Re: The Frame Problem

From: Dye, Matt (mdye@cogsci.soton.ac.uk)
Date: Wed Mar 05 1997 - 14:50:34 GMT


> Christos wrote:
>
> Phin, maths is a man made phenomenon, hence if you are
> taught only part of the whole, this will be the part you
> will be able to deal with. It's like making up a game, not
> telling you all the rules and expecting you to be able to
> play it. Having some grasp of maths however it would
> probably be possible for you to grasp the problems and
> work out the meaning of the symbols given enough time and
> practice. So in a way I believe it is similar to the frame
> problem but does not indicate that we are cognitively bound
> by a "frame", just that we do not possess knowledge of all
> man made rule systems. Another example not being able to
> automatically understand all languages, you still have the
> ability to learn, so expand your "frame".

OK. SO why is the "frame problem" just that - a problem. If we also
suffer from it, why is it a problem that computers suffer too?

Think about all the things you know.

2+2=4

My name is Christos

When people yawn it can mean they are tired or bored

If it's a lecture at 9 in the morning, it could be tiredness or boredom

If the lecture is in the afternoon and is on statistics, it is more
likely to be boredom

And so on, and so on .... you know an awful lot of things. How would
you go about giving a computer all this information ... expressing
everything in terms of symbols and rules? How long would it take? Is
it the case that everything you know you were told and just remembered?
If you were to be told everything you know in the same way as a symbolic
computer needs to be told, would you have been able to know all the
millions of things you know by the age you are now?

It seems to me very unlikely that I could have acquired the knowledge I
have now just by being taught millions of rules. What do others think?

This is the essence of the frame problem. Also think about how many
times you "trip up" - that is are unable to respond or give an answer to
something. Usually we can make inferences and have an educated guess.
Could a computer whose knowledge is just in the form of rules we have
given it do these things?

Matt
_________________________________________________
Matthew Dye

Department of Psychology
University of Southampton
Highfield
SOUTHAMPTON SO17 1BJ
U.K.

Tel. +44 (0)1703 594584 (Daytime)
Tel. +44 (0)1703 346906 (Evenings and Weekends)

Email: see header
WorkWeb: http://www.soton.ac.uk/~coglab/coglab/matt.html
FunWeb: http://www.soton.ac.uk/~mwgd/
________________________________________________



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:51 GMT