<[Bill Leach 951114.19:20 U.S. Eastern Time Zone]
[Shannon Williams (951114)]
How I have a serious problem with trying to conceive of how we would
use "fuzzy" logic to "understand what information in the world the
control loop needs to perceive".
Bill P. talked of a matrix that cooresponded to 'just left of center',
etc. 'F' logic is the theory behind this matrix. 'F' logic allows you
to build systems without using explicit equations and algorithms.
I think that part of our problem in discussing this is that the
mathematics necessary to describe what happens is orders of magnitude
more complex than the mathematics necessary to emulate the process in a
computer.
How would you describe the process by which a control loop is activated?
At this point in my understanding, I wouldn't! This has been the subject
of many many hundreds of messages. A number of different proposals have
been suggested for how this might come about.
How would you describe how an activated control loop responds?
An activated control loop "responds" by controlling its perception to the
reference value. I know this probably reads like a "smart ass" answer
but it is most definately not meant to be such.
The point is that for at least some perceptions the reference is not a
"bi-directional" pair. For some others, there appears to exist a "dead
band" where some observable deviation in the perception seems to not
result in any change in the error signal level but deviation beyond some
additional amount is resisted. These would both be examples of control
loops that would be active but might not appear to the observer to be so.
In addition, some discussion has occurred about the idea that control
loops for some perceptions might "inhibit" the operation of certain other
perceptual control loops. Even though all of these examples deal with
simple control loops the effect of how they are "wired" quickly becomes a
complex matter to examine.
Do your two quoted statements above, mean different things?
I could be quite wrong in what I thought that you were saying... but I
still don't think so.
What I am trying to say might best be expressed with a generic sort of
example:
A control system has a reference for a perception "P" of "P greater than
100 units". As long a perception "P" exceeds the reference (no matter
what the cause), the control loop will appear to an observer to be
"inactive". If for some reason perception "P" drops below 100 units then
the control loop output will act upon the environment to return the
perception to its reference.
To the observer, it might appear that the change in "P" "triggered" the
activation of the control loop (particularly if the gain of the loop was
rather high). The change in "P" did NOT stimulate the control loop into
operation, the control loop was operating all along.
A difficulty here is that any form of suggestion that the environment
"caused" the behaviour is highly misleading. Changes in the environment
only "cause" changes in behaviour to the extent that the changes, through
physical laws, disturb controlled perceptions. The organism is always
open to changing the references or even the perceptions to be controlled
and to do so independently of what the environment might be "doing".
The usual response to what I just said above is that to do so might
result in the destruction of the organism (effectively self-destruction)
and that is both true and irrelevant.
-bill