Learning Theory

The development of learning theory

General approaches

Phenomena-centered theories

Current work in traditional areas

BIBLIOGRAPHY

Since its emergence as a relatively distinct topic, learning theory has played a central role in psychology. Historically, many psychologists interested in the scientific understanding of behavior have worked with learning phenomena, while psychologists with major interests in areas other than learning, as well as workers in related disciplines, have considered learning to be a pervasive process that enters into quite diverse aspects of behavior. Widespread interest in learning theory has followed the recognition that the theoretical integration of facts and laws is an integral part of what is meant by scientific understanding and that theory serves useful organizational and conceptual functions. Objections to the theory aspect of learning theory have generally reflected disagreement concerning the schedule of the theorizing relative to the empirical development of the field, rather than questions of the ultimate desirability of theorizing about learning.

At the present time the kinds of activities subsumed under the rubric of learning theory present a rapidly changing and expanding pattern of interests. Because of this situation it is impossible to define or specify learning theory in any simple way. Indeed, neither “learning” nor “theory” is a term that is used with consistent meaning by those active in the area. Commonly, learning has been considered to be a process which results from practice and which is reflected as a more or less permanent change in behavior. In many traditional learning theories, learning has been carefully specified to be some sort of associationist process as distinguished from motivational, maturational, inhibitory, and fatigue processes. While learning is thus defined in some cases, even a cursory survey of those systems called learning theories reveals that they include motivational and inhibitory factors. In most such formulations, with the exception of mathematical theories of learning, consideration of motivational variables greatly overshadows concern with the more narrowly defined learning process itself. Thus, many learning theories are in reality theories of behavior, with the term “learning” more or less limiting the range of behavior included.

Similarly, the word “theory” is used in many ways, with an even greater range of meaning than is the case with “learning.” At one extreme are theories which represent nonspecific verbal systems that better serve the “psychology of discovery” of the theorist than satisfy the basic requirement of theories with respect to the integration of data or generation of testable predictions. In this respect, it should be noted that the present discussion applies the term “theory” to a wide variety of formulations, without requiring that they meet criteria of testability, usefulness, or scope. At the other extreme are very specific uses of the term deriving from its usage in mathematical logic. In more restricted uses of the term, there are theories which represent, in varying degrees of quantitative elaboration, hypotheses about the interrelationships of systems of constructs, such as those of the intervening variable type. Further complicating the picture is the increasing use of the term “model” (Lachman 1960), which has a similar variable meaning.

It is obvious that no simple classification of learning theory is possible, and instead of an attempt to develop an arbitrary classificatory scheme attention will be directed, first, toward the development of learning theory and, second, toward a characterization of present approaches and formulations.

The development of learning theory

In the United States, systems or theories specifically concerned with learning and motivation began to emerge in the late 1930s and early 1940s. Coming from a background of “schools” of psychology, for example, structuralism, functionalism, gestalt psychology, and behaviorism, learning theory took the form of systematic positions organized around individuals who promulgated systems of constructs, principles, and research strategies in attempts to account for varying ranges of learning phenomena. Closely connected with this development were the controversies which arose about the basic nature of learning and reinforcement. These controversies, which came to dominate much of the activity in learning at that time, furthered the establishment and growth of the individualistic systems.

Major systems

The major systematic positions were the subject of Hilgard’s influential book Theories of Learning (1948), the book itself being instrumental in establishing the term “learning theory” in common usage. Among the systems described by Hilgard were Thorndike’s early connectionism, Guthrie’s emphasis on contiguous conditioning as a basic principle of learning, Hull’s attempt to develop a highly rigorous quantitative theory based on data from simple learning situations, and Tolman’s cognitive, gestalt-influenced theory, which stressed “sign-learning.” The contributions of Pavlov and Bekhterev certainly must be listed here also because of their tremendous influence on the development of learning theory and because of the importance of present-day neo-Pavlovian theory in the Soviet Union. [See the biographies of Bekhterevand Pavlov.]

The learning theories of this period were characterized by Spence (1951) as being divided on two major issues: first, the nature of the concepts used to represent the hypothetical changes taking place in learning, and, second, the conditions believed to be necessary for these changes to take place.

Sign-signiftcate versus stimulus–response. With respect to the first of these issues the comparison was between those (for example, Koffka, Köhler, Lewin, Tolman) who considered learning to reflect some kind of a perceptual reorganization or restructuring of the subject’s cognitive field which corresponded to the stimulus relationships present in the environment; and those (for example, Guthrie, Thorndike, Hull) who conceived learning to be a modification of the strength of associations, habits, or response tendencies. The former were called S–S (sign–significate) theorists, the latter S–JR (stimulus–response) theorists. The emphasis was directed, respectively, toward the effects of field conditions and other variables on perceptual organization and the relations between sensory events and toward the factors influencing the strength of associations, whether the associations were represented as empirical functional relationships or defined theoretical constructs. [See the biographies of Guthrie; Hull; Lewin; Thorn-dike; Tolman; Watson.]

Reinforcement versus contiguity. The second division was termed the reinforcement–contiguity issue. Here the distinctions concerned whether or not environmental aftereffects of behavior operated in some manner to change the strength of the learning process. Also involved were issues about the usefulness of special theories of reinforcement that attempted to identify the nature of the reinforcement or the manner in which the learning association was changed. Mention must also be made of two factor theories that generally postulated a classical–instrumental or autonomic–skeletal breakdown in which different kinds of learning were involved. These dual theories, which became increasingly popular, were held at various times by B. F. Skinner, Harold Schlosberg, O. H. Mowrer, and Kenneth Spence, among others. While many variations were proposed, contiguity principles were commonly paired with classical conditioning or with the conditioning of autonomicnervous-system responses, and reinforcement theory was usually paired with instrumental-skeletal responses.

Learning controversies

The learning theories and issues discussed above resulted in a great many disputes and controversies regarding the nature of learning, especially discrimination learning. Absolute and relational views of discrimination learning represented one such issue. The absolute position held that discrimination learning involves the strengthening or weakening of the response of approaching different aspects (discriminanda) of the total stimulus configuration, as a function of reinforcement and nonreinforcement; the relational position viewed discrimination learning as depending upon inherent perceptual-organizing tendencies, with the response always being to certain relational properties inherent in the stimulus configuration. This distinction also appeared in views of the nature of generalization and in analyses of transposition phenomena.

Another important controversy of the period centered on continuity and noncontinuity interpretations of discrimination learning, interpretations that were concerned with the question of whether an animal learns about environmental events which are being differentially reinforced but to which he is not responding differentially. The continuity position’s answer to this question was, Yes; the noncontinuity answer, No. Interest in the problem declined because of the difficulty in testing the alternative positions that developed, although in a modified and less controversial form the general question of attention in discrimination learning remains an active area of interest. Finally, mention should be made of disputes regarding “latent learning,” place (cognitive) versus response learning, and insight versus trial and error learning. [See Learning, article ondiscrimination learning; Perception, articles Onperceptual developmentanddepth perception.]

The shift from major systems

It must be recognized that these positions and controversies occupied the major attention of learning theorists for a relatively long period during the growth of interest in learning phenomena and that much of the research of this time was in the context of these issues. Thus, the shift away from these formulations that began to be evident in the early 1950s marked a distinct change in both the direction and content of learning theory. This transition, which is clearly evident in a comparison of Hilgard’s first (1948) and second (1956) editions of Theories of Learning, was due to a number of factors. Without detracting from the historical importance of these approaches or the recognition of their essential contribution to all aspects of current learning theory, it can be recognized that the problems and limitations of the systematic and controversy-oriented theory of that period were such as to lead to change. As psychology became more sophisticated about applying testability criteria to theories, the demand increased that concepts and constructs (or the systems and models into which they were incorporated) should have empirical reference. It became obvious that many of the positions and controversies that had been the focal points of debate were not formulated in a way that would provide clear-cut empirical predictions. In other words, with some exceptions these were systems at the verbal level that could not be translated into the clear-cut experimental manipulations necessary for unambiguous testing. While adequately representing general approaches and serving certain heuristic functions, many of the systems did not serve a desired function of theories—that of integrating and predicting laws. It was also recognized that one possible reason for their lack of specificity was the range of behavior included. It was found that with the existing state of empirical knowledge, learning, as an area, was much too complex to be adequately handled by these broad approaches.

By and large the systems were concerned with simple learning situations; for example, classical conditioning, instrumental learning, and simple discrimination learning. While many learning psychologists recognized the strategy of building from the simple to the complex, they were not satisfied with the pace, were skeptical of the system–controversy aspect of the activity, or felt that important variables were being neglected. These workers, therefore, applied some of the techniques and concepts to other areas of interest or became involved in different kinds of theorizing. The changes as they have developed have represented a distinct turn toward “smaller” theories that are more closely tied to empirical data and that often deal with a single or closely related group of phenomena. Thus, these “smaller” theories have not been based on the theoretical predictions of a few dominant theorists, instead, they have proliferated as phenomena or areas of investigation have been developed or have caught the interest of investigators. Concurrently, there has come a great increase in the use of the model and in the utilization of mediation process notions in all areas of theory construction.

