Applications of Conditioning Theory

Graham Davey

• Publisher: Taylor & Francis
• Year: 2017
• Print ISBN: 9781138574946
• eBook ISBN: 9781351273060
• Edition: 1
• Topic: Psychology
• Subtopic: Applied Psychology
• Index: Psychology

The area of applied psychology known as behaviour modification or behaviour therapy had progressed remarkably in the ten years, prior to publication. Illustrative of this progress is the variety of therapeutic and behaviour management techniques now available to the applied psychologist. This volume, originally published in 1981, describes some of the important characteristics of this development, and in particular, the relationship between behaviour change techniques and the principles of conditioning theory that generated them.

This brief gives rise to three main themes. First, the book describes some of the reasons underlying the adoption of the conditioning paradigm and the epistemological advantages of the paradigm for behaviour modification. Second, a number of chapters discuss the current trends in specific areas of applied psychology where conditioning principles play an important heuristic role. These chapters deal with the uses made of conditioning theory in the areas of mental handicap, psychiatric therapy, work organizations, and the treatment of brain injury. Third, later chapters discuss some of the more recent theoretical developments in the field of behaviour modification/therapy, in particular the drift from strict behaviouristic applications of conditioning principles to more cognitive ones.

1 The experimental analysis of behaviour and its relevance to applied psychology

D. E. Blackman

This chapter discusses in a general way the relationships beteen one systematic approach to experimental psychology and an orientation to problems confronted by applied psychologists. It is often asserted that 'the experimental analysis of behaviour' has provided the scientific foundation for 'applied behaviour analysis' and for the techniques of 'behaviour modification'. However, the links between basic and applied research in these fields have also sometimes appeared to be strained. To what extent, then, do they share common assumptions, aims and methods? In what respects do they diverge?

In order to consider such questions, it is necessary to define our terms carefully and resist being sucked into a loose use of words which has allowed them, particularly 'behaviour modification', to mean different things to different people. In doing this, it is not of course possible to claim to be an arbiter of correct language, for the meaning of a word is in its usage and the usage of words evolves over time. Nevertheless, it can at least be asserted immediately that the contemporary movement within psychology known as 'the experimental analysis of behaviour' is merely one approach to investigating behaviour experimentally, that the term 'applied behaviour analysis' is but one system of analysing and interpreting behaviour in applied settings, and that not all methods (not even all effective methods) of changing behaviour are to be subsumed under the rubric of 'behaviour modification'. Perhaps these bald statements made at the outset may serve to assuage to some extent any expectation that by embarking on this discussion we are engaging in an imperialistic crusade which seeks to define psychology, both pure and applied, in a particular way, insensitive to the aspirations and achievements of psychologists of different persuasions.

The experimental analysis of behaviour

First then, what are the distinctive characteristics of 'the' experimental analysis of behaviour? It has been suggested previously (Blackman, 1974) that there are three identifiable and logically separate aspects of this approach. These are: (i) the use of the experimental techniques of operant conditioning; (ii) the search for empirical statements about behaviour which are valid and reliable but which relate to individuals rather than to differences between groups; (ii) the broad philosophical position within psychology known as radical behaviourism. When these three elements are combined together, they result in a coherent and effective strategy in experimental psychology. However, since the three strands are logically distinct, each deserves careful consideration in its own right.

