Amie Thomasson

Metaethics has long included debates about the function of moral discourse. Some have argued that moral statements express our attitudes, others that they serve as prescriptions for how to act, still others that they describe moral facts or properties. This article argues that we can best address these questions about function by attending to work in empirical linguistics. Tracing the development of moral language through different grammatical forms, and noting the additional functions added at each stage of development, enables us to give a more empirically grounded and fine-grained answer to questions about the functions of moral discourse. This analysis shows how multiple functions come to be layered up through different grammatical forms. Interestingly, it also preserves the different grains of truth picked up on by historically “rival” views—including prescriptivist, expressivist, and even descriptivist views. In so doing, it helps develop a comprehensive neo-pragmatist approach to understanding moral discourse.

This book aims to change how we think about metaphysics, what it can do, and why it matters. Traditionally, metaphysics has been presented as aiming to discover deep truths about the world. But this leads to a rivalry with science, epistemological mystery, and a despairing skepticism about how we can gain knowledge in metaphysics. Rethinking Metaphysics (2025) diagnoses the problems with many prior approaches to metaphysics as arising from a problematic assumption that all discourse functions in the same way. By drawing on work in linguistics, it shows how to develop a richer view of linguistic functions that enables us to see why this assumption leads us astray. Better understanding the plurality of linguistic functions enables us to disentangle ourselves from many old metaphysical problems, e.g. about properties, numbers, morality and modality. In place of the traditional model, we should think of metaphysics as work in conceptual engineering—including both reverse engineering projects and constructive engineering projects. Rethinking metaphysics in this way enables us to avoid the problems of traditional metaphysics, while also demonstrating the perennial importance of metaphysics to human life. For how we think and talk is centrally important to how we live, organize our investigations, and run our societies.

This chapter continues the theme of the analytic or a priori status of the causal theory. It is suggested that rather than conceiving of the causal theory as a product of conceptual analysis, it should be considered an exercise in what has become known as conceptual ethics. Just as some behaviourists took their view not to expose the concept they already had, but as a suggestion for conceptual reform, Armstrong’s causal theory might be taken as a suggestion for conceptual reform in the light of scientific and other discoveries which might make one doubt whether there is anything in the world corresponding to some of the pre-reform concepts of the mental. The chapter argues that while Armstrong certainly never explicitly understood what he was up to in these ways, it’s certainly clear that in his later writings he became unhappy with the idea that the causal theory was a straightforward analytic truth about existing concepts—even though he had no clear idea about what the alternative might be.

How ought we to do work in conceptual ethics? Some have thought that conceptual choice should itself be guided by (heavyweight) metaphysics—for we should be sure that our concepts pick out things that exist or should aim to choose concepts that really ‘carve the world at its joints’. An alternative is to take a pragmatic approach to conceptual ethics. But pragmatic approaches are often criticized as unable to account for intuitions that some conceptual choices are objectively better than others, and intuitions that the world is structured. As a result, the fear is that a pragmatic approach leaves conceptual choices arbitrary and insusceptible to critique. This chapter confronts such worries and develops a pragmatic method for conceptual ethics that clearly avoids these problems. As a result, we need not rely on heavyweight metaphysics and become entangled in its epistemological mysteries to do conceptual ethics.

The easy approach to ontology suggested by Stephen Schiffer’s work on pleonastic transformations holds that some disputed existence questions can be answered by trivial inferences from uncontroversial premises. This chapter aims to clarify the relations among different forms of the easy approach and the consequences of the approach. Defenders and critics often assume that the entities we come to accept on the basis of trivial inferences are in some sense “cheap” or “lightweight,” or that the existence claims should be taken as somehow deflated. The author argues, however, that easy arguments actually give us a reason to accept straightforward first-order realism about the entities in question. The approach, however, also leads to a meta-ontological deflationism that shows something to be amiss in hard ontological debates. Properly understood, the easy approach to ontology may present the biggest threat to serious ontology, and the greatest hope for clarifying the epistemology of metaphysics.

After more than fifty years, metaontology is back in fashion. But in most recent discussions, the original Carnapian deflationist position has been missed. How could this be so? And what difference would rediscovering it make to contemporary discussions in metaontology? Those are the questions this chapter aims to answer. It argues that Carnap’s original position was wrongly dismissed by association with verificationism and anti-realism. It was then put aside and forgotten, given the common assumption that Quine had won the Carnap–Quine debate. Ever since, metaontological disputes have largely focused on quantifier variance. Each of these moves went wrong. Taken together, they led the original and most promising deflationary position to be largely overlooked. The chapter concludes with a contemporary neo-Carnapian form of deflationism, arguing that it is largely untouched by recent defenses of serious ontology, and that its prospects are rather promising.

There has been growing interest in the idea that model descriptions should be thought of as similar to stories, and model systems should be thought of as akin to fictional characters. But if model systems were (like) fictional characters, what would they be? Two prominent approaches to fictional discourse have been pursued in the literature on models: realist approaches, which take models to be abstract objects that (in some sense) fit the model descriptions, and anti-realist approaches, which typically hold that the relevant discourse involves pretense in a way that enables us to deny that we ever refer to models. Both of these views have problems well known in the literature on fiction. These problems have motivated a third, increasingly popular approach to the ontology of fiction: an artifactual approach, according to which our (external) discourse about fiction refers to abstract artifacts. This approach has been little considered in the literature on scientific models—but this chapter argues that it has important advantages over the familiar alternatives. Most notably, an artifactualist approach can retain the advantages of the pretense view while giving a far more straightforward account of external historical, theoretical, and critical discourse about models. In short, bearing in mind the full range of discourse about models gives us reason to accept that there are model systems, where these are considered as a kind of abstract artifact. The main perceived drawback to artifactualist views is their supposed “ontological costs.” In closing, the chapter suggests why ontological qualms of this sort should be discounted.

Various forms of deflationism have been defended in various arenas. On the semantic side, there is deflationism about truth, reference, and meaning, all of which are popular, perhaps even orthodox. On the metaphysical side, there is deflationism about existence, and meta‐ontological deflationism about ontological debates. Unlike their semantic counterparts, these views have gone unnoticed or been widely attacked. The purpose of this chapter is to bring discussion of these sorts of deflationism together. It is argued that semantic deflationism leads to a form of existence deflationism. As a result, it is incoherent to accept semantic deflationism but reject existence deflationism. Moreover, it is argued that existence deflationism leads to a form of meta‐ontological deflationism. Finally, it is argued that once we see how these views form part of a global deflationary package, it becomes evident that they are more defensible than they have often seemed when examined in isolation.

The cluster of theses that underpin the reflective common sense worldview defended in this book has significant consequences regarding the proper methods and limits of metaphysics. This chapter argues that given those theses, the metaphysical side of questions about identity and persistence conditions, ontological status, and existence must be addressed by a form of conceptual analysis whose proper methods and limits are very different from those of the empirical sciences. As a result, radically revisionary answers to any of these questions must be met with suspicion. Moreover, if we properly understand what we are doing in asking questions about existence, we must suspect that many apparent debates in ontology are pseudo-debates in which the disputants talk past each other or try to answer unanswerable questions. The book closes with a sketch of some of these important metaontological consequences that fall out of the work of this book.

This chapter begins by bringing out three common mistakes that lie behind the various arguments against ordinary objects discussed in this book. It goes on to suggest that all of these mistakes may be avoided by adopting a unified picture, based on the thesis that our singular and general nominative terms have a basic conceptual content in the form of frame-level conditions of application and coapplication collectively established by competent speakers. It shows how to develop a workable ontology of ordinary objects — one that can avoid all of the standard arguments against such objects — from that basis, and defends the claim that this provides a common sense view. The closing sections consider some prominent objections to the positive view, including that it posits too many objects and that it leads to some form of anti-realism.

If we don't need ordinary objects in our causal explanations, it is often held, Occam's razor enjoins us to eliminate them. This chapter examines these arguments from parsimony, suggesting first that Occam's razor does not generalize to cases in which there are analytic entailments between existence claims. Moreover, it is argued that eliminativists, such as van Inwagen and Merricks who seek to paraphrase claims, for example, about baseballs in terms of claims about atoms arranged baseballwise, don't really offer a more parsimonious theory, since their paraphrased claims may be pleonastically transformed into claims that are explicitly committed to ordinary objects. This provides the basis _for_ an argument for the existence of ordinary objects, and against Quine's criterion of ontological commitment. In closing, this chapter considers whether this also entails commitment to extraordinary objects, and whether more severe forms of eliminativism can do better.

The divergence between the world descriptions provided by physical science and by common sense has led to some of the oldest and most persistent arguments for eliminating ordinary objects. Some (inspired by Eddington) allege that there is a conflict between common sense and physical science, while others (such as Sellars) hold that there is rivalry between the scientific and manifest images, as each purports to offer the _true_ and complete description of the world. This chapter addresses these two claims in turn, arguing that both at bottom rely on the idea that a generic, category-neutral use of “thing” can be made sense of. If, as has been argued in previous chapters, we reject that idea, then neither claims of a conflict nor a rivalry between the two can be sustained.

This chapter investigates what is involved in asking general ontological questions such as: “what exists?” and “how many things are there?” It is argued that well-formed existence and counting questions are “specific” existence questions that specify a certain category or sort of entity enquired about. Purely “generic” existence and counting questions (asking about “anything whatsoever”, where “thing” is not used sortally) are ill-formed unanswerable questions. This gives us reason to think that quantificational claims likewise presuppose certain categories of entity quantified over. It also suggests that many of the questions driving ontological debates are based on pseudo-questions, and does so without subscribing to anti-realism or quantifier variance. The chapter closes by considering whether we can revive a form of universal quantification by way of a “covering” use of “thing” that generalizes over a range of category-specific terms.

Peter van Inwagen, Terence Horgan, and Matjaž Potrč argue against ordinary inanimate objects on grounds that no view that accepts them can provide a satisfactory answer to the special composition question: the question of when a plurality of things composes some other thing. That question involves the existence question of when there is some thing composed (by other things). If “thing” is used in a sortal or covering sense, however, the argument against ordinary objects does not go through. But if “thing” is used generically in asking the question, then (given the results of chapter 6) this is an ill-formed question to which we can expect no uniform answer. However we understand it, then, the special composition question cannot be used to argue against ordinary objects.

Unger's sorites-style argument against ordinary objects relies on the vagueness of the associated terms. While standard solutions to the problems of vagueness are often accused of being _ad hoc_, this chapter argues that the above view of reference provides an understanding of the source of vagueness that suggests that supervaluational solutions and Tye's indeterminist solution are entirely suitable and not _ad hoc_. Others reject ordinary objects given worries that these objects would have to be vague, since ontic vagueness is often considered incoherent. This chapter also addresses Horgan's and Evans' reasons for rejecting ontic vagueness, arguing that the kind of vagueness in the world that we must posit to accept ordinary objects is a kind of anodyne vagueness that explains the vagueness in the world in terms of vagueness in our representations, but that even serious ontological vagueness does not present the difficulties it is thought to.

Those who accept an ontology of ordinary objects face the problem of colocation: holding that different objects — such as a lump of clay and a statue — occupy the same place at the same time or are composed of the same parts. Anti-colocation arguments appear in several forms: rhetorical appeals that surely there is nothing “over and above” the clay, claims that this violates the plausible principle that no two objects may share the same space or parts, and worries that this falsely leads to “doubling up” property instances (such as weighing ten pounds). This chapter aims to defuse all these related worries. We are then left with the grounding problem: how _can_ two objects share all the same parts and non-modal, non-sortal properties, yet differ in their identity and persistence conditions? It is argued that the modal conceptualist view of Chapter 3 provides the key to resolving this problem.

The hybrid theory of reference developed in Chapter 2 has important consequences for our understanding of metaphysical claims about identity, persistence, and modality, which are drawn out in this chapter. Specifically, it leads to the view that the most basic conditions of existence, identity, and persistence for the objects we refer to are discoverable by a kind of conceptual analysis, and the most basic claims about these conditions are analytic. This in turn leads to the conceptualist view that the most basic modal claims are likewise analytic. Conventionalist views are often said (e.g., by Michael Rea and Crawford Elder) to lead to objectual anti-realism, but it is argued that modal conceptualism clearly does not. Moreover, modal conceptualism is independently appealing, since it can help soften epistemic and ontological worries about modality.

Causal redundancy arguments allege that all the causal work attributed to ordinary macroscopic objects is really performed (jointly) by their microscopic parts, making ordinary objects epiphenomenal. Trenton Merricks holds that this shows there are no such things, since, if there were ordinary objects such as baseballs, they would be causally efficacious. It is argued, however, that Merricks' argument is either invalid or unsound, depending on how one understands “overdetermination”. This chapter makes the case that standard prohibitions against overdetermination do not apply where there are analytic entailments between the causal claims, so we can accept the causal claims of, for example, a baseball and of simples arranged baseballwise, without them being rivals in claims to causal efficacy or yielding redundant “double causation”.

This chapter defends the idea that there are analytic entailments among our sentences against two sorts of prominent objections. In response to Quine's rejection of the analytic/synthetic distinction, it suggests that analytic interrelations may be understood as grounded in (tacit, collective) legislations of the rules of use for our terms. Against objections based on a Kripkean approach to reference, it argues that those inclined to causal theories should accept a hybrid theory of reference in order to overcome both the _qua_ problem and the problem of singular nonexistence claims. This hybrid theory acknowledges that our general and singular nominative terms have determinate reference only to the extent that they are associated with frame-level conceptual content, in the form of application and coapplication conditions that fix the category of entity to be referred to. This minimal conceptual content can underpin the analytic entailments discussed in Chapter 1.