Current learning theory

Current activity in learning theory cannot be simply classified along orthogonal dimensions, such as type of theorizing employed, type of learning phenomena involved, or experimental situations used. Rather than attempting to develop and justify a complex classificatory system, the following section will use broad categories that are intended only to serve a loose organizing function. Considered first are more general theories that have a relatively close relationship to the older systems; second, those which deal with classes of behavior or specific variables. In addition, theoretical activity concerned with traditional experimental learning situations, for example, classical conditioning, instrumental conditioning, discrimination learning, and verbal learning, is discussed, along with a brief consideration of more complex learning situations. It should be recognized that considerable overlap exists between these divisions and that many individual theories and areas of theoretical activity are omitted.

General approaches

While it is obvious that present-day learning theory is not primarily engaged in the elaboration of the theoretical structures of previous systems, the influence of older formulations is clearly represented in current work, and some theoretical activity has been rather directly derived from the “classical” positions. The latter has been the case more for S–R theory than for S–S approaches. There are several theoretical systems that are closely related to Hullian theory, and, similarly, the relationship of stimulus-sampling models to Guthrie’s position is obvious. On the other hand, the influence of S–S theory is primarily evident in approaches that stress perceptual and cognitive variables, for example, perceptual models of discrimination learning and notions of cue utilization.

Modifications of S–R approaches

Miller. The changes in S–R approaches are exemplified by Miller’s discussion of the “liberalization of S–R theory” (1959), which includes a consideration of the application of S–R concepts to central processes and the role of cybernetic-type feedback systems and attentional mechanisms in behavior. While admitting that postulation of central processes within an S–R framework reduces the difference between S–R and cognitive theory, Miller describes the S–R position as one that tends to apply the same laws to central processes as to peripheral stimuli and responses; this is in contrast to cognitive theory, which is characterized as being less specific about the laws involved. A characteristic of Miller’s work has been his attempt to apply laboratory-developed theories and concepts to complex social-behavior situations, the application of his theory of approach–avoidance conflict behavior to diverse and complex human behavioral situations being a case in point. In a sense this is a “model” approach, since the attempt is to find isomorphism between the systems developed in simple animal learning situations and complex human behavior. This approach is to be contrasted with that which attempts to expand theories dealing with a restricted range of simple phenomena by gradually integrating variables and laws from other behavioral situations. [See Conflict, article onpsychological aspects; Cybernetics.]

Spence. Spence’s theory (e.g., 1956) developed from early collaborative work with Hull. Starting with a quantitative S–R account of discrimination learning, Spence has developed in his later work a system that is more systematic than Hull’s theory and much more closely tied to empirical data. In contrast to Hull’s broader, less empirically based approach, is Spence’s detailed concern with such topics as (1) the fractional anticipatory goal response, which is proposed as the mechanism underlying incentive; (2) the role of frustration in partial reinforcement and extinction; and (3) his theory of emotionally based drive. A major objective of Spence’s theorizing has been to develop formulations that would allow for the derivation of empirical relationships found in a variety of learning situations and that could, with the addition of “composition rules,” extend to more complex learning phenomena. [See Drives, article onphysiological drives.]

Mowrer. Another theorist to be considered in this section is Mowrer. His most recent formulation (1960) assigns a central role to the classical conditioning of implicit responses or emotional states, which are called hope, disappointment, fear, or relief, depending upon the nature of the reinforcer (positive or negative) and the relation of the stimulus to the reinforcer (signaling its presence or imminent onset, or absence or approaching cessation). Mowrer is one theorist who does not follow the trend toward more restricted theorizing; rather, he proposes that his basic explanatory principles will encompass a wide range of human learning phenomena.

Mathematical theories

Perhaps the most rapidly expanding area of learning theorizing is that of mathematical (stochastic) theories of learning. Two principal lines of development have generally been distinguished. Statistical learning theory, or stimulus-sampling theory, has used conceptions of the environmental stimulus situation to obtain learning axioms and theorems about the changes that occur in response probabilities as a consequence of environmental events. Operator models, on the other hand, have been primarily concerned with those properties of response sequences that are a result of various transformation rules; assumptions about outcome effects and response classes appear in the nature of the particular model. Both approaches share similar features, such as the assumptions that the environmental outcomes associated with response alternatives change the distribution of choice probabilities and that probabilistic mechanisms govern response selections. Mathematical representations of learning have been quite successful in handling the data from some learning situations. Often, however, these situations have been specifically arranged to lend themselves to mathematical treatment and do not represent paradigms commonly used by other theorists. Further problems have been the relative difficulty of deciding when a particular formulation is appropriate and the fact that there are often a number of alternative assumptions or models that lead to essentially the same results. While promising advances have been made, the future of this approach will be determined by its success in overcoming obstacles and arriving at transsituational mathematical representations of basic learning processes. [See Models, mathematical; for a survey of this area, see Sternberg 1963.]

Phenomena-centered theories

Turning to theories which tend to deal more with certain kinds of behaviors or classes of variables, brief mention will be made of several areas in which the general shift toward phenomena-centered theorizing is evident.

