The Dancer and the Dance

From Stefan Balke (960723.1500 CET)

Rick Marken (960717.0621) --

The Dancer and the Dance:
Methods in the Study of Living Control Systems

Thank you very much for your beautyfull text. I think this text is a very
good proof of your intention (and ability) to make it easy for a reader to
understand the basic PCT ideas. I read it with great advantage and, although
I already knew the most ideas, I detected some new perspectives to look at
them. And of course, now I know why you spoke about Hollywood and the beauty
of PCT; one of your figures should contain an attractive passer-by to test
the phenomenon, no one should be able to turn the page within the next five
minutes :wink:

Behavioral Illusion

      The behavioral illusion is seen when an
environmental variable appears to be causing the
actions of a control system. For example, we see
this illusion when a person's head turns to see an
attractive passer-by. It looks like the passer-by
causes the person's head to turn but the person is
actually controlling the image of the passer-by,
trying to keep it centered on the fovea. The
movement of the passer-by is a disturbance to the
position of the image on the eye and head-turning is
the action that counters this disturbance. It looks
like movements of the passer-by (IV) act on the
person to cause proportional movements of the
head (DV). In fact, the causal path from IV to DV
runs through the environment, not the person; the
proportional changes in the DV in response to
changes in the IV are a result of the laws of optics
(that relate the passer-by to the image on the eye)
not the laws of behavior (that relate the image on the
eye to head turning).

Sorry, but I can't see that their is a causal path from IV to DV running
through the environment. Could you explain it to me? To exclude
misunderstandings, I do understand that their is not a single causal path
from IV to DV running through the person, but a complete control system in
action, which includes an interaction of several causal pathes.

      A familiar example of an everyday use of
The Test (familiar to movie goers) is determining
whether you are being followed. The hypothesis is
that the driver in the other car is controlling a
perceived relationship: "behind your car". You Test
this hypothesis by applying disturbances; randomly
and quickly turning up one street and down another.
If the car remains behind yours after a number of
these disturbances have been applied, it's a good bet
that you are being followed. If one disturbance is
effective, however (that is, if you turn and the car
that was following is no longer behind you) then the
apparent "following" was a coincidence; the
perception of being behind you was not under
control.

Must not be a coincidence. If you suffer from paranoia or have other good
reasons to think that someone is following you, you could come to the
conclusion, that the follower has noticed the application of The Test
(because quick turnings can only make that sense to the follower) and
changed his strategie to remain undiscovered. You will assume that he will
have a complice with another car and a handy, waiting for you at the next
street corner. This points to the problem of the practicability of The Test
in the case of changing intentions (means) at one level and a stabel
intention at a higher level.

BTW, did you discuss the paper at the meeting and what were the coments?

Best,

Stefan

[From Rick Marken (960723.0900)]

Stefan Balke (960723.1500 CET) --

Thank you very much for your beautyfull text.

Thank you for your helpful comments.

Sorry, but I can't see that their is a causal path from IV to DV running
through the environment. Could you explain it to me?

I tried to make it clear in Figure 4 (which you didn't get, of course,
because I sent only the text version). My claim that the causal path between
IV and DV runs through the environment, not the organism, when the organism
is a negative feedback control system, is based on the following equation
relating the output of a control system (DV) to a disturbance (IV) to the
variable controlled by that system (I call it S in the paper):

DV = -1/g(IV)

where g() is the environmental feedback function that characterizes the
relationship between system output (DV) and controlled input (S). This is
just a surprising fact about the way control loops work; it means that what
we are seeing when we look at the relationship between IV and DV in
experiments on control systems is the environmental function, g(), that
relates DV to S, not the organism function (call it f()) that relates S to DV.

Me:

If one disturbance is effective, however (that is, if you turn and the car
that was following is no longer behind you) then the apparent "following"
was a coincidence; the perception of being behind you was not under
control.

You:

Must not be a coincidence. If you suffer from paranoia or have other good
reasons to think that someone is following you, you could come to the
conclusion, that the follower has noticed the application of The Test
(because quick turnings can only make that sense to the follower) and
changed his strategie to remain undiscovered

You don't have to be paranoid to come to suspect that this is the case. You
have just hypothsized a different controlled variable-- "following
undetected" as opposed to just "following". That may, indeed, be the
variable that is actually controlled. To test it, you would have to monitor
THAT variable under disturbance, not just whether or not the car is being
followed.

...This points to the problem of the practicability of The Test in the case
of changing intentions (means) at one level and a stabel intention at a
higher level.

The Test can be quite difficult even when intentions are stable. Try the coin
game with a couple of friends; see how difficult it can be to determine what
another organism is controlling for. The Test requires creativity (in
generating hypotheses about possible controlled variables) skill (to test
those hypotheses), persistence (to keep testing as hypothesis after
hypothesis fails) and patience (since the first few hypotheses are rarely
right). The Test is not easy; but it's the only way to learn what is most
important about the behavior of a living control systemn: the perceptions
it is controlling.

BTW, did you discuss the paper at the meeting and what were the coments?

Yes. Most of the comments were suggestions for clarification which I will try
to incorporate into the next version of the paper which will be submitted for
publication.

Best

Rick

[From Bruce Gregory (960805.0925 EDT)]

Rick Marken,

I very much enjoyed your paper on methods in the study of living
control systems. It completely lives up to your standards of
clarity and economy. (Just to show you that I _am_ paying
attention, I note that on page 14 you have a reference to
Powers, 1991, 41-54, but there is no corresponding paper in the
references list.)

Regards,

Bruce Gregory

[From Rick Marken (960805.0740)]

Bruce Gregory (960805.0925 EDT) --

I very much enjoyed your paper on methods in the study of living control
systems.

Thank you. Thank you!

(Just to show you that I _am_ paying attention, I note that on page 14 you
have a reference to Powers, 1991, 41-54, but there is no corresponding paper
in the references list.)

I'm impressed; you made it all the way to end:-) The Powers reference should
be 1992, not 1991. I've found other errors since I sent the paper to you --
as well as things I want to change (say differently) or add -- but that
always happens _after_ I submit something for publication. I'm just dying to
know whether the journal editor will send the paper out for review, period.
After all, the not very subtle subtext of the paper is that the experimental
method, as currently used in the behavioral sciences, tells us nearly nothing
about the behavior of organisms and thatthe only way to find out about
behavior is by Testing to determine the variables organisms control. I would
be thrilled if _Psychological Methods_ would even _consider_ publishing such
a paper.

Best

Rick