Attributes of Memory (Psychology Revivals)

Peter Herriot

• Publisher: Psychology Press
• Year: 2013
• ISBN:
  9781135984601
• Edition: 1
• Topic: Psychologie
• Subtopic: Histoire et théorie en psychologie

First published in 1974, Attributes of Memory rejected the prevalent stress on the structure of memory. It suggests that the view of memory as a sequence of stores through which information passes is mistaken. Instead, the author emphasizes the coding process of memory by which the nominal stimulus, the stimulus as presented, is transformed into the functional stimulus, the stimulus as coded. Dr Herriot proposes that there are many different forms of coding, and that efficiency of recall or recognition performance is a function of the nature of coding employed. He suggests that the subject's linguistic system is the most frequently employed linguistic device; that is, that the underlying attributes and rules of language are used automatically when material is verbal. Since the basic function of language is to communicate meaning, those forms of coding which are meaningful in nature are most effective in memory.

The book cites a great deal of experimental evidence, including many studies of the time. As well as stating a point of view, it should be useful to undergraduate and postgraduate students as a review of the early literature, read in its historical context.

Chapter 1

Introduction

1.1 Experimental psychology of memory

1.1.1 Experimental and ‘natural’ memory

The function of this first chapter is to orient the reader towards the concepts employed in the rest of the book. These will be defined more fully and evidenced in subsequent chapters. The first chapter aims at being a reasonably neutral introduction to conceptual terminology; the remainder presents a particular point of view.

Most experimental psychologists investigating memory have presented material of one sort or another to subjects and required them to recognize or recall it subsequently. The criterion of success is the recall or recognition of the correct material. Correctness is determined by the experimenter; in the recall task, the degree of correctness required is usually verbatim; that is, the subject has to recall the material in the form in which the experimenter presented it. More recently, some psychologists have required subjects to produce verbal or other behaviour which is derived from their general store of past experience rather than from a particular presentation by the experimenter. In this case there are seldom ‘correct’ responses. The objective is to obtain evidence concerning the structure of the subject’s knowledge.

It is clear immediately that the way in which we use the words ‘remember’ and ‘memory’ in ordinary language is considerably wider than the experimental psychologist’s usage. When we speak of someone remembering what another person said, we are usually talking about successful recall of the gist of the utterance, not a verbatim record. Moreover, we do not only speak of remembering something. We also remember how to do something, or when and where something happened. And we usually remember as a means to an end rather than as an end in itself. Why do the psychologist’s and the layman’s uses of the word differ so?

It may be asked, however, whether it is useful to have any labels at all for so-called areas of psychology. The existence of labels has led to the decoupling of one area from others (Reitman, 1970). Psychological processes have been termed learning processes, memory processes, or perceptual processes. Some of these distinctions may be valuable — for example, it may be useful to distinguish attention and perception from memory. However, there is always the danger that the areas are simply rationalizations of the use of different experimental techniques. The apparent conclusion to be drawn from such terminology is that, for example, memory processes and learning processes are different in kind from each other. Such a conclusion may not be justified.

Moreover, the decoupling of memory has resulted in an insularity in terms of both theory and practice. Until recently, for example, researchers into memory paid little attention to psycholinguistics. If a memory task requires subjects to employ linguistic skills, it is surprising if the experimenter does not take any account of what is known about them.

One alternative approach to the use of terms such as memory and learning is to adopt exclusively the analogy of the human being as the computer, the processor of information. Within the last fifteen years there has been a paradigm shift in psychology (Kuhn, 1962). The language or terminology of stimulus and response has been replaced by that of information processing. This has resulted in the strategy of using the input and the output of the human ‘machine’ as evidence from which to infer how the machine works. Less attention is paid to the question of why the machine works as it does, since this is regarded as logically subsequent. The computer analogy has been more successful in describing what is known than was the stimulus-response terminology. It has also given rise to a greater variety of research questions. However, mathematical information theory and measurement has not proved the unifying tool that it was at first expected to be. As a result, areas of research within the information processing area have become isolated. In particular, initial information processing and subsequent information processing have become very distinct from each other. Using older terminology, attention and perception have become distinct from memory and learning.

1.1.2 Pre-theoretical assumptions

There are, however, a great many more determinants of the contents of this book than the computer analogy. Experimental psychologists carry out their work with many, often unspoken, pre-theoretical assumptions. The very fact that they find it possible to use the analogy of a machine to describe at least part of the activity of human beings indicates certain assumptions. They are particularly concerned with making inferences about how the ‘machine’ works from what goes into it and what comes out of it. They will, therefore, pay particular attention to controlling the input by controlling the experimental situation, and to measuring the output, the behaviour of the subject. When they make inferences about how the ‘machine’ works, they will adopt a nomothetic rather than an idiographic approach; that is, they will be more concerned with making generalizations about how the human organism functions than with individual differences between people (e.g. Warr, 1973).

It must be stressed, however, that the use of the machine analogy does not imply any particular ethical or philosophical position. It is used purely as an heuristic device; that is, something which adequately describes the results of existing research and generates fruitful hypotheses. Experimental psychologists do not necessarily adopt a completely environmentalist view of human behaviour; they do not all assume that the machine produces a certain output of behaviour solely as a result of previous input. Nor is their purpose the sinister one of prediction and control of the subject’s general behaviour. Their prediction is the statement of an expected outcome of a particular experiment, and their control is of the externals of the experimental situation.

The danger of the experimentalist’s assumptions are all too clear. Because the input and the output are in the same form, he runs the risk of underestimating the extent to which the input has been transformed on its way through the machine. For example, given that he presented a subject with a list of words, and the subject subsequently recalled a large number of those words, it was not necessarilysome physical copy that the subject stored and retrieved. There is an important distinction to be drawn, therefore, between the nominal and the functional stimulus. The former is the stimulus as the experimenter presents and defines it, the latter is the stimulus as coded by the subject. The subject acts upon or transforms the physical stimulus, and this is termed coding (see 1.3.1).

As a result of the underestimation of the difference between nominal and functional stimuli (i.e. of the extent of the subject’s coding) a further unwarranted assumption has followed. It has been assumed that the limitations of the system are more important than its potential. Experimental psychologists have asked why we forget more often than they have asked how we remember. They have made tasks difficult by presenting unrelated meaningless items at a fast rate, instead of giving the subject the opportunity to use the immense resources at his disposal in the leisured perusal of a meaningful whole. It is the main purpose of this book to redress the balance a little.

1.2 The human computer

1.2.1 Structure and process

Within the framework of the analogy of the human being as a processor of information, two basic emphases may be detected. The first concentrates upon the structure of the system, the second on the processes which must underly its successful function. In computer language, structure implies the hardware, the nature of the machine itself and the system it presupposes; process has affinities to the operations the computer performs, the sequence and identity of which is determined by the programme.

The structural emphasis places particular stress upon the nature of the system. It suggests that the nature of the system places constraints upon the rate of flow of information through it. The system, it is supposed, is of limited capacity. There are certain points in the flow diagram representing the system at which these constraints are strongest. Particular structures are hypothesized at these points in the flow diagram. For example, short-term and long-term stores are distinguished and placed in that order in the flow diagram. Moreover, the very use of a flow diagram may imply a temporal sequence, with information being passed through the system from one structure (e.g. short-term store) to another (e.g. long-term store). Thus the basic concepts of the structural emphasis are the structural constraints upon the transmission of information, which is conceived as a temporal sequence.

The second emphasis is on process. Within this emphasis we may distinguish two apparently contradictory components. The first is closely connected with the structural emphasis (Atkinson and Shiffrin, 1968). If the system is of limited capacity, particularly at certain points, then control processes, usually consciously applied, may be postulated which have the function of minimizing these limitations. So, for example, one of the functions of rehearsal of items in short-term memory is to retain information in temporary store until the system has free capacity to pass it along to a more permanent storage location. Similarly, a control process of selection is hypothesized. This has the function of filtering incoming information at an early stage in the system, so that irrelevant information is discarded as soon as possible. Clearly, both these control processes are made necessary by the supposed limitations on information flow imposed by the nature of the task. And both may be consciously applied in an effort to compensate for these limitations.

However, there is a second component in the process emphasis, an elaborative component which cannot be explained in the same way. For example, the multiple nature of the coding of items cannot be explained in terms of reducing load. It seems that subjects can and do encode material in many and various ways. Not only may they code it in terms of many aspects of the situation in which it was presented (e.g. ‘What item was item x next to?’, ‘Was the voice that uttered the item male or female?’, etc.). They may also code it in terms of a wide variety of other features or attributes. These may be the product of their previous experience organized into a cognitive and semantic system. So, for example, they may code the word ‘dog’ in terms of its attribute as a domestic animal. Such coding may not always be a consciously controlled process, but rather a swift and automatic one (Posner and Warren, 1972). Its function is to provide a set of attributes by which the item can be uniquely defined. Then, when recall or recognition is required, the item can be retrieved or reconstructed without being confused with any of the innumerable related items within the subject’s experience. The emphasis is thus on richness, not on limitations; the richness of coding, and the richness of experience.

