Re: Aaron Sloman's "The Irrelevance of Turing Machines to AI" article

From: Allan C Cybulskie (allan.c.cybulskie_at_yahoo.ca)
Date: 08/07/04 

Date: Sat, 7 Aug 2004 07:37:04 -0400

BTW, I seem to have lost the last post you made to me before I could reply
to it, so if you want something in it answered and can repost it that would
be great.

"Wolf Kirchmeir" <wwolfkir@sympatico.ca> wrote in message
news:haPPc.23036$Vm1.458145@news20.bellglobal.com...
> Sergio Navega wrote:
> But mostly, categorisation is language use. The question is, how do I
> get you to "understand" a new use of the language? How do I arrive at
> that new use? The first question is easier to answer than the second - I
> train you to use the language the same way I do, by, for example,
> "agreeing with what you say" in repsonse to my speech. **

This is an interesting view, but I think it's ultimately flawed. Think
about teaching someone a new language. The first thing you do is start with
simple words that you can relate to something they are already aware of. So
you either teach them what the word means in terms of a shared, existing
language, or terms of physical references to things really in the world.
The "agreeing with what you say" part really only is the final cemeted
confirmation that understanding has been achieved, and it is unnecessary for
learning. I can learn what a word means without anyone ever confirming my
understanding. At that point, it's just the case that I THINK I know what
it means, but I'm lacking the final confirmation of "Yes, that's right,
that's what it means". But this isn't required to actually understand a new
use of the language.

>
> The second question is harder. Introspection and reports on "how I feel"
> suggest there are unpleasant feelings attached to certain language uses,
> and that these feelings may change to pleasant ones when a new use is
> produced. Euphemisms seems to work this way, for example. (That is, the
> discriminator in this case is a feeling. That feelings act as
> reinforcers should be obvious.)

Certainly, feelings can act as reinforcers, but it would seem odd to say
that you couldn't understand or learn a new word without a feeling of
satisfaction about it since we seem to be able to pick up new words kind of
by osmosis. For example, I read a lot of books and so pick up a lot of new
words from their usage in them. But to my knowledge there's no feeling of
satisfaction or pleasure attached to my learning of the word while reading.
My attention is focused on enjoying the experience of the book, and the
words are just things that I come across in context and thus learn to
understand (often this is automatic and there's no struggle to understand
the word to catch the pleasure of the book). You might claim that this is
conditioning, but it is not reinforced conditioning in the sense that I
think you want it, and so I wonder where your view of "the discriminator"
fits in these cases.

>
> Metaphor is a little harder, but it seems to be a case of experimenting
> with different usages until that positive feeling is evoked. The fact
> that "poetry programs" can produce interesting and striking metaphors by
> semi-random combinations of words supports this explanation: we, the
> readers of thsoe random collocations, judge them as ointersting/etc. So
> do reports by poets, who say they "search for" the right
> phrase/image/etc, ie, they try out different words and phrases until one
> "sounds right." That's a remarkably behaviorist account, allowing for
> the non-technical terminology.

I think you are impressed by behaviourism so much because you give it WAY
too much credit for things that either it does not necessarily espouse, or
that it is not the only theory that can handle or posit such an explanation.

Metaphor in poetry as you've described is not necessarily behaviourist, and
I'd probably argue the exact converse. When a poet "tries out" words, all
that we have is a complete demonstration of the fact that ultimately the
effect of poetry is dependent entirely on the actual experience of the words
and lines. When a poet "tries out" metaphors, what they are doing is
looking for a word or set of words that "sounds right" in the sense of: a)
fits the rhythm of the poem (most peotry has fairly strict rules on rhythm)
and b) invokes the mental experience that best represents the point they are
trying to get across. You can't tell that without actually invoking the
experience, and that is what they try to do. And to judge which "sounds
right" is simply a relation from the image invoked to the image they wanted
to invoke. Now, all of this is clearly not behaviourist, but I fail to see
why it is not as acceptable -- if not more acceptable -- an explanation than
yours.

Relevant Pages