ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Lost conceptions in decomposition. Thirds, Triadic r

To: Steven Ericsson-Zenith <steven@xxxxxxx>, "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "Barkmeyer, Edward J" <edward.barkmeyer@xxxxxxxx>
Date: Thu, 28 Mar 2013 11:37:21 -0400
Message-id: <63955B982BF1854C96302E6A5908234417DCB79511@xxxxxxxxxxxxxxxxxxxxxxxxxx>
Steven,    (01)

You wrote:    (02)

> One further point of clarification. I assume that you refer to Donald Davidson
> when you say "Daniel Davidson." As in:
>
>       http://plato.stanford.edu/entries/davidson/
>
> This will explain some differences in our approaches.    (03)

Yes.  I see that I keep making that mistake.  It is caused by (a) my having had 
a professor who was "Daniel Davison", and (b) my writing emails without proper 
attention.  It is now apparent to me that I have some brain cross-wiring that 
is not easy to fix in my old age.  :-)    (04)

And yes, that is the difference in our approaches.  In my case, it is 
compounded by my recent re-education in natural language processing, which 
leads me to believe that almost all active NLP projects use a Davidsonian model 
of events/occurrences.  This is in part a consequence of the (IMO ill-founded) 
academic love affair with RDF triples.  (As I recall, in the 1970s, there was a 
similar popular notion that parses of formal languages would all be triples, 
with arguments about linked graphs vs. Polish notation structures.  But there 
was a camp, led (if I recall correctly) by Jeffrey Aho, who argued for "quads", 
as a representation powerful enough to capture the substance of any 
computational structure without requiring dyadic subterfuges.)    (05)

-Ed    (06)

