[From RIck Marken (930825.2200)]
Well, let's try to untangle some knots.
Martin Taylor (930825 19:000 --
You misunderstand again. Is it too much of a disturbance to you to
perceive me agreeing with you?
No. And I loved having you agree with me. I was just using your comment
about generosity as an excuse to rail about "political" concerns that
really had nothing (and everything) to do with what anyone in particular
had posted. I was addressing my comments to god -- not you. Although I
don't believe in god, he's nice to talk to sometimes.
So I have no serious dispute with the rest of what you say. I LIKE
having you agree with me and having me agree with you.
Michael Fehling (930825b) --
I have been aware from the first of your aggressive penchant.
I'm a pussy cat once you get to know me.
For example, when you ask "Is there a big transparent retina hanging over
the factory deriving perceptions of things...", I have only one answer--Of
course not!
When one asks (as Oded did) "What is the factory controlling for" then
I think of "the factory" as the system doing the controlling. A control
system must be able to perceive the variable(s) to be controlled and it
must be able to act in order to influence the environmental correlates
of those variables. Therefore, if "the factory" is the controlling system,
it must have a perceptual input function. I could not think of what might
be considered the perceptual input function for "the factory" as a whole
so I asked what that function might be. You could have answered this
apparently silly question exactly as you answered Oded's intelligent
question about "What is the factory controlling", as follows:
the model we are building seems consistent with PCT in that
an organization's management team is making business-process decisions
based on
(a) their shared perception of the world as they would like it to be
(a reference perception), and
(b) their perception of past, present, and (most importantly) future
conditions, these preceptions being a function of inputs to
the organization as a whole (e.g., changes in a global data base)
or to organizational subgroups or to individuals.
So (b) tells me that "the organization" is not controlling anything. There
is a "management team" (a subset of the individuals in the organization);
each individual in this team is controlling perceptions that are a function
of "inputs to the organization". Each individual on the team has one
reference perception -- and it is exactly the same for each individual.
The individuals apparently control their perceptions by making "business
process decisions". So it sounds like your model is a perceptual control
model; the controllers are individual organisms who act in order to bring their
perceptions of the state of the organization to their reference levels.
So "the organization" isn't doing any controlling -- just the individuals
in that organization. The state of the organization (as perceived by
each individual as well as by you, the omniscient observer) will change
as a result of the controlling done by each individual; but the organization,
per se, is not controlling anything.
From the organization's viewpoint, control can only be an
attempt to bring perceptions (e.g., situational awareness) into correspondence
with reference perceptions (e.g., desired situations).
How can "the organization" have a viewpoint? The organization is a collection
of individual control systems. Each control system is trying to bring it's
perception(s) to reference levels -- so the viewpoint you describe is that
of each individual in the organization. Behavior in real organizations
suggests that the perceptual variables controlled by each individual
are not -- and often should not -- be the same. But even when the perceptions
that are controlled are about the same, they are almost certainly controlled
relative to varying reference levels that are, at any time, unlikely to be at
the same level for any two individuals.
But, our model goes
beyond this subjective perspective by seeking to clarify how this
organizational process bears upon changes in the external environment.
I should hope so; isn't the whole goal of the research to see how the
controlling done by the individuals in the organization influences
the state of the organization (which is the external environment
with respect to your modelled individuals).
To the
extent that PCT won't allow the interaction between a system and its
environment to be investigated then it's one of those "radically subjective"
theories that I've already explored and rejected.
Fear not -- PCT "allows" you to study the interaction between a system
and its environment. In fact, you MUST study this interaction to know
what the controller is "doing". The goal of PCT research is to find
the aspect(s) of a system's external environment that correspond to
the perceptual signals being controlled by that system. We can't
perceive what the system perceives -- but we can find aspects of the
system's external environment (using the test for the controlled
variable) that correspond to those perceptions. The term "controlled
variable" typically refers to an external environmental correlate of
what, from the system's perspective, is a "controlled PERCEPTUAL
variable".
In sum , I offer Oded's questions (that you say you intended to trash) to be
quite evocative and highly conducive to good communication. He took care to
express them in terms of specific issues I raised in my original post.
Frankly, I also found them pleasantly insightful. Answering them forces me to
"put my cards on the table" about what is being modeled.
But not all of them. I have a couple more non-insightful questions to
see what else is in your hand:
1. What, precisely, are the external environmental variables controlled
by the model individuals?
2. How do the actions (decisions) of the individuals influence these
variables.
3. How many levels of control are there in each individual? What variables
are controlled at each level?
4. If the reference perceptions change as a result of changes in
perceptual input variables (as you say above) then don't these
references soon become quite different for each individual as the
simulation progresses?
I think a diagram of one of your model individuals (including environmental
connections, of course) would help clarify things quite a bit. Here's
how we would diagram a simple control loop for a model individual:
r
>
v
---|C|---
^ |
p e
> v
_ _
ยทยทยท
f> >h> System
______________________
> > Environment
q<--g---o
^
>
d
Everything above the line is system (organism); everything below
the line is external environment. q is the controlled environmental
variable (like the position of your finger); its value is determined
by environmental disturbances (d) like gravity, as well as system
outputs (o) like the forces exerted by tensing muscles. f is
a perceptual function (in the organism) that turns q into a neural
signal, p. So p is perceived finger position. p is continuously
compared (in the comparator, C) to another neural signal, r. The
comparison is a subtraction and it results in another neural signal
called the "error",e. The error signal is VERY small; it is amplified
considerably by the output function,h, which, in this case, transforms
a small neural signal, e, into a BIG effect on the environment, o (the
force exerted by the muscle). The muscle forces influence finger position
according to the feedback function, g. This closes the loop. When
all the parameters of the loop are set up right, the system will
"force" p to stay equal to r (and q [finger position] -- the only
variable that is actually visible to an external observer -- will
stay equal to 1/f(r)).
That's just one little control system out of thousands in your typical
human being; these systems control perceptions as simple as finger
positions and as complex as political positions (at least, according
to PCT) -- and they control them all at the same time.
Best
Rick