Anandi Hattiangadi

Polarisation is a widespread societal issue, dividing populations into opposing groups whose beliefs may span multiple, seemingly unrelated, topics. This phenomenon creates distinct cultural groups with internally consistent yet mutually opposing beliefs, strengthening group identities and deepening societal divides. We develop a mathematical model to investigate how polarisation and emergent associations between unrelated beliefs occur. Unlike prior models assuming irrationality, informational segregation, or similarity biases, our model employs an epistemic distinction between two filtering mechanisms: content and source filtering. In content filtering, agents evaluate beliefs based on intrinsic coherence, rejecting information opposing their current beliefs. Source filtering, in contrast, involves adopting beliefs based on the perceived credibility of information sources, where credibility corresponds to mutual consistency between individuals’ belief systems. We further incorporate external signals through varying innovation and loss rates to reflect empirical biases from real-world data. Simulations show that source filtering rapidly clusters beliefs into two dominant, highly polarised groups. Within each group, beliefs become tightly correlated, creating strong opposition between the groups across all cultural traits. Thus, new associations emerge spontaneously between previously unrelated beliefs. Arbitrary and initially independent beliefs transform into ideological bundles serving as clear group signals. External signals further reinforce these associations, allowing advantages of particular beliefs (e.g. empirical evidence) to transfer to associated beliefs. Content filtering can also produce polarised groups, but beliefs remain aligned with preexisting logical or factual structures. Source filtering mirrors algorithmic biases such as collaborative filtering prevalent on digital platforms, where content exposure depends heavily on shared user preferences rather than intrinsic coherence. Our findings illustrate how these biases amplify polarisation by reinforcing homogeneous belief packages, intensifying ideological divides within cultural systems.

According to doxastic normativism, belief is essentially governed by a norm of correctness, which states, ‘Your belief that is correct if and only if is true’. The normativist further holds that this condition of correctness has normative import, meaning that it entails claims about what one ought to believe. Bykvist and Hattiangadi argue that at least one of these two claims made by the normativist is false; that is to say, having truth as its condition of correctness is either not constitutive of belief, or does not have normative import. They argue that ‘correct’ is not an essentially normative term, and that the purported norm of truth fails to distinguish belief from other propositional attitudes. Moreover, no version of normativism can avoid the unacceptable implication that we ought to believe blindspot propositions, which are true but nevertheless could not be truly believed.

This chapter investigates what we disagree about when we disagree about logic, on the assumption that judgments of logical validity are normative. If logic is normative, then the popular anti-realist thesis that there are no normative facts or properties generalizes—it entails that there are no logical facts or properties. When faced with this anti-realism, it is tempting to endorse a pluralist thesis, according to which two people who disagree about the validity of an argument can both say something true. This chapter explores the limitations of three prominent forms of pluralism: contextualism, relativism, and expressivism. It argues that none of these forms of pluralism gives an adequate account of what we disagree about when we disagree about logic.

This chapter discusses the application of formal methods from social choice theory to the metasemantic question of whether radical interpretation is possible. Radical interpretation involves deducing semantic truths from non-semantic truths by appeal to certain a priori principles or criteria, such as the principle of charity. A familiar view is that the intended interpretation is the one that best meets a combination of constraints. It is suggested that this situation can be modelled as follows: each constraint determines a binary relation on the set X of interpretations (‘_x_ is at least as good as _y_ with respect to the _i_th constraint’) that is transitive and complete. The radical interpreter’s task is to determine an overall ordering as a function from the profile of individual orderings. The application of Arrow’s theorem in this context is discussed.

In his recent book, _Meaning and Normativity_, Allan Gibbard argues at length that meta-ethical expressivism can be profitably extended to semantic and intentional language: meta-linguistic discourse about meaning, reference, content, and the like. This chapter argues that the extension of expressivism to semantic discourse is unprofitable and—worse still—in a certain sense self-undermining. It is unprofitable because it sheds no light on the problem of intentionality; and it undermines itself because many of the sentences that make up the expressivist’s theory are semantic sentences, and if these are understood to express non-cognitive attitudes of some kind, the expressivist’s explanations are spurious.

