Exploring Working Memory

Selected works of Alan Baddeley

Alan Baddeley

• Publisher: Taylor & Francis
• Year: 2017
• Print ISBN: 9780367735784
• eBook ISBN: 9781351618526
• Edition: 1
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

In the World Library of Psychologists series, international experts present themselves career-long collections of what they judge to be their finest pieces - extracts from books, key articles, salient research findings, and their major theoretical and practical contributions.

Alan Baddeley has an international reputation as an eminent scholar and pioneer in the field of human memory, and is principally known for the theory of working memory, devised with Graham Hitch. This model continues to be valuable today in recognising the functions of short-term memory. This volume includes a specially written introduction by Alan Baddeley which gives an overview of the start of his career and his entry into the field of Psychology. Throughout the book he also provides introductions to the selection of works included and contextualises them in relation to changes in the field during this time. Exploring Working Memory includes the author's most influential publications on topics including short-term memory, the distinctions between short and long-term memory, the theory of working memory, the phonological loop, the concept of the central executive, and the episodic buffer. This exceptional selection concludes with an article giving a broad overview of the author's current views on working memory and its relation to other theories in the field.

Through his outstanding work Alan Baddeley has become known as a world-leading expert on human memory. Exploring Working Memory is a unique collection which will be of great interest to both students and researchers interested in human memory from psychology backgrounds.

Part I
How many kinds of memory?

My work on postal coding was supervised by Conrad (who like a certain fictional detective preferred to avoid his first name). Conrad was a brilliant applied psychologist, much valued by the government postal and telecommunications department. It was he who optimised the postal code that was eventually adopted and that subsequently proved highly successful. Our next assignment was to undertake a further quite different research project concerned with improving measures of the sound quality of telephone lines. This was typically assessed by speaking a series of potentially confusable words and checking the accuracy with which they were detected by a listener. It was suggested that requiring the listener to then process the words in some way might impose an additional cognitive load which could make the task more sensitive to the quality of the telephone line. I was assigned this project while Conrad left for a year’s sabbatical in the United States.

As part of the project, the Post Office were due to provide a system that would take speech and distort it in a way that resembled the loss of information during telephone transmission. This equipment, however, took some time to develop, and meanwhile I settled for a simple white noise generator, rather like the sound of a waterfall, where the intensity could be adjusted appropriately. I began by inventing games on topics that required conversation between participants under quiet or noisy conditions. This left me with large amounts of recorded verbiage and few ideas as to how to analyse it. I decided instead to simplify the task, presenting lists of five words for immediate serial recall. I could check the accuracy of simply hearing the words and see if errors increased when, in addition, they had to be remembered. To make the task a little harder, I had one condition in which the words were similar in sound (e.g. man, can, cat, mat, map versus pit, cow, sun, day, top). Since it was only a pilot experiment, I did not worry about the niceties of psychoacoustics, simply assembling a group of about 20 people, playing the words on a tape recorder either with or without concurrent background noise.

There proved to be a few more errors on the similar lists, but this tendency was not substantially increased by the requirement to remember, suggesting that was not the answer to the Post Office problem. What I did find however was that the similar sequences were hugely more difficult to remember than the dissimilar items, whether in noise or quiet. This was expected since Conrad had already demonstrated that similar sounding letters (b, g, t, p, c) were harder to recall in the correct order than dissimilar ( f, k, m, r, l ). Conrad (1964) related his findings to the proposal of a short-term memory system that differed from long-term memory, arguing that the system was based on an acoustic code. Having absorbed a good deal of standard verbal learning theory during my time at University College London and Princeton, I realised that verbal learning theorists typically did not discriminate among types of similarity and hence might reasonably object that this did not prove that the store was acoustic as the result might be the same, regardless of whether the similarity was acoustic or not. I therefore created a set of parallel conditions, again involving remembering sequences of five words, but this time varying the degree of semantic similarity comparing sequences like huge, big, wide, long, tall with old, late, wet, thin, hot. The results were clear, with a tiny though still significant drop in performance based on semantic similarity.