>
> Steven
>
>
> On Mar 27, 2013, at 3:42 PM, Steven Ericsson-Zenith <steven@xxxxxxx>
> wrote:
>
> >
> > {ppy: I have changed the subject line - though I may be the wrong
> > person to do this :-)}
> >
> >
> > Dear Ed,
> >
> > Before you go ...
> >
> > What is your definition of the term "semantics?" The formal definition of
> the term (derived from Carnap) is the set of rules by which a language may
> be transformed.
> >
> > I find that there is a great deal of ambiguity in the usage of this term in
> computer and software engineering, both in academia and industry. It
> appears to often incorrectly associated with the equally ambiguous term
> "meaning."
> >
> > For completeness then, for me, the term "meaning" is the pragmatic one
> (due to Peirce). It is the behavior that is the conceivable consequence of
> apprehending the sign, it is the difference that a term/phrase/sentence may
> conceivably make in the world. Although, in fact, in my work I go further than
> Peirce by arguing that it is more exact to say that the meaning of a sign is 
>the
> actual behavior that is the product of its apprehension.
> >
> > I just want to clarify that I have understood you correctly.
> >
> > Best regards,
> > Steven
> >
> > --
> > Dr. Steven Ericsson-Zenith
> > Institute for Advanced Science & Engineering http://iase.info
> >
> >
> >
> > On Mar 27, 2013, at 10:05 AM, "Barkmeyer, Edward J"
> <edward.barkmeyer@xxxxxxxx> wrote:
> >
> >> Steven,
> >>
> >> You wrote:
> >>
> >>> Dear John,
> >>>
> >>> As is often the case in dyadic thinking you and the others that have
> >>> responded find the third in yourself (the apprehender) and not
> >>> objectively in the syntax. In other words in a triadic statement
> >>> decomposed to dyads something is lost.
> >>>
> >>> You knowing that lost thing, that lost conception, and you find it
> >>> immediately when reviewing the dyadic statement but cannot see that
> >>> it is lost (it's a mental act not, in fact, present in the
> >>> syntax/representation). Another human - from a similar background -
> >>> might also have the same experience (indeed, we rely upon it) but a
> computing machine will certainly not.
> >>
> >> On what evidence do you base this assertion?  A computing machine has
> no conception of anything.  A particular software product creates a machine
> with the concepts that are built into the software.  If the software
> 'understands' ternary relations, then it is not necessary that there be a 
>dyadic
> rewrite of the statement.  If the software accepts dyadic syntax and converts
> it internally to the ternary relation, then the unit triadic concept is
> maintained, not lost.  The software model, of course, is not the world it
> represents, but neither is the human conceptual model.  The human
> conceptual model is doubtless richer, but whether that richness is either 
>valid
> or relevant to the communication at hand is 'quite another thing entirely'.
> The software may also have some additional richness in the form of axioms
> and facts about the verb and the nouns involved.
> >>
> >>> This is
> >>> what Peirce was talking about in my earlier quote when he says to
> >>> Victoria Welby that Russell's dyadic logic relies upon the very thing 
>that it
> omits.
> >>
> >> This may well be, but it is not clear to me that semanticists in the last 
>100
> years generally agree with Peirce on the subject of the relationship between
> conceptual content and syntax, or on the idea that relationships beyond
> binary are necessarily unitary in concept.  I fully agree that there are 
>ternary
> relations that are conceptually unitary, but there are also ternary (and
> quaternary, etc.) relations that have a meaningful factorization, or perhaps
> multiple meaningful factorizations.  I agree that "John gives the book to
> Mary" is conceptually a unit.
> >>
> >>> So, for example, in the case of the statement "A gives B to C," -
> >>> that I claim Peirce shows is impossible to decompose as you have
> >>> been suggesting (certainly "conventional means," I accept). However,
> >>> to be exact Peirce says (CP 1.474): "Thus, A gives B to C becomes A
> >>> makes the covenant D with C and the covenant D gives B to C."
> >>
> >> That may be what Peirce suggests, but Daniel Davidson suggested an
> alternative, to wit, that one describes a unitary 'event' -- a 'giving', and 
>then
> describes roles that are played in that event.  The unitary concept is
> maintained as the event object -- what happens -- not some strange
> factorization (covenant D) that has no real meaning.  The nature of a giving
> event involves three roles, and those roles can be represented by three
> dyadic relationships that assign role players to those roles.  This is a 
>useful
> general model for the description of conceptual events, whether or not they
> can be described as instances of a single natural language verb.
> >>
> >> The problem with this approach is the need to support truly meaningful
> binary verbs with a different model, and the value decisions in factoring
> larger conceptual situations into interconnected subordinate situations.  "The
> airship industry declined because the Hindenburg disaster demonstrated the
> danger of using hydrogen for buoyancy, and the substitute -- helium -- was
> too expensive."  Humans must decompose that thought in order to
> conceptualize it, but we also impose viewpoint in determining how to
> organize the interrelations of the subordinate concepts.
