Functional network activation during a subtraction task.

Arithmetic processing recruits fronto-parietal and temporal networks with frequency-specific dynamics modulated by operation and complexity. However, spectral and connectivity dynamics underlying individual differences in arithmetic performance remain poorly characterised. Here, we quantify spectral power and mutual information (MI) during serial subtraction to examine task performance correlates. Beta power in the left frontal lobe and delta power in the right temporal region were positively correlated with arithmetic score. MI between the left frontal and right temporal lobes in the beta band, as well as between the right frontal and posterior regions in beta and gamma bands, showed negative correlations with arithmetic score. In the alpha band, MI between bilateral frontal and left temporal regions was positively correlated with arithmetic score, whereas MI between the left frontal and right central derivations was negatively correlated. Overall, increased left frontal activation coupled with interhemispheric functional disconnection from contralateral posterior regions in the beta band was associated with higher arithmetic score, suggesting that selective activation and segregation from task-irrelevant activity are associated with efficient working memory and arithmetic processing. Additionally, MI between the left temporal and frontal regions in the alpha band was positively associated with improved performance, supporting attention and default mode network modulation as correlates of arithmetic task performance.