I duly submitted my paper to the Journal of Experimental Psychology whose referees failed to be convinced, although Arthur Melton the Editor and leading figure in the verbal learning community at the time was sufficiently intrigued to replicate our results and report his successful replication at a meeting in Cambridge a year or so later. They were duly published in a UK journal, followed by a paper in which we studied the effect of similarity on long-term memory by doubling the length of the lists and presented for several trials, finding exactly the opposite pattern. This time semantic similarity was crucial (Baddeley, 1966). In a series of experiments with my friend and colleague Harold Dale we subsequently re-demonstrated our dissociation using paired-associate learning in a series of papers that were indeed accepted by North American journals.

Showing a clear difference between a short-term system that appeared to rely on acoustic cues and a long-term system that reflected semantic coding, had implications for the hot topic of the moment, namely whether memory comprised a single unitary system as was widely assumed at the time, or whether it was necessary to assume more than one type of memory. A challenge to the unitary view first came in the late 1950s with the demonstration by John Brown (1958) in the UK and Peterson and Peterson (1959) in Indiana that quite small amounts of information such as a consonant triplet could be forgotten over a matter of seconds, unless some form of rehearsal was possible. Both studies were interpreted in terms of a fading short-term memory trace, while accepting the general view that forgetting in long-term memory depended on interference rather than decay. Other studies provided evidence that certain tasks might involve both a long-term and a short-term component, again suggesting separate long-term memory and short-term memory systems. A good example of this was free recall in which a list of unrelated words is presented and must be recalled in any order. When recalled immediately, the most recent items are very well recalled, the recency effect. This recency advantage completely disappears, however, after a filled interval of as little as 5–10 seconds suggesting that the recency effect may be based on a more fragile short-term memory system (Glanzer & Cunitz, 1966).

Perhaps the most convincing evidence for separate systems, however, came from neuropsychology with the demonstration by Milner (1966) that a densely amnesic patient HM appeared to have excellent immediate memory as demonstrated, for example, by digit span, coupled with grossly impaired performance on a wide range of long-term memory tests. Other patients showed exactly the opposite pattern of impaired short-term memory with preserved long-term memory (Shallice & Warrrington, 1970).

The three papers chosen in this section reflect my own introduction to the long-term memory/short-term memory debate. The first describes the initial discovery of a clear acoustic-semantic difference in verbal short-term memory. The second illustrates further complications in our interpretation of this distinction while the third describes my introduction to neuropsychology and its implications for the long-term memory/short-term memory controversy.

Although our semantic-acoustic similarity differences between short-term memory and long-term memory proved very replicable, our initial simple assumption of separate long and short-term systems based on different codes came under increasing pressure as we expanded our research on coding across a wider range of situations. It is clearly the case that phonological long-term memory must exist, otherwise how could we learn new words or unfamiliar names? The fact that memory span for sentences extends to around fifteen words rather than five unrelated words, provides strong evidence for the potential influence of meaning in immediate serial recall. The third paper in this section describes an experiment to tease apart the influences of sound and meaning, showing that people will use semantic coding, if it is easy to do so, as is the case for meaningful sentences. With unrelated words; however, it is typically too difficult to create the necessary meaningful bonds between the words on a single brief trial with the result that semantic encoding that captures the order in which they are presented will only develop over successive attempts. The relationship between semantic and acoustic coding is complex with both codes potentially operating at the same time; although, the acoustic code appears to be less durable. Chapter 2 shows that this can lead to a paradoxical tendency for performance to improve after a delay when people switch from a defective phonological to a more stable semantic code. The link between memory systems and coding was proving rather more complex than we at first thought.

The third chapter stems from an invitation from Elizabeth Warrington, a neuropsychologist at the National Hospital in London, to study a group of densely amnesic patients with otherwise well-preserved cognitive abilities. I was initially reluctant, arguing that it was unlikely that patients would have their lesions in a sufficiently convenient location to allow theoretically convincing experiments to be run. I did, however, accept the invitation and was immediately struck by the apparent purity of the deficit in the first patient I encountered. The paper illustrates our attempt to combine the methods that were evolving in mainstream cognitive psychology with the study of carefully selected patients. It presents evidence for a distinction between impaired long-term memory and pre...

Table of contents

• Cover
• Title
• Copyright
• Contents
• Permissions acknowledgements
• Introduction
• PART I How many kinds of memory?
• PART II A multicomponent model
• PART III The phonological loop
• PART IV The visuo-spatial sketchpad
• PART V The central executive
• PART VI The episodic buffer
• Index