X-Message-Number: 9156
Date: Thu, 12 Feb 1998 09:34:28 -0800
From: "Joseph J. Strout" <>
Subject: Re: CryoNet #9153 (brainstuff)

In Cryomsg #9153, Thomas wrote:

>Your answer raises further questions. If this reconstructed machine into which
>your memories have somehow been implanted is at the same scale as you are,
>then I do not understand just why you believe we can work so finely (nano-
>finely, so to speak) with this machine but somehow cannot do the same with
>biological objects such as brains.

It depends on what you mean by "work with".  The relevant issue, I think,
is what can we construct?  It is rather easy to construct artificial
devices on a scale comparable to, or smaller, than, neurons; this is
because we can use whatever materials and techniques are best suited for
the job.  For example, we choose materials which are not squishy and
collapse under their own weight.  I have confidence that if we do not
constrain the engineers to particular materials or techniques, computers
will continue to get smaller -- and after all, we're talking about building
a specialized type of computer.

But to build a biological brain, out of proteins and lipids and water --
this is a very different thing!  Objects have to be built gradually, so how
exactly does half a cell maintain its shape?  You'd have to do build it
frozen, perhaps, but then you have the problem (well known in this forum)
of thawing it out without killing it.  The only way around it that anyone
has proposed (AFAIK), is to have Nanomachines infiltrated throughout the
entire thing, watching for ice crystals and repairing damage as fast as it
happens, while hopefully not doing even more damage themselves.

All this may be possible, but it is much Bigger Magic than simply building
an artificial device -- out of silicon, or carbon, or whatever works --
that is functionally equivalent.

(An alternative is to grow a brain, as is done in development, but then you
have very little control over where the connections form -- the result
would be a new brain, not a copy of the original.)

>One initial thought was the idea that
>perhaps you felt that we understood matter at that scale, for electronics
>purposes, well enough or close to well enough already, but did not understand
>how brains worked at that scale. However there is a problem there: we can
>hardly make an electronic machine containing your memories unless we know
>how to read off the memories inside brains --- and for cryonics purposes,
>damaged brains, too.

Yes.  I'm not saying this has anything to do with understanding the brain
vs. understanding electronics; I refer to purely practical problems of
building with such squishy materials.

>There is, just to add a bit of likelihood to this picture, a growing
>sense that long term memory formation occurs by the creation of new
>synapses and new connections.

Yes, this is my working assumption, which I believe is backed by a good
deal of evidence these days.  If it turns out that smaller-scale changes
are important for long-term memory, then uploading becomes more difficult.

>If we create a device capable of imitating your brain
>on the scale at which your memories are stored, then that device will have
>to have some ability to grow and change, and not in any fixed mode (ie.
>turn on connections already there, for instance).

Well, I wouldn't presume to say how a brain emulator must work in so much
detail.  An optical computer, for example, could create and destroy
connections in the form of light paths, but still have no moving parts.
But you may be right, and the best answer may turn out to be little moving
connections.  It would still be easier to build them out of more durable
stuff than biological tissue.

>So why is it that you believe we can make electronic (or light, or other)
>networks of connections to match that of our brain, but cannot do so with
>biological tissue? I know we can't do this now, but the question is that
>of what is possible IN PRINCIPLE.

I should reread my earlier message -- I never meant to say building a
biological brain from scratch is impossible even in principle.  Just that
it's likely to be far, far more difficult than building an equivalent brain
using unconstrained materials and techniques.

>Of course one major feature of biological tissue is that it is floppy, and
>wet, and moves around all the time (when it's alive). But then I find it hard
>to imagine remaking YOU (or me) into something which does not have such
>features. And the insights in Aplysia memory suggest that such behavior
>is essential to us, not just a side effect of the way we are made.

The connections are changing, which has functional significance, but there
are myriad ways the same functionality could be implemented.  Nature
developed using the most convenient tools available to it, which were
floppy bags of mostly water.  When we attack the same problem, we may find
more convenient tools are available to us.

>It does have its advantages, as you know --- even on a simple level. If I get
>dings and scratches they heal over, while if your car, television, or
>computer gets dings and scratches they stay there forever.

But you'll note that dings and scratches on a computer do not (generally)
affect its behavior.  And when it gets too old and beat up, you replace it
-- after copying all its information content to the new hardware, of course!

>And please be
>careful: I am not claiming here that creatures able to carry out such
>self repair must necessarily be composed of the same materials and systems
>as we are. But it IS an advantage, especially if we want to survive a
>long time.

I must shrug here -- as long as the information content is intact and can
be easily backed up and restored, then wear and tear are mostly irrelevant,
just as they are to computers today.  And here's perhaps the biggest reason
to use non-biological materials, even if we had big enough magic to build
wet brains: there may be no possible way to upload a wet brain without
destroying it.  So if you always have a wet brain, you cannot make live
backups, and you run the risk of dying permanently in a plane crash or
whatever.  (OK, in principle you could purposely destroy and rescan your
brain every few years, but it would be a much bigger chore...)  An
artificial brain, on the other hand, will no doubt be built with a simple
input/output procedure.  Weekly backups will be no problem, and if you want
to restore, or fax your brain pattern to Mars, etc., then it's a simple
matter of attaching a cable and pressing "download".

To summarize: yes, it may someday be possible to build a biological brain
to the exact specifications of a previous one, but I think it will be
extremely difficult and relatively pointless.

I wonder, though, whether your questions are idle curiousity about my view
-- or would you feel more comfortable, for some reason, in a biological
brain rather than one made of other stuff?  If so, is this a personal
identity issue, or do you simply doubt that other stuff could be made
functionally equivalent?

Best regards,
-- Joe

,------------------------------------------------------------------.
|    Joseph J. Strout           Department of Neuroscience, UCSD   |
|               http://www-acs.ucsd.edu/~jstrout/  |
`------------------------------------------------------------------'
        [ Help stop spam: http://www.imc.org/ube-sol.html ]

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=9156