Operant conditioning

The experimental methods of operant conditioning have developed from Skinner's pioneering investigations (1938). He defined operant behaviour as that which organisms emit and which is affected by its consequences, intending thereby to distinguish it from the more stereotyped and limited reflexes which are simply elicited by antecedent stimuli and which formed the basis of earlier experimental studies in classical, or Pavlovian, conditioning. Operant conditioning is the empirical study of how such emitted and potentially variable behaviour may nevertheless adjust to different arrangements of environmental circumstances and consequences. The prototypical experiment certainly requires little exposition here, so well-known has it become in psychology. In general, laboratory animals such as rats or pigeons are exposed to consistent arrangements of environmental events which are carefully controlled by the investigator. This can be achieved by testing the animals in constrained chambers (Skinner boxes) in which extraneous and uncontrolled environmental variations can be kept to a minimum, and in which the animals can remain for substantial periods of time without being disturbed by the experimenter. Within such environments, a limited number of readily controlled events can be introduced by the experimenters. These generally take the form of lights or noises (either of which may be sustained or relatively brief), the delivery of small amounts of food or water, and in some experiments the occasional delivery of brief electric shocks. Within these limited arrangements, the animal is free to behave in a variety of ways at any time. However, there is usually some device in the test chamber which can be readily operated by the animal, such as a key at which a pigeon may peck or a lever which a rat may press. If such patterns of behaviour occur, they can be easily recorded in a consistent and objective manner, usually by the automatic sensing of the closure of a microswitch or contact relay, though more sophisticated recording systems have also been used. Normally these patterns of behaviour, which are arbitrarily selected on the basis of their convenience to both the experimenter and the animal, provide the examples of operant behaviour whose occurrence is studied in the different conditions provided by different arrangements of lights, noises, food-presentations, etc.

These examples of operant behaviour are usually called responses, though it must be emphasized that this term defines the unit of behaviour studied and should not be taken to imply that the behaviour, which the animal is normally free to emit or not at any time during an experimental session, is bound to some eliciting stimulus as are the reflexes studied in classical conditioning. A good deal of unnecessary confusion might have been avoided if the units of operant behaviour had been termed 'bits' or 'acts', but the use of the term operant responses has become entrenched over time. A similar point might be made with respect to the environmental events such as the noises or lights normally employed in operant conditioning experiments. These have been termed stimuli although they rarely elicit an 'automatic' reaction on the part of the animals and they therefore do not have the properties of a goad which characterize the stimuli used in classical conditioning experiments.

The highly controlled experimental arrangements used by operant conditioners have made it possible to study effectively the ways in which simple patterns of behaviour are influenced by different arrangements of environmental events. At the most basic level, of course, it can be shown that the lever-pressing of rats becomes more frequent if it is followed by certain events; for example by the presentation of food, especially if the experimental animal has been deprived of food for a short time. It is required by the definition of operant behaviour that it should be affected by its consequences, and when the frequency of lever-pressing (operant responding) increases in this way, the consequence is defined as a reinforcer and the strengthening of behaviour is termed reinforcement.

Operant conditioners have investigated the ways in which different rules for delivering reinforcers affect behaviour. They have demonstrated (some might say ad nauseam!) that the intermittent delivery of reinforcers can exert very powerful effects and that characteristic patterns of operant behaviour over time emerge as a function of different schedules of reinforcement to which an animal is exposed. The early work of Ferster and Skinner (1957) showed, for example, that when the number of reinforcers to be delivered is expressed as a proportion of the number of responses emitted (ratio schedules) high rates of responding are typical, whereas when the availability of reinforcers is determined by the passage of time (interval schedules) lower rates of operant responding occur. Similarly, rates of responding are more consistent over time when reinforcement is unpredictable (variable schedules) then when it occurs after a specified number of responses or becomes available after a specified period (fixed schedules). One enduring interest in operant conditioning has been to identify exactly the important aspects of schedules of reinforcement which lead to these characteristic patterns of schedule-controlled behaviour in experiments with animals (for a recent set of such experimental and theoretical analyses, see Zeiler and Harzem, 1979).

