X-Message-Number: 21392
Date: Tue, 11 Mar 2003 23:26:55 -0700
From: Mike Perry <>
Subject: Symbols and Awareness

Thomas Donaldson writes:

>Your discussion of awareness in computers has several empty spots.
>First of all, where in the Turing test is awareness even discussed?

It isn't. I brought up the Turing Test because (1) a system that passes it 
at least will *seem* to have awareness, but also (2) to note that, in the 
example I considered, something *more* than passing the Turing test is 
involved.

>Second, you clearly believe that there is no valid test for
>awareness.

Depends on the circumstances. Also, if I am said to "believe" something, it 
is not the same as "believe dogmatically"--I could be wrong (yes, even more 
often than water freezes in the hot sun ;-).

>Yet we already have (very poor focus) ways to read
>brains as they work. This ability is improving. So who says I
>could not find out whether or not you were aware?

I'm not disputing that there could be ways of testing, "behind the scenes," 
whether a brain is showing awareness or not. The real question is whether 
some other system that seems, on the face of it, to be aware, really is so 
or not. Finding out how a brain works would not, in and of itself, prove 
that something that works differently could not be aware. Suppose we have a 
system, call it S, that simulates the brain at a deep level, though S 
itself is computational and symbolic. S, then, passes reasonable tests for 
awareness just like the real brain. Not only does it communicate, but 
internal brain structures and their associated functioning: neurons, 
synapses, molecular architectures, and so on, are all isomorphically 
reproduced. Again, though, S is "only" a (very advanced and sophisticated) 
computational device, and, at its *own* deepest level, is clearly not 
functioning just like a brain. But I maintain that, under the circumstances 
and barring some fundamental new discovery about reality, there would be no 
reasonable proof or compelling argument that S does not have true awareness.

>I'm not saying that our brains work by some magical process at
>all. They work by processes we are coming to understand; and
>their working is much more like the operation of a piston
>or the slow burning of a fire than like the inevitably symbolic
>operations of a computer.

You seem to draw a sharp distinction between "inevitably symbolic 
operations" and other processes such as fire or a piston's pushing. But 
consider this possibility. Nature herself is, at the deepest level, also 
"symbolic." Why would we suspect this? Because that seems to be the 
take-home message of quantum mechanics. Things happen in discrete jumps, 
rather as they do in a computer. So, even though fundamentally "unsymbolic" 
processing seems to exist, this is really only an illusion caused by the 
scale and complexity of what goes on. In the same way, a gas seems to be a 
smooth fluid, infinitely divisible, but we know it isn't that at all, but 
resolves into discrete units (molecules) on a fine scale.

>That is why I have said and are saying
>that computers by their nature cannot be aware.

I don't see it that way, and, if things are "symbolic" at the deepest level 
as suggested above, it especially undercuts your claim.

>...And so far as you can build a machine which works like a brain
>rather than a computer, working not with symbols at all but
>with the world,

Once again, though, it seems the world itself is just an ongoing 
computation using "symbols" of a sort; David Deutsch makes this point in 
*The Fabric of Reality*.

Another thought is that the term "symbol" itself may be a stumbling block 
for the points I've been trying to make, because we naturally think a 
symbol is something that has to "stand for" something else, that is to say, 
we have to assign a meaning. But really that's not so. Instead of "symbols" 
I could have used "letters"--understood to be chosen from a finite 
alphabet--or just "bits"--which are the most convenient since there are two 
letters in the alphabet and two only. Individually, a bit (0 or 1, say) can 
be called a "symbol" but it really doesn't have to mean anything, except 
that it is different from its opposite (0 and 1 are different that is). 
Here it's not the individual bits that have meaning, but the *patterns* 
they are formed into, and these patterns are not just arbitrary. We 
immediately note the difference, for example, between a pattern that looks 
random and one that is structured in some simple way, such as, for example, 
having all 0's or all 1's. The patterns may have assigned meanings, it's 
true, but could reasonably suggest their own meanings also, as if, for 
example, we were to represent the prime numbers according to some simple 
encoding. Other math concepts could similarly be represented, and you could 
go from there.

Mike Perry

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=21392