> >>
> >>> It is necessary to introduce D in the decomposition to a necessary
> >>> "triadic tetrad" - which none of the proposals made here so far have
> >>> done - and so information was lost in the naive triadic syntax and
> >>> dyadic forms you have both proposed.
> >>
> >> It not necessary, nor even semantically appropriate, to introduce the
> nonsense 'covenant D', as the Davidsonian model that John and I have
> described shows.  And when you describe the triadic relation as 'na´ve', I
> wonder on what basis you think the natural language syntax: A gives B to C is
> any less 'na´ve'.
> >>
> >>> In short you cannot take language naively as evidence but must be
> >>> aware of the nature of apprehension and the semeiotic process
> >>> (semeiosis) that is active across what Peirce called "the living
> >>> mind." It is for this reason that Peirce and Ladd-Franklin objected
> >>> to the "Russellization of logic" - which sanitized all 20th Century logic 
>of
> such broader consideration.
> >>
> >>> The resulting formal logic is a pure mathematics that is not
> >>> suitable to be mapped to natural language.
> >>
> >> That is probably a fair assessment of Russell's approach, yes.  But  I 
>think
> the record will show that logic in the latter half of the 20th century, and 
>its
> relationship to semantics, departed significantly from Russell.
> >>
> >>> I am confident that any modern logician must accept the inverse to
> >>> be true. For modern practical needs it is more effective, and no
> >>> doubt cheaper, for humans to simply adopt formal logic as natural
> >>> language.
> >>
> >> I disagree completely.  The function of language is to communicate.  The
> language must be chosen with attention to the content being communicated
> and the nature of the audience.  Formal logics have their place in the
> spectrum of languages, along with English, classical Sanskrit, and road signs.
> >>
> >>> Peirce's view of logic, indeed much of logic before Russell, is
> >>> logic as semeiotic theory - it deals with logic both as the natural
> >>> science of conceivable consequences and as a pure mathematics (the
> >>> science of necessary conclusions).
> >>>
> >>> If you wish to understand natural language and the behaviors of the
> >>> individuals using it (i.e., complex biophysical structures) then it
> >>> is this broader conception of logic that you must consider.
> >>
> >> I doubt you will find much disagreement with this in the ontology
> community.  We know that we are making models, and those models are
> necessarily inaccurate in whatever ways they (intentionally) fail to represent
> the whole of the domain they describe.  That is the nature of models.
> >>
> >>> I imagine some software smarty-pants will think this question can be
> >>> resolved quickly by writing clever software and taxonomies. But they
> >>> will be wasting their time because no matter how exhaustive you try
> >>> to be in implementing what I have described above the problem is
> combinatorial.
> >>
> >> It is more than "combinatorial".  In general, models made from different
> perspectives may present conflicting assertions, precisely because they
> ignore aspects that are important differentiators in other perspectives.
> >>
> >>> Binary relations are not your friend when dealing with the real
> >>> bindings across these complex conceptual manifolds the ghost of
> >>> which is merely reflected by natural language.
> >>
> >> And yet, European languages have only unary (intransitive) and binary
> (transitive) verbs.  All of the rest of the conceptual environment is made up
> by what our Latin educational heritage calls "adverbial phrases" that are said
> to "modify" the verb, or to assert "general roles" of noun phrases (agent,
> patient, instrument, purpose, location, etc.).  The problem of conveying
> "complex conceptual manifolds" did not originate with logic languages.
> >>
> >>> You cannot compete in modern computing machines with the low power
> >>> dynamics in biophysical structure that has no need to store results
> >>> or move input data from its natural path.
> >>
> >> I have no idea what this means.
> >>>
> >>> There is an important distinction in what I have said before that
> >>> has been missed by most responders. When I speak of "the third" I am
> >>> referring to Peirce's "thirdness" and not to the syntax of triadic forms.
> >>>
> >>> As Peirce says in the following a third always deals with the
> >>> general, and not an individual, the general was lost in the earlier
> >>> example. It's difficult avoid quoting Peirce, so forgive me - I know
> >>> that his 19th century language can seem cryptic at times - the
> >>> following is taken from CP 1.475 - CP 1.480 on the subject of TRIADS:
> >>>
> >>> == Peirce quote ==
> >>> "It may be said that it is a psychical fact. This is in so far true,
> >>> that a psychical fact is involved; but there is no intent unless
> >>> something be intended; and that which is intended cannot be covered
> >>> by any facts; it goes beyond anything that can ever be done or have
> >>> happened, because it extends over the whole breadth of a general
> >>> condition; and a complete list of the possible cases is absurd. From
> >>> its very nature, no matter how far specification has gone, it can be
> >>> carried further; and the general condition covers all that incompletable
> possibility.
> >>>
> >>> There, then, we have an example of a genuine triad and of a triadic
