It is time for a change in how we think about metaphysics. Or so I have argued.Footnote 1 On the mainstream conception, metaphysics is conceived of as a philosophical discipline, which aims to discover ‘deep’ facts about reality. These discoveries are supposed to be a matter for philosophical work – and can't simply be handed over to the empirical sciences or answered through conceptual means. Doing metaphysics is thought of as a matter of formulating metaphysical ‘theories’ about how the world is – theories that are to be assessed by the same sorts of criteria for theory choice used in assessing scientific theories.
The time for reevaluating this conception has come. For the mainstream conception of metaphysics has proven undeniably problematic.Footnote 2 First of all, there have long been concerns about a rivalry with the empirical sciences, if metaphysics really thinks of itself as discovering deep facts about reality. Second, in metaphysics, unlike the empirical sciences, we have nothing like a convergence on the truth – instead, we are embarrassed by an ever-increasing proliferation of views. This proliferation, in turn, leads some – both within and outside of philosophy – to a despairing skepticism. For (as our undergraduate students often lament to us) it seems that we can never know the answers to these metaphysical questions – and if not, it seems like we might do better to just give up, and put our efforts somewhere more useful. Skepticism also arises from epistemological mysteries about how we could come to know the facts metaphysics purportedly aims to discover. Most mainstream metaphysicians deny that such facts may be known empirically – often all parties to metaphysical disputes agree that no empirical facts could settle who is right. Yet they almost uniformly deny that their questions can be answered merely by ‘conceptual analysis’. For, as they often insist, these are questions about the world, not questions about our language or concepts. What is the epistemology for metaphysics, then? It has become commonplace to appeal to the idea that metaphysical issues can be settled by appeals to the theoretic virtues or to the idea that metaphysical views are confirmed with our scientific theories.Footnote 3 But these views of the epistemology of metaphysics face formidable problems.Footnote 4
In the face of these worries, metaphysics threatens to make itself suspect, obscure, and irrelevant. And indeed, to many outsiders, at least, that is how it has come to seem. Debates about the existence of trout-turkeys and snowdiscallsFootnote 5 are likely to strike outsiders as worthless, while arguments that we should deny that there are tables, persons, or numbers show up merely as evidence of how far philosophy can go wrong.Footnote 6
On the other hand, many are rightly hesitant to throw the baby out with the bathwater. Even if some recent debates seem (at least to outsiders) worthless, and some positions crazy, far fewer would say that we should commit all of metaphysics to the flames. Classic metaphysical debates about whether we have free will, what the conditions are for personal identity, or what art is may frustrate our students with inability to ‘find the answer’, but they seldom strike people as worthless or silly.
1. The Deflationary (Re-)conception of Metaphysics
I share the sense that there is much in metaphysics that is worth preserving. Although I am a critic of metaphysics, I am not a slash and burn critic. Rather than committing metaphysics to the flames, we should change our conception of what metaphysics is, and how we can do it – assimilating the best of it, or transposing it to a new key. This involves developing a deflationary approach to understanding and addressing metaphysical questions.
Where existence questions are concerned (and where we take them in what Carnap would have called an ‘internal’ sense), I have defended the ‘easy’ approach to ontology.Footnote 7 I call an approach to answering (a particular range of) existence questions an ‘easy’ approach provided it shares the following two features: 1. It relies on nothing more than empirical and conceptual work in answering existence questions that are well-formed and answerable (requiring nothing ‘epistemically metaphysical’),Footnote 8 and 2. It allows that at least some disputed existence questions may be answered by means of trivial inferences from uncontroversial premises. Of course, given that empirical and conceptual work can be difficult, adopting the ‘easy’ approach does not mean that existence questions can all be answered over a cup of tea – only that answering them need not involve any methods beyond these familiar and non-mysterious approaches. Nonetheless, the approach does enable us to answer many ontological questions by trivial inferences from uncontroversial premises. For example, we can start from the undisputed truth that May was born on a Monday, and conclude that a birth (an event) occurred on a Monday, and thus that there are events. Similarly, a competent speaker who has mastered the use of property language is in a position to start from the undisputed observation that Beyoncé’s dress is red and move to the conclusion that the dress has the property of redness, and so that there are properties, without reviewing metaphysical debates about properties.
If existence questions can be answered straightforwardly and non-mysteriously, then endless debates about whether there ‘really’ are events or properties seem out of place, not because the answers can't be found, but rather because the existence questions can be easily answered: Yes.
Metaphysical modal questions about the existence conditions, identity conditions, persistence conditions, etc. of things of different kinds (provided they are answerable questions, taken ‘internally’) I have argued, can likewise be addressed by making use of our conceptual competence, inferential skills, and empirical information.Footnote 9 For, as I have argued elsewhere, metaphysical modal claims should not be seen as aiming to describe or track modal features of reality, or facts about other possible worlds.Footnote 10 Instead, I have argued, the function of having metaphysical modal vocabulary is that it enables speakers to convey semantic rules (or their consequences)Footnote 11 in particularly useful ways, by formulating them as object-language indicatives. Conveying rules in this way has various advantages, including making the regulative status of what is said more explicit than it is in the other forms of expression, enabling us to express conditionals connecting them to other rules, making explicit our ways of reasoning with them, and enabling permissions as well as requirements to be expressed. Since, on this view, metaphysical modal claims fundamentally serve a normative function, I call the view ‘Modal Normativism’.
