sn13+@andrew.cmu.edu (S. Narasimhan) (02/08/90)
There have been a lot of postings in comp.ai regarding symbols and semantics.The more I think about what semantics really means the more I'm convinced that they are nothing but symbols.What the number 1 is to human is the same to a
computer too. It is only a symbol for both of them. Then why should 1 mean more to us than to a machine? But does it? Atleast to me it does. No.1 means a lot
of things. It could the number of my house, my position in an organisation and so on. But for a computer 1 is just that - only one.So I'd say 1 is semantically
meaningful to me, but not so for a computer.
But, is the above argument a fair one ? Obviously not. When we say number 1
we immediately attach a group of symbols to it which according to our notion
has a meaning. But for the machine it is a single symbol with no other symbols
attached.
be
This can shown in another way. Can we write down on a paper what the
symbol 1 means^ to us. ? Surely we can. If we can write down semantics in
terms of symbols then they are nothing but group of symbols.In fact if my
sentences in the previous paragraph related to you the semantics I attach
to the number 1 I have obviously shown that semantics are symbols.
An immediate reaction to this would be : Symbols express semantics but
cannot be semantics themselves.My answer would be : How come the symbols
can express semantics? Imagine this : As you read these lines the input
to you is only symbols in terms of characters. How is it that you attach
a meaning to these characters? The answer is simple. You do that by attaching
to these symbols other symbols that you think are relevant to these symbols.
But what does one mean by relevance? Relevance in a sense means similar semantics. Is 1 relevant to 1 ? No, for they are just symbols.Is first relevant to 1?
Yes. Why ? Since the symbol "first" has semantics associated with it we say they are relevant.
Thus, semantics are group of symbols which are relevant to one another.In turn, we say two symbols are relevant if they have similar semantics. This is a
hopeless recursive definition. Why does this occur? Because we think semantics
are different from symbols.
Also note that relevance is subjective and so is semantics. I'd claim that
the symbols are retrieved first and then we reason about the semantics. In
other words what I'm saying is that remindings that occur in a person are the
CAUSE for semantics and the not the vice-versa. Remindings are nothing but
symbols. The input symbols cause symbol retrieval from the memory and the
reasoning part of the brain invents a semantics for this.That is,it attaches
new symbols to the already existing group to form a new symbol group and
thought results or rather a chain of thought results.
But yet how does the brain know what to retrieve ? Consider the following
algorithm:
1. Read input symbol, say A.
2. Try to retrieve a symbol from memory, say B.
3. If nothing is retrieved, reject previously stored symbol.
and go to 2.
4. If something is retrieved store the previous symbol.
5. Include the newly retrieved symbol in the symbol group.
6. Go to 2.
Narasimhan.