From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!sun-barr!ames!haven.umd.edu!darwin.sura.net!cs.ucf.edu!barros!gomez Mon May 25 14:06:32 EDT 1992
Article 5777 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!sun-barr!ames!haven.umd.edu!darwin.sura.net!cs.ucf.edu!barros!gomez
>From: gomez@barros.cs.ucf.edu (Fernando Gomez)
Subject: Grounding and Symbols
Message-ID: <1992May20.181548.7296@cs.ucf.edu>
Sender: news@cs.ucf.edu (News system)
Organization: University of Central Florida, Orlando
References: <60703@aurs01.UUCP> <78417@netnews.upenn.edu> <1992May20.170019.26095@kbsw1>
Date: Wed, 20 May 1992 18:15:48 GMT
Lines: 55

I have trouble with the idea of "grounding" as presented  by
Harnad  and others.  I read his paper in JTAI. The source of
my difficulty is that most symbols (I should say concepts  -
should   he?)   cannot  be  grounded  in  reality.  Consider
"bachelor." How do you ground this one? It is clear that you
do  not  find  out that somebody is a bachelor by looking at
his face, or touching him, etc.  Stimuli do not  help  here.
"Bachelor" gets its meaning from other symbols which in turn
are clearly ungroundable: "male" and "unmarried."  (Uhh!  On
second  thought, I think one could ground "male.") In under-
standing "bachelor" we are acting not much differently  from
Searle's  man  in the chinese room.  Now if you take another
familiar symbol  (concept),  say  "book,"  you  perhaps  may
ground  one  of  its  meaning, the physical object book, but
certainly you can not ground its other meaning: book as  the
set of ideas produced by some human. Here again at least one
of the senses of "book" obtains its  life  from  symbols  to
symbols.   (It  is interesting to observe the representation
tricks that such semmingly innoccous concepts such as "book"
play  on us! But, that will take us back to the Penrose dis-
cussion.)

What is meant by "grounding?" Is it what Quine calls empiri-
cal meaning?  Quine defines it as: "what remains when, given
discourse together with all its stimulatory  conditions,  we
peel  away  the verbiage," Meaning and Translation).  So, is
the strand connecting symbols to the physical world a clear,
more  or less straight line?  Let us Quine answer for us (we
are in good hands):

" ...Roughly imaginable sequences of nerve hits can  confirm
us  in  the  statement  that  there  is a brick house on Elm
Street, beyond the power of secondary associations to add or
detract.  Even where the conditioning to non-verbal stimula-
tion is so firm, however, there is no telling to what extent
it  is  original  and  to  what  extent  it  results  from a
shortcutting, by transitivity of conditioning, of  old  con-
nections  of  sentences with sentences" (Word and Object, p.
13, MIT Press)

I may add that not only the line that connects symbols (con-
cepts)  to  the physical world is spaghetti-like, but also a
thin one.  Probably, one could  ground  some  of  the  first
rudiments  of  human language, when it was forming. But, the
case is hopeless for, say, 20th  century  English  language.
And  of course, this is one of the problems with symbols. At
first sight, they seem almost nicely anchored in the  physi-
cal  world. But, soon, as in the sorcerer's apprentice, they
gain life of their own and take  off  leaving  the  physical
world  far, far behind.  You may want to ground them, or you
may want to reduce them to numerical weights; but sooner  or
later  they  cling  to you with a revenge claiming a life of
their own.

Fernando Gomez


