We model the evolution of concepts, i.e. how members of a social group associate properties to concepts. Our Agent Based Model (ABM) is based on Conceptual Agreement Theory (CAT), which states that individuals can only infer the conceptual state of others when communicating. Through communication agents develop a conceptual structure which is influenced by three variables: the size of the group, the number of possible properties that may describe each concept and the rate at which agents learn. In general, the results show that these three variables non-linearly interact and that the larger the group and number of available properties, and the slower the learning process, the richer the conceptual structure that emerges from agents’ interactions.