| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
RESEARCH ARTICLE |
National Institute on Aging, National Institutes of Health, Bethesda, Maryland.
Address correspondence to Jeffrey W. Elias, PhD, Dean's Office, UC Davis School of Medicine, 2921 Stockton Blvd., Suite 1400, Sacramento, CA 95817. E-mail: jeffrey.elias{at}ucdmc.ucdavis.edu
 |
Abstract |
|---|
THIS special issue of the Journal of Gerontology: Psychological Sciences is devoted to further defining the future for cognitive intervention/training research. The special issue developed out of a symposium on cognitive training for older adults sponsored by the National Institute on Aging. In March 2004, 11 leading cognitive researchers and colleagues met in Bethesda, Maryland, to address the scientific and methodological issues that define this domain of cognitive research. The general goal of the symposium was to address the following: (a) the state of the art in development of cognitive interventions and the readiness to turn research findings into practice; (b) key research and methodological issues that the consumer, practitioner, and the researcher should be aware of when considering the concept of cognitive intervention; and (c) guidelines for judging cognitive intervention as successful. These are major issues to address. One collection of writings cannot fully address all of the concerns, but it can help move the field forward.
Underlying the examination of each of these issues and the future of intervention/training is the contemporary context and the background against which cognitive research is developed. To quote from The Aging Mind (Stern & Carstensen, 2000
, p. 21), "To understand cognitive functioning, it is necessary to pay attention to the context of cognition. This context includes not only evolutionary and biological constraints and affordances, but the cultures in which minds reside." The evolving context for cognitive research serves as a rich background for intervention/training research, and there are specific aspects of cognitive research that are of particular importance to cognitive intervention and training. One of the contexts for intervention research is that plasticity, life-course malleability, and compensation are well-recognized concepts of life-span development (Settersten, 1999) that fit well with the notion that cognitive training and interventions in middle to late adulthood can offset the potential for cognitive decline in later adulthood.
|
COHORT EFFECTS AS BACKGROUND TO COGNITIVE TRAINING AND INTERVENTION |
|---|
; Schaie & Hofer, 2001; Shock et al., 1984) illustrates that there are cohort effects for cognition that are particularly important with respect to interpreting cross-sectional and longitudinal studies. Although not normally thought of as intervention research, these studies demonstrate that there are potential powerful cohort effects against which contemporary intervention studies take place.
Likewise, researchers infrequently think of normative data sets as representations of intervention data, but they can be viewed in that context. As shown in Table 1 (Au et al., 2004), there were significant improvements in cognition for individuals in the Framingham Offspring Study (N = 1,841; 1,063 women) relative to those in the original parent study (N = 1,805; 988 women), resulting in the need for new norms for the offspring. As with other studies of cohort effects, researchers attribute normative differences between cohorts to historical improvements in methods (possibly delivery) of education and improvements in health care (potentially encompassing delivery and/or literacy). Investigators often employ education and health (frequently self-reported health) as correlates or covariates in most cognitive studies, thereby tacitly recognizing the powerful intervention aspects of education and health. Researchers do not know the specific pathways by which an individual's life history influences cohort health and education (Cagney & Lauderdale, 2002; Farmer, Kittner, Rae, Bartko, & Regier, 1995; Karp et al., 2004; Le Carret et al., 2003). Nor do we know if there are inflection or plateau points where education or health contributes little variance to sustained cognitive performance. There is always the potential that at some point cognitive performance, health, and education have reciprocal effects. Given the encompassing nature of education and health, it is tempting to suggest that the influences on cognition are nonspecific. Nevertheless, despite the importance of education and health to cognition, one cannot specifically conclude that all waters rise in cognitive performance with higher levels of education and better health (Arbuckle, Maag, Pushkar, & Chaikelson, 1998). We could say much more about the variable of education than we could cover in this brief overview. A single number or range of numbers can parsimoniously represent education, but in reality this belies the qualitative complexity of a variable that likely not only changes with cohort, but would be more accurately assessed against a background of regional, cultural, and cohort/historical potential.
|
|
OCCUPATION AS CONTEXT FOR COGNITION |
|---|
; Karp et al., 2004; Le Carret et al., 2003). Assigning a value to occupational status is difficult (particularly if there have been several occupations), although researchers occasionally use socioeconomic status or years of education as a proxy for occupational context (Koster et al., 2005). Very often investigators sum education with other measures (e.g., U.S. Census Bureau codes and standard occupational codes) to determine socioeconomic status.
A contemporary and potentially major organizing concept for the cognitive influence of work is cognitive complexity. This concept, and its potential status as a recognized covariate for cognitive research, shows significant variability in definition throughout the literature. Complexity per se is not the only important variable in the work complexity concept. Frequently referenced research on occupation by Schooler, Mulatu, and Oates (1999, 2004) embraces a significant degree of personal control and self-directedness relative to the stimulus and demand characteristics of the environment. An environment that requires more decisions of a less well-defined nature is more complex, but if the complexity is to be stimulating there has to be reward in the management of the complexity. This reflects the ability to cope by meeting the demands of the environment (Pearlin & Schooler, 1978).
Supporting the importance of control relative to domains of everyday functioning is research by Lachman and Weaver (1998), who, via multivariate analysis of variance, reported greater feelings of control in older groups for work, finances, and marriage, and less control within the domains of child relationships and sex life (N = 3,032, aged 25–75). The age differences favoring older individuals in perceived control for work, finances, and marriage were modest relative to the larger mean age differences in the domains of children and sex life, where there were clear trends of decreasing control from young to middle-age to older individuals. A recent study by Lachman and Andreoletti (2006) observed a modest to moderate relation, respectively, between greater perceived control over noun–list recall within middle-aged (r =.30) and older adults (r =.53), but not young adults (r =.13). The use of clustering categorization strategies to aid in better list recall showed similar age patterns of correlations with perceived control (rs = –.04,.27, and.33, for young, middle-aged, and older adults, respectively; N = 335, aged 21–83). West and Yassuda (2004) also reported a positive relation between feelings of control and cognitive performance. When presented with a 24-item shopping list, individuals with a high sense of control at baseline showed better memory performance and maintenance over trials. The researchers defined high or low control as high or low control over the factor that influenced performance most (e.g., strategy, concentration). For those who noted lower feelings of control, establishing goals improved performance; but for those who expressed feelings of better personal control, goals did not improve memory performance. Age (aged 18–22 [N = 64] and 62–80 [N = 70]) was a significant predictor of performance at baseline (R2 =.19), but when combined with control beliefs the age effect was attenuated and the overall predictability increased significantly (R2 =.50). It appears that there is potential for personal perspective on degree of personal control, goals, and potential for reward to interact with occupational complexity to make it more than just cognitive stimulation.
There is much more to know about the importance of the work place and occupation as it influences cognition. The value of occupation-related cognitive stimulation for maintenance of cognitive function also may be mediated by the ability to reduce stress during the work day or maintain energy levels. Sufficient sleep prior to work could contribute to the benefits of cognitive stimulation not only by virtue of consolidation of memory, but by ability to withstand daily stress (Stickgold & Walker, 2005). Subjection to, and ability to manage, Internet interruptions or diversions may be a contemporary cohort effect interacting with age in occupations where computer use is a central part of the occupation.
Recognition of work and occupation as a significant background against which to develop cognitive interventions is reasonable; knowing how to measure occupation influence, particularly as it contributes to maintenance of cognition in adulthood beyond the influence of education, is a daunting but important task. Occupational influence may have a general component to it, but research of Maguire and colleagues (2000) showing London taxi cab drivers (N = 16, aged 32–62 years, time as taxi cab driver 1.5–42 years) to have hippocampal volume correlated with the amount of time spent as a taxi driver (positive association with the posterior [r =.50, age-adjusted] and negative with the anterior hippocampus [r = –.60]) pointed to specific as well as general occupational influences on the brain.
|
INTERVENTIONS JUDGED RECIPROCAL WITH COGNITION |
|---|
 TOPAbstractCohort Effects as Background...Occupation as Context for... Interventions Judged Reciprocal... Allostatic Load, Genetics,...Advances in MeasurementCultural and Existential Support...References |
|---|
; Colcombe & Kramer, 2003; Kramer et al., 1999) and social activity. One can also consider the latter as engagement in overall activity. That is, some part of social activity is contributing to overall activity and vice-versa. Both constructs can also serve as a background or context against which researchers can judge the effectiveness of cognitive interventions (Craik, Byrd, & Swanson, 1987; Hill, Wahlin, Winblad, & Bäckman, 1995; Luszcz, Bryan, & Kent, 1997; Wilson et al., 2005).
The general (all cognitive waters rise) or specific intervening nature of these constructs on cognition remains to be disentangled (see Hess, 2005, for review of cognition/activity relation). Although easily understood as general constructs, both constructs show variability of operational definition and are likely to act in reciprocal fashion via a number of pathways with cognition. Individuals who function at a higher level of cognition may be energized in the domains of physical and social activity. Even though social activity and physical activity are likely to have their influence on many domains of cognition, establishing an agreed-upon appropriate control group for a focal physical activity intervention is much easier than establishing control groups for social interaction. It is much harder to derive an activity value from a social intervention and the degree of separation between the level of activity in a social intervention and normal life. Social activity intervention studies may need to reconsider the popular notion that the control group should be a group that has little similar activity. Perhaps the goal should be to compare such interventions against well-known and controlled interventions where the effect sizes are reasonably clear. The clinical trial results and the continuing data collection from such training studies as the ACTIVE study (Advanced Cognitive Training for Independent and Vital Elderly; Ball et al., 2002) provide this kind of data for the latent constructs of memory, reasoning, and speed of processing (e.g., visual search skills and identification and location of visual information in a divided-attention format). We should note that researchers may define speed of visual processing differently throughout the literature (e.g., digit symbol, letter comparison, pattern comparison; Salthouse & Ferrer-Caja, 2003), and it is not clear how much variance the differing measurements share.
There is some evidence that physical activity levels may maintain protective effects on cognition over several years. Richards, Hardy, and Wadsworth (2003) observed an association between any physical exercise at age 36 to change in memory (15-item word list administered at Test Time 1 and 2) between 43 and 53 years of age (standardized beta weight =.44 [0.01–0.87 95% confidence interval]; N = 1,119). Yaffe, Barnes, Nevitt, Lui, and Covinsky (2001) studied 5,925 women older than age 65 and measured physical activity as self-reported blocks walked per week (1 block approximate to 160 m), as total kilocalories (energy) expended per week in recreation, and as stairs climbed. Six to eight years later, with a modified Mini-Mental State Examination (Folstein, Folstein, & McHugh, 1975) as the measure of cognitive decline (3-point drop or greater), decline occurred in 17%, 18%, 22%, and 24%, respectively, of those in the highest, third, second, and lowest quartile of blocks walked per week. The authors found similar results for total kilocalories expended. Given the correlational nature of this kind of research, it is possible that potential for cognitive decline first appears as reduced activity years before cognitive measures show decline. Nevertheless, these preliminary correlative studies should encourage intervention if needed by at least middle age. The U.S. Department of Health and Human Services (2002) reported that roughly one third of persons aged 65 or older lead a sedentary lifestyle. It also reported that 54% of men and 66% of women aged 75 and older engage in no leisure-time physical activity. Alarmingly, as a potential cohort effect, the Department of Health and Human Services reported that one third of young people in Grades 9–12 do not regularly engage in vigorous physical activity and that activity levels fall off during the course of adolescence. The idea that one can "bank it" early with respect to physical activity is an attractive one and may link to the concept of building a physical–cognitive reserve relation (Scarmeas & Stern, 2003). Alternatively, or additionally, early-in-life activity levels may influence adult habits and lifestyles.
If there are effects of activity on cognition beyond the reduction of potential health-risk factors, researchers will need to establish the degree and duration of activity (Kramer, Colcombe, McAuley, Scalf, & Erickson, 2005) along with the ability of age groups to achieve and sustain recommended activity levels. Neither long-term nor current social or physical activity has gained the status of basic covariate for other interventions. But if researchers can show that activity and social activity have significant effects as interventions, then eventually these factors should serve as individual difference covariates for other interventions contingent upon finding an agreed upon and representative value for such adjustments.
Dietary practices and supplements have received recent attention as factors that individuals can modify to influence cognitive function. Several studies in both animals and humans have implicated the consumption of fruits and vegetables rich in antioxidants as important for the maintenance of cognitive function in older age (Cartford, Gemma, & Bickford, 2002; Joseph et al., 1998; Kang, Ascherio, & Grodstein, 2005; Milgram et al., 2005). Furthermore, research has shown that introduction of these antioxidant-rich foods later in life reverses age-related declines in cognition (Joseph et al., 1998). In older beagles the combination of antioxidants and environmental enrichment was more effective in improving measures of learning than was either intervention alone (Milgram et al., 2005). Research has shown that hyperhomocysteinemea, which is linked to low levels (and perhaps low consumption) of B vitamins and folate, is associated with reduced cognitive performance in older age (Miller et al., 2003) and increased risk for Alzheimer's disease (Quadri et al., 2005). Thus, another matter to consider for the context or background in which to assess the effectiveness of a cognitive intervention in humans would be individual dietary intake patterns upon entry into a study and even the change in dietary intake during the course of a study. The developing interest in the way the components of food interact with genes to influence behavioral phenotypes, referred to as nutrigenomics or nutritional genemomics (Trujillo & Milner, 2006), should provide significant insight into the individual response to dietary intervention, even as it introduces greater complexity to researchers' understanding of the process.
Likewise, the pharmacological environment of the individual, often compounded in older adults due to polypharmacy, may impact cognitive function transiently or chronically and may become a factor in accurate assessment of the effectiveness of a cognitive intervention. The development of compounds for cognitive enhancement in older age carries with it considerations of global versus local impact of this category of drug on brain function and thus cognitive performance. Ramos and colleagues (2003) demonstrated this recently. Intracellular signaling mechanisms were manipulated by using drugs to stimulate protein kinase A (PKA) in rats and monkeys performing spatial working memory tasks that depended on the frontal cortex. Activation of PKA in adult rats by direct infusion of pharmacological agents into the prefrontal cortex (or in monkeys by systemic delivery of similar agents) markedly impaired cognitive behaviors mediated by the prefrontal cortex. By contrast, research has shown that activation of PKA in rodent models enhances memory consolidation (Bernabeu et al., 1997) and long-term memory function mediated by the amygdala and posterior cortical regions (Huang, Martin, & Kandel, 2000; Schafe & LeDoux, 2000). This example of opposing actions of a single drug type on different cognitive domains highlights the need for careful consideration of, and screening for, regional actions of systemically administered pharmacological cognitive interventions.
|
ALLOSTATIC LOAD, GENETICS, HEALTH INFLUENCES, INTERCEPTS, AND SLOPES |
|---|
). This physiological wear-and-tear concept is not exposure per se to biological stressors (psychological stressors are manifest in biology), but exposure relative to an overall reduction in capacity to respond (i.e., system dysregulation). This is an emerging concept with respect to definition and measurement, and advances in biomarker assessment will push this field forward. The benefit of this construct to intervention studies as a background variable is contingent upon finding an agreed-upon and representative value, for it seems unlikely that a single biomarker can be predictive, although clearly biomarkers and early assessment can have significant predictive ability for cognitive decline at different points in the life cycle (Elias et al., 2005; Seplaki, Goldman, Weinstein, & Lin, 2004).
Research will reveal how allostatic load relates to cognitive training/intervention. The degree of allostatic load might affect the initial performance status where an intervention begins (i.e., the intercept), or it might affect the rate of growth in performance over time (i.e., the slope of improvement or gain) to include growth, plateau, and subsequent rate of decline. Secondary data analysis studies from the ACTIVE trial (Ball et al., 2002) found that a preliminary diagnosis of diabetes or hypertension has its primary effect on the intercept (Kuo et al., 2006), not the slope (response to the intervention). Overall, however, there are not enough studies as yet to provide an accurate prediction of how allostatic load influences cognitive intervention.
|
ADVANCES IN MEASUREMENT |
|---|
). Designs for disentangling the nature of a cognitive construct from other cognitive constructs that might share variance are gaining recognition. Salthouse and Ferrer-Caja (2003) noted that age-related influences can be shared across what the variables within a design have in common, and, in addition, there may be unique age-related influences (e.g., age-specific influences for speed and memory). Researchers could examine the effects of interventions or training in similar fashion. These are not the typical analytical designs for intervention studies, but one can see from such designs that a targeted intervention in one cognitive domain (such as speed of visual processing) could influence the relation among other cognitive variables (e.g., memory, spatial relations) and their relation to speed as well as their relation to more distal measures of functioning (e.g., instrumental activities of daily living). One can find a good example of this general approach in research demonstrating the importance of age changes in related speed of processing to changes in other age-related cognitive processing domains (Salthouse, 1996).
A design that is attracting more attention for the assessment of interventions in developmental work is growth mixture modeling (Cuijpers, van Lier, van Straten, & Donker, 2005; Li, Duncan, Duncan, & Hops, 2001; McArdle, 2006; McArdle & Nesselroade, 2002; Muthen et al., 2002). These models allow the combination of categorical and continuous latent variables into the same model. The growth mixture concept allows heterogeneity in the sample and different individuals to belong to different subsamples, rather than assuming homogeneity in growth parameters from a single population. With respect to interventions, the model would permit identification of subpopulations that respond differently to an intervention defined in terms of the outcome itself (i.e., response to intervention). Investigators could interpret this as who benefits most, followed by attempts to trace the individual differences contributing to classification in a subpopulation of response. In cognitive interventions with older individuals there would be classification by growth (improvement) and then classification by decline from the pinnacle of growth, with the possibility that the factors contributing to each process would be quite different.
|
CULTURAL AND EXISTENTIAL SUPPORT FOR INTERVENTION RESEARCH |
|---|
|
TOPAbstractCohort Effects as Background...Occupation as Context for...Interventions Judged Reciprocal...Allostatic Load, Genetics,...Advances in MeasurementCultural and Existential Support...References |
|---|
), in discussing potential funding for the National Institute on Aging (NIA), noted, "The committee also encourages NIA to work collaboratively with other institutes and the [Centers for Disease Control and Prevention] to educate Americans about the ways they can maintain their brain as they age," and, "The committee encourages NIA to consider next steps in research to develop cognition enhancing interventions and to report back on efforts." As consumers, practitioners, and researchers, we readily embrace the general notion that cognitive systems and ways of thinking remain variably plastic across the life span and that this plasticity is manifest in performance and behavioral change. Even in the event of diagnosed pathology, the notion of finding the right cognitive intervention for slowing decline comes into play. Both the readers and authors of this special issue provide the motivation for this domain of research.
|
References |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
