[From Bill Powers (920920.0900)]
Greg Williams (920919-2) --
Until Pat finds a way of perceiving the actual effects of her
nutritional scheme, she is operating open loop.
No. From the first time I gave this example, I have said that Pat is
trying to control for her perception that the kids are eating healthy
food.
We're talking about several levels of control here. One level is the
control of what is put on the table before the kids. This level is
completely under Pat's control. She's in control of whatever she wants
to put there, whatever she calls it.
While still satisfying the serving-food goal, she can vary the kind of
food from healthful to unhealthful; that's an independent dimension of
perception. She wishes to serve food she perceives as healthful, and
she can also control for that because she can see the kind of food
she's serving. But unless she can monitor the actual effects of the
food, she can't control for its effects on health. Without a means of
perceiving the way the state of health varies as the type of food
varies, there's no way she can control health itself. All she can
perceive is that she's doing something that theoretically should be
good for the kids. To say that she serves "healthy food" is to assert
that the effect of the food is known. What's really going on would be
much plainer if you'd say that she serves food that she believes ought
to be good for the kids.
This is a language problem, not a theoretical problem. Language lets
us sneak in assertions without coming right out and stating them.
Before you can apply PCT to any situation, you have to take the
description apart to make all the hidden assertions explicit.
Lest you think this is a cop-out, consider the case where Pat is
attempting to control for her perception that the kids are healthy.
Again, the definition of healthy MUST be hers -- as the slogan goes,
it's all perception! How she came up with her definition of health is
not an issue here. In this case, if she sees the kids NOT being
healthy >(her definition), she performs actions which result in that
error signal being corrected. If she performs an action and sees the
kids >becoming LESS healthy (her definition), she performs different
actions.
Now you're talking about actual control. If Pat has ways of perceiving
the state of health of the children (her way, but what else have we?),
and knows how to vary the offered diet to counteract changes in the
indicator variables, then she isn't "trying" to control her perception
of their health; she's controlling it. The implication is that she
could make their health match any reference level for it by varying
their diet. If she can actually see how varying the kind of food
changes the kids' state of health, then she can control the state of
health. Unless, of course, they're already controlling it themselves.
I see no fundamental difference in control which involves other
persons (of whatever developmental stages) and control which involves
non->living entities, EXCEPT that in the former kind of control,
(some of) the other persons' wants must be taken into account for the
control to succeed without generating conflict in the others' control
systems.
I agree that control is control, whatever is being controlled. But
that little difference, I claim, makes all the difference.
We're pulling in different directions here. You're looking for
generalizations in which differences don't make a difference. From my
point of view, all this does is obscure the differences that DO make a
difference. I'm interested in how control of people differs from
control of inanimate things. To me, that is more "fundamental" than
ways in which such relationships can, by ignoring details, be viewed
as the same. When you say you see no fundamental difference here, to
me this looks like going UP the ladder of abstraction, which leads
AWAY from the "fundamentals" and in the direction of less and less
useful generalizations.
Anyway, it isn't sufficient to know SOME of the variables under active
control by the other person. You must know ALL of them. You can
control a person's reaching out by holding up something you have
reason to believe that the person wants. But you have to understand
that in reaching out, that person is shifting the center of gravity of
the body, so you must be sure that there is room for the person to
lean backward or shift a foot forward, and you must take care that in
leaning backward the person won't touch some wet paint, or that in
shifting a foot forward, the person won't pitch headlong down a flight
of stairs. You have to know how the person feels about having to reach
out to get what you're offering; the King, for example, might take it
amiss if you make him reach too far for the wine glass he wants.
Of course you may be unaware of the extent to which you're disturbing
another's hierarchy by trying to control some innocent-seeming aspect
of that person's behavior. To appreciate the full extent of the
effects, you'd have to use the Test systematically, with a far more
organized and extensive effort than anyone knows how to mount now.
There are more effects than people realize, in no small part because
in our culture, it's considered demeaning or weak to allow others to
see just how much their efforts at control disturb you. A person may
be perfectly aware of having his or her actions manipulated, but will
go along anyway simply to keep the manipulator from knowing the
murderous impulses that the controllee is holding in check. Robbers
and con-men often seem puzzled at what seems to them an overreaction
to what they did, when they're caught. They know a little, but not
enough, about what the victim wants. It's their ignorance of human
nature that gets them into such professions in the first place.