Dropping the assumption that metaphysical modal talk serves a descriptive function gives us hope of avoiding some of the well-known epistemic and methodological problems of modality, and makes addressing metaphysical modal questions (again, taken in a Carnapian ‘internal’ sense), like addressing existence questions, ‘easy’. We avoid epistemological mystery, for if we do not think of modal terms as functioning to track modal features (or possible worlds), then we needn't proceed in a ‘metaphysics first’ fashion to somehow (how?) discover what the modal properties (or features of other possible worlds) are, and thereby determine which of our modal utterances are true. Instead, the route to modal knowledge, on this conception, is a matter of moving from mastery of the relevant semantic rules to ability to convey these (and their consequences) explicitly in the useful form of object-language indicatives. For example, one can exploit one's conceptual competence in determining that a painting could not survive being burnt to ashes (that would no longer ‘count as’ a situation in which there was a painting) and combine this with empirical knowledge (that temperatures above 500 degrees Fahrenheit would burn canvas and oil) to determine that no painting could survive such conditions. This way of rethinking modality thus brings modal questions under the same umbrella as existence questions: as questions that can be straightforwardly resolved by making use conceptual analysis, often combined with empirical information.
But despite its virtues in clarifying the epistemology of metaphysics, to fans of metaphysics the deflationary approach left something to be desired. For it seemed to be unable to capture the feeling that metaphysics is deep, important, and world-oriented. In response to this, I have argued for reconceiving of some of the most difficult, interesting and persistent of metaphysical debates (considered now in the mode of what Carnap would have called ‘external questions’) as implicitly engaging in metalinguistic negotiation: that is, pressing for adopting, preserving, modifying or rejecting elements of our conceptual (or linguistic) scheme by using the relevant terms, in the object language.Footnote 12 Engaging in such metalinguistic negotiations is a worldly matter: because they are conducted in the object language, because we often must appeal to worldly facts in determining what conceptual choices make sense, and because the results have great worldly relevance to how we live and what we do.
Some philosophical arguments wear an element of conceptual negotiation on their sleeves: look to Ruth Millikan's work on function, Sally Haslanger's on race and gender concepts, the work of Bernard Gert et al on death, Joshua Gert's work on color, David Davies’ view that works of art are not objects but performances, and so on.Footnote 13 In other cases, the element of conceptual negotiation is implicit, but can be seen in the sorts of considerations that are raised for or against various views. For example, hard determinists are not swayed by observations that we tend to in fact hold people responsible just when they were (without encumbrance) doing what they wanted. Instead, they press for refocusing on the question of what concept of freedom we should adopt, given the facts of determinism.Footnote 14 Similarly, debates about personal identity often appeal to what we care about in identifying people over time, and suggest modifying our criteria accordingly. John Locke, in proposing his view that we should identify people over time based on ‘continuity of consciousness’, acknowledges that this does not fit with the standard practices of the time, but justifies the alteration by noting that ‘person’ is a ‘forensic term, appropriating actions and their merit, and so belongs only to intelligent agents, capable of a law, and happiness, and misery’.Footnote 15 But, he argues, we are only rightly punished for actions we can attribute to our same consciousness, and so personal identity should be measured by a continuity of consciousness: ‘In this personal identity is founded all the right and justice of reward and punishment’.Footnote 16 In a great many such debates, the disputants simply use the contested terms in the object language, and speak about what freedom is or what a person is. But in doing so (and in the criteria they appeal to for these metaphysical views) we can see them as engaged not in reporting metaphysical discoveries, but rather as negotiating for how we ought to use the relevant terms. Since what concepts and terms we use matters to how we organize our lives together, to what we do, to who and how we punish, to how we reason, and to what we value, these disputes are deep, important, and worldly.
Beyond the interpretive claim, that many past debates can be seen as explicitly or implicitly engaged in conceptual negotiation, is the normative claim: that we should come to reconceive metaphysics as centrally concerned with what concepts we should use and how we should use them – that is, with work in conceptual ethics and conceptual engineering.Footnote 17 Both existential and metaphysical modal questions (taken in what Carnap would have considered an ‘external’ sense) may be seen in this light. By conceiving of the work of metaphysics in this way, we can still demystify the methods of metaphysics by appealing to nothing more mysterious than empirical and conceptual work, (now including under that both descriptive and normative conceptual work). We can also avoid the problem of a rivalry with science. Moreover, we can avoid being embarrassed or driven to skepticism by the proliferation of metaphysical views. Instead, what we may have – at least sometimes – is a range of alternative solutions to an engineering problem; solutions that might have different merits given different goals, settings, and external constraints. Most deeply, once we take on this conception of metaphysics, we shall be better able to see why – at least in many central cases – working on these normative conceptual issues is important, not at all pointless or a waste of time, and why simply ‘giving up’ would be exactly the wrong move. For we have to employ some concepts and terms to get around in the world, and which ones we use can make an enormous difference to how we live and what we value. Surely it is worth taking the time to think about which ones we should employ, and how we should employ them.
