re:re:Re: SUO: Continuants and Occurrents in 4D
>On Mon, 12 Mar 2001 14:13:07 -0600 pat hayes <phayes@ai.uwf.edu> wrote:
>
>### Pat pls note this is a reply to an older interesting msg by you.
>Sorry for the length, but it did require somewhat of a think.
>
> >John Sowa:
> >
> >> >I would agree. And in fact, I think that the distinction
> >> >between the terms "continuant" and "occurrent" can be quite
> >> >nicely defined in Whitehead's terms: A continuant is something
> >> >that we can recognize at multiple encounters. An occurrent is
> >> >something that does not have enough distinctive characteristics
> >> >that we can be sure whether another encounter is with "the same"
> >> >or "a similar" entity.
> >
> >Robert Meersman:
> >
> >>### a most enlightening definition I must say. But in it, and in
> >>fact in all of the above, who are these "we" and "I" who do the
> >>recognizing and identifying... I tried to observe before that all
> >>of semantics constitutes an agreement among cognitive agents
> >
> >Im afraid I disagree, and moreover think that this view is profoundly
> >misleading (please don't quote Peircian chapter and verse in
> >response.) Here's why, in brief: it confuses *semantic* issues with
> >*epistemic* issues; it assumes that meaning is ultimately concerned
> >with what 'we' (or maybe 'they') can know, and how they come to know
> >it. I think this is deeply and profoundly ass-backwards, since until
> >we have some reasonably clear account of what it is that we (or they)
> >are knowing, it is both premature and rather hopeless to try to think
> >about how it is that it (whatever it is) can be known. Meaning is not
> >reducible to agency; rather, agency presumes meaning.
>
>### Pat,
>
>### First of all I am not a Peircian (I think) and have never read
>anything by him so far --in fact only yesterday in the Stanford bookstore
>I bought a first book with some of Peirce's writings... will start it on
>the plane home, and tell you afterwards whether he agrees with me :-)
>.
>.and not to worry, I will put a Tom Clancy jacket around it while
>reading it in public :-)
>
>### But in the meantime surely you don't want to imply that the only "true"
>meaning must be something "absolute", independent of everything such as
>observers, agents, users, ...?
No, I didnt say anything about 'one true,' or about meaning being
'absolute'. I meant only that to talk about *epistemic* notions -
what people, or maybe agents, know and believe, and what they mean by
the signs they use, and so on - without first having a theory of
*meaning itself*, seems to me to be approaching the topic backwards.
For example, you talk of "semantics through agreement". But ask
yourself: what are these people agreeing ABOUT? If all one can say
about their use of signs, is that they come to agreements to use
signs in some way, one might as well be talking about the agreements
they come to use cars in some way, or use lawnmowers in some way.
There is nothing in this picture of agents making agreements to use
things (in some way, with some ends in mind, you can keep going...)
which ever gets to grips with the essentially *semantic* nature of
signs, such as the fact that signs *refer* to things, or that they
*express propositions* about things. I can't see any route from a
purely pragmatic mode of analysis to semantics, since there isn't
anything particularly semantic about pragmatics. Everything people
do is pragmatic: digging ditches is pragmatic, but it's nothing to do
with semiotics. Now, you might object that lots of our pramatic usage
of signs is in fact intimately tied up with their meaning, and of
course I would agree; but my point is that one cannot get a useful
*analysis* of the semantic nature of sign usage without first having
some account of the semantic nature of the signs themselves.
>### For my position on semantics through agreement I don't yet feel the need
>to hide behind some long-dead philosopher's back but would rather want to
>report from a long experience with data modeling that in practice
>--a qualifier
>to which I return below-- there is no such thing as "one truth", but *every*
>registered fact, rule, event etc. is the result of an explicit or implicit
>agreement.
That may well be true, in some ultimate sense and with a rather
attenuated sense of 'agreement', though it is often debated. But some
'agreements'- such as the referential conventions of the language
that our brains learn for us before we even know it has happened -
are essentially semantic in nature, others are not. Without some
account of meaning, one cannot even begin to make this distinction.
>Until the "ultimate" laws of physics are known, physicists and
>everybody else will still need to agree on the observed outcome of experiments
>and e.g. agree whether they conform or contradict the current ontology
>and its laws. Actually, in modern physics such agreements become ever
>less evident to make, as I am sure you are well aware of.
>
>### Now even *if* these laws will *ever* be known, the reductionism required
>to describe everyday things and events in them will likely render them
>useless for any practical (again) ontology. Who knows, there even may be
>real information-theoretic limits for this, such as not enough matter in
>the universe to construct an intelligence that is able to interpret them.
>(Probably not a useful hypothesis, as there will be no matter left to plug
>it in a wall outlet but it sure is fun to talk & think at this scale,
>isn't it *grin* ).
>
>### So the question comes around; how can we establish meaningful
>communication about anything at all *without* such an absolute ontology...
>as e.g. humans seem to have achieved notwithstanding and to a reasonable
>extent over a few 100K years? This is where in my opinion the practical
>aspect must come into play. As I said before, if the purpose of the SUO
>effort is to capture all of physics, logic, and other science, etc. first,
>before it will/can address pragmatic issues such as "how will one be able
>to use it", *ahem*, this may be philosophically very satisfying and great
>fun (I confess also to me in fact) but risks to appear as a waste of time
>to many.
I agree that it would be, and that has never been my view of the role
of the SUO.
>
>### (goes on further down)
> >
> >> (e.g.
> >>two persons who agree they are pointing at the same occurrent; or
> >>one agent who agrees with himself that he is looking at an instance
> >>of the same continuant after a while, etc etc. Is this too trivial
> >>to mention, or am I too obtuse, in hypothesizing that we include
> >>these cognitive agents, AND perhaps the procedure by which they
> >>arrive at their agreements, AND the contexts that must restrict or
> >>qualify these agreements, as first-class citizens in any ontological
> >>design process?
