[Top] [All Lists]

Re: [ontolog-forum] Context, at last!

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Nicolas F Rouquette <nicolas.rouquette@xxxxxxxxxxxx>
Date: Wed, 15 Jun 2005 16:46:42 -0700
Message-id: <42B0BDE2.2030608@xxxxxxxxxxxx>
Chris Menzel wrote:    (01)

>On Mon, Jun 13, 2005 at 10:18:01AM -0700, Nicolas F Rouquette wrote:
>>'de dicto' and 'de re': Following [Hayes04], the contrast between de
>>dicto (of the speech) and de re (of the thing) can be illustrated by
>>the distinction between direct quotation of speech, as in "Louis said,
>>'Superman is Clark Kent' " vs.  "Louis said that Superman is Clark
>>Kent" . The first, de dicto, reports Louis' actual words (and is
>>false, in the story) while the second, de re, reports what she said
>>about someone, using the speaker's words ....
>This example is a better illustration of the distinction between direct
>and indirect discourse -- reporting what speaker literally said vs
>reporting the *content* of what is said.  That is, of course, a very
>interesting distinction, but it's not coextensive with the de re / de
>dicto distinction.  The dr/dd distinction usually arises in modal or
>other so-called "intensional" contexts, and typically involve the
>interplay of modal or intentional concepts and quantification.  For
>example, if I tell you "I'm worried that someone is watching me", what
>I've said is ambiguous depending on whether there is some specific
>individual that worries me (de re), or whether my concern is simply that
>someone or other -- I don't know who -- has me under surveillance (de
Which field of study talks about "intensional" contexts?    (02)

>The logical interplay noted above in this case has to do with the scope
>of the existential quantifier vis-a-vis the intentional verb "worried".
>The quantifier has "wide" scope on the de re reading, and "narrow" scope
>on the de dicto; in a KIF-ish syntax:
>De re: (exists (?x) (IsWorried Menzel [IsWatching ?x Menzel]))
>De dicto: (IsWorried Menzel [exists (?x) (IsWatching ?x Menzel]))
Thanks for making illustrating this distinction in clear, understandable 
terms.    (03)

>One of the interesting things about intentional (hence de re and de
>dicto) contexts that can be seen here is that their proper
>representation often seems to require reifying *propositions* -- the
>things worried about, believed, feared, desired, etc.  For this reason
>worry, belief, etc are often called "propositional attitudes".  I've
>used square brackets above as a mechanism for forming a
>proposition-denoting term out of a sentence.  (There are other ways to
>do represent intentional contexts, but this one is clear and pretty well
Coincidentally, your point about reifying propositions sounds very close 
to the
point DOLCE makes in their approach to formalizing the notions of 
and "situation" in their ontology. For example, their view of a 
"situation" is an instance
of a concept that has unitarian entity (i.e., something that we can 
point to and therefore requires
enough reification to do so) where the "unitary" criteria constitutes 
the "description" of that concept.
This parallel w/ DOLCE should be about right notwithstanding plausible 
bugs in my understanding of DOLCE.    (04)

