Principles of Learning and Memory

Classic Edition

Robert G. Crowder

• Publisher: Psychology Press
• Year: 2014
• Print ISBN: 9781848725027
• eBook ISBN: 9781317599845
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

In this landmark volume from 1976, Robert Crowder presents an organized review of the concepts that guide the study of learning and memory.

The basic organization of the book is theoretical, rather than historical or methodological, and there are four broad sections. The first is on coding in memory, and the relations between memory and vision, audition and speech. The second section focuses on short-term memory. The third is loosely organized around the topic of learning. The final section includes chapters that focus on the process of retrieval, with special attention to recognition and to serial organization.

Crowder presumes no prior knowledge of the subject matter on the part of the reader; technical terms are kept to a minimum, and he makes every effort to introduce them carefully when they first occur. It is suitable for advanced undergraduate and graduate courses.

1
Background Comments and Three Analytic Concepts

The experimental study of learning and memory is almost 100 years old as this book is being written. In this same period others of the life sciences, biology for example, have undergone mind-boggling changes—imagine the confusion of a nineteenth-century naturalist walking into a modern college class in cell biology. Gains within psychology have been more modest than that since the close of the 1870s, when Ebbinghaus first undertook to study the formation of associations in the mind, but still the psychologist has no reason to feel apologetic: Any science, biology included, has had to pass through a preliminary period of simple observation, quantification, and classification, with inevitable false starts, before reaching a rate of evolution with the potential for boggling many minds. As a specific example, research on the cause of cancer seems at present to be going through such a groping phase (although, of course, the treatment of cancer is another story).

In fact, impatience with the progress of psychology in such areas as learning and memory gives way to excitement at the possibilities for the future when it is appreciated how little total effort has been invested so far. As a measure of the youth of our enterprise, consider that there are now scientists still active whose graduate training included reading all of the experimental literature on learning and memory that had ever been published. It is no longer remotely feasible, even if it were desirable, to read all of the literature on even the subset of learning and memory research to be covered in this book. Nor is this book to be a guide to that literature in any comprehensive way. Instead, the intention here is to lead the reader through the most significant conceptual developments, with due consideration to the experimental foundations of these ideas and also to the historical context from which they have emerged.

The first chapter is focused on three somewhat abstract principles having to do with how learning and memory should be analyzed. These analytic principles are generally transcendental with regard to theoretical points but they provide an organized context from which to approach theory. The ideal topical outline for this book would be a three-dimensional solid defined by these analytic principles: stage analysis, coding analysis, and task analysis. A book must necessarily follow a linear ordering, however, so we shall try to weave them through the material wherever possible. First, however, a preview of what these principles entail is necessary.

Stage analysis denotes the separation of learning and memory processes into (1) acquisition—the placing of information into memory storage in the first place, (2) retention—the persistence of memory over passing time, and (3) retrieval—the extraction of information from memory storage when it is needed. Traditionally, the term “learning” has been assigned to experimental operations where primary focus is on the first of these stages and the term “memory” to the second and third stages.

Coding analysis addresses the problem of which aspects of experiences get recorded in memory. When we “remember” a dinner from the previous week, for example, are we remembering the actual flavors as such, the names of the dishes served, or both? The assumption is that all possible aspects do not invariably become the basis for memories of those experiences and therefore coding analysis is needed to cover the principal ones.

Task analysis is the process of decomposing complex skills into simpler constituent skills in the belief that these subskills may be more tractable to theory than the global task. For example, there are two logically separable tasks in associating names of strangers with their faces. First, it is necessary to learn the new name itself (which can pose problems with names from another language) and, second, there has to be a hooking up of the learned name with the proper face; it is at least possible (and testable through experimentation) that these two processes are independent.

We shall devote considerable time to going through concrete examples of each analytic principle from the published literature on human learning and memory. This survey has several goals. The experiments that have been selected are particularly vivid illustrations of the analytic principles, but they also serve to introduce the reader to many of the major research methods in the field as well as to some of the important basic results. First, however, a few more preliminary remarks are in order.

Learning, memory, and the scope of the book. Any reasonable definition of “learning” includes many phenomena that lie outside the domain of the present book—habituation, the organization of receptive sensory fields, acquisition of strategies in problem solving, operant training, perceptual learning, the processes of abstraction in concept learning, and, of course, the data and theory that have come from the animal laboratory. These subjects are important and interesting and their exclusion from a book on learning is neither an act of derogation nor imperialism: Rather, the subject matter of this book is directed toward instances of learning that are (1) exclusively human and (2) largely symbolic, because these two restrictions define an area that is internally coherent and has historical continuity. The internal coherence of this field will hopefully become apparent through the organizing power of stage, coding, and task analysis and also through numerous theoretical and methodological concepts common to the various chapters and subheadings to be covered. The historical continuity of human symbolic learning and memory extends back to the preoccupation of ancient philosophers with the “higher mental processes” and their faith that the key to understanding the mind was in the analysis of these processes.

