As a novel approach to conceptualizing executive functions, this study applied network analysis to a common battery of executive function tests administered to a sample covering the life span. Participants (N = 3,944; age: M = 20.8 years, SD = 19.6, range: 3-85; maternal/self education: M = 12.9 years, SD = 2.6; 53.3% girls/women, 46.7% boys/men; 61.1% White, 18.2% African American, 14.0% Latinx, 6.8% other races/ethnicities) completed tests of inhibition, shifting, and updating/working memory. Zero-order and partial correlation network models were calculated for divided age groups, with network parameters compared between groups: edge weights, corresponding to zero-order or partial correlations between two executive functions; expected influence, quantifying centrality; and global strength, quantifying differentiation. Executive functions differentiated from childhood to adolescence and dedifferentiated during young adulthood, with further dedifferentiation at older adulthood. Shifting emerged as more central than other abilities in adolescence and adulthood versus childhood, with a mediational role of shifting between inhibition and updating/working memory. A network approach can appropriately capture the unity and diversity of executive functions, by which unity reflects the reciprocal engagement between diverse abilities to produce goal-directed behavior. The engagement between abilities, and a mediational role of shifting between inhibition and updating/working memory, may be necessary for the emergence of effective goal-directed behavior. Through a network approach, the unity of executive functions represents an emergent property of the dynamics between multiple abilities (e.g., inhibition, shifting, and updating/working memory) that, when working effectively in tandem, lead to integrative processes (e.g., problem solving) that contribute to successful executive behavior and goal attainment.