Denis Bonnay

In this paper we compare different models of vagueness viewed as a specific form of subjective uncertainty in situations of imperfect discrimination. Our focus is on the logic of the operator “clearly” and on the problem of higher-order vagueness. We first examine the consequences of the notion of intransitivity of indiscriminability for higher-order vagueness, and compare several accounts of vagueness as inexact or imprecise knowledge, namely Williamson’s margin for error semantics, Halpern’s two-dimensional semantics, and the system we call Centered semantics. We then propose a semantics of degrees of clarity, inspired from the signal detection theory model, and outline a view of higher-order vagueness in which the notions of subjective clarity and unclarity are handled asymmetrically at higher orders, namely such that the clarity of clarity is compatible with the unclarity of unclarity.

The standard relation of logical consequence allows for non-standard interpretations of logical constants, as was shown early on by Carnap. But then how can we learn the interpretations of logical constants, if not from the rules which govern their use? Answers in the literature have mostly consisted in devising clever rule formats going beyond the familiar what follows from what. A more conservative answer is possible. We may be able to learn the correct interpretations from the standard rules, because the space of possible interpretations is a priori restricted by universal semantic principles. We show that this is indeed the case. The principles are familiar from modern formal semantics: compositionality, supplemented, for quantifiers, with topic-neutrality.

The problem of logical constants consists in finding a principled way to draw the line between those expressions of a language that are logical and those that are not. The criterion of invariance under permutation, attributed to Tarski, is probably the most common answer to this problem, at least within the semantic tradition. However, as the received view on the matter, it has recently come under heavy attack. Does this mean that the criterion should be amended, or maybe even that it should be abandoned? I shall review the different types of objections that have been made against invariance as a logicality criterion and distinguish between three kinds of objections, skeptical worries against the very relevance of such a demarcation, intensional warnings against the level at which the criterion operates, and extensional quarrels against the results that are obtained. I shall argue that the first two kinds of objections are at least partly misguided and that the third kind of objection calls for amendment rather than abandonment

The standard semantic definition of consequence with respect to a selected set X of symbols, in terms of truth preservation under replacement (Bolzano) or reinterpretation (Tarski) of symbols outside X, yields a function mapping X to a consequence relation ⇒x. We investigate a function going in the other direction, thus extracting the constants of a given consequence relation, and we show that this function (a) retrieves the usual logical constants from the usual logical consequence relations, and (b) is an inverse to—more precisely, forms a Galois connection with—the Bolzano-Tarski function

We take Carnap’s problem to be to what extent standard consequence relations in various formal languages fix the meaning of their logical vocabulary, alone or together with additional constraints on the form of the semantics. This paper studies Carnap’s problem for basic modal logic. Setting the stage, we show that neighborhood semantics is the most general form of compositional possible worlds semantics, and proceed to ask which standard modal logics (if any) constrain the box operator to be interpreted as in relational Kripke semantics. Except when restricted to finite domains, no modal logic characterizes all the Kripkean interpretation of P. Moreover, we show that, in contrast with the case of first-order logic, the obvious requirement of permutation invariance is not adequate in the modal case. After pointing out some known facts about modal logics that nevertheless force the Kripkean interpretation, we focus on another feature often taken to embody the gist of modal logic: locality. We show that invariance under point-generated subframes (properly defined) does single out the Kripkean interpretations, but only among topological interpretations, not in general. Finally, we define a notion of bisimulation invariance — another aspect of locality — that, together with a reasonable closure condition, gives the desired general result. Along the way, we propose a new perspective on normal neighborhood frames as filter frames, consisting of a set of worlds equipped with an accessibility relation, and a free filter at every world.

