[From Rick Marken (960108.1530)]
Bruce Abbott (960108.1055 EST) --
When I said I had a little bombshell to present, I didn't mean something
that would challenge PCT, but only something that might be surprising given
the pronouncements you and Bill made about Killeen's model in December.
This is still true in that neither of you recognized the control system
implicit in this implementation.
I have to question the value of a control system model that is _so_ implicit
that nothing can be learned from it about how organisms behave . If Killeen's
model really is a control system model (I doubt it but I'm willing to give
him the benefit of the doubt), he apparently didn't learn anything from it if
one of the three basic principles of his theory of behavior is that:
(1) Incentives [reinforcements] excite responding.
This statement is completely inconsistent with a control theory understanding
of behavior. In a control loop, it is the _difference_ between the perception
of environmental variables (like "incentives") and the reference
specification for that perception that "excites" responding. The result of
this "excitment" is control of perception (the perception of the "incentive"
is brought to and maintained at the reference level).
If a control model accounts for the behavior of organisms, then we know that
understanding behavior is a matter of understanding what perception(s) the
organism is controlling (or trying to control). I saw nothing in Killeen's
discussion of his model that gave the slightest hint that a control system
acts to control its own perceptions -- a pretty significant omission, I
think.
If we credit everyone who has built an implicit (or even an explicit) control
model with having a control theory understanding of behavior, then we would
have to give credit to people (like Killeen) whose verbal (and mathematical)
descriptions of behavior are often completely inconsistent with the notion of
control of perception. These verbal descriptions are important because they
reflect people's understanding of the nature of behavior and they influence
the kinds of things people want to know about behavior in order to understand
it. I don't think Killeen's verbal descriptions of behavior would
lead anyone to suspect that the most important thing you could know about
behavior is the kinds of perceptions the organism controls.
I am familiar with work done by some folks who built what they called "S-R
models" of bugs following moving objects. Although these model bugs were
called "S-R" devices they were actually control systems. They were S-R in
that they sensed the object (stimulus) they were following and responded to
this stimulus with propulsive forces from two rear "engines". But these
bugs were actually control systems because 1) there was a continuous effect
of their responses on the sensed position of the stimulus (closed loop
relationship between bug and environment 2) there was a negative effect of
sensed stimulus on response, creating a negative feedback loop and 3) there
was dynamic stability provided by responding that was the time integral
of the sensed stimulus.
So these modelers believed that they had built S-R systems (they called them
that) that could follow objects (control the relationship between their own
position and that of a target object). These modelers had built control
systems that they _called_ S-R systems. But I would not credit these modelers
with having made a significant contribution to our understanding of how
organisms "follow" objects because they did not understand that the behavior
of their bugs was organized around the _control of perception_. The model
organisms behaved the way they did because of what they were trying to
perceive (sensed position of other object at zero); not (as the modelers
thought) because of the rule relating sensed S to response R.
PCT is based on two important insights about the application of control
theory to living systems. The first is that the events we call "behavior"
consist of intentionally produced results of action: behavior IS control. The
second is that the results that organisms intend to produce are perceptual
representations of an external reality; behavior is the control of
perception.
William T. Powers had these insights after he took a VERY careful look at
what it means to "behave" and at how control systems control. Clearly, it is
possible to apply control theory to behavior, consciously (as manual control
theorists have done for years) or unconsciously (as Killeen might have done)
and remain oblivious to the fact that behavior is the control of perception.
But, for my money, if you don't know that behavior is the control of
perception (and what that means) then you have no idea what organisms are up
to and you are not on the trail of PCT.
Best
Rick