Spreadsheet, Ladders, Categories

[From Rick Marken (931101.0900)]

Bill Powers (931030.1230 MDT) --

Even in Rick's three-level spreadsheet there are really no rules
for producing outcomes. The reason is that Rick also has
disturbances acting on the outcomes, so if the same output is
applied twice, the outcome will be different from one occasion to
the next.

Yes. And now there are disturbances that vary during each iteration
of recalculation. All levels still control exquisitely; the outputs
at all levels are just continuously varying to protect perception.
Very nifty.

In the example of gain-reversal, at first this may require
reorganization. But after a while, the reversal takes place
automatically and immediately when the sign-reversal happens.
...
Right, Rick?

Right.

Re: Nasrudin story (I lost the original post).

I, too, celebrate Greg's willingness and interest in going to our
"neighbors" to ask about borrowing the ladder. I am just against
telling the neighbors that we actually returned what we borrowed
earlier(a flashlight?) when we didn't really didn't return it. I
don't think Greg advocates doing this -- I just want to make sure
that this doesn't happen (by accident) as a side effect of our
efforts to control for acceptance of PCT.

PCT has some bad news for the neighbors; we've discovered some
unpleasant facts about the nature of behavior. I'm not afraid
to go and discuss this with the neighbors but I'm also not
interested in dissembling about PCT in order to make the message
more palatable. We might have many things in common with
our neightbors (same neighborhood, same langauge, etc) but
eventually we're going to have to deal with the sore spot --
we borrowed something and didn't return it (ie. we found out
that behavior is control OF perception not controlled BY
perception) and at some point or another we're going to have
to deal with this problem.

If we didn't return what we borrowed then I say that we fess up.

Avery Andrews (931101.1105) --

The last BBS of 1992 (15:4) is a particularly good place to start for
looking at the actual literature on motor-control-related topics:
there are good guys ...and someone who expresses dissatisfaction
with the idea that feedback systems control what they perceive.

Since I won't be able to get to a BBS for a while, could you just
say who the latter person is, his/her e-mail address and why s/he
is dissatisfied with the idea that feedback systems control what they
perceive. We rarely get "neighbors" who are so explicit in their
criticism of PCT (ie. none of our neightbors ever come by and prod
us about returning what we borrowed -- ah, love those metaphores).

Martin Taylor (931031 10:10)--

(1) "natural kinds" versus constructed categories"

The flip-flop mechanism depends on the existence of mutually exclusive
sets of category perceptions.

Doesn't your flip-flop mechanism DEFINE (not depend on) the existence
of these perceptions?

From about here on I find it difficult to understand your description

of a "natural kind". It sounds to me like a "natural kind" is the output
of a perceptual function -- the outputs of such a function cannot be
in two states (have two values) at the same time. Is this it? Or is a
"natural kind" a property of the world beyond our senses?

Martin Taylor (931029 22:00) --

I said:

My point (and Bill's too, I think) was that
"categorization" is what perceptual functions do, in the sense that
the output of ANY perceptual function could be the result of many
possible inputs.

Martin replies:

Categorization is what people looking at the output of a many-to-one
function can do if they want. It is not what the function does, unless
it is a categorizing function.

OK. We are using the word "categorization" in two different ways. I
was using "categorization" to refer to ANY many-to-one mapping. You
are using it to refer to what the "categorizing function" does; the
categorizing function being a hypothetical perceptual function that
produces a perception at the hypothetical "category level" of
perception. Now the question is "what does the categorizing function
do?" -- ie. what is unique about category level perceptions? I take it
that your next remarks address this:

Not all subspaces are categories of the
space within which they reside. At a minimum, it would seem necessary
that the subspace occupied by a category should have a measure greater
than zero. In other words, specification of a measure by a point on the
real number line is insufficient to define a category in itself.
For a category to be valuable as a category perception, there must be
a probability greater than zero that another member of the category will
be perceived during the (finite) lifetime of the perceiver.

But I'm afraid I really don't understand this. I think it's an attempt
to explain why the notion of category as many-to-one mapping is not
an appropriate definition of categorization (or the resulting category
perception). But, if it is, it still doesn't help me understand what a
category IS (as distinct from the categorical outputs of any perceptual
function). The last statement, for example, seems to be satisfied by
my example of a sum function. Many members of the category "4", ie.(2,2)
(3,1) (4,0) have a probability greater than zero of being perceived
during the (finite) lifetime of the perceiver.

Mach bands may not be category perceptions, but "Mach bands"
certainly are.

Yes. Very nicely put. I agree. What I am trying to get at is what
differentiates this kind of category (like "Mach bands", ie. THIS
is a Mach band and THAT is a Mach band) from the categorical nature
of perception in general. For example, I can near many variants of
the p in spin as the same phoneme -- p. So I am categorizing
disciminable sounds (THIS is a p and THAT is a p) as one phoneme.
I have the feeling that this kind of categorization is different
from the categorization that goes on when we categorize different
stripe patterns as "Mach Bands" (or different people as all
"Canadians" or different books as all "Fiction").

I proposed one approach to experimentally determining whether or not
there might be a "category" level. It involved doing the same kind
of experiment as the "sequence control" experiment I described at the
last CSG meeting. The idea was to take what seemed like what "we" agree
is a category (I suggested vowel letters) and see if the rate at which
the perception of this category can be controlled is about what would
be expected from the current location of this perceptual level in the
hierarchy -- slower than sequence, faster than program. No one commented
on this suggestion but I would appreciate anyone's opinion.

Best

Rick