Metaphysics as Fairness

Metaphysics as Fairness

Abstract What are the rules of the metaphysical game? And how are the rules, whatever they are, to be justified? Above all, the rules should be fair. They should be rules that we metaphysicians would all accept, and thus should be justifiable to all rational persons engaged in metaphysical inquiry. Borrowing from Rawls’s conception of justice as fairness, I develop a model for determining and justifying the rules of metaphysics as a going concern.

1

Introduction

ETAPHYSICS is a norm-governed enterprise. But are there, or do there need to 1 I believe so, for three reasons. First, there is substantial disagreement over what the norms of metaphysics are. Whether we ought to obey Ockham’s razor, whether we should use intuitions as evidence, whether we are obliged to render metaphysics continuous with science, are all normative issues that engender a great deal of dispute. Such disagreement is troubling. Debates in metaphysics that rely on these norms are likely to stall until some determination has been made concerning what the norms of metaphysics should be. Determining what the norms should be, however, is partly a matter of identifying those rules that express the right norms for doing good metaphysics, and legislating accordingly.

M be, rules for doing ‘good’ metaphysics?

1 The distinction of interest between ‘rules’ and ‘norms’ is between the norms as they actually exist and rules that express possible norms, which may be legislated to ensure compliance with some ideal. For instance, it may be that the manner in which wealth is distributed within a society is unfair, despite being norm-governed. Accordingly, if fairness in wealth distribution is valued, then the norms must be overturned for a set of rules that do better. The situation is similar in metaphysics. Metaphysics has aims. Exactly what these aims are depends on what metaphysics is. I won’t take a stand on this issue yet (though see §3.1). The point is just this: for a given set of aims we can imagine various rules, each of which expresses a possible norm geared toward ensuring the efficient achievement of defined metaphysical ends. We may then wonder whether the norms of metaphysics best ensure the successful attainment of those ends, or whether there is some possible set of rules that would do better.

1

Second, it is a common refrain that metaphysics makes no progress, or at least progresses very little. As I see it, there are (roughly) three possible explanations for why metaphysics makes no progress: (i) what it is trying to do is extremely difficult; (ii) what it is trying to do is impossible; (iii) there’s something wrong with the manner in which it’s being done. There is substantial pressure on the metaphysician to rule out explanations (ii) and (iii). One way to rule out (iii) is to identify rules for doing ‘good’ metaphysics and check to see if they express the norms that currently exist. If they do, then (iii) is not a plausible explanation. If they do not, then we may be forced to legislate. Of course, progress within metaphysics – in some sense of ‘progress’ – may be achieved with a merely arbitrary set of rules. What we want, however, is real progress in metaphysics, rather than pseudo-progress, and to do that we need to come up with the right set of rules. Finally, we need a rule-book for metaphysics because sometimes when a move is made in the context of a metaphysical debate it feels like cheating, despite the absence of any prohibition against the move. For instance, invoking deities at certain points in a debate, or taking certain concepts to be intuitively primitive, seems to be fundamentally parochial in this sense: one thereby excludes, from subsequent discussion, those who give low credence to deities or those who don’t share the relevant intuitions about conceptual primitivity. Cheating in metaphysics, however, only makes sense against the backdrop of a set of rules, and so the question arises: what are those rules? The rules of metaphysics should not, themselves, be parochial; they must be rules that would be endorsed by all fair-minded, truly disinterested fans of rational collective inquiry. Furthermore, the rules should be in the best interests of metaphysics as a discipline. In what follows I advance a model for determining the rules of metaphysics as a going concern, one that is built on the foundations of justifiability. My aim, however, is not to come up with a definitive list of rules, and thus to outline the rule-book for metaphysics. That is a much larger project. My aim, rather, is the more modest one of developing a framework within which this more general project can be profitably developed. Although developing such a framework is my primary focus, I will nevertheless briefly apply the framework and offer advice on what some of the rules of metaphysics might be.2 The framework draws heavily on the work of John Rawls. Rawls (1999) outlines a model for identifying the principles of justice that we as rational, mutually disinterested agents would agree to be bound by. The Rawlsian framework, though powerful in its own right, is also quite flexible: a similar framework can, in principle at least, be developed for any system of practices involving rational agents, where 2 The Rawslian model is not an account of what the methods of metaphysics are and so does not compete with any such account; it is rather a decision procedure for selecting a set of methodological norms.

