22 January 2010
REPORT: Epistemology, Context and Formalism
Posted by Rasmus Rendsvig under: Conference and workshop reports .
In November, the Maison des Sciences de l’Homme at the University of Nancy hosted the workshop on Epistemology, Context and Formalism. EpiConFor offered an impressive line-up of participants with wide-ranging interests: epistemologists, computer scientists, logicians and linguists all contributed to a tremendously rich and informative workshop. Every day was opened and closed by an invited talk and two or three presentations of contributed papers took place in between.
The conference was kicked off by Peter Baumann, who proposed a way out from a seeming contradiction in the position of epistemic contextualism. The contradiction is derived from basic tenets of contextualism together with disquotation, factivity of knowledge and epistemic closure under modus ponens. Peter’s solution, based on a special form of contextualism in which the closure principle is slightly modified stirred interest in the audience and questions revolved mainly around the issue of the relation between validity and indexicality.
Erich Rast followed suit with a stimulating discussion of the many issues raised by the attempt of describing context as the set of assumptions mutually shared in a group of epistemic agents. Erich delineated the framework needed to make such an attempt, pointing out that a viable model requires iterated belief revision, default reasoning and the representation of qualitiative beliefs.
The next paper was a fascinating contribution by Brad Armendt linking context and degree of belief. The main idea in Brad’s insightful analysis is that one’s belief in p may be influenced by what’s at stake given that p turns out to be true or false. He argued that, when context is not specified, (syncronic) belief states should be modeled as mixed states, varying on a continuum depending on a parameter relative to the pragmatic interest in the truth of the entertained beliefs.
The first day came to a close with Richmond Thomason‘s discussion of epistemic agency as a society of mind. According to Richmond, agents’ beliefs are better represented not by a unique belief operator, but rather by a family of operators, shifting according to context. The idea has powerful implications, and Richmond covered many of those in his thought-provoking lecture. Some were related to classic philosophical puzzles (Kripke’s Pierre) or to new ones (the intricate relation between belief and what’s at stake analyzed earlier in Armendt’s talk). Some were related to linguistics (the tailoring of common ground in conversational contexts) and some to the issue of knowledge-based architectures (the dynamics of the interactions of intra-agency modalities).
The next morning the audience was treated to a masterful overview of a wealth of techniques in representing epistemic attitudes and their dynamics. Johan van Benthem‘s profound and delightful lecture focused on the importance of modeling knowledge, information and its dynamics in interactive frameworks. The “scandal” of the non-informativeness of deduction receives new light when epistemic modalities are understood as related to agents’ information and inference as an action through which agents come to entertain, or to be aware, of a proposition. A variety of epistemic dynamics were surveyed–form the logic of public announcements and dynamic epistemic logic to more distant cousins as, for instance, dynamic semantics.
Julien Dutant offered an intriguing synthesis of logic and epistemology. Agents’ knowledge is determined by the methods they use and, crucially, the semantics allows for the composition of different methods. The idealization typical of Kripke semantics are given here a new interpretation, since the method-based semantics distinguished between axioms that are justified by mere introspective properties and axioms that require excellence of methods. Questions focused on the distinction between type and token processes.
In Roger Clarke’s talk the context-sensitivity of belief represents the central premise for the proposed solutions to two long-standing formal and philosophical puzzles. Context-sensitivity helps reconcile degrees of belief and all-or-nothing beliefs through high-probability operators (threshold operators). Moreover, context-sensitivity helps solve the preface paradox, since the context in which the preface is assessed and the contexts in which it was written are different.
Dylan Dodd analyzed simple arguments involving probable knowledge and used intuitions about their validity relating them to a principle entertained by contextualists and contrastivists alike. On such a basis, Dylan argues that the principle (that a subject knows p only if she can eliminate all the not-p possibilities) applies not only to subject’s knowledge, but also to her belief.
Day two ended with the energetic and captivating talk delivered by Vincent Hendricks. Vincent started off by offering an overview of his theory of epistemic agency obtained by mixing together formal learning theory and epistemic logic. In Vincent’s theory of agency, agents have two basic methods of inquiry: discovery (a function from evidence to hypotheses) and assessment (from evidence and hypotheses to {T,F}). Defining knowledge as limiting convergence, one sees the relation between the epistemic methods of inquiry adopted by agents and specific classic axioms of epistemic logic. Building on this account, Vincent went then on to show that the epistemic phenomena of knowledge transmissibility and pluralistic ignorance can be readily understood and modeled in the framework of his theory of epistemic agency.
The third days started with Hans Kamp‘s rich lecture on Dynamic Semantics and Discourse Representation Theory. Hans looked at the referential-attributive distinction, modeling the notion of the context of an utterance (as discussed by Kaplan and others) in the framework of Dynamic Semantics. One relevant outcome of the operation is that the distinction between choice of linguistic expression (on part of the speaker) and interpretation of linguistic expression (on part of the audience) come across neatly as distinct processes in the formal framework, enabling a better understanding of the challenges related to the referential-attributive distinction.
Paula Sweeney followed up with her talk on epistemic modals. Paula’s paper sees content-relativism as the most promising semantic theory for epistemic modalities. By assuming that the content of epistemically modalized sentences varies across contexts of utterance, Paula shows that the unpalatable notion of truth being relative to utterances is avoided, while embeddings under epistemic operators are not particularly problematic.
In the last contributed talk, Giacomo Sillari tackled the topic of peer disagreement. He stressed the importance of the technical literature on Aumann’s agreement theorem to the recent peer disagreement debate, arguing that dynamic version of the agreement theorem highlights the importance of higher-order knowledge. Giacomo then presented a formalization of a vexed example widely discussed in the peer disagreement literature and showed how the agreement theorem offers new insights to it.
The conference was closed by John Hawthorne‘s sweeping overview of problems, challenges and insights relative to contextualism. John’s enchanting discussion covered eight aspects of contextualism, showing deep connections, at both the formal and the conceptual level, among many of the themes discussed during the EpiConFor workshop, whose impeccable organization was due to Manuel Rebuschi and Franck Lihoreau. Marion Renauld, part of the efficient local organizing team, helped with this report.
Workshop website: http://poincare.univ-nancy2.fr/Activites/?contentId=5657&languageId=1
Loading... 
