Knowledge-based communities bring together persons who collaborate and interact in order to produce knowledge. The tasks involved in this production are usually described in terms of “epistemic communities”: groups of individuals working on shared topics in pursuit of a common goal of knowledge production. Focus is thus placed on the formation, evolution and interactions of these communities – in general, on processes of collaboration. The objective is twofold: on one hand, describe how epistemic communities emerge and overlap at various hierarchical levels and, on the other, expose the determinants of collaboration between members of these communities. In an attempt to formalize the analysis of distributed cognition, an epistemic community is defined as the largest set of agents sharing and, working on, the same set of concepts. This formal definition helps us describe the distributed structure of knowledge production; understand the impact of this structure on the perception and arrangement of activities related to distributed cognition and appraise dynamically the stability over time of the distribution of tasks within the network, in spite of a continuous turnover among members. An empirical study of a community of embryologists illustrates this approach.