[From Bruce Abbott (970929.1915 EST)]
Bill Powers (970929.1158 MDT) --
Bruce Abbott (970929.1050 EST)
A lot of silliness could be
avoided when comparing PCT to EAB if this difference were kept firmly in
mind. What happens when one insists on using the PCT definition of
"control" to interpret the meaning of EAB sentences like "the environment
controls behavior," or "this is an example of stimulus control"?
These sentences would sound quite different to my ears if they were changed
to "the environment influences behavior" and "this is an example of
stimulus influence." Such sentences would acknowledge, in my terms, the
fact that behavior can change independentsly of the environment or stimuli.
But do you think that other EABers would be happy to agree that "influence"
or "affect" are exactly equivalent to their usage of "control?" When an
EABer demonstrates shaping the behavior of a pigeon so it will peck on a
key, doesn't the EABer think of this as controlling the behavior of the
pigeon?
That would depend on what you were talking about. Does behavior vary as a
function of some variable? Then to that degree the behavior is "controlled"
by it. If the relationship is weak, then "control" by this variable would
be viewed as poor. In "stimulus control," some aspect of behavior is
"brought under control" of the stimulus. This means that when the stimulus
is presented, that aspect of behavior will change to some predictable value,
pattern, etc., that is characteristic of behavior under that stimulus. For
example, a pigeon may be trained to peck at a key on a fixed ratio schedule
of reinforcement when the key is illuminated red, and not otherwise. When
the key color is changed to red, the pigeon immediately turns to the key and
begins to peck at it, and the pattern of pecking will be that typical of
pecking under fixed ratio schedules. Such a stimulus is said to "set the
occasion" for the particular value, pattern, etc.; it is not viewed as "the
cause" of it. The pattern may not occur if other necessary conditions have
not been established (e.g., deprivation).
And doesn't this usage of control imply that the pigeon's behavior
is brought to some particular state by the experimenter's _varying_ the
influences applied to the pigeon?
Maybe yes, maybe no. I don't know what you have in mind.
Suppose I dismissed this demonstration by
saying, "Oh, you didn't make the pigeon do anything in particular -- you
influenced its behavior, but you didn't determine its final state. You
didn't know what the pigeon was going to end up doing until it did it, and
after you saw what it did, you claimed that you made it do that."
I think that this conclusion would be unwarranted, because in the scenario
you present, the _experimenter_ is clearly using her knowledge and skill to
bring about a certain outcome -- like pecking at the key, or turning circles
in the chamber when a tone is sounding. In other words, the experimenter is
attempting to control (PCT meaning) the bird's observable behavior by making
appropriate use of known relationships between the manipulable conditions
and the ensuing behavior.
I suspect that the EABer would object strenuously to this view of his
shaping demonstration, and that in doing so he would show that his concept
of control is exactly the PCT concept, and not merely "influencing"
behavior. The experimenter sets a reference condition for the behavior of
the pigeon, and then varies the giving of rewards however necessary
("improvising," as Skinner put it) to make the observed behavior match the
intended behavior.
A person may use certain techniques to bring about an intended pattern of
behavior on the part of the pigeon. This might be done by a pet owner or
commercial animal trainer, or by an experimenter as a preliminary to an
investigation (e.g., getting the pigeon to peck the key as a preliminary to
investigating how behavior varies with schedule parameters). But in the
latter case, whatever that relationship turns out to be, it was not
preconceived by the experimenter, and is not a relationship she is
attempting to produce.
I do not believe that EABers would want to substitute "influence" and
similar words for control, and that their reason for not wanting to do so
is not just that they are used to the word: they really mean what PCT means
by control.
When they are talking about the _experimenter_ controlling the pigeon's
behavior, they do indeed: they are using known influences (EAB control) to
bring about PCT-type control over their perceptions of the pigeon's
behavior. However, they probably wouldn't describe it this way; they would
probably just say they are getting the pigeon to, e.g., peck the key. But
when they are talking about the _environmental_ control of behavior, they
most definitely do not mean that the environment varies its actions so as to
oppose the effects of disturbances and keep the behavior near some reference
value. The environment has no intentions. They mean that certain behaviors
are very likely to be observed under certain environmental conditions.
Those conditions are thus said to "control" the behavior.
Regards,
Bruce