This account of the legitimate role of ‘deep’ metaphysics as examining what concepts and terms we ought to use, and how we ought to use them, leaves metaphysics something deep, worldly, and important to do. But it's not a matter of making quasi-scientific ‘discoveries’ about what ‘really exists’, and it doesn't require anything more mysterious than empirical work, combined with descriptive and normative conceptual work, to do it.Footnote 18
Some have objected to the easy approach on grounds of doubts that our terms have sufficient conceptual content to ‘do the job’ of answering all these metaphysical questions. But this belies a misunderstanding. First, the deflationist isn't committed to the view that all metaphysical questions are answerable – the fact that some aren't is part of the point. Second, nothing in this metaontological position commits the deflationist to any very specific views about how much conceptual content our terms have, or to how widely it is shared, how stable it is, and so on.Footnote 19 If, as Peter Ludlow has argued, there is very little to a ‘common coin’ of meaning,Footnote 20 that will only serve, from the deflationist's perspective, to help account for the ongoing difficulties and failures of agreement, and to lend plausibility to the idea that the much of what we have done and can do in metaphysics should be read not in the internal mode of explicating the rules there are, but rather in the external mode, of lobbying for those rules the speaker thinks there should be. In short, how much (shared, stable) conceptual content our terms have will make a difference not to whether or not the deflationary position is tenable, but rather to how much of our work in metaphysics may be thought of as engaged in addressing internal, and how much in addressing external, questions.Footnote 21
But if the most interesting parts of metaphysics involve determining what concepts we should use, and how we should use them, how should we do this work? Elsewhere I have argued for a pragmatic approach to normative conceptual work, which begins with a kind of reverse engineering, aiming to determine what function(s) it serves to have this sort of vocabulary (biological kind terms, moral terms, modal terms, mathematical terms…) in a language. Then, if a function or functions can be determined, we may take a step back to assess whether this is a function we should aim to preserve – this is a matter of work in conceptual ethics. If so, then different choices about how terms or concepts should be used may be justified by appeal to whether or not such changes would enable them to better fulfill their function (this is work in conceptual engineering). In each case, these typically require empirical work. For it is often an empirical matter whether the changes advocated will actually enable the term to better fulfill its function. Moreover, technological and empirical changes in the world often challenge the boundaries of our concepts, forcing us to make new conceptual choices or reevaluate old ones (consider debates about whether persons can survive replacements of various parts with artificial substitutes, or about what the identity conditions are for works of internet art). If, on the other hand, we find that the relevant function is one we should reject, then we have further decisions to make about whether we should retain the terms and assign them new functions or reject the terms altogether.
2. What difference does it make?
That, in all too brief form, gives an overview of the deflationary approach to metaphysics that I have developed elsewhere. In this paper, I do not aim to defend any of these views. Instead, I aim to address a different question: Suppose we do make this change in how we think about metaphysics. What difference would it make to how we do metaphysics?
What difference will it make to which problems in metaphysics we take seriously, and think of as central and worth solving – and which we reject or brush off? What difference will it make in terms of how we go about evaluating metaphysical debates – which criteria are (and are not) relevant to assessing positions? Finally, what difference will it make in terms of which positions we take seriously – and which we should not?
2.1. Which debates are worth having?
Reconceptualizing metaphysics in this deflationary spirit draws the line just where we wanted it: justifying the feelings that some debates are worthless, and some positions implausible, while still preserving the sense that other areas of metaphysics really matter.
Consider, for example, what has become the poster child for silly disputes in metaphysics: debates over the existence of particular mereological sums, such as trout-turkeys, or the sum of my nose and the Eiffel Tower. Taken in the mainstream spirit, this is presented as a matter of ‘discovering’ whether there ‘really are’ such things. Translated into the new mode, it can be taken in two ways. Taken internally, it can be given an ‘easy’ answer: given the rules of use that introduce the vocabulary of mereological sums, if there is A and there is B, then there is a mereological sum of A and B. So (provided we have a trout and a turkey), there is a trout-turkey. But this is too easy to be a matter for interesting, extended debate. Taken externally, we must transpose the question to ask (explicitly) whether we ought to make use of the term. And pretty clearly, this particular term (‘Trout-Turkey’) is one for which we have no foreseeable use. The broader, more interesting question, however, is whether in general we should retain the language of mereological sums, or a conceptual scheme like that outlined in General Extensional Mereology. To answer this, we must ask what the function of that scheme is. If it is to provide a neutral and fruitful formalized theory of wholes and parts, perhaps we will find a justification for retaining it. If, on the other hand, it is employed primarily so that those who have ‘ontological qualms’ about the existence of sets or other abstracta could still do some of the same work without ‘accepting these abstracta’, then, from the deflationary point of view (which finds such qualms out of place), it might be justly abandoned. Put briefly, the response to the question ‘are there mereological sums’ is, ‘well, if you're going to adopt that terminology, of course we should say there are – but why would you want to do that?’ This way of being torn between what to say on the debate conceived as internal versus external can also help explain the mixed feelings about whether the right response to the question is ‘yes’ or ‘no’ – and indeed deflationists themselves have divided on this point.
Debates about snowdiscalls or incarsFootnote 22 can be similarly analyzed: although, given the associated criteria, we may be perfectly (easily) entitled to conclude that there are snow discalls and incars, once we have adopted the associated conceptual scheme, there is no clear reason why one would want to be able to distinguish and separately refer to a squashed snowball, or a car-in-a-garage. These were concepts introduced ad hoc to make a philosophical point – not concepts that serve an ongoing function in our lives. If metaphysics, reconceived, aims to engage in conceptual ethics and conceptual engineering in ways that matter to our lives, we do indeed have reason to ignore debates involving these rather peripheral, useless terms.
Such debates can aptly be contrasted with those that seem central, important, and worth engaging in. Such terms as ‘freedom’ and ‘person’ have crucial uses in our lives. Attributions of freedom are central not only to our ways of understanding ourselves and others, but to our attributions of responsibility, praise, and blame, to our appropriate feelings of guilt, gratitude and resentment, and to our practices – to how we punish, teach, and evaluate ourselves and others. Identifications of someone as a person, or as the same person over time, have relevance to our provision of rights, protections and medical care, to our attributions of debts, credits, and responsibility, to what and whom we care about. But both of these are heavily contested concepts, and in their everyday uses exhibit a great deal of vagueness and indeterminacy. Whether we ought to use ‘freedom’ roughly as the compatibilist does, or should require ‘higher standards’ involving indeterminacy and/or agent causation makes an enormous difference to our lives – and is precisely what a great many participants in the free will debate have been arguing about. So similarly, criteria that identify A and B as the same person over time make an enormous difference to our lives and to our legal, medical, familial and economic practices. Even if we think of resolving these debates not as a matter of ‘discovering’ the ‘true’ criteria but rather as deciding what criteria we ought to adopt, these are debates very much worth having, and with great worldly relevance.
Other, more specific, terms likewise play central roles in our practices that ensure that debates about whether and how we ought to use these terms really matter. Consider, for example, debates about what art is, how we should understand disability, what species concept we should use (or which we should use in different areas of biology), debates about who counts as a woman, or what causation is. These problems can be pressed in the material mode (‘what is a woman?’ ‘what is disability?’) or in the linguistic mode (‘who should we count under the label of ‘woman’?’, ‘should we employ a medical or social concept of disability?’) – but either way, they are worth undertaking. In sum, while some metaphysical problems might fall by the wayside on this re-conception, it seems that they are the right ones to leave behind. Instead, we can focus our attention on a range of crucial problems, the importance of which does not diminish if we explicitly (re-)conceive of them in pragmatic conceptual terms.
2.2. Which criteria should we make use of?
Even if many metaphysical problems remain, however, won't our ways of evaluating them change radically if we adopt the deflationary approach?
The approach used in many debates about identity, persistence, and other metaphysical modal matters will remain largely intact and relevant – in some cases more relevant than ever. For, given the normativist approach to metaphysical modality, we can finally have a vindication for the use of conceptual analysis and thought experiments in addressing metaphysical modal questions. By contrast, justifying such traditional approaches is very hard if we take these questions in the spirit of serious metaphysics – for then it is unclear why we should think that our concepts, intuitions or imaginative experiments should have anything to tell us about the deep metaphysical facts of the world.Footnote 23
Of course, not all positions on these questions are driven by conceptual analysis – there are also revisionary views about personal identity or the natures of works of art, for example. But, as mentioned above revisionary views (such as Locke's on personal identity) are often driven by considerations of ‘what we really care about’ in making attributions of personal identity. A similar example can be seen in David Davies’ view that works of art are not objects but performances – a view he justifies by appealing to what we care about in critically evaluating these works of art.Footnote 24 These can likewise be easily – indeed best – accommodated if we see their advocates as engaged in conceptual negotiation rather than as presenting metaphysical ‘discoveries’. In short, if we adopt this reconception of metaphysics, much of the way modal metaphysics is done will remain intact – it is just that we must then become more explicit about the reasons behind adopting the relevant conceptual choices.
Matters are more complicated when we address debates about whether or not entities of various sorts exist – and these debates will often come to seem more problematic. Seen from the deflationary perspective, I will argue, some criteria remain apt, but many others should be tossed aside as based on inappropriate generalizations. Let us consider some commonly used criteria in turn.