Curiosity behaviors and reinforcement

One trend that has developed since the 1950s has been the increasing attention directed toward exploratory, manipulatory, and curiosity behaviors; and it is not surprising to find corresponding theoretical formulations which attempt to integrate the data of this area. One such theory is represented by the work of Berlyne (1960), who considers four variables to be of primary importance in stimulus-selection processes: novelty, uncertainty, degree of conflict, and complexity. The organism is presumed to direct attention both by central processes and by exploratory behavior (orienting responses, loco-motor exploration, and investigatory responses) that alters the stimulus field. These variables are integrated with arousal-level concepts and further tied to reinforcement, for example, in that arousal reduction may be reinforcing.[see Attention; Stimulation drives.]

Concern with the nature of the reinforcing event is characteristic of formulations dealing with the effects of novelty, exploratory behavior, curiosity, and similar stimulus variables and response patterns. To a large extent this concern has represented dissatisfaction with the tendency of older theories to expand their motivational and reinforcement notions from a single drive or drive mechanism and with their disposition to concentrate upon a few biogenic drives, for example, hunger. Also contributing to this interest has been the demonstration of the high reinforcing value of visual and manipulatory exploration. This work, in which Harlow has played a major role, has forced learning to attend to a new class of variables in a manner similar to the way in which gestalt psychology focused attention on a previously ignored set of perceptual phenomena. While theory concerned with novelty, curiosity, and similar variables has generally not reached the degree of specificity associated with some other areas of theorizing, it is an active and promising area that will undoubtedly become integrated with theories that presently do not deal with these variables to any great extent.

A similar situation exists with respect to investigations of the orienting reflex. Starting with Pavlov’s original work, the orienting reflex has proved a rich topic for research in the Soviet Union and has been the focus of a great increase of interest in the United States. The orienting reflex, which is considered to be a functional, centrally organized and integrated system of somatic, visceral, and cognitive reactions, is evoked by changes in stimulation or “novel” stimuli. Sokolov, the most prominent worker in this field, has elaborated a neuronal model concerned with the properties of the orienting reflex (1960).

Verbal processes

Of similar interest is the concern with the role of verbal processes in learning. While these processes play a major role in some theories of discrimination learning, interest in verbal processes also serves as a more general framework within which many theoretical formulations are being made. Work in the Soviet Union is particularly noteworthy in this respect. Coming from the separate but related traditions of Pavlov and L. S. Vigotski, Soviet researchers have increasingly been concerned with the second (verbal) signaling system and its relationship to learning. Luria (1961) among others has been quite active in theorizing about the verbal regulation of behavior, especially voluntary movements, with an emphasis on developmental factors both in normal and abnormal children. Note should also be made of the growing interest in relating conceptions of orienting reflex and feedback to theories of the development of voluntary action. [See Concept formation; Learning, article Onverbal learning.]

Punishment

Another area that has seen a large increase in theoretical activity is that concerned with the effects of punishment on behavior. It has become clear that punishment can have a wide variety of facilitatory, inhibitory, or suppressive effects depending upon the behavioral, situational, and punishment parameters involved, and a number of theorists have attempted to integrate these effects into existing theoretical structures or to develop principles which will link the various experimental findings. Thus, some theorists have discussed the conditioning of anticipatory punishment cues or have considered punishment to be a special case of avoidance learning, while others have emphasized the role of fear, the nature of the skeletal responses elicited by the punishment, or the stimulus properties of punishment. [See Learning, article onavoidance learning.]

Developmental psychology

Another increasingly active, area with import for learning theory is developmental psychology. The recent trend in this area has been a de-emphasis of normative, naturalistic observation and an increasing use of the experimental method. Correspondingly, there has been a turning away from the “grand” developmental theories as theories, although their utilization as a source of ideas continues. Thus, the conceptual framework and insightful observations of Piaget have occasioned intense interest in developmental psychology, and considerable effort is underway to translate the system and specific ideas into experimentally testable form. As this sort of experimental activity continues, developmental psychology seems destined to have closer ties to other areas of psychology. Indeed, those working in various content areas have also been moving toward developmental concepts. It appears obvious that learning theory must utilize, include, or become integrated with specific experimentally based developmental theory if it hopes to make significant progress in the future. [See Developmental psychology.]

Neurophysiology

Brief mention should also be made of the current work on the neurophysiological basis of learning. Although learning theory of the past has not emphasized physiological constructs to any great extent, this situation may well change as progress is made in understanding the neural basis of learning. It should also be noted that physiological theorizing has generally taken the form of hypotheses about the nature of the physiological or biochemical mechanisms involved. This is in contrast to learning theory in the United States, which has been much more inclined toward the use of systems involving defined concepts or constructs. Theorizing in both areas has felt the impact of the “model” approach, and some rapprochement may occur because of this. Learning theory in the Soviet Union has been much more closely tied to neurophysiological concepts. [See Learning, article onneurophysiological aspects; Nervous system.]

