Assessment Sensitivity: Relative Truth and Its Applications
Forthcoming from Oxford University Press in early 2014. Here’s a recent draft. Note that the text may still change slightly as a result of copy-editing. You should not cite or quote it without permission.
This book is about how we might make sense of the idea that truth is relative, and how we might use this idea to give satisfying accounts of parts of our thought and talk that have resisted traditional methods of analysis. Although there is a substantial philosophical literature on relativism about truth, it has tended to focus on refutations of the doctrine, or refutations of these refutations, at the expense of saying clearly what the doctrine is. The aim here is to start by giving a clear account of what it is to be a relativist about truth, and then to use the view to give satisfying accounts of what we mean when we talk about what is tasty, what we know, what will happen, what might be the case, and what we ought to do. The book seeks to provide a richer framework for the description of linguistic practices than standard truth-conditional semantics affords: one that allows not just standard contextual sensitivity (sensitivity to features of the context in which an expression is used), but assessment sensitivity (sensitivity to features of the context from which a use of an expression is assessed).
Chapter 1: A Taste of Relativism
Reflecting on the meaning of “tasty” gives some prima facie motivation for an investigation of truth relativism. Some desiderata for a satisfactory account are proposed, and it is argued that none of the standard approaches—objectivism, contextualism, and classical expressivism—can meet them. A relativist approach is sketched that promises to meet the desiderata, but much work will be required to state it clearly and show it to be coherent.
Part I: Foundations
Chapter 2: The Standard Objections
This chapter surveys the standard objections to relativism about truth: that it is self-refuting, that it makes disagreement impossible, that it is incompatible with the equivalence schema, that propositions cannot be the bearers of relative truth values, and that we do not understand what the relativized truth predicate means. From the discussion of these standard objections, we extract a list of problems that any truth relativist must address. These help set the agenda for Chapters 3-6.
Chapter 3: Assessment Sensitivity
After surveying various ways in which truth relativism has been characterized in the literature, this chapter argues that it is best understood as the view that some of our thought and talk is assessment-sensitive. Assessment sensitivity is understood by analogy with ordinary contextual sensitivity—or, as it is called here, use sensitivity. Just as the truth of uses of ordinary context-sensitive sentences depends on features of the context in which they are used, so the truth of uses of assessment-sensitive sentences depends on features of the context in which they are assessed. We describe how standard semantic frameworks must be changed to make room for assessment sensitivity.
Chapter 4: Propositions
This chapter extends the notion of assessment sensitivity from sentences to propositions. Following Kaplan, we take propositions to have truth values relative to circumstances of evaluation. The issue of which parameters comprise a circumstance of evaluation is shown to be distinct from the issue of assessment sensitivity: in particular, taking propositional truth to be relative to parameters besides possible worlds (and possibly times) is neither necessary nor sufficient for assessment sensitivity. Content relativism is distinguished from truth-value relativism, and truth-value relativism from nonindexical contextualism. Various objections to the idea that propositions can be the bearers of relative truth are considered and rejected. Semantics for the ordinary monadic truth predicate are given, and the equivalence schema is shown to be compatible with assessment sensitivity.
Chapter 5: Making Sense of Relative Truth
This chapter addresses the substantive philosophical question that remains: what does it mean to talk of truth relative to a context of assessment? It does this by explaining how a semantic theory that assigns assessment-relative truth values is connected to norms for assertion and retraction. Relativist and nonindexical contextualist theories make the same predictions about what may be asserted, but differ in their predictions about when previous assertions must be retracted. The combined theory of Chapters 3-5 allows us to formulate relativist semantic theories and derive from them substantive predictions about language use, so that they can be compared with and evaluated against non-relativist alternatives. This is sufficient to have “made sense of relative truth” and warded off apriori objections to its intelligibility.
Chapter 6: Disagreement
Disagreement has been the crux of debates between relativists, contextualists, expressivists, and objectivists. The simple view that there is disagreement just in case one party rejects what the other accepts is rejected. Several varieties of disagreement are distinguished—practical and doxastic noncotenability, preclusion of joint accuracy, and preclusion of joint reflexive accuracy—and it is shown how the choice between the different semantic approaches we have considered can be reduced to an issue about what kind of disagreement there is in the domain under discussion. The notion that there can be “faultless disagreement” is subjected to critical scrutiny.
Part II: Applications
Chapter 7: Tasty
This chapter returns to the motivating example of Chapter 1: the thin evaluative term “tasty.” It is argued that each of the standard views (objectivism, contextualism, and expressivism) is motivated by a genuine insight that it accommodates in a one-sided way, and that a relativist account is needed to accommodate them all. Compositional semantic issues concerning the interaction of “tasty” with tense, modality, quantifiers, attitude verbs, and factive verbs are explored, and differences between the account recommended here and the relativist accounts of Lasersohn, Egan, and Stephenson are highlighted. The relativist view is also compared to an expressivist account, inspired by Gibbard’s work, which uses the same compositional semantics.
