<[Bill Leach 940721.10:08 EST(EDT)]
NET
Once again, I should be "about other things" but thoughts about PCT are
winning for the moment.
I was thinking some more about the nature of some of the discussions here
on CSG-L and a remembered experience with an electronic control system
from many years ago "popped" into my consciousness.
I remember thinking to myself at the time (worded a bit differently
today however), that a control system that consists of many individual
control loops each with a wide dynamic range for output is very complex
to troubleshoot when problems occur.
Indeed, not only is it difficult to determine how a particular control
loop failure may alter the behaviour of the control system but it might
be almost impossible to even determine that a control loop failure has
occurred at all through the use of only non-invasive observation.
I can't help but think that (at least the more complex) living control
systems are really made up of nearly uncountable control loops of this
nature. We (quite correctly) talk about a "single perception" a "single
error" signal and a single control loop. We are also correct (at least
over what is probably a significant range) to say that Behaviour is the
control of perception and that splitting the system into its' constituent
control loops does not alter the truth of that statement.
However, as soon as we try to discuss "real world" examples (such as the
dog and the cat) that have not been modeled and tested against real world
examples we can quickly loose whatever common understanding of terms that
we may have had. The biggest problem that I can think of is the
definition of the "perception(s)" for the example.
When a dog chases a cat, the actual number and nature of perceptions
under control that are directly involved in that activity would probably
shock even the most thoughtful PCTer much less one of us newcomers.
My personal experience with some electronic control systems tells me that
occassionally and under certain circumstances, the systems own control
loops "disturb" each other. This is not necessarily bad and in the case
that I am think about, it is definately not bad but it can quickly cause
one that is "trying to analyze" how the system will behave in the face of
disturbance very difficult.
Another example that is often used but is quite useful for my purposes
here is the example of "opening a door". Control loops within the
control system disturb each others perceptions. When an organism sets a
reference for a door's position as being open and acts to achieve the
perception (and the acting WILL occur because the reference was set AND
the organism has all of the requisite control loops and methodolgy for
such control - ie: it has learned how to open a door sometime in the
past), the actions WILL perturb other controlled perceptions. A person
might even fall over which would be a control failure but not of the
original perceptual reference's system. This example illustrate some of
the difficulty in "arm chair" analyzing such a complex control systems.
Even when applying the TEST, the conclusion concerning the "controlled
variable" could be incorrect if the variable disturbed by the test is a
disturbance to more than one controlled perception. The organism's
actions then would (in effect) be the vector sum of the actions of the
relevent control loops.
I believe that the proper term for such a phenomenon might be the
"incidental effects" and could be quite a problem when the researchers
choice for a variable is closely related to one or more perceptions but
is not the direct object of control to any of them. As I see it, such
errors whether made be me or a PCT researcher in discussion here are not
significant as far as the theory itself is concerned. Simple tests have
shown the phenomenon of control and observations of truely "macro" human
behaviour such as Bill made in B:CP are again strong evidence of control
system operation. The rest is "details", important to understand but
errors are not critical to the theory unless actual proof can be offered
that control phenomenon is NOT operative.
This posting is not an attempt to say that there is anything wrong with
what sorts of discussions that we have, nor that we should not use
examples and analogies but rather to recognize the truth of statements
made by Tom (and others):
This is why it is so crucial that anyone who wants to understand PCT
look beyond the words they see on this net, or in our sometimes hard-to-
locate publications. A deep understanding of PCT can only come from
grasping the significance, if not the computations, of the quantitative
modeling. The PCT model is not about the words; the words are always
inadequate for expressing the quantitative model, and the necessarily
linear structure of language can never convey the simultanaeity and
continuity of the model in action.
-bill