Transfer of Learning

Contemporary Research and Applications

Stephen M. Cormier, Joseph D. Hagman, Stephen M. Cormier, Joseph D. Hagman

• Publisher: Academic Press
• Year: 2014
• ISBN:
  9781483297378
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

Since the mid-1970s, scientific and educational research has left a gap in the field of basic and applied research on transfer of learning. This book fills the gap with state-of-the-art information on recent research in the field, emphasizing methodological paradigms and interpretive concepts based on contemporary cognitive/information processing approaches to the study of human behavior. Issues discussed include how transfer is measured, how its direction and magnitude are determined, how training for transfer differs from training for acquisition, and whether different principles of transfer apply to motor, cognitive, and meta-cognitive processes.

CHAPTER 1

Introduction

STEPHEN M. CORMIER¹

JOSEPH D HAGMAN

Publisher Summary

This chapter discusses the issues of current transfer research which have relevance beyond the confines of any single methodological approach. In essence, transfer of learning occurs whenever prior-learned knowledge and skills affect the way in which new knowledge and skills are learned and performed. When later acquisition or performance is facilitated, transfer is positive; when later acquisition or performance is impeded, transfer is negative. Transfer can be general, affecting a wide range of new knowledge and skills, or specific, affecting only particular knowledge and skills within a circumscribed subject matter. The four generic issues important to a comprehensive description of transfer, both as a learning phenomenon and as an event with substantial importance to real-life situations are: (1) how transfer should be measured, (2) how training for transfer differs from training for rapid acquisition, (3) how direction and magnitude of transfer are determined, and (4) whether different principles of transfer apply to motor, cognitive, and metacognitive elements.

Since the domain of transfer phenomena is so large and varied, all the more need exists for a methodical way of exploring and organizing the research findings. This chapter presents a brief discussion of what we consider to be the important issues of current transfer research which have relevance beyond the confines of any single methodological approach.

In essence, transfer of learning occurs whenever prior-learned knowledges and skills affect the way in which new knowledges and skills are learned and performed. When later acquisition or performance is facilitated, transfer is positive; when later acquisition or performance is impeded, transfer is negative. Transfer can be general (i.e., content independent), affecting a wide range of new knowledges and skills, or specific (i.e., content dependent), affecting only particular knowledges and skills within a circumscribed subject matter (see Mandler, 1962).

In our view, there are four generic issues important to a comprehensive description of transfer, both as a learning phenomenon and as an event with substantial importance to real-life situations. These issues are: (a) how transfer should be measured, (b) how training for transfer differs from training for rapid acquisition, (c) how direction and magnitude of transfer are determined, and (d) whether different principles of transfer apply to motor, cognitive, and metacognitive elements.

I THE MEASUREMENT OF TRANSFER

The measurement of transfer has long been a contentious issue. Actually, there are several aspects to this problem. First, the degree of transfer is usually defined in relative terms where performance on a transfer task (i.e., Task 2), for example, is compared between an experimental group receiving some prior task training (Task 1) and a control group receiving no such Task 1 training. Under this paradigm, the experimental group generally outperforms the control group if Task 1 transfers positively to Task 2. How much better this performance is, however, depends upon which of a half dozen formulas are used, each of which has its own particular strengths and weaknesses. Thus an understanding and awareness of the particular formulais) used across different transfer studies is necessary for a correct interpretation of their results.

Second, it is important to be aware of the kind of performance being measured, for example, overall Task 2 learning rate or accuracy, or initial (first-trial) Task 2 performance. Even then, there is really no one “correct” performance measure to use in all cases. As a result, the purpose of the transfer study will many times dictate which transfer criteria are most appropriate. The point we wish to make here is simply that the criteria used will limit conclusions that can be made about transfer and should be kept in mind when comparing the results of different studies. Just because Task 1 learning may enhance first-trial performance on Task 2, for example, does not mean it will necessarily have the same effect on learning rate over trials, or vice versa.

Third, one should be concerned with the reliability and validity of experimental and control-group performance. Degree of Task 1 learning, for example, must be the same for both groups in order to make relative comparisons about Task 2 performance, yet degree of Task 1 learning is often difficult to assess with complete confidence. Ceiling effects, for example, may mask degree of learning once performance has reached 100% on the dependent variable scale, for example, number correct; percentage correct (Underwood, 1964). Besides ceiling effects, other significant sources of measurement error in transfer experiments include small sample sizes, floor effects, insufficient practice, and inadvertent differential treatment of groups, to mention only a few.