Other phenomena

Examples of theorizing can also be seen with learning phenomena of more limited scope. Thus, the effects of partial reinforcement in acquisition and extinction have served to trigger the development of” small” theoretical formulations that attempt to isolate the effective parameters in the experimental situation and to derive the effects from more basic learning phenomena, for example, stimulus generalization, or as special cases of more general learning theories. In some cases, empirical findings that have countered common sense expectations or the simple derivations of theory have served as the focal points of theoretical activity.

In these examples the phenomena-centered nature of current learning theory is evident, in contrast to the older formulations which tended to start with general principles or postulates concerning the nature of the learning or reinforcement process.

Current work in traditional areas

Classical conditioning

Contemporary theories concerned with classical conditioning have been summarized by Grings (1963). With some exceptions, for example, Razran’s detailed schema, the primary theoretical work has been directed more toward various aspects of the conditioning process than toward the development of a general explanatory system of classical conditioning per se. Much of the interest in classical conditioning has been in its postulated role in other, presumably more complicated, kinds of behavior; for example, theorizing regarding incentive variables, such as the fractional anticipatory goal response, has made use of classical conditioning processes, or as Lachman (1960) discusses it, classical conditioning provides the model, that is, the inference rules, for fractional anticipatory goal response theory. Similar theorizing has developed with respect to

(1) the consequences of frustration, with frustration being defined either as the blocking of an ongoing response or as the omission of a reward, as in partial reinforcement or extinction; and

(2) behavioral situations involving the use of punishment. Thus, in many situations where events are conceptualized to mediate overt behavior, classical conditioning is postulated to play some role or to serve as an inference model for the mediation theory. It should be noted that this use of conditioning is in the S–R rather than the cognitive tradition. [See Drives, article onphysiological drives; Learning, article onclassical conditioning.]

Soviet work. In the Soviet Union, theorizing about classical conditioning has remained a major interest since the pioneering work of Pavlov (e.g., see Anokhin 1955). The characteristics of Pavlov’s theorizing are well known, and while many of his specific notions regarding physiological structure and function have been abandoned or modified considerably, present-day learning theory in the Soviet Union retains a physiological orientation. The current Soviet emphasis on interoceptive conditioning, semantic conditioning, and the orienting reflex is reviewed in detail by Razran (1961). Especially noteworthy, in the present context, has been the theoretical development coming from Soviet interests in configural conditioning, the role of conditioning in verbal behavior, and the ontogenetic implications of their work.

Instrumental conditioning

Theorizing involving instrumental conditioning has taken several forms. The effects of various parameters upon instrumental learning have been of interest to theorists who have attempted to integrate empirical relationships into a more comprehensive theory, or who have tried to use the laws obtained in these simple situations to develop formulations which would make it possible to derive the data previously obtained from more complicated selective-learning paradigms. The relative simplicity of instrumental conditioning has been attractive to those who have found the more complicated situations difficult to analyze in a precise manner. Phenomena-centered theorizing is quite evident here, as in other behavioral situations. To mention only two such efforts: certain partial reinforcement effects in acquisition and extinction have been of theoretical interest, and Abram Amsel’s analysis (1962) of frustrative nonreward effects has been based on data from instrumental conditioning. [See Learning, article oninstrumental learning.]

Avoidance learning. Perhaps the greatest amount of theorizing concerning instrumental behavior has taken place with respect to avoidance learning. These theoretical formulations have generally been concerned with the mediating role of anxiety or fear in avoidance behavior and with the reinforcement principles operating in the dual learning processes of (1) fear or anxiety and (2) the instrumental skeletal response. A comprehensive overview of this area is presented by Solomon and Brush (1956). More recent theorizing has been concerned with specific avoidance phenomena, again demonstrating the trend toward more molecular theorizing. [See Learning, article onavoidance learning.]

Selective learning

The original work in selective (discrimination) learning was largely concerned with controversies regarding the nature of learning and reinforcement processes. While some of this type of work is still found, current interest has largely shifted to the various processes—attention al, verbal, etc.—which presumably mediate the learning. A closely related development has been the emphasis on phylogenetic and ontogenetic considerations, in terms of discrimination-learning performance and the relative use of the postulated mediational mechanisms. One of the most important developments in the discrimination-learning area has been the learning-set work of Harlow (1959). This research focused attention on discrimination procedures, demonstrated the relevance of this sort of research for more complex learning (for example, concept formation), and provided behavioral techniques that have proved useful in comparing human and infrahuman learning. Theoretical activity has been concerned with the nature of interfering tendencies or error factors.