>The dr/dd distinction is also prominent in modal logic.  Typically (and
>very informally), a de dicto proposition involves the attribution of a
>modal property (notably, necessity or possibility) to a general
>proposition, whereas a de re modality involves the attribution of a
>modal property to a specific individual.  Here's a simple example from
>medieval philosophy concerning an argument that God's foreknowledge of
>the future is incompatible with human free will.  Consider any simple
>action that ostensibly seems free, say, whether or not one sits at a
>given moment.  God, of course, being omniscient, knows today who is and
>who is not sitting tomorrow at noon.  Consider, then, the following
>1. Whoever God sees to be sitting at noon tomorrow is necessarily
>   sitting at noon tomorrow.
>2. God sees that Bill Clinton will be sitting at noon tomorrow.
>3. Therefore, Clinton is necessarily sitting tomorrow.
>4. If Clinton is necessarily sitting tomorrow at noon, he is not free 
>   to refrain from sitting tomorrow at noon.
>5. Therefore, Clinton is not free to refrain from sitting tomorrow at
>   noon.
>Of course, the sort of argument can be framed for any individual and any
>action.  So it looks like God's foreknowledge is incompatible with human
>St Thomas responded to this argument by showing that it hinges upon a de
>re / de dicto ambiguity.  Specifically, Premise 1 is ambiguous between:
>1a. (Nec (forall (?person)
>                 (=> (Sees God [SittingAtTime ?person TomorrowNoon])
>                     (SittingAtTime ?person TomorrowNoon))))
>1b. (forall (?person ?time)
>             (=> (Sees God [SittingAtTime ?person TomorrowNoon])
>                 (Nec (SittingAtTime ?person TomorrowNoon))))
>(Note we have a similar sort of scope ambiguity here as in the first
>example, this time between the modal operator "Nec" and the universal
>quantifier.)  1a is just an instance of the logically trivial (given
>that God is omniscient, of course) de dicto proposition that,
>necessarily, whatever God sees will be the case will, in fact, be the
>case.  1b, by contrast, is the dubious de re proposition that if God
>sees that someone will be sitting tomorrow, then that person is
>*necessarily sitting tomorrow*; i.e., that person is sitting tomorrow in
>*every possible world*.  Unlike 1a, 1b is implausible; but 1b is the
>reading of premise 1 that is needed for the argument to go through.
>Thomas therefore concluded that free will was still safely in tact.
>Of course, there are lots of examples not involving medieval
>philosophical theology, but they are rather bland and joyless by
>comparison. :-D
Thanks again for this entertaining explanation.    (05)

I mentioned the dr/dd distinction because there is some relationship to 
context I couldn't quite explain.    (06)

Would it be fair to say then that precise reasoning over an ontology 
w.r.t. "context"
would require :    (07)

a) reifying the propositions whose interpretation is potentially subject 
to a context of some kind    (08)

In your example:    (09)

De re: (exists (?x) (IsWorried Menzel [IsWatching ?x Menzel]))    (010)

De dicto: (IsWorried Menzel [exists (?x) (IsWatching ?x Menzel]))    (011)

we could say that the interpretation of "IsWatching" is subject to a context of 
some kind,
namely, an individual (?x in de-re), or another proposition (IsWorried) which 
itself have a larger "context" sensitvity of its own.    (012)

b) specifying the attribution of the modality of such "context" propoerties    (013)

Perhaps you said it best with:    (014)

Typically (and
very informally), a de dicto proposition involves the attribution of a
modal property (notably, necessity or possibility) to a general
proposition, whereas a de re modality involves the attribution of a
modal property to a specific individual.     (015)

Without (a) and (b) done across an entire ontology, then we have an 
over-hanging ambiguity
w.r.t. a given proposition has any context sensitive interpretation at 
all and if so, relative to what.    (016)

Once we have (a) and (b), then, we could further require that the 
modality of "context" properties
should be a well-defined concept, kind of like what you hinted at w.r.t. 
(1b) as a "dubious" context
compared to (1a) which is reasonably well-defined.    (017)

Does this make sense?    (018)

What kind of worries me is the idea that people who have studied "modal 
propoerties" for dr/dd
purposes seem to be content w/ the view that "context" is about the 
scope of the quantifiers
and that we have to be careful in placing these quantifiers. Ok, I can 
appreciate that and understand
that these things matter w.r.t. automated reasoning and the sensibility 
of the inferences we get from it.    (019)

However, there is a larger concern at the level of using ontologies as a 
way to facilitate knowledge
capture and representation in which it would be helpful to have some 
awareness about the relevance
of clarifying what quantification scopes are meant and whether the 
dependencies between quantifiers
and propositions and individuals or something else are really what is 
"meant" with the way an ontology
is actually written.    (020)

Would you have suggestions w.r.t. where these concerns have been already 
discussed in the litterature
or perhaps analyzed w.r.t. how to reason about them?    (021)

-- Nicolas.    (022)

>Chris Menzel
>Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>Shared Files: http://ontolog.cim3.net/file/
>Community Wiki: http://ontolog.cim3.net/wiki/ 
>To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>    (023)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (024)

<Prev in Thread] Current Thread [Next in Thread>