A life in grey areas: cognitive gerontology from 1957 to 2007 Pat Rabbitt

A life in grey areas: cognitive gerontology from 1957 to 2007

Pat Rabbitt

DOI:10.1093/acprof:oso/9780199228768.003.0017

Abstract and Keywords

This chapter examimes changes and developments in the field of cognitive gerontology during the period from 1957 to 2007. It describes personal experience in studying psychology and conducting research on cognitive gerontology. This interest in cognitive gerontology was significantly influenced by the works of Alan Weber, Donald Broadbent, C. Brinley, and Tim Salthouse.

Keywords:   cognitive gerontology, psychology, Alan Weber, Donald Broadbent, C. Brinley, Tim Salthouse

The history of science is written about, and sometimes by, remarkable people who have radically changed our understanding of the world, but the footnotes of pedestrian scientists can tell how it felt to live through these major transformations without temptations to structure chaos or edit muddle into prescience.

During 1956–1957 the psychology department at Cambridge University was a uniquely good place to observe the transformation of a subject that could be taught, and perhaps even understood in its entirety, by any one of its practitioners into mutually incomprehensible specializations. After taking Part 1 of a degree in almost any other subject one could graduate, after a single further year, allegedly equipped to begin research in any branch of‘psychology’. Oliver Zangwill had recently succeeded Sir Frederick Bartlett as Head and, with the newly appointed Larry Weiskrantz, had begun to shift the ethos of the department from Bartlettian, behaviour-based. human experimental psychology to physiological psychology and neuropsychology. But Zangwill not only taught the neuropsychology of Henry Head enlivened with descriptions of patients he saw at the National Hospital in Queen's Square, he also gave us some of the little Social Psychology we got. Most junior staff had been appointed by Bartlett, and his genial signed photograph hung on almost all of their office walls. The other Zeitpoltergeist haunting grubby beige corridors was Bartlett's charismatic and untimely deceased protégé, Kenneth Craik, who had been appointed by Bartlett as first director of the Medical Research Council's Applied Psychology Research Unit (MRC APRU). Applications of psychology were key to Bartlett's legacy. He is currently most famous for his contributions to the theory of memory and of social psychology, but it was leadership in military applications, both on submarine hydrophone detection systems during World War I and again as a national coordinator of applied psychological research in World War II, that won recognition from practitioners of the established sciences, credibility with government, and representation on (p.220)committees of UK funding agencies. The wartime colleagues he appointed to lectureships exemplified the Cambridge ethos that, at a pinch, a competent scientist could teach all of his or her subject—and perhaps much of any other. So William Hick, a practising psychiatrist, gave disturbing seminars illustrated by scratchy tape—recordings of confrontations with cowed patients protesting that they really had seen Red Indians in suburban Cambridge—and also intimidating lectures on applications to psychology of Shannon and Weaver's (1947) information-processing theory, to which he had made a seminal personal contribution (Hick 1952). Derek Russell Davis, another psychiatrist, had worked with Kenneth Craik on a primitive aircraft simulator, the ‘Cambridge Cockpit’, providing some of the first contributions to understanding the effects of fatigue on pilot performance and of the effects of life stress on human errors in driving and mines. His clinical lectures and case demonstrations were tranquil. Richard Gregory's humour, brilliance, track record of applied research in submarines, and remarkable contributions to Vision Science are evident from his chapter in this book (see Chapter 9). His lectures were, in effect, notes for the famous Eye and Brain, now in its fifth edition (Gregory 1997).

“C.” Grindley, who had begun his research career as a physicist, was a paradigmatic example of the Cambridge idea of universal academic deployment. While working in Bristol as a colleague of Paul Dirac, he had also given lectures in psychology to prisoners in Dartmoor. His career in the Cambridge Cavendish Laboratory was curtailed abruptly by Ernest Rutherford with the advice: ‘Grindley, if I were you I should take up Psychology’. Obediently he crossed Downing Street to work with Lord Adrian, in Physiology, on human detection thresholds for rates of changes in tactual pressure and, perhaps as a sequel, to mutual mapping of the sensitivity of the entire human male skin surface with Oliver Zangwill (allegedly) to the alarm and distress of an intruding cleaning lady. He had also anticipated Skinner's work on operant conditioning by training a guinea-pig to turn its head appropriately for rewards of carrot. During the war he had introduced effective reforms of haphazard processes of aircrew selection. His colleagues maintained that the hospitality of RAF messes launched his career as an irredeemable, gentle, and charming alcoholic whose lectures on Animal Behaviour were interrupted by strange entrapments, such as failures to break out of physical demonstrations of how Ant Lions delve their burrows by persistent circling and by problems in distinguishing between cigarettes and sticks of chalk. In spite of these difficulties, with the support of Valerie Townsend he continued to publish, well regarded papers throughout the 1960s (e.g. Grindley and Townsend 1966, 1968), and before his retirement he published the first demonstration that humans can control attentional (p.221) sensitivity in their visual fields without making eye movements, a groundbreaking study that inspired elegant work by Mike Posner and many others (Grindley and Townsend 1970—then known as the Big G and T paper).

My time as a research student from 1957 to 1961 was made more tolerable by long afternoons in his company following his solitary lunches in the Castle pub on Regent Street. This was an education in the exhilaration of playful scientific speculation, in how formidable cleverness can be coupled with extreme sweetness and gentleness of nature, and in the grace and gaiety of his self-mockery and disdain for vanity, self-regard, or pomposity.

Graduating in 1957,I was certain that psychology was the only thing that I wanted to do but, with a mediocre second-class degree, this seemed impossible. Anne Taylor, now Anne Treisman, then aptly nicknamed ‘needle’ by Richard Gregory, gained her predicted outstanding First and went to a research studentship in Oxford. I mooched around Cambridge sandpapering trucks for painting at Marshall's Airport, swallowing chagrin and wondering what next. As on very many occasions in my career, John Morton rescued me from depression. Oliver Zangwill had recognized his undergraduate brilliance and negotiated a Department of Scientific and Industrial (DSIR) research studentship to keep him at Cambridge. However, because he was otherwise occupied producing review sketches, his degree class did not allow him to take this up. Zangwill rapidly found him alternative support with Margaret Vernon at Reading, puzzled to find another candidate, discovered that I was still hanging around and made the offer. I eagerly accepted and was told that I would be supervised by Donald Broadbent, who had been my sole undergraduate tutor, and set the 24 weekly essays that completely summarized my pawky knowledge of Psychology. Oliver's encouragement that I ‘obviously got on well’ with Donald was a pleasant surprise. Brilliant mentors can be exhilarating, but also baffling and sometimes painfully abrasive. A whine that I would like to work on schizophrenia with the gentler and vaguer Derek Russell-Davis was ignored.

I was luckier than I could then understand. During 1957–1958, Donald was completing his masterpiece, Perception and Communication (Broadbent 1958), running an exciting personal research programme, directing applied work for the RAF, and assuming directorship of the MRC APRU. Consequently our encounters during my first research year were few, brief, and decisive. At the first he gave me a research project based on recent demonstrations that people can recognize words as members of particular semantic categories (typically as ‘obscene’ or ‘threatening’), even when they are too briefly presented to be identified as individual words, such as ‘fellatio’ or ‘pain’. Donald suggested that people might be taught to make categorical discriminations (p.222) between emotionally neutral items, such as letters and digits below the item recognition threshold. This required a tachistoscope, but none was available. Richard Gregory thought up a splendidly unfeasible device synchronizing a gravity-driven shutter and a variable-sized empty sector in a rotating aluminium disc. My second and third meetings with Donald were brief because I could only report that the contraption did not yet work. At the fourth I reported that it now sometimes worked but that my results made no point. Donald threatened to terminate my scholarship. Thus braced, I abandoned the machine and reversed the problem by timing people as they sorted visiting cards marked with digits and capital letters of the alphabet into randomly assigned categories. The number of items in each of two categories made no difference to sorting times, but with four or more categories times slowed as a multiplicative function of category number and size. At our fifth meeting Donald remarked that I was ‘getting along like a house on fire’. The idiom was alarmingly unfamiliar, but proved to be a reprieve. Reassuringly the results were publishable (Rabbitt 1959), and Donald and I contentedly avoided each other for two years until he rang to ask whether I had yet written a thesis and, if so, what its title might be.

My career in the study of ‘Reaction Times’, had begun, but so, insidiously, had another in Cognitive Gerontology. I ran out of young adult acquaintances who would tolerate my dull experiments and began to solicit affable, late middle-aged porters around the Downing Street Science site. I was unsurprised that they were slower than young adults, but distressed that they were never indifferent to category size (Rabbitt 1965). I was saved by Alan Welford—a very different personality and intellectual influence. Possibly because of his second job as Chaplain of St John's College, his lectures were, for that department at that time, uniquely well prepared and tightly organized, but also somnolently sonorous. With kindness, intelligence and patience he taught me to be intrigued rather than dismayed that there were marked individual differences in performance that were entirely neglected by current models for decision processes. Oriented by this advice and by the appearance of Ageing and Human Skill, his account of work of the Nuffield Age Research Unit that he had directed (Welford 1958), I began to be impressed—by individual differences and by the extreme variability of human cognitive performance.

In 1962 I was hired by Reuben Conrad as staff member at the MRC APRU, and during my first year spent eight hours a day teaching postmen (higher grade) to use a ten-finger keyboard intended for letter-sorting machines. As many of these were middle aged and some, unfortunately, could not be trained because they had fewer than ten fingers, resources and time were available for modest work on age comparisons and human variability. (p.223) This supported further conversations with Alan Welford and with his sabbatical visitor Jim Birren, who kindly offered me a temporary post-doctoral position for 1963–1964 in his age research unit at the National Institutes of Health (NIH), Bethesda, Maryland. There I encountered a much richer gerontological tradition.

The agenda of Welford's Nuffield Unit had been to improve the efficiency of the British workforce by exploring how, and if at all possible why, even those skills that people have brought to remarkable levels during their working lifetimes deteriorate as they age. The focus on applications echoed the wartime work of Bartlett and his colleagues, and its current extensions at the MRC APRU. In the 1950s, without computers, it was impossible adequately to simulate the tasks and systems studied by the simple laboratory paradigms on which the limited models then current in academic human experimental psychology had been based. The obligation to find some ways in which to help humans to cope with complicated equipment and scenarios undermined credulity in the disingenuously simple academic ‘models’ of human performance of that time, but usually left investigators only with specific recommendations for improvement of particular systems but with no idea how these could generalize to plausible descriptions of human mental processes. Like Broadbent's (1958) contemporary masterpiece, Perception and Communication, Welford's (1958) Ageing and Human Skill describes how he and his many talented associates used problems posed by complex tasks to transcend the limitations of the current psychology by inventing an entire new language for modelling human performance. For example, E. R. F. W. (Ted) Crossman was the first to suggest ‘neural random noise’ as an explanation for age-related slowing of decisions. Although he did not work in Welford's unit, Richard Gregory extended this idea to ageing sensory systems. Harry Kay produced seminal work on age and motor skills. Jacek Szafran moved to the USA where he continued applied work with ageing aircrew. The MRC recognized the success of the unit as an engine for useful research and a school for scientific talent, and offered to fund its continuation in Cambridge with Welford as director. Zangwill declined to house it. Welford declined to leave Cambridge to direct it elsewhere. The MRC funded two units to research the effects of age on workforce effectiveness and mobility, directed by Alaistair Heron in Liverpool and by Huwell Murrell in Bristol. In 1966 Welford did at length leave Cambridge for Adelaide, South Australia, where he built up a successful department with a much wider research agenda than cognitive gerontology.

Jim Birren's research was driven by broader concerns than industrial or military applications. Influenced by the psychophysical tradition of‘Smitty’ Stevens he saw sensory psychophysics (Birren 1959) and the investigation of (p.224) putatively ‘elementary’ processes such as times to make simple choices (reaction times; RTs) as tools to understand age-related changes in neural processing (Birren 1956, 1965). During 1963–1964 his lab supported work on ageing rats as well as his personal work on human hand tremor, heart rate, psychophysical judgements, and, increasingly, on RTs, with past and current collaborators such as Jack Botwinnick, Joseph Brinley, and Joe Robbins. In particular Brinley (1965) has the distinction of a clever insight that thoroughly confused discussions of age and information processing for the next forty years. He found that, independently of task difficulty, mean RT for groups of older people can be accurately estimated by multiplying means of groups of younger adults by a simple constant (from 1.1 to 1.5 depending on the difference in group ages). Birren was eager to regard this ‘general slowing’ across tasks as evidence that RTs provided a unique insight into the effects of age on the central nervous system, and it became his obsession from the 1960s (Birren 1965) to the early 1980s (Birren 1979; Birren et al. 1980). A brilliant meta-analysis by John Cerella (1985) validated, extended, and formalized Brinley's discovery, and Tim Salthouse made much deeper new experiments showing that even on memory tasks, in which decision speed is neither measured nor constrained, part of the variance between individuals that is associated with differences in their ages disappears, or is greatly attenuated, when individual variance in their decision times is also taken into account. Tim concluded that slowing of information processing in the ageing brain directly causes all observed effects of age on mental abilities (Salthouse 1985, 1991, 1996).

Salthouse's ‘general slowing’ theory was the first attempt at an overall theory of cognitive ageing. It was a necessary and timely provocation because it exposed the major weakness of research in the psychology of ageing: thoughtlessly atheoretical use of each newly emerging experimental paradigm of ‘mainstream’ cognitive psychology to compare older and younger groups. However, the suggestion that ‘general slowing’ may not just be one aspect of changes in mental abilities but actually the common functional cause of all age-related changes in mental abilities seemed to leave nothing left to explain and to block further research. In some it caused irritable apathy for a variety of logical and empirical reasons tediously rehearsed elsewhere (Rabbitt 1996, 1999, 2002). Others found more constructive responses. Gus Craik and his associates quietly continued investigating aspects of cognitive functions for which general slowing can have no explanatory value, such as the nature and relative efficiency of different kinds of memory representation and encoding process (see Chapter 6 of this volume). A more direct challenge was doggedly to seek exceptions to the annoyingly ubiquitous ‘Brinley-Cerella plot’ (p.225) (Myerson and Adams 1998; Myerson et al. 1990, 1992). Another was to seek alternative models for the functional basis of age differences in decision speeds. Hasher and Zacks and their associates proposed that differences in decision times between young and older adults could be better understood in terms of qualitative rather than quantitative differences between highly practised ‘automatic’ and novice or ‘controlled’ processing (Hasher and Zacks 1979), to which Shiffrin and Schneider (e.g. 1984) had first drawn attention. A different move was to try to find tasks on which age differences were greater than ‘general slowing’ might predict. Hasher, Zacks, and their associates, intrigued by accumulating evidence that age affects frontal and temporal cortex earlier than other areas of the brain (Albert 1993; Gur et al. 1987; Haug and Eggers 1991), began to speculate, like later authors such as West (1996), that age might affect performance on ‘frontal and executive’ tasks earlier than on others. Hasher, Zacks, and associates were chiefly concerned with ‘inhibition’ of unwanted information in perception and memory as a key example of executive control dependent on frontal lobe integrity (e.g. Hasher and Zacks 1988; Hasher et al. 1997, 1991; Kane et al. 1994).

On the whole, results of age comparisons on ‘frontal’ tasks have been mixed, although more recent work, requiring complex and sensitive statistical analyses, suggests that while age indeed may affect frontal tasks more than others, differences are very small and the amount of age-related variance accounted for by general slowing is at least an order of magnitude greater than that associated with changes in frontal or executive function (e.g. Verhaegen and Barsak 2005). A more direct strategy was to question the statistical validity of analyses using ‘Brinley functions’ (e.g. Ratcliff et al. 2000) and to develop and test more rigorous models for analysing decision processes (Ratcliffe et al. 1999). A further strategy was to illustrate that differences in mean decision times are only part of what has to be explained because increases in RT means can be largely accounted for in terms of increasing moment to moment variability (Hultsch et al. 2000, 2002; Rabbitt et al.2001b).

These various endeavours eroded confidence in ‘general slowing theory’ but, although it was clearly logically unsatisfactory, the theory could not be disproved by behavioural evidence alone. As with all powerful scientific paradigms, comfortable disbelief required new kinds of data.

Throughout the twentieth century a reckless, but necessary and sustaining, optimism for human experimental psychologists was that models based entirely on behavioural comparisons would eventually provide useful guides to neural functions supporting cognition. Brain imaging has abruptly brought about the changes that other authors in this book have termed ‘the neuropsychological revolution’ and, inevitably, show up the naivety of late twentieth (p.226) century models. The careful work of Salthouse and his colleagues convincingly showed that behavioural data alone do not disassociate the effects of age from slowing of information-processing speed on most cognitive tasks. But recent work has shown that gross brain changes such reduced cerebral blood flow, age-associated atrophy, and white matter lesions account for all of the differences in speed between people that is associated with differences in their ages; they account for little or no differences in intelligence or in performance on frontal tasks (e.g. Rabbitt et al. 2008c).

Birren's preoccupation with RTs was moderated by his location in the NIH, then possibly the largest medical research centre in the world. This exposed him to discussions of how biological factors, health, and demographics determine ‘successful’ old age (crf. Birren et al. 1963). Birren's extraordinary warmth, congeniality, and energy won him collaborations across many disciplines. His resulting broad interests are evident from the editorial work that he has sustained over his lifetime, particularly the Handbooks of Human Ageing, which tracked and shaped directions of work in the field. His medical and epidemiological interests followed his friend and mentor, Nathan Shock, who had, in Baltimore, initiated the first, and now the longest, longitudinal study of age-related changes, with a main emphasis on the collection of data on pathological and physiological changes. During the 1960s and 1970s the methodology of longitudinal studies was markedly improved by a series of theoretical contributions by Paul Baltes, John Nesselroade, and Werner Schaie (e.g. Baltes 1968; Baltes and Nesselroade 1970; Baltes et al. 1979; Schaie 1965; Schaie and Baltes 1975), and by the steady output of analyses from Schaie's Seattle study (Schaie 1983) and from many other large and well -conducted medically oriented studies. Growing interest in these studies as a way of obtaining information about ways to cope with the surge in the proportions of elderly in the populations of industrialized nations revived the interest of funding agencies in the UK in ageing research.

During the late 1960s, the UK MRC had become disappointed with the Liverpool Age Research Unit, in spite of Alaistair Heron and Sheila Chown's (1967) large, excellent, but short longitudinal study based on volunteers from Merseyside industries. The MRC closed the Liverpool and Bristol Units, and funding for cognitive ageing became hard to obtain. In 1968 I moved to Oxford, where I could only find support for work on RTs and selective attention in young adults until 1979 when a revival of public interest in the looming problems of ageing populations allowed a successful bid to begin a longitudinal study. A brief start in Oxford was curtailed by a move to Durham in 1982, but funding continued there for a study based in Newcastle with a parallel cohort in Manchester when I moved there in 1984 to direct the Age and (p.227) Cognitive Performance Research Centre. This study extended until 2004, supported by a patchwork of grants by the UK Economic and Social Research Council (ESRC), MRC, European Commission, Wellcome Trust, Unilever plc, and the University of Manchester.

Perhaps the broadest questions about cognitive changes that longitudinal data can answer are: When do changes first appear? How fast do they then proceed? Do all cognitive abilities change at the same rates or do some decline earlier than others? Do individuals show different rates of cognitive changes? Do individuals show different patterns of cognitive changes? Can we identify factors that accelerate or slow rates of cognitive change or that produce some patterns of changes rather than others? During the 1980s and 1990s it became clear that none of these questions could be approached without development of new statistical methodologies. For example, in the 1960s Baltes and Schaie had noted, as an inconvenient theoretical possibility, that if participants take the same or similar cognitive tests, even at long intervals, improvements with practice may disguise declines of performance with age. Ways of separating these practice effects from age-related declines were not then available, so, like all my colleagues, I was glad to argue that they must be negligible if tests were re-administered at intervals as long as four years. Our accumulating data confuted this expedient optimism and further suggested that older and less able individuals might gain less from practice and so appear to decline faster than the young and more able. We could confidently publish only cross-sectional analyses, comparing people of different ages, until Peter Diggle came to our aid (Rabbitt et al. 2001a, 2004). Mary Lunn showed us how to confirm our guesses that improvements with practice were smaller in the old and less able (Rabbitt et al.2008a), and also that true rates of decline with age, or with the onset of pathology, cannot be correctly estimated unless dropouts and deaths are also taken into consideration (Rabbitt et al. 2005). It becomes clear that rates of decline accelerate during the eight years preceding death or dropout, and that healthier individuals show very slight changes even during their eighth decade (Rabbitt et al. 2008b).

This general picture is consistent with experiments by investigators adopting the ‘medical model’, that most of the variance in amounts of decline in individuals can be attributed to differences in general health, and in the burdens of pathologies that accumulate with age (e.g. Houx 1991; Knottmerus and Jolles 1998; van Boxtel et al.1998). Recent improvements in brain imaging now show both how patterned age-related changes in brain mass and structure are correlated with cognitive changes and, even more interestingly, how age appears to affect the number and kinds of brain area involved in different, particular kinds of task (Raz 2000; Raz et al.1998) The aetiology of these (p.228) changes can be related to the increasing general burden of pathologies and to specific effects of particular conditions. It now seems no longer worthwhile to carry out large and prolonged longitudinal studies that collect only behavioural data. Studies are useful only when they include neurophysiological and neuropsychological data, particularly imaging data, detailed information about health and the number, kinds, and progress of pathologies, and focused batteries of cognitive tests. Such studies need involve only small numbers of carefully selected and thoroughly documented individuals over relatively short periods of time—five rather than twenty years or longer. After spending half a working lifetime on a large behavioural study, I am delighted that my younger colleagues will not have to go through all that again.