[From Rick Marken (940201.1020)]
JIM DUNDON (940201 0026) --
My question was in response
to your description of two people "losing control" (your
description) when they decided to fight over an issue.
I was thinking in terms of what I understood to be one of
the major points of PCT namely that all bahavior is
purposive.
Let's look at this in terms of the basic PCT diagram of
a control system:
SYSTEM | ENVIRONMENT
>
r -->|C|<--p |I|<--q.i<--d
> > ^
> > >
e----->|O|--->o
>
>
Capital letters are functions, small letters are signals
(r,e,p) inside the system or variables (o,d, q.i) outside
the system. As you know, a properly functioning control
system keeps its perceptual signal, p, matching its reference
signal, r, continuously over time. The closer the on-going
match between r and p, the better the control. Control could
be measured by RMS deviation of p from r; control is perfect
if this deviation is continuously 0 (or very close to 0);
control is poor to the extent that the RMS deviation is
continuously much greater than 0. When I say that a system
"loses control" I mean that the RMS deviation of p from r
is continuously MUCH greater than zero.
Of course, we cannot measure RMS deviation of p from r
because these signals are inside the system (person). We can
only infer how well the system is controlling by measuing the
deviation of the environmental correlate of p, the
variable q.i, from what we guess to be its reference
value. The probelm of measuring control is non-trivail
because it is not possible to "see" the reference value
of q.i (the value of q.i that corresponds to r). We usually
assume that we know the reference value because we tell the
subject to keep q.i at position that corresponds to what we
want r to be. In a tracking task we do this by asking a person
to keep the cursor (q.i) next to (r=0) a target. But we
don't know what the person's r value REALLY is. An excellent
description of an APPROACH to determining a person's actual
r in a tracking task where r possibly variable is described
by Powers in Hershberger's (Ed) "Volitional Action" collection
(North Holland, 1989?).
The term "behavior" is ambiguous in PCT. In general, it refers
to changes in variables in a control system's environment,
specifically o and q.i. o is the output variable and it is
the MEANS by which the system influences the controlled
variable, q.i, whose reference value is the intended END of
controlling. Variations in o are NOT purposeful -- they
are determined by the effects of disturbances (d) to the
controlled variable. So variations in output variables are
non-purposeful behaviors in PCT. The change in the diameter
of the pupil that results from shining a light in the eye, for
example, is NOT a purposeful behavior. Variations in q.i
ARE purposeful -- they are determined MAINLY by variations
in the internal specification for the perceptual correlate
of q.i -- r. Variations in r function as variations in
intention (or purpose) regarding the (perceptual) results
to be produced.
So, in the environment of a control loop there is purposeful
behavior (of q.i, the controlled variable) and non-purposeful
behavior (of o, the systems influence on the controlled variable).
There is also one other kind of behavior: "irrelevant side
effects". These are behaviors of variables that are outside
of the control loop that have nothing to do with the operation
of the loop. They are not shown in the diagram above. An
irrelevant side effect is an effect that o or q.i has on another
variable, call it i.r, that is not in the control loop. For
example, when I move the mouse (o) to keep a cursor (q.i)
alighed (r=0) with a target (d), the mouse movements create
friction and hence, varying amounts of heat (i.r). So i.r =f(o),
but it is an irrelevant side effect because it is not perceived
and controlled.
So PCT does not say that all behavior is purposive. In fact,
PCT starts well before we start talking about how we might
explain behavior. It starts by asking "WHAT IS BEHAVIOR?" and
it shows that the word "behavior" is ambiguous; behavior means
intentional results of output (purposeful behavior or control),
outputs that influence intended results ("influential" behavior
or "action") and irrelevant side effects of control ("accidents").
PCT shows what we mean by behavior AND it explains how ALL these
different kinds of behavior occur. But, of course, the most
intersting kind of "behavcior" is purposeful behavior -- CONTROL.
Right I would have lost control of that reference signal
It makes no sense to say that you lost control of a reference
signal. The reference signal determines the intended state of
a perception (the state of being "upright" for example). The
system loses control of the PERCEPTION to the extent that
it is unable to get the perception to match the reference
signal.
