Brief Lecture Notes for Unit 4 (revised August 22, 2005)
Introduction -- The behaviorist challenge
The history of psychology over the past 125 years or so can be viewed in terms of a pendulum swing between -- or, to use a different analogy, a turf war between -- two competing general views of the field. These two views are usually called "cognitivist" versus "behaviorist", but think of those terms in the broadest possible sense. "Cognitive" psychology in this broad sense includes not only the cognitive school of thought that we'll be considering in unit 6, but all approaches that make reference to mentalistic constructs that that, of necessity, rely on self-report. Personality theory in general is dominated by cognitive approaches in this broad sense. For us, then, behaviorism represents the "loyal opposition", a completely different approach to the field that begins by denying (or at least downplaying) the existence of personality, or the usefulness of the mentalistic construct of personality, and diverging from there.
A behaviorist is, in general, an individual who eschews mentalistic constructs and attempts to construct a psychology oriented purely around the analysis of outwardly observable behavior. This means a general rejection of the merits or utility of self-report information. It also means the general belief that all behavior is learned or acquired through experience. It means reductionism and determinism, at least in the traditional, classical forms of behaviorism. Let's examine each of these propositions in turn to get a sense of their implications.
1. Rejection of mentalistic constructs
Behaviorists can be divided into two subgroups, "radical" and -- there's no consensual term for the second group, though I've always thought "sane" wouldn't be too bad a choice. Radical behaviorists deny the very existence of the mind (leaving open the thorny question of how they are denying it, if they have no mind with which to do so). True radical behaviorists are few, and are probably philosophical agents provocateurs rather than literal advocates of a serious point of view. Nonradical behaviorists comprise the bulk of the behaviorist camp; they concede that human beings have minds, but deny that this fact has any significance or scientific usefulness. In other words, to them, the mind is an epiphenomenon: it reflects the reality of behavior, but it does not influence behavior in any meaningful way, and in any event can't be studied scientifically. In practice, their objections to the study of the mind fall into two categories: methodological and metaphysical. The two are not necessarily linked, but in practice, they often are:
a. Methodologically, behaviorists argue that self-report data is so unreliable, so subjective, and so incapable of independent verification that it is scientifically useless. Better to scrap it entirely, they argue, and build the science of psychology around what can be observed, verified, and quantified: outwardly observable behavior. If this requires the assumption (which is inherently unprovable, of course) that the mind "does not matter", so be it. To the methodological behaviorist (who may or may not be metaphysically neutral), this is a price worth paying, even though it narrows the range of topics that can be studied by the psychologist, because scientific objectivity is gained as a result. (The underlying debate -- is it better to be precise about things that may not really matter, or vague about things that definitely do matter? -- is, by all accounts, a difficult one.)
b. Metaphysically, the behaviorist may think (falsely in my view) that science requires the belief that all events in the universe are linked through chains of impersonal, observable causes and effects. This point of view has enjoyed much success within the natural sciences (physics, chemistry, biology, and the like), though it is not in fact actually required by them (the proportion of scientists who adopt this point of view has not changed much -- about 50% -- in the past eighty years according to one recent longitudinal study); perhaps envying the success and prestige of the natural scientists, some psychologists are motivated to become like them, while others may honestly share the same set of philosophical assumptions for other reasons. The human mind, of course, is neither impersonal nor observable, so to conclude that human beings shape their own behavior by means of their choices, motives, and the like is anathema to the metaphysical behaviorist, whose view of the universe requires that all such causes be excluded. (But note the circularity of the argument: the honest behaviorist must argue that she is not choosing a point of view, since her point of view is that choice is impossible -- even for behaviorists. Taken to extremes, this means concluding that logic and rationality are also illusions; if they are, isn't science also an illusion?)
2. Belief in the empirical acquisition of behavior
Behaviorists by definition are empiricists (as opposed to nativists who believe that there are significant numbers of innate, inborn, or genetically pre-programmed patterns of motivation or action). Here the behaviorist is on firmer ground (it is unlikely that human beings have any true instincts in the biologist's sense of the term), yet taken to extremes, this point of view ignores the influence of genetic factors on personality and behavior, which in many instances is no doubt quite profound. Recent twin studies highlight the astonishing degree of concordance and heritability found in many human traits of personality and behavior. Still, the role of learning cannot be denied, and one does not have to be a behaviorist as such to say so.
3. Reductionism and determinism
By definition, a consistent behaviorist must be a reductionist (the apparent complexity of human personality can be explained fully in terms of the accumulation of specific behaviors that are learned by means of simple, mindless processes) and a determinist (such concepts as freedom and choice are illusory and have no place in a truly "scientific" view of the person). In reality, however, many "neo-behaviorists" (who really have one foot in the behaviorist camp, the other in the cognitivist) are quite willing to apply a "soft reductionism" or "soft determinism" to human behavior; the usual term, as we'll see in a later module, is "reciprocal determinism", though this is in fact an oxymoron. (If I have any degree of free choice, my freedom is not "determined" in any sense, though it is no doubt constrained, limited, and influenced by situational factors.)
