The human brain is a hierarchical system where molecular, cellular, and corticallevel components interact to produce myriad mental states. Thus, brain networks can broadly be characterized in a hierarchical fashion, such that macroscale cortical functions result from accumulated microscale events ranging from the cellular to the molecular levels. However, modern neuroimaging techniques are limited in the extent to which they can resolve these interactions. In the following set of experiments, we employ a combination of imaging epigenetics and statistical machine learning techniques to decode individual differences in epigenetic makeup from multiple modalities of macroscale brain network data. A convenient target for this approach is the oxytocin system, which is modulated by epigenetic modifications to the oxytocin receptor gene (OXTR). Differential levels of DNA methylation in OXTR are thought to have downstream effects on both social behavioral phenotypes and the development of neural networks supporting such behaviors. Using data obtained from a large sample of healthy young adults, we describe: 1) The identification of epigenetic fingerprints in macroscale neural network architecture;2) Spatially-heterogenous relationships between epigenetic factors and spontaneous BOLD dynamics; and 3) An entropy-based model that links epigenotypes and behavioral phenotypes through patterned network dynamics. Our approach offers both novel applications of previously-existing techniques and new tools for quantifying dynamics in functional brain networks. Together, these methods have the potential to illuminate complex interactions across multiple levels of brain systems in numerous contexts, from social behavior and beyond.