2.2.1 ‘It turns out that nothing (or no uniquely best candidate) meets the associated conditions’
A classic form of argument for elimination is simply the claim that the relevant conditions associated with the term or concept are not (in fact) met. Consider, in this mode, Kwame Anthony Appiah's arguments that (if we follow an ideational approach) what it would take for there to be races is for there to be ‘significant correlations between the biological and the moral, literary, or psychological characters of human beings; and that these be explained by the intrinsic nature… of the members of the race’. But there have turned out to be no such significant correlations – and so we should conclude that nothing meets the criteria for being a ‘race’ in the traditional sense of ‘race’ assumed by Thomas Jefferson and Matthew Arnold.Footnote 25
Such arguments work as well on the deflationary approach as elsewhere. Where terms have worldly application conditions, the main way for the deflationist (treating these as internal existence questions) to show that the corresponding entities don't exist is to argue (as Appiah does) that those conditions are not met. This is how the easy ontologist can (while accepting most philosophically disputed entities) legitimately come to deny that there are witches or Vulcan.
Nonetheless, on this model, finding that nothing (or something) meets a set of associated criteria (even assuming we have identified these correctly) does not yet determine the answer to the external question – of whether we should reject (or retain) the concept. For even if nothing meets the associated criteria, there may yet be reasons to retain the concept, which justify amending the criteria, and/or amending its function. Sally Haslanger, for example, accepts Appiah's analysis of traditional race concepts and their inapplicability, but argues that we should adopt social constructionist replacement concepts of race, to serve the function of identifying and fighting racially-based injustice. On the new definition:
A group is racialized iff its members are socially positioned as subordinate or privileged along some dimension… and the group is ‘marked’ as a target for this treatment by observed or imagined bodily features presumed to be evidence of ancestral links to a certain geographical region.Footnote 26
These criteria are met, and so we can say that there are racialized groups in this sense. Debates about the existence of races, then, can proceed in the latter vein, as we examine Haslanger's proposal and aim to determine whether we should adopt modified race concepts along these lines.
The approach of determining whether the associated criteria are met is often employed in addressing everyday existence questions, and is perfectly well preserved on the deflationary reconception of metaphysics. But it is not so common in what are thought of as ‘metaphysical’ existence questions – where other criteria are often used for ‘eliminating’ ‘suspicious’ entities.
2.2.2. ‘The concept is inconsistent, so nothing could meet the associated conditions’
In metaphysical discussions, arguments that certain conditions are not in fact met are less common than arguments that the relevant criteria could not be met – as the relevant concept turns out to be inconsistent. And that is another criterion that remains relevant on the deflationary conception. Arguments for ‘eliminating’ various kinds of entities, including ordinary objects, fictional characters, persons, qualia and more often allege to find contradictions in the concept. From the internal perspective, the deflationist can of course acknowledge that if the concept is indeed inconsistent, then there can be nothing that meets the associated criteria. Mainstream arguments purporting to show that concepts such as baseball, person, or qualia are contradictoryFootnote 27 have to be taken seriously on the new approach to metaphysics, just as on the old.
Nonetheless, as I have argued elsewhere,Footnote 28 if the relevant concepts have been playing an important and useful function in our lives, we should examine these claims of inconsistency with suspicion. I have also tried elsewhere to debunk various claims that certain ordinary concepts are inconsistent.Footnote 29 However, if such claims of inconsistency can be made out, then (from the internal perspective) deflationists have reason to deny that there are such entities.
Suppose we shift to the external perspective – of pragmatic conceptual engineering. On this approach, too, we have reason to be concerned about and take seriously claims about a concept's inconsistency. For a concept that can lead us into contradiction is not a well-functioning concept. Nonetheless, we might, on the new conception, take a somewhat different attitude, and engage in different responses even if the concept is shown to be inconsistent. For if they are very useful concepts, and the defects peripheral or unlikely to cause problems in their standard uses, we might well have reason to revise the concept rather than reject it. In some cases (as Alexis Burgess has argued for the concept of truth Footnote 30) we might even have reason to leave it be.
So, while on the new approach to metaphysics we should still care about consistency, we do not just regard it as a cut-and-dried matter that if we find an inconsistency we should throw out the concept. Even where claims of inconsistency are borne out, we should not rush to declare that such things don't exist, or to reject (positive atomic) use of the terms, but rather (from the external perspective) undertake a studied decision about what we should do from here – whether we should reinterpret, revise, or reject the relevant concept, or replace it with another in the vicinity which could do much needed work. And that decision will also require attention to questions about what functions the concept has served and might legitimately be used to serve going forward.
2.2.3. Criteria that won't carry over (as across-the-board)
So far, I have argued that traditional arguments that appeal to whether or not associated conditions have been, or could be, met remain relevant on the revised conception of metaphysics. Beyond these, however, there may be little we can say across the board about criteria for handling existence questions, considered externally as a matter of evaluating pragmatic conceptual choices.