Chapter 8: Knows
Invariantist accounts of knowledge attributions force an uncomfortable choice between skepticism, dogmatism, and rejecting closure. Contextualist accounts offer a way out of the trilemma, but yield faulty predictions about disagreement, retraction, and truth attributions. This chapter argues that a relativist account can combine the attractive features of contextualism and invariantism, breaking a deadlock in the contemporary debate. It is argued that the subject-sensitive invariantism advocated by Stanley and Hawthorne, which also tries to find a middle course, gets the wrong dimension of variability. Nonindexical contextualist and expressivist accounts are also considered and rejected. Replies are given to objections to the relativist account by Stanley and Montminy.
Chapter 9: Tomorrow
How should we assign truth values to contingent statements about the future in an objectively indeterministic world? The classical solutions—Peircean, three-valued, Ockhamist + thin red line, Ockhamist + supervaluations—are considered and rejected. It is argued that only a relativist approach, on which future contingents are assessment-sensitive, can satisfy all the desiderata that motivate the various classical views. Objections that the relativist account gives bad predictions about norms for asserting future contingents, and that it cannot explain attitudes of wondering or partial belief with future-directed contents, are developed and answered.
Chapter 10: Might
Epistemic modal sentences like “Joe might be in Boston” are usually thought to say something about what is is possible given what is known. But known by whom? Contextualists hold that the relevant body of knowledge is determined by facts about the context of use. This chapter argues that the contextualist account fits poorly with the way we use epistemic modals, and that a better account takes the relevant body of knowledge to be determined by facts about the context of assessment. The relativist semantics not only explains the puzzle cases, but yields a simpler treatment of the data that motivated the complexities in the standard contextualist semantics. A compositional semantics is given to explain how epistemic modals behave when embedded in conditionals and other constructions, and the account is compared to expressivist alternatives.
Chapter 11: Ought
In the literature on ethics and practical reason, it is widely held that “ought” is ambiguous between objective and subjective senses. In the semantics literature, it is widely held that different uses of “ought” must be evaluated relative to different bodies of information. This chapter argues that a relativist treatment, on which “ought” is contextually sensitive to the information relevant at the context of assessment, can better account for all the uses of “ought” in deliberation and advice, undercutting the motivation for the distinction between subjective and objective senses (or uses) of “ought”. A compositional semantics is given and several puzzles about “oughts” in conditionals are resolved.
Chapter 12: The Rationality of Relativism
One might object that, even if we can understand what it would be for a community to use words with assessment-sensitive meanings, and even if we in fact do use some words that way, such practices are irrational. Once we see what we are committing ourselves to in asserting assessment-sensitive contents, we should give up talking this way and find assessment-invariant alternatives. Taking an engineering perspective, this chapter argues that it makes sense for certain expressions to be assessment-sensitive, given plausible assumptions about their purposes.
What Does it Mean to Say that Logic is Formal?
PhD Dissertation, University of Pittsburgh, 2000. You can download a PDF.
Much philosophy of logic is shaped, explicitly or implicitly, by the thought that logic is distinctively formal and abstracts from material content. The distinction between formal and material does not appear to coincide with the more familiar distinctions between a priori and empirical, necessary and contingent, analytic and synthetic—indeed, it is often invoked to explain these. And although there are clear notions of formality that advert to schematic inference patterns, syntactic rules, and grammar, none of these is capable of demarcating logic. What does it mean, then, to say that logic is distinctively formal?
Three things: logic is said to be formal (or “topic-neutral’’)…
in the sense that it provides constitutive norms for thought as such,
in the sense that it is indifferent to the particular identities of objects, and
in the sense that it abstracts entirely from the semantic content of thought.
Though these three notions of formality are by no means equivalent, they are frequently run together. The reason, I argue, is that modern talk of the formality of logic has its source in Kant, and these three notions come together in the context of Kant’s transcendental philosophy. Outside of this context (e.g., in Frege), they can come apart. Attending to this history can help us to see the sources of our disparate intuitions about logicality, and more importantly to sort these intuitions into central and adventitious ones. I argue that we have largely lost sight of the notion of formality (1) by which logic was demarcated in a central tradition from Leibniz through Frege—the intellectual home of most of the philosophical projects for which it matters how logic is demarcated.
This historical perspective is especially useful in evaluating contemporary debates about the demarcation of logic, which often seem to turn on opposing but equally brute intuitions about logicality. As an illustration, I examine the popular permutation-invariance account of logicality, which is commonly motivated by appeal to sense (2) of formality. I present the account in a way that reveals a hidden lacuna, and I show how this lacuna might be filled by appealing to formality in sense (1).