Operant psychology and PCT

[From Bill Powers (950621.0730 MDT)]

Dennis Delprato (950620) --

I do appreciate your thoughtful post on Skinnerian psychology and its

     It seems that someone ought to put together all the challenges to
     Skinner's ideas that arose from his own camp.

This, as well as a sympathetic but firm discussion of the whole field of
behavior analysis and Skinnerian theory, would make a book of
outstanding importance. There are few people I know about -- you and
Bruce Abbott are the only two who come to mind -- who have had a deep
involvement in the Skinnerian movement and also have the advantage of an
outside viewpoint other than that of conventional psychology from which
to speak. You could correct many misconceptions about radical
behaviorism, and at the same time work toward leading that field into a
more advanced conception of modeling behavior. Isn't it about time to do


     They use 'history' in a temporal sense--what happened previously in
     time. If you are saying that what happens to an organism
     historically is a function of the organism's behavior, operant
     theorists would agree.

This is partially what I mean. The rest of what I mean has to do with
the _interpretation_ of the role of historical events. There is far more
involved in the Skinnerian interpretation than a mere listing of
spatiotemporal processes over a period of time. There is a deep-seated
_causal interpretation_ that has the same roots as do other ideas in

The whole thrust of psychology as a new science was to interpret the
behavior of organisms as the behavior of a mechanism like any other
physical mechanism. I'm sure this orientation was consciously adopted
very early in the history of psychology, as a way of catching up with
physics, the Big Brother of science. It was adopted also as a
counterthrust against what scientists have always regarded as the enemy
of progress: imprecision, superstition, mysticism, and magic. During all
the formative years of scientific psychology, there was no known
alternative to the physical-science approach that did not in fact entail
some form of mysticism and magic.

All this began to change in the middle of the 20th Century when the
automation revolution and the computer revolution began. Old concepts of
mechanism began to crumble as amazing discoveries were made about the
capacities of artificial computing devices and control mechanisms. The
firm boundary between "mental" and "physical" phenomena began to waver,
until now nobody who is knowledgeable about the new machines takes the
distinction seriously.

This is one reason, Dennis, that I see your firm insistence on
"spatiotemporal" explanations as somewhat old-fashioned. It has the
flavor of the old conception of mechanism, in which mental phenomena are
considered ghosts in the machine rather than realities of the physical
universe. What we have discovered about the capabilities of artificial
computers and control systems has taught us that organization and
function have a reality just as certain as the reality of matter and
energy. Not only that, we have found that organization and function,
while residing in physical matter, can transcend physical matter in the
sense that the same organizations and functions can be realized in an
endless variety of physical instantiations. The particular physical
particles and physical laws that are used to implement computing or
control functions are unimportant; every year more and different
physical means are found for storing bits and performing logical
functions. What matters is how the various physical elements are
_related_ to each other, not what the physical elements are. And these
relationships are not spatiotemporal in nature. They are functional and
organizational -- what used to be called "mental."

Arthur C. Clark was the one who said "The products of any sufficiently
advanced civilization would seem to us to work by magic." This is how
the models of any sufficiently advanced theory of human behavior would
appear to a conventional scientific psychologist, even a "radical" one.
A great part of Skinner's emphasis was on showing the superficiality of
mental explanations of behavior, explanations that invoked unobservable
entities and intervening variables. For every mental explanation, he
tried to supply an equivalent explanation that involved only externally
observable relationships among physical variables. Among the
explanations he sought to displace were those that relied on attributing
purpose and intention to the organism. He saw these as simply more
intervening variables in the same class as traits, dispositions,
expectancies, aspirations, fears, and thoughts. As far as he could
possibly have known, all these concepts were just fictions, unprovable
and untestable by their very nature. Skinner could not distinguish
between a bad model of brain function and a good model, because he had
never seen a good model.

Skinner's great mistake, which he was far from the only one or the first
one to commit, was to assume that everything necessary to give a full
and accurate account of behavior could be found in the external
circumstances surrounding an organism. He assumed that if he just
recorded very carefully what happened, he would be on firm ground
because anyone, at any time in the future, could record the same events
and verify his analysis. This is the faith of the empiricist; that a
careful, honest, and well-trained observer cannot make any substantial
mistakes. It is also the faith of the naive realist, who would be
extremely surprised and skeptical about any claim that others, equally
well-trained and equally honest, would not necessarily see exactly what
he sees.

Remember my example of the lawn ornament on which a little wooden man
busily turns a crank that turns the vanes of a windmill to cause the
wind to blow. In the absence of any reason to doubt the interpretation,
one could trace cause and effect point by point through the linkages,
and show that this interpretation makes perfect sense. One could even
show that when circumstances changed, the same interpretation still
worked: when the little man stops cranking, the wind stops, too.