These Greek philosophers—and their descendants down through the nineteenth-century British empiricists (and ultimately including many of the scientists represented in this book) may have been wrong about the essentially experiential (learning and memory) and essentially symbolic (verbal and imaginal) basis for mind. Alternatively, they may have been correct in these two beliefs but their followers in current psychology may be wrong in expecting experimentally derived solutions for problems of mind (as opposed, for example, to solutions derived from such nonempirical fields as philosophy or artificial intelligence). In our present condition of basic ignorance about how the mind works, we can hardly decide between the ultimate usefulness of these different approaches. However, these considerations do provide a clear rationale for the content and approach of a book such as this one, to wit, the propositions (1) that psychology has a special responsibility to elucidate the mental processes of human beings among all of the other species, (2) that one distinguishing feature of human mental processes is their symbolic nature, and (3) that the main basis of symbolic mental processes is in learning and memory. These propositions have guided the content of this book.

There are some matters of terminology and definition that cannot be put off until later: One is what we mean by “learning” and “memory.” Formulating a definition of learning is an instructive exercise, pondering the various inclusions and exclusions that must be added to any straightforward statement, but for the present purposes we may simply describe it as a change in the organism that occurs at a particular time as a function of experience. (Experience, of course, may also produce many other changes that are not learning.) The change in the brain that constitutes learning corresponds to what gets entered into memory. However, this change, learning, cannot itself be observed directly and therefore some indirect performance test must be used to infer that learning has occurred. The basic form of such a test is a comparison of performance that could reflect the learning experience with control performance, in which the critical experience is missing. Learning is inferred from a difference as a function of the experience. To recapitulate and anticipate the present sense of stage analysis, therefore, the acquisition process (learning) may be studied only through performance in a memory test (retention plus retrieval).

The term memory is often used in two different senses. When reference is to “a memory,” the intention is to talk of the brain change that results from learning, the memory trace, that is, the hypothetical, unobservable product of experience to be inferred from performance. Memory as a process, in contrast, refers to the dynamic mechanisms associated with the holding of a memory trace over time and the retrieval of information about it in performance. In other words, the term “memory” is used for both the product of learning and the process of retention and retrieval. In this book the term “memory trace” is used for the product sense of the term. Forgetting, incidentally, has a technical reference to the loss of information in the retention stage; information is not forgotten if it has never been learned in the first place and it is not considered forgotten if it is just inaccessible to retrieval (although available in memory).

Stage Analysis

There must have been the inevitable Greek scholar who first conceived of separating acquisition, retention, and retrieval processes, but modern sources for this crucial distinction are Kohler (1947, p. 165) and especially Melton (1963a). If performance in a memory task is perfect, of course, one can be assured that information has been successfully learned by the subject, that it has been retained or held in storage from the time of acquisition until the time of attempted retrieval, and that the retrieval stage has contained no block itself. However, if performance in memory situations were always perfect, the world would be quite a different place.

The central problem occurs when there is a failure of retrieval. Such a failure brings up a very fundamental ambiguity: One has no idea whether the information has been (1) acquired adequately and retained adequately but is for some reason inaccessible at the time of attempted retrieval; (2) acquired adequately but then lost (forgotten) during the time elapsing between acquisition and retrieval; or finally (3) acquired inadequately in the first place so that there is nothing there to retain or retrieve. The big challenge is to tease possibilities apart through experimental means.

One of the most common applications of inferential problems related to stage analysis comes from the assessment of forgetting, the amount of information that does not persist between acquisition and a test of retrieval. It is a universal experience to try to recall a name or a word, without success, and to conclude that the missing item has been forgotten. However, the test of recall—which for the moment we can define as trying to reproduce information “out of the blue”—is only one of several, fallible measures of what has been retained. Often when the “forgotten” material is later discovered or somehow otherwise presented, people experience immediate, high confidence that they recognize it or find it familiar. Because recognition itself is a measure of retrieval, however, and therefore a performance test of retention, this must mean the information had indeed been retained. Sometimes, after years of disuse, people lament having forgotten a whole language, perhaps one they used as a child but not since; however, often the ostensibly forgotten material can be relearned at substantial savings, more quickly or easily than a totally new language. By this still more sensitive measure of retention, relearning, still more deeply inaccessible memory traces may therefore be recovered. From these commonplace demonstrations of the logic of stage analysis we turn now to examples from the laboratory.

