This study examined the association between aerobic fitness and cognitive flexibility, which is one aspect of cognitive control, in preadolescent children using the TMT. The main finding indicated that higher-fit children exhibited superior task performance relative to their lower-fit peers on the TMT-A, which requires minimal cognitive control. This result corroborates previous findings indicating the general nature of the relationship between fitness and cognition in preadolescent children (Buck et al. 2008; Hillman et al. 2009). In contrast, task performance did not differ between lower-fit and higher-fit children for the TMT-B. Given that prior research indicated that differences in task performance between lower-fit and higher-fit children (i.e., superior task performance for higher-fit children) was disproportionately larger for a task condition requiring greater cognitive flexibility (Pontifex et al. 2011), we hypothesized that differences in task performance between groups would be disproportionately greater for the TMT-B relative to the TMT-A. The current finding did not support our hypothesis.

One possible explanation for the unexpected finding might be that the TMT-B was too difficult for both lower-fit and higher-fit children. This possibility is supported by previous studies examining the relationship between fitness/physical activity and cognitive flexibility using task-switching tasks in young adults. A typical task-switching task consists of pure and mixed task conditions. The pure task condition includes a single rule set (i.e., AAAAAA… or BBBBBB…) and asks participants to respond based on the stimuli presented. The mixed-task condition consists of two or more tasks and requires participants to change rapidly between different tasks (e.g., AABBAA… or ABABAB…). The mixed task condition requires greater amounts of cognitive control such as cognitive flexibility, working memory, and inhibition (for a review, see Monsell 2003). Hillman et al. (2006a) used the task-switching task to examine the relationship between physical activity level and cognitive control in younger and older adults, and indicated that physically active individuals exhibited a shorter reaction time (RT) relative to inactive individuals irrespective of age. Kamijo and Takeda (2010) showed, using the task-switching task, that physically active young adults had a shorter RT for the mixed task condition relative to their inactive peers, whereas no such difference was observed for the pure task condition. These results suggest that physical activity level is selectively associated with task performance for task conditions requiring greater cognitive flexibility. Thus, it is likely that a higher level of physical activity is associated with superior cognitive flexibility even during young adulthood.

In contrast, Scisco et al. (2008) observed no relationship between aerobic fitness and task performance during the task-switching task in young adults. One possible source of this discrepancy relates to differences in task difficulty among these studies. In Hillman et al. (2006a) and Kamijo and Takeda (2010), single digit numbers (digits 1–4 and 6–9) were used for the task-switching task. Specifically, participants were required to judge whether the number was odd or even, or whether the number was greater or lesser than 5. In Scisco et al. (2008), participants were asked to judge whether double digit numbers (digits 11–99) were odd or even, whether the numbers were greater or lesser than 50, whether the sum of two digits was odd or even, or whether the sum of two digits was greater or lesser than 10. Task-switching tasks including four tasks as well as calculation (Scisco et al. 2008) must undoubtedly be more difficult than those involving only two tasks (Hillman et al. 2006a; Kamijo and Takeda 2010). Thus, it is speculated that the relationship between fitness/physical activity and cognitive control might blur if the cognitive task is too difficult, even though the task requires variable amounts of cognitive control based on task conditions.

Previous research, which investigated the relation between aerobic fitness and cognitive function in younger and older adults using task difficulty manipulation during an oddball task, supports this speculation (Pontifex et al. 2009). Specifically, the oddball task required participants to respond to rare target stimuli (i.e., oddballs) and withhold their response to frequent nontarget stimuli. The target stimuli were 5.5 cm diameter white circles in both easy and difficult task conditions, whereas the diameter of the nontarget stimuli (i.e., white circles) were 3.0 cm and 5.0 cm in the easy and difficult condition, respectively. Results indicated that higher-fit individuals exhibited shorter RTs relative to their age-matched lower-fit counterparts for the easy, but not for the difficult, task condition. Thus, it is likely that the positive association between aerobic fitness and cognitive function is affected by task difficulty. Taken together, it is plausible that, in this study, task performance did not differ between the child groups for the TMT-B since this task condition was too difficult for preadolescent children.

It is noteworthy that differences in task performance between children and young adults were larger for the TMT-B (about 40 ms) relative to the TMT-A (about 20–30 ms). Given that there is no doubt that young adults have superior cognitive flexibility compared to children, the differences in task performance as a function of age and the task condition indicate that cognitive control demands were successfully manipulated between the TMT-A and TMT-B in this study. Nonetheless, a positive association between aerobic fitness and cognitive function was selectively observed for the TMT-A requiring minimal cognitive control. Given that accumulating evidence has demonstrated that the association is selectively and disproportionately greater for higher-order cognitive functions (i.e., cognitive control) not only during adulthood (e.g., Colcombe and Kramer 2003; Hillman et al. 2006a; Kamijo and Takeda 2010; Kramer et al. 1999) but also during childhood (e.g., Kamijo et al. 2011; Pontifex et al. 2011), the result of no association between fitness and task performance on the TMT-B might not be due to its cognitive control demands, but due to its task difficulty for the current participants (i.e., preadolescent children) as discussed above.

In sum, aerobic fitness was associated with task performance for the TMT-A, but not for the TMT-B in preadolescent children. These findings support the general nature of the relationship between fitness and cognition during childhood and suggest that this relationship can be underestimated or even disappear if cognitive tasks are too difficult. Thus, in conclusion, we propose that researchers should select cognitive tasks with caution when examining the association between fitness and cognition. This caution should include not only aspects involved with cognitive control demands, but also aspects dealing with task difficulty.