For (as I have argued elsewhere) we must, at least in part, assess concepts based on their ability to perform their function.Footnote 31 But our terms and concepts may serve a variety of functions. It is plausible that some noun terms function to track and co-vary with certain features of the world, perhaps by tracking distinctions that are especially relevant to explanations and predictions, say, in chemistry and biology. But for many terms, it is at the least a philosophical open question whether the terms are even supposed to be used in predicting and explaining, or to be tracking worldly features at all. Consider moral terms. On an expressivist view, for example, the function of having these terms in our vocabulary is not to track moral features of reality, but rather to enable us to express and coordinate our plans or attitudes in specific ways that enable us to better live together. If this is on the right track, then having moral terms as part of our conceptual scheme is very important, and it would be a mistake to jettison such terms or concepts on grounds of their failure to track joints of reality. And there are many similar views that ascribe some function other than joint-tracking to certain philosophically central terms. Consider Stephen Yablo's view that mathematical terms serve the function of enabling us to express scientific laws more succinctly, Gilbert Ryle's view that dispositional terms and mental predicates serve to license inferences, or Robert Brandom's view that modal talk serves to make explicit certain norms governing the use of our non-modal terms.Footnote 32 Even logical terms, as I have argued elsewhereFootnote 33 arguably serve a very different function than tracking joints (logical or otherwise) in reality – namely, enabling us to make assertions with and reason using other terms (some of which may serve a joint-tracking function).
Once we take things from the perspective of engaging in pragmatic conceptual choice, and acknowledge the possibility of functional pluralism, most of the criteria that have commonly been used in metaphysical debates turn out to be inappropriately generalized. Although they may be apt in certain contexts, they will be seen as illegitimate in almost all of the cases in which they have been applied by metaphysicians as arguments for eliminativism. Let us examine some frequently used criteria in turn.
2.2.4. ‘We should accept only those entities that are required to make true our best scientific theories.’
The dominant approach to metaphysical arguments over the last several decades is generally traced to Quine, who famously argued that we are ontologically committed only to those entities over which we need to quantify to render true the statements of our best scientific theories.Footnote 34 This is often taken as a directive: that we should go about choosing the best scientific theory, and only accept the entities we need to quantify over to make its statements true.
From an internal perspective, of course, the easy ontologist does not care whether a term is part of a scientific theory or not. One may positively answer the question ‘are there hats?’ by noting that the application conditions for ‘hat’ are met, even if our best scientific theories fail to quantify over hats. Moreover, the fact that we do not need to quantify over properties (say), since we could say ‘the dog is white’ rather than ‘the dog has the property of whiteness’ (and thereby quantify only over particulars, not properties), I have argued (following William Alston and John Searle) is also irrelevant.Footnote 35 For such paraphrases really are just a matter of rejecting certain terms and concepts – not the commitments that are entailed by accepting the original (‘uncontroversial’) premise.
Could we reconstruct some relevance to criteria like these, taken from the external perspective, then – taken as pragmatic criteria for engaging in conceptual choice? Explicitly transposed into this mode, it would be something like: ‘Accept only those terms and concepts that range over things you need to quantify over to make true your best total scientific theory’. This is probably not so far off from how we should read the historical Quine, who writes ‘Our ontology is determined once we have fixed upon the over-all conceptual scheme which is to accommodate science in the broadest sense’.Footnote 36
But if we lay this out explicitly as a general criterion for engaging in conceptual choice, it becomes immediately clear that we should hesitate to accept it. Clearly science is not the only place where we do and should have use for concepts. Many of our concepts play many other roles than figuring in our scientific theories – they figure in our moral views, in our laws, in our pedagogy, in our art and culture, in our social lives. Should we eliminate all of these, if we don't need them in our best scientific theories? This surely would be rash, and eliminate much that is useful to us. If, on the other hand, we limit the advice to ‘accept only those scientific terms and concepts that you need to formulate your best total scientific theory’, it seems like no more than a triviality.
In short, considered either in the internal or external mode, if we accept the deflationary approach to metaphysics, then we should reject generalized arguments in neo-Quinean style, that we should not or need not ‘posit’ certain entities because we need not quantify over them in our best scientific theories.
2.2.5. ‘We should only accept those entities (or properties or kinds) that are perfectly natural’
This is a criterion that has been used most commonly in determining which properties to accept. David Armstrong famously presses for a ‘sparse’ view of universals, accepting, ‘It is to natural science… that we should look for knowledge, or perhaps just more or less rational belief, of what universals there are’.Footnote 37 David Lewis severed Armstrong's approach from its tie to a theory of universals, and argued for a related general distinction between those properties and relations that are (and are not) ‘natural’; those that do (and don't) ‘carve at the joints’.Footnote 38
From the internal perspective, the easy ontologist of course accepts the existence of properties regardless of whether they are ‘natural’ in this sense – for we can perfectly well make easy inferences from ‘The sculpture is rococo’ to ‘The sculpture has the property of being rococo’, to ‘There is a property of being rococo’, without concern for whether that property is endorsed by the natural sciences, or carves the world at its joints.