2

those agents possess motivations appropriate to the system in question. In particular, as I will demonstrate, the framework can be extended to determine and ultimately justify the rules of good metaphysics. While I focus on the rules of metaphysics, I recognise that it is somewhat artificial to do so. There may not be any rules peculiar to metaphysics that are not simply rules of philosophy more generally. While I believe that metaphysics does require its own particular rules I admit that not everyone will agree. In the event that there are no rules peculiar to metaphysics, the framework developed in this paper has merit as a toy model, one that provides the foundations for producing a larger method for justifying and determining the rules of philosophy tout court. I foresee two further benefits of the Rawlsian approach. First, the lack of an explicit account of the rules of metaphysics, along with an account of why we should follow them, may contribute to the existence of ‘stalemates’. Stalemated debates in metaphysics are often intractable in a way that stalemated debates in, say, science are not. One reason for this, perhaps, is that such stalemates are a direct result of questionable moves made by one or both sides; moves that have not been properly criticised for being unfair in the context of the enterprise in which they occur. This is not to say that all stalemated debates are like this; but I believe that some are,3 and so determining the rules of metaphysics promises to provide new tools for reassessing old, deadlocked disputes.4 Second, care for the rules of metaphysical inquiry is good for the health of the discipline.5 Metaphysics as a practice is increasingly scrutinised by scientists; by the public; by university administrators looking to make funding cuts; by grant bodies looking for cutting edge research and, of course, by philosophers outside of metaphysics looking to make their mark. If we metaphysicians can’t agree on the norms of metaphysics, then it appears we are at a relative disadvantage. Without the resources to properly defend our discipline, we are thus apt to find ourselves squirming under the increased scrutiny of others. A rule-book for metaphysics would therefore constitute a bedrock for addressing sceptism about metaphysics as a discipline. 3

Debates that may be stalemated include the debate between presentism and eternalism; the debate over persistence and the debate over unrestricted versus restricted composition. I discuss some of these debates in §4. 4 A similar point is made by Wilson (forthcoming). 5 As Wilson (forthcoming) notes, our understanding of metaphysical methodology is still rudimentary. Of course, inroads have been made. There is an important project investigating the role of scientific virtues in metaphysical theory choice (Baker (2003), Nolan (1997, forthcoming)). There is also a recent debate over the status of methodology in ontology (see Schaffer (2009)). And there is the project of defending metaphysics against positivist and neo-positivist scepticism (for the latest incarnation of this debate, see Ladyman and Ross (2007) and Maclaurin and Dyke (2012, 2013)). This work does not, however, speak to the development of a framework for justifying the norms of metaphysics. Rather, the literature cited focuses on the relationship between science and metaphysics and/or on whether or not metaphysics is a legitimate form of investigation.

3

2

Justice as Fairness

Although many will be familiar with the Rawlsian picture, its uptake has not been universal and so I’ll start there. I should note, though, that in considering Rawls’s view I will outline only those features of his theory of justice that are relevant to the issue at hand, and that are thus transferable to the case of metaphysics. In particular, I will focus on the model that he develops for determining and justifying the principles of justice, extracting its core features. I will thus, with some regret, set aside the richer details of Rawls’s position. The model begins with the original position. The original position is a hypothetical scenario, in which agents come together to make agreements about the principles by which society will be organised. Principles are put forward, discussed and voted upon. The aim of the original position is unanimity: agents must unanimously agree on the principles of justice, since it is only then that the principles will be justifiable to everyone and thus fair. In addition, agents in the original position are subject to certain epistemic limitations; they are behind a veil of ignorance. They are ignorant of their social status; class position; fortune in the distribution of natural assets; abilities; intelligence; strength and so on. Agents are not entirely ignorant, however. They still possess general information. In particular, they know whatever general facts are needed to make decisions about principles of justice. They therefore know general facts about human society; about political affairs and economics; the basis of social organization; the laws of human psychology and general information about laws and theories. The veil of ignorance is crucial. It is only by stripping agents of particular knowledge about their position in society that they are guaranteed to arrive at agreements not based on natural contingencies, which are irrelevant from the standpoint of justice. As Rawls (1999, p. 17) puts the point: The aim [of the veil of ignorance] is to rule out those principles that it would be rational to propose for acceptance, however little the chance of success, only if one knew certain things that are irrelevant from the standpoint of justice. For example, if a man knew that he was wealthy, he might find it rational to advance the principle that various taxes for welfare measures be counted unjust; if he knew that he was poor, he would most likely propose the contrary principle. To represent the desired restrictions one imagines a situation in which everyone is deprived of this sort of information. If we do not abstract away from this information then the principles offered as principles of justice by agents in the original position, and thus the principles that have the potential to gain acceptance, are unlikely to be justifiable to everybody. The

