Newsgroups: comp.ai.philosophy
Path: utzoo!utgpu!watserv1!watdragon!mpdevine
From: mpdevine@watdragon.waterloo.edu (Michel P. Devine)
Subject: Introspection
Message-ID: <1991Feb4.224101.18846@watdragon.waterloo.edu>
Organization: University of Waterloo
Date: Mon, 4 Feb 91 22:41:01 GMT
Lines: 110


In article <3035@yarra-glen.aaii.oz.au> dnk@yarra-glen.aaii.oz.au (David Kinny) writes:
>What is it that leads you to believe that introspection is one of
>the greatest sources of information about our cognitive processes ?
>I would regard it as a most unreliable source of information about the
>true nature of those processes.  

The objective study of aspects of cognition, while important and possibly 
beneficial, differs substantially from the experience which is by necessity 
purely subjective.  The attempt to separate the observed phenomena from
the observer works well in domains where pure deterministic reasoning is
applicable, but my experience is that people and even animals are 
infuriatingly unpredictable, except at the coarsest level of functioning
(for example, living beings run away from danger) and therefore the 
black box or machine model of life may be inappropriate.

I should preface my comments on this important topic with a little background.
I once held with the AI clan with the fanaticism of the hardcore believer,
a "faith" born of despair at the failure of philosophy and religion to
enlighten me on the true nature of the universe.  Viewing the world as a 
Great Machine is comfortable for the cognoscenti and leads to a feeling
of vast power and (usually) superiority over less illuminated and gullible
masses.  My fall from grace came about with my attempt to make music 
intelligible to computers.  After consulting scores of so-called music
experts, it became increasingly clear that, in contradiction with their claims, 
they were most certainly *not* in the possession of clear-cut rules for music
appreciation.  The resulting angst built over the span of years, a gnawing
suspicion, then a fear, finally terror, that I was WRONG in a basic way. 
I should be careful to point out that I do not repudiate the machine
model in toto, but that I consider it incomplete.  It is true that it is 
possible to assign specialized functions to parts of the brain, but it is 
pure speculation to conclude that we are purely mechanistic.  The destruction 
of my most dearly held beliefs has been cataclysmic but necessary (I should 
note that the process is still ongoing).

Perhaps the collapse of the hierarchy of so-called "Levels of Being" is 
at fault; I think the major weakness of AI research is essentially an 
optimistic and enthusiastic over-simplification of nature.  
I will very briefly present a simple breakdown of the aforementioned LoB: 

1. Material 
2. Life
3. Consciousness
4. Self-awareness

Most "classical" philosophical systems make use of some such structure to
explain nature.  Plants are typically thought to reside at level 2, animals
at level 3 and humans at 4, although it may be argued that dolphins may
also qualify (how would anyone verify or discount this?).  Self-awareness
amounts to separating the program from the programmer: we can change 
our opinions, our beliefs at any time.

The levels represent qualitative differences in thinking ability or 
"intelligence".  For example, the ability of plants to orient their 
photoreceptors with the sun seems to be the action of some kind of
thought, although not necessarily consciousness.  Animals seem to be free
of most neuroses by virtue of their lack of self-consciousness, living
entirely in the present, but without the ability to forecast the effect
of their actions.  My point is that the only reality to which we have direct
access is our own, and even then from a very limited vantage point, namely
we may come to know ourselves, but we shall never truly know anyone else.
It is our predilection as scientists to search for all-encompassing
paradigms, explanations and simplifications.  Assuming that we are nothing 
more than ambulatory formal systems leads to logical contradiction as pointed
out in a previous posting.
f
It seems therefore only rational to conclude that our basic assumption is 
incorrect, or at least dubious.  So, what is the truth?  I don't know, but
I am not satisfied with the purely mechanical answer.  It does not address
the most mysterious aspects of intelligence, namely intuition, creativity
and so on, and therefore it is naive, and in fact degrading.  Who really
wants to believe that he is no more than a complex automaton?  No AI
researcher truly applies such reasoning to himself, his family or co-workers
which thereby indicates a split between belief and action that is most telling..

>It seems to me that introspection produces subjective and unreliable
>information about a tiny subset of our higher level cognitive processes,
>If we could understand the workings of a lizard mind it would be a
>major step towards understanding human cognitive processes.
>

I don't understand how you can relate the lizard mind to the human mind.
I can I know how lizards think, without becoming a lizard?  Certainly,
I can dissect lizards, make a list of parts and try to assemble one
from my kit, but is it possible that I have thereby misplaced the crucial 
ingredient "life"?  On a related note, isn't it interesting that our medical
experts pursue the understanding of life solely from studying dead material?
Perhaps that is why most hospitals are rotten places to go when one is sick...

I think that introspection provides subjective and crucial information about
a process most of us are almost studiously ignoring.  Introspection is the 
*only* way to know for certain what is really going on in your brain, 
whether it agrees with any given theory or not.  
Most of our activity is below our level of awareness, but it is possible 
(I am tempted to say "essential") to become *more* conscious, to acquire
more control.  

I should stress that I have barely skimmed the surface of objections to the 
AI viewpoint regarding its ultimate aim.  There exist many books that one
can consult, many techniques which with to experiment.  I recommend E.F.
Schumacher's "Guide to the Perplexed" as a good starting point, since it
fleshes out my argument and transcends it. Ken Wilber's books are also
useful. 

Michel Devine
-- 
--------------------------------------------------------------------------------
mpdevine@watdragon.waterloo.{edu|csn}            (519) 884-7123 Michel P. Devine
mpdevine@watdragon.uwaterloo.ca				   CS Dept., U. Waterloo
{uunet|utzoo|decvax|utai}!watmath!watdragon!mpdevine       Waterloo, Ont. N2L3G1