Transposed into the external perspective instead, this would be the view that we should only accept those terms or concepts that carve the world at the joints. Thus expressed, this brings us very close to the view recently defended by Theodore Sider, who aims to ‘reconceptualize metaphysics’ in terms of the notion of structure.Footnote 39 On Sider's view, this requires that – at least when doing ultimate metaphysics – we ask, ‘which notions carve perfectly at the joints’.Footnote 40 The goal of inquiry is not merely to state truths (which Sider admits can be stated using concepts that do not carve at the joints); it is ‘to use the right concepts, so that its conceptual structure matches reality's structure’.Footnote 41 The evidence for a concept's being joint-tracking is its figuring in our best theory. Those that are perfectly joint-carving, and so mark fundamental structure, he thinks of as ‘certain concepts of physics, logic, and mathematics’.Footnote 42
In some limited but important contexts, the deflationist can see something like the ‘joint carving’ criterion as entirely apt. For those terms (like natural kind terms) whose function is to serve in our explanatory and predictive scientific theories, ability to track similarities and differences that figure prominently in explanation and prediction will be a crucial desideratum. Those terms that do so are those we naturally think of as picking out ‘natural kinds’ or as ‘carving at the joints’. This enables us to distinguish terms like ‘lithium’ as relatively natural; those like ‘lithium on earth but not on Mars’ as relatively unnatural, and worthy of rejection for the purpose.Footnote 43
But again, we can generalize this to an across-the-board criterion only if we rule out or ignore the possibility of functional pluralism. For it is at least a live option to think that moral, modal, mathematical, and other terms are not even supposed to track worldly features, ‘carve the world at its joints’, or serve in our explanatory and predictive theories. If they aren't, it would be entirely misguided to rule out these concepts on grounds of their failing to carve at joints or track features of the world they never even aimed to track. On this approach to metaphysics, then, something like joint-carving (understood not in heavyweight metaphysical terms but rather as tracking similarities and differences that particularly help in explanation and prediction) can be a virtue. But it is only a virtue for terms with some sorts of function – joint-carvingness is not a desideratum across the board.Footnote 44
2.2.6. ‘We should not accept any vague entities, or entities without determinate identity conditions’
If we adopt the deflationary approach, then common arguments to the effect that we should not accept any vague entities in our ontology, or that we should reject entities that lack clear and determinate identity conditions, likewise turn out to be inapplicable to the vast majority of cases in which they are used to support eliminativist views. Taken in the internal mode of easy ontology, we can of course legitimately conclude that clouds, for example, exist – without regard to whether they would have to be ‘vague’ or lack determinate identity conditions.
Transposed into the mode of normative conceptual work, the question would become whether we should accept vague concepts, or sortal concepts that lack determinate identity conditions. And here again we can see that the answer is variable depending on the function of the relevant terms or concepts. A certain degree of precision may be a desirable criterion for concepts to be used in many scientific, legal and medical contexts. Yet for use in everyday concepts, it may only undermine usability and introduce arbitrarily sharp cut-offs if we, say, judge the end of childhood by the second. So again, as with ‘carving at the joints’, the deflationist will reject the general principle that we ought to avoid concepts that are imprecise, or be eliminativist about any corresponding objects that would lack fully determinate existence or identity conditions.Footnote 45
2.2.7. ‘We should only accept those entities that have (distinct) causal powers’
David Armstrong famously promotes the so-called ‘Eleatic Principle’ – ‘Everything that exists makes a difference to the causal powers of something’.Footnote 46 This is a principle widely adopted, and utilized in contexts as different as debates about what properties we should accept, about whether we should accept the mental, whether we should accept ordinary objects, and many more besides.Footnote 47
But again, once we accept the deflationary approach, this will be seen as a principle too quickly generalized. From the internal point of view, if possession of causal powers is not part of the application conditions for a term, then failure to possess them is no mark against the existence of the relevant entities.Footnote 48 As a result, Eleatic arguments that we should reject the existence of numbers or other entities that are supposed to be abstract and not causally relevant are out of place, though they may remain relevant for entities that we do think of as causal (perhaps scientific posits and ordinary objects).
From the external point of view, we would have to transpose the question: Should we accept terms or concepts where the alleged referent would lack causal powers? Again, one can see why it is sometimes a relevant criterion. If a term or range of terms is to function in tracking environmental features, then there must be causal powers in the entities to be referred to in order for us to stand in a proper tracking relationship to them. Where we are aiming to track and discuss birds, plankton, or auras, there had better be causal powers to enable the term to function properly in tracking them.
