Selected Newsgroup Message
From: "Sergio Navega" <snavega@attglobal.net>
Subject: Re: Thinking without language?
Date: 20 Dec 1999 00:00:00 GMT
Message-ID: <83k5qi$mmf$1@mulga.cs.mu.OZ.AU>
Approved: ai@cs.mu.oz.au (Moderator comp.ai)
References: <80v9fc$741$1@mulga.cs.mu.OZ.AU> <83cglk$g3a$1@mulga.cs.mu.OZ.AU>
X-Date: Fri, 17 Dec 1999 09:46:05 -0200
X-Mod: ?
Organization: Intelliwise Research and Training
Followup-To: comp.ai
Newsgroups: comp.ai
steve phelps wrote in message <83cglk$g3a$1@mulga.cs.mu.OZ.AU>...
>I vote for 'we use language' whilst thinking. Perhaps we might believe we
>can think and make decisions even not so intelligent decisions but it must
>have some form of language. Perhaps the language isn't our native language
>but it's some form of symbols strung together formulating phrases. The
>symbolic phrases we use to make decisions.
>
But Steve, there are thoughts that cannot be done linguistically.
Here's an example that I recently wrote in a similar discussion
in comp.ai.philosophy:
> Draw in your mind a square. From the upper right corner, draw a
> diagonal that meets the lower left corner. Now take the middle
> point of this diagonal and from it drop a vertical line to the
> base (inferior) side of the square. Take the middle point of this
> vertical line and draw a straight horizontal line over it, extending
> all the way over both sides. Now answer: how many lines does this
> horizontal line crosses? The answer (4) cannot be obtained by
> purely linguistic methods.
I think this idea was first coined by Herb Simon. There are two
currents today to explain what happens: Pylyshyn and Kosslyn. Zenon
Pylyshyn says that we solve this kind of problem propositionally,
which means, the image is 'transformed' into propositions and then
inference is applied over those propositions. This appears to
suggest that we think propositionally even when doing visual
thinking, which could support the idea of "all linguistically"
related activities.
However, Stephen Kosslyn argues that we deal with pictorial
representations of images and that we process over these pictorial
representations. I find Kosslyn's arguments more convincing and
more directly supported by available evidences (see his book
'Images and Brain').
>Infants during the stage of not having acquired a language use a form of
>language. It's very rudimentary yet it's a language. You may check
>Chomski's research further in this matter.
>
The issue here is even more convoluted. While there are some that
defend strong innatist positions for language, I prefer to side
myself with those who are not so radical.
The important point here is that 'semantics precedes syntax'. No
child can solidly learn grammar without having sensory notions of the
concepts involved. Before uttering her first words (around 18 months
or less) a child is full of sensory (and semantic) notions of her
world. These notions are obviously a form of knowledge and are
not linguistic!
This is very important to AI, because an intelligent system should
also start this way. Any natural language processing program that
intends to be useful and human-equivalent should have a solid
semantic ground in which to plant grammatical structures.
>So if we have a basic form of language when we come into this world, it's
>easily understandable why humans created a language, it's innate. I think
>it's in our nature to use a language whilst thinking.
Innateness of language is still a highly debated issue, with a chasm
between the two competing sects. In my view of the problem, we're
constantly fooled to think that we reason in linguistic terms. But
some studies of cognitive psychology show that we have a huge amount
of processes going on below the conscious level (these are topics
explored in a discipline known as 'Implicit Learning').
What appears reasonable to us may be only a small and incomplete vision
of the whole process. I suggest that most of what symbolic AI did in
the past decades was the result of such misleading introspection.
> Furthermore
>decisions probably have a different level of language. There is a
>difference in making a decision based on fear (less lingual thinking :) and
>making a well informed decision either way some form of symbolic thoughts
>occur.
>
>But as for lower life forms, it's not neccessary. I think that's where the
>ambiguity comes from. Language is very natural to us; it would be very
>unnatural to try to think without formulating some level of a language.
>
Neuroscience is showing us that the thing is a bit different than this.
When you hear 'Grab that ball over the table', there is a huge activity
in your brain before you start commanding your muscles. Interestingly,
if you're previously instructed not to act, just to listen to the phrase,
your motor cortex will show signs of activity, even without moving a
finger. Just "thinking" of grabbing something makes you use several
of the areas required for the physical action to take place. Language
is important, but it is the only visible part of the cognition. There's
a whole world under that rug.
Regards,
________________________________________________________
Sergio Navega
Intelliwise Research
http://www.intelliwise.com/snavega
---
[ comp.ai is moderated. To submit, just post and be patient, or if ]
[ that fails mail your article to <comp-ai@moderators.isc.org>, and ]
[ ask your news administrator to fix the problems with your system. ]
Back to Menu of Messages Sergio's Homepage
Any Comments? snavega@attglobal.net