A great deal of current activity is concerned with discrimination-learning “transfer” situations. A number of paradigms have been used to (1) compare the learning processes in human and infrahuman organisms, (2) explicate the nature of postulated mediating mechanisms, and (3) examine the mediational processes ontogenetically. Theoretical approaches that should be mentioned here include those of Luria (1961) and Kendler and Kendler (1962), both of which have tended to identify mediational processes with verbal behavior, and that of Zeaman and House (1963), which has developed from work with retarded children and which, although mathematical in nature, emphasizes the importance of attentional responses to stimulus dimensions. These formulations demonstrate several trends in discriminationlearning theory. First, the use of multistage mediational models to account for the data; second, the increasing use of normal and retarded children as subjects; and third, as in the case of Zeaman and House, the emphasis on observing or attentional responses. Other theories proposed for selective learning include the observing–response formulation of Wyckoff (1952) and the analyzer-mechanism approach of N. S. Sutherland (1959). A succinct phrase which describes the basic process of concern in a number of these formulations is “selective attention,” which can be conceived of in terms of stimulus—response relationships and laws or perceptual–cognitive processes. Approaches which attempt the integration of verbal-behavior relationships with discrimination-learning processes are also popular.

Verbal learning

An excellent discussion of the nature of theory in verbal learning is available in a paper by Gough and Jenkins (1963). These authors point out that verbal learning—the “rote” learning of material under laboratory conditions— did not develop from the learning theories of the 1940s but from the work of Ebbinghaus and the functionalist school of psychology. As an area, verbal learning has always been very closely tied to empirical data and methodological considerations, with little in the way of broad systematic theory. The theories that have developed have been concerned with specific verbal learning or retention phenomena and, as Gough and Jenkins point out, often have been called “analyses” rather than “theories.” This lack of broad systematic formulations has led to the development of “small” testable theories, which have been quickly modified to reflect new experimental evidence. A listing of some recent theoretical formulations or analyses gives the flavor of the work. It has been proposed by Underwood and Schulz (1960) that paired-associate verbal learning can best be considered as a two-stage process involving response learning and stimulus-esponse associative stages. A considerable amount of research has demonstrated the usefulness of this conceptualization. Underwood (1957) has demonstrated that the role of proactive interference in laboratory learning is much greater than was previously assumed, an advance that has contributed to the understanding of forgetting and has led to important changes in methodology. Finally, the development of the interference theory of retroactive inhibition has witnessed the introduction of the concepts of differentiation or the discrimination of list membership and unlearning or extinction to account for discrepancies between interference theory and the obtained data. The close interplay between data and theory is apparent. [See Forgetting; Learning, article onverbal learning.]

Complex learning situations

Concept learning, skill learning, and problem solving are areas that have generally been considered to be more complex and harder to handle in terms of theory than those previously discussed. There are several dimensions of this complexity, for example, the ease or difficulty of dealing with discrete units of behavior or a limited number of basic processes and the necessity of considering sequential relationships. Until relatively recently, theorists were reluctant to deal with these situations, as demonstrated by the early discarding of problem-box situations for the simpler classical-conditioning and instrumental-learning procedures. The resurgence of interest in and of theorizing about these more complex areas has led to attempts to extend S–R and cognitive approaches to these phenomena, and, in some cases, it has led to new conceptual frameworks which bear little obvious relationship to traditional learning approaches.

Concept learning. Concept learning has, perhaps, remained closer to conventional learning research than the other areas mentioned. Kendler (1964) has pointed out that various learning models have been applied to concept learning. He lists S—R conceptions, operant conditioning, clustering, Piaget’s methods of investigation, computer simulation of cognitive processes, and mathematical models as methods and models for the analysis of concept learning. [See Concept formation.]

Skill learning. In comparison to concept learning, theoretical activity regarding skill learning has been more removed from traditional learning theorizing. While issues such as the relative role of specific associations or cognitive sets in skill learning seem closely related to learning theory, the conceptual framework is not. Thus, the language of many models of skill learning is couched in terms of communication models, involving (1) notions of information processing, with subcategories of information translation, transmission, reduction, collation, and storage; (2) control-system models emphasizing feedback systems; and (3) adaptive-system models, with programs and memory systems which allow changes in the characteristics of the model with experience. [See Learning, article onacquisition of skill; see also Fitts 1964.]

Problem solving. Problem solving has long represented an area of controversy with S–R oriented theorists opposing gestalt-cognitive approaches. The S–R approach attempts to use such concepts as mediated generalization, response hierarchies, verbal mediators, and fractional anticipatory goal responses to account for problem-solving phenomena. Gestalt theory, on the other hand, emphasizes perceptual reorganization processes within the problem. One recent formulation or suggested framework, more in the gestalt than in the S–R tradition, has been the ahistorical, relatively rationally derived notions of Miller, Galanter, and Pribram (1960), which involve informational and feedback processes. [See Problem solving.]