Fourth, it needs to be recognized that the applied environment often poses obstacles to the implementation of particular transfer paradigms or methodologies regardless of their accepted validity. The amount or kind of training that can be conducted on simulators, for example, is often constrained by operational training requirements or internal needs of the organization. Industry or military training systems usually cannot suspend or even substantially modify operations in order to permit either the required experimental control over treatment conditions or the measurement of a particularly diagnostic aspect of performance, although many times, suitable compromises can be arranged. Consequently, the results of applied research should undergo careful scrutiny, not because applied researchers are unaware of particular experimental pitfalls, but because they are unable to avoid these pitfalls due to operational constraints.

These and other performance measurement problems are not unique to the study of transfer. They all, however, can affect the interpretability and replicability of results. Fortunately, substantial progress has been made in the area of performance measurement during recent years because of increased sophistication of experimental methodologies and a better understanding of what variables are likely to affect transfer. Even with these advances, the issue of performance measurement will still be a matter for concern for years to come until we achieve our ultimate goal of accurate performance prediction on the basis of sound psychological principles (see chapters 8–10).

II TRAINING FOR TRANSFER VERSUS RAPID ACQUISITION

Trade-offs between desired goals are a common occurrence in everyday life, and many examples exist in the psychological domain as well. Perhaps the best known is the speed-accuracy trade-off, but another trade-off occurs between training for transfer and training for rapid acquisition. In other words, initial training experiences which tend to increase the amount and kind of observed transfer seem to have negative effects on the rate at which the tasks are learned to some criterion level.

This effect is undoubtedly related to the relationship between the training on the original Task 1 and the target Task 2. By training for rapid acquisition, we are making the implicit assumption that the training tasks are the target tasks. To this end, Task 1 training will involve more time spent on fewer task elements. Conversely, training for transfer involves not only training individuals on the original task, but also preparing them for related but distinct target tasks.

The extent to which acquisition or transfer considerations should predominate in designing Task 1 training will depend on the target tasks to be performed. At the conceptual level, the goal is to specify which training procedures can optimize transfer or acquisition and from that develop functions which permit trade-off estimations to be made for particular situations.

At the practical level, this trade-off must be taken into account in evaluating the success or failure of particular training programs. A training program may take more time and have lower trainee performance than a second program and yet result in higher target task performance. Of course, this result is likely only in a carefully designed training program which has set transfer as its goal. Nevertheless, this illustrates the care which must be taken in choosing appropriate criteria for transfer evaluation in real world settings (see chapters 2, 8, 9, and 11).

III THE DIRECTION AND EXTENT OF TRANSFER

A DIRECTION OF TRANSFER

Predicting whether transfer would be positive or negative from one task to another was a dominant preoccupation of human learning researchers for many years (e.g., Osgood, 1949). Explanations of transfer were typically based on the theory of identical elements and couched in the stimulus-response (S-R) language of interference theory.

The body of interference theory research uncovered many important phenomena about transfer (primarily negative interference) and forgetting (primarily retroactive interference). Positive transfer and proactive interference were less thoroughly examined. Despite the methodological brilliance of the investigators, the conceptual understanding of transfer did not seem to advance as fast as the empirical findings accumulated. As one of the foremost researchers in this area put it:

It is fair to say that the total picture is complex and beset by uncertainties … with respect to the general conceptualization of the underlying mechanisms of interference. One cannot help but wonder why after so many years of patient experimental effort interference theory today finds itself entangled in so many empirical inconsistencies and theoretical complications. (Postman, 1976, p. 179)

In effect, we still have difficulty predicting whether transfer will be positive or negative in particular cases because of incomplete knowledge of what was learned originally and how the transfer task is represented. Perhaps part of the problem can be attributed to the narrowness of the methodological approaches employed, such as arbitrary word lists, and the absence of much critical attention to the assumptions embodied in interference theory. An implicit adoption of the investigatory methods of the physical sciences can be seen, in which domain a factor can be manipulated precisely while keeping other factors at least relatively constant. In this situation, the progressive refinement of a methodology designed to measure one factor makes sense. However, in psychology, it is much more difficult to manipulate one factor without affecting other factors as well. In addition, a narrow methodology increases the odds that some critical factors will be omitted from consideration and experimental manipulation.

One of the by-products of an information-processing approach to the study of human learning is the wealth of different methodologies used, thereby creating the opportunity to arrive at solutions to various problems by way of converging operations. As the reader will see, there is a tremendous variety in the methodological procedures used to study transfer and a greater concentration on the conceptual basis of the factors hypothesized to underlie transfer phenomen. Information-processing theories have more explicit descriptions of the processes which are hypothesized to contribute to...