Another important interest has been in discovering the circumstances in which events will serve as reinforcers. Since a reinforcer is defined in terms of its effects on behaviour, it is an empirical matter to specify when such events as the delivery of food or the delivery of other stimuli such as noises or lights or even electric shocks will serve to increase the frequency of the behaviour to which they are related (see Gollub, 1977; and Morse and Kelleher, 1977, for recent reviews of these topics). Operant conditioners have also investigated the effects on established behaviour of withholding reinforcers (extinction: see Ferster and Skinner, 1957) and of delivering response-dependent events which lead to a decrease in the frequency of that behaviour (punishment: see Azrin and Holz, 1966), They have also studied the effects on behaviour of schedules which maintain behaviour by reducing the frequency with which an event is delivered ('avoidance' schedules: see Hineline, 1977). The details of such work are not relevant here, except in so far as it is possible to claim that operant conditioners have done a great deal of experimental work to investigate and identify the effects on behaviour of different forms of consequences and different arrangements for their delivery. Their experiments have taken psychologists a very long way from a simple awareness that some patterns of behaviour can be affected by their consequences.

A parallel, but sometimes less emphasized, interest in operant conditioning has been the experimental investigation of the discriminative control of behaviour. At its simplest, this develops from the observation that aspects of the environment which accompany specified schedules of reinforcement may come to set the occasion for patterns of behaviour appropriate to that schedule. For example, if a pecking key is illuminated with a green light whenever a variable-interval schedule of reinforcement is in operation but is lit with a red light when no reinforcers are scheduled, pigeons will begin to peck at the key at a moderate sustained rate whenever the green light is present but will cease responding in the presence of the red light. In such a situation, the experimenter finds himself in a position to turn on or off at will the key pecking behaviour of the pigeon simply by presenting the green or red light on the key. Once again, a great deal of experimental work has been conducted to identify the circumstances in which an event (a discriminative stimulus) may acquire such facilitatory or inhibitory control over operant behaviour (see Terrace, 1966).

Operant conditioning has been a particularly active part of experimental psychology for more than two decades. Like other fields of experimental endeavour, it has become increasingly technical and doubtless ever more intimidating to the newcomer. Current work in this area is characterized by investigations of increasingly complex patterns of behaviour. For example, the ways in which animals allocate responses between alternative schedules which are simultaneously available have provided a method of studying 'choice' in experimental animals. The field is also marked by the use of increasingly complex mathematical expressions as operant conditioners attempt to improve the precision in their ability to quantify behaviour in relation to measurable aspects of the environment in which it occurs (see Bradshaw, Szabadi and Lowe, 1980; Zeiler and Harzem, 1979). Furthermore the experimental work of operant conditioners has in recent years become better integrated with the theoretical and experimental concerns of other psychologists, as for example with species-differences in behaviour (e.g. Seligman and Hager, 1972) and varying theoretical interpretations of learning (e.g. Mackintosh, 1974), and for this reason operant conditioning has become more diffuse from a theoretical point of view. However, for the purpose of the present chapter enough has perhaps been said to identify the principal features of operant conditioning as a component in the experimental analysis of behaviour: its principle concern is the empirical investigation of the relationships between emitted behaviour and the environmental milieu in which it occurs. It has domonstrated the power of consequences and setting conditions to exert characteristic control over patterns of behaviour which are free to occur at any time. To this extent, operant conditioners have provided empirical evidence that the behaviour of animals in controlled experiments is a delicate function of the environmental conditions to which they are exposed, and they have therefore made a significant (but not all-embracing) contribution to experimental psychology in general.

Single subject vs. group studies

The second element in 'the' experimental analysis of behaviour concerns the logical nature of the statements which psychologists working within this tradition seek to make. All empirical psychologists, of course, aspire to knowledge about behaviour which is valid and reliable, and which has some degree of generality. That is to say, their data should be correct and should support the interpretations which are placed on them, they should be likely to emerge once more if a direct replication of a study is carried out, and they should normally have some general significance rather than being limited to the particular subjects used or to the specific situation investigated. Beyond such common goals, however, Bakan (1967) has distinguished two different kinds of statement. He describes these as aggregate propositions and general propositions. The former assert something which is presumed to be true of samples of subjects considered as aggregates, while the latter assert something which is presumed to be true of each and every member of a designable class of subjects. Either kind of statement can emerge from appropriate experimental investigations, but workers in 'the' experimental analysis of behaviour overtly aim for general propositions through their studies of limited numbers of individual subjects.