Providing a principled characterization of the distinction between logical and non-logical expressions is a longstanding issue in the philosophy of logic. In the Logical Syntax of Language, Carnap proposes a syntactic solution to this problem, which aims at grounding the claim that logic and mathematics are analytic. Roughly speaking, his idea is that logic and mathematics correspond to the largest part of science for which it is possible to completely specify by "syntactic" means which sentences are valid and which are not. Despite a renewed interest in the philosophical benefits of analyticity, both inside and outside of Carnap scholarship, Carnap's definition of logical expressions seems to have drawn too little attention. I shall argue that it is worth a second look. More precisely, my aim will be to defend this idea against some technical problems faced by Carnap's way of implementing it and against Quinean attacks on syntax-based conventionalism. Section 1 presents Carnap's definition in the context of Logical Syntax of Language, that is, how exactly the definition works, and why Carnap needs it. In section 2, I review three challenges that have been raised in the literature, and I propose to revise the definition accordingly. I argue that its modified version is immune to the previous challenges, and, to some extent, immune to new challenges as well. In the last section, I suggest that the definition has a philosophical interest of its own, because standard Quinean objections are not as conclusive as one might think when attention is paid to the fact that Carnap requires complete syntactic specification of validities

Consider any logical system, what is its natural repertoire of logical operations? This question has been raised in particular for first-order logic and its extensions with generalized quantifiers, and various characterizations in terms of semantic invariance have been proposed. In this paper, our main concern is with modal and dynamic logics. Drawing on previous work on invariance for first-order operations, we find an abstract connection between the kind of logical operations a system uses and the kind of invariance conditions the system respects. This analysis yields (a) a characterization of invariance and safety under bisimulation as natural conditions for logical operations in modal and dynamic logics, and (b) some new transfer results between first-order logic and modal logic.

Since the ground-breaking contributions of M. Dummett (Dummett 1978), it is widely recognized that anti-realist principles have a critical impact on the choice of logic. Dummett argued that classical logic does not satisfy the requirements of such principles but that intuitionistic logic does. Some philosophers have adopted a more radical stance and argued for a more important departure from classical logic on the basis of similar intuitions. In particular, J. Dubucs and M. Marion (?) and (Dubucs 2002) have recently argued that a proper understanding of anti-realism should lead us to the so-called substructural logics (see (Restall 2000)) and especially linear logic. The aim of this paper is to scrutinize this proposal. We will raise two kinds of issues for the radical anti-realist. First, we will stress the fact that it is hard to live without structural rules. Second, we will argue that, from an anti- realist perspective, there is currently no satisfactory justi cation to the shift to substructural logics

The dual character of invariance under transformations and definability by some operations has been used in classical works by, for example, Galois and Klein. Following Tarski, philosophers of logic have claimed that logical notions themselves could be characterized in terms of invariance. In this article, we generalize a correspondence due to Krasner between invariance under groups of permutations and definability in L∞∞ so as to cover the cases that are of interest in the logicality debates, getting McGee’s theorem about quantifiers invariant under all permutations and definability in pure L∞∞ as a particular case. We also prove some optimality results along the way, regarding the kinds of relations which are needed so that every subgroup of the full permutation group is characterizable as a group of automorphisms.

What is a logical constant? In which terms should we characterize the meaning of logical words like “and”, “or”, “implies”? An attractive answer is: in terms of their inferential roles, i.e. in terms of the role they play in building inferences. More precisely, we favor an approach, going back to Dosen and Sambin, in which the inferential role of a logical constant is captured by a double line rule which introduces it as reflecting structural links (for example, multiplicative conjunction reflects comma on the right of the turnstyle). Rule-based characterizations of logical constants are subject to the well known objection of Prior’s fake connective, tonk. We show that some double line rules also give rise to such pseudo logical constants. But then, we are able to find a property of a double line rules which guarantee that it defines a genuine logical constant. Thus we provide an alternative answer to Belnap’s requirement of conservatity in terms of a local requirement on double line rules.

Philosophy of science studies the methods, theories, and concepts used by scientists. It mainly developed as a field in its own right during the twentieth century and is now a diversified and lively research area. This book surveys the current state of the discipline by focusing on central themes like confirmation of scientific hypotheses, scientific explanation, causality, the relationship between science and metaphysics, scientific change, the relationship between philosophy of science and science studies, the role of theories and models, unity of science. These themes define general philosophy of science. The book also presents sub-disciplines in the philosophy of science dealing with the main sciences: logic, mathematics, physics, biology, medicine, cognitive science, linguistics, social sciences, and economics. While it is common to address the specific philosophical problems raised by physics and biology in such a book, the place assigned to the philosophy of special sciences is much more unusual. Most authors collaborate on a regular basis in their research or teaching and share a common vision of philosophy of science and its place within philosophy and academia in general. The chapters have been written in close accordance with the three editors, thus achieving strong unity of style and tone.