4

rich person’s principles of justice won’t be acceptable to the poor person and vice versa. Agents must therefore agree on principles of justice that they would agree to be bound by regardless of the actual position in society that they ultimately come to occupy, once the veil of ignorance is lifted. Agents in the original position are rational and mutually disinterested. They are concerned to maximise their own share of the primary social goods. These are the goods that a rational person would desire regardless of whatever else she desires: rights, liberties, opportunities, income and wealth. A rational agent in the original position is only interested in her own gains. She remains neutral on the gains or losses of other agents. As such, she does not seek to maximize or minimize the difference between her gains and the gains of others. Rather, she seeks only to get for herself as much as possible. Put in terms of a game, a mutually disinterested rational agent seeks to attain the highest possible score. As Rawls (1999, p. 125) notes, the analogy with a game is imperfect, as mutually disinterested agents do not seek to win. Winning implies the attainment of a higher score than other players, and agents in the original position are not interested in other players’ scores. Rational, mutually disinterested agents must choose the principles of justice that they would agree to be bound by, once the veil of ignorance is lifted, from a list of possible principles of justice.6 Rawls’s central contention is that rational, mutually disinterested agents would agree upon a single, unique set of principles from behind the veil of ignorance. It is these principles, then, that are the principles of justice that we ought to follow. According to Rawls, agents within the original position would select two principles of justice in particular, over all other alternatives. These two principles are stated as follows (Rawls 1999, p. 53): (1) Each person is to have an equal right to the most extensive scheme of equal basic liberties compatible with a similar scheme of liberties for others. (2) Social and economic inequalities are to be arranged so that they are both (a) reasonably expected to be to everyone’s advantage, and (b) attached to positions and offices open to all. One useful way to think of the hypothetical scenario that Rawls asks us to consider is in terms of optimisation. Optimality models include three components: the thing one is trying to optimise, the options for optimisation, and any relevant constraints on the situation. So, for example, consider the evolutionary optimality model of why it is that hive-bees produce hexagonal honeycomb (see, for discussion, Lyon 6 Ideally, individuals would have a list of all possible principles of justice. For obvious reasons, however, Rawls (1999, pp. 126–127) restricts his statement of the original position to a list of the most likely principles of justice.

5

and Colyvan (2008) and Baker (2012)). The reason that hive-bees produce hexagonally structured honeycomb is that partitioning the surface of the honeycomb using hexagons will allow the swarm to produce the largest nectar-storing cells, for the least expenditure of energy in producing wax. This, in turn, will enable those bees to outcompete other swarms and so honeybees have evolved to exploit this optimal solution. This outcome is predicted by an optimality model in which one seeks to maximise the amount of nectar stored for the least energy expended. The constraints on the model are typical environmental constraints on hive-bees and the choices involve different regular and irregular polygons. Similarly, the two principles of justice outlined above can be seen as optimal solutions to an optimisation problem, one that Rawls models using the original position. Rawls seems to suggest this interpretation when he writes: Suppose, for example, that it is rational for the persons in the original position to prefer a society with the greatest equal liberty. And suppose further that while they prefer social and economic advantages to work for the common good they insist that they mitigate the ways in which men are advantaged or disadvantaged by natural and social contingencies. If these two features are the only relevant ones ... the two principles are the optimum solution. (Rawls 1999, pp. 106–107) The model stated in this quotation is a simplified version of the case at hand, since it assumes only that rational, mutually disinterested agents are concerned to optimise two features (i) the equal liberty in society and (ii) the degree to which any advantages serve the common good. The choices in this simplified case are, as before, potential principles of justice (see Rawls (1999 pp. 126–127) for the relevant options), and the central constraint is that the optimisation must mitigate the advantages or disadvantages afforded by natural and social contingencies. It is thus because principles (1) and (2) represent the optimal solution to this problem, that they would be selected by rational mutually disinterested agents in the original position. It is because they would be so selected that they are justifiable to everybody. I raise the parallel between Rawls’ hypothetical scenario and optimisation merely as a heuristic for better understanding the scenario in question. What really matters for present purposes is that the scenario can be distilled into three core features, which I set out as follows: Motivation

