[From Fred Nickols (970928.10:20 ET)]
Rick Marken (970927.0900)
Fred Nickols (970927.0933 ET)
Fred:
Communicating via the written word is very difficult.
Rick:
But not impossible. I have no trouble at all communicating
with Bill Powers, for example. It seem to me that the people on
CSGNet who express the most concern about the limits of
communication are those who (like B. F. Skinner) seem to think
that science is a matter of selecting the right turn of phrase.
I am confident that you and Bill have many of the same referents
for the words and phrases you use, a circumstance that is bound
to facilitate communication between the two of you. However, if
you are to "spread the PCT gospel" (my own figure of speech for
getting others to understand and support the PCT way of looking at
things) you will have to find ways of communicating with those who
don't hold the same referents. If you don't, your efforts to
spread it will fail.
Rick:
I'm reluctant to respond your post because I would have to use
words to do it and you seem to have a different concept of what
words are about than I do. You seem to believe that there is some
magic set of words I could line up to "correctly" communicate
my thoughts. I don't think this is the way communication works
at all. I think we vary the words we say in an attempt to communicate
an _idea_; it's the idea that matters, not the words. I think
we continuously try to determine whether or not our words have
gotten the idea across by listening to the replies we get and
seeing whether those replies evoke, in us, something like the idea
we have in mind. We iterate this process until we are satisfied
that we (the people involved in the commumication) are "of one
mind".
Except for the first two sentences in the paragraph above, I agree
with you. Regarding the first two sentences, why don't you describe
for us all a PCT-based experiment in which you would test the idea?
Rick:
I have no doubt the Bill Powers and I are of one mind
about the idea of PCT. I know this because Bill rarely tries to
correct my comments about PCT.
You might want to revisit the reasoning in the paragraph immediately
above. It sounds like you're saying that Bill's silence constitutes
agreement--which might indeed be true--however, as a general rule,
I'm not sure silence signifies assent, let alone concurrence. A test
of your proposition, and I won't even pretend to label it an instance
of "the test," would be for Bill to answer this question:
Has Rick made statements about PCT with which you have
disagreed but not voiced your disagreement in ways that
others would know you disagree?
Rick:
More importantly, I know it becuase
I can check my understanding of PCT against actual perception. I
am able to produce perceptions of PCT model behavior and actual
human behavior that correspond exactly to my ideas (imaginations)
about PCT -- ideas that were oriignally communicated to me in words.
That's good. That's also where I'm headed.
Rick:
Communication is difficult, but it is certianly not impossible --
unless, of course, one of the people involved in the communication
doesn't want to understand the idea being communicated.
The question, of course, is which one?
Fred
If behavior controls perception, and if we are dealing with
causal-loop or closed-loop feedback systems, it would seem
equally true that perception controls behavior.
Rick:
But it is not, once you understand the _idea_ of control. I can
use words and formulae to try to get this idea across but you
are the one who has to get the idea and be able to apply it.I will use some words now to explain, briefly, why it is _not_
true that perception controls behavior. I know this is probably
just a waste of bits but maybe some of those listening in will
get the idea.
I believe you said above that communication is an iterative process,
so I doubt that another attempt at it is a waste.
Rick:
"Control" refers to the phenomenon where a variable is brought
to and maintained at a particular value, protected from the
influence of other variables (disturbances). In a negative
feedback control loop, the only variable that is under control
in this way is the perceptual variable. This control is achieved
by the operation of the feedback loop as a whole; not by one
or another variable in the loop. When we say "behavior controls
perception" we mean that the variables that are typically called
"behavior" (actions and side effects of those actions) are part
of the process -- the closed loop control process -- that keeps
the perceptual variable under control. To understand the idea
communicated by the words "behavior controls perception" or
"behavior is the control of perception" you have to understand
how a control system _works_. The operation of a control system
is the _idea_ behind these words. Once you understand how a control
system _works_ you will see why the statement "perception controls
behavior" is nonsense.
Being an ex-firecontrol technician, I think I have a pretty good
idea as to how a control system works, but let's check that out.