Classical behaviorism
To understand behaviorism, we must begin by understanding the nature of learning. Learning can be defined as the (a) more or less permanent change in (b) behavioral potential (c) that results from an individual's experience. Three parts of the definition are needed because (a) forgetting is possible (did you remember that?), (b) learning is not always manifested in outward behavior, (c) not all change is learning (e.g., genetic maturation).
We can speak of a "ladder of learning" beginning with the simplest (most reductionistic!) forms of learning and moving toward the most complex. The bottom rungs (0-3) of the ladder can be -- and have been -- studied extensively by behaviorists without any reference to mental or cognitive concepts. The top rungs (4-6) cannot be (and thus will not be extensively discussed in this unit).
0. Reflex behavior
1. Habituation
2. Respondent (or "classical") conditioning
3. Operant (or "instrumental") conditioning
4. Observational learning
5. Concept learning
6. Insight learning
0. Reflex behavior
At the bottom of the ladder, we have something that is not learned at all, a reflex. While we are born with many reflexes that are, by definition, unlearned, aspects of reflex behavior can be modified by experience as explained below, and these comprise levels 1 and 2 of the ladder of learning.
1. Habituation
In this simplest type of learning, the strength or intensity of a reflex is modified by how frequently the eliciting stimulus is encountered. The learning is represented by the fact that a stimulus which originally generates (for instance) interest or fear will come to elicit a less and less intense response over time as the individual learns that the stimulus is harmless or boring. Note that this simplest form of learning does not involve the generation of any new responses, but simply a change in the intensity of an existing response.
A word about levels 2 and 3 on the ladder of learning
The next two forms of learning are both examples of conditioning, or learning by association. They involve two different forms of behavior, respondent versus operant, which are most easily thought of as reflexive versus voluntary. In other words, a respondent behavior can be automatically elicited by presenting an appropriate stimulus, but an operant behavior must be spontaneously emitted. More on that below.
2. Respondent conditioning
a. Respondent acquisition
In respondent acquisition, an initially neutral stimulus (conditioned stimulus or CS) is paired with a second stimulus (unconditioned stimulus or US) which produces an unlearned reflexive response (unconditioned response or Ur). (Note the use of the small letter r to represent respondent behavior. We will reserve the capital letter R for operant behavior later on.) With repeated CS-US pairings, the CS alone will come to elicit the reflexive response (conditioned response or Cr) even when presented without the US. In extreme cases, even a single CS-US pairing can produce the Cr. For instance, a child who has no built-in fear of dogs (hence, a dog is a neutral stimulus or CS), if bitten (US) by a dog, may well develop a conditioned fear (Cr) of dogs from that point on.
b. Extinction and counterconditioning
Once learned, a Cr will generally persist indefinitely (the mere passage of time does not diminish the intensity of conditioned responses), but they can be eliminated in either of two ways. In extinction, the CS is repeatedly presented alone (without the US), gradually reducing the intensity or magnitude of the Cr. However, the extinction is initially more apparent than real, since the passage of time often makes the Cr recur (spontaneous recovery). Extinction works, but it is slow and inefficient; it is subject to spontaneous recovery; and a single CS-US pairing often restores the Cr to its full intensity. For this reason, the alternative approach of counterconditioning often works better. In counterconditioning, the CS is paired with a different US which elicits an incompatible response. For instance, opposite emotions (like relaxation and anxiety) are usually viewed as incompatible.
c. Generalization and discrimination
An automatic by-product of acquisition is generalization: if a Cr is learned in response to one CS, it will likely also automatically occur (though with lesser intensity) to similar stimuli that have not been paired with the US. (If you are in love with one person, others who look like him or her will likely also generate similar feelings, for instance.) The magnitude of the generalized response is a function of the perceived similarity between the original CS and the new stimulus (and thus this methodology can be used in perceptual research, see Unit 4). This tendency can be eliminated by means of discrimination, which is the simultaneous application of acquisition (to one stimulus) and extinction (to other similar stimuli).
3. Operant conditioning
a. Operant acquisition
Operant behaviors are spontaneous, not reflexive, so a different form of learning is involved. The basic concept is that of reward and punishment: if you want a given behavior to occur more frequently in the future, you follow it with a reward, but if you want it to happen less often in the future, you follow it with a punishment. Rewards and punishments are technically known as response consequences. From these facts, you can see three key differences between respondent and operant acquisition:
1. Respondents are reflexive (automatically elicited); operants are voluntary (spontaneously emitted).
2. In respondent conditioning, the eliciting stimulus comes first, then the elicited response, so the main behavior sequence is always: S -> r. In operant conditioning, the spontaneously emitted response comes first, then the response consequence, so the main behavior sequence is always: R -> S.