The distinction between these two kinds of statements may appear vague and can sometimes be overlooked. It is therefore worth considering a hypothetical example. Suppose, for example, we wish to examine the proposition that people are happier in Italy than in Norway. We could aim to check the validity of this statement as an aggregage proposition by asking random samples of people who live in Italy and in Norway to complete our well-constructed questionnaire on happiness which leads to a valid and reliable 'happiness quotient'. Having collected our data, we might compare the means and standard deviations of the happiness quotients from the two samples, using the established techniques of inferential statistics to guide us to a decision as to whether the distributions of the scores are significantly different from discrepancies which we might expect between the scores of two random samples by chance alone. If a significant difference can be established, then we have supported our hypothesis expressed as an aggregate proposition, one which refers to the difference between samples as aggregates. Of course, there may very well be some people in Norway who have higher happiness quotients than some in Italy, but overall it is possible to detect a difference between the two samples taken as wholes: the happiness quotients are higher in Italy on average. However, we may also legitimately interpret the hypothesis that people are happier in Italy than in Norway in terms of what Bakan describes as a general proposition, one which is presumed to be true of each and every member of a designable class (in this case, 'people'). Now we would need a random sample of people and it would be necessary to expose each of them to Italy and to Norway, administering happiness tests to them in each country. Our question is now whether persons A to Z have higher happiness quotients in Italy than in Norway. To the extent that all these subjects do have higher scores in Italy, we have established empirical support for our hypothesis as a general proposition. Before leaving this facetious example, it is worth noting that there is no logical reason why the truth of the statement as an aggregate proposition should necessarily imply its truth as a general proposition (or vice versa). For example, our test of the aggregate proposition might quite reasonably be based on samples consisting almost entirely of Italians in Italy and Norwegians in Norway, and the differences might perhaps be the outcome of genetic differences between these groups. On the other hand, our sample for testing the general proposition might, for example, consist of one arbitrary member of every nationality in the world, and it is quite conceivable that in this case they might be happier in Norway than they are in Italy. If so, this does not contradict the validity of the aggregate proposition that people are happier in Italy.

There is no need here to discuss which of Bakan's two types of propositions is the more important in psychology: it is surely the case that psychology has need of both. However, it is a little disconcerting to realize how much of our psychological knowledge arises from studies which by their very nature can lead only to aggregate propositions and which therefore relate to groups, rather than to individuals. Thus a great deal of experimental psychology is based on group designs in which subjects are allocated to experimental conditions and control conditions at random and inferential statistics are used to compare the scores of the groups. It is also disconcerting to note Bakan's (1967) suggestion that psychologists sometimes slide inadvertently from studies of groups to statements about individuals.

The movement known as 'the' experimental analysis of behaviour is one area of experimental psychology (but not the only one) which is overtly directed towards uncovering propositions which are true of individual members of a designable class. Indeed, its principal organ, Journal of the Experimental Analysis of Behavior, was set up in 1958 'primarily for the original publication of experiments relevant to the behavior of individual organisms', and not for reports of operant conditioning experiments as such. However, operant c...

Table of contents

• Cover
• Title
• Copyright
• Original Title
• Original Copyright
• Contents
• Notes on the Contributors
• Editor's introduction
• 1 The experimental analysis of behaviour and its relevance to applied psychology
• 2 Applied behaviour analysis: intervention with retarded people
• 3 The token economy
• 4 Application of behaviour modification in the treatment of traumatically brain-injured adults
• 5 Behaviour modification in organizations
• 6 Pavlovian principles and behaviour therapy
• 7 Establishing behaviour: the constructional approach
• 8 Self-instructional training and cognitive behaviour modification: a behavioural analysis
• 9 Conditioning principles, behaviourism and behaviour therapy
• Name index
• Subject index