1.2.2 Short- and long-term stores

The emphasis on structural limitations and reductive coding will be considered first. Leaving aside the registration of input by the senses, two memory stores have been commonly distinguished, primary memory and secondary memory, or short- and long-term stores (Baddeley, 1972). These should be distinguished from short-and long-term memory tasks, which are defined by the length of time between presentation and recall. The distinction is between overt features of the task and postulated structures in the system. Primary memory and secondary memory have been distinguished in three ways. First, they are assumed to differ in capacity. Primary memory has a limited capacity, both in terms of amount of material and time of retention, while secondary memory has an essentially unlimited capacity. Second, they are assumed to differ in the type of coding which is applied to the information they contain. Primary memory has been assumed to favour a coding in terms of the physical or phonological characteristics of an item, whereas secondary memory favours one in terms of its meaning. Third, they are assumed to differ in the way items are lost from them, or forgotten. Primary memory loses items because they are overloading it, secondary because they are interfered with by other material.

The capacity limitations of primary memory are associated with the postulation of the control process of rehearsal. Rehearsal in its broader sense refers to the maintenance of material in a relatively uncoded form in order to minimize forgetting. It does not always result in overt vocalization; rather, it is a process that must be inferred from various other sorts of evidence.

1.2.3 Storage and retrieval

Another result of the structural approach, and in particular the notion of a long-term store, is the distinction between storage and retrieval. Storage is supposed to occur as the result of the selection and retention of presented material, while retrieval from store is considered a necessary condition for recall. Is this distinction justified?

One piece of evidence that it may be is the finding of Tulving and Pearlstone (1966). They discovered that there is a lot more memorised material available at the time of recall than is actually accessible (i.e. can actually be retrieved). By providing subjects with category cues at recall, e.g. ‘animals’, they increased considerably the number of items recalled successfully. This indicates that there is a discrepancy between what is stored and what is retrieved, and implies that the task of retrieval is worthy of investigation in its own right.

A second type of evidence often taken to indicate that it may be useful to distinguish between storage and retrieval is the superiority of recognition over recall performance. The recognition task requires the subject to identify items as having been presented earlier; the recall task requires him to reproduce such material. In the former, the subject is given some items and asked whether he saw them earlier. In the latter, he has to produce the items himself. Some psychologists (e.g. Kintsch, 1970a) have assumed that recognition performance is an index of the extent to which subjects have stored the material successfully. Therefore the inferiority of recall to recognition represents the extent to which storage and retrieval are more difficult than storage alone.

Finally, there are occasions when retrieval fails, and subjects retrieve attributes of to-be-recalled words only, or words related to them.

1.3 Coding

1.3.1 Reduction and elaboration coding

Now we consider the process, as opposed to the structural emphasis. In view of its centrality, the concept of coding processes requires further introduction and elaboration. The concept demands the distinction between the nominal and functional stimulus outlined earlier. It refers to the operations the subject performs upon the nominal stimulus in order to transform it into the functional one.

The nominal stimulus, the stimulus as presented, is first registered by the sensory systems. These retain a fairly liberal representation of the stimulus material for a short period of time, but during this period a great deal of material may be retained. The function of this sensory registration may be to retain material for a sufficient period for coding processes to act upon the physical stimulus and transform it.

The two basic aspects of coding processes are, first, the structural limitations upon the amount of information that can be processed at any one time; second, the variety of the coding processes available. These two aspects are reflected in the distinction between reduction and elaboration coding (Baddeley and Patterson, 1971). Reduction coding reduces the amount of material the subject has to process; elaboration coding adds to it. Both types of coding serve to assist recall performance: reduction coding by easing the information load on the system, elaboration coding by making material more distinctive from other material, and therefore more easily retrieved.

Reduction coding may take the form of selecting one from among many attributes of a presented stimulus item (see 3.4.1). If, for example, the nominal stimulus is a consonant-vowel-consonant (CVC) nonsense syllable, e.g. VUP, the functional stimulus as coded might be a phonemic representation of the second consonant, P.

A second form of reduction coding involves the rewriting of several items into a single coding (see 3.4.2). This coding therefore contains more information than each item individually, and may thus be used at recall to enable the items to be retrieved. The classic example in the experimental literature is given in Miller’s well-known paper (1956). Material which consisted of long sequences of the binary digits 0 and 1 was recoded so that the digits 0 to 8 were used as codes for sequences of three binary digits. Thus for example, the sequence 000 was coded as 0, 001 as 1, 010 as 2, 100 as 3, etc. The recoding does not have to be in terms of the same symbol system as the items, however, or even the same modality. Both types of reduction coding have in common the function of easing the load on the processing system. They therefore tend to be used by the subject when the material is presented at a fast rate, and w...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Original Title Page
• Original Copyright Page
• Table of Contents
• Preface
• 1. Intorduction
• 2. Surface forms of coding
• 3. Deeper forms of coding
• 4. Conditions of coding
• 5. Organizational coding
• 6. Semantic memory
• 7. Past, present and future
• Appendix
• Bibliography
• Index