COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	normal[w85,jmc]		Normally birds fly
C00005 ENDMK
C⊗;
normal[w85,jmc]		Normally birds fly

Normally birds fly, but Tweety's wings have been clipped.

Suppose someone tells me, ``Normally elephants are monogamous''.  On the
one hand, this assertion can be heard, understood, stored, believed or
challenged.  In this it retains its unitary character.  In its
application, however, it gets expanded into some construction involving
$ab$ and is used via a circumscription relative to a certain collection of
sentences.  Its verification conditions may also be peculiar.

Feb 20

normally birds fly
isn't translatable
as
∀x.bird x ∧ normal x ⊃ flies x
because if we also have "normally birds lay eggs"
we don't want
∀x.bird x ∧ normal x ⊃ oviparous x
and if we learn that a particular bird can't fly, we don't
want to also lose our conjecture that it lays eggs.

One way out is given above.  Translating all the normally's in
the facts taken into account into a formalism involving  ab.
Another way is to write
∀x.bird x ∧ normal(x, context1) ⊃ flies x
where  context1  includes the syntactic context, so that
we also have
∀x. bird x ∧ normal(x, context2) ⊃ oviparous x.
Then if we learn that Tweety is not normal in context1
we cannot conclude that Tweety is not normal in context2.
Bah.  normal(x,context1)  seems exactly the same
as  ¬ab aspect1 x, although it may have some technical advantages.
e.g. it is easier to make the contexts all different than to make
the aspect functions all different.