[From Rick Marken (950929.0840)]
i.kurtzer (950928.2245) --
I thought your stream of consciousness description of the nature of control
was absolutely brilliant! It was as though James Joyce had rewritten my paper
on "The nature of behavior: Control as fact and theory".
Really nice, Isaac.
Bruce Abbott (950929.0825 EST) --
Bruce! Get ahold of yourself, you're behaving like a two-year-old!
Don't listen to Bruce, Bruce. He's behaving like a control system.
Me:
There is one interesting way to prevent a control system from becoming
active...you can prevent the control system from being active (acting to
bring r-p to 0) by getting p into a range where r-p produces no change
from 0 error (the flat part of the error curve).
Dick Robertson (950928.2300CDT) --
This is fascinating. Will you explain it in more detail?
Rick, how can I simulate that?
The simulation is available as a HyperCard stack so you are welcome to a
copy if you have a Mac; I could make a Pascal version pretty easily, I think;
then you could run the compiled version on a PC.
The demo is very simple; there are three cursors on the screen; the computer
is controlling all three; it is controlling one in the usual way (basic
control model) and it is controlling the other two using versions of
the "range of control" model.
You disturb all three cursors by moving the mouse. All disturbances to the
cursor controlled by the ordinary control system are resisted; the computer
keeps that cursor on target. Disturbances to the "range of control" control
system (the one with a comparator function like the one I posted) are
resisted as long as they don't move the cursor (the controlled perception)
outside the range of control. Once you move the cursor out of the range of
control it is no longer controlled; the control system does nothing to try to
move it back to the reference state. If you move the cursor back into the
range of control it is "captured" by the "range of control" control system
and disturbances to the cursor are again resisted (as long as they don't move
the cursor back out of the range of control).
It is definitely a thought-provoking demo -- and VERY simple to program. If
you think of the cursor as the perception of sweets and your mouse
disturbances as the location of an apple pie, then you can see how, when the
apple pie moves the perception of sweets into the range of control, the
perception of sweets is "goobled up" to its reference level. As long as the
apple pie disturbance is located so that the perception of sweets (cursor)
is uncontrollable, the sweets control system remains happy no matter what
happens to the perception of sweets.
I happen to love Nestle's chocolate bars (I think its the foil wrapper); and
if they are in my refridge (I like them cold) I eat them (control for the
taste). But I can keep myself from eating them (well, even perfect people
have to worry about their weight;-)) by keeping them out of my range of
control; so another control system in me keps me from buying them at the
market and bringing them home and into my range of control.
Best
Rick
PS. I've just gotta do a reference signal calibration test here:
Is the OJ defense team a bunch of assholes or what?