IS THERE AN ULTIMATE REFERENCE SIGNAL
There are ultimate reference signals (plural). We have no
idea what they all are but most of them probably have to
do with what you call "survival". They are references
for physiological variables (like O2, CO2, glucose, blood
pressure, etc etc) that MUST be at certain levels in order
for control systems to work at all. They are the "highest" order
variables because all learned perceptual control systems
presumably exist as they do in order to keep those variables
at (or as near as possible to) their (probably genetically
determined) reference levels.
How about abberant behavior. Is it controlled, that is
directed behavior, chosen? How about mental illness
is it controlled? The beauty of PCT as I originaly
understood it. is that it explained these behaviors
as purposefull.
What you see as "aberrant behavior" or "mental illness"
is YOUR perspective on purposeful behaviors, yes. But
my discussion above should make it clear that "behavior"
is not a simple concept. What one perceives as "aberrant
behavior" in another person includes your perceptions of
variables that the person is controlling (q.i's) but it also
includes your perceptions of the person's outputs that are
the means of controlling q.i's that you might not be aware of
AND irrelevant side effects of the behavior of controlled
variables and outputs. PCT shows that terms like "aberrant
behavior" and "mental illness" are over-simplifications that
come from an era when people thought of "behavior" as something
you could watch people doing. PCT shows that "behavior" is
ambiguous (it refers to three phenomena -- purpose, influence
and accident). It also shows that "purposeful behavior" --
control -- is NOT something you can watch people doing
because it involves control of PERCEPTION. My "Mind reading"
program illustrates this point. In order to know what a
person is doing purposefully (what they are controlling)
one must do the TEST FOR THE CONTROLLED PERCEPTUAL VARIABLE:
but that's another story.
I said:
Try to perceive your finger pointing in two different
directions at once ....which is impossible.........
neither of the two control systems......
You say:
WHOOOOOA!
What two control systems? You just said it was impossible.
I said it is impossible to keep one variable in two different
states: q.i cannot equal 2 AND 4 AT THE SAME TIME. Imagine
two control systems (like the one I above) BOTH trying to
control q.i relative to their OWN r values. Here's a
diagram.
SYSTEM1 | ENV | SYSTEM2
r1-->|C|<--p |I|<----q.i---> |I|-->p-->|C|<--r2
> > ^ ^ | |
> > > > > >
e----->|O|-->o1 o2<--|O|<-------e
Note that both systems are perceiving the same variable
(q.i) in the same way (via function I). Both systems
also influence q.i via their outputs, o1 and o2. There
is no problem as long as r1 = r2. That is, if both systems
want to keep q.i at the same value, everything is OK. (By
the way, I have not shown disturbances but they are also
influening q.i). If, however, r1<>r2 then there is conflict;
and if the systems are good controllers the conflict will
be WORSE than if they are poor controllers. Both systems
can get what they want (r1 and r2) only by getting q.i
to TWO DIFFERENT VALUES. That what is impossible.
What will happen is that NEITHER SYSTEM will get it's
perception to match it's reference. Both will end up putting
out their maximum possible output and q.i will end up frozen
between r1 and r2. If there are now distrubances added,
neither system will be able to resist them -- they are too
busy resisting each other.
In this conflict neither system is able to continuosly keep its
perceptual signal matching it's reference signal (RMS deviation of
p from r for both systems is large). That's why I say that both
systems in conflict have LOST CONTROL. They have lost control precisely
becuase they are GOOD CONTROLLERS. This is why conflict is such
a tremendoous problem for control systems.
This makes me realize that word that people should find "ugly"
is not "control" (and its cousins "purpose", "grace", "skill" and
"cooperation") but "conflict" (and its cousins "competition",
"winning", "fighting the good fight", etc).
Where's that commitment to excellence I love about PCT'rs
How'd I do?
Getting back to two people fighting. Are they losing control
or controlling within the limits of their chosen reference
perceptions.
After all of the above, I leave the answer as an exercise.
Best
Rick