|
There
were a variety of flavors of S - R psychology. The important
terms of S - R psychology were:
Now that seems pretty simple
doesn't it? And it is, although it seemed so simple that many
characterized it as simplistic!
The
basic concern shared by all flavors of behaviorism was that psychology
concern itself only with observables. Thus, the label
behaviorism since on this view behavior is that component of
an organism's activity that is "observable." The
first and second terms; the S for stimulus and the R for response
were the observable anchors to this S - R theory.
Now, on the "radical behaviorism"
view, the third term, the '-' is an empty term. And this is how
it should be since we can not observe what goes on in the organism
from the point in time that the organism is exposed to the S
and emits an R. The organism is simply viewed as a "black-box"
. This is the attitude most of us adopt most of the time to the
technology around us. We don't care how the calculator works
as long as it returns the correct output to the values
that we input to it. However, note that in the example
of the calculator,we can precisely describe the input to the
calculator and the output provided by the calculator. We know
that the calculator will respond to the numbers entered and will
not be sensitive to which finger we used to enter the number,
or whether we used the eraser tip of a pencil, or whether they
were entered quickly or slowly, etc. And, we know exactly where
and when to look at the calculator's screen to find the output.
But are things really this simple
with living organisms? Can we apply the formula:
|
|
|
Here
we have used standard mathematical notation which reads that
"R is some function of S." This seems fairly
innocuous. But note that as soon as we give the statement mathematical
pretensions, we begin to suspect that what started out as a slogan
will require some scrutiny before we can claim that it is anything
more than a slogan.
Functions
take some mathematical object (like a number) and return a similar
mathematical object as the value of the function. But we aren't
entirely sure what a stimulus or a response is much less how
to assign numbers or some other suitable mathematical object
to them. Now you begin to appreciate Skinner's solution....just
forget about all of these niceties.......a response is what the
experimenter decides to measure and a stimulus is what
the experimenter decides to vary in some systematic fashion.
The justification is in the results. If the "response"
varies in some systematic way with the "stimulus",
then the experimenter made some wise choices and 'truth'
has been found. Forget about mathematically describing anything;
simply publish the curve that shows the relation between "stimulus"
and "response".
|
| Thus,
this type of S - R Theory was a theory about why one didn't need
or want a theory. Other S - R psychologists were less comfortable
with this approach. They were looking for a way in which to make
theoretical statements while ensuring that those theoretical
statements were really about observables. There "slogans"
went something like this: |
|
| The
kind of internal physical events they had in mind were things
like the motivational level of the organism (if it was dozing
off it probably was not behaving in quite the same way as when
it was wired!) and changes in the strength of associations, and
the like. But recall, like any upstanding behaviorist, these
folks wanted to only talk about observables. (Those of
you with some philosophical background can relate this position
to that of the positivist movement that was quite prominent in
the U.S. at this time.) But what if internal events are solely
a function of external events? Then we can talk about internal
events by talking about external events! So now on to operationalism
- which is kind of a fancy way to justify Skinner's position
that a response is what I say it is and the same goes for a stimulus
(Skinner was a fan of Gertrude Stein ..."...a rose is a
rose is a rose...." so it probably all makes sense in some
sense of sense.) Operationalism says that it is the scientist
responsibility to 'operationally define' any term that the scientist
is going to use. A term is operationally defined when the rules
for making the observations that the term ultimately refers to
are specified. Thus, a response might be defined as any depression
of a bar that is sufficient to cause the recorder to make a mark
on a strip of paper. Notice that the response is now something
that is recorded by some instrument. And, we can define 'associative
strength' (or what these types called 'habit strength') as a
term that increases as some function of the number of reinforced
pairings of a stimulus with a response. And, voilà! ...
an internal event; i.e., learning is now under the "control"
of the experimenter because it is the experimenter that controls
the number of times a stimulus and response pairing is reinforced.
Thus 'habit strength,' an internal event, becomes operationally
defined as a value that increases as some function of the number
of trials where the stimulus and response pair is reinforced. |
|
| Similarly,
the motivation or 'drive' as it was referred to could be defined
to increase as some function of the per cent ad libitum
body weight of the animal. Some of the main terms used in this
S - R theory (often referred to as the Hull-Spence Theory after
its two main proponents) are shown below. |
|
| H
or habit strength and D or drive have already been
mentioned. One of the major preoccupations of the hay day of
this theory was to determine how the values of these terms combined
to yield the observed behavior. Analysis of Variance, a relatively
new statistical technique at the time, was used to attempt to
answer questions concerning how the values of these "theoretical"
terms combined to determine the observed behavior. We will take
up this use of the Analysis of Variance in more detail in the
section on mental chronometry. As a result of a great many experiments,
it was argued that the motivational level, D, and the habit strength
of a particular S - R connection combined multiplicatively.
But notice that a term referred to as reaction potential,
E, is introduced in the equation below. |
|
| Finally,
the equation that mentions the observed response, R, is
shown below. The response is equal to its reaction potential,
E, minus a term referred to as oscillatory inhibition.
This term is essentially a random variable and is assumed to
be a normal random variable. Recall that it is assumed that there
are a set of responses that are connected to a particular stimulus.
Each of these responses has some associative strength to the
stimulus. Thus, they can presumably be (partially) ordered based
on this strength. This partial order is what they refer to as
the response hierarchy. Note that for a fixed value of
D the response with the highest value of habit strength will
always be associated with the highest level of reaction potential,
E. Thus, the organism would always do the same thing. But not
if we add a random variable as is done in the equation below.
Now, the dominant response(s) will still have the greatest probability
of being observed, but they will not always be observed because
sometimes the random variable will assume a value that causes
the dominant response to, at the moment, have a lesser value
than some other response. |
|
| There
is much more that could be said of this version of S - R Theory.
It eventually gave way to much more sophisticated mathematical
models of learning such as Stimulus Sampling Theory as developed
by Estes and others. And the basic idea that feedback from the
environment can be used to learn an "arbitrary" function
has been developed in exciting and extraordinary ways within
an area commonly referred to as "Connectionism." And,
there are still quite lively debates as to whether any of this
has anything to do with human reasoning. However, one thing is
quite certain; we now possess a rather deep mathematical understanding
of the requirements for and implications of this style of learning. |