In this paper, we offer a simple argument for the conclusion that the outputs of large language models (LLMs), such as OpenAI’s ChatGPT and Google’s Gemini, are meaningless. Our argument is based on two key premises: (a) that certain kinds of intentions are needed in order for LLMs’ outputs to have literal meanings, and (b) that LLMs cannot plausibly have the right kinds of intentions. We defend this argument from various types of responses, for example, the semantic externalist argument that deference can be assumed to take the place of intentions and the semantic internalist argument that meanings can be defined purely in terms of intrinsic relations between concepts, such as conceptual roles. We conclude the paper by discussing why, even if our argument is sound, the outputs of LLMs nevertheless seem meaningful and can be used to acquire true beliefs and even knowledge.

In Making it Explicit, Brandom aims to articulate an account of conceptual content that accommodates its normativity—a requirement on theories of content that Brandom traces to Wittgenstein's rule following considerations. It is widely held that the normativity requirement cannot be met, or at least not with ease, because theories of content face an intractable dilemma. Brandom proposes to evade the dilemma by adopting a middle road—one that uses normative vocabulary, but treats norms as implicit in practices. I argue that this proposal fails to evade the dilemma, as Brandom himself understands it. Despite his use of normative vocabulary, Brandom's theory fares no better than the reductionist theories he criticises. I consider some responses that Brandom might make to my charges, and finally conclude that his proposal founders on his own criteria.

This book provides a response to the argument for meaning scepticism set out by Saul Kripke in Wittgenstein on Rules and Private Language. Kripke asks what makes it the case that anybody ever means anything by any word, and argues that there are no facts of the matter as to what anybody ever means. Kripke's argument has inspired a lively and extended debate in the philosophy of language, as it raises some of the most fundamental issues in the field: namely, the reality, privacy, and normativity of meaning. The book argues that in order to achieve the radical conclusion that there are no facts as to what a person means by a word, the sceptic must rely on the thesis that meaning is normative, and that this thesis fails. Since any ‘sceptical solution’ to the sceptical problem is irremediably incoherent, the book concludes that there must be a fact of the matter about what we mean. In addition to providing an overview of the debate on meaning and content scepticism, this book presents a detailed discussion of the contributions made by Simon Blackburn, Paul Boghossian, Robert Brandom, Fred Dretske, John McDowell, and Crispin Wright, among others, to the controversy surrounding Kripke's argument. The issues considered include the normativity of meaning and its relation to the normativity of moral judgments, reductive and non-reductive theories of meaning, deflationism about truth and meaning, and the privacy of meaning.

In this chapter we explore the topic of logical disagreement. Though disagreement in general has attracted widespread philosophical interest, both in epistemology and philosophy of language, the general issues surrounding disagreement have only rarely been applied to logical disagreement in particular. Here, we develop some of the fascinating semantic and epistemological puzzles to which logical disagreement gives rise. In particular, after distinguishing between different types of logical disagreement, we explore some connections between logical disagreements and deep disagreements over fundamental epistemic principles; we discuss several semantic puzzles that arise on various accounts of the meanings of logical terms; we investigate how such disagreements relate to Kripke’s so-called “Adoption Problem”; and we probe epistemological puzzles that arise from disagreements about logic in the light of central principles from the peer disagreement literature.

In this paper, I present an explanatory gap argument against the view that the semantic facts are fully grounded in the physical facts. Unlike traditional explanatory gap arguments, which stem from the failure of analytic reductive explanation, the explanatory gap I point to stems from the failure of metaphysical explanation. I argue for the following theses. (i) Physicalist grounding claims are metaphysically necessary, if true. (ii) To be explanatorily adequate, these grounding claims must be deducible from facts about essence. (iii) Semantico-physical grounding claims are possibly false, not (only) because they are conceivably false, but because they cannot be deduced from facts about essence. (iv) Semantic properties are essentially weakly normative: it lies in their natures to have correctness conditions and subjectively rationalize—rather than merely cause—behaviour. This gives rise to an explanatory gap that indicates that the semantic facts are not fully grounded in the physical facts.