3. The goal of respondent conditioning is to change the intensity, strength, or magnitude of the response. But the goal of operant conditioning is to change the frequency, likelihood, or probability of the response.
Response consequences can be either unconditioned (you do not have to learn to find them rewarding or punishing) or conditioned (you do have to learn these properties). For instance, food is an unconditioned reward (babies are born liking to eat), but money is a conditioned reward (to a baby, there is nothing inherently rewarding about a $100 bill). Usually, you can think of unconditioned response consequences as being associated with viscerogenic motives, and conditioned response consequences with psychogenic motives. We use the following (slightly less than logical) notation to refer to the four kinds of response consequences:
SR+ unconditioned reward
Sr+ conditioned reward
SR- unconditioned punishment
Sr- conditioned punishment
To jump-start the process of operant acquisition, it is often necessary to use the method of shaping (in which we reward successively closer approximations to the desired behavior). Do you see why shaping is entirely unnecessary in respondent acquisition?
b. Eliminating an undesirable operant behavior
There are four ways to eliminate an undesirable operant.
1. Extinction: failing to reward the behavior (simply ignoring it)
2. Counterconditioning: rewarding a different, incompatible behavior
3. Punishment: following the behavior with an aversive stimulus
4. Satiation: directly meeting the need that motivates the behavior
c. Generalization, discrimination, and chaining
Discrimination can occur with operant as well as with respondent behavior. A stimulus can precede the operant behavior as a signal that the response either will or will not be rewarded. These signals are known as discriminative stimuli. The following symbols are used:
SD reward indicator
SΔ nonreward indicator
Training a sequence of behaviors in order involves the process of chaining. Behavior chains are learned in reverse order. First, the last response is acquired. Then, it is keyed to a discriminative stimulus (SD) which then takes on reward properties (as an Sr+) for the next behavior to be learned. Do you see why the intermediate stimuli are Sr+'s and not SR+'s?
Neobehaviorism
Unlike "classic" behaviorists, neobehaviorists are not strict reductionists or strict determinists. Rather, they emphasize a process of reciprocal determinism in which any of the three elements of the environment (E), the person (P), and behavior (B) can influence, and be influenced by, the other two. For instance, my existing attitudes about tuna fish (P) will interact with the current cost of tuna (E) in determining how likely it is that I order and eat a tuna sandwich (B). This behavior will in turn alter or confirm my attitudes (depending on how good the sandwich is) and will indirectly impact the environment as well (if enough people, along with me, eat tuna, its price will change).
To a neobehaviorist, most human behavior is learned through observation and direct or indirect observation (Level 4 on the aforementioned ladder of learning). In some cases this learning is deliberate; in other cases it is adventitious ("just happens" without any conscious intentionality or effort). The extent to which our behavior is likely to be influenced by the actions of a model are impacted by:
a. The characteristics of the model. For instance, we are more likely to emulate the behavior of those whom we see as similar to ourselves, and/or those whose characteristics we admire.
b. The attributes of the observer. For instance, in many cases, lowered self-esteem (whether a trait or a state) will increase the probability of imitation.
c. The reward consequences associated (objectively or subjectively) with imitative behavior.
In other words, unlike classic behaviorists, neobehaviorists do not believe that learning depends on direct external reinforcement. Vicarious reinforcement (that is, observed reinforcement of the model) can produce learning through an incentive effect. In addition, various forms of self-reinforcement can play a significant role in learning.
The latter is key to the development of the trait of self-efficacy, or the belief that one can successfully perform the behaviors required to cope with the demands of everyday life. Self-efficacy is related to the kinds of attributions (see below) that we make about our own and others' behavior. The general pattern of attributions we make over time leads us to develop either an internal or an external locus of control. The locus of control is our general belief that either we ourselves (through our actions and choices), or something outside of ourselves (circumstances, or the actions of others), primarily determine the outcomes we experience. (Of course both are relative only; everyone would agree that many outcomes are outside the individual's control.) Our fundamental locus of control influences our expectancies about the outcomes we will experience in the future, which through a feedback loop influences our own choices and actions.
An attribution is a mental (cognitive) explanation for an observed event. Attributions do not have to be accurate to influence behavior. Attributions can be classified along three independent dimensions:
1. Attributions are either internal/dispositional or external/situational. With an internal disposition, a person's behavior is explained in terms of "the kind of person they are" or something about them as a person (their interests, motivations, personality traits, qualities, attitudes). With an external attribution, a person's behavior is explained in terms of "the kind of circumstances they are in" or something about their environment, situation, or social context. For instance, "I failed the exam because I'm stupid" is an internal attribution (something about me); "I failed the exam because it was too hard" is an external attribution (something about the situation, in this case, the exam).
