Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

ONT Re: Inquiry Driven Learning Environments




o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o

IDLE.  Discussion Note 2

o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o

JA = Jon Awbrey
MA = Murray Altheim

JA: http://suo.ieee.org/ontology/msg05148.html

MA: You threw me for a bit of a loop.  Is there a direct corollary
    relationship between the two triples of abstract components or
    schematic categories, and perhaps in Peirce's principles, as in:

      deductive   <------->   syntactic   <------->   Firstness
      inductive   <------->   semantic    <------->   Secondness
      abductive   <------->   pragmatic   <------->   Thirdness

MA: Or am I jumping the gun and the answer to this coming up?

Murray,

This is a paper whose first draft was written 10 years ago.
In fact it was a project report for a graduate AI course
that I was taking at the beginning of my Systems Engine
phase, still idling on, if somewhat chuggily of late.

I'm not quite sure what I believed at that time.
I believe that one can justify a parallelism of
the above shape, but I can also tell that I was
being careful not to be too simplistic about it.
I can see a way to rationalize the first set of
correspondences, and perhaps even the final set,
but I have watched ranglers butting heads about
the latter association until the cows come home,
and even then, rationalizing one correspondence
does not mean that there can be only one.  Long
story too late to be short time, the utility of
a given correspondence generally depends on the
fine structure of what conducts the parallelism.

JA: It is a working hypothesis of the philosophy that I am taking up here
    that these three dimensions are irreducible and exhaustive, at least
    when they are taken at their own proper level of approximation, but
    the character of their relative independence as dimensions does not
    seem to involve that of interacting trivially with each other.

JA: In fact, the evidence of my own experience with implementing
    some fraction of their capabilities leads me to the following
    provisional conclusion:  In order to build a live, real-time
    capacity for "understanding" it will be necessary to integrate
    all three of these dimensions of intelligence so thoroughly about
    a common family of data structures that each of the distinct aspects
    will operate more like an alternate perspective on the same facts than
    as a separately embodied module of code.  Along the way, partly in support
    of this deliberately short list of functionalities, a particular slant on the
    aims of AI will be advanced.

MA: This was what I was referring to in my response to Richard Cooper

    http://suo.ieee.org/email/msg11374.html

    in stating that knowledge systems cannot be built without such
    a core.  I like the idea that all three are essentially facets
    of the same essential facts, inseparable, a holism.  So I'm
    eager to see this one fleshed out.

This is sometimes called a "limits of modularity" thesis.
It's one of the things that I learned about the hard way,
against all of my initial intuitions, when I was working
on my Theme One program all through the 80's, since this
forced me to integrate empirical and rational capacities,
or what's more or less the same, induction and deduction.

Jon Awbrey

o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o