Back on the fantail of the destroyers I used to ride sat a 5" gun mount
(See how old I am?). Up forward, on top of the bridge was a "director,"
a rotating and tilting large metal box containing three seats, an optical
rangefinder, some joy-stick boxes, and a radar antenna on its top. Below
the director was the radar room, which contained most of the radar equipment
and the target designation system (TDS), a device that is very reminiscent
of your joystick tracking demos. Down in the hull is a space called the
plotting room, in which is housed a computer, the radar console, a
gyroscopic component known as "the stable element," some switchboards, and
other odds and ends.
A target is "perceived" and "tracked" by way of the radar and the circuits
that control the positioning of the radar. Variations in the signal
returned from the target are converted into error signals suggesting the
radar beam is "too far" in front of, behind, above, or below the target.
The director moves in response to signals fed to its motors, which serves to
keep the target more or less centered in the beam (but always with some
slight error signal because you never really know where the target is, only
where it just was).
Down below, the computer processes information about the target's position
over time and, coupled with information about the likely behavior of a shell
blasted out the barrel of the gun mount, calculates the position the gun
mount would have to assume in order to put that shell in the general
vicinity of the target at some future point in time. The computer issues
what are known as "gun orders." These gun orders describe a continuously
changing position in planes that are known as "train" (the gun mount
rotating on its base) and "elevation" (barrel raising or lowering). These
"gun orders" are electrical signals -- inputs to amplifiers controlling the
gun mount's motors. There are devices in the gun mount that provide
information about the gun mount's actual position in the form of electrical
signals to the same amplifiers. The output of the amplifier is an error
signal defining the difference between the gun mount's actual and its
ordered position. This error signal causes the motors in the gun mount to
operate, turning the gun mount right or left and raising or lowering its
barrel until the gun mount's actual position corresponds with its ordered
position. At some point, someone says, "Commence fire," and someone else
pulls a trigger, and electrical charges detonate powder casings and the
resulting explosion forces projectiles out the barrel, and so on and so on.
Interestingly, there are two very different kinds of feedback at work in
those weapons systems. First, there is the closed-loop negative or
error-reducing kind of feedback that serves to bring the gun mount to
correspondence with its ordered position. Second, there are positive and
negative kinds of feedback associated with the error signals driving the
motors that move the gun mounts. Positive, or in-phase, feedback made the
mount's movement more responsive. Too much, of course, made it jittery.
Conversely, negative feedback dampened its responsiveness. Too much of that
made it sluggish. The trick was to find the right balance between positive
and negative feedback in the amplifiers so as to get a rapid, smooth
response from the gun mount. In short, we sought to control both the
position and the "behavior" of the gun mount.
I think I know how control works, Rick, from entire weapons systems right
down to single little servomechanisms that make the dials move -- which is
why I was attracted to PCT in the first place.
In any event, I think it's fair to say that the radar's "perception" of the
target controls the behavior of the gunfire control system. I also think
it's fair to say that the orders issued from the computer to the gun mount
control the gun mount's position. I also think it is just as fair to say
that the gun mount's position is the result of the error signal out of the
amplifier that compares its actual with its ordered position.
COMPLETELY MISSING FROM THAT WEAPONS SYSTEM IS INTENT! IT DOESN'T INTEND A
DAMN THING; IT SIMPLY BEHAVES AS DESIGNED AND BUILT (PROVIDED I'D KEPT IT IN
FINE FETTLE). THAT'S THE BIG DIFFERENCE BETWEEN ME AND THAT GUN MOUNT OR,
FOR THAT MATTER, BETWEEN ME AND THAT ENTIRE GUNFIRE CONTROL SYSTEM -- I HAVE
INTENTIONS; THE WEAPONS SYSTEM HAD NO SUCH QUALITY.
So, I've always thought that in trying to figure out how what I knew about
control might apply to people and behavior, that allowances had to be made
for intentions, for purposes. People aren't simply servomechanisms. They
are not gun mounts to be issued orders and expected to comply with great
alacrity. Nor are they amenable to having their positive feedback loops
adjusted to increase their responsiveness, or to having their negative
feedback loops adjusted to improve their stability in the same way I used to
adjust the positive and negative feedback loops in the amplifiers that
controlled so many servo motors.
Again, I think I understand the "phenomenon of control," Rick. But, if
there's something wrong with my understanding of it, please let me know.
[
Fred Nickols
Chief Fire Control Technician
United States Navy (Retired)
nickols@worldnet.att.net
Source: