What is Instrumental Learning?

The word “classical” suggests that classical conditioning was the first type of learning studied by experimental psychologists. But this is not the case. Although scientific research on classical conditioning began soon after the turn of the twentieth century in the laboratory of the famous Russian physiologist, Ivan P. Pavlov, scientific research on a second type of learning — instrumental learning — had already been started during the 1890s in the work of an American graduate student named Edward L. Thorndike (1874-1949). According to Wozniak (1999b), Thorndike's 1898 dissertation for the Ph.D. degree, Animal Intelligence: An Experimental Study of the Associative Processes in Animals:

is widely considered to be one of the most influential publications of the first half century of psychological science. In addition to offering a conception of animal intelligence couched solely in terms of the organism's ability to form new associations, it described ingenious apparatus for the observation of animal learning and demonstrated the use of such apparatus in systematic laboratory research.

For example, Thorndike performed research (summarized in Thorndike, 1911) on the ability of cats to learn to escape from a "puzzle box" (see Figure 1). Cats placed within the box had to learn to push a lever, pull on a wire loop, pull on a string, turn a "button," lift a latch, or push aside a door, in order to escape from the box. In some experiments, the cat had to perform two or three of these actions sequentially before the door would open. In still other conditions, the door opened only after the cats licked or scratched themselves. Cats were rewarded for these behaviors by food, which was placed outside the box. In addition, because cats typically do not like being confined in small enclosures (which you already know if you've ever tried to put a cat into a small animal carrier for a visit to the veterinarian), Thorndike's subjects also were rewarded by escaping from the tight confines of the box.

Figure 1. Thorndike's (1911) Illustration of a "Puzzle Box"
Used in His Research on Instrumental Learning in Cats.

We can use this brief description of Thorndike's method to show that the type of learning demonstrated by the cats is similar to and also different from the type of learning demonstrated by subjects in classical conditioning studies. The most important similarity is that, with both classical conditioning and instrumental learning, subjects learn to associate paired events. In classical conditioning, subjects learn that the presentation of one stimulus is followed by the presentation of a second stimulus that reflexively elicits an involuntary response.

The degree to which a response is voluntary is best represented as involving a continuum:

Involuntary <-------------------------------------------------> Voluntary

• Mostly or Fully Involuntary. It is virtually impossible to voluntarily experience a complete panic attack (one that includes most of the characteristic physiological and psychological events) if asked to experience a panic attack, beyond mimicking behaviorally a panic-stricken person; and it is virtually impossible to not experience a panic attack if put in an extreme situation that is terrifying.
• More Involuntary Than Voluntary. It is virtually impossible not to salivate when asked to inhibit this response after food has been placed on one's tongue, although one may be able to voluntarily salivate, perhaps by thinking of food, when asked to do so.
• More Voluntary Than Involuntary. People with mild tic disorders (such as mild Tourette's Disorder), can easily perform their tics when asked to do so and, with more difficulty, to not perform their tics when asked not to do so, although eventually they will feel compelled to perform their tics.
• Mostly or Fully Voluntary. It is quite easy for most people to move their left thumbs when asked to do so; and to not move their left thumbs when asked not to do so.

The development of an association between the two stimuli is indicated by the development of a reflexive response to the first stimulus. In instrumental learning, subjects learn that, when placed within a particular situation, the performance of a voluntary (nonreflexive) response[∂] to the situation will be followed by a rewarding (or punishing) consequence. The development of an association between the nonreflexive response and the rewarding consequence in that situation is indicated by an increase in the speed of performance of the response when placed in that situation in the future.

The most important difference between the two types of learning is that, in classical conditioning, the CR is elicited reflexively by a preceding stimulus (the CS), whereas in instrumental learning, the learned response (called the instrumental response) is performed because it is followed by a stimulus (the reward). Furthermore, in classical conditioning, the CR develops because of the association formed between the CS and UCS, whereas in instrumental learning, the instrumental response develops because of the association formed between the instrumental response and the reward.

According to Wozniak (1999), Thorndike's approach to understanding instrumental learning foreshadowed the theoretical approach of behaviorism:

Thorndike situated himself theoretically within the long tradition of associationism. Unlike his associationist predecessors, however, he construed association not as linking one ... [mental] element ... with another nor even as linking [mental elements] with movements. Rather, for Thorndike, associations exist between [external] situations in which an organism finds itself and [internal] impulses in the organism to action. In this regard, Thorndike took a step beyond traditional associationism in the direction of the [behavioristic] stimulus-response approach that would eventually come to dominate the field.

Thorndike's "prebehavioristic" approach, according to Wozniak, is evident in two assertions he made:

The first, that psychology could be viewed as the science of behavior continuous with physiology, anticipated arguments soon to be advanced by John B. Watson in his famous behaviorist manifesto [Watson, 1913]. The second, that the study of 'consciousness for the sake of inferring what a man can or will do, is as proper as to study behavior for the sake of inferring what conscious states he can or will have,' anticipated the general approach to consciousness that would become common among early behaviorists.

The problem of how best to measure learning in his animal subjects was a central concern of Thorndike in planning and performing his research. Thorndike's solution to this problem has influenced the methodology of scientific psychology to the present day. Let's look more closely at this problem and Thorndike's approach to solving it.

How Should Psychological Concepts Be Measured?

As defined in Section 1, a concept[∂] is a mental category that refers to a group of meaningfully related events (objects, activities, situations, processes, or the products of these processes). In other words, a concept is a mental representation of a group of events in terms of an association among them — an association developed (passively or actively) by the people who use the concept. For example, the concept of desk refers to a group of objects associated in at least the following ways: (a) they have flat or sloped surfaces resting on a base (often with legs); (b) one generally performs work-related activities at them that include reading, writing, drawing, studying, calculating, and typing; and (c) they typically have drawers to hold objects important for the work-related activities. The concept of desk differs from the concept of table. The latter concept refers to a group of objects associated in at least the following ways: (a) that have flat tops on which objects are placed; (b) one performs a variety of nonspecific activities at them that include (but are not limited to) eating, working, and playing; and (c) they typically have legs. Although it is difficult to express in words the precise distinction between the concepts of desk and table — especially since they both can be used for the same activities and have similar structures — it rarely is difficult to classify a particular item as being either a desk or a table. In general, it is obvious to most people whether an item of furniture is a table or a desk.

It tends to be much more difficult, however, to classify events related to mental phenomena in terms of psychological concepts. This is because of two characteristics shared by most psychological concepts. First, unlike concepts referring to physical objects (such as desks and tables), psychological concepts (such as learning and memory) refer to unobservable mental phenomena. We must make inferences[∂] about psychological concepts based on people's behavior as well as any other relevant events that are directly observable by our physical senses. In other words, we must make inferences about unobservable mental phenomena based on empirical data that consist of observable physical phenomena. For example, we attempt to infer something about the psychological concept of intelligence based on observations of people's scores on tests of intelligence.

The second characteristic of psychological concepts that creates difficulties for classification involves the fact that a particular psychological concept may mean different things to different researchers. For example, the concept of intelligence has meant different things to different researchers since it first was studied scientifically during the late-nineteenth century by Francis Galton (see History of IQ at Test Cafe; also see Human Intelligence web site). And we saw in Section 1 that even the concept of psychology meant something different to the behaviorists, on the one hand, and the structuralists and functionalists who came before them.

So how do scientific psychologists deal with the problem of classifying mental events in terms of psychological concepts? The work of Thorndike provided an ingenious solution to this problem — a solution that has had an enormous influence on psychological research since Thorndike's time. Wozniak (1999) stated that:

Thorndike was ... a methodological innovator, developing a general experimental technique that was to revolutionize the psychological study of animal behavior. As described by a prominent modern researcher, Thorndike's approach to method 'was objective: it minimized the influence of the observer ... quantitative: the course of learning could be measured accurately in terms of the time taken for the appearance of the correct response on each trial ... reproducible: the work of one investigator could be repeated and verified by others ... flexible: the responses required could be varied in kind and complexity ... natural: ... the problems presented ... were not too remote from the animal's ordinary course of life ... (and) convenient: a large enough sample of animals could be studied to provide a representative picture of each of a variety of species.'

In short, Thorndike was concerned that his measures of the concept of learning be objective, reliable, valid, and verifiable; and that his results be generalizable. Let's look at each of Thorndike's concerns one at a time.

Objective Definitions of Concepts
As stated in Section 1-4, an objective definition of a concept involves defining a concept in terms of a specific set of observations that can be verified by any competent observer. By objectively defining concepts, researchers make relevant observations that, in principle, give the same results regardless of who is making the observations. This is because observations based on objectively defined concepts do not depend on the subjective biases, preferences, or attitudes of individuals. Introspection, which was the main method used in experimental psychology when Thorndike began his work, provided "measurements" that were subjective. Furthermore, it couldn't be used to study the intellectual functions of animals.

Measurement Reliability
Measurement reliability is the degree to which an instrument designed to measure a stable characteristic provides repeated measurements of an individual that are similar to each other. A stable characteristic is one that changes very little over a specified period of time, such as the length of a plank of wood. If an instrument designed to measure the lengths of physical objects, such as a tape measure, shows that a plank of wood is 2' 3" one day, 1' 10" the next day, and 3' 1" the following day, then there must be something wrong with that tape measure: it is not measuring lengths accurately. In a similar way, if an instrument designed to measure a stable mental characteristic, such as a test designed to measure the amount of intelligence in individuals, shows that a person is highly intelligent one day, profoundly mentally retarded the next day, and of average intelligence the following day, then there must be something wrong with that intelligence test. In other words, the test is not valid.

Measurement Validity
Measurement validity is the degree to which an instrument measures what it was designed to measure. For example, bathroom scales are designed to measure the concept of weight. If a normal man who is 6' 2" tall and a normal child who is 3' 1" tall both weigh the same according to a scale, then that scale cannot be measuring accurately their weights. Whatever it is measuring, it is not weight. In a similar way, if a profoundly mentally retarded man and a woman who is a professor of astrophysics both receive the same score on an intelligence test, then that test cannot be measuring accurately their levels of intelligence.

Although it typically is a relatively simple matter to determine if a test designed to measure a physical characteristic actually is measuring that characteristic accurately, it can be extraordinarily difficult to determine if a test designed to measure a mental characteristic actually is measuring that characteristic accurately. This is because mental characteristics cannot be observed directly: they have no physical existence (we say that they are mental constructs[∂]). Instead, they must be inferred from individuals' behavior. For example, there is no way to observe directly a person's level of depression. Instead, we must infer how depressed the person is from his or her behavior, and it is best if that behavior is measured objectively. Thus, depression scales (for example, see here) have been developed that provide scores meant to indicate people's levels of depression.

Verifiable Measurements
As stated in Section 3-1, behaviorists argued that, in order to have a scientific psychology, its data (its observations/measurements of psychological concepts) must be verifiable[∂]. In order to have verifiable data, a psychological concept must be objectively defined, and the objective definition must lead to observations/measurements that are reliable and valid.

Generalizable Results
Let's say that you were so busy with work, school, and friends that you had only 45 minutes to study for the first test in a communications course. When the tests were returned, you were very surprised to find that you had received an A because you had studied so little. As the second test approached, you again were able to devote less than an hour to studying; and again was surprised when you received an A. At this point, it seems reasonable to predict that, for future tests in that course, you probably would need to study very little to get an A. Of course, as with any prediction, this one could turn out to be wrong; but it probably would be a good idea to test the prediction when studying for the next test.

Would it be equally reasonable to predict, based on these two experiences, that other courses in communications probably will require a similar amount of study time per test? It might be if you signed up for the same instructor and the course was at the same level. Nevertheless, because it involves a different communications course, this prediction is less likely than the first to be correct.

Finally, would it be reasonable to predict, based on these two experiences, that the chemistry course you signed up for next semester also will require less than one hour of study time per test. This prediction probably is an unjustifiable extrapolation[∂] from your experiences with the communications course: even if the chemistry course is at the same level as the communications course, the prediction involves a different discipline — a discipline that is one of the "hard sciences" — and a different instructor. This would represent an over-extrapolation of your observations: it is unreasonable to infer what is likely to be true for a chemistry course based on observations made in a communications course because the two situations are very different.

But an important goal of psychological research is to extrapolate discoveries made in research situations to natural situations in everyday life. In other words, researchers want their findings to be generalizable[∂]. This was Thorndike's goal. Thus, he tried to make the problem that the cats had to solve similar to something that they might encounter in a natural situation: finding themselves enclosed in a small space from which they had to escape. Nevertheless, the research situation still is fairly artificial because of the need to control for extraneous variables. In general, the more that one controls for the effects of extraneous variables, the less natural the research situation becomes. If the research situation is unlike any natural situation, it may be difficult to apply the results obtained in the research to the real world. This problem will be discussed more fully later in this book.

Study Questions

1. What was the first type of learning studied scientifically?
2. How did E. L. Thorndike study learning in cats?
3. How is classical conditioning similar to instrumental learning?
4. How is classical conditioning different from instrumental learning?
5. What is associated in classical conditioning?
6. What is associated in instrumental learning?
7. In what way(s) did Thorndike foreshadow the behavioristic approach?
8. How would you define "concept" in your own words?
9. What are two examples of physical concepts not mentioned above?
10. What are two examples of psychological concepts not mentioned above?
11. Why is it more difficult to measure instances of psychological concepts than it is to measure instances of physical concepts?
12. How would you define an "objective measurement" in your own words? (In your answer, please give an example of an objective measurement that is not mentioned in the readings.)
13. How would you define a "reliable measurement" in your own words?
14. How would you define a "valid measurement" in your own words?
15. How would you define a verifiable measurement in your own words?
16. What is required for a measurement to be verifiable?
17. What does it mean to have generalizable results?
18. What is the relation between control of the research situation and the generalizability of its results?

What is Operant Conditioning?

Thorndike's research on animal learning before and after the turn of the twentieth century had an enormous influence on the direction taken by experimental psychology after that time. It influenced John Watson's promotion of the behavioristic approach, and the eventual transformation of experimental psychology from a science of the conscious mind into a science of behavior.

Probably the best known experimental psychologist of the twentieth century was B. F. Skinner (1904-1990). Skinner continued Thorndike's work on instrumental learning but renamed it operant conditioning because, Skinner explained, individuals learn new behaviors that "operate on" the environment — behaviors that cause the individuals to experience environmental stimuli. For example, in Thorndike's puzzle-box experiments, the cats' behaviors operated on the environment by allowing them to escape from the small enclosure and to experience the sight, smell, and taste of food.

Figure 2. A Rat Being Operantly Conditioned to Press a Lever in a Skinner Box
(from http://www.has.vcu.edu/psy/psy101/forsyth/learn/12.htm)

Skinner apparently enjoyed building mechanical devices to use in his research (Bjork, 1993), which eventually led him to develop what now are generally referred to as "Skinner Boxes" (see Figure 2). Skinner boxes are fully automatic conditioning devices: a rat or pigeon (the animals that Skinner used in most of his research on operant conditioning) is placed inside the box and learns to press a lever or push a button in order to receive stimuli such as food or water. The lever press or button-push leads to the consequence, however, only when preceded by a light, tone, or other sensory stimulus. This antecedent stimulus (a stimulus that precedes something else) indicates that the behavioral response of pressing the lever or pushing the button is likely to be followed by a consequent stimulus (a stimulus that comes after something else), such as food or water. Presentations of the antecedent stimulus, the recording of responses, and presentations of the consequent stimulus are all mechanized and, therefore, an experimenter need not be present. The general operant-conditioning procedure is illustrated in Figure 3:

Figure 3. The Sequence of Antecedent Stimulus, the Learned Response to
this Stimulus, and the Consequence that Causes the Response to be Learned.

In operant conditioning, the learned response is called the operant response. The pulling of a wire loop in a puzzle box and the pressing of a lever n a Skinner Box are examples of operant responses:they are responses to the antecedent stimulus and eiither increase or decrease in frequency depending upon the nature of the consequent stimulus. A consequent stimulus that strengthens the operant response it follows is called a reinforcement. The food that Thorndike's cats ate after pulling the wire loop and the water that Skinner's rats drank after pressing the lever are examples of reinforcements. A consequent stimulus that weakens the operant response it follows is called a punishment. For example, rats first might be reinforced for pressing the lever in a Skinner Box, which over time should lead to a high frequency of lever pressing. Then, we might begin to shock these rats after pressing the lever, which over time should lead to a low frequency of lever presses that eventually would stop. The electric shock would be a punishment. The antecedent stimulus, which is called the discriminative stimulus, serves as a cue that signals the probable consequence of an operant response (that is, it signals whether the operant response will be reinforced or punished). In a Skinner Box, the discriminative stimulus might be a light that, when turned on, indicates that a lever press is likely to be followed by a reinforcement or punishment. Figure 4 uses these terms to illustrate the general operant-conditioning procedure:

Figure 4. The Sequence of Discriminative Stimulus,
Operant Response, and Reinforcement or Punishment.

Let's look at some examples of operant conditioning to help you learn how to apply these terms to actual situations.

For many people, drinking alcohol often is followed by pleasurable feelings or relief from anxiety. This is an example of operant conditioning because people are learning to perform a behavior because of the consequences of this behavior. What is the discriminative stimulus, the operant response, and the consequence (reinforcement or punishment)? The answers are provided in Figure 5:

Figure 5. The Operant Conditioning of Drinking Alcohol.

When many people see a bottle of alcohol and/or smell it, they will experience an increase in positive feelings or a decrease in negative feelings.

In bungee jumping, a person jumps off a tower (or some other high place) while connected to elastic cords. In this example, what is the discriminative stimulus, the operant response, and the consequence (reinforcement or punishment)? Think about this example for a minute before looking at the answers in Figure 6:

Figure 6. The Operant Conditioning of Bungee Jumping

Here, you see that there are two sets of answers. In both sets, the discriminative stimulus is the sight of the tower and the operant response is jumping off the tower. But some individuals will experience reinforcement of the operant response and others will experience punishment. We know that the operant response was reinforced if the individual shows increased bungee-jumping responses in the future (regardless of what the individual says after the experience); whereas we know that the operant response was punished if the individual shows decreased bungee-jumping responses in the future (again, regardless of what the individual says after the experience). Thus, the behavioristic approach explains individual differences in nonreflexive behaviors in two ways

• individual differences in what is experienced;
• individual differences in which stimuli reinforce and punish operant responses.

A variety of factors determine whether a person finds a stimulus to be reinforcing or punishing, such as physiological factors, past experiences, one's current mood, etc. The only way to determine whether a person is reinforced or punished by a consequent stimulus is to see if the operant response is more likely or less likely to be emitted in the future.

[paraphilia example: masochism]

Study Questions

19. How are instrumental learning and operant conditioning related?
20. Why did B. F. Skinner call the type of learning he studied "operant conditioning"?
21. What is a "Skinner Box" and what is it used for?
22. What is being associated in operant conditioning?
23. How do you know when an association has formed in operant conditioning?
24. How does operant conditioning differ from classical conditioning?
25. How would you define "discriminative stimulus" in your own words?
26. How would you define "operant response" in your own words?
27. How would you define "reinforcement" in your own words?
28. How would you define punishment" in your own words?
29. What are two examples of operant conditioning that you've experienced recently? (Note: In your examples, label the discriminative stimulus, the operant response, and the reinforcement/punishment.)
30. How do individual differences arise in what is learned through operant conditioning?

What is the Difference Between Punishment & Extinction?

While driving, you've probably encountered a situation in which, regardless of your speed, the person behind you wants to go even faster. He probably communicated his intention to you by driving just a couple of feet behind your car. If you wanted to discourage him from behaving in this way, which is endangering both of your lives, what does your knowledge of operant conditioning tell you to do? Before you can answer this question, you first will need to identify the discriminative stimulus and the operant response. What behavior has the person behind you learned — the behavior that is endangering both you and him? The learned behavior (the operant response) is tailgating. What is the discriminative stimulus — the stimulus that tells the person that, if he tailgates, some consequence is likely to occur? The discriminative stimulus is any car that is driving slower than his. Has his operant response been punished or reinforced in the past? Since he is performing the operant response, it is likely that it has been reinforced in the past. What would this reinforcement have been? The operant response of tailgating most likely was followed by the slower car speeding up or letting him pass. The operant conditioning of this person's tailgating is outlined in Figure 7:

Figure 7. Operant Conditioning of Tailgating Response

There are two things you could do to decrease the person's tendency to tailgate. The slower way involves the process of extinction. With operant conditioning, extinction is defined as the decline of the operant response to the discriminative stimulus when the reinforcement is removed. In this example, extinction would involve not speeding up when the man tailgates. In other words, if everyone who is tailgated by this man agrees not to speed up, then his tailgating response eventually would be eliminated. Extinction, however, requires multiple presentations of the discriminative stimulus followed by no reinforcement of the operant response. This isn't going to help you if you want a more rapid decline in the operant response.

Thus, you might consider punishing his tailgating response by slowing down when he begins to tailgate. Again, however, it is likely that multiple pairings of the operant response and punishment would be required before the man stopped tailgating. Furthermore, his immediate response to your slowing down probably would involve frustration and anger (especially if he realized that you were trying to punish his behavior), which could endanger your life even more. Therefore, it might be best to refuse to speed up and to get out of his way as soon as possible. This would partially reinforce his tailgating response, but you would be safe.

