Newsgroups: comp.ai.philosophy,sci.cognitive
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!uunet!in1.uu.net!cs.utexas.edu!utnut!utgpu!pindor
From: pindor@gpu.utcc.utoronto.ca (Andrzej Pindor)
Subject: Re: Grounding Representations
Message-ID: <D6BI9K.2rB@gpu.utcc.utoronto.ca>
Organization: UTCC Public Access
References: <harnad-1503952148320001@sm1.psy.soton.ac.uk> <pautler-2403950925390001@pautler.ils.nwu.edu> <D64BM4.7qF@gpu.utcc.utoronto.ca> <pautler-2703951710270001@pautler.ils.nwu.edu>
Date: Fri, 31 Mar 1995 18:34:32 GMT
Lines: 54
Xref: glinda.oz.cs.cmu.edu comp.ai.philosophy:26452 sci.cognitive:7128

In article <pautler-2703951710270001@pautler.ils.nwu.edu>,
David Pautler <pautler@ils.nwu.edu> wrote:
>In article <D64BM4.7qF@gpu.utcc.utoronto.ca>, pindor@gpu.utcc.utoronto.ca
>(Andrzej Pindor) wrote:
>
>> In article <pautler-2403950925390001@pautler.ils.nwu.edu>,
>> David Pautler <pautler@ils.nwu.edu> wrote:
>> >How many symbols do there have to be in such a universe for each of
>> >them to be grounded?  Two?  A hundred?  Four hundred thousand?
>> >What is it about crossing this numeric threshold that *causes*
>> >the symbols to be grounded?
>> >
>> In a universe describable by two symbols (a very simple universe indeed)
>> if the two symbols have some relation to each other, they are both fully
>> grounded.
>
>I think you are begging the question by using the term 'describable by'
>in the premise.  That is, if the universe is describable by only two
>symbols, it doesn't matter if they are related or not.  If the symbols
>can be said to describe, then they are necessarily grounded.

Please note that I've said "describable" not "described". Grounding is not an
abstract objective (whatever it might mean) property but requires a mind
which attempts to model the outside world.
The case of 'two symbol universe' was meant to indicate a limititing situation
opposite to a case of an infinite universe. 

>
>> The more 
>> relations a given symbol has to other symbols the better grounded it is.
>
>It sounds to me like an infinite regress of IOUs.
>
Why infinite regress? Rather a case of 'bootstraping'.

>Let me propose a possibly wildly misleading metaphor from groundedness
>to contentfulness: Think of all the Web pages that currently just
>point to other Web pages without adding "content".  Is a collection
>of content-less, pointer-rich Web pages contentful?
>
'Contentful' is not a yes or no attribute. Content of such a system would be 
determined by the richness of the structure of relationships. Think for 
instance about prinicipal component analysis as some way of making sense of 
relationships inherent in data. One problem with the case under discussion
is that dimensionality of the data is so unimaginably huge.

>     -dp-

Andrzej
-- 
Andrzej Pindor                        The foolish reject what they see and 
University of Toronto                 not what they think; the wise reject
Instructional and Research Computing  what they think and not what they see.
pindor@gpu.utcc.utoronto.ca                           Huang Po
