Our ability as human beings, to retain and manipulate information over short periods of time, is fundamental to an enormous number of everyday tasks. Cognitive psychology has put some considerable effort into making sense of the means by which we can accomplish this feat, and the notion of a working memory has made a substantial contribution in providing a framework for advance in the area. Alan Baddeley's book Working Memory brings together, in one source, a summary and an evaluation of 12 years of literature, its antecedents and hopes for the future within the context of recent advances. It is a personal account by the main proponent of the working memory framework, and as such betrays an insight that would be lacking in a more biographical treatment.
The book is arranged around four sections; 'Precursors', 'The concept of a general working memory (WMG)', 'A specific model of working memory (WMS)', and Applications of working memory'. In two chapters, 'Precursors' sets the context for the initial working memory model. There is a very useful and concise summary of the issues in the study of short-term memory in the late 1950s and 1960s, with the interference-decay debate leading to the premature claim by Arthur Melton in the early 1960s that interference theory in a unitary memory system provided a complete explanation of all forgetting. The first chapter concludes with a brief description of Atkinson and Shiffrin's modal model of short-term memory.
Chapter 2 discusses the problems with the modal model in some detail, with its ultimate demise seen as a result of disuse by its main proponents. There follows a discussion of the Levels of Processing framework and its associated difficulties. Baddeley views Levels of Processing as a good 'rule of thumb rather than a feasible model, and working memory is introduced as a natural and adequate successor to the modal model.
The second section contains Chapters 3 and 4, which address 'The concept of a general working memory'as a notion with which few researchers would disagree. The notion is ofa work space which is responsible for temporary storage and manipulation of information during human information processing. The intention is to describe a system that 'should be limited in capacity, and should operate across a range of tasks involving different processing codes and different input modalities'. Chapter 3 examines the role of working memory in learning. There is first a discussion of the methodology used in the area, and particularly the secondary task procedure. Baddeley provides a useful treatment of the dificulties associated with the interpretation of the results of such procedures, in particular the 'attentional trade-off problem'. This is one major source of criticism of the working memory approach. There is only limited consideration of the problem at this stage, and Baddeley has chosen to raise this and related issues at appropriate points later in the book.
Chapter 3 goes on to describe a series of studies that examine the effect on learning, of loading working memory with a secondary task. The basic conclusion is that there is a clear effect on learning, but that the effect on retrieval is less clear-cut. The effects on retrieval, if any, are small and unreliable. Chapter 4 describes a similar approach to comprehension and reasoning. In both cases there is a clear role for a general working memory. These demonstrations are fairly fundamental to the claim for a system that is involved in everyday information processing, and these two chapters lend substantial support to the idea.
The end of Chapter 4 introduces a more specific working memory model, with specialized subcomponents as described in the seminal 1974 paper by Baddeley and Hitch. The third section of the book, 'A specific model of working memory', devotes two chapters to a detailed summary of the articulatory loop and the visuospatial sketch pad, while a third chapter deals with the relationship between working memory and recency.
Chapter 5 describes the concept of an articulatory loop; a system thought to be involved in temporary storage of speech-based material. The articulatory loop hypothesis has been the component of working memory that has received by far the greatest degree of research effort, and represents the greatest impact as a result. There is a logical and convincing progression through the variety of converging data which support the idea; the effects of phonological similarity, of word length, of suppression, and the revision of the articulatory loop model in the light of the effects of unattended speech and neuropsychological evidence. Given the range of sources of consistent evidence, the case for an articulatory loop is fairly strong.