This paper takes issue with an influential interpretationist argument for physicalism about intentionality based on the possibility of radical interpretation. The interpretationist defends the physicalist thesis that the intentional truths supervene on the physical truths by arguing that it is possible for a radical interpreter, who knows all of the physical truths, to work out the intentional truths about what an arbitrary agent believes, desires, and means without recourse to any further empirical information. One of the most compelling arguments for the possibility of radical interpretation, associated most closely with David Lewis and Donald Davidson, gives a central role to decision theoretic representation theorems, which demonstrate that if an agent’s preferences satisfy certain constraints, it is possible to deduce probability and utility functions that represent her beliefs and desires. We argue that an interpretationist who wants to rely on existing representation theorems in defence of the possibility of radical interpretation faces a trilemma, each horn of which is incompatible with the possibility of radical interpretation.

This chapter investigates the view that meaning is normative. Meaning is understood here in a broad sense to include such semantic properties as sense, reference, truth‐conditions, content, and the like. Normativity can either be viewed as a property of representations or as a feature of the world. The view that meaning involves rule‐following or a normative judgment of some kind is untenable, and in any case, has no bearing on the hard problem of intentionality. However, the view that meaning is a source of normativity is implausible, and there is little evidence that known difficulties with the reductive analysis of meaning can be resolved by adding normativity to the explanation. Finally, the chapter reviews Gibbard's recent suggestion that the concept meaning is normative is implausible and his proposed expressivist resolution of the hard problem of intentionality appears to be untenable.

In this paper, I take issue with a core commitment of logical conventionalism: that we impose a logic on ourselves by adopting general linguistic conventions governing our use of logical terms, thereby determining the meanings of the logical constants and which of our inferences are valid. Drawing on Kripke’s ‘adoption problem’, I argue that general logical principles cannot be adopted, either explicitly or implicitly. I go on to argue that the meanings of our logical terms, and the validity of our inferences, cannot depend on our adoption of logico-linguistic conventions.

John MacFarlane has recently argued that his brand of truth relativism provides the best solution to the puzzle of future contingents: assertions about the future that express propositions that are metaphysically neither necessary nor impossible. In this paper, we show that even if we grant all of the metaphysical, semantic and pragmatic assumptions in terms of which MacFarlane sets and aims to solve the puzzle, his truth relativism is not apt to solve the problem of future contingents. We argue that the theory fails to vindicate the intuition that future contingent propositions are neither true nor false, leaving the theory open to a charge of Reductio. We show that these problems cannot be answered while preserving the core tenets of truth relativism.

It is disputed what norm, if any, governs assertion. We address this question by looking at assertions of future contingents: statements about the future that are neither metaphysically necessary nor metaphysically impossible. Many philosophers think that future contingents are not truth apt, which together with a Truth Norm or a Knowledge Norm of assertion implies that assertions of these future contingents are systematically infelicitous. In this article, we argue that our practice of asserting future contingents is incompatible with the view that they are not truth apt. We consider a range of norms of assertion and argue that the best explanation of the data is provided by the view that assertion is governed by the Knowledge Norm.

In his recent book, Meaning and Normativity, Allan Gibbard argues at length that meta-ethical expressivism can be profitably extended to semantic and intentional language: meta-linguistic discourse about meaning, reference, content, and the like. This chapter argues that the extension of expressivism to semantic discourse is unprofitable and—worse still—in a certain sense selfundermining. It is unprofitable because it sheds no light on the problem of intentionality; and it undermines itself because many of the sentences that make up the expressivist’s theory are semantic sentences, and if these are understood to express non-cognitive attitudes of some kind, the expressivist’s explanations are spurious.

ABSTRACTIn a recent paper, Alexander Greenberg defends a truth norm of belief according to which if one has some doxastic attitude towards p, one ought to believe that p if and only if p is true. He responds, in particular, to the ‘blindspot’ objection to truth norms such as da: in the face of true blindspots, such as it is raining and nobody believes that it is raining, truth norms such as da are unsatisfiable; they entail that one ought to believe p, but if one does believe p, they entail that it is not the case that one ought to believe p. In this paper, it is argued that Greenberg’s response to the blindspot objection is unsatisfactory.