The epistemic function of replication: Mapping the domains of the debate