[From Bill Powers (970930.0646 MDT)]
One last remark about the phenomenon of reinforcement.
There are two ways to view what a rat in an operant conditioning cage does.
1. The rat behaves randomly, until it accidentally produces the action that
operates the apparatus to produce a food pellet. The food pellet has some
effect on the rat that increases the likelihood that it will again perform
the act that produced the food pellet. Continued reinforcement increases
the likelihood of the critical behavior to nearly unity, and that behavior
is then maintained by continued delivery of the food pellets.
2. The rat desires to obtain food. When it lacks a source of food, it
searches for one, varying both its actions and the locations in which it
acts until an action produces some food. The rat then focusses on a
narrower range of actions around the one that worked, varying its actions
until it finds one that will reliably produce the food. It then uses this
behavior to keep itself supplied with food to the extent possible.
These descriptions obviously emerge from completely different concepts of
what makes the rat behave.
In the first one, the food pellet is imagined to be the cause of a change
in behavioral organization, and the cause of maintaining behavior in a
certain form. By manipulating the environmental conditions and observing
the changes in behavior, the experimenter can control the rat's behavior by
changing its environment, and thus demonstrate the effect of the
environment on the rat.
In the second interpretation, the rat is seen as an active agent that can
find the means to satisfy its own wants, and then to use that means to keep
those wants satisfied. The experimenter varies the environmental
conditions, but the rat has the ability to vary its behavior to nullify the
effects of those changes on its ability to provide for itself what it wants
(to the extent that is possible for it). This demonstrates the _lack_ of
effect of the environment on the rat. The rat continues to provide for
itself what it wants, despite changes in the environment.
In the case at hand, the common element is the food pellet. Under the first
view, the food pellet is interpreted as having some effect on the rat that
shapes its behavior toward the form that produces the food pellet. Under
the second view, the food pellet is viewed as being produced by the rat in
order to satisfy its need for food, and the variations in behavior are seen
as caused by the rat and being aimed at producing that food. In short, the
first view interprets the change in behavior as a passive consequence of
the reinforcing effect of the food, while the second views the change in
behavior as intentional, and as having the goal of producing enough food.
With this difference in interpretations in mind, we can look at a simple
situation.
In a standard place-setting at a dinner table, the fork is placed on the
left and the knife and spoon on the right. When a knife is required, the
person seated at the table will reach with one hand toward the position of
the knife and pick it up. Why does that happen?
Under the first interpretation, this happens because reaching to the right
side is reinforced by success in picking up the knife, while reaching
toward any other place will soon be extinguished because no knife will be
picked up. The position of the knife reinforces only those actions that
pick it up.
Under the second interpretation, the hand reaches toward the position of
the knife because the intention is to pick up the knife, and the person
produces whatever action will have this result.
Now what will happen when the place-setting is reversed? Under either
interpretation, the person will reach toward the customary position of the
knife. However, under the first interpretation, this behavior must first be
extinguished by repeated failure to find the knife there, and a random
sequence of behaviors must then be emitted until one behavior results in
the hand reaching the knife and picking it up. Under the second
interpretation, failure to find the knife in the standard position will
result in a search for the knife, and as soon as it is located it will be
picked up.
What is the difference in these interpretations? It can be seen in the
answer to the question, "Why does the reaching change from the right side
to the left side?" Under the first interpretation, it changes because (a)
reaching to the right side ceases to be reinforcing, so that behavior
extinguishes, and reaching to the left side becomes reinforcing, so that
behavior increases in probability. Under the second interpretation, the
explanation is, basically, "Because the person intends to pick up the
knife, wherever it is."
So the difference is in the part of the observable phenomenon that is
considered the cause, and the part that is considered the effect. In the
first view, the behavior is the effect, while the consequence is its cause.
The picking up of the knife on the left or right is what causes the
reaching to be toward the left or toward the right. In the second view, the
behavior is the cause and the consequence is its effect: The behavior of
reaching causes the knife to be picked up, whether the reaching must be to
the left or to the right.
The underlying mechanisms that are imagined are also different. In the
first view, the underlying mechanism is a special effect of the consequence
on the production of specific responses. In the second view, the underlying
mechanism is the ability to compare a perceived consequence with an
intended consequence, and to use the difference as the basis for varying
behavior in the way that will eliminate the difference.
Conclusion: the EAB and PCT interpretations of the observed phenomena are
fundamentally different.
Best,
Bill P.