We applaud Valian's (Reference Valian2014) thoughtful analysis of the cross-sectional and longitudinal studies that have thus far contributed to our knowledge about the role of bilingualism in cognitive aging. In evaluating the literature as summarized by Valian, we think it is useful to distinguish between the following four research questions and representative analytic approaches: (1) cross-sectional associations between bilingualism and executive function (e.g., regression), (2) longitudinal associations between bilingualism and change in executive function (e.g., growth curve modeling), (3) bilingualism as a predictor of dementia incidence (e.g., time-to-event analysis), and (4) bilingualism as a source of cognitive reserve (e.g., comparisons of brain pathology between bilinguals and monolinguals matched on cognitive performance).

We agree with Valian's conclusion that most studies show positive cross-sectional associations between bilingualism and executive functioning in older adults – research question (1), above. We do not agree with her use of the terms “benefit” or “protective role” of bilingualism, even in retrospective studies, because these terms imply causation where there is only association.

We agree with Valian's conclusion that when prospective data are available, research shows no relationship to change in executive function or dementia incidence (Zahodne et al., Reference Zahodne, Schofield, Farrell, Stern and Manly2014; Crane et al., Reference Crane, Gibbons, Arani, Nguyen, Rhoads, McCurry, Launer, Masaki and White2009; Sanders, Hall, Katz & Lipton, Reference Sanders, Hall, Katz and Lipton2012; Yeung, St. John, Menec & Tyas, in press). Short of randomized studies, we feel that this is the best evidence regarding research questions (2) and (3), above. We note that the one study (Kavé, Eyal, Shorek & Cohen-Mansfield, Reference Kavé, Eyal, Shorek and Cohen-Mansfield2008) Valian cites, as an exception to this pattern of null results, only conducted cross-sectional analyses – despite having three time-points of data – and did not examine rates of cognitive change or dementia incidence. Thus, we feel this study can only be used as evidence for research question (1). We believe it is essential for researchers to distinguish between cognitive level and cognitive change, as the former reflects the confluence of a lifetime of experiences rather than the exclusive underlying process of interest: aging or age-related neurodegeneration. For example, educational attainment is clearly associated with higher cognitive level but not rates of cognitive decline in modern studies of older adults (Zahodne et al., Reference Zahodne, Glymour, Sparks, Bontempo, Dixon, MacDonald and Manly2011). We argue that it is not necessarily true that experiences that are associated with better executive function will also attenuate cognitive decline or protect against dementia incidence. Prospective longitudinal studies, with direct standardized observation of cognitive performance and determination of dementia onset, should be the gold standard. Such studies require measures that are sensitive to change over time, can assess performance over a wide range of abilities, and are not biased by cultural variation. Retrospective studies often rely in part or in whole on informant reports of functional decline without considering demographic and cultural influences on these reports among families or caregivers of monolingual versus bilingual people.

We also point out that results of studies addressing research question (4) – i.e., about bilingualism as a source of cognitive reserve – are contradictory. For example, one study (Schweizer et al., Reference Schweizer, Ware, Fischer, Craik and Bialystok2012) found more brain atrophy on clinical CT in bilinguals with dementia, compared to monolinguals with equivalent cognitive performance. This is in line with the theory of cognitive reserve. In contrast, another study (Luk, Bialystok, Craik & Grady, Reference Luk, Bialystok, Craik and Grady2011) found higher white matter integrity in bilingual older adults, compared to monolinguals with equivalent executive functioning – in direct opposition to the theory of cognitive reserve.

A crucial point in Valian's review is the possibility that in certain contexts, bilingual older adults may be antecedently different from monolingual older adults. In isolating and quantifying the independent effect of bilingualism on executive function, cognitive decline or dementia incidence, we must strive to disentangle bilingualism from other related variables such as general intellectual ability and socioeconomic status. There has not been a true experiment in which individuals are randomized to bilingualism or not. Nor has there been a study in which instrumental variables, or natural experiments, were used to approximate randomization (Glymour, Kawachi, Jencks & Berkman, Reference Glymour, Kawachi, Jencks and Berkman2008). While Valian states that individuals in some communities are “born into bilingualism,” most of the older adult literature on bilingualism comes from regions in which bilingualism is confounded by a variety of variables such as socioeconomic status, education, and/or immigration history. Interestingly, regions with large numbers of people “born into bilingualism” do not report lower rates of dementia compared to other primarily monolingual regions with comparable environments (Canadian Study of Health and Aging Working Group, 1994).

Finally, while Valian defines bilingualism as “knowledge of any number of languages beyond one,” we suggest that research on bilingualism and cognitive aging will benefit when studies are able to examine the skills and experiences of bilinguals on a continuum. This approach may also mitigate the problem of never being able to fully account for confounders in observational studies. Similarly, future studies should characterize details about when a second language was attained and under what circumstances. Historical changes in immigration policies and conditions in the native country may be powerful tools to understand the conditions and experiences that travel along with bilingualism.

In conclusion, it is our position that until we are able to isolate the effects of bilingualism from other, significant confounds, current empirical evidence favors Valian's logical possibility 2 (i.e., that there is no cognitive benefit of bilingualism) with respect to cognitive decline with age and dementia incidence.