Cognitive reserve (CR) is the name given to the hypothesized construct that serves to buffer cognitive decline in normal and pathological aging. As described in the present article it “helps to explain the mismatch between expected cognitive decline and observed maintenance of cognitive functioning in older age.” Presumably the concept has a neural basis (or bases) such as well-maintained frontal functioning or superior integrity of cognitive control networks, although this has not yet been established definitively. Lifestyle factors that apparently promote CR include education, high-status occupation and high levels of social, intellectual and physical activity. In turn, individuals with supposedly high levels of CR will exhibit higher levels of various cognitive functions (e.g., memory, attention, thinking) than might be expected from the general state of their brains (e.g. amount of atrophy and buildup of amyloid plaques). As the authors state, it is not yet possible to measure CR directly, so various ‘proxy’ measures have been used to estimate individual differences in the concept. Given the growing importance of CR to understanding cognitive decline, the field badly needs some ‘gold standard’ to serve as a reliable marker, and to assess such a marker’s relations to other cognitive measures, to lifestyle factors and ultimately to indices of neural integrity.
The present article suggests that improvements over 10 trials on the digit symbol substitution test (DSST) may serve as such a gold standard marker of CR, and the article reports correlations between individual differences on this measure and a variety of other cognitive tasks and psychological factors. In the first group the proposed index of CR correlates significantly with word list learning, with the difference between trail-making tests A and B, with matrix reasoning, and marginally with Stroop interference. This pattern confirms previous findings that DSST improvement relates to good performance on memory and executive function tasks and serves to further characterize the nature of CR as a buffering agent. Surprisingly, however, the present study found no relations between the authors’ index of CR and various measures of quality of life and leisure activities. The authors suggest that this may be because all of their participants had high levels of education and reported high levels of cognitive leisure activities, but the absence of a relationship is somewhat worrying. The main question, after all, is how do we define CR, and that has to be in terms of unexpectedly slow cognitive decline in old age, either relative to age-related norms or to the state of the person’s brain. So – given that quality of life and quality of leisure activities (physical, social and cognitive) have been shown to be major contributors to CR “in real life,” the alternative possibility is that the current DSST index is not in fact a very valid measure.
My main criticism of the article is that this alternative is not really considered. Other than this major point I think the study and the article were well done. The sample of participants was large and well chosen, the tests used were sensible and covered the territory well, the analyses and results are very believable. In general terms, the study adds to the literature on CR, and my only major suggestion is that the authors acknowledge that they have a problem in that their index of CR does not relate to the main established contributors to CR in real life. I like the idea of a dynamic index in a testing-the-limits framework but its validity is not “confirmed” (p. 6) by this study in my opinion. In my view the “confirming” statements on p. 6 should be stated more tentatively. The study does “confirm” that the DSST gain index is related to memory and executive processing, but that’s not the same as being related to CR in the real-life clinical setting. The proof of that would lie in a study in which older adults of say 70 years were given the DSST gain test and then followed for 10-15 years to see if the index predicted protection from dementia. This would be a good study for the authors to consider!
Other than this main point, I have some minor issues: First, I think that the testing-the-limits notion needs more explanation (briefly) for readers unfamiliar with the concept. It would also be useful to explain why a dynamic index might be more valid than static indexes. Second, I think I understand that the standard German version of the DSST was used in the battery of tests, and that a new (different) version was made up for the CR index – but also that the same new version was used in all 10 trials of the CR index? If this is correct, perhaps the details could be spelled out a little more clearly. Third, I was not clear what the term “population maximum” refers to, and lastly the abscissa legend in Figure 1 should be “Trial.”
In general I found this to be an interesting study with potentially important results and implications for clinical testing given further evidence that the proposed CR index does indeed relate to real life observations of CR.