Electrophysiological signatures of dedifferentiation differ between fit and less fit older adults

Abstract

Cardiorespiratory fitness was found to influence age-related changes of resting state brain network organization. However, the influence on dedifferentiated involvement of wider and more unspecialized brain regions during task completion is barely understood. We analyzed EEG data recorded during rest and different tasks (sensory, motor, cognitive) with dynamic mode decomposition, which accounts for topological characteristics as well as temporal dynamics of brain networks. As a main feature the dominant spatio-temporal EEG pattern was extracted in multiple frequency bands per participant. To deduce a pattern's stability, we calculated its proportion of total variance among all activation patterns over time for each task. By comparing fit (N = 15) and less fit older adults (N = 16) characterized by their performance on a 6-min walking test, we found signs of a lower task specificity of the obtained network features for the less fit compared to the fit group. This was indicated by fewer significant differences between tasks in the theta and high beta frequency band in the less fit group. Repeated measures ANOVA revealed that a significantly lower proportion of total variance can be explained by the main pattern in high beta frequency range for the less fit compared to the fit group [F(1,29) = 12.572, p = .001, partial η² = .300]. Our results indicate that the dedifferentiation in task-related brain activation is lower in fit compared to less fit older adults. Thus, our study supports the idea that cardiorespiratory fitness influences task-related brain network organization in different task domains.

Supplementary information: The online version of this article (10.1007/s11571-020-09656-9) contains supplementary material, which is available to authorized users.

Keywords: Cardiorespiratory fitness; Dynamic mode decomposition; Electroencephalography; Older adults; Spatio-temporal coherent patterns.

Fig. 1

Experimental setup of the sensory (a, b), n-back (c), and motor task (d)

Fig. 2

DMD main mode features during rest with eyes closed and the three different task conditions (motor, sensory, cognitive) in the frequency ranges of θ (4 to < 7 Hz), α (7 to < 12 Hz), low (12 to < 16 Hz) and high β (16 to < 30 Hz) as well as their source representation. Each row represents a condition and each column represents a frequency band, thus there are four topographic maps per condition. Maps represent the mean over all participants

Fig. 3

a–d Statistical t-maps of significant differences of DMD main mode features in θ, α, low- and high β between the conditions divided in fit (green) and less fit (grey) participants. Only t-values with corresponding corrected p value < .05 are visualized. The opposite side of each group served as second term in the t-test for each group. e–h DMD main mode goodness of fit expressed in % of variance they explain as group mean and standard deviation. * indicates significant pairwise comparisons of the main effect group. # indicates significant pairwise comparisons of the main effect task