Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!elroy.jpl.nasa.gov!ames!amdahl!kp
From: kp@uts.amdahl.com (Ken Presting)
Newsgroups: comp.ai
Subject: Re: Chess, Reductionism, Probablistic Determinism.
Summary: How Interpretations Span Ritual Abstractions
Keywords: normative, abstraction, interpretation
Message-ID: <16ai029B9byy01@amdahl.uts.amdahl.com>
Date: 9 Apr 90 21:49:12 GMT
References: <491fffd5.1a4d7@cicada.engin.umich.edu> <2080@aipna.ed.ac.uk> <365@ntpdvp1.UUCP> <366@ntpdvp1.UUCP>
Reply-To: kp@amdahl.uts.amdahl.com (Ken Presting)
Organization: Amdahl Corporation, Sunnyvale CA
Lines: 169

In article <366@ntpdvp1.UUCP> sandyz@ntpdvp1.UUCP (Sandy Zinn) writes:
>> (Ken Presting) responds:
>> > Fantasy and ritual are the appropriate models for thought . . .
>
>Why are rituals more appropriate?  Because they are phenomenological,
>as opposed to being logical?  This is what I think you are saying.

I got such a hard time for so long from Stephen about excessive formalism
that I decided if I'm going to get pounded for having a silly and
simplistic opinion, I might as well take a pounding for the silly opinion
that I actually do hold!  (The truth will out ...)

The attribute of rituals that gets my attention is their *concreteness*,
combined with their *futility*.  (Ahh, that explains everything! :-)

Not that all rituals are always futile - this is precisely why ritual
is fascinating.  Why would grown organisms with important things to
do like gathering nuts and berries, and dragging off the females (:-)
engage in behavior with *no discernable practical consequences*?  This
has got to be, without question, the strangest phenomenon in the animal
kingdom.  (I include posting netnews within this phenomenon - no joke)

To anticipate Sandy's later remark, I see no alternative to the
hypothesis that symbol manipulation originated historically in ritual
behavior, and that what we today describe as symbol manipulation remains
an enactment of a ritual.

I should admit that I am overstating a point for effect when I say "NO
practical consequences".  If there were no selective advantage to having
a tendency to ritual, then the tendency would probably disappear.  But
selection for a tendency does NOT entail that all consequences of having
the tendency are advantageous.  The gene which confers resistance to
malaria in West African populations also confers susceptibility to
sickle-cell disease.  It is an elementary fallacy to suppose that every
trait in a *phenotype* confers a selective advantage, and there are many
subtler fallacies lurking for selectionist explanations.

Having said this, I have finally made good on a promise I gave back
during the last big Chinese Room debate - I've finally defined "symbol
manipulation".  My definition does have the virtue of blasting Searle
out of the water, but I will understand if some readers continue to
reserve judgement!  (Posted or e-mailed objections requested, natch)


(Sandy makes her first mistake:)
> . . .  You are a superlative logician; . . .

In Sandy's defense, I should note that she must have composed this article
before I claimed that logic and calculation are also rituals.  (:-)

I did once take a course from Leon Henkin, but was utterly defeated by
ultrafilters.  (Don't ask me what they are)

>... a propositional mode, which elevates the concepts of separateness,
>levels, abstractions, distinctions of types, etc.

This brings up some interesting points.  I would say that *arguments*
*must* be conducted in terms of propositions, their relations, etc.  This
has very important implications for the symbolist vs. connectionist
issue.  I would still deny that any non-ritual thought is propositional,
that is, any thought that is not related to argumentation.  But if the
practical criteria for *evidence* of thought is to be communication,
then the capacity to apply recursive rules (which may be difficult for
NN's) is essential.


With regard to levels:  In order to be able to discuss the truth of
propositions, and other semantic issues, it is indispensable to use
seperate languages (ie abstractions), and let one of them refer to the
other.  If this is not done, then the discussion is conducted in a
semantically closed language, and argument is pointless.  (Other than
this object language/metalanguage distinction between levels, I would
de-emphasize any hierarchy of abstractions.)

This is not a pedantic triviality.  The Greeks had a terrible time with
arguments that depended on paradoxes to trip up opponents.  Some of these
were (more or less) well intended, as in the case of Zeno. But to conduct
an argument that reached an interesting conclusion (ie, anything other
than "change remains the same and changes" et. al.) required dedicated
collaboration between disputants.  Schools of thought tended to be more
like religions than intellectual investigations.

Aristotle finally cleared up the mess with the logic of syllogisms, which
are adequate for most of geometry and some number theory, but not for
as much mathematics as we do in Fregean logics.  The point is that some
set of rules is indispensable to control communication in disagreement,
or disagreement will become inescapable.

IMO, this is the beginnings of a justification of the survival value of
the logic ritual, part of an explanation of why we continue to perform it,
part of the motivation for learning the ritual, and most importantly,
for *internalizing* it.  Thought *is posterior* to negotiation.

(Simplified diagram:)
                       X      X
             .........X......X......
               .     X      X    .
              .     X      X    .
             .     X      X    .
            ......X......X..........
                 X      X

>Those X-bars are  your darned normative properties, which I, living in
>dotland, believe are really just more dots,
>just like the dot-planes (levels of abstraction),

Each normative property is part of an abstraction all of its own.  Truth
is a good example - it is part of the whole abstraction of Logic.  A
normative property is always exemplified by the dots, but can never be
defined in terms of dots.  Again, truth is the perfect example - truth
is not definable in terms of syntactic properties (Tarksi's Theorem).

>just like the dot-planes (levels of abstraction), which as you see I can
>move between, with or without your normative properties (though I will
>concede that all my bridges may well be labeled as such).

The dot-bridges correspond to *implementations*.  The relation between
normative properties and implementations is very close.  Both depend on
homomorphisms of logical structure.  But normative properties are more
"fluid" than implemented properties, because a normative property is most
important when it is *imperfectly* exemplified.  For example, people's
assertions are never the whole truth and nothing but.  Normatives
depend on *interpretations* which are always tentative.

>
>In dotland, all the patterns I see seem to pulse in this rhythm: 
>    propagation, selection     propagation, selection
>
>The resulting patterns are very different, but the essential processing 
>is the same.  Between levels, within levels, outside the levels: patterns
>begetting patterns.

Natural selection is an excellent analogy for the process of
interpretation, because the propagation is always tentative.  I think it
is undeniable that a process with similar logical structure is
responsible for scientific progress, though the degree of similarity is
open to question.

Also, I am personally convinced that some similar processes occur in the
brain during (at least) early cognitive development, as well as during
creative thought, throughout life.  I think there is a very suggestive
analogy between the operation of non-deterministic automata and, say,
the proliferation of antibodies.  How far this analogy can be pressed
is a good question.

>                  Maybe I like the idea of emergence because a syncretic
>reality sees levels as a coalescence phenomenon, whereas a propositional
>reality has built-in levels, and thus sees "emergence" as silly.

You're confusing me with a logician again!  Bertrand Russell suggested
that reality was composed of "facts".  I think reality is composed of
us and the rest of the stuff.  Whatever it is, I don't care, as long as
there's only one kind of stuff, and we are made of it.  I call this view
"indifferent monism".

I see emergence as unnecessary.  Normative properties can be clearly
defined, argued about, calculated, and are lots of fun to be with.
Emergence is vaguely slimy, or slimily vague, and I just don't trust it.


>You wrote that the relation between description and perception is fantasy.
>Literally. -- Let me propose that dotland is fantasyland.  Homomorphs,
>isomorphs, allomorphs -- they're all here, wildly procreative but lawfully
>selective.

I'm confused about this, I guess.  Where is Reality?


Ken Presting   ("I can't these squirms out of my mind")