HF & PCT; categories

[From Bill Powers (931020.1215 MDT)]

Greg Williams (931020) --

Greg, I think what you're apparently trying to do is a good thing
-- getting the general objections to the Human Factors (HF)
approach converted to specific chapters and verses. I, too, would
like to see specific quotations from the literature that reveal
differences in the concept of control. That would be more
valuable than generalizations that aren't accompanied by specific
evidence.

It's not always easy to get evidence in such a form. When you're
in the midst of a group of people and are receiving a generally
hostile treatment, the reason for the hostility isn't often made
crisp and clear. When I discuss PCT ideas with non-PCTers, the
reaction is often just a general defensiveness, a pushing-back
against a disturbance that's not clearly identified. Any boat-
rocker knows the sense of general opposition that arises, even
when there haven't been any specific disagreements. I think
people perceive, at a higher level, that the general form of what
you're saying just doesn't fit the familiar grooves in which they
are used to thinking; the resistance isn't against PCT (How could
it be? They don't understand it yet) but simply against this
strangeness and refusal to fit. I think there's a strong
component of fear in the reaction: take that odd idea away, it's
threatening what I believe! All this happens without any real
discussion of the ideas; it makes a real discussion almost
impossible.

I've had in mind for some time an idea for a demo of the
underlying phenomenon, but so far it's been too vague to put into
practice. The basic idea would be to set up some sort of
moderately complex control process in which there are rules that
have to be discovered. The rules should be discoverable, but
complex enough that some real effort is required to find them and
put them into practice. When the participant masters the task,
something that would count as a real achievement would result.

Then you alter a rule. This doesn't become apparent right away,
but suddenly what the participant now knows isn't exactly right
any more. The behavior of the environment now starts looking a
little strange. The question is, how long will it take for the
participant to realize that something is wrong, and what will the
participant try to do about it?

A pertinent variant would be to have one person learn the task,
and then ask that person to help another one whose task looks the
same but is actually subject to a slightly different set of
rules. How hard will the first person work to convince the other
one that the first set of rules applies, even though it doesn't?
What sorts of arguments will develop?

Maybe even more pertinent would be to have the first person
develop an explanation for successful behavior in a task where
there are no disturbances of the outcomes, so an S-R explanation
would seem to work. Then, for the alternate experiment, introduce
disturbances and see what happens to the hypothesis. Judging from
experience, I would expect the person to rationalize like mad,
invent all sorts of unjustified mechanisms, and in general try to
perserve the original explanation. And it would be exceedingly
interesting to see what would happen if the first person then
tried to help another person doing the task with disturbances
present from the start.

Any social scientists out there who think this might be doable?

ยทยทยท

---------------------------------------------------------------
Rick Marken (931020.0930)--

Rick, you have taken the words out of my mouth; I was just
warming up to say something about category perceptions when you
said it all:

While thinking about your emphasis on "categoricalness"
a question occurred to me that you might be able to answer: the
<question is "isn't ALL perception categorical?".

....

That is, any perception is "invariant" with respect to certain
transformations of its components -- just like "pin" and "bin".

...

In general, given the perceptual model of PCT -- p = f(s1,
s2...sn) -- a particular perceptual signal value will be
invariant with respect to certain transformations of the
inputs, s1, s2 etc. Which transformations leave a partcular
perception invariant depend on the nature of f(). So, for

perceptions of many different "types" (as defined in HPCT)

we would expect to find that a particular state of that
perception (value of p) is "categorical" in the sense that it
is the same (invariant) for many different combinations of
input (s1, s2..).

This is what I meant a while back about not being in too much of
a hurry to get to the category level. The reason I introduced
categories was that I needed a way to connect symbols to the
things symbols stand for. I needed that so there could be
elements available for constructing lists and programs that work
with symbols as inputs and outputs. Symbols stand for classes of
things, not analog quanties; that's just the nature of a symbol.
You can't have a different symbol for every way a dog can look or
sound. You just refer to "a dog" and that stands for one or two
of the main aspects of this infinitely-variable kind of
perception. When you work with symbols you settle for a very
sketchy representation of the world of experience. It takes
buckets-full of symbols to get across a simple idea.

Anyway, that's all I was after. I should have known that
"category" is a big meaningful word in some circles, that has had
the hell studied out of it and that has probably come to mean
dozens of things I don't mean. I should have thought of some
other word, except that all of our words are shopworn from being
used to mean other things.

When "category" can be used to mean any perception at any level
of organization, that indicates that we don't understand category
perceptions, even while we're using them. We see something there
that reminds us of category-ness, but can't step back far enough
to see what we mean by that.

All I know is that a lot of kinds of perceptions can be
controlled on an analog basis and without the use of symbols, and
that at some level we suddenly shift to using symbols in lists
and programs in lieu of the analog perceptions. If the perceptual
operation that takes place at that level isn't "categorizing,"
then it is SOMETHING, and we ought to be able to identify it
SOMEHOW. It won't help to define categorizing in a way that could
apply at any level.
---------------------------------------------------------------
Bruce Nevin(931020)

What kind of mouse? Help is on the way.
---------------------------------------------------------------
Best to all,

Bill P.