Methodology in the Experimental Analysis of Behavior
When thinking about the implications of a science of behavior,
two questions that immediately come to mind for many students
are:
- Can we really learn anything of importance about human
behavior from studying animals, since animals and humans are so different?
- How can a scientific analysis of behavior disregard internal processes such as thinking and feeling?
These questions are related, since the way in which most of us think about our differences from other species is precisely in the area of thinking and feeling.
Animal Research
While quite a bit of research in conditioning and learning is
conducted using human subjects in applied settings, most of it is
conducted using rats, pigeons and non-human primates (monkeys and
apes) as subjects. To understand why this research would be
meaningful, we need to go back to some of the assumptions of behaviorism. If we are only quantitatively
different than other species, and if our behavior follows the
same lawful relationships as other species, and if complex
behavior follows the same lawful relationships as simple
behavior, then what we learn from studying other animals will
tell us something about human behavior. As one student in class
remarked last semester, this is kind of a "leap of faith." In one sense it is
(in that these are assumptions), but in another sense it is
not--if in fact we can derive laws from the laboratory and use
them to predict human behavior with some degree of confidence,
then the efficacy of using non-human animals in research is an
empirical fact.
There's really only one reason for using animal subjects in
behavioral research-- one can control their past history to a
degree that would be both impossible and unethical with human
beings. The importance of controlling the past history of the
organism under study cannot be overemphasized. If, as we've been
saying, we can understand an individual's current behavior by
looking to past interactions with the environment, then building
a known past history (which, in essence, is what experimentation
in conditioning is all about) gives us a tremendous opportunity
to derive lawful relationships based on this known past history.
Why rats, pigeons and monkeys?
Rats. Rats are used more frequently than any other
species, and for good reason. They are inexpensive, easy to
maintain, and they have some remarkable similarities to us: they
are omnivorous, exploratory, learn fast, and perhaps most
importantly, have adapted to the same environments as us. On the
downside, they interpret the world very differently than us. They
are not particularly visual, and rely much more on olfaction than
we do. Researchers are becoming increasingly aware that they need to pay attention to the sensorial experiences that different species have, their umwelt. Much of the research on rats and visual discrimination from the 30's and 40's has been questioned recently.
Pigeons. Pigeons may seem like an unlikely choice for
researching general laws of behavior--of the three, they are
least like us evolutionarily. They have little tiny bird brains and,
as someone who did his doctoral dissertation using pigeons as subjects, I
can attest to the fact that they are quite astonishingly stupid.
But, they are inexpensive and easy to maintain, and they do one
thing very well--they can see like nobody's business. Their
visual acuity is superior to ours, and they are very visually
dominant organisms. In that regard, they are similar to us, and
in experiments with visual discriminative stimuli (it's primarily
in stimulus control studies that pigeons are used), they learn in
a manner very much like us.
Monkeys. In many respects, monkeys and apes are wonderful
subjects--they are evolutionarily most like us, they interpret
the world very much like we do (a similar balance of sensory
intake), they are primarily omnivorous, learn fast, and can learn
very complex repertoires of behavior. Unfortunately, they are very
expensive and difficult to maintain. Also, due to their
similarity to us, there is more of an ethical concern about using
them as subjects of experimentation.
A Typical Experimental Setup
The diagram to the left
shows a typical operant conditioning chamber, or Skinner
Box. In this diagram, a pigeon is being conditioned to peck
a key--a 1" round plastic disk located about 9" above the floor
of the conditioning chamber. When the pigeon's beak hits the
key with enough force, it closes a microswitch placed behind the
key, and electronic equipment records the response. To the left
of the pigeon is a speaker for auditory stimuli. The pigeon is
reinforced by a few seconds access to grain through the opening
located right below the key. For visual discrimination experiments, the key can
be transilluminated from behind with colored lights and/or visual
patterns.
Skinner Boxes for rats are similar, but the response apparatus is
usually a bar that can be pressed, and reinforcement is usually a
pellet of food delivered into a food trough, or a few seconds
access to water.
Shaping. When the pigeon is first
placed in the Skinner Box, it needs to be conditioned to peck the
key. The procedure used is called successive approximations, or
shaping. (As an aside, in the late 1960's, an automatic way of shaping
pigeons was developed, with considerable theoretical
implications for behavior theory. We will discuss this in a future unit.)
The first task of the experimenter is to get the pigeon
used to getting grain when a light goes on in the opening below
the key and the apparatus allowing access to the grain operates
(it makes a fairly loud noise). Pigeons learn this pretty
quickly, and learn to rapidly go to the opening when the
light/noise occurs. The experimenter's task is to
reinforce the pigeon for coming closer and closer to the key,
until it eventually pecks it.