What is the Difference Between Positive & Negative Reinforcement?

As already stated, a reinforcement is any stimulus that strengthens the operant response it follows. But what kinds of events are reinforcing? It should be obvious that presenting a stimulus that a person experiences as pleasant after he or she has performed an operant response would reinforce that response. For example, if a child sees a cookie and asks, "may I please have the cookie?," and you praise her for using the word "please" and give her the cookie, she has been reinforced both for asking for the cookie and for using the word "please." Another example: let’s say that one of your instructors asks a question to which you know the answer, and you blurt it out without even thinking. The instructor smiles and says, “Great answer!” It is likely that your question-answering behavior has been strengthened (that is, you are more likely to answer such questions in the future). This type of reinforcement is called positive reinforcement, which is defined as a stimulus that strengthens the operant response when it is presented after the response.

The word "positive" in positive reinforcement refers to the fact that the stimulus is presented after the operant response rather than removed after the response. It does not mean that the stimulus was something that the person found to be pleasant, although most positive reinforcements are stimuli that are pleasant, or at least not unpleasant. Sometimes, however, a stimulus that strengthens the operant response when it is presented after the operant response is experienced as unpleasant. For example, when a pimple is just beginning to develop (before a bump has appeared), a person may find the "proto-pimple" by pressing his finger on the spot, which causes mild pain. The mild pain that follows the finger-pressing response may lead him to press the spot again and again. In this case, the mild pain is a positive reinforcement because it has strengthened the operant response when it occurred after the response. (If the pain had been intense after the sensitive spot was pressed, it is likely that the pain then would have been punishing.) Figure 8 outlines the positive reinforcement of this behavior:

Figure 8. The Positive Reinforcement of Finger-Pressing Behavior.

Let's look at another way that reinforcement of an operant response can occur. Let's say that a very irritating sound (fingers scratching chalkboards) is presented to you through earphones until you press a button that stops the sound for 10 seconds. It is very likely that you would quickly learn to press the button whenever the sound began again. In this case, the operant response (pressing the button) is strengthened by removing a stimulus (the noise) after the operant response has been performed. This type of reinforcement is called negative reinforcement, which is defined as a stimulus that strengthens the operant response when it is taken away after the response. Another example: think of a time when you hadn’t studied much for a test. Perhaps you were very anxious on the day of the test because you expected that you would fail the test. So, what did you do? You may have called the instructor and told her that your car wouldn’t start (even though there was nothing wrong with your car) and that you wouldn’t be able to make it to class to take the test. To your surprise, the instructor told you not to worry — that you could take the test the next day. After hanging up the phone, you felt relieved (anxiety reduction). Your “excuse-making behavior” was just negatively reinforced: your feelings of anxiety were taken away after you had performed the operant response. The operant conditioning of "excuse-making behavior" is presented in Figure 9:

Figure 9. The Negative Reinforcement of Excuse-Making Behavior.

Negative reinforcement typically involves taking away something you don’t like after you have performed an operant response.

How is Operant Conditioning Important for Mental Disorders?

We can understand the development of many mental disorders by using the concepts of positive reinforcement and negative reinforcement. For example, many people in college consume large amounts of alcohol more frequently than is good for them. People drink alcohol because of the reinforcing consequences of this behavior and, therefore, it seems likely that many cases of lifelong drinking problems begin during the college years when students are operantly conditioned to drink excessively and frequently.

Which type of reinforcement — positive or negative — do you think best explains the development of drinking behavior? It probably depends on the person and the situation. Some people drink heavily because alcohol gives them a pleasant feeling that helps them to enjoy themselves more in social situation. In this case, drinking is positively reinforced. On the other hand, some people drink heavily because alcohol removes unpleasant feelings associated with stresses caused by trying to work, go to school, and meet interpersonal obligations, all at the same time. In this case, drinking is negatively reinforced. Since there are plenty of opportunities during the college years for both positive and negative reinforcement of drinking behavior, students may begin the process of operant conditioning that can lead to severe and lifelong problems with alcohol. Drinking may seem innocent enough when it begins, but it soon can explode into a significant problem unless one is aware of the situations that reinforce drinking and finds ways either to avoid those situations or to deal with them in more adaptive ways. The positive and negative reinforcement of alcohol-drinking behavior are illustrated in Figure 10:

Figure 10. The Operant Conditioning of Alcohol-Drinking Behavior.

