Re: Schyns Comms 10-12 french gauker grossberg

From: HARNAD Stevan (harnad@coglit.soton.ac.uk)
Date: Sun May 03 1998 - 16:09:55 BST


> From: Beck Alexandra <aeb195@psy.soton.ac.uk>
>
> Schyns demonstrates which French & Weaver acknowledge:
>
> > that there is a aquisition of new features during category learning.
>
> French & Weaver do not follow that the aquisition of new features is a
> typical part of category learning. French & Weaver argue that the
> widespread assumption of fixed feature catagory learning experiment is
> supported in the general case of adult category learning.

Are we sure that even adult category learning does not require the
construction of new features? After all, people are typically adults
when they start to learn to sex chickens, classify biopsy slides in
hospitals, and so on.

> 11. Gauker
>
> Schyns theory argues that the concepts of the features need not be
> developmentally more basic. They don't explain how novel features might
> be formed.

Kid-sib does not understand what you mean.

> Yet all agree that on one thing :
>
> > a theory of conceptual development has
> > to explain how children learn to apply
> > category words to roughly the same sort
> > of things as others who have learned language.
>
> No two people's abstract general concepts form exactly the same class
> of instances. Yet people agree on categories. These categories, it is
> being argued, may be innate or shared principles of abstraction.

What are "principles of abstraction"?

People sort and name things the same way, even though they didn't learn
to do it on exactly the same examples. Fine. So how do they do it?

You have to be kid-sib not only in how you explain to us what the commentator
said, but also in how you read the commentator: if it does not make
sense, say so!

> 12. Grossberg
>
> Prototypes are used to match other objects. If the matched object is
> good enough the system learns to incorporate a novel boundary of
> groupings into learned categories.

Are all or even most category members really just better and worse
"distortions" of of one and the same prototype? Think of examples of
categories -- living and nonliving, natural and artificial, concrete
and abstract; use a dictionary if need be. But test whether all or most
could be recognised as variants of a prototype.

On the other hand, if a "prototype" is just a set of features, and the
system just detects those features, then how does Grossberg's theory
agree or disagree with Schyns on fixed vs. flexible fetaures (or with the
underlying "Whorfian" vs. "NonWhorfian" view of categories)?

--------------------------------------------------------------------
HARNAD Stevan harnad@cogsci.soton.ac.uk
Professor of Psychology harnad@princeton.edu
Director, phone: +44 1703 592582
Cognitive Sciences Centre fax: +44 1703 594597
     
Department of Psychology http://www.cogsci.soton.ac.uk/~harnad/
University of Southampton http://www.princeton.edu/~harnad/
Highfield, Southampton ftp://ftp.princeton.edu/pub/harnad/
SO17 1BJ UNITED KINGDOM ftp://cogsci.soton.ac.uk/pub/harnad/



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:20 GMT