[From Rick Marken (970422.0920)]
Me:
...the sensory variable is the additive result of disturbance and
output (as in a tracking task), p = f(o + d).
Martin Taylor (970422 00:00) --
I think you'd better look at Bill's diagram, in which he makes
quite clear that the "disturbance" (qd) acts through the
disturbance function, to form Fd(qd), which is what acts on the
controlled environmental variable qi. qi = Fd(qd) + Fe(qo).
My equation above is equivalent to qi = Fd(qd) + Fe(qo) where Fd and
Fe are constant multipliers of 1 and the perceptual function (which
transforms qi into p) is also a constant multiplier.
You don't even see the difference between qd and the variable
that is added to the output effect on qi. How can you possibly
make coherent statements about what conveys information about
what?
It's a gift;-)
I said:
knowing p tells you nothing about the state of d.
Hans Blom (970422) says:
Let me give you a simple counterexample that demonstrates that
Martin is right and you are wrong: there _is_ information about
the disturbance in the perception.
Given your definition of "information about the disturbance in the
perception", I agree. If p = f(u) + d and you are given p and f(u) then
I have no doubt that even you could solve for d. You say that this means
that there is information in p about d. I say that this means that
algebra works. When I say that there is no information
in p about d I mean that, given only p, there is no way to determine d.
In order for a control system to be able to "extract" the information
about d from p it must know f and u. You claim that:
a model-based controller knows f more or less perfectly.
I think it is extremely unlikely that living controllers "know" f
(actually, the many different f's that relate them to the many different
variables the control). But I have my integral control
demo handy if you ever want to _test_ your fantasies.
Clark McPhail (970422) --
For those of us who are interested in what you do, and for others
who are unfamiliar with what you and others have done, could you
provide reference citations to the studies to which you refer
above of people controlling: "...more than one variable at a time" .
Two papers in the "Coordination" section (6) of "Mind Readings".
. . "several people [controlling] several variables at the same
time"
I was thinking of the many experiments on multi-person interaction done
by Tom Bourbon. One of Tom's papers on this was published in
the ABS issue; another is in Hershberger's "Volitional Action" (I
believe). Some of Tom's most interesting studies were reported at
CSG meetings and at a Midwest Sociology Assn. meeting that _you_
chaired.
....[and]..."people [controlling] complex variables (like a
sequence or a program)."
Most of these are unpublished studies. They are based on work
described in the "Hierarchical Behavior of Perception" (HBP) paper,
which is available at my web site (http://www.leonardo.net/Marken).
After two years of trying to get HBP published I realized that
conventional psychology was just not interested in PCT. I was working on
control of sequence and program experiments at the time HBP was being
rejected (for saying "nothing new") and I realized that write-ups of the
experiments I was doing would never get published either.
I thought (in all modesty, of course;-)) that HPB was a great paper;
I also thought the experiments on sequence and program control
looked promising; they were a "step beyond" the simple tracking stuff
I had been able to publish. But I could tell from the reviews of HBP and
the complete lack of response to my published tracking research that the
conventional psychology community was not ready for PCT; heck, they had
no idea what the "simple tracking experiments" were about; they didn't
understand (and, thus, had no interest in) the basic phenomenon of
control. So there is really no point (I think) in trying to publish
research that "goes beyond" what your audience (in my case, the
psychological community) doesn't yet understand -- control. I'll just
present this stuff a CSG meetings.
Best
Rick