Meaningfulness and forgetting. Most laymen accept as self-evident the notion that meaningful material is forgotten more slowly, over time, than nonsense material. Pleas for “meaningful” as opposed to “rote” learning in the field of education are partly defended on this supposition. However, such truisms of folk psychology should not be accepted uncritically by psychologists. This particular problem yields, in an interesting way, to stage analysis, as explained originally by Underwood (1964).

For purposes of the present discussion, meaningfulness may be defined as the extent to which verbal materials resemble sensible English words or familiar letter sequences. Three-letter groups high in meaningfulness are ABC, POT, and THO, whereas comparable items low in meaningfulness are XVQ, HJI, and YWU. It should come as no surprise that subjects indeed perform better with items high on the meaningfulness dimension than with items low on it. The question is how to understand this advantage in terms of stages. Better performance on high-meaningful items may result from an advantage at any one of the three stages, from any pair of them or from all three. Our concern here is with separating acquisition and retention effects. It has been known since the earliest experiments on learning (Ebbinghaus, 1964, originally published in 1885) that meaningfulness bestows an advantage in learning (acquisition) but untangling the influence of meaningfulness on retention is more difficult.

Clearly it will not do to present high- and low-meaningful materials to two groups of subjects for a constant amount of study time, to excuse them for some retention interval, and then to call them back for a test of memory. In this hypothetical experiment the memory test would surely show a performance advantage for high meaningfulness, but because the fixed study time would have allowed better learning of the high-meaningful materials, than of the low-, it would be impossible to know whether there had also been a retention advantage.

A better solution is to measure the progress of acquisition in groups receiving high- or low-meaningful materials and to interrupt study at a fixed criterion, say 80% of all the information to be learned. Then, if all subjects are stopped and excused when they have just reached the 80% criterion, it may be supposed that acquisition has been equated and that any differences that show up after some subsequent time period are assignable to retention losses over that period. However, Underwood (1964) has shown that even this method can lead to incorrect conclusions; the argument is rather too involved for coverage here, but it is explained carefully in his article.

The main problem is in knowing exactly how much each group—high and low meaningfulness—really knows at the beginning of a retention interval. Without this information it is not possible, then, to estimate what they have lost during the interval. The most elegant solution is to divide each experimental group into two subgroups, testing one subgroup at the beginning of the retention interval, right after the termination of study, and the other after the retention interval. Forgetting is estimated by comparing initial and delayed retrieval scores for each type of material.

Underwood’s experiment. The experiment reported by Underwood (1964) illustrates this approach and answers our question about whether meaningfulness affects forgetting. Underwood’s subjects were required to learn paired-associate lists of either high or low meaningfulness. We should pause to describe the task before continuing with the experiment. In paired-associate list learning, a list of double items is presented in such a way that the subject has to learn, given one of the items in a pair, to provide the other member on his own. The given item is called the stimulus term and the item to be retrieved in response to it is called the response term. This is the same task as learning a foreign language vocabulary lesson such that, given an English word, the corresponding Spanish word can be reproduced. A trial in paired-associate learning consists of one learning experience for each pair on the list. Usually this experience includes both a test, where the stimulus term is given alone for a few seconds with the subject trying to supply the response himself, and an immediate feedback or study period in which the correct pairing is displayed. These trials are continued until all the pairs are correctly given by the subject, or until some fraction of them are given, at the discretion of the experimenter.

In Underwood’s experiment, the items forming the pairs to be learned were single letters, not foreign-language equivalents. One paired-associate list consisted of easy, meaningful items, such as V-W and A-B; the subjects’ job was to learn to say “W” given “V” or “B” given “A.” On the other list, the low-meaningful one, the pairs did not reflect real-world associations (Z-J, N-Q, for example). There were four groups of subjects, two learning the low-meaningful list and two learning the high-meaningful list. All subjects were interrupted in acquisition just after the trial on which they first correctly anticipated (in response to the test with only a stimulus item showing) six of the nine pairs.

To reach the criterion of si...

Table of contents

• Cover
• Title
• Copyright
• Dedication
• CONTENTS
• Introduction to the Classic Edition
• Preface
• 1 Background Comments and Three Analytic Concepts
• 2 Iconic Memory
• 3 Echoic Memory
• 4 Recoding by Speech in Short-Term Memory
• 5 Nonverbal Memory
• 6 Primary Memory
• 7 Forgetting in Short-Term Memory
• 8 The Interference Theory of Forgetting in Long-Term Memory
• 9 The Effects of Repetition on Memory
• 10 The Organization of Memory in Free Recall
• 11 Retrieval
• 12 Serial Organization in Learning and Memory
• References
• Author Index
• Subject Index