The introductory chapter lays out the motivations for defending a common sense ontology and provides an overview of the various eliminativist arguments against ordinary objects that will be discussed in the book. It also describes the argument strategy of the book: the early chapters (2 and 3) argue for central views about reference and modality — namely, a hybrid approach to reference, and a conceptualist understanding of modality. But the later chapters provide the most important evidence for these views by demonstrating their ability to defuse all of the arguments against ordinary objects, and to make sense of our common sense world view nonproblematically. The book closes by sketching important metaontological results of the above work regarding which metaphysical questions are and are not answerable, and what methods should be employed in pursuing them.

In Ontology Made Easy (2015), I defend the idea that there are “easy” inferences that begin from uncontroversial premises and end with answers to disputed ontological questions. But what do easy inferences really get us? Bueno and Cumpa (this journal, 2020) argue that easy inferences don’t tell us about the natures of properties—they don’t tell us what properties are. Moreover, they argue, by accepting an ontologically neutral quantifier we can also resist the conclusion that properties or numbers exist. Here I address these two issues in turn—in ways that help clarify both the scope and results of easy ontology. First, it is important to see that easy inferences were never intended to address modal questions. Modal questions are addressed by a different part of the total deflationary view—modal normativism. So understood, metaphysical modal questions nonetheless do not provide a remaining area for serious metaphysical inquiry. Second, I argue that we have reason to resist adopting an ontologically neutral quantifier, if we aim to answer ontological questions (without begging the question). Addressing these issues helps to clarify both what does (and does not) follow from easy inferences, and how they form part of a larger deflationary metametaphysical view.

In this text, I reply to the other papers this volume, which include discussions of my work on fiction, metaontology, and modality. Regarding my artifactual theory of fiction, I respond to issues about the reference of fictional names, and discuss whether the artifactual theory of fictional characters I develop can be extended to characters in films, or to the theoretic entities of scientific theories. On metaontology, I defend my metaontological views about the ontology of art, and defend an easy ontology of facts, while also discussing what ontological discourse does for us. Finally, on the topic of metaphysical modality, I respond to accusations that my views about metaphysical modality unduly rule out substantive metaphysical debates, and discuss whether these views can be extended to handle counterfactuals and talk of possible worlds, and how they may be relevant to debates about the meanings of artifactual and natural kind terms.

This paper addresses several foundational questions in conceptual engineering: When is conceptual engineering needed? When we engage in conceptual engineering, should we think of ourselves primarily as aiming to change concepts or language – and how should we think of either of those? Finally, how is implementing the changes recommended by conceptual engineering possible? I begin by outlining a number of different sorts of circumstances in which conceptual engineering is desirable, bringing out the commonalities across many classic projects in the history of philosophy and contemporary, socially-conscious, work in conceptual engineering. I then argue that some prominent ways of understanding concepts and meanings are unhelpful for conceptual engineering. We can do better if we think first and foremost of engineering words, considering words as a certain kind of abstract cultural artifact, which (like other artifacts) have functions and norms of use. Then we can begin to understand better how words do change naturally, and use that as a way to investigate how they (like other cultural artifacts and social norms) can also be changed artificially.

Dan Zahavi raises doubts about the prospects for combining phenomenological and analytical approaches to the mind, based chiefly on the claim that phenomenology is a form of transcendental philosophy. I argue that there are two ways in which one might understand the claim that phenomenology is transcendental: (1) as the claim that the methods of phenomenology essentially involve addressing transcendental questions or making transcendental arguments, or (2) as the claim that phenomenology is committed to substantive theses of antirealism and the like, which are sometimes thought to follow from a transcendental approach. I argue that while (1) is appropriate, it in no way leads to conflicts with analytic work in philosophy of mind. Moreover, adopting this method and practicing phenomenology in no way commits us to claims of type (2) that might be thought to conflict with common assumptions in analytic philosophy of mind.

It is often noted that institutional objects and artifacts depend on human beliefs and intentions and so fail to meet the realist paradigm of mind‐independent objects. In this paper I draw out exactly in what ways the thesis of mind‐independence fails, and show that it has some surprising consequences. For the specific forms of mind‐dependence involved entail that we have certain forms of epistemic privilege with regard to our own institutional and artifactual kinds, protecting us from certain possibilities of ignorance and error; they also demonstrate that not all cases of reference to these kinds can proceed along a purely causal model. As a result, realist views in ontology, epistemology, and semantics that were developed with natural scientific kinds in mind cannot fully apply to the kinds of the social and human sciences. In closing I consider some wider consequences of these results for social science and philosophy.