PCT Puzzle

[From Dan Palmer (2000.08.07.1143)]

One aspect of PCT has puzzled me for a while, and I was hoping someone could
help.

In verbal and diagrammatic descriptions of PCT's basic unit (the control
system), there is typically a line separating the system from its environment.
The control system is said to keep a perceptual variable (internal) matching a
reference variable (internal) by protecting a controlled variable (external)
from disturbances (external). As I understand it, the line that distinguishes
the internal from the external in the case of human control systems is
equivalent with the skin of the organism. Please correct me if I am wrong
here, as I assume the above in what follows.

The thing that puzzles me is that if a system is defined as something like "a
whole consisting of elements, the interaction of which makes possible its
existence or action" (from Jarvilehto, 2000), and if some of those elements lie
outside the skin (e.g., the controlled variable), then how can those elements
be properly said to be outside the system, if we are to remain consistent with
our earlier definition of a system?

The two approaches were contrasted by Jarvilehto (2000), and also by Bateson
(1976, available at http://www.oikos.org/forgod.htm) who said "computer science
is input-output. You�ve got a box, and you�ve got this line enclosing the box,
and the science is the science of these boxes. Now, the essence of Wiener�s
cybernetics was that the science is the science of the whole circuit."

Thanks in advance,
Dan Palmer

Jarvilehto, T. (2000). The theory of the organism-environment system: IV. The
problem of mental activity and consciousness. Integrative Physiological and
Behavioral Science, 35, 35-57. Available at
http://wwwedu.oulu.fi/homepage/tjarvile/art4.htm

[From Bill Powers (2000.08.07.0638 MDT)]

Dan Palmer (2000.08.07.1143)--

The thing that puzzles me is that if a system is defined as something like "a
whole consisting of elements, the interaction of which makes possible its
existence or action" (from Jarvilehto, 2000), and if some of those

elements lie

outside the skin (e.g., the controlled variable), then how can those elements
be properly said to be outside the system, if we are to remain consistent

with

our earlier definition of a system?

Which is more important, to be consistent with one's definition of an
abstract concept, or to be precise about the physical situation one is
trying to describe? What is it that makes any physical organization work:
the actual interactions among its parts, or the abstract class to which it
belongs?

As I define "control system," it is a physical organization consisting of
an input function, a comparator (with an input from a reference signal),
and an output function which converts error to action. Placed in any
environment with approximately the right properties, this -- pardon me --
system will control its own perception of a variable in that environment.
The nature of the controlled variable is determined by the organization of
the control system's input function, not by the environment.

A person who has acquired the control system for driving one stick-shift
car can drive any other stick-shift car with similar properties. You might,
if you wish, say that the cars are each part of a different control system;
the driver, however, remains the same while the cars are interchangeable.
It is the nature of negative feedback control systems that they can
maintain good control of a variable over a range of variable properties in
the environment even without any adaptation, just as they can maintain
control even in the presence of unpredictable disturbances also without any
adaptation (where by adaptation I mean a change in organization or
parameters).

This being the case, one has a choice between defining a single control
system that can work in a large variety of different environments, or
redefining the control system every time the organism's surroundings change
to any discernible degree. The latter, while it fits one rigid definition
of "system," multiplies entities endlessly and offends against Occam. In my
opinion, it makes more sense to define the organismic part only as the
control system, even though for it to work the environment must be
restricted to a certain range of properties. If we speak of adaptive
control systems, the range of controllable environments becomes much greater.

One other thing. If you define the system so it includes the environment,
where do you stop the inclusiveness? A car's behavior is affected by the
wind, by bumps in the road, by hills, and by mechanical defects, as well as
by the driver. Are winds, bumps, hills, and defects to be included in the
definition of the system? And to take just one further branch, winds are
affected by insolation, by terrain, by high and low pressure systems, by
the rotation of the earth -- do those variables, too, become part of the
definition of the system? If we include in the system definition everything
that plays a part, do we not end up including essentilly the whole universe
from the time of the Big Bang? I think there is no way to define a system
in a completely general way that does not end in absurdity.

Best,

Bill P.

[From Rick Marken (2000.08.07.0820)]

Dan Palmer (2000.08.07.1143)--

The thing that puzzles me is that if a system is defined as something
like "a whole consisting of elements, the interaction of which makes
possible its existence or action" (from Jarvilehto, 2000), and if some
of those elements lie outside the skin (e.g., the controlled variable),
then how can those elements be properly said to be outside the system,
if we are to remain consistent with our earlier definition of a system?

This is a good point; the controlled variable is certainly part of
the "system" if the "system" includes all the variables that are
linked in a closed loop. In PCT diagrams it would probably be clearer
to use "organism" rather than "system" to name the entity above the
environmental boundary line. When we are talking about an artifactual
(rather than a living) entity then it's probably better to refer to
this entity as the "control system" rather than the "system". This is
really more of a verbal "puzzle" than a conceptual puzzle. It's usually
pretty clear, when you do modeling, which variables are in the environment
of the control system and which are inside the control system (organism)
itself.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
MindReadings.com mailto: marken@mindreadings.com
www.mindreadings.com