What is the Two-Factor Theory of Mental Disorders?

Other behavioral problems also can be explained by the concepts of positive and negative reinforcement. For example, positive reinforcement probably is important for the development of fetishistic disorders. In Section 3-1, it was stated that the development of an erotic attraction to an inanimate object can be understood as occurring because of the pairing of the inanimate object with an object that automatically elicits sexual arousal (typically a potential partner), which would involve the classical conditioning of sexual arousal to the inanimate object. After this classically conditioned sexual arousal has been acquired, the operant conditioning of approach behavior — behavior that brings the person into contact with the fetish — would occur. In this case, the person engages in voluntary behaviors that allow him to experience the fetish as much as possible. In the example from Section 3-1, the man with the panty fetish began to collect them by stealing panties from family members, friends, and acquaintances. This “collecting behavior” was voluntary and, thus, would have been learned through operant conditioning. The consequence — the feelings of pleasure when coming into contact with panties — was a positive reinforcer because the feelings occurred only after he had performed the operant response:

            Discriminative Stimulus            Operant Response                        Negative Reinforcer
             sight of panties                            picks them up                                feels pleasure

In this case, the development of fetishism occurred in two steps: the classical conditioning of sexual arousal to an inanimate object followed by the operant conditioning of an approach response to the fetish (contacting and, perhaps, collecting it). This example illustrates the two-factor model of fetish development. The first factor is the classical conditioning of sexual arousal and the second factor is the operant conditioning of approach behavior.

In a similar manner, the development of a phobia also may occur in two steps. First, as discussed earlier, there is the classical conditioning of a fear response to an event, which then is followed by the operant conditioning of escaping from the event or avoiding the event. Thus, the two-factor model (the classical conditioning of an emotional response to some stimulus followed by the operant conditioning of a set of voluntary bodily movements involving this stimulus) also can explain the development of phobias. In the case of phobia development, however, negative reinforcement is involved in the learning of the escape behavior. For example, in the case of Danielle described earlier, a classically conditioned fear of spiders was developed after Danielle saw her relatives become terrified in the presence of spiders. Afterwards, whenever she saw a spider, she would run away from it. In fact, many people with such phobias completely avoid situations in which a spider might be present, such as dark and damp places. Whenever they experience such a situation, their anxiety increases until they escape from the situation. The “escape behavior” is an operant response that is negatively reinforced by anxiety reduction:

            Discriminative Stimulus            Operant Response                        Negative Reinforcer
             sees spider, gets anxious                runs away                                    anxiety is reduced

The two-factor explanation for the development of phobias has been a popular one. There are some problems with it, however. One problem involves the fact that we cannot always identify an experience that can explain the development of a fear response to an object. In fact, some people just seem to be naturally fearful, and they tend to develop fears of all sorts of objects without having any negative experiences involving them. A biological explanation of their phobias may be a better one: perhaps they have a nervous system that reacts strongly to many stimuli, even harmless ones, causing them to feel afraid in a wide range of situations. Another problem is that some people have terrible experiences with various objects and, yet, they still do not develop a phobia. Again, a biological approach may explain why this occurs: perhaps they have a nervous system that reacts mildly to most stimuli, even relatively harmful ones, causing them to feel calm and relaxed in a wide range of situations.

Study Questions

31. How would you define extinction in operant conditioning?
32. In what way is punishment similar to extinction in operant conditioning? (Note: Please think of examples from your life that illustrate the similarity.)
33. In what way does punishment differ from extinction in operant conditioning? (Note: Please think of examples from your life that illustrate the difference.)
34. In what way is positive reinforcement similar to negative reinforcement? (Note: Please think of examples from your life that illustrate the similarity.)
35. In what way does positive reinforcement differ from negative reinforcement? (Note: Please think of examples from your life that illustrate the difference.)
36. How could the development of addictive behaviors be explained with operant conditioning?
37. What is an example from your own experience of the operant conditioning of abnormal behavior?
38. What are the two factors in the two-factor theory of phobic disorders?
39. How might a phobia of enclosed spaces (such as a closet) develop according to the two-factor theory of phobic disorder?