> >>> conception. But what is the general description of a genuine triad?
> >>> I am satisfied that no triad which does not involve generality, that
> >>> is, the assertion of which does not imply something concerning every
> >>> possible object of some description can be a genuine triad. The mere
> >>> addition of one to two makes a triad; and therein is contained an
> >>> idea entirely indecomposable into the ideas of one and two. For
> >>> addition implies two subjects added, and something else as the
> >>> result of the addition. Hence, it is wrong to define two as the sum
> >>> of one and one; for according to such a definition, two would
> >>> involve the idea of three. The idea characteristic of two is other. The
> corresponding idea characteristic of three is third. ...
> >>>
> >>> The genuine triad contains no idea essentially different from those
> >>> of object, other, third. But it involves the idea of a third not
> >>> resoluble into a formless aggregation. In other words, it involves
> >>> the idea of something more than all that can result from the successive
> addition of one to one. This "all that can"
> >>> involves the idea of every possible.
> >>>
> >>> The world of fact contains only what is, and not everything that is
> >>> possible of any description. Hence, the world of fact cannot contain
> >>> a genuine triad. But though it cannot contain a genuine triad, it
> >>> may be governed by genuine triads. So much for the division of
> >>> triads into the monadic, dyadic, and triadic or genuine triads.
> >>>
> >>> Dyadic triads are obviously of two kinds, first, those which have
> >>> two monadic subjects, as a high perfume and a burning taste are
> >>> united in many essential oils, and secondly, those which have [for] all
> their subjects individuals.
> >>>
> >>> Genuine triads are of three kinds. For while a triad if genuine
> >>> cannot be in the world of quality nor in that of fact, yet it may be
> >>> a mere law, or regularity, of quality or of fact. But a thoroughly
> >>> genuine triad is separated entirely from those worlds and exists in
> >>> the universe of representations. Indeed, representation necessarily
> >>> involves a genuine triad. For it involves a sign, or representamen,
> >>> of some kind, outward or inward, mediating between an object and an
> >>> interpreting thought. Now this is neither a matter of fact, since thought
> is general, nor is it a matter of law, since thought is living."
> >>> == End Peirce quote ==
> >>>
> >>
> >> I must here confess my ignorance.  I have no idea what the subject of the
> above excerpt is -- the "genuine triad" that seems to elude clear definition 
>or
> description by Peirce, if the excerpt is any indication.  And I cannot imagine
> how it might relate to the development of ontologies.  So, Steven, you have
> convinced me to stop reading this thread.
> >>
> >> -Ed
> >>
> >>
> >>
> >>> Best regards,
> >>> Steven
> >>>
> >>> --
> >>> Dr. Steven Ericsson-Zenith
> >>> Institute for Advanced Science & Engineering http://iase.info
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>> On Mar 26, 2013, at 3:28 AM, sowa@xxxxxxxxxxx wrote:
> >>>
> >>>> Steven,
> >>>>
> >>>>> I'm not sure that I Peirce would accept the functional notation as
> >>>>> triadic, but - as you say - he would prefer a a diagram/graph...
> >>>>
> >>>> The basic arithmetic operators -- add, subtract, and multiply --
> >>>> take two
> >>> inputs and generate one output.  That is certainly triadic.
> >>>>
> >>>> You can reduce the triads to dyads by Currying, but those dyads
> >>>> only
> >>> represent intermediate results.  You still get triadic connections
> >>> in the complete diagrams.
> >>>>
> >>>> Divide taks two inputs (dividend and divisor) and generates two
> >>>> outputs
> >>> (quotient and remainder).  But it can be represented by two triads
> >>> -- one that takes the two inputs and generates the quotient, and the
> >>> other that takes the same inputs and generates the remainder.
> >>>>
> >>>> If you use a purely relational notation (as Peirce did for his
> >>>> algebra of 1885
> >>> or his existential graphs of 1897), all those functions must be
> >>> represented by triadic relations.
> >>>>
> >>>> As another example, look at Prolog.  It represents the basic
> >>>> arithmetic
> >>> operators by triadic relations.
> >>>>
> >>>> John
> >>>>
> >>>
> >>>
> >>>
> __________________________________________________________
> >>> _______
> >>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> >>> Config Subscr:
> >>> http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> >>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> >>> Shared Files: http://ontolog.cim3.net/file/ Community Wiki:
> >>> http://ontolog.cim3.net/wiki/ To join: http://ontolog.cim3.net/cgi-
> >>> bin/wiki.pl?WikiHomePage#nid1J
> >>>
> >>
> >>
> __________________________________________________________
> _______
> >> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> >> Config Subscr:
> >> http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> >> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> >> Shared Files: http://ontolog.cim3.net/file/ Community Wiki:
> >> http://ontolog.cim3.net/wiki/ To join:
> >> http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> >>
> >    (07)


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (08)

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