We don't make this mistake with the lawn ornament because of a vast
network of understanding and observational detail that surrounds our
observations of it. But when that vast network is missing, as it has
been for the behavior of organisms, we can easily think we are simply
reporting a causal relationship when we have it exactly backward. That
is what happened to Skinner with the concept of reinforcement.

He was almost doomed to make this mistake because of a strong prior
assumption: that all behavior was caused, ultimately, by the
environment. A reinforcer and a change in behavior are two observable
events. If the environment causes behavior, then clearly the reinforcer
must cause the change in the behavior. The basic assumption permits no
other interpretation even if one is inclined to make it. So what Skinner
reported as a a trained observer was exactly what he was trained to

The most frustrating (to me) aspect of Skinner's early work was not this
interpretation, which was really standard among scientific
psychologists, but the fact that he then proceeded to set up an
apparatus in which a reinforcer would never appear unless the behavior
occurred first. The sequence of events always began with a behavior,
followed by a reinforcement. This causal relationship was right there in
the environment, visible even to an untrained observer.

Once an experiment got under way, however, reinforcements and behaviors
occurred in a continuous stream, so as with the lawn ornament one could
choose either way of perceiving causes and effects. As the animal worked
toward mastery of the task, one could see the gradual increase in
reinforcement rate as causing a gradual increase in behavior rate, or
the gradual increase in behavior rate as causing a gradual increase in
reinforcement rate. Once the rates had leveled out, one could see either
the reinforcements maintaining the behavior, or the behavior maintaining
the reinforcements. In the absence of any larger network of
understanding of behavior, either interpretation would fit the

Skinner, unfortunately, understood that the causes of behavior always
lie in the environment. He therefore had no option but to see the
reinforcement as the cause and the behavior as the effect, both during
acquisition of the behavior and during maintenance of the behavior.

Along with the basic method came the so-called schedule of
reinforcements, implemented by an automatic apparatus or a human being
working strictly to a rule. The role of this schedule could also be
interpreted in two ways. It could be said that the behavior, acting as
input to the schedule, caused an output in the form of periodic
reinforcements. Or it could be said that the contingency embodied in the
schedule, in combination with the effects of the reinforcements, caused
the pattern of behavior. Once again, Skinner's basic assumption left him
no choice as to how to report his observations. Clearly the contingency
and the reinforcement combined to control the behavior. The trained
observer once again observed what he had trained himself to see.

All subsequent interpretations of experimental results were required to
remain consistent with the initial Gestalt resulting from the choice
between the two possible directions of causation. If the initial choice
was wrong, then necessarily all subsidiary and subsequent
interpretations had to be wrong, too. The numerical records of
reinforcement and behavior rates, of contingency ratios and combinations
of ratios, would remain quite accurate, but in every instance their
interpretation would depend critically on the initial interpretation of
the relationships among behaviors, contingencies, and reinforcements.

The integrity of radical behaviorism, therefore, rests on the
correctness of Skinner's initial interpretation. Every conclusion about
behavior drawn by every subsequent radical behaviorist as a result of
every subsequent experiment depends for its validity on whether Skinner
saw the lawn ornament in the right way. If Skinner got it backward, then
so has everyone who followed him, from the very beginning to this very

I think that in his or her heart of hearts, every radical behaviorist
who has come up against PCT knows that this is the basic threat of PCT.
For PCT does not portray reinforcers as causes and behaviors as effects.
Quite the contrary: PCT shows behavior as the means by which the
organism acts on its environment to produce consequences that are to its
liking. If the reinforcer is not there, the organism finds a way to
bring it into being. If there is not enough reinforcer, the organism
changes its behavior to make more of it. If there is too much
reinforcer, the organism acts to reduce it. It is the organism that
determines how much reinforcer there is to be, and that learns to create
whatever action is required to produce that much reinforcer. In fact,
the name "reinforcer" totally loses its meaning as something that can
make the organism do something. The reinforcer is an effect, not a

So if Perceptual Control Theory is right, Radical Behaviorism is wrong.
It is as simple as that. The fundamental assumptions about the direction
of causality are diametrically opposite; there is no reconciliation
possible, no meeting in the middle, no compromise that will satisfy both
theoretical structures. As with the lawn ornament, one must choose an
initial interpretation; from that initial interpretation _all else

So, Dennis, the historical record of which you speak does not speak for
itself. It must be interpreted, and how it is interpreted makes all the
difference. I think I have strong evidence, incontrovertible evidence,
that Skinner chose the wrong interpretation and sent thousands of his
followers down the wrong path. To correct this error it is necessary to
go back to the initial bifurcation and see what happens if we start with
a different interpretation of the same events.
Best to all,

Bill P.