[From Bill Powers (980806.1406 MDT)]
Bruce Gregory (980806.1230 EDT)--
What does it mean to say that someone is controlling for a perception at low
gain?
It means that there is a reference signal and a perception being compared
with it, and an output function that produces actions to oppose the error,
but that the amount of opposition is very small even for large errors.
Am I controlling my perception of "freedom of speech" even when I am
completely unaware of doing so?
Could be -- I don't know of any general principle that says you MUST be
controlling it, but you could be doing so unawaredly while your attention
is on something else.
I ask this because I have always been
puzzled by the status of perceptions that I could be controlling, but
apparently am not. For example, I have a preference for Butler's Irish
Chocolates, but I am not acting to make what I perceive match that desired
perception.
This is similar to the question about long-range goals. If we take the
control model literally, as I prefer to do until given a reason not to, we
would have to explain why huge errors don't always and immediately lead to
very energetic actions. The simplest answer is to accept the implied
observation that they usually don't. That means that the output gain is low
-- in fact, it is just enough to account for the amount of action that does
occur for an error of the observed (or deduced) size.
But there's another possibility you may like more. When we make plans,
we're operating in the imagination mode, matching perceptions to reference
signals, but internally synthesized perceptions, not real ones. In this
configuration, the control systems can have very high sensitivity and tight
control, keeping each (imagined) perception very close to its reference
level. So as long as we deal with the goal via imagination, we can keep the
errors very small.
In fact, I can't see any other way we can handle goals that we think are
very important but which can't possibly be satisfied immediately. We
imagine how it would be to have reached the goal. The result, of course, is
low error and no action. If we do act (as for example in getting a college
degree), it is not because of the vastly important goal of graduating,
which is being satisfied in imagination, but because of more mundane
day-to-day goals like keeping up with the homework assignments and lab
projects.
It seems odd to say that I am controlling this perception at low
gain. I don't seem to be controlling it at all. Yet under the right
circumstances, I could be controlling it with a high gain.
I think we learn how to avoid being driven to painful amounts of effort by
large, hard-to-correct, important errors. One way, of course, is to give up
all important goals, or to learn how to ignore them and leave their
reference signals set at zero until the time comes when action might have a
chance at success. Another is to turn down the gain, so large errors don't
imply strenuous action. And another is to satisfy the goal in imagination
-- daydream -- so as to make the error less.
I remember how hard it was, as a child, to stop wanting something that I
desperately wanted, like a bicycle. It was terribly painful to want
something so much and know that I couldn't have it. My mother explained to
me that since I really couldn't have one just then, I would go on being
miserable unless I could learn how to turn off the wanting, and not want it
any more, at least for now. I guess this was an early lesson in going up a
level. So I did learn to do that, more or less. And it was strictly because
I couldn't stand being in that state of constant large error any more (as I
would describe it now).
I think the lesson here is that unless we learn how to deal with them,
large errors do indeed lead to large attempts to correct them, and to
painful frustration when they can't be corrected right away. I think that
most people have such experiences from time to time, and have to learn some
way to avoid having them again. While there are certain approaches that
would work, the solution found by any one person is up to that person.
Best,
Bill P.