[From Rick Marken (970426.0900)]
I recommend that fans of conventional psychological research (Bruce
Abbott, Martin Taylor, Hans Blom, etc) read the following VERY carefully
(note, in particular, the very last statemeant; I have said this over
and over again for the last 10 years or so but maybe if
it comes from Bill, wrapped in such a nice, rigorous package, you
will pay attention). If, after reading this, you still believe that
conventional psychological research contributes to our understanding of
behavior then I think there is nothing anyone can do for you.
Best
Rick
···
-----------------
Bill Powers (970424.0601 MST) --
You are talking about a case in which failing to recognize a regular > relationship leads to the _impression_ that there is a low
correlation, when in fact there is a very high one. This is what
I call "using the wrong model."
The most pertinent example, of course, is that of S-R investigations > of the behavior of a control system, where the controlled variable
is not taken into account. The manipulated variables are actually
disturbances, and the responses are actually actions that cancel
the effects of the disturbances on the controlled variable. Not
knowing that, however, the experimenter does not use the right
measure of the disturbances: he uses qd instead of Fd(qd);
neither does he use the right measure of the action: he uses qo
instead of Fo(qo). Furthermore, he does not realize that there are
many qd's that could affect the same controlled variable, so he
does not control the environment so as to eliminate random effects
on qi -- he can't, because he has not identified qi. And of course
he will probably include variables that have no effect on qi. The
net result is a low correlation between qd (the "stimulus") and
qo (the "response"), singly or in bunches.
If qi had been identified reasonably well, the experimenter could
have made sure that nothing else but qd (aside from qo) could
affect it, and the true connections between qd and qi, and qo
and qi, could be determined by physical examination of the
environment. This would provide an exact model of the environment,
and now the relationship between qd and qo would become much more
nearly exact, most of the variance disappearing. In fact, multiple
disturbances could be used, and multiple linear correlation analysis
would now show that there is a very high (negative) correlation in
the relationship
Fo(qo) = -[Fd1(qd1) +Fd2(qd2)+ ... Fdn(Qdn)].
It's only through identifying controlled input variables that one
can distinguish between environmental variables that ought to
relate to actions and those that ought NOT to relate to them.
The key is to understand what the controlled variable is.
[From Bruce Gregory (970426.1800)]
From Rick Marken (970426.0900)]
Bill I understand, but your are an enigma. What is an "emmis"?
(It's not in the OED.)
Bruce
[From Rick Marken (970426.1800 PDT)]
Bruce Gregory (970426.1800)--
Bill I understand, but your are an enigma. What is an "emmis"?
Emmis (not sure of the spelling) is a yiddish word that (if it could
be translated) would mean something like "plain, honest talk". Here
is more emmis:
Bill Powers (970424.0726 MST)
One test [comparing PCT and MCT]is to see how people actuallybehave
> when current inputs are lost. The MCT model is set up to continue
producing an output that would be right if the disturbance remained
as predicted and the properties of the environment did not change.
In Jeff Vancouver's experiments, we did not see this happening,
although there was _some_ ability to keep acting. That ability,
however, could be explained by controlling the relationship
between the felt and seen positions of hand and mouse and the
position of the moving target. Rick showed that when an integral
was inserted between the mouse and the cursor, this could no
longer be done. I showed that in compensatory tracking, where
the target is stationary and gives no clues as to the right
movments, "control" is even worse. If there is any model-based
control in this situation, it is very rudimentary. At higher
levels of control, when we get around to testing them, people
may do better...
The main test [comparing PCT to MCT] is to see whether real
behavior _fails_ under the same conditions where the models
fail. I think there is plenty of evidence that at the lowest
levels of control, the MCT model predicts success under loss
of input where we observe profound and often permanent failure.
Deafferentation instantly destroys skilled control (as it would
for the PCT model), which returns only over a period of months,
if ever. The MCT model would predict that there should be no
immediate effect; poor control would begin to appear only as
the model gradually becomes outdated. The evidence is pretty
clear that at the lower levels of organization, the MCT model
is not correct. This does not rule it out at higher levels, of
course.
L'chaim
Rick
[From Bruce Gregory (970426.2100 EST)]
Rick Marken (970426.1800 PDT)]
Emmis (not sure of the spelling) is a yiddish word that (if it could
be translated) would mean something like "plain, honest talk". Here
is more emmis:
Thanks. I'll have to get a copy of the O.Y.D. 
Bruce
[From: Oded Maler (970427)
Rick probably meant EMET which is "truth" in Hebrew, pronnounced
as EMES in Yiddish.
I thought the title was intendeded to inidcate Rick's association
with the Amisch community by refusing to use modern control theory 
--Oded
···
--
