Argumentation and nonmonotonic/default logics have long been considered as two important forms of commonsense reasoning, side by side, with multiple interactions but essentially different characteristics, in particular with respect to semantics. The outputs of argumentation systems have been evaluated according to the standards of nonmonotonic reasoning, and, in particular, answer set programming (presumably the most successful implementation of nonmonotonic logics) has been used to implement argumentation semantics. Changing argumentation systems arose as a vital topic of research only recently, and again, standards of belief change like AGM theory have been adapted to guide and evaluate such changes. Our project aims at combining both frameworks on a much deeper, syntactic and semantic level, overcoming the still present gap between nonmonotonic semantics, which is often based on preferential models, and argumentation semantics, which is based on the notions of attack and defence. The Abstract Dialectical Frameworks (ADFs), developed by Brewka and Woltran, are an important first step in this direction. However, ADFs use the syntax of classical logic, with a three-valued logic on a meta-level. Our project builds upon ADFs, but we will use three-valued conditionals right from the beginning to implement arguments. This allows for representing the non-classical (three-valued) nature of argumentation at the object level and hence for a novel semantic view on arguments, which we explore in this project. But most importantly, this establishes a basic connection between argumentation and reasoning from conditional knowledge bases that opens up many ways to transfer features and results from the rich scope of methodologies from either side to the other, leading to novel methods and techniques on either side, and a better understanding of the relation between argumentation and conditional reasoning, ideally resulting in a global picture that shows (at least parts of) both frameworks to be "two sides of the same coin".