[Martin Taylor 920928 16:45]
The following question was posed to me by a colleague, and I thought it of
sufficiently general interest to communicate it and my answer to CSG-L.
On the first day, David moves the garbage can into the middle of the ailse.
Martin moves it back.
On the second day, David moves the garbage can into the middle of the aisle.
Martin moves it back.
On the third day David moves it again.
Martin leaves it.
Is Martin controlling for the position of the garbage can?
No, he stopped.
Yes, he has determined that if he ignores the disturbance, it will
go away, eventually.
How does David test whether Martin is controlling for the position of the
You are talking about a phenomenon I label, with great originality, "giving up."
When two ECSs are in conflict, it is impossible for both to bring their errors
to zero simultaneously. That's almost a definition of conflict. Typically,
both control systems are inside one body, and in many real-world situations
there is more benefit to the body from allowing one of the control systems
to satisfy its reference, than from leaving the real-world variable at a
compromise level. Sometimes, the world is such that there exists no compromise.
If one ECS, for example, has a reference to perceive the body as bicycling to
work, and the other has a reference to perceive it as driving, there is no
compromise intermediate position (possibly a motorcycle, but let's suppose
none is at hand). The mechanism is that one ECS gives up attampts to control.
Either the person cycles, or the person drives, but not part of each.
(There are other reactions to persistent error, including reorganization, which
could lead to the possibility of both references being satsfied, the problem
being seen in a new way that eliminates the conflict).
The "giving up" mechanism is needed in a related situation, a shift of
attentional focus in response to an alerting signal. The alert causes some
ECS that was not controlling to attempt to exert active control. This means
that some ECS that had been controlling must "give up." If that did not happen,
the body would not survive very well--imagine a primitive man eating berries
when an alerting system showed him a tiger about to pounce. If the berry-eating
percept HAD to be satisfied, or even if it caused error in the perceptual
control for escape, there would be a nice dinner for the tiger. And this might
happen, if the berries were particularly tasty so that the man hesitated about
the escape (conflict situation).
So, we must have a built-in mechanism for giving up in the face of difficulty.
It can't be too strong, of course, and I suspect that in humans it is too weak
for our own good (leading to war and similar bad things). But it is there.
If attempts to control a percept are persistently thwarted, there are two
possible reactions--try harder or give up, fight or flight, for example.
Now to your original question, how to apply The Test. I don't think you can,
unambiguously. The very first time you move the wastebasket, and I move it
back, you know that I am controlling for some percept that includes (at that
moment) having the wastebasket in its original place. But you don't know
just what it is that is really being controlled for. Perhaps the wastebasket
is blocking a draft that makes my feet cold, and you could have replaced it
with something else that would have kept me happy. Perhaps I want to see a
rigidly aligned set of objects in the office, and you could have satisfied
my reference by moving two or three other things into the aisle. All you know,
and this you do know, is that I am controlling now for a percept that includes
the wastebasket where it was.
My claim is that The Test is always ambiguous. P can tell that Q is controlling
for a percept that incorporates some CEV that P has disturbed, but P can
never tell that the percept Q is controlling for corresponds exactly to the
CEV that P's percept corresponds to. The CEV is in the world, but only
percepts can be controlled. The percepts are private, so in a sense the
CEVs are, too. But the correspondences of CEVs can be tested, and that is
what you do when you try variants on moving the wastebasket, such as putting
some other obstruction in the aisle, or moving the wastebasket to some other
unobtrusive place. Such experiments can never give exact answers (pace
Rick Marken and Bill Powers) but can only be evaluated statistically.
The fact that on the third day I leave the wastebasket where you moved it is
irrelevant. You got your information on the first trial. On the third day,
I obviously am not controlling for a percept that includes the wastebasket
in its original position, or if I am, there is an internal conflict in me
that causes some other controlled percept to override that one. For example,
I may perceive that by persisting in moving it back I will cause you to get
angry, and this would create error. My reference for seeing you satisfied
has overridden my reference for having the wastebasket where it was.
Reference levels change. What we control for at one moment may not be what
we control for at another moment. Why not? Because some other percept changed
at a higher level, altering the error in this higher ECS, and this changed
the reference levels in all sorts of ECSs at levels below it. Another reason
might be reorganization, which naturally changes reference levels by virtue
of the changed interconnections of ECSs. The changes in reference level are
an independent reason why experimental tests can only be statistical, and
why one can never be sure to what degree one's percepts coincide with the
percepts of other people.