As has been extensively documented, age-related cerebral and cognitive declines are not uniform but quite differentiated (see [43
] for reviews). Similarly, most of the recent literature on aging, fitness, and cognition has shown that chronic exercise would result in selective improvements in executive functions rather than general benefits [13
]. However, contemporary theoretical models of executive functions posit that the construct of EF is not uniform but encompasses fractionated executive subprocesses [2
]. It is then important to examine whether chronic exercise benefits are homogeneous or selective to some particular executive subprocesses. Using this established theoretical and methodological framework, the aim of this study was to examine differences in cognitive performance as a function of chronological age and long-term regular practice of swimming by measuring younger and older adults' information processing speed and performance on three well-known EFs, inhibition, updating, and shifting. Globally, the results revealed a clear pattern of age-related decline in all cognitive functions. On the other hand, older adults who practice regular swimming were shown to have a positive relation with improved performance on five of eight experimental tasks and all EF components. VO2
max level was only positively related to executive performance for two tasks. These results provide a clear argument for the use of a multitask approach when studying the effects of chronic exercise on cognitive functions.
Before examining the effects of aging and swimming practice on cognitive performance, it is important to discuss the multitask approach used in this study. Because it is impossible to find a “pure” task to assess an isolated cognitive function, particularly an EF, multiple measures were used to rule out “task impurity” (see [2
]). As such, we wanted to assess the EF construct at the level of a general or latent variable, which reflects the postulated cognitive function more strongly than each individual experimental task alone. However, as revealed by the standardized Cronbach alphas and the matrix of correlations displayed in , it appears that the construct validity of our three EFs was quite low. Indeed, the correlations between tasks supposed to reflect a particular EF component were often low or even nonsignificant and sometimes smaller than the correlations between two tasks thought to assess different EF components. Such results have already been reported by previous studies from our own team [2
] and others [11
] and raise a question of the low specificity of the tasks used to assess EFs, which will be discussed in the limitations section.
Consistent with a large body of research on cognitive and cerebral aging, we found that chronological age has a detrimental effect on information processing speed and on the three evaluated EFs. The age-related decline in the RT tasks is in agreement with numerous studies on the effect of aging on processing speed [2
]. The performance deficits observed in our older adults in the three EF domains also agree with the literature examining this question with a similar approach [2
]. Our results also suggest that the deleterious effects of aging are more pronounced for processing speed (69% of the explained variance, see partial η2
in ) than for updating (48% of the explained variance), shifting (37% of the explained variance), and inhibition (23% of the explained variance). However, it is important to note that taking into account the educational level (as measured by the number of years of academic education) decreased the age-related effects on all the cognitive tasks and even eliminated the effects of chronological age on the letter running-span task and the WCST. This finding strengthens the importance of taking into account education as an important moderator of cognitive aging which can be interpreted in the more general context of cognitive reserve [46
]. As we will discuss below, it may be that PA (in the present study, swimming) could also be viewed as a proxy of cognitive reserve beyond educational level.
In the older population, the differences in cognitive performance between swimmers and sedentary people were not homogeneous across all the studied cognitive processes. First, there was no significant difference between active and sedentary participants in the measures of information processing speed, as assessed by the SRT and the CRT tasks (see ). This result contrasts with some previous results [48
] but agrees with others [20
], particularly with those of Hawkins and coworkers [31
] in the same context of swimming practice. These contradicting findings may be due to specific differences between particular PAs practiced by different samples of participants. The impact on processing speed may be different between racket sports or hand-ball, such as in Spirduso's study [49
], and swimming such as in our study. Another, more likely possibility is linked to the assertion that chronic exercise exerts selective effects on cognitive processes involving executive control and not on less-controlled processes, such as the stimulus-driven information processing involved in basic reaction time tasks (see [13
]). Our results favor this second possibility and strengthen this selective benefit of PA on EF processes.
Second, our results indicate prima facie that the impact of swimming practice was quite homogeneous across all three EF components, assuming that our groups only differed at the level of swimming practice. All three MANOVAs contrasting the performance between older swimmers and sedentary participants were significant for these cognitive functions. However, the percentage of variance explained by swimming practice was more important for Shifting (42%) than for Updating (30%) and Inhibition (16%), as revealed by partial eta squared (see partial η2
in ). Our results of a strong relationship between swimming practice and Shifting confirm the previous conclusions of Hawkins et al. [31
] on the positive effect of water aerobics on attentional flexibility. However, as pointed out in the Results section, the examination of speed and accuracy measures in the two switching tasks underscores that the two groups of participants may not have used the same strategies to perform these tasks. The swimmers made more errors and emphasized speed to the detriment of accuracy. Consequently, the speed-accuracy tradeoffs observed in the two switching tasks weaken the strength of the positive relationship between swimming practice and Shifting.
Moreover, five out of eight tasks were strongly sensitive to differences between swimmers and sedentary participants, but the performance of two tasks out of three (Stroop task and signal-stop task) assessing inhibition and one task out of three (dimension-switching task) assessing shifting did not significantly differ between regular swimmers and sedentary participants. As previously stated, we think that this result underscores the importance of using a multitask approach when studying the effects of chronic exercise on cognitive functions and may help to resolve some discrepant results in this domain. For example, the absence of significant effect of exercise training reported by other researchers on some cognitive tasks [21
] may not reflect an absence of real effects but instead a lack of sensitivity of the experimental tasks. For instance, our results conflict with those reported by Smiley-Oyen et al. [21
], because they found an effect of aerobic exercise on the Stroop task and no effect on the WCST. Several reasons can be hypothesized to explain the discrepancies, such as differences in study design, experimental tasks, modes of response, and selected dependent measures, but it is always difficult to definitively conclude which is the correct explanation. One of the strengths of the present study is to combine the use of several different experimental tasks to examine the influence of PA within the same sample not only at the level of individual tasks but also at the more general level of the explored cognitive functions. This procedure allowed us to show that regular swimming was clearly related to better executive performance on the three examined EFs, despite the absence of significant effect on three of the eight tasks. This may indicate that some tasks could be more sensitive to the effects of PA than other tasks. Additional work should verify the consistency of our results, as our findings are far from conclusive. A recent study from our group [51
] used different measures of PA and explored the same executive functions with different tasks or variations of the same tasks as used in the present study and showed that PA exerted a selective effect on inhibition but not on updating or shifting and only for the oldest adults of the sample (71 years and older). These different results highlight the lability of the effects of exercise on cognitive performance according to tasks and characteristics of the participants. Clearly, more work is needed to draw a definitive conclusion on the selectivity of the effects of chronic exercise effects on EFs.
The role of cardiorespiratory fitness in the relationship between PA and cognitive performance in older adults remains under debate (see [4
]). Accordingly, we were interested in examining the relationships between VO2
max level and cognitive performance. As shown in , only performance on the verbal running span task and the dimension-switching task was significantly correlated with VO2
max level. Of particular interest, one must remember that there was no significant effect of PA on the global switch cost of the dimension-switching task (P
= .11, see ). Thus, there is only one task (the verbal running span task) showing PA-related benefits that could be mediated by cardiorespiratory fitness level and one task (the dimension-switching task) showing no relationship with swimming practice but a significant relationship with VO2
max level. These results underscore the differences between a behavioral measure assessed through questionnaires (PA) and an estimated physiological measure assessed through field test (VO2
max), and their different influences on different cognitive functions [21
]. Distinguishing between PA and cardiorespiratory fitness, which were weakly related (r
= .35; P
= .054) in the present study, seems of particular importance because they may exert their influence on cognitive functioning by different mechanisms. A challenge for future studies will be to understand and characterize the mechanisms underlying these selective effects. One may hypothesize that PA could be viewed qualitatively as a proxy of cognitive reserve, as defined by Stern and collaborators [46
], allowing physically active older adults to make flexible and efficient use of available brain reserve and to demonstrate better cognitive performance. In other respects, better cardiorespiratory fitness could induce anatomical and neurophysiological changes in the brain [6
] that could reflect and enhance brain reserve [46
There are potential limitations in the present study that should be addressed. First, the low specificity of the tasks used to assess the construct of EF in this study limits the generalizability of our conclusions and strengthens the importance of looking for and using well-designed tasks to reflect the postulated cognitive functions. Second, our cross-sectional design precludes inferences about causality in the relationship between chronic engagement in swimming and executive performance. Well-designed randomized-controlled trials are now needed, using the same multitask approach, before a definitive conclusion can be made. Third, the level of education of the participants has been taken into account as a covariate in the statistical analyses of the present study. However, other important moderators of aging related to lifestyle, such as income, socioeconomic status, food habits, previous careers, and lifelong leisure activities, have not been measured in our experiment. These different variables can explain part of the variance in cognitive aging and may lead to an overestimation of the relationship between present level of physical activity and cognitive performance in cross-sectional designs, as could have been the case in the present study.
To summarize the principal findings, this study demonstrated the validity of using a multitask approach in examining the potential benefits of regular swimming on the aging of cognitive function. Such a theoretical and methodological approach allowed us to show that chronic swimming practice is related to better executive functions in older adults and that these benefits are seen on the three EFs studied but that some tasks are less sensitive to detecting these benefits. Finally, the demonstrated benefits of PA were not necessarily linked to better cardiorespiratory fitness, showing the potential relative independence of these behavioral or physiological moderators on executive performance.