>>scientific arguments, they would have used them.
>>
>>
>
>That is a non sequitor. People generally need a reason to do
>work that your not paying then to do, they don't normally need a reason to
>avoid working for nothing.
>
>
### Some scientists offer unsolicited opinions about cryonics without
facts and numbers - why? Nobody is paying them to do so.
--------------------------------
>>### Well, here is how I imagine cryonics might work for me:
>>
>>Cryonic vitrification very soon after death (a few hours, hopefully a
>>few minutes) will preserve the brain structure down to the level of
>>synapses, with intact synaptic protein levels (which define the synaptic
>>strength), and the levels of other proteins, including transcription
>>factors in the nucleus, and most RNA and protein in the cytoplasm and
>>the ECM.
>>
>>
>
>Minutes vs hours?
>
### There is some mild ultrastructural damage within 20 minutes of warm
ischemia (mitochondrial swelling), and after 24 hours there is visible
retraction of a significant number of synapses. Therefore, loss of
personality-defining information occurs somewhere during that time,
probably no less than 5-6 hours. Anything less than 30 minutes should be
perfectly safe, based on imaging of vitrified samples.
---------------------------------
> Death as determined how?
>
### By cessation of heartbeat.
-------------------------------
> Vitrification how?
>
### Using Alcor's current procedure.
------------------------------
> Down
>to (and including the synapses I presume) - you'd want your memories.
>To capture you personal synapse pattern you'd need nanoscale
>resolution.
>
### The resolution provided by a near-field scanning optical microscope
is more than sufficient (goes down to 32 nm) to image synapses (500 -
5000 nm). A confocal scanning microscope has lower resolution but still
sufficient to image synapses (and even actin filaments).
-----------------------------------------
> As you know neurons are not nicely rounded cells like one
>might find in a high school text book - they are more like tree or root
>structures and the structure matters.
>
### I do not get my information about neurons from high school textbooks.
-------------------------------
> Proteins levels alone won't do
>it you need to know where the proteins (and not just the proteins)
>were when they were in their natural state.
>
>
### This is why you need to use a confocal or near-field scanning
microscope, as described. Non-proteinaceous molecular species can also
be detected by antibodies. In the unlikely case of nucleic acids assays
being necessary, oligonucleotide in situ hybridization can be used
(FISH). You can actually see single molecules in a regular microscope in
this way.
----------------------------------
>
>That evaporate layers of tissue at what temperature?
>
### Up to 10 000 kelvin, same as in a commercial lasik or other tissue
machining system.
----------------------------
> How do you stop
>the tissue below the surface layer from heating up and information getting
>lost before you determine it?
>
>
### Using very short laser pulses of the correct frequency. Published
data describe no visible ultrastructural damage beyond a few hundred
nanometers with conventional lasik, and that is much less than the depth
of focus for a scanning confocal microscope.
--------------------------------------
>3D reconstruction onto what substrate?
>
### 3D computational model.
--------------------------------
>Surely not the same sort of
>organic substrate as originally - how would you put it together without
>it decomposing. And if on some other substrate how would you translate
>the infromation from the first substrate (unique info remember as memories
>can't be templated out) onto another substrate?
>
>
### What do you mean by "templated out"?
-------------------
>
>
>>All the above steps use existing technologies, and reasonable extensions
>>of them (e.g. the antibodies to all important molecules are not yet
>>available, but will be once the molecules are cataloged).
>>
>>
>
>You don't say how you will do important steps. You don't talk about
>algorithms
>for storing information,
>
### Algorithms? The intermediate scan data can be stored as any other
form of digital imaging data.
----------------------------------
> for translation that information into something
>that
>could be ported to another sunstrate.
>
>
### Explain?
------------------------
>Do you break the brain sized stating material down into smaller pieces?
>
>
### Do you mean if I work on the whole brain or first cut it up? No,
this method would not require prior dissection.
-----------------------
>How do you prevent loss of info due to cracking?
>
### Since nanoscale information about the structure of the crack
surfaces would be available, the cracked surfaces can be apposed in the
3D model, just like freeze-fractured cell surfaces can be matched in the
freeze-fracture scanning electron microscope.
------------------------------------------
> How do you move
>pieces around?
>
>
>
### Pieces?
------------------------
>>Then a sufficiently powerful computer will construct a neural network
>>replicating the connectivity pattern and the synaptic strengths, as well
>>as the rules of modification of the synaptic strengths in the course of
>>information processing, producing a device which will be behaviorally
>>sufficiently similar to me as to satisfy my desires regarding future
>>states of the world (to silence critics I don't even need to say that I
>>have been "reincarnated" or "brought back to life", or "survived", or
>>any such rigmarole - I only say that both I today and the device in the
>>future are satisfied with this particular outcome, and consider the
>>cryonics money well-spent).
>>
>>
>
>
>
>>Does this describe a sufficiently developed idea?
>>
>>
>
>No. It really, really doesn't. If you want numbers and technical analysis
>in a critique (see your statement above) then you'd have to do a lot more
>work yourself. And before you can put numbers on things you have to
>do more than just name a few potential tools.
>
### Wait - you wanted to have an outline of "how cryonics might work for
me", to prevent misunderstandings and ambiguities, and not discuss an
idea so insufficiently developed that it would start mutating under
analysis, like a religious concept (at least this is how I read your
question). The outline I gave suffices to define my proposed cryonics
approach. Now, if are you ready to give me a technical critique yourself
, tell me which additional data do you need, and which numbers that I
provided above and below do you disagree with.
---------------------------
>
>You haven't really engaged with *any* key "how to" engineering problems
>at all.
>
### I engaged:
- how to freeze
- how to collect information
and I hinted at how the information could be used. What other key
engineering problems do you see here?
-----------------------------------
>>I could come up with some ballpark
>>estimates of the file sizes, numbers of antibodies, speed of laser
>>machining, but these are all mere technical details (nothing that would
>>be unreasonably expensive or time-consuming, as far as I can tell).
>>
>>
>
>"as far as I can tell" is a very big statement. Unless you wade into the
>problem space with a serious engineering frame of mind you aren't even
>starting to think about the sort of engineering problems that you'd have
>to solve.
>
>
### OK, you have about 1700 cubic centimeters of tissue (actually less
if you can find methods for substituting generic information for some
subcortical brain areas not involved in personal episodic memories, the
main type of memories I want to have analyzed). At a resolution of 50
cubic nanometers (more than sufficient to show synapses) there would be
13.6 x 10^12 voxels. The imaging procedure would need to reliably image
probably no less than 70% of synaptic connections (given the known
redundancy of the brain, but Anders could perhaps give us more details),
and estimate the synaptic strengths to within 10% of their true values
(there is some degree of drift of synaptic strengths in the living
brain, which means you don't need to be very precise in measuring, but
exactly how much you can drift and still get largely the same behavioral
effects is unknown). The range of values for synaptic strength should
not be significantly above 8-bit depth. Given that there are many
hundreds of types of synapses, the actual number of bits per voxel might
be as high 20 to 25. Some of the data would be needed to describe
cellular membranous structures, as well as post-translationally modified
proteins. All in all the uncompressed image of the structural substrate
of personality might be on the order of 10^20 bit. This is only a few
thousand petabytes, not an unreasonable data volume to work on in 20
years from now. The computing capacity needed to transform this
uncompressed image into a functionally equivalent neural network
simulation is quite large by today's standards, but given the estimates
of Moravec, the processing power needed to actually run the neural
network is unlikely to be many orders of magnitude more than 100
teraflops/s, a capacity available even now.
Criticize technically now, please : )
Rafal
More information about the extropy-chat
mailing list