But generalizing this criterion is again out of place in cases where terms do not serve a tracking or co-variation function. Terms that serve rather to simplify our statements of scientific laws, to engage in certain kinds of generalization, or to enable us to coordinate our planning or attitudinal states have no reason to be held to the standard that their referents have causal powers.
2.3. Which positions are worth seriously considering?
Just as some debates will show up as not worth engaging in, once we explicitly adopt the deflationary reconception of metaphysics as engaged in pragmatic conceptual choice, so similarly certain positions will show up as hardly worthy of consideration. Wherever we have a term in common use that is playing a useful and harmless function in our language and lives, it will be a hard sell to convince us that we should eliminate it. Suppose, for example, that Stephen Yablo is right that having noun terms for numbers serves a useful function, enabling us, for example, to state in finite form scientific laws that would otherwise require an infinite series of infinitely long disjunctions.Footnote 49 If number terms do indeed serve this useful function, and if (as I have argued above) we should pay no heed to arguments that demand our ‘posited’ entities have causal powers, there seems to be no reason to eliminate these terms from our vocabulary (and go through the trouble of a Field-style nominalization),Footnote 50 and every reason to keep them. Much the same goes for everyday terms for ordinary objects – talk of tables and their kin does not seem socially or ethically problematic, and plays a useful and central function in the organization of our everyday lives and economies. Once we can put aside arguments for rejection based on constraints to admit only natural or fundamental entities, or entities lacking vagueness and indeterminacy, there seems little reason to countenance undertaking such upheaval and rejecting our terms for ordinary objects. Some issues may, of course, remain: we will still have to take seriously claims that there is an internal contradiction in the concept, or a conflict in our conceptual scheme owing to problems with material coincidence of the table and lump of wood. But seen from the perspective of pragmatic conceptual choice, we will have far more reason to try to disentangle these portions of our conceptual scheme (whether by showing that the apparent contradictions aren't real, or finding ways to minimally revise and avoid them) than to discard them.Footnote 51
Other issues will be far more contestable and unclear. Do we do better to retain a notion of freedom that entitles us to hold responsible, praise, blame, and punish, or would we do better to drop that whole way of thinking, and the range of practices that go with it, aiming more for explanation and education than blame and punishment? Supposing we retain the concept of person, what criteria of personal identity should we adopt? What should we count as the beginning and end of the life of a person? Could the concept of person ever apply to an ‘intelligent’ machine? Do we do better to reject all of our harmful traditional race concepts, or to reconstruct them to serve a new social purpose of identifying and fighting racially-based injustice? What concept of causation should we employ in the natural sciences, and is it the same as we should employ in the social sciences, or in legal contexts? What concept of ‘art’ should we employ, to best capture what has been done under that name, while also allowing for innovation and growth, and preserving a sense of the importance and value of art?
These and many other issues more remain difficult, requiring thoughtful philosophical work, often combined with the best empirical information we can get our hands on. In such cases, as far as I can see, there are many views worth seriously considering. Taking the deflationary approach won't make them ‘easy’ in anything but the technical sense. And the work of metaphysics, taken in this vein, will never be finished, since there will always be new social contexts and empirical situations that put pressure on us to refine, revise, or expand our old conceptual repertoire.
3. Conclusion
As the above discussion has made clear, adopting this deflationary approach to metaphysics will not change whether we do metaphysics, but it may lead to profound changes in how we do it. Some parts will remain relatively unscathed. Modal metaphysics will require little change in our practices, and in fact we will be able to get a clear justification for engaging in venerable practices that use conceptual analysis and thought experiments in doing modal metaphysics. Revisionists will, however, be pressed to make it clear that they are not reporting metaphysical discoveries (say of what art or persons ‘really are’), but rather making conceptual recommendations – and will be pressed to make the reasons for these recommendations transparent.
Recent ways of addressing existential questions, however, will require greater change. Pragmatic conceptual analysis encourages us to think in terms of accepting, revising, or rejecting a concept based on how well it fulfills its function, and (more deeply) based on whether this is a function we think should continue to be served. Once we acknowledge the possibility of functional pluralism, however, most of the standard criteria for elimination turn out to be reasonable in limited contexts, but irrelevant where they are most commonly used. As a result, one of the major results of this change in our meta-metaphysics would be that the majority of first-order metaphysical arguments for eliminating entities of various sorts will turn out to be inappropriate. In any case, the hope is that making explicit issues about what function these concepts serve – or we want them to serve – and examining what effects various conceptual moves would have on their ability to serve the relevant functions, will clarify and redirect debates. This in turn should make matters far more transparent and tractable than they can be when each disputant simply claims to have be discovered the ‘metaphysical truth’, and clothes their recommendations in the guise of discoveries.
By turning to focus on issues that matter, positions that are worth considering, and using methods that are transparent, the hope is that metaphysics can be both relevant and non-mysterious. If we adopt this reconception of metaphysics, we do not commit it to the flames, but instead focus on doing what as philosophy at its best always has done: assessing and sometimes changing the way we think and live.
