The Test and Conflict

[From Rick Marken (961007.1330)]

Avery Andrews (961004)--

Conflict is clearly a feature of human societies... but I don't see
conflictive control, as I understood (or maybe mis-understood) Kent as
describing it to be a major factor. I'd suggest `strategic thinking' as a
much more important factor.

I think this is an excellent point.

Before this discussion of conflict goes away I want to mention something
that came to mind while I was Testing some of Bruce Abbott's controlled
variables. Mark Abrams called last week and suggested that I might have
avoided a "conflict" with Bruce if I had gone about the Testing differently.
I argued that there was no conflict; I was just "pushing" on suspected
controlled variables and Bruce was "pushing back" to the extent that I was,
in fact, disturbing a controlled variable. Nevertheless, I think this is
worth discussing becuase I think that there is some confusion about when we
are seeing a conflict and when we are seeing plain old disturbance resistence.

There is typically no conflict when you Test for Controlled Variables
because when you do such a Test you are not trying to _control_ the suspected
controlled variable relative to a particular reference value. When I did The
Test with Bruce I was varying (controlling) a disturbance variable (the
disturbance variable was what I said to Bruce about the value of conventional
psychology) and _watching_ to see the effect of my disturbances on the
hypothetical controlled variables. But I was not trying to control that
variable myself. That is, I was not trying to control my own perception
of the value of conventional psychology.

The situation in "The Test" looks something like this:

     Testee | Testor
      p1<--- |
      > \
r1 -->C q -->p2
      > /\ / |
      e--->o1 d C<--r2
                \ |

The Testee is the system on the left; the Testor is on the right. q is the
hypothesized controlled variable. The Testor applies disturbances, d, to q.
These disturbances are controlled by the Testor. If the Testee actually is
controlling q (as shown in the diagram) then q will be protected from the
effects of d. Another system in the Testor (not shown in the diagram)
monitors the relationship between d (actually, p2) and q (actually a
perception thereof); if that system sees that q is protected from the effects
of variations in d then the Testor concludes that q is under control.

There is no conflict between the Testor and Testee in this situation. The
Testor is not actively working against the Testee's efforts to control q.
The Testee will resist the Testor's disturbances to q, and if q is
important to the Testee, he or she will aggressively resist all disturbances
to it. It might look like the Testee and Testor are fighting when this occurs
- - that is, it might look like they are in conflict. But there is no
conflict; just vigorous disturbance resistance.

Compare the situation that exists when you do The Test (above) to the
situation that exists when you are in conflict (below). In The Test (above),
Testee and Testor control two _different_ variables; the Testee controls q
while the Testor controls d. In a conflict (below), Testee and Testor
control the _same_ variable, q. Moreover, they control this variable relative
to two different reference levels: r1 and r2.

   Testee | Testor
      p1<--- --->p2
      > \/ |
r1 -->C q C<-- r2
      > /\ |
      e--->o1 d<---e

In a conflict, the Testor's disturbance to the controlled variable (d) is
part of the Testor's efforts to bring q to the Testor's reference level for
q, r2. This effort is a disturbance with respect to the Testee because it
tends to move q away from the Testee's reference level for q, r1. When there
is no conflict, the Testee's action (o1) joins with the disturbance to keep
q under control. When there _is_ conflict, the Testee's action is a
disturbance to the Testor's effort's to keep q under control; the Testor's
output (d) increases which leads to an increase in the Testee's output (o1);
both outputs continue to increase up to the limits of strength of each
system. If the systems are of approximately the same gain and strength, q
remains somewhere between r1 and r2; neither system keeps q under control.

The difference between what goes on in The Test and what goes on in conflict
is the difference between control and conflict. In The Test, the Testee can
keep q under control while the Testor keeps d under control. In conflict,
neither the Testee nor the Testor may be able to keep q under control. You
can see the difference between control and conflict using the rubber band
demo. When you Test to see if the subject is controlling knot position, you
control the position of your finger in one rubber band (d) and monitor the
position of the knot. You don't try to control the knot (ie. move it to any
particular position). If you did, you would quickly find yourself in conflict
with the subject, pulling your rubber band violently to compensate for the
Testee's efforts to keep the knot where he or she wants it.

It is not the apparent "violence" of the response to disturbance that
indicates conflict; the response to a disturbance -- especially a sudden,
large disturbance -- can be quite "violent" (such as the knee jerk reflex)
even when there is no conflict at all. Conflict is indicated when two (or
more) systems try to control the same variable relative to different
reference levels. Conflicts like this might not appear violent, especially if
the systems involved are low gain relative to the variable under mututal
control. But there is a conflict nevertheless.