Common threads and fault lines

Alan Baddeley and Graham Hitch introduced the term working memory to a meeting of the UK Experimental Psychology Society in 1972 to point out that it is inadequate to consider 'short-term memory' as a simple receptacle for fast-fading traces of recent perceptions. To interact with a rapidly changing world we need an active system that integrates current with just-recorded information, predicts what new information is necessary to make sense of these integrations, shifts attention to seek and obtain this new information, sets goals for future action, makes plans to achieve them and, in short, is more concerned to foretell the immediate future rather than to record the immediate past. An unkind tease from the EPS audience was that 'working memory seems to be anything you happen to be working on at the time'. The diversity of topics covered in this collection of six essays shows that, 33 years later, a weak jibe has become marvellously prescient and working memory has indeed become the nearest candidate for a 'Grand Theory of Everything' in cognitive psychology.

Tim Salthouse discusses how nice studies, mainly his own, show that old age impairs reorganisation of recently perceived with new information. His further interesting psychometric analyses show that the concepts of fluid intelligence, (gf) and working memory are closely related, both in terms of the verbal descriptions we use to define both these capabilities and by performance on the behavioural tests that we use to ostensively define each of them.

Angela Kilb and Moishe Naveh Benjamin discuss another element in our job description for working memory: the maintenance and management of attention. Their explanation of increasing inefficiency of these processes in late life is that brain changes deplete the amount of 'central resources' available for these tasks. As David Navon once pointed out, if it is predicated only on behavioural data, this idea of 'global resources' becomes so vague as to be a 'theoretical soup-stone'. However, it does become useful and illuminating if we think of it as global degradation of neuronal functionality, associated with losses of connectivity and increasing deaths of neurones, marked by gross brain changes picked up by brain scans, such as increasing incidence of white-matter lesions. These neurophysiological changes, and their relations to behaviour, are documented by Rebecca Charlton and Robin Morris in a review of own, and others' recent work.

So a common theme in the chapters by Salthouse, Naveh-Benjamin, Charlton and Morris and other contributors is their departure from the initial assumptions made by Baddeley and Hitch and others of their intellectual generation. The early assumption was that by designing sufficiently clever behavioural experiments we can identify subsystems responsible for each of many different capabilities attributed to working memory: such as 'binding' and associating different attributes of perceptions, separately retaining and analysing visual and auditory input, etc., etc. We can then track relative age changes on each of these behaviourally defined systems by comparing performance of old and young people across the tasks that we have used to define them in the first place. If we have done our job properly, we can further hope to relate differential changes in performance on the behavioural tasks we have used to define these putative systems to changes in correspondingly distinct neuroanatomical structures.

A logical problem in using this approach to study individual differences is the circularity of defining systems in terms of scores on different tasks and then using differences in scores on these tasks as evidence for differential changes. A methodological problem is how to be sure that the different behavioural tasks on which we compare old and young people are equally difficult. A further problem, as Salthouse's psychometric analyses illustrate, is that scores on the various tasks we have used as definitions of putatively separate capabilities are, actually, closely inter-correlated. Logie et al. compare the amounts of age-related decrements in performance on behavioural tasks that have been used as ostensive definitions of various putative working memory subsystems, such as auditory and visual storage, binding, etc. etc. They hope to avoid the problem of incommensurability of difficulty of these different tasks by standardising scores for older adults on average baseline scores by young adults. I do not see that this solves the problem. I think that the sensitivity of tasks to detect change must also depend on the number of possible gradations of scores that each can yield rather than solely on differences between simple average scores. In addition there remains the major issue that scores on these putatively independent tasks are, in fact, highly inter-correlated. A more convincing statistical treatment to avoid all of these problems might be to derive a factor structure for data from all of these tasks and to compare age loadings across these factors.

So, a fault-line between assumptions underlying these essays is the gap between the idea proposed by Logie et al. that the way forward is to compare scores of old and young adults on behavioural tasks defining putatively distinct information-processing subsystems and the contrasting psychometric treatment of aggregated task data by Salthouse, the 'general resource decline' models of Naveh-Benjamin and, even more interestingly, the global, and task-undifferentiated, changes in connectivity and computing power suggested by the neurophysiological data collected by Charlton and Morris. My personal bets are on the success of these latter, newer, descriptions rather than on the outdated cognitive models on which Bob Logie and I were brought up.

The remaining chapters are a fine and useful review, by Stigdotter-Neely and Nyberg, of evidence for and against the idea that 'brain training' on selected working memory tasks can benefit performance on other tasks that make quite different ('distant') rather than closely comparable ('near') demands. Also a concluding, masterly, conceptual overview of the field by Nagel and Lindenberger.

I have enjoyed this collection of essays, and believe that I have learned more from the diversity of approaches and assumptions they represent than from any common thread between them. I warmly recommend them to all who are interested in individual differences in working memory, even if they are not, at all, interested in the tiny and unfashionable field of cognitive ageing. 

Psychology Press; 2014; Pb £29.99
- Reviewed by Patrick Rabbitt who is Emeritus Research Professor of Gerontology and Cognitive Psychology, University of Manchester