> >
> >That is not too trivial by any means, but I think it is a very bad
> >methodology. How will the project get started? You need to 'include'
> >a theory of agency, etc., into the ontology. Now, how does one
> >include anything in an ontology? One writes axioms about it (or them)
> >with the intention of representing in those axioms enough information
> >about the things in question that relevant questions can be answered
> >by drawing conclusions from those axioms. So the proposal amounts to
> >begin by writing an axiomatic theory of cognitive agents, procedures
> >by which they arrive at agreements, and contexts in which they
> >perform. Now, how will you get started on writing those axioms? This
> >is where we came in, of ocurse, but doesnt that seem like a rather
> >bad place to *start*? Cogntive agents are rather complicated things
> >compared to, say, processes arising in the oil industry, or clay
> >being fired in kilns. And you still need to know if these cognitive
> >agents and their doings are going to be described as continuants or
> >4-d histories or whatever.
>
>### I disagree that you have to be able to do this first and even less
>that it would constitute bad methodology; why cannot we establish the
>role of the cognitive agents as axioms and first-class players in an
>ontology paradigm. You don't need to model them (at first) since their
>behavior by definition occurs *outside* the ontology.
Wait! You just said what looks like a direct contradiction, so maybe
we are misunderstanding each other. An ontology *consists* of axioms.
If these agents are "in the ontology" in any sense I know, they will
be there by virtue of being described in axioms of the ontology;
which is another way of saying they will be modelled in the ontology.
>They can even be
>made implicit: the set of cognitive agents associated with a part of the
>ontology may be exactly those that agree on the facts and rules in it,
>i.e. hold them in their beliefs.
Oh well, of course. This goes without saying. Ontologies are made to
be of use to people, or more usually groups of people, and particular
ontologies often have user communities. In this respect they are like
almost every other human artifact. But I don't quite see how
focussing on this is going to help us make progress. Cars run on
roads, but one doesn't do vehicle design by looking at tarmac.
>BTW movement of "individual agents"
>between sets will be of no concern to us (at first), and it would also
>allow for instance multiple internally consistent but mutually
>contradictory belief sets to co-exist in the same ontology.
I fail to see how.
>Naturally
>this leads to interesting logical and representational issues; I don't
>carry my literature database with me now but I remember asking a prominent
>logician a few years ago about possible logics for this and got a
>negative at the time. It occurred to me that several of the categorization
>arguments in SUO could and maybe should in fact be decided by the eventual
>(future) ability of agents (i.e. ontology designers) to move more easily
>between belief sets. For instance, it may feel easier to retract a 4D fact
>than a 3D fact since one can just qualify its validity by a time interval.
>Or, modeling things as occurrents may prove to allow easier identification
>between things occurring in different contexts but alleged to be the same.
>But I readily admit I don't have a formal argument for this as yet. If one
>at all exists...
>
>### And of course this brings me back to the issue of future practical
>use of an ontology. Qualifying belief sets (or contexts, or what have you)
>by "their" cognitive agents allows them to start off in the middle of the
>"reductionist scale", by defining facts as high level axioms without
>worrying (initially) about consistency with other agents' contexts
>--but of course maintaining intra-context consistency.
Ah, yes. Some kind of 'context' mechanism might be quite useful. But
it depends on what it can do. If the contexts are just separate boxes
with no connection between them, then the SUO isnt a single O at all,
and the entire enterprise is pointless. If it does allow some kind of
connection, what kinds does it allow? If it is going to account for
inter-translation automatically, then it is going to have to work a
miracle. So we need to know more about this context mechanism, and
what makes something 'high-level' with respect to it
> Note this has
>been happening for 50 years now as database designers were making their
>little avant-la-lettre "ontologies", called data models, for specific
>applications; problems appeared only in the 90s when interoperability
>requirements became important, of course. In fact I think some work that
>has happened in the database field on view and schema integration may be
>relevant to the SUO enterprise at one point; one of the things discovered
>was that generalizing everything to "entity relates_to entity", while
>always true and therefore *very* interoperable rarely yielded useful
>data models.
>
> >
> >Pat Hayes
> >
> >PS. John's suggested definition (above) of continuant/occurrent is
> >faulty on at least three counts. First, it doesn't actually capture
> >the meaning, since one can easily recognize an occurrent when it is
> >seen a second time, eg if you ever go to use the rest room during a
> >football match, you will have no trouble recognising that you are at
> >the same match when you get back. Second, even if it was adequate,
> >it isn't a definition of continuant, but a way to transcribe the
> >occurrent notion into a 4-d ontological framework without
> >acknowledging the continuant identity criteria, which we already know
> >is easy; and third, it uses scare-quotes, which is just a way of not
> >actually giving a real definition, but saying something that looks
> >like a definition, but isn't really: as Russell said (in a different
> >context), it has all the advantages of theft over honest toil.
>
>### hmmm... could be, but frankly the more I hear this discussion the
>less I understand the notions involved. Are we sure this discussion is
>not being held for its own sake?
No, it is central. Try actually writing some axioms about change and
you will rapidly discover how real these differences are. One can't
just smurdge them together and pretend it isn't an issue, and I think
it would be a mistake to declare either view the 'standard'. But Ive
suggested a way to bridge this particular gap, and there are plenty
of other problems to solve. Anyone got any good ways to describe
infinitesimals?
Pat Hayes
---------------------------------------------------------------------
IHMC (850)434 8903 home
40 South Alcaniz St. (850)202 4416 office
Pensacola, FL 32501 (850)202 4440 fax
phayes@ai.uwf.edu
http://www.coginst.uwf.edu/~phayes