I think it's the SAME story if Pat wants to see ME eating healthy
(her definition) food. If she wants to succeed, she'll try to figure
out something I want which she can "link" to healthy (her definition)
food.
Again we have that language problem: "healthy food." Calling food
"healthy" is classifying it by its hoped-for effect. There's no
question that you can control for giving another person a particular
kind of food, while avoiding kinds the other person rejects. But if
you can't monitor its effects, there's no way you can say you're
controlling for the effect you hope it will have. As I commented a few
posts back, not entirely at random, hoping for something or wishing
for something is not the same as controlling for it.
I have a problem with terms like "linkable wants." What good are such
terms? They just make us try to interpret them in PCT terms. Why not
simply express what you mean in PCT terms in the first place? Is
"linking" a new phenomenon that requires a new explanation? To me, it
seems a step backward into vagueness and ambiguity. The general term
"linking" can apply in a lot of specific situations. You could "link"
my eating what you put in front of me to treating me nicely; i.e.,
make your nice treatment of me contingent on my wanting what you want.
If you didn't mean THAT kind of linking, then you shouldn't use
"linking."
Advertising writers appeal to consumers' patriotism to sell them
dinnerware at gas stations; who is controlling the higher-level
perception?
The consumers.
Why are you not controlling for being amused by seeing your cat drink
from the faucet, and in order to successfully control for that,
PURPOSIVELY providing the opportunity for the cat to get what you
think >(model) him to want, namely drinking from the faucet.
Providing an opportunity is like serving healthy food. What I can
provide on purpose is a physical situation under my direct control
that I think of as an opportunity to drink. But it doesn't matter to
the cat how I think of it. What matters is whether the cat uses this
situation as a means of drinking. I have no control over whether my
physical manipulation amounts to an opportunity or not. Half of the
time the cat switches its tail at me, ignores the sink, and drinks out
of his cup. It's not as if I can control what the cat's going to do.
That's not the relationship. I can't control for my amusement, because
I can't make the cat "take advantage of the opportunity." There's a
critical part of the process that's under the cat's control, not mine.
The relationship between me and the cat is not one of control.
It's control of a variable of a special kind: one that needs two
independent control systems to be controlled.
It's that, also. IN ADDITION to being (I think) an example of
simultaneous purposive influence by two interacting organisms.
Show me, in PCT terms. How are the two people carrying the bed
upstairs making each others' actions be in a desired state? What are
the controlled variables? The reference levels? The actions taken to
affect the controlled variables?
(from my 920219)
This isn't a case of symmetrical purposeful influence _of the
other's behavior_. Whatever mutual influences remain, they are only
disturbances which the other's control system can handle without
difficulty.
I need to think some more about your suitcase-lifting example. Can
you expand a bit on your final remarks?
I'm controlling an upward force at a reference level of 35 pounds,
independently of position. The LOL is controlling for position by
_varying_ an upward force, which need be only 15 pounds to move the
50-pound suitcase, as I am now cancelling 35 pounds of its weight. As
she moves the suitcase she will be disturbing my control system
somewhat, and my system will respond by bringing my upward force back
to the required level. The LOL will bring the suitcase to whatever
height above the ground she wishes; I'm not controlling for that.
At a higher level, I'm helping her with the suitcase because she let
me know she wanted help, and I decided, for numerous reasons, to
comply. I've got my mitzvah; she has the suitcase where she wanted it.
I lent her some lower-level systems to use in controlling the variable
she wanted to control -- not me, but the suitcase's position. It was I
who rearranged my momentary wants to allow her to use my muscles in
her control action. I could have said no.
···
----------------------------------------------------------------------
IBMers could probably build systems with identical blocks (amps,
integrators, delays, etc.) as those used in STELLA, so what could be
passed from IBMers to MACers and back could be lists of the blocks
and their connections and parameters.
That's a definite possibility. I haven't heard anything from Hammond
yet. Stella would have to be modified to accept a setup language. And
I think the language should be both generated by and interpreted by
the programs, to avoid errors and for the sake of people just starting
to learn.
-----------------------------------------------------------------
Best,
Bill P.