Genetic control of dynamic brain network reconfiguration during working memory

Working memory is fundamental to human cognition, yet the genetic contributions to dynamic brain network states during task remain poorly understood. Here, we examined static and dynamic functional connectivity in monozygotic (MZ) and dizygotic (DZ) twins performing a 2-back versus 0-back working memory task, isolating cognitive-load-specific neural processes. Static functional connectivity exhibited moderate heritability (A ≈ 0.30), predominantly in heteromodal association cortex, including the Default Mode Network (DMN) and fronto-cingulate control regions, whereas primary sensory-motor networks showed minimal genetic influence. Dynamic connectivity further distinguished mean and variance components: mean dynamic connectivity demonstrated moderate heritability (A ≈ 0.23-0.30) across DMN-attention and DMN-frontal control interactions, whereas variance of dynamic connectivity showed the strongest genetic effects (A ≈ 0.25-0.44), reflecting highly heritable moment-to-moment neural flexibility within temporal, cingulo-opercular, and executive networks. Using k-means clustering, we identified two recurring dynamic brain states. State 2 was dominant (occupancy 75.6%) and stable (mean dwell ≈ 3.98 windows), whereas State 1 was transient (occupancy 24.4%, dwell ≈ 1.29 windows). Notably, neither state occupancy nor switching dynamics showed significant heritability, indicating these temporal characteristics are largely driven by environmental or task-related variability. Collectively, these findings reveal a gradient of genetic influence: static connectivity captures a moderately heritable trait-like baseline; mean dynamic connectivity reflects intermediate task-modulated coordination; and dynamic variability represents the most genetically influenced phenotype, highlighting neural flexibility and adaptability as heritable traits. These results suggest that variability in dynamic connectivity may serve as a sensitive endophenotype for cognitive and psychiatric traits, emphasizing the importance of temporal network dynamics in capturing genetically mediated individual differences in working memory.