I argue that to know that a proposition is true one must have justification for being certain that the proposition is true. That is, one must have infallible epistemic justification for believing the proposition. It is widely accepted among epistemologists that we rarely, if ever, have such strong justification for our beliefs. It follows that there is precious little that we know. That conclusion is unacceptable to many philosophers. I argue that the positions that lead to the skeptical conclusion are well-motivated and that the skeptical conclusion is implicitly accepted by ordinary speakers. My dissertation has three main components: a metaphilosophical position, an epistemological position, and an error theory. First, the metaphilosophical position. One very important part of philosophy is the analysis of our ordinary concepts. Analysis of our concepts begins with reflecting on what we are inclined to say about various actual and possible cases when considered under specific descriptions. This traditional method has recently come under attack, due in part to the rise of semantic externalism. I agree with externalists that if ‘meaning is reference’ then there is little reason to think reflecting on our concepts from the armchair will provide insight to the nature of the concepts we investigate. I defend a version of semantic internalism which grounds meaning in factors with which subjects are directly acquainted. That view supports the traditional methodology. Furthermore, as the goal of philosophical analysis is to accurately describe concepts of philosophical interest, the only kind of objection that could be decisive against a proposed analysis is that it does not correctly describe our concept. That opens the door to a skeptical analysis of knowledge. Second, the epistemological position. I argue that the unacceptability of sentences of the form “S knows that p but it is possible for S that not-p” is best explained by the hypothesis that our concept of knowledge requires having justification for being certain that what one believes is true. I offer as a criterion of justified certainty the idea that when one knows a proposition is true, one is in a position to decisively answer questions about one’s knowledge of that proposition. I survey a number of competing theories of knowledge and show that they allow for the possibility of knowledge when one fails my criterion of decisive answerability. Those views fail my criterion because according to those views there is nothing the subject is aware of that guarantees for the subject that the allegedly known proposition is true. On that basis, I contend that knowledge is direct awareness of the factors that constitute the truth of the proposition one believes. Third, the error theory. Of course, we rarely have direct awareness of the factors that constitute the truth of the propositions we believe. So, our knowledge attributions are generally false. Yet, they are overwhelmingly natural to make. I argue that competent speakers are often quick to recognize knowledge attributions as a kind of ‘loose talk’ akin to the way we loosely ascribe geometric properties to ordinary objects. We regularly call objects ‘square’ that are not even close to being square, and we are quick to recognize this when challenged. I argue that we do this because we are so accustomed to communicating using strict language to make false claims that we often do not attend to the literal falsity of what we are saying. While we accept very demanding standards for knowledge, the phenomenon of recognizing that our knowledge claims are regularly false is as pervasive as our use of loose speech.

This work has four aims: (i) to provide an overview of the current debate about the semantics of knowledge attributions, i.e. sentences of the form ⌜S knows that Φ⌝; (ii) to ground the debate in a single semantic-pragmatic framework; (iii) to identify a methodology for describing the semantics of knowledge attributions; (iv) to go some way towards describing the semantics of knowledge attributions in light of this methodology, and in particular to defend moderate invariantist semantics against its main current rivals. Aims (i) and (ii) are largely clarificatory; in §1 I set out a single semantic-pragmatic framework and over the course of this work show that it can be modified to explain and distinguish the various theories of the semantics of knowledge attributions currently on offer. Aim (iii) is also met in §1. I argue that a theory of the semantics of knowledge attributions T must be able to account for at least some ordinary speakers’ intuitions about the felicity or infelicity of utterances of the sentence ⌜S knows that Φ⌝ (felicity intuitions) purely in terms of its semantics. I also identify a number of theoretical considerations about knowledge and argue that if T conflicts with any one of these considerations, we should presume that T is false. Aim (iv) is met over the course of this work. According to moderate invariantism ⌜S knows that Φ⌝ is true if and only if S confidently believes the proposition expressed by , this proposition is true and S’s epistemic position with respect to this proposition meets a moderately high epistemic standard. In §§2 – 5 I argue that the main current rivals to moderate invariantism – attributor contextualism, contrastivism, subject-sensitive invariantism and assessor relativism – conflict with at least one of the theoretical considerations identified in §1. In §6 I argue that moderate invariantism accounts for some ordinary speakers’ felicity intuitions purely in terms of the semantics of ⌜S knows that Φ⌝; I also argue that it is consistent with all of the theoretical considerations identified in §1. Moreover, in §§2 – 6 I argue that no theory is capable of accounting for all felicity intuitions purely in terms of the semantics of ⌜S knows that Φ⌝, and that only moderate invariantism can successfully explain why speakers have all of these intuitions. In §7 I conclude that moderate invariantism correctly describes of the semantics of knowledge attributions, or at least does so better than its main current rivals.

Ce mémoire présente une analyse critique de la solution contextualiste de David Lewis au problème sceptique. Pour Lewis, il s’agit d’un problème de possibilités d’erreur à éliminer. En effet, puisque l’on ne peut pas éliminer diverses hypothèses sceptiques, il semble que la connaissance soit impossible. Ce travail est d’abord une présentation du problème sceptique et de la solution que Lewis lui apporte. On verra que son objectif est de rendre compte de la plausibilité de l’argument sceptique tout en protégeant certaines intuitions au sujet de la connaissance. Ainsi, la théorie de Lewis explique quelles sont les présuppositions qu’il est légitime d’entretenir dans les attributions de connaissance. Ces attributions sont les énoncés de type « S sait que P ». On verra que cela prend la forme d’une théorie contextualiste des possibilités d’erreur pertinentes. Ce travail vise ensuite à situer la thèse de Lewis dans la littérature sur le contextualisme épistémique, c’est-à-dire en général, l’idée selon laquelle le contexte de conversation joue un rôle dans la détermination de la valeur de vérité d’une attribution de connaissance. Nous verrons que les solutions contextualistes au problème sceptique peuvent avoir des résultats significativement différents. Enfin, ce travail présentera des critiques importantes sur Elusive Knowledge. Elles prendront la forme d’un retour sur la manière dont Lewis conçoit le problème du scepticisme et celui de l’élimination des possibilités d’erreur. / This thesis is a critical analysis of David Lewis’s contextualist solution to the skeptical problem. The skeptical problem, for Lewis, is a problem of relevant possibilities of error : given that we cannot rule out various skeptical hypotheses, it seems that knowledge is impossible. The first chapter introduces Lewis’s solution to skepticism. We’ll see that his aim is to account for the plausibility of the skeptical argument, while at the same time to protect some of our intuitions concerning knowledge. To that end, Lewis’s solution points to the specific possibilities which can be properly ignored in our knowledge attributions. Knowledge attributions are sentences like « S knows that P. » We’ll see that this explanation takes the form of a contextualist theory about relevant possibilities of error. The second chapter’s task is to locate Lewis’s theory within the literature on epistemic contextualism. Epistemic contextualism, in general, is the claim that the truth value of knowledge attributions is strongly determined by the context of the conversation. But, as this chapter should emphasize, several solutions to skepticism may share the « contextualist » label, while yielding significantly different results. The final chapter will address some important objections to Elusive Knowledge. They are related to the reasons why, according to Lewis, skepticism is a genuine problem for epistemology; on the other hand, they question the notion of elimination of error possibilities.

This dissertation is about the nature of perspectival thoughts and the context-sensitivity of the language used to express them. It focuses on two kinds of perspectival thoughts: ‘subjective' evaluative thoughts about matters of personal taste, such as 'Beetroot is delicious' or 'Skydiving is fun', and first-personal or de se thoughts about oneself, such as 'I am hungry' or 'I have been fooled.' The dissertation defends of a novel form of relativism about truth - the idea that the truth of some (but not all) perspectival thought and talk is relative to the perspective of an evaluating subject or group. In Part I, I argue that the realm of ‘subjective' evaluative thought and talk whose truth is perspective-relative includes attributions of knowledge of the form 'S knows that p.' Following a brief introduction (chapter 1), chapter 2 presents a new, error-theoretic objection against relativism about knowledge attributions. The case for relativism regarding knowledge attributions rests on the claim that relativism is the only view that explains all of the empirical data from speakers' use of the word "know" without recourse to an error theory. In chapter 2, I show that the relativist can only account for sceptical paradoxes and ordinary epistemic closure puzzles if she attributes a problematic form of semantic blindness to speakers. However, in 3 I show that all major competitor theories - forms of invariantism and contextualism - are subject to equally serious error-theoretic objections. This raises the following fundamental question for empirical theorising about the meaning of natural language expressions: If error attributions are ubiquitous, by which criteria do we evaluate and compare the force of error-theoretic objections and the plausibility of error attributions? I provide a number of criteria and argue that they give us reason to think that relativism's error attributions are more plausible than those of its competitors. In Part II, I develop a novel unified account of the content and communication of perspectival thoughts. Many relativists regarding ‘subjective' thoughts and Lewisians about de se thoughts endorse a view of belief as self-location. In chapter 4, I argue that the self-location view of belief is in conflict with the received picture of linguistic communication, which understands communication as the transmission of information from speaker's head to hearer's head. I argue that understanding mental content and speech act content in terms of sequenced worlds allows a reconciliation of these views. On the view I advocate, content is modelled as a set of sequenced worlds - possible worlds ‘centred' on a group of individuals inhabiting the world at some time. Intuitively, a sequenced world is a way a group of people may be. I develop a Stalnakerian model of communication based on sequenced worlds content, and I provide a suitable semantics for personal pronouns and predicates of personal taste. In chapter 5, I show that one of the advantages of this model is its compatibility with both nonindexical contextualism and truth relativism about taste. I argue in chapters 5 and 6 that the empirical data from eavesdropping, retraction, and disagreement cases supports a relativist completion of the model, and I show in detail how to account for these phenomena on the sequenced worlds view.