Bill Powers writes in response to my query about internal states.
<snip>
We haven't been much concerned with modeling internal states (although
they're important). PCT is basically concerned with _behavior_ that others
can observe, and with understanding it as a control process. The control
loops with which we're concerned always pass through the environment where
they can be observed. The models we use propose only the simplest internal
functions: perception, comparison, and conversion of error to action.
Simple models based on these functions reproduce human behavior (in
simulation) quite accurately in simple situations.
<snip>
Jim Dix here...
Thanks for the welcome. Yes, I recognize the difficulty of modeling
control systems when one does not have much to go on -- akin to trying to
figure out how a cruise control system works without a block diagram.
One of less scientific portions of Control Theory is the artful practice
of determining a viable model without knowing its design. It is this I'm
trying to tap into. However, if the emphasis is merely on behavior of a
"black box" in this group, I am undoubtedly out of place. My apologies
for intruding.
Respectfully,
Owleye