Jeffrey conditioning tells an agent how to update her priors so as to grant a given probability to a particular event. Weighted averaging tells an agent how to update her priors on the basis of testimonial evidence, by changing to a weighted arithmetic mean of her priors and another agent’s priors. We show that, in their respective settings, these two seemingly so different updating rules are axiomatized by essentially the same invariance condition. As a by-product, this sheds new light on the question how weighted averaging should be extended to deal with cases when the other agent reveals only parts of her probability distribution. The combination of weighted averaging and Jeffrey conditioning is a comprehensive updating rule to deal with such cases, which is again axiomatized by invariance under embedding. We conclude that, even though one may dislike Jeffrey conditioning or weighted averaging, the two make a natural pair when a policy for partial testimonial evidence is needed.

Hintikka makes a distinction between two kinds of games: truthconstituting games and truth-seeking games. His well-known game-theoretical semantics for first-order classical logic and its independence-friendly extension belongs to the first class of games. In order to ground Hintikka’s claim that truth-constituting games are genuine verification and falsification games that make explicit the language games underlying the use of logical constants, it would be desirable to establish a substantial link between these two kinds of games. Adapting a result from Thierry Coquand, we propose such a link, based on a slight modification of Hintikka’s games, in which we allow backward playing for ∃loïse. In this new setting, it can be proven that sequent rules for first-order logic, including the cut rule, are admissible, in the sense that for each rule, there exists an algorithm which turns winning strategies for the premisses into a winning strategy for the conclusion. Thus, proofs, as results of truth-seeking games, can be seen.

La logique est une théorie normative du raisonnement, qui vise à caractériser la classe des arguments déductifs valides en déterminant si la conclusion est conséquence logique des prémisses. Mais, selon la définition sémantique devenue classique, la caractérisation de la relation de conséquence logique dépend elle-même de la caractérisation de la classe des mots logiques, ces mots qui, comme « non », « et », « tous » ou « certains » servent à articuler nos raisonnements. J’examine dans cet article à quelles conditions une analyse conceptuelle des propriétés sémantiques des mots logiques permet d’éclairer et de compléter une analyse conceptuelle des propriétés distinctives de la relation de conséquence logique.Logic is a normative theory of reasoning, which aims at characterizing the class of deductively valid arguments by determining whether a given conclusion logically follows from certain premisses. According to the standard semantic definition, the characterization of the relation of logical consequence depends itself on a characterization of the class of the so-called logical constants, those words, such as « not », « and », « all » or « some » which provide the logical structure of our arguments. In this paper, I shall make precise the conditions under which a conceptual analysis of the semantic properties of logical constants can complete, and shed light upon, a conceptual analysis of the distinctive properties of logical consequence

La logique IF prétend constituer une alternative à la logique classique du premier ordre : en libéralisant les schémas de dépendance entre quantificateurs, elle mènerait à leur terme les idées sous-jacentes à la logique classique. Mais les jeux de Hintikka ne constituent pas la seule manière possible de fournir une sémantique pour l’indépendance : on pourrait au contraire vouloir le faire dans le cadre d’une sémantique récursive avec des quantificateurs de Henkin. Nous présentons ici quelques arguments techniques et philosophiques en faveur de IF, en montrant pourquoi son concept d’indépendance, élargi aux connecteurs, peut prétendre être pleinement général, et en montrant en quel sens la logique IF traite l’indépendance de manière analytique. Ce dernier point est réalisé à travers une explicitation du contenu épistémique de IF, sous la forme d’une traduction partielle dans la logique modale.

La vision par ordinateur est un des domaines de l’intelligence artificielle qui connaît les succès les plus fulgurants. Depuis une vingtaine d’années, les machines n’ont cessé de progresser dans leur capacité à extraire des informations à partir d’images et à identifier des objets. Mais faut-il en conclure que ces machines sont littéralement des machines voyantes, ou ne s’agit-il que d’une façon imagée de décrire des capacités de détection? Le présent article se propose de fournir les bases d’une réponse raisonnée à cette question. Nous examinons successivement trois types de conditions d’attribution de la vision, et soutenons que le critère fonctionnaliste, en termes de représentations internes de stimuli visuels exploitables à diverses fins, doit être préféré aux critères behavioriste ou phénoménologique. Ce critère fonctionnaliste est ensuite appliqué à une des technologies les plus utilisées dans le domaine de la vision par ordinateur, celle des réseaux de neurones et plus particulièrement des réseaux convolutifs. Nous soutenons alors que de tels réseaux sont bien capables de voir, précisément dans la mesure où ils sont entraînés pour produire des représentations à partir de données visuelles. Ces représentations ont des traits spécifiques associés à un entraînement réussi : elles sont hiérarchiques, versatiles et robustes. De telles propriétés ont un intérêt théorique au-delà de leur domaine d’origine : nous suggérons pour finir qu’elles pourraient être utilisées pour spécifier la nature de la fonction de représentation et résoudre certains des problèmes classiques auxquels font face les théories téléosémantiques de la représentation comme celle de Dretske.

Quine est célèbre pour sa critique de la notion d’analyticité, mais il en a également proposé des substituts définissables en termes behavioristes. Cet article examine la question de savoir si de tels substituts peuvent ou non jouer un rôle épistémologique, en les comparant avec des tentatives récentes de réhabilitation de l’a priori. Il apparaît que la caractérisation de ce qu’est une définition acceptable en termes behavioristes est cruciale, et qu’un élargissement de la classe des comportements linguistiques pertinents peut ouvrir la voie à une réhabilitation substantielle de l’analyticité.Quine is famous for his critics against the notion of analyticity used in epistemology and philosophy of science by philosophers of the Vienna circle. However, Quine has also proposed alternative definitions of the notion, which would be acceptable from a behaviorist perspective. In this paper, I address the question whether such alternative definitions can play any epistemologically significant role, by comparing them with recent attempts at a new defense of a priori knowledge. The upshot is that the answer cruccially depends on what is an acceptable definition in a behaviorist perspective. It might well be that an epistemologically significant notion of analyticity can be based upon a broad enough characterization of linguistic behavior

Hintikka makes a distinction between two kinds of games: truthconstituting games and truth-seeking games. His well-known game-theoretical semantics for first-order classical logic and its independence-friendly extension belongs to the first class of games. In order to ground Hintikka’s claim that truth-constituting games are genuine verification and falsification games that make explicit the language games underlying the use of logical constants, it would be desirable to establish a substantial link between these two kinds of games. Adapting a result from Thierry Coquand, we propose such a link, based on a slight modification of Hintikka’s games, in which we allow backward playing for ∃loïse. In this new setting, it can be proven that sequent rules for first-order logic, including the cut rule, are admissible, in the sense that for each rule, there exists an algorithm which turns winning strategies for the premisses into a winning strategy for the conclusion. Thus, proofs, as results of truth-seeking games, can be seen.

La logique est un compagnon naturel de la philosophie. Qu'est-ce qu'un raisonnement correct? Qu'est-ce qu'une preuve? Peut-on definir le concept de verite? Que faire face aux paradoxes? Ces questions sont debattues par les philosophes depuis l'Antiquite; et la logique moderne, usant de langages formels, developpe une analyse rigoureuse de ces concepts les plus fondamentaux. Les onze textes classiques reunis ici proposent un retour reflexif sur cette discipline et sur la signification philosophique de ses achevements. Ils s'adressent a quiconque souhaite prendre la mesure des enjeux conceptuels de la logique, et a tous les etudiants desireux de completer leur apprentissage de la discipline par une reflexion epistemologique sur ses fondements.

L’A. revient, dans le prolongement de son ouvrage L’anthropologie logique de Quine (qui plaçait l’accent sur la traduction, la logique et le principe de charité chez Quine) sur un élément souvent négligé de la thèse d’indétermination de la traduction, à savoir la reconstruction/identification des connecteurs logiques par l’assentiment et le dissentiment. Après avoir examiné diverses objections à l’utilisation par Quine des « fonctions de verdict » dans la traduction, l’A. en vient à la véritable motivation de Quine dans son recours à l’assentiment, qui émerge dans l’ouvrage méconnu the roots of reference. Quine revient, par cet usage de l’assentiment, à la question de l’analyticité et à sa critique des « Deux dogmes », qu’il nuance, dans son oeuvre tardive, par la suggestion d’une analyticité non seulement naturalisée, mais socialisée.