Once the pigeon is pecking the key reliably, the experimenter is
free to impose whatever conditions she wishes, and record the resulting
changes in behavior.
What operant conditioning experimentation is all about is
the experimenter controls the conditions that confront the organism, records the resulting
behavior, and compares this behavior to the behavior of organisms under other conditions, or this same organism before these conditioning were in effect.
The experimenter is looking for similarities and differences that might lead to more general principles summarizing
the lawful relationship between the environment and behavior.
At this level, the goals of the experimental analysis of behavior are no different than those of any other scientific
pursuit--control conditions and look for orderly relationships.
What any conditioning experiment is about, fundamentally, is building in a known past history
of interactions with the environment. If the basic goals
of conditioning theory are viable, then knowing this
past history (the reason for building it in) should allow the experimenter to predict what the pigeon will do
in the future. At this point, the tenets of conditioning theory become testable.
| Question--how confident do you think an
operant conditioner might be in predicting the behavior of an organism which has been studied for a while? |
Some historical context for a science of behavior.
[This section under construction.]
The methodology used in a science of behavior derives directly from its historical context. Historically, a scientific analysis of behavior has been called Behaviorism. While most scientists studying behavior find this term to be overly simplistic, we can use it as a shorthand term for the present. Behaviorism is firmly within the context of the development of science, in that most sciences have progressed when they stayed within the context of observable events, and tried to make testable laws based solely on the relationships between observable events. Obviously, this has been more difficult when the subject matter is us--our philosophical beliefs about who we are and where we came from are oftentimes at odds with this colder, more detached view of human behavior.
One tradition from British Empiricism that influenced the early behaviorists (e.g.-John B. Watson) is the concept of a tabula rasa--the idea that humans (as well as other organisms) are born with minds that are blank slates (a strict translation of the Latin tabula rasa). The idea is that our experiences with the environment then write those experiences on our blank slate, and we are the sum total of these experiences. This tradition has been heuristically valuable for the development of a scientific analysis of behavior, in that it has led scientists to look to past experience for explanations of current behavior. However, as we talk about in Misbehavior, this view is overly simplistic -- our inheritance has a major role on what we do when, and no one takes it to be literally true that we are born with blank slates.
Disregarding internal processes such as thinking and feeling.
The fact is, a scientific analysis does not ignore thinking and feeling. It does, however, attempt to look at thinking and feeling within the confines of the assumptions, procedures, and goals of all science. Let's briefly look at how a scientific analysis deals with these two processes, starting with the easier one, feeling (emotion).
Emotion. As we saw in the last unit, Pavlovian conditioning is primarily of interest because it summarizes our motivation and emotional connections with the environment. A scientific analysis would focus more on the objective components of motivation (e.g., "What is the probability of the individual behaving in a particular way in this situation?") and emotion (e.g., "What outward signs of fear or attraction does the individual show?"), rather than using more subjective labels such as desire, love, hate, or envy.
Thinking. Thinking presents a major challenge to any scientific understanding of human behavior. First, it is a private event, making it largely inaccessible to all but the thinker. Since science is about accounting for observable events and relationships, thinking's role is going to be difficult to ascribe. Second, it is viewed by most people as the distinguishing characteristic of being human. The language of thinking and humanness are inextricably intertwined in western thought. Because it is so closely tied to our identities, there is a great deal of emotion attached to discussions of thinking, making a dispassionate analysis difficult. Anyone who tries to make a purely objective analysis runs the risk of being viewed as trying to discount what makes us special. B. F. Skinner, even though he wrote extensively about thinking and tried to make sense of it, was frequently and erroneously accused of denying that humans thought. Even intellectual giants such as former Vice-President of the United States, Spiro Agnew, accused Skinner, in his attempts at a scientific analysis of thinking, as "trying to take away our freedom and dignity," and Skinner is still viewed with hostility by many who misunderstand his position. (As an interesting sidelight, Agnew's charge came shortly before he resigned from office for accepting bribes, another emotion-laden term that Skinner was often charged with promoting.)
The reason for much understanding of a scientific analysis is rooted in the reason why it prefers "scientific" explanations. When people are pressed to
explain why they think and feel as they do, they usually go back to
things that happened to them in their past. These past interactions between the environment and the individual can
be studied empirically (e.g., "I'm afraid of dogs because when I was young, I cut across my neighbor's yard and her dog bit me."), which is what a scientific analysis of behavior is all about. Again, this focus on observable relationships doesn't mean that thinking isn't important or not going on, it's
just that it necessarily lead back to public events if it is to be studies scientifically.
Before going on, be sure to Make Up Your Own Question!
Email Tom
Last modified on August 29, 1998.