Agents in the original position are rational and mutually disinterested, and are thus motivated to achieve the greatest share of primary social goods for themselves as possible, without trying to maximize or minimize the differences between themselves and others.

6

Ignorance

Agents in the original position are ignorant of their social status; class position; fortune in the distribution of natural assets; abilities; intelligence; strength and so on.

Justifiability

Agents in the original position must agree upon those principles of justice that are justifiable to everyone. This is represented through unanimous agreement on the principles in question.

It is these features of the model that will become important as I seek to extend the Rawlsian paradigm to metaphysics.

3

Metaphysical not Political

Let us call a rule that constrains the manner in which metaphysics qua practice is carried out, a principle of metaphysics. In order to extend Rawls’s model for determining and justifying the principles of justice to the principles of metaphysics, an original position for this particular set of practices must be elucidated. As before, the original position is a hypothetical scenario. In this case, however, it is a scenario in which metaphysicians come together in order to determine the principles by which metaphysics will proceed. As such, for the sake of clarity, let us call this: the metaphysical original position.

3.1

The Metaphysicians

Who exactly are the metaphysicians? One way to answer this question is by citing canonical examples. Current candidates include Elizabeth Barnes, Sally Haslanger, Carrie Jenkins, Kit Fine, Kristie Miller, Laurie Paul, Jonathan Schaffer and Ted Sider. Looking back: Anscombe, Foot, Lewis, Russell and Quine surely count as metaphysicians (if meta-ethicists) in some sense. And in the early twentieth century, we find De Beauvoir, Heidegger, McTaggart, Neitszche, Hegel and Husserl. The early modern period boasts (at least) Berkeley, Cavendish, Descartes, Locke, and Kant and in ancient times: Hypatia, Plato, Socrates and Aristotle. Now, I’m sure my canon is going to raise a few eyebrows: many of the names cited would not play well together. Still, all of these philosophers have at least some claim to the title ‘metaphysician’. Citing canonical examples can take us quite far: we all know a metaphysician when we see one and so we roughly know who’s coming to the party. Knowing who’s coming to the party, even roughly, is perhaps enough for the Rawlsian model 7

I outline below to gain dialectical traction. The trouble, however, with my canon is that it is quite tightly tied to the institutional arrangement of disciplines, in this sense: the ‘default’ metaphysicians are ‘officially’ philosophers. Not everyone who should qualify as a metaphysician must be officially a philosopher, however. Many scientists are metaphysicians too. So it behoves me to say a little more. Very roughly, a metaphysician is a rational agent who is motivated to achieve the aims of metaphysics and who is disposed to act upon those motivations. Exactly who counts as a metaphysician on this account therefore depends entirely on what metaphysics is, since an adequate account of the nature of metaphysical inquiry will be one that has a story to tell about its aims. I make two assumptions about the nature of metaphysics. First, I assume that metaphysics is a single discipline, distinct to other forms of rational investigation, and possessing a single, unified set of aims. Second, following Paul (2012) I assume that whatever else metaphysics might be, it is the search for general and fundamental truths about reality.7 These assumptions produce a crisp account of who the metaphysicians are. Paul, herself, would clearly count as a metaphