Dance is a joyful complex activity combining physical exercise with cognitive, social, musical, and artistic stimulation34. The benefits of dance intervention stem from the recruitment of higher-order cognitive functions that require enhanced engagement and coordination of large-scale brain networks35. This study followed up on our previous findings showing that DI elicits distinct motor10 and cognitive improvements11; it focused on the moderating effects of CR proxied by education on the DI-induced rs-FC changes involving the large-scale brain networks, namely the SMN and the DAN-aDMN.
We observed that the rs-FC increase of DAN-aDMN was dependent on the DI, while the changes in the SMN intra-network connectivity depended on the interaction between the DI and education. In other words, the follow-up changes within the SMN network were not significant across the whole DI group but only in those with higher CR. This result is discussed in more detail in the context of CR moderating effects in the text below.
All the baseline behavioral (fitness and cognitive) scores as well as their follow-up improvements were mutually correlated, sharing approximately 10% of their variance. This variance can represent a shared component of psychomotor speed, which declines throughout the lifespan36.
Regarding our neuroimaging results, previous literature showed that dance practice may modify brain plasticity as evaluated by structural MRI12,37, but little is known about the rs-FC changes induced by the DI. By comparing professional and naïve dancers, Burzynska and colleagues34 demonstrated differences between both groups in the engagement of the general motor learning network, including major nodes of the SMN, basal ganglia structures, and frontoparietal regions. The current study employed rs-fMRI and for the first time explored the CR moderation of brain plasticity changes resulting from the DI. Our major finding supported the significant CR moderation of the DI effect on the rs-FC changes within the SMN. Specifically, the observed increase of the SMN rs-FC was dependent on ≥ 13 years of education (which equals the minimum of secondary education with graduation in the Czech schooling system). This effect dissolved in moderate levels and significantly reversed in low levels of CR (≤ 10 years of education). Note that only 4.41% of the cases had such a low education level, and thus this latter result cannot be further interpreted. The observed DI-induced increase of rs-FC of the SMN is clinically relevant as it was associated with improved performance in dynamic balance and mobility, known to decline with aging8. The SMN is particularly engaged in motor learning and execution of specific motor actions38; although the SMN has not been assessed in the context of DI, studies on aerobic exercise interventions have consistently reported a reactive increase of rs-FC of the SMN in healthy24 and diseased subjects39, as well as significant differences in its structural connectivity among professional ballet dancers in comparison to a control group40. The relation between education level and motor network involvement may seem peculiar; nevertheless, associations between motor aspects and education levels have been demonstrated. For instance, in Parkinson’s disease patients (i.e. the typical patient group with a movement disorder) the CR (proxied by education) was inversely correlated with motor symptom severity despite greater reductions in dopamine levels41,42.
In contrast, the moderation model of the DAN-aDMN rs-FC changes revealed that the DI alone, without CR contribution, is a significant predictor of its change. Although the results yielded the most prominent effect for those with secondary education, the interaction effect was not significant, and therefore, this latter finding has to be taken with caution. The DMN-DAN connectivity plays an important role in cognitive control and working memory27,29, which is significantly altered with aging43. Anthony and Lin44 speculate that individual hub seeds of the DMN, including the anterior cingulate region, underlie the core hub of neural reserve in the context of CR, while the DAN regions are rather related to neural compensations (i.e. engaged in brain maintenance to compensate for brain pathology). Therefore, our results are in line with the notion that by increasing the DAN-aDMN crosstalk, dancing may facilitate neuroplasticity and the preservation of CR45. A prospective 21-year study demonstrated that regular participation in dancing was the only physical activity among the 11 studied (e.g. bicycling, playing tennis or swimming) that was associated with a lower risk of dementia in an elderly cohort, presumably by increasing plasticity and CR46.
Interestingly, while higher levels of CR were related to better baseline behavioral scores, they were not correlated with their follow-up changes. Even though ours is the first study to observe such discrepancies resulting from an intervention, many longitudinal observational studies, and particularly those conducted on samples with a degenerative brain disease, found that CR (proxied by education, occupation, or premorbid IQ) was related to baseline behavioral outcomes, but not to their changes47,48. For instance, higher education among PD patients predicted lower incidence of high Hoehn-Yahr stage, better cognitive and motor baseline scores as estimated by MMSE and gait speed with UPDRS-III respectively, but not their annual progression of 6 years49. This phenomenon has been dubbed a passive reserve hypothesis and highlights the CR contribution to better cognitive and motor performance scores resulting from the persistence of differences that appear at younger ages, rather than from ongoing changes (e.g. lifestyle or pathology) that influence differential rates of cognitive decline49.
There are limitations to our study. We used a static proxy of CR which may not be reflective of the dynamic nature of the CR and its pathology-induced depletion50. Estimating dynamic CR using a latent or residual CR index51,52 in future research might deepen our understanding of its reactive nature. Besides, educational attainment is contaminated with socioeconomic factors, such as income, access to health care, gender, and healthy lifestyle habits. Finally, adaptive testing is a more sensitive approach to training or evaluation in uncovering post-intervention effects.
In conclusion, the protective effects of cognitive reserve in nondemented older adults have been suggested by several lines of research. We showed that an intensive six-month DI can induce clinically-relevant changes in brain plasticity, physical fitness, and cognition, and importantly, that some of the brain plasticity changes depend on education, a proxy of CR, suggesting that higher capacity for plasticity applies to better intervention outcomes. Our study also demonstrated that the DAN-aDMN rs-FC, a potential neural representation of CR, can be modulated by DI. Future studies should employ multimodal comprehensive programs to benefit people across different CR levels36. Despite our clinically-relevant results, it is unknown whether short-term engagement in any set of activities is sufficient to elicit changes that last several months or even years after the intervention completion. Therefore, long-term behavioral outcomes of such interventions should be examined and long-lasting moderation effects of CR should be tested.