In this talk,  i will discuss how coherence-based models can be seen as an alternative to logic-based BDI and argumentation models for the reasoning of normative agents. In particular i'll introduce a coherence based argumentation model for agents to deliberate on possible norms. First a deductive coherence model will be presented, in which the coherence values are derived from the deduction relation of an underlying logic which makes it possible to identify the reasons for why a proposition is accepted or rejected. Then I show how coherence-driven agents can generate candidate norms for deliberation, after which i introduce a dialogue protocol for such deliberations.