A General Theory of Manipulation

From Greg Williams (920828 - 2)

Bill Powers replying on "influence" (920828)

OK, Bill, I could counter your lurid family-relations examples showing the
ineffectuality of influence with lurid con-game and advertising examples
showing the effectuality of influence, but what's the use? Instead, why not
join with me in constructing a general treatment of manipulation from the
point of view of PCT -- a general treatment which will show the scope and
limits of manipulation? Here's a start. The axioms are supposed to arise from
orthodox PCT.

PROLEGOMENON TO A GENERAL THEORY OF MANIPULATION

Definition 1. Control is acting so as to maintain a match between reference
signals and perceptual signals in the face of disturbances which tend to alter
the perceptual signals.

Axiom 1. Comparing reference signals and perceptual signals occurs ONLY within
organisms, including humans.

Axiom 2. Given current and forseeable technology, other persons CANNOT directly
("causally") alter a given person's current structure of reference signals.

Axiom 3. Other persons CAN, in general, directly ("causally") generate
disturbances which alter a given person's current perceptual signals.

Lemma 1. A person can ONLY control his/her own actions, not others. In
particular, other persons cannot control a given person's actions. (By Def. 1
and Axioms 1-3.)

Definition 2. Influence is one person's controlling his/her actions so as to
alter another person's perceptual signals.

Lemma 2. Influence is possible. (By Def. 1 and Axiom 3.)

Definition 2. Manipulation is a given person's (the influencer's) altering of
another person's (the influencee's) current perceptual signals so as to affect
the influencee's current control processes in such a way that the influencee
acts so as to enable matching of certain reference signals and perceptual
signals of the influencer (that is, to enable certain control processes of the
influencer).

Lemma 3. Manipulation is possible. (By Defs. 1 and 2 and Axioms 1, 2, and 3.)

Axiom 4. Information about the influencee's current control structure (that
is, structure of reference signals) cannot be obtained directly, given current
and forseeable technology, but can be obtained indirectly, by the Test for the
Controlled Variable.

Axiom 5. Successful manipulation as set forth in Def. 2 would require
information about the influencee's current control structure so as to avoid
conflicts within the control structure and to assure consonance between the
current control processes and acts of the influencee which enable control
processes of the influencer.

Theorem. By Lemma 3 and Axioms 4 and 5, the chances of successful manipulation
are enhanced IF the influencer performs the Test.

This would be extended in a full treatment to mapping out how and why
manipulation has limits, the possible manipulation of others' FUTURE control
structures by altering their PRESENT perceptual signals, and so forth. Are you
game?

Greg

[Martin Taylor 920829 13:15]
(Greg Williams 920828 - 2)

I think a lot of the problem in the discussion of "controlling" other people
comes from a dual use of the word "control," which was much clarified by Bill's
posting of yesterday (reference not at hand).

Using Greg's set of axioms indirectly, or Bill's discussion, we have a
statement that the only legitimate use of "control" is the control of
perception, specifically the provision by an ECS of output signals that
by some unknown means bring the percept of that ECS nearer to the reference
level of that ECS. Any other use of the word "control" is analogic and
imprecise.

Any percept in an ECS corresponds to a CEV (complex environmental variable).
It is an estimate, more or less precise, of the current state of that CEV.
The word "control" is often used to indicate effects on the CEV, but that is
only legitimated by the assumption that the percept is a precise and immediate
representation of the exact state of the CEV. One should never say that one
"controls" the location of the car within its lane on the road. One should
say that one controls the percept of the location of the car....

The word that has been used to describe external effects has been "influence,"
but this seems inappropriate, because it has the S-R connotation that a
force has been applied, but the result not noted. We need a different word
for what happens to the CEV corresponding to a controlled percept. For
lack of a better word, I propose here to use X-control (externalized control).
One can X-control passive CEV's in the real outer world. One cannot control
(P-control?) them.

Bill pointed out that one does incorporate the actions of other people in
control loops. We want the child to be in the garden, not in the middle of
the busy street. If one action does not bring this desired percept to its
reference, another might, and we continue acting until the percept is within
its reference. We have controlled the percept of the child's location. We
have X-controlled the child's location. We may have done this by restricting
the child's range over which it is capable of controlling its own perception
of its own location, or by influencing it to change its reference level for
its percept of its own location. Whatever happened, we X-controlled the
child in at least one aspect. From our point of view, it may be more difficult
to X-control a child than to X-control a car, but the principle is the same.

I hope this clarifies something.

Martin