Upalevel; language

[From Bill Powers (921027.0700)]

Rick Marken (921026.0930) --

You're right that the Blind Men paper shows that the three Blind Men
all have it wrong. They don't have it "partly right." How can you have
part of a control system right? Unless you put all the parts together,
you don't have a control system at all.

It's occurred to me that perhaps what we need is an "upalevel" paper.
This should be written by you and Gary and perhaps other real
psychologists. Instead of battering away at the wall, trying to get
past the objections, why not start writing about the wall? "Look,
folks, we have a new theoretical approach that has excited many
respectable scientists from all over the world. But we can't get
papers on it published in mainstream journals, simply because it
contradicts accepted wisdom and because you have to understand it
rather well before you can see what is new about it. We've offered
papers on many subjects of interest to psychologists, trying to show
how control theory can apply over the whole range of human behavior.
But each time we do this, referees with narrow interests see it as
wrong, misguided, outmoded, or old stuff. Referees with many different
and even mutually-contradictory theoretical stances have said that
their own field already takes control theory into account -- that not
only is control theory compatible with their views, but they already
know everything it has to offer. Or, at the other pole, the
behaviorist says "You sound like a cognitivist," while the cognitivist
says "You sound like a behaviorist."

"How can this be? Wouldn't you suspect that a theory that seems to
mesh with widely differing schools of thought might have something of
great generality and truth in it? But referees do not see it this way,
because each of them judges from one narrow point of view. A theory
that fits behaviorism as well as cognitive psychology might seem to
offer a unifying principle of which these different schools are only
special cases. But to see that, one has to see all the applications,
not just one. One has to grasp the new theory as it is, not as it is
imagined. How can we control theorists get past this barrier against
publication so that psychologists in all fields can become aware of
the potential of this approach?"

And so on.

Variable aspects of sentences, for example (like meaning,
structure, inflection, etc) are controlled INPUTS, not generated

I want to make a modest push to make sure our linguists really truly
get this point. They may actually get it, but I don't think we're
hearing the result. Sentence construction is not construction of some
object out there, or in some vague conceptual space; it's construction
of an input, a perception. The mere fact that we know of a sentence
shows that it's a perception. The same goes for grammar, for any
regularity we PERCEIVE in language. The relationships between
different levels of analysis of language are relationships among
levels of perception, not levels of output production. We have to
guess what the production processes are, because they're outputs and
we don't perceive outputs. We perceive only their perceptual
consequences: language is perception.


I'm passing on social control for the time being.
Best to all,

Bill P.