[From Rick Marken (970818.0830)]
Here is a modest proposal for a paper I would like to see
written by someone (like Bruce Abbott, perhaps) who is a
recognized expert in both operant behavior analysis and PCT.
The paper could be called something like:
Reinforcement: The Control of Perception
The paper would be about the fact that reinforcement -- the apparent
strengthening and maintenance of behavior by certain consequences
of that behavior -- is an illusory side effect of the precess
of perceptual control. The paper would begin by discussing the
point Bill Powers (970818.p0415 MDT) made this morning; the
reinforcement phenomenon (increase in behavior that is followed
by an increase in the reinforcing consequence) is really only
observed in a special subset of operant conditioning situations:
it occurs when a deprived animal first hits on the response that
produces the "reinforcing" consequence and it happens when the
animal is "redeprived" and allowed to use the same means to
produce the consequence.
But there are many (far more common) situations where we see an
increase in behavior that is followed by a _decrease_ in
the reinforcing consequence (as when "schedules" are changed)
and where we see a _decrese_ in behavior that is followed by an
_increase_ in the reinforceing consequence (as when non-
contingent reinforcement is given along with the contingent
reinforcement).
The paper would go on to show that all these relationships between
behavior and its contingencies can be explained by assuming that
the organism is trying to control a perceptual representation
of what has been called the "reinforcing consequences" of its
behavior. This part of the paper would show how a simple control
model account, in principle, for all these of the relationships
observed in operant conditioning studies. It would show how
consequences will appear to have a "reinforcing" effect when an
animal is first placed in a situation where it can control a
variable that it wants to control (as in Bill Powers
(970818.p0415 MDT) example); it will show how schedules can
appear to influence behavior when, in fact, they are only
influening the effect behavior can have on a controlled
consequence; it would show how all operant conditioning data can
be explained, in principle, by looking at behavior -- not just
the behavior followed by "reinforcement" -- as the means by which
the organism controls its own perceptions.
The paper would make it very clear that, according to the control
model, what we are seeing in operant behavior is not environmental
control of the organism but, rather, the organism's control (and
attempts to control) its environment (as perceived).
The paper could show some detailed predictions, where possible
(as in Bill's analysis of the rat shock data and the Staddon/
Motherall data). But the paper would have to explain why these
analyses are suspect; there were no appropriate tests done to
determine what variables were actually under control. The paper
could also include a note about how Bruce Abbott's analysis of
pressing rate makes it likely that the animals in many operant
conditioning experiments are trying to control but not being
allowed to succeed.
Finally, the paper would describe how one could start testing
the idea that animals actually are controlling _perceptual_
variables in operant conditioning experiments. Examples of
operant studies that used techniques that had elements of
the test for controlled variable -- such as studies where non-
contingent consequences were added to the presumed controlled
consequence -- could be discussed.
The paper could conclude by showing that control theory provides
a completely new approach to understanding and studying the
behavior of organisms. Control theory sets us on the path to
studying organisms as agents that select the consequence of their
behavior rather that as machines whose behavior is seleced by its
consequences.
And I'd submit it to the heartland of operant research: JEAB.
It would be worth it just to read the reviews.
Best
Rick
···
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken