1989
	Abstract for Washington University Technical Report

	DEFEAT AMONG ARGUMENTS II




This paper discusses a system of non-monotonic inference that treats multiple
inheritance situations in a way that many have desired.  The system takes a
form new to AI; defeat among competing arguments is the mechanism that leads
to non-monotonicity.  The system does in FOL something like what Touretzky's
system does for inheritance in semantic networks [Touretzky85]; it does more
of what Nute did [Nute85]; it develops more carefully what Poole attempted
[Poole85].  The system owes its form to philosophers' work on defeasible
reasoning (esp. [Kyburg82]).

This paper updates the ideas in [Loui87a], and it responds to Hanks and
McDermott's criticism [Hanks, McDermott87].  Finally, it considers what should
be done about the CMU-school's "clash of intuitions" [Touretzky, Horty,
Thomason87].