learning and knowledge

[from Joel Judd 920630]

Martin (920729 & 0630) and Bill (920629 & 0701),

Sorry to assume too much, but I didn't want to write too much in the
original post:

Knowledge can certainly be represented, but interactions do not require the

kind of representation >that implies regress.

Right; *transmission* requires the kind of representation that involves
regress. Transmitting encodings requires that both receiver and sender
understand what the encodings represent.

I also agree with your description of ECS environmental interaction, and
what it means for an ECS to "know" what's going on. The latter part of my
comments referred to how an ECS knows that it's getting "better"; how does
it "learn"?

With reference to education, I guess the crude way to express it is to say
that in many cases the whole is more than the sum of its parts. A house is
NOT just the things you mentioned--it's an organized combination of them.
The contracter knows what that means. The rookie apprentice does not. Sure,
he's seen pictures and been inside other houses and watched movies of
construction, etc., but he's never actually made one. Not having the
details of BLC theory handy, I would say that not even point (2) can be
taught. All three points require learner experience; the teacher can only
evaluate evidence of learner.

I guess I should be happy that interactivism (i.e. organism interacting w/
environment) as the basis for learning is apparently assumed by PCTers--it
certainly isn't by all educators. Bill mentioned the dilemma I was
alluding to--Meno--which Hugh Petrie attempts to resolve in his book.
Again, I agree that the solution necessitates action, blind action, but
that's not what the transmission analogy implies. It implies that one can
learn without making mistakes; BVSR says in effect that we learn FROM our
mistakes (hence the title of Perkinson's book). However, it's interesting
that many learning "errors" NEVER occur. One of the standard lines in
language acquisition goes something like "But how do you account for the
fact that children never say XYZ?" I suppose that a PCT response (assuming
that "never" turns out to be accurate) would be along the lines of "Well,
at such and such a point in development, the particular PC hierarchy is
such that XYZ is not the kind of random variation a normal system will
produce; there are constraints on the blind variation."

So instead of continuing to look for more molehills, can I say that the
following approximates a PCT epistemology:

New organisms are born with genetically transferred intrinsic variables.
These allow the organism to interact, from the start, with its environment
in order to preserve itself. Disturbances to these variables provoke
reorganization in an effort to reduce error and maintatin reference levels
necessary to life. In a given environment, however, the actions commenced
by reorganization involve (of necessity) the sensory-motor capabilities and
limitations of the organism. Perceptions which reduce intrinsic error
sufficiently are remembered, and remain in place until or unless they fail
to reduce future intrinsic error, in which case reorganization recommences.
As sensory-motor systems unfold in their genetically pre-determined manner,
reorganization finds newer, more efficient and more sophisticated ways of
satisfying intrinsic error--perhaps even anticipating such error through
memories of past experiences, vicarious learning, and extrapolation. By the
time a human being is just a few years old, such massive foundational
learning has taken place in the sensory-motor systems (visual-aural
discrimination, tactile development, figure-ground, conservation, etc.)
that future learning often takes for granted such development, and /or
assumes that it must be "hardwired." Adding to the confusion is the seeming
convergence in development by members of the same culture and linguistic
community--barring damage or abnormality EVERYONE born in particular group
will grow up able to function as a member of that group.

But, after a certain "maturation" point, this reorganization system becomes
somewhat suspect. For example, in the case of language, after about 6-7
years--certainly after puberty--the results of reorganization in a L2
become highly variable. There is no longer the convergence showed by
children in an L1, now all kinds of perceptions seem to satisfy the
intrinsic error of these adult systems.

Am I just preaching to the converted?

P.S. Martin--do you know something about Canada we don't? Are things that
bad?

[Martin Taylor 920702 15:30]
(Joel Judd 920630)

Knowledge can certainly be represented, but interactions do not require the
kind of representation >that implies regress.

Right; *transmission* requires the kind of representation that involves
regress. Transmitting encodings requires that both receiver and sender
understand what the encodings represent.

No it doesn't, unless you use "coding" in the sense of a one-to-one mapping.
PCT says you don't do that. Layered Protocol theory says you don't do that.
We had some discussion about it last year (from Bruce Nevin, mostly). Who
uses that sort of coding except spies? and if they lose the codebook, they're
toast. Transmission of information does not require encodings that involve
infinite regress. It requires some recursive beliefs that each hold about the
other's belief about... And ultimately the belief system is grounded in
the belief that each has some commonality of interactive experinece with
the world.

Martin

PS. If you don't know that Canada is in the middle of a constitutional
crisis, your news organizations must be pretty poor. If, by July 15, the
provinces and the federal government cannot come up with an agreed
Constitution that is acceptable to Quebec, then in October Quebec will
have a referendum on whether to separate and become an independent country.
So far, there is no sign that the provinces and the feds can agree unanimously
(apart from Quebec), and even if they do, it is unlikely to be an agreement
that Quebec politicians will deem acceptable. It's much the same as
Czechoslovakia (sorry, the Czech and Slovak Federated Republic; sorry,
Czechia (or something) and Slovakia (next year)). They are discussed in
your news services?