From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!csd.unb.ca!morgan.ucs.mun.ca!nstn.ns.ca!bonnie.concordia.ca!ccu.umanitoba.ca!access.usask.ca!kakwa.ucs.ualberta.ca!unixg.ubc.ca!ubc-cs!uw-beaver!micro-heart-of-gold.mit.edu!w Thu Apr 16 11:34:32 EDT 1992
Article 5097 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!csd.unb.ca!morgan.ucs.mun.ca!nstn.ns.ca!bonnie.concordia.ca!ccu.umanitoba.ca!access.usask.ca!kakwa.ucs.ualberta.ca!unixg.ubc.ca!ubc-cs!uw-beaver!micro-heart-of-gold.mit.edu!w
upost!cs.utexas.edu!sun-barr!news2me.ebay.sun.com!exodus.Eng.Sun.COM!orfeo.Eng.Sun.COM!silber
>From: silber@orfeo.Eng.Sun.COM (Eric Silber)
Newsgroups: comp.ai.philosophy
Subject: Categories
Message-ID: <kum8beINNsv3@exodus.Eng.Sun.COM>
Date: 14 Apr 92 18:16:14 GMT
References: <1992Apr14.143822.10246@psych.toronto.edu>
Organization: Sun Microsystems, Mt. View, Ca.
Lines: 15
NNTP-Posting-Host: orfeo


 The issue of how to define, both operationally and conceptually,
 what "categories" are,  seems to lead to some interesting disjunctions.
 In using neural-nets , for example, to elicit from a computational system
 the invariants which characterize its ability to classify a certain set
 of inputs, one may arrive at a satisfactory model for the
 "categorization" of external objects.  However, there is also the
 problem of the categorization of internal objects.  I assume there
 is a resolution to this problem by appeal to internal communication
 between subsystems.   The practical example case is a human
 eye-brain system.   In response to external visual input, an eye-brain's
 architecture "finds" invariants which correspond to the 
 parameters of some feature-recognition paradigm for classifying
 external objects.  But the eye-brain is also producing
 endogenous , "imaginary" entities which are also classifiable.