The question as to what extent new language and methodological approaches are needed for theorizing in these complex areas remains to be determined. It seems evident, however, that information processing and feedback concepts of some sort will greatly influence learning theorizing in the future.

Leonard E. Ross

[Directly related is the entry Learning, especially the articles onclassical conditioning, instrumental learning, reinforcement. Other relevant material may be found in Drives; Gestalt theory; Motivation.]

BIBLIOGRAPHY

Amsel, Abram 1962 Frustrative Nonreward in Partial Reinforcement and Discrimination Learning: Some Recent History and Theoretical Extension. Psychological Review 69:306–328.

Anokhin, P. K. (1955) 1961 Features of the Afferent Apparatus of the Conditional Reflex and Their Importances for Psychology. Pages 75–103 in N. O’Connor (editor), Recent Soviet Psychology. New York: Liveright.

Berlyne, D. E. 1960 Conflict, Arousal, and Curiosity. New York: McGraw-Hill.

Fitts, Paul M. 1964 Perceptual–Motor Skill Learning. Pages 243–285 in Symposium on the Psychology of Human Learning, University of Michigan, 1962, Categories of Human Learning. New York: Academic Press.

Gough, Philip B.; and Jenkins, James J. 1963 Verbal Learning and Psycholinguistics. Pages 456–474 in Melvin H. Marx (editor), Theories in Contemporary Psychology. New York: Macmillan.

Grestgs, William W. 1963 Classical Conditioning. Pages 495–526 in Melvin H. Marx (editor), Theories in Con-temporary Psychology. New York: Macmillan.

Harlow, Harry F. 1959 Learning Set and Error Factor Theory. Pages 492–537 in Sigmund Koch (editor), Psychology: A Study of a Science. Volume 2: General Systematic Formulations, Learning, and Special Processes. New York: McGraw-Hill.

Hilgard, Ernest R. (1948)1956 Theories of Learning. 2d ed. New York: Appleton.

Kendler, Howard H. 1964 The Concept of the Concept. Pages 211–236 in Symposium on the Psychology of Human Learning, University of Michigan, 1962, Categories of Human Learning. New York: Academic Press.

Kendler, Howard H.; and Kendler, Tracy S. 1962 Vertical and Horizontal Processes in Problem Solving. Psychological Review 69:1–16.

Koch, Sigmund (editor) 1959 Psychology: A Study of a Science. Volume 2: General Systematic Formulations, Learning, and Special Processes. New York: McGraw-Hill.

Lachman, Roy 1960 The Model in Theory Construction. Psychological Review 67:113–129.

Luria, Aleksandr R. 1961 The Role of Speech in the Regulation of Normal and Abnormal Behavior. New York: Liveright.

Miller, George A.; Galanter, E.; and Pribram, K. H. 1960 Plans and the Structure of Behavior. New York: Holt.

Miller, Neal E. 1959 Liberalization of Basic S–R Concepts: Extensions to Conflict Behavior, Motivation and Social Learning. Pages 196–292 in Sigmund Koch (editor), Psychology: A Study of a Science. Volume 2: General Systematic Formulations, Learning, and Special Processes. New York: McGraw-Hill.

Mowrer, Orval H. 1960 Learning Theory and Behavior. New York: Wiley.

Razran, Gregory 1961 The Observable Unconscious and the Inferable Conscious in Current Soviet Psycho-physiology: Interoceptive Conditioning, Semantic Conditioning, and the Orienting Reflex. Psychological Review 68:81–147.

Sokolov, Eugene N. 1960 Neuronal Models and the Orienting Reflex. Pages 187–276 in The Central Nervous System and Behavior: Transactions of the Third Conference. Edited by M. A. B. Brazier. New York: Macy Foundation.

Solomon, Richard L.; and Brush, Elinor S. 1956 Experimentally Derived Conceptions of Anxiety and Aversions. Volume 4, pages 212–306 in Marshall R. Jones (editor), Nebraska Symposium on Motivation. Lincoln: Univ. of Nebraska Press.

Spence, Kenneth W. 1951 Theoretical Interpretations of Learning. Pages 690–729 in Stanley S. Stevens (editor), Handbook of Experimental Psychology. New York: Wiley.

Spence, Kenneth W. 1956 Behavior Theory and Conditioning. New Haven: Yale Univ. Press.

Sternberg, Saul 1963 Stochastic Learning Theory. Volume 2, pages 1–120 in R. Duncan Luce et al. (editors), Handbook of Mathematical Psychology. New York and London: Wiley.

Sutherland, N. S. 1959 Stimulus Analysing Mechanisms. Volume 2, pages 575–609 in Teddington, England, National Physical Laboratory, Mechanisation of Thought Processes: Proceedings of a Symposium. London: H.M. Stationery Office.

