We report an ABM that simulates agents that communicate and experience agreement with other agents. "Observer" agents carry in their minds one of several versions of the same concept, and observe the behavior of "actor" agents. When an actor provides evidence consistent with an observer's conceptual content, then the latter agent feels agreement. Concepts that promote agreement are useful for communication and are strengthened in agents' minds, while concepts that do not promote agreement are not useful and are weakened. Blind to individual agents, agreement may be of two kinds. True agreement (labeled p(a1)) means that actor and observer really share versions of the same concept. Illusory agreement (labeled p(a2)) means that actor and observer have different concepts and that apparent agreement occurs only due to the probabilistic nature of how conceptual content is distributed among agents. This ABM exhibits an interesting dynamical behavior for which we have developed mathematical formulations which turn out to be consistent with the system's outcomes. Several interesting conclusions were drawn from the models. Particularly, individual agents are blind to whether concepts promote true or illusory agreement among them, but even so, as a collective, the multiagent system is able to keep the concepts that generate true agreement and weed out those that promote illusory agreement. We believe that this characteristic of the ABM is an emergent property of it and its study may shed light on similar cognitive processes that occur in social groups and their results may be useful in the development of better GUI's for search engines and multiagent communication frameworks.