2. Attributions are either stable or unstable. With a stable attribution, the perceived cause is something fixed and unchanging, something that will remain constant over time. With an unstable attribution, the perceived cause is something temporary and changeable, something that is likely to alter with time. For instance, "I failed the exam because I'm stupid" is a stable attribution (I've always been that way, and I always will be), while "I failed the exam because I didn't study" is an unstable attribution (next time, maybe I will study harder).
3. Attributions are either global or specific. With a global attribution, the perceived cause influences many different aspects or domains of life, not just the specific behavior under question. With a specific attribution, a specific behavior or situation is influenced, but not much else. For instance, "I failed the exam because I'm stupid" is a global attribution (being stupid likely influences all aspects of your life), while "I failed the exam because psychology is a hard subject for me" is specific (I might be smart at everything else, so only this one class is affected).
From the above examples, you can see that any attribution can be classified on all three dimensions simultaneously. Thus, "I failed the exam because I'm stupid" is an internal, stable, global attribution. But "I failed the exam because it was unusually difficult" is an external, unstable, specific attribution... do you see why?
People who persist at difficult tasks, and who are therefore more likely both to succeed frequently and to recover quickly from failure, are those who make internal, stable, global attributions for successes in their lives, but internal, unstable, specific attributions for failures... do you see why? Hence, at-risk students, employees, and so forth can be helped to change their attributional patterns and lower their risk of persistent failure as a result. This attributional therapy approach has been used in middle school settings, for instance, with great effect.
Clinical applications of behaviorism
The key idea, not surprisingly, is one of situational or environmental determinism: behavior is caused by outward, not inward, factors (environmental contingencies, such as stimulus cues or reinforcers).
There really is no way to talk about the "structure of personality" because "personality" refers to something inward and unobservable, something behaviorists are unwilling to countenance. Instead of focusing on "personality", these therapists focus purely on outward, observable, concrete behavior patterns.
Unlike the other two schools of thought, which presume that there are unique processes of development that underlie abnormal behavior, behaviorists would say that there is really no distinction between "abnormal" and "normal" behavior: all behavior is developed and maintained in the same way. Problem behaviors continue to exist because they are rewarded in some way, because they have never been adequately extinguished or counterconditioned, and so on: there is nothing special or different about abnormal behavior in this sense, no causes or processes that are different from those that determine normal behavior.
Behavior therapy relies on three significant steps:
a. Defining the problem -- and the goal, or desired change or solution -- in specific, measurable, observable, concrete, behavioral terms. (This is sometimes very easy to do, as when a client wants to stop smoking -- and sometimes very hard to do, as when a client wants to get over being depressed.)
b. Identifying the specific environmental (situational or stimulus) conditions that underlie, maintain, or reinforce the problem behavior.
c. Developing and implementing an intervention plan that changes the environment (adding or subtracting from the existing environmental contingencies), thus changing the behavior.
Criticisms include difficulties in applying the above program to problems that are vague, verbal, or clearly cognitive or affective (subjective) in nature; the problems of maintenance (it's easier to produce short-term than lasting long-term change) and of symptom substitution (getting rid of one problem, only to find that another problem has taken its place); and the reductionistic, mechanistic, and potentially dehumanizing nature of the behaviorist assumptions.
Study Guide
1. Discuss the nature of the "behaviorist challenge", including the distinction between metaphysical and methodological behaviorism.
2. What is learning? State a formal definition of learning and explain why all three parts of the definition are necessary and important.
3. Discuss the ladder of learning and its connection to reductionism.
4. What is habituation? Give some examples, proving that habituation is learning as defined above.
5. What is conditioning? How do two types of conditioning differ? State three ways that respondent and operant conditioning can be distinguished.
6. What are four ways to eliminate an operant behavior? Which also work to eliminate respondent behaviors?
7. What is generalization? What role does it play in conditioning? What role does it play in perceptual research?
8. What is a response consequence? What are four types of response consequences, and how do they differ?
9. What is shaping? Why is shaping often necessary in operant acquisition? Why is it not necessary in respondent acquisition?
10. What are discriminative stimuli? What is their role in chaining?
11. How do classic behaviorism and neobehaviorism differ? What is meant by reciprocal determinism? Give some examples.
12. What is social learning? What are two subtypes? What are three kinds of influences that can alter the probability that social learning will occur?
13. What is meant by vicarious reinforcement? Self-reinforcement? Self-efficacy? What is the relationship between self-efficacy and locus of control? What role do attributions play in this process?
14. What are three steps in behavior therapy?
15. How do the assumptions of behavior therapists differ from those of other, more cognitively or mentally oriented therapists?
16. What are some common criticisms of the behaviorist approach?