COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Comments on Dennett's "Beyond Belief"
C00013 ENDMK
C⊗;
Comments on Dennett's "Beyond Belief"
5:2:4 Tentatively, I don't agree that it is necssary to have definite
criteria for deciding what counts as two different propositions or about
what a proposition is.

5:∞:1 I certainly prefer constructing programs that treat propositions
as having an abstract syntax.  Do you know about abstract syntax?

6:"(2) It seems to me that all the virtues of considering a proposition
as a set of possible worlds can be obtained by treating a proposition
as determining a set of possible worlds but possibly having other
properties as well.  Having identity conditions for propositions only
leads people into mischief.

6:"(3)" Entities of the third kind also exist and should be related to
the first two.  Humans and programs must make do with whatever grips
on the outside world they can find.

7:"(b)" Why must an intension BE an extension determiner?  Why can't
it determine extensions and have other properties as well?

9: The issues on this page seem to overemphasize unary predicates and
functions.

[If Dennett doesn't take the design stance at some point, he is
inconsistent with the Brainstorms doctrine.  We needn't choose one
of the notions of belief.  We can say that people, dogs and computers
have certain relations to sentences in English and various formal
languages, relations to abstract entities corresponding to the
various notions of proposition, etc.  All of these relations may
be explored.  We would like to find   really good abstract entity,
but to choose it, we must take the design stance.  We must ask what
kinds of abstract entities and what kinds of relations between actors
and them and objects and events external to the actors best enable us
to describe the phenomena connected with believing].

24 Focussing on the organismic contribution in isolation is what Putnam
calls methodological solipsism.

26:"6" Changing proper names will work, but most likely interchanging
first and third person singular won't work.  There are two many cases
where the context redundantly determines the context.  Also cryptography
is generally applied to texts with word boundaries removed, so that
making words out of chapters won't disguise the content either.

40 The view of computer languages seems a bit schematic, but I don't
have a specific criticism other than that perhaps he or his reader
may confuse programming languages with the information in data
structures.

44::8 Computer languages are usually not sentential.

47:-2:-4 I don't see that equivalence of computer programs is a "vexing
question".  Different equivalence relations are useful for different
purposes.  Admittedly computer scientists fall into the trap of
defining and equivalence with one purpose in mind and then try to
inflate it to "the notion of equivalence".

48:2 I like the idea of starting with what notional attidudes are
expected to accomplish.

52: From my point of view, this paper would benefit from re-ordering
so that Dennett's ideas of notional worlds would come on p.5 rather
than on p.50.  I think that 50 pages of arguing the inadequacy
of other models will lose more readers through boredom than it
will gain through attacking a hypothesized smugness about the
rival ideas.

52::-1 cryptography to the rescue again

56 It is perhaps vaguely relevant that any chemist can tell you that
there can be no compound made out of the known elementary particles
that a layman couldn't tell from water and still different from water.
Gold can be faked pretty well but not water.

60:2 Correspondences other than that in which the centers agree are
worth considering too.

61::4 You believe that salt is sodium chloride because you heard it
indirectly from a chemist.  You don't have everything he said or
presumed his hearers knew, but maybe the correspondence isn't so
hopeless to establish.  The chemist learned that salt is sodium
chloride on your level in grade school and was not conscious of
a change in that particular fact as he learned chemistry.

63::4 I don't know that notional worlds can't have contradictory
properties.  I think notional worlds are partial worlds, and these
can have properties that have no consistent extension to a full
world.  For example, when I consider someone wondering whether
189 is prime, it may be worthwhile to imagine a possible partial
notional world in which it is.

72 I cannot decide whether AI requires a solution of the problem
of the two Shakey's Pizza Parlors.  It might make a good example
for non-monotonic reasoning applied to intensions.  Thus at a certain
point the situation turns out to be less clear than was supposed
and certain names must be replaced by two names.  Perhaps whenever
the same name is used in two places there is a circumscription saying
that since there is no reason they should designate different things
in this case, then we should conclude that they designate the same
thing.

section 5 with re-initialized page numbers

7 Herb believes that all Iranians in California should be deported.

Sam is an Iranian living in California but not an acquaintance of Herb.

raises good questions.

10 Is "The shortest spy is a woman" an expression of belief about
the shortest spy - whoever he may be?

[It seems that second order definitions would help in defining notional
attitudes.  In the first place, there may be only one ascription that
meets the criteria of the second order definition, and in the second place,
we may often be more interested in whether an ascription is ok than
in finding one ab initio.  It is interesting to contemplate a situation in
which a suitable ascription has been found and a search for others has only
turned up a few fringe variants, e.g. the meaning of proper names occurring
only once.  Almost certainly, it won't be practically possible either
to find a substantially differrent interpretation or to prove that
none exists.

There is a kind of belief analogous to knowing a telephone number, but
it doesn't have a good name in English.  We can use circumlocutions
like "Pat thinks he knows John's telephone number" or "Pat has a belief
about John's telephone number".  In this connection, we also have
value(Pat, Telephone John) to use the notation of my "First order concepts
paper.].
[Dennett's examples should be tried using non-monotonically obtained
extensions of naive formalizations.  It is also clear that approximate
theories come in here.  I guess  there may be a general relation
between approximate theories and non-monotonic reasoning.  In fact,
the approximate theories are obtained by non-monotonic reasoning
from prototheories that don't make the distinctions and have all
the names of the general theory.  Partial models.  Also the notion
of a body of knowledge without commitment to a specific language.
Also doing metamathematics by going down into an inner model rather
than by going up to a metalanguage.].