Subject: Black Box - RKC

from Bob Clark

to Bill Powers.

Re: the 4-port black box, your note of February 11. Of course one
can describe the situation in terms of its being "affected by the
state of some external variable." However, if one is curious about
some particular box, it must be intentionally "disturbed" if one is
to discover the nature of one (or more) external variables it may be
controlling. While discussion in terms of "states of variables" is
certainly possible, I find a more experimental approach more useful.
I also think some people are unfamiliar with the word "state" in this
sense.

You suggest a difference between the case where there is an external
link between terminals #4 and #2 versus an independent variable
applied to #2. The experimenter, being outside the box, can observe
any external connections. And any disturbance from the experimenter
is (really, by definition) an "independent variable."

I am intrigued by the question of the boundary between the Box and
its environment. To the experimenter, it is quite clear. The
environment includes everything except what is inside the Box. But
what does the Box perceive? Its Behavior is the Control of its
Perception. But what does it perceive? Only the disturbances of
those connections that cross the walls of the Box. And its output
actions reflect the differences between disturbances affecting
terminals 2) and 3). In addition, some kind of connection must exist
outside the Box, not perceptible by the Box but observable by the
experimenter. If these relations are not found, the Box does not act
as a Negative Feedback Control System.

These considerations seem to me to be applicable to many non-living
systems, and the use of non-living systesm as extensions of human
systems is impressive.

Enough for now. Regards, Bob Clark