Toward Justifying Actions with Logically and Socially Acceptable Reasons

53 notions has significance as an alternative method to value-based decisions. In fact, compared to value-based approaches, reason-based approaches have attractive features that they give the way we normally think and talk about choices, the way suited to handle complex and real world decisions, and the way to understand the conflict characterizing the making of decisions, although it has disadvantages in weak rigor [3]. The aim of this paper is to formalize argument-based reasoning for actions supported by believable reasons in terms of nonmonotonic consequences and desirable reasons in terms of Pareto optimality and maximizing social welfare functions. Using modal propositional language with defaults and defeasible inference rules for practical reasoning, we give three kinds of structured practical argu-mentation frameworks for nonmonotonic consequences, for Pareto optimality, and for social welfare, respectively. We combine these frameworks into a layered practical argumentation framework to evaluate actions supported by Pareto optimal , social welfare maximizing, and nonmonotonic consequence reasons. Our contribution is summarized as follows. First of all, we extend argument-based reasoning so that it can formally combine not only logical properties, i.e., non-monotonic consequences, but also about social properties, i.e., Pareto optimality and social welfare functions on a specific logical language. Second of all, we propose a unified method to evaluate actions with logically and socially acceptable reasons by proposed layered practical argumentation framework. In [4], the authors apply Pareto optimality and social welfare functions for evaluating desirable extensions, i.e., argumentation consequences, given by Dungean semantics. Whereas this is the first report introducing these notions in argu-mentation, it does not address justifying these notions by means of argument-based reasoning. Game theoretic approaches tend to assume fixed and immutable agents' utility functions [5]. Contrary to human decision makings, however, the approaches do not tolerate utility changes caused by new information. For instance , when I set out to buy a car, I may initially decide that I want a car with an electric sun roof. However, if I subsequently read that electric sun roofs are unreliable and tend to leak, then this might well change my preferences [6]. Furthermore, people may not understand an explanation as a sequence of complex equations provided by the approaches. Our argument-based approach allows agents to justify their beliefs by providing mechanisms for making counterargu-ments and its explanations are reasons we normally use when we think and talk. In [7], the authors give pessimistic and optimistic criteria for …