Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!seismo!lll-lcc!ames!ucbcad!ucbvax!SAIL.STANFORD.EDU!JMC From: JMC@SAIL.STANFORD.EDU (John McCarthy) Newsgroups: comp.ai.digest Subject: re: AIList Digest V5 #181 Message-ID: <8707170622.AA15750@ucbvax.Berkeley.EDU> Date: Fri, 17-Jul-87 01:07:00 EDT Article-I.D.: ucbvax.8707170622.AA15750 Posted: Fri Jul 17 01:07:00 1987 Date-Received: Sat, 18-Jul-87 11:37:10 EDT Sender: daemon@ucbvax.BERKELEY.EDU Distribution: world Organization: The ARPA Internet Lines: 18 Approved: ailist@stripe.sri.com [In reply to message sent Tue 14 Jul 1987 22:58-PDT.] The distinction I had in mind between natural kind and cluster is the presumed existence of as yet unknown properties of a natural kind. When I said "doubtful cases are rare", I left myself open to misunderstanding. I meant that in case of chairs in Timothy's experience doubtful cases are rare. Therefore, for a child to presume a natural kind on hearing a word or seeing an object is advantageous, and it will also be advantageous to built AI systems with this presumption. Finally, a remark concerning the "symbol grounding" discussion. My problems with it were mainly quantitative - there was just too much to follow. I suspect that Stevan Harnad's capacity to follow very long discussions is exceptional. I would welcome a summary of the different points of view by someone who did follow it and feels himself sufficiently uncommitted to any single point of view.