X-Message-Number: 3766
From: 
Date: Fri, 27 Jan 1995 19:48:33 -0500
Subject: SCI. CRYONICS brain as net

It is heartening that neuroscientist Joseph Strout (Cryonet #3754) has
understood at least
part of my attempt to clarify what I mean by the "self circuit" and why it is
important--although not yet why I say it MAY not be possible to "upload" it
into (say) a silicon-based computer.

Since many readers are bright and well informed and usually open-minded
(Confucius say, man with hole in head have open mind) , and since I have
reason to believe my writing is usually clear, I can only attribute the
comunication difficulty to the extreme slipperiness of the subject matter and
the recalcitrance of certain mind-sets. Well, let me try yet again to clarify
the reason we cannot ASSUME that a mind, or some subset of it--especially the
self circuit--could be realized in an arbitrary medium.

1. First an example. It may (for some people) not be well chosen, and
analogies are always tricky, but let's try it anyway. (I posted something
like this recently, but it apparently didn't go through.)

THE SOUL OF A BULLDOZER:

We might with some plausibility say that the the heart or soul of a
bulldozer--its most essential or characteristic part--is the blade. Of course
it is useless and helpless without its platform and power source and guidance
etc.; but it is the blade that sets it apart from other self-propelled
machinery (or robots; let's think of a roboticized or computerized
bulldozer).

Now, you can't realize a bulldozer blade in tissue paper. Maybe the guidance
algorithm--perhaps it could be a Turing tape--but not the blade, which
requires steel or something else hard and tough. No matter how thoroughly you
understand the nature and working of bulldozer blades, if you have nothing to
work with but tissue paper you are not going to build one (or the platform or
engine either).

2. WHAT'S IN A BRAIN?

Many will instantly reject the bulldozer analogy by claiming that there is no
parallel, that a bulldozer pushes dirt and a brain pushes information--and
that all information processors are essentially equivalent or
interchangeable, at least in potential. But this--brain as just an
information processor--is only an UNPROVEN ASSUMPTION. 

The assumption is seductively plausible, since much of what the brain does
(outside of housekeeping) is indeed information processing.  This notion is
reinforced by the common practice in recent decades of calling computers
"electronic brains," and by the attempts to make computers imitate some of
the activities of brains, including parallel processing, trial-and-error
learning, etc. Nevertheless, to say that the brain is
essentially "nothing but" a computer or "nothing but" an information
processor is simply a false premise, or at the very least an unjustified
premise.

Of course, if you accept this premise, then you automatically accept the
conclusion, that a mind could be uploaded into a computer. The conclusion is
little more than a restatement of the premise.

3. THE ISOMORPHISM FALLACY: This is repetitive, but apparently necessary for
many people. 

Is an analog (an isomorph) as good as the original? Depends on what you want
it to be good for--and how far the isomorphism extends. 

In some respects, a map is BETTER than the territory--easier to understand,
usually, if you are looking for routes. A blueprint is better than a
building, if your purpose is to begin to organize the construction of a
similar building. An electric circuit schematic is better than the circuit
itself, if you are calculating its characteristics or learning to understand
it.. 

On the other hand, you can't drive on a map or live in a blueprint or light a
house with a schematic. 

All this is so obvious that it may sound patronizing; yet it seems to be
necessary as a prelude to the crunch.

4. SIMULATION AND EMULATION

"Simulation" is often used to mean imitation or mimicry of SOME functions of
a system; "emulation" usually means mimicry of ALL functions of the
system--or at least all the functions that are important in the context of
the discussion. 

A silicon-based computer can certainly simulate some functions of the brain.
Could it simulate ALL of them SIMULTANEOUSLY? 

There is NOTHING that can emulate an iron atom--do EVERYTHING an iron atom
does. In fact, doing EVERYTHING an iron atom does is what DEFINES the iron
atom. 

If you claim to be able--even potentially--to emulate a human brain in
silicon, then you are claiming to KNOW all of the important functions of that
brain. Perhaps you respond that, while you don't know all the brain's
functions, you do have some general knowledge of their character,
specifically that they constitute information processing.  But now we are
back to postulates that are not at all self evident.

A computer/program that describes the nature of a water drop, and predicts
its behavior, nevertheless is not wet; one that describes a flame is not hot,
etc. The partial analog cannot be substituted for the thing itself--at least
not for all purposes.

5. BEING AND DOING

John Clark and others say any system that "does me" IS me. They mean
essentially that our minds are our brain processes, hence any system that
imitates my brain processes with sufficient fidelity must be me. This is
slippery language--plausible-sounding, but  deceptive.

Suppose e.g. that part of your brain (and this may even be true in some
animals) depends on iron atoms (in magnetite) to produce or/and detect
magnetic fields. If an important part of you will not work without some tiny
magnets in exactly the right configurations, then to some extent those
magnets ARE you and you ARE those magnets. Process is NOT necessarily
everything; substance may be something too.

Maybe the self circuit is a little like a standing wave, but of a very
special kind, capable of novel feedbacks and responses to stimuli. It may be
CRITICAL for it to have very specific dimensions, frequencies, response
times, interface capabilities--who knows what. That something very
different--a silicon-based computer--could fill the bill is not even
sensible, let alone self-evident.

Of course, the die-hard uploader will merely fall back on his first response:
"All that matters is the information processing." But that is only a mantra,
not an answer.

6. THE BRAIN AS NEURAL NET

Dr. Strout assumes that the essential brain is in the neurons and their
interconnections and interactions. "If a brain is reconstructed neuron for
neuron...then we have realized
the brain device in electronic.....media."

Conceivably, sure--but not necessarily. Perhaps an electronic simulation
could, indeed, do everything important--but we don't know. Maybe there are
important brain features outside of the neural net. (After all, lower animals
that have no neurons sometimes exhibit behavior suggestive of feelings.) 

Suppose we aren't dealing merely with a "network" that sends simple signals
from one node to the next, opening and closing logic gates. Suppose the
chemistry is important, as well as the electricity and magnetism. Suppose the
SHAPE of the neuron is important and changeable. 

GOING TO E XTREMES

Of course, the die-hard uploader (I am not suggesting that Dr. Strout is one)
will again say that we can simulate all that stuff--and now we are back to
the iron atom and emulation vs. simulation etc. Well, one last reminder that
this road leads to Hans Moravec.

If I understand him correctly, Dr. Moravec--the ultimate uploader--thinks we
are essentially abstractions. If information is everything, and can be
represented in an arbitrary medium, then it exists/has existed/will exist
with or without the presence of matter of any kind. If you are just a pattern
of activity, then you have existed always and everywhere, at least in
potential (somewhat as the number 5 "exists" or a triangle "exists" or a
 sine wave "exists")  and will continue to do so, no matter what happens to
your meat. A book or algorithm describing an iron atom "is" an iron atom. It
doesn't even matter if the book has been written yet; after all, it could
be!..........My suggestion: Don't give up your meat.

As usual, a  lot of rambling, and probably minimal impact, but it will look
better in the book.

Robert Ettinger
Cryonics Institute
Immortalist Society




 








































Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=3766