Underwood, Benton J. 1957 Interference and Forgetting. Psychological Review 64:49–60.

Underwood, Benton J.; and Schulz, Rudolph W. 1960 Meaningfulness and Verbal Learning. Philadelphia: Lippincott.

Wyckoff, L. B. 1952 The Role of Observing Responses in Discrimination Learning. Part 1. Psychological Review 59:431–442.

Zeaman, David; and House, Betty J. 1963 The Role of Attention in Retardate Discrimination Learning. Pages 159–223 in Norman R. Ellis (editor), Handbook of Mental Deficiency. New York: McGraw-Hill.

Learning theory

Theory about how people learn and modify pre-existing thoughts and behavior.

Psychologists have suggested a variety of theories to explain the process of learning. During the first half of the 20th century, American psychologists approached the concept of learning primarily in terms of behaviorist principles that focused on the automatic formation of associations between stimuli and responses. One form of associative learning— classical conditioning—is based on the pairing of two stimuli. Through an association with an unconditioned stimulus (such as meat offered to a dog), a conditioned stimulus (such as a bell) eventually elicits a conditioned response (salivation), even when the unconditioned stimulus is absent. Principles of classical conditioning include the extinction of the response if the conditioned and unconditioned stimuli cease to be paired, and the generalization of the response to stimuli that are similar but not identical to the original ones. In operant conditioning , a response is learned because it leads to a particular consequence (reinforcement ), and it is strengthened each time it is reinforced. Positive reinforcement strengthens a response if it is presented afterwards, while negative reinforcement strengthens it by being withheld. Once a response has been learned, it may be sustained by partial reinforcement, which is provided only after selective responses.

In contrast to theories of classical and operant conditioning , which describe learning in terms of observable behavior, intervening variable theories introduce such elements as memory , motivation , and cognition . Edward Tolman demonstrated in the 1920s that learning can involve knowledge without observable performance. The performance of rats who negotiated the same maze on consecutive days with no reward improved drastically after the introduction of a goal box with food, leading to the conclusion that they had developed "cognitive maps" of the maze earlier, even in the absence of a reward, although this "latent learning" had not been reflected in their observable behavior. Even earlier, Wolfgang Köhler , a founder of the Gestalt school of psychology, had argued for the place of cognition in learning. Based on experiments conducted on the island of Tenerife during World War I, Köhler concluded that insight played a role in problem-solving by chimpanzees. Rather than simply stumbling on solutions through trial and error, the animals he observed seemed to demonstrate a holistic understanding of problems, such as getting hold of fruit that was placed out of reach, by arriving at solutions in a sudden moment of revelation or insight.

The drive-reduction theory of Clark L. Hull and Kenneth W. Spence, which became influential in the 1930s, introduced motivation as an intervening variable in the form of homeostasis, the tendency to maintain equilibrium by adjusting physiological responses. An imbalance creates needs, which in turn create drives. Actions can be seen as attempts to reduce these drives by meeting the associated needs. According to drive-reduction theory, the association of stimulus and response in classical and operant conditioning only results in learning if accompanied by drive reduction.

In recent decades, cognitive theories such as those of social learning theorist Albert Bandura have been influential. Bandura is particularly known for his work on observational learning, also referred to as modeling or imitation . It is common knowledge that children learn by watching their parents, other adults, and their peers. According to Bandura, the extent to which children and adults learn behaviors through imitation is influenced not only by the observed activity itself but also by its consequences. Behavior that is rewarded is more readily imitated than behavior that is punished. Bandura coined the term "vicarious conditioning" for learning based on the observed consequences of others' actions, listing the following requirements for this type of learning: attention to the behavior; retention of what is seen; ability to reproduce the behavior; and motivation. Cognitive approaches such as Bandura's have led to an enhanced understanding of how conditioning works, while conditioning principles have helped researchers better understand certain facets of cognition.

Computers play an important role in current research on learning, both in the areas of computer-assisted learning and in the attempt to further understand the neurological processes involved in learning through the development of computer-based neural networks that can simulate various forms of learning.

Further Reading

Bower, G. H., and E. Hilgard. Theories of Learning. 5th ed. Englewood Cliffs, NJ: Prentice-Hall, 1981.

Grippin, Pauline. Learning Theory and Learning Outcomes: The Connection. Lanham, MD: University Press of America, 1984.

Norman, D.A. Learning and Memory. San Francisco: Freeman, 1982.

learning theory The formulation of laws or principles of learning, learning theory is usually identified with behaviourist ‘stimulus-response’ models of learning, centred on the process of conditioning. However, psychologists now view learning as a process involving cognition and information handling, and as a rule do not attempt to develop highly generalized theories of learning. See also AGGRESSION.