Adaptive logics for defeasible reasoning. Applications in argumentation, normative reasoning and default reasoning (Q2517310)

Adaptive logics provide a general dynamic framework for nonmonotonic reasoning designed to steer a course between a logic, the ``lower limit logic'', that seems too weak for its purposes, missing important inferences, and a logic, the ``upper limit logic'', that seems too strong. While it captures those inferences, it may have untoward consequences in exceptional cases. In adaptive logics such inferences are treated provisionally, being regarded as normally as possible, and allowed to stand until the information of their premises appears abnormal, whereupon they are withdrawn, even while other inferences of the same form may go through, and they may even be reinstated with further development of that information. The classic, accessible sources [\textit{D. Batens}, Log. Anal., Nouv. Sér. 44, No. 173--175, 45--68 (2001; Zbl 1047.03017); Log. Univers. 1, No. 1, 221--242 (2007; Zbl 1116.03019)] explicate these notions rigorously. The present work, however, also presents a thorough account of the motivation and mechanics of adaptive logics, all that is required to maintain the subsequent development of the book. This book applies the adaptive logic framework to aspects of three areas of defeasible reasoning, default reasoning, Dung's argumentation theory, and certain problems in deontic logic. The book itself is a revision and slight expansion of the author's PhD dissertation. Most of the chapters are versions of previously published papers, several with co-authors; these papers are indicated below. Part I presents the general picture of adaptive logics as a framework for defeasible reasoning. Chapter 1 is a brief introduction to motivate the picture. Chapter 2 provides the details of adaptive logics in their standard format, both semantically and proof-theoretically. Chapter 3 shows how adaptive systems can be sequentially combined; this is crucial for some of the later applications. Chapter 4 (cf. [\textit{D. Batens} et al., Log. Anal., Nouv. Sér. 52, No. 207, 281--304 (2009; Zbl 1190.03002)]) describes criteria for determining when premise sets are equivalent in adaptive logics. Chapter 5 generalizes the standard format. Part II applies adaptive logics to default reasoning. Chapter 6 shows how to apply modus ponens adaptively to default conditionals (cf. the author's paper, [``Adaptively applying modus ponens in conditional logics of normality'', J. Appl. Non-classical Log. 22, 125-148 (2012)]). This calls for introducing an operator into the object language that marks information as atypical or exceptional; modus ponens may apply (provisionally) when the antecedent is not so marked in context. Chapter 7 (cf. [the author, in: The many sides of logic. Selected papers of the simultaneous conferences ``CLE 30 -- 30th Anniversary of the Centre for Logic, Epistemology and the History of Science'', ``XV EBL -- 15th Brazilian logic conference'' and ``XIV SLALM -- 14th Latin-American symposium on mathematical logic'', Paraty, Brazil, May 11--17, 2008. London: College Publications. 47--67 (2009; Zbl 1227.03036)]) provides an adaptive logic for rational closure. Part III applies adaptive logics to P. M. Dung's abstract argumentation theory (e.g., [\textit{P. M. Dung}, Artif. Intell. 77, No. 2, 321--357 (1995; Zbl 1013.68556)]), with Chapter 8 (cf. [the author and \textit{D. Šešelja}, J. Log. Comput. 21, No. 2, 133--156 (2011; Zbl 1234.03014)]) offering a proof-theory for that framework, and Chapter 9 extending that account to accommodate joint attacks on arguments. Part IV applies adaptive logics to three problems in deontic logic. Chapter 10 (cf. [the author et al., Log. Anal., Nouv. Sér. 55, No. 219, 477--506 (2012; Zbl 1288.03018)]) addresses the problem of normative conflicts and how to avoid deontic explosion; it defines adaptive extensions of certain systems of this reviewer (e.g., [J. Appl. Log. 3, No. 3--4, 461--483 (2005; Zbl 1079.03009); Lect. Notes Comput. Sci. 3065, 74--113 (2004; Zbl 1169.03335)]). This repairs a problem in the book's author's earlier paper. Chapter 11 (cf. [the author, Log. Log. Philos. 19, No. 1--2, 95--128 (2010; Zbl 1229.03020)]) considers how versions of rational monotony can be treated in adaptive logics of conditional obligation, especially when they allow for normative conflicts. (This too takes off from the reviewer's [Lect. Notes Comput. Sci. 3065, 74--113 (2004; Zbl 1169.03335)].) Chapter 12 (cf. [the author, J. Appl. Log. 9, No. 1, 61--80 (2011; Zbl 1229.03021)]) takes up the question of ``factual detachment'' in logics of conditional obligation, and applies methods similar to those of Chapter 6, though the atypicality operator is used here to mark conditional obligations rather than factual information or antecedents. Through all of these applications, we see that each calls for its own particular adaptive logic, tailor-made for its purpose, but often in tension with the others. It is the framework of adaptive logic in general that is supposed to provide a unifying perspective on defeasible reasoning. The more technical demonstrations of results are presented in a series of Appendices following the chapters. Because most of the chapters are adapted from previously published papers, there is much overlap among them, e.g., to explain the methods of adaptive logics, and each chapter contains its own list of references. There is, however, a comprehensive, if somewhat short, Index.