Path: utzoo!attcan!uunet!lll-winken!ames!nrl-cmf!ukma!rutgers!att!whuts!homxb!genesis!hotlr!dave
From: dave@hotlr.ATT ( C D Druitt hotlk)
Newsgroups: comp.ai
Subject: Re: Are whales fish? (was Re: Cognition vs formalizations)
Message-ID: <519@hotlr.ATT>
Date: 16 Jan 89 21:31:44 GMT
References: <681@cogsci.ucsd.EDU> <2959@uhccux.uhcc.hawaii.edu> <965@husc6.harvard.edu> <179@calmasd.GE.COM> <47459@yale-celray.yale.UUCP> <31266@tut.cis.ohio-state.edu>
Reply-To: dave@hotlr.UUCP (54246 - C D Druitt hotlk)
Organization: AT&T-BL Holmdel NJ - Lab 5431
Lines: 31

In article <31266@tut.cis.ohio-state.edu> tanner@saqqara.cis.ohio-state.edu.UUCP (Mike Tanner) writes:
 > In article <47459@yale-celray.yale.UUCP> Krulwich-Bruce@cs.yale.edu (Bruce Krulwich) writes:
 > >In article <179@calmasd.GE.COM>, wlp@calmasd (Walter L. Peterson, Jr.) writes:
 > >>biologist are in
 > >>agreement as to what creatures are members of the Class Aves.
 > >Maybe a historian on the net can describe studies showing that before the age
 > >of biological classification people didn't use the word "bird" to refer to a
 > >class of animals.
 > length on whether or not to call whales fish.  He says that whalemen, and
 > pretty much everybody else, call them fish.  In fact the only people who care
 > 
 > Is this the distinction being made here?  There are formally defined
 > categories and everday common sense ones.  A classification task that is hard

To take this sense even further: in the UK, a "bird" is an attractive
human female. In the USA, a bird is a scatalogical gesture. In the USa,
a "fish" can be a female (i.e. many fish in the sea) and a "bird" can
be an airplane or a rocket. BTW, the mouse's biological relative is the
elephant.
Seems to me that both cognitive and formal categories need to be modularized
so that, as with so many other "systems", there can be a layered or
hierarchical approach. Then the only real machine intelligence needed
would be that of determining which components of which modules should be
grouped to form a "useful" system for the task at hand. How many homunculi
would be necessary? Maybe one for each module (to maintain components) and
one to maintain the mechanism for selection and use of the modules.

Dave Druitt
(the NODES)

"I don't know and neither do you - but we can find out!"