re:Re: SUO: Continuants and Occurrents in 4D
Pat,
I'm doing all of this in a purely first-order framework with
no "psychologism" or any other reference to anything that
requires any kind of "peering" into inner mental processes.
>John, you seem to live a double life. On the one hand, you have
>argued vigorously and effectively for the centrality of FOL as a
>representational formalism for ontology use, and as you know I agree
>with you. On the other hand, you seem to now be arguing that any kind
>of assertional logic cannot possibly be adequate, since such logics
>do not (either in their syntax or their sematnics) pay even passing
>attention to any notion of "agency", and certainly do not "include
>the agency in the formalism", whatever that means. I am left
>wondering what your position actually is.
What I mean is something along the lines of the paper on
contexts that I am working on:
http://www.bestweb.net/~sowa/ontology/contexts.htm
That paper is now complete up to the beginning of Section 6
on "nested graph models", which gives the formal definition
and shows how NGMs can be used to represent a Tarski-style
model, a Kripke-style model for propositional modal logic,
and an extention of Kripke models for quantified modal logic
with the notion of "counterparts" for objects in "other worlds".
The next part, which I haven't yet finished, completes the
circle by showing how any nested graph model can be
"flattened" by using segmented names (a la Unix file systems)
to create globally unique identifiers. Then you can turn any
NGM into a good old-fashioned Tarski-style model M=(D,R), where
the domain D contains the union of all entities (with their
unique URI's) and R contains all the relations.
The advantage of the nested graph models is that they allow
you to modularize things, have metalevel discussions about
what is contained in any nest, etc. But to prove that all that
power doesn't really take you outside the FOL realm, I show
that you can flatten everything Tarski-wise.
And I don't say anything about "agency" that can't be
expressed in FOL (although I do use triadic relations, such
as legislate(A,L,X), which says that agent A legislates L as
a law for context X). That provides a way to slip Dunn's
semantics for modality into the mix so that you can show how
the agent A "causes" things in context X to turn out the way
he or she "intends" by legislating the "laws" or "axioms" of X.
All of this provides an interpretation for verbs such as
"intend" in a way that can be axiomatized at the metalevel,
but still be reduced to a flat Tarski-style model in which
all that mess of stuff can be proved to be consistent in
terms of nothing more complex than a model M=(D,R).
>I agree with Barry on this one, and I will dismiss Peirce. (He did
>neat work on notations and deserves to have his historical influence
>acknowledged, but all that is an aside. I dismiss 'pragmatism' as a
>guide for ontology.)
Don't blame Peirce. I quote him in Section 5 of the paper,
but I don't extract any "deep thoughts" from him beyond what
I quote explicitly. There's nothing in the nested graph
models that goes beyond what you can prove to be consistent
in terms of a Tarski-style model.
John