This thesis explores the extent to which the doxastic ‘should' is information-sensitive and the implications of this for a number of debates in normative epistemology. The doxastic ‘should' is a special case of the deontic modal ‘should' and occurs in sentences such as ‘You shouldn't believe everything you read online'. In the recent semantics literature, it has been suggested that the deontic ‘should' is information-sensitive, meaning that sentences of the form ‘S should do A' are relativized to information-states. After a short introductory chapter, I survey the relevant semantics literature in chapter 2 and provide a simplified contextualist semantics for the doxastic ‘should', according to which the truth-conditions of sentences containing the doxastic ‘should' vary with the information-state provided by their context of utterance. In chapters 3 to 6, I discuss the different kinds of information-states the doxastic ‘should' can be relativized to and how the respective relativization matters for normative epistemology. Chapter 3 argues that the doxastic ‘should' has a subjective and an objective sense and that this distinction solves the apparent conflict between subjective epistemic norms and the truth norm for belief. Chapter 4 addresses the question of how one should react to misleading higher-order evidence. I propose that two seemingly opposing views on this issue, Steadfastness and Concilliationism, are both correct. In a sense of ‘should' that is relativized to one's first-order evidence, one should remain steadfast in the face of misleading higher-order evidence, but in another sense, which is relativized to one's higher-order evidence, one shouldn't. In chapters 5 and 6, I argue that when we advise others on what they should believe, we talk about what they should believe in light of their and our joint evidence. Chapter 7 concludes this thesis with a defence of contextualist semantics for the doxastic ‘should' against truth-relativist challenges.

Notre travail s'attache à examiner la spécificité ainsi que la pertinence du mode d'explication cognitiviste des croyances, tel qu'il est défini par Raymond Boudon, à travers, tout d'abord, une analyse de la notion de "rationalité épistémique" et un examen de certaines des objections qui peuvent lui être faites. Nous interrogeons ensuite la compatibilité de ce type d'explication avec l'idée, propre à un certain nombre de travaux en sciences sociales, selon laquelle les intérêts sociaux et/ou les mécanismes fonctionnels sont aptes à expliquer la formation des croyances. Enfin, nous réfléchirons à la question de savoir si la présence éventuelle de facteurs explicatifs mixtes doit nous inciter à repenser la question de la rationalité des croyances, c'est-à-dire ici de la contribution explicative des raisons au sein de leurs modes de formation. Cette interrogation nous mènera alors à une tentative d'élucidation de la notion de "contexte épistémique" ainsi qu'à un examen de sa pertinence explicative. / My work aims at examining the specificity and the relevance of what Raymond Boudon calls 'the cognitivist model of belief explanation', through an analysis of the notion of 'epistemic rationality' and an examination of some objections against it. Then I ask the compatibility between this kind of explanation with the common idea in social science that social interests and functional mechanisms are relevant to explain the belief formation process. Finally, I will study the question of the influence of the plurality of explanative factors on the rationality of social agent's beliefs. This interrogation will lead us to an elucidation of the notion of 'epistemic context' and its explanative relevance.

No description available.

The ability to change one's beliefs in a rational manner is one of many facets of the abilities of an intelligent agent. Central to any investigation of belief change is the notion of an epistemic state. This dissertation is mainly concerned with three issues involving epistemic states: 1. How should an epistemic state be represented? 2. How does an agent use an epistemic state to perform belief change? 3. How does an agent arrive at a particular epistemic state? With regard to the first question, note that there are many different methods for constructing belief change operations. We argue that semantic constructions involving ordered pairs, each consisting of a set of beliefs and an ordering on the set of "possible worlds" (or equivalently, on the set of basic independent bits of information) are, in an important sense, more fundamental. Our answer to the second question provides indirect support for the use of semantic structures. We show how well-known belief change operations and related structures can be modelled semantically. Furthermore, we introduce new forms of belief change related operations and structures which are all defined, and motivated, in terms of such semantic representational formalisms. These include a framework for unifying belief revision and nonmonotonic reasoning, new versions of entrenchment orderings on beliefs, novel approaches to withdrawal operations, and an expanded view of iterated belief change. The third question is. one which has not received much attention in the belief change literature. We propose to extract extra-logical information from the formal representation of an agent's set of beliefs, which can then be used in the construction of epistemic state. his proposal is just a first approximation, although it seems to have the potential for developing into a full-fledged theory. / Computing / D.Phil.(Computer Science)

The ability to change one's beliefs in a rational manner is one of many facets of the abilities of an intelligent agent. Central to any investigation of belief change is the notion of an epistemic state. This dissertation is mainly concerned with three issues involving epistemic states: 1. How should an epistemic state be represented? 2. How does an agent use an epistemic state to perform belief change? 3. How does an agent arrive at a particular epistemic state? With regard to the first question, note that there are many different methods for constructing belief change operations. We argue that semantic constructions involving ordered pairs, each consisting of a set of beliefs and an ordering on the set of "possible worlds" (or equivalently, on the set of basic independent bits of information) are, in an important sense, more fundamental. Our answer to the second question provides indirect support for the use of semantic structures. We show how well-known belief change operations and related structures can be modelled semantically. Furthermore, we introduce new forms of belief change related operations and structures which are all defined, and motivated, in terms of such semantic representational formalisms. These include a framework for unifying belief revision and nonmonotonic reasoning, new versions of entrenchment orderings on beliefs, novel approaches to withdrawal operations, and an expanded view of iterated belief change. The third question is. one which has not received much attention in the belief change literature. We propose to extract extra-logical information from the formal representation of an agent's set of beliefs, which can then be used in the construction of epistemic state. his proposal is just a first approximation, although it seems to have the potential for developing into a full-fledged theory. / Computing / D.Phil.(Computer Science)

Dans les logiques épistémiques, la connaissance est généralement modélisée par un graphe de mondes possibles, qui correspondent aux alternatives à l'état actuel du monde. Ainsi, les arêtes entre les mondes représentent l'indistinguabilité. Connaître une proposition signifie que cette proposition est vraie dans toutes les alternatives possibles. Les informaticiens théoriques ont cependant remarqué que cela a conduit à plusieurs problèmes, à la fois intuitifs et techniques : plus un agent est ignorant, plus elle a d'alternatives à examiner ; les modèles peuvent alors devenir trop grands pour la vérification de système. Ils ont récemment étudié comment la connaissance pourrait être réduite à la notion de visibilité. Intuitivement, l'idée de base est que quand un agent voit quelque chose, alors elle sait sa valeur de vérité. A l'inverse, toute combinaison de valeurs de vérité des variables non observables est possible pour l'agent. Ces informations d'observabilité permettent de reconstituer la sémantique standard de la connaissance : deux mondes sont indistinguables pour un agent si et seulement si chaque variable observée par cet agent a la même valeur dans les deux mondes. Notre objectif est de démontrer que les logiques épistémiques fondées sur la visibilité constituent un outil approprié pour plusieurs applications importantes dans le domaine de l'intelligence artificielle. Dans le cadre actuel de ces logiques de visibilité, chaque agent a un ensemble de variables propositionnelles qu'elle peut observer ; ces visibilités sont constantes à travers le modèle. Cela accompagne une hypothèse forte : les visibilités sont connues de tous, et sont même connaissance commune. De plus, la construction de la connaissance à partir de la visibilité entraîne des validités contre-intuitives, la plus importante étant que l'opérateur de la connaissance distribue sur les disjonctions de littéraux : si un agent sait que p ou q est vrai, alors elle sait que p est vrai ou que q est vrai, parce qu'elle peut les voir. Dans cette thèse, nous proposons des solutions à ces deux problèmes et les illustrons sur diverses applications telles que la planification épistémique ou les jeux booléens épistémiques, et sur des exemples plus spécifiques tels que le problème des enfants sales ou le problème du bavardage. Nous étudions en outre des propriétés formelles des logiques que nous concevons, fournissant axiomatisations et résultats de complexité. / In epistemic logic, knowledge is usually modelled by a graph of possible worlds, representing the alternatives to the current state of the world. So edges between worlds stand for indistinguishability. To know a proposition means that that proposition is true in all possible alternatives. Theoretical computer scientists however noticed that this led to several issues, both intuitively and technically: the more an agent is ignorant, the more alternatives she must consider; models may then become too big for system verification. They recently investigated how knowledge could be reduced to the notion of visibility. Intuitively, the basic idea is that when an agent sees something, then she knows its truth value. The other way round, any combination of truth values of the non-observable variables is possible for the agent. Such observability information allows us to reconstruct the standard semantics of knowledge: two worlds are indistinguishable for an agent if and only if every variable observed by her has the same value in both worlds. We aim to demonstrate that visibility-based epistemic logics provide a suitable tool for several important applications in the field of artificial intelligence. In the current settings of these logics of visibility, every agent has a set of propositional variables that she can observe; these visibilities are constant across the model. This comes with a strong assumption: visibilities are known to everyone, and are even common knowledge. Moreover, constructing knowledge from visibility brings about counter-intuitive validities, the most important being that the knowledge operator distributes over disjunction of literals: if an agent knows that p or q is true, then she knows that p is true or that q is true because she can see them. In this thesis, we propose solutions to these two problems and illustrate them on various applications such as epistemic planning or epistemic boolean games, and on more specific examples such as the muddy children problem or the gossip problem. We moreover study formal properties of the logics we design, providing axiomatizations and complexity results.

The quest for epistemic liberation remains an important project in the post-colonial era of Zambia, and entails challenging the existing inequalities in knowledge representation at the epistemic front in the country. At the core of this quest is the position that the continued dominance of Western knowledge and the corresponding marginalisation of indigenous knowledges amount to an epistemic injustice that affects the contemporary existence of the peoples in the country. This study critically examines the problem of epistemic injustice in Zambia while reflecting on the country’s uncompleted project of decolonisation. It traces the problem from the theoretical assumption of modernity that Western knowledge is universal and that it should, therefore, be applied to all societies in the world. It is submitted that the current education system in Zambia is based on this assumption, and, consequently, favours Western knowledge to the exclusion of indigenous knowledges. This practice is identified as a conduit for accelerating epistemic injustice and its intensity in the country. The study approaches this problem from an African philosophical standpoint, and draws its current from the history of the political struggle against domination on the continent. To adequately confront the problem of epistemic injustice in Zambia, the study suggests parity and equilibrium in representation between indigenous knowledges and Western knowledge in the country. / Philosophy, Practical and Systematic Theology / M.A. (Philosophy)

No description available.

This dissertation investigates the expression of the indirect evidential subdomain in two languages in contact, i.e. the northern variety of Central Aymara and the variety of Spanish spoken in La Paz (Bolivia). For this aim, the study uses first-hand data collected in La Paz and El Alto (Bolivia) during 2014 and 2015. Data was elicited through: the “Family Problems Picture” task (San Roque et al. 2012), formulated by the members of the Max Planck Institute for Psycholinguistics and created specifically for the activation of cognitive categories such as evidentiality and mirativity; the “Pear Story” designed for Wallace Chafe, professor at the University of California, to collect narrative texts that show how humans perceive, elaborate and verbalize experience; and, finally, personal narratives, traditional narratives and interviews. Thirty-three recordings (12h 48’) of 48 Spanish-Aymara bilingual speakers (17 males, age range: 18-64) were fully transcribed and annotated. The resulting corpus consists of 33 transcriptions of which 14 are in Aymara (c. 19 154 words), whereas 19 are in Spanish (c. 46 245 words). The dissertation is built around four research questions. First, the dissertation shows the functions of the forms identified in the data in both languages. The study identifies for each form both evidential and non-evidential functions. Indirect evidential functions are systematically analyzed and classified by combining Willett’s (1988) and Aikhnvald’s (2004) classifications. The analysis shows evidential functions of forms that have not been previously studied as such, i.e. digamos and diciendo in Spanish and sañani and sapxi in Aymara, but it also reveals unnoticed evidential functions for previously described forms. Second, the dissertation provides a clear view of the relationship between the evidential and the epistemic modal domain involved in the use of the forms identified. Two types of correlation are found. Both languages, indeed, show forms that only point out the way in which speakers acquired information and forms where the two domains overlap. Third, the dissertation investigates speakers’ epistemic stance, in terms of commitment, towards information involved in the use of the evidential forms identified. The study shows that the forms which convey merely evidential information express mainly a medium-high commitment degree, whereas the forms in which the distinction between the evidential and the epistemic modal domain is blurred indicate a low degree of commitment. Forth, the dissertation sheds light on the relationship between the expressions of the indirect evidential subdomain in the two languages. The study proposes a quantitative and qualitative analysis of the evidential types and subtypes in both languages. The results show a high degree of convergence between the two languages, suggesting also situations of influence of one language on the other.

This thesis examines some epistemic defences of democracy put forward by David Estlund, Michael Fuerstein, Cheryl Misak, and Fabienne Peter, as well as a critique of democracy raised by Jason Brennan. It then develops an epistemic defence of a moderately non-egalitarian system, which it proposes to call liberal trusteeship. According to the proposed theory, the power to draft laws ought to be separated from the power to enact those drafts into law. The former power ought to be vested in trustees, who are essentially specialists that have inquired extensively into a given matter, and the latter power ought to be vested in a democratically elected parliament. Subsequently, this thesis argues that parliament should nevertheless have the prerogative to ultimately override trustees on ethics and pass its own legislation regulating moral matters; that the criteria for selecting trustees should be determined by jury courts; and that parliament and jury courts should be given some powers to influence the composition of trustee committees, so that the political process can guard against the risk that trustees might be biased or corrupt. The above proposal is grounded on three principal claims. Firstly, this thesis argues that moral authority and legitimacy ought to be reserved for the political system that strikes the best balance between competence and equality. Secondly, it argues that liberal trusteeship is more likely than democracy to determine correctly what ought to be done in light of the progress of open and vigorous inquiry into a given matter. Thirdly, and as a result, it argues that liberal trusteeship is likely to exercise power sufficiently more competently than democracy, such that its moderate deviation from political equality will be justified. In the light of this, the thesis concludes that liberal trusteeship would strike a better balance between competence and equality than democracy.