The Test, Hal's "Theory"

[From Rick Marken (930930.0930)]

Avery Andrews (930930.1210) --

You might want to take a look at:

  Kelso, Tuller, Vaikoitis-Bateson and Fowler (1984), `Functionally
   Specific Articulatory Cooperation Following Jaw Perturbation During
   Speech: Evidence for Coordinative Structures', Journal of Exp.
   Psych: Human Perception and Performance 10:812-832.

Yes, excellent; thanks. A perfect "almost" test; there is
definitely a hypothetical controlled variable (lip aperature)
that is maintained against disturbance by changes in the
subject's actions. I have another article by Abbs and Gracco
(Trends in Neuroscience, vol. 6, 1983) where they give the
lip traces when a sudden load is applied. I think both groups of
researchers would have learned a lot if they had VARIED the
disturbance smoothly during the utterance (something they
could have done easily with the equipment, it seems). In
that case the maintenance of the time course of lip
aperature during the utterance (as it is in the undisturbed
case) would have been quite evident. Despite the obvious
control displayed in the Abbs/Gracco study I find this
statement (right next to the figure showing controlled
lip aperature):"the compensatory responses cannot be
ascribed to a closed-loop, feedback process, especially
as these multiple strutures are controlled independently".
There is clearly a general neurosis about closed loop
control in the behavioral sciences; it's really quite
pathological. What are these people afraid of?

Hal Pepinsky (930930)--

Thanks very much for the description of your model of human
nature. I think this can be the basis of a helpful discussion
of the difference between a PCT type theory and the type of theory
you describe (your type being, I think, by far the most common
in the behavioral sciences). I think you will see that PCT does
not compete with your theory. I would call your type of theory
"descriptive" while PCT is "generative".

The problem with your theory (from my perspective) is that it
"explains" in terms of the very phenomena that it is designed
to explain. For example, you say:

Two contending sides of human nature are described in every
religious/political theory tradition I have encountered--one side that
hurts others unless it is punished into submission, and one side that
operates on the belief that compassion pays.

So the theory is that there are two "sides" to a person. These
sides have no properties other than those that correspond to
what we observe: people hurt each other and people show compassion.
Why do we observe these phenomena (hurtfulness and compassion)?
Because people have hurtful and compassionate "sides"; the
phenomenon (hurtfulness and compassion) is explained in terms of
itself. This is quite a different approach to theory than that
taken by PCT. In PCT, the phenomenon to be explained (control)
is not part of the model that produces it; the model consists
of components (perceptual and output functions, neurons carrying
perceptual, reference and error signal's) that, when hooked up
in a particular way, produce control. The PCT model generates
the phenomena we observe (the phenomena of control) from com-
ponents that, themselves, do not exhibit these phenomena. When the
model generates the same behavior as the system being modelled,
we say that the model is an explanation of this behavior.

In pure form, the retributive side
entails having unchanged motives, while the compassionate side entails
changing one's motives as a function of where one finds referent
motives in others, not to mimic them, but to accommodate them.

This is starting to sound more like a generative model; "motives"
are theoretical constructs that might be responsible for what we
see ((hurtfulness and compassion) without, themselves, exhibiting
this phenomenon. But your description says nothing about how
motives are "hooked together" to produce the phenomena we see;
all that happens is that a "side" of a person (which already
behaves as we observe -- hurtfully or compassionately) "entails"
motives. It is not clear how this "entailing" happens or why it
is necessary since we already know that people are hurtful or
compassionate becuase they have these "sides".

Compassionate interaction is two-sided: either party, which may be an
individual or a group, may respond to my bid to have motives orbit
around one another by persisting in a straight line. A retributive
response would seek for my motive to prevail over the opposition.

This is a description of the phenomena to be explained; comapassion
and hurtfulness. It is good to have such a description; now we
need a model to explain how it happens. Your model says that it
happens becuase people have compassionate and hurtful "sides";
that is, it happens because it happens. Most behavioral scientists
seem quite content with this kind of theory; that's fine, but it
really has nothing to do with the type of model we are trying
to build with PCT.

In my inquiry, I ask what results people get for taking the one
path or the other.

So it sounds like you are not trying to test any theory; you are
just looking at the consequences of acting with hurtfulness or

I suppose that the more we learn (and indeed find the more
I learn) the more attractive the choice of trying to engender
compassionate interaction becomes, simply because we have the safety
of knowing how others have succeeded in such attempts.

This "theoretical" approach is a good example of what we call
"generalization" -- a very common approach to "theorizing" in
the behavioral sciences. Generalizatoin type "theories" are
based on the belief that "if something happened in the past
then it will probably happen again -- probably". So you say
that compassionate interaction has virtually always succeeded
(in some sense) in the past; so you expect that this kind of inter-
action will succeed in the future. Unfortunately, such generalizations
can lead to quite unfortunate results in individual instances;
this is why real sciences have moved past generalization to
generative models (ie. understanding). PCT is an attempt to develop
a generative model of human nature.

PCT falls into the realm of modeling what people do when their
motives are supposed to be independent of others'. Hence, it
describes one side of human nature to me. What say?

Well, as you can see, I don't think PCT "describes" any "side"
of human nature (a "side" is not part of the PCT model); PCT is
a generative model that produces the same kind of controlled
results that we see being produced by people (and other living
systems). So your analysis of the realm into which PCT modeling
falls is (from my perspective) based on a misunderstanding
of the nature of the PCT model; PCT is NOT a descriptive model
or a set of generalizations; it is a model of the processes that
we believe are responsible for ALL purposeful behavior.

There is a problem with passing off generalizations about human
behavior (like yours) for explanations (like ours). The problem
is that people might mistake your generalizations from explanations
and try to base their actions towards others on such generalizations.
I think that many of the social ills that you (and I) abhor are a
direct result of the fact that many people deal with each other
on the basis of such generalizations. PCTers are trying, as best
as we can, to disabuse people of their reverence for generalizations.
But the belief in generalization (among professional behavioral
scientists as well as the lay public) is pretty deep. The only
solution is education -- ergo, CSGNet.