The objectives of this chapter were to provide a coarse review of the literature on age changes in WM across the lifespan and to show how similar were the hypotheses brought forward in cognitive development and in cognitive aging. A long-standing issue in the developmental literature has been to ask whether there are general mechanisms driving cognitive change with age. If such mechanisms are truly general, they should be operational across the lifespan. Even though change in older adulthood is not a full mirror of what happens in childhood, the same processes should be at play, while their respective weight and interplay may vary. A lack of full symmetry between childhood and older adulthood inevitably points to the existence of several underlying processes. It is therefore crucial to adopt a multivariate approach to test both whether the hypothesized processes are general and whether several processes are at work.
The study briefly reported in this chapter was a first attempt on our part to provide preliminary empirical backup to the proposal that both processing speed and inhibition influence age differences in WM across the lifespan. It was not possible to study age in a continuous manner and the study was limited to a comparison of children aged 8 to 12 years of age, young adults, and older adults. The young-adult group was considered as an anchoring group to which compare children on the one hand and older adults on the other hand; the focus was therefore not placed on a comparison of children and older adults. The first question raised was whether all three constructs (or tasks supposed to tap this construct) undergo change with age. The response was clearly affirmative: age differences were significant in all tasks, as well as in the summary scores used in the present chapter. I emphasize, however, that change was steepest for processing speed; the difference between children and young adults, as well as between young and older adults, amounted to several standard deviations, whereas it was much smaller in inhibition tasks. This could, of course, also be due to a greater homogeneity of the young-adult sample in some types of tasks. Nevertheless, the effect of age (variance accounted for by age differences) was much larger for speed and working memory than for inhibition. Note also that the effect of age on inhibition was larger in older adults than in children, relative to young adults.
The second question raised dealt with the generality of each of the three constructs. Correlations among WM tasks, on the one hand, and among processing-speed tasks, on the other hand, were high; a single latent variable for each of these two constructs was easily defined. In contrast, correlations among the inhibition tasks were much lower, and a latent variable accounted for very little variance. It is noteworthy that a model with a single latent variable could nevertheless yield an adequate fit, because other researchers (e.g., Park et al., 1996; Salthouse & Meinz, 1995) encountered difficulties when attempting to include inhibition tasks in confirmatory-factor analyses or in structural-equation models. It should be stressed that inhibition scores reflected only the amount of interference and inhibition, controlling for most other processes that are probably at work in the tasks. In contrast, WM or processing-speed scores were based on raw performances in tasks that certainly also require other processes, such as strategical ones; it is therefore not surprising that they correlated more highly. Nevertheless, the hypothetical generality of inhibition did not receive much empirical support; further studies are obviously necessary before accepting or rejecting a hypothesis according to which inhibition is a general mechanism, all the more so as only two tasks (three in some of the analyses) could be retained.
The third question concerned the possibility that both inhibition and processing speed act as mediating variables between age and WM. Commonality analyses and structural-equation modeling showed that even though processing speed exerted a larger influence, the role of inhibition was not negligible. However, inhibition apparently played a more important role in older adulthood than in childhood. Once again, all conclusions pertaining to inhibition have to be taken with extreme caution and will require replication. Most interesting for our hypotheses was the finding that the same model applied to both the group composed of children and young adults and the group composed of young and older adults, but that the numerical values of the parameters were different. This seems to imply, in line with our hypotheses, that similar processes are at work as far as age changes are concerned, but that their interplay does not remain identical across the lifespan. Moreover, the relative role of inhibition was apparently somewhat stronger in older adulthood.
Because of the complexity of the questions raised, the present study should be considered only as the beginning of a research agenda. A number of issues remain unsolved. First, the meaning and the generality of inhibition need to be clarified by using a greater number of tasks simultaneously. Even though the construct of inhibition is proposed by many researchers as a mediating variable, it is perhaps less general than processing speed or working memory (both of these constructs also requiring further theoretical clarification). A more extensive study is now under way in our laboratory, using more inhibition tasks. Second, other studies using a similar design should be conducted. Third, groups with greater age continuity than the groups studied here should be studied. And last but not least, more detailed theoretical analyses of inhibition tasks should be conducted in order to formulate more precise hypotheses as to which tasks are likely to tap similar processes, and therefore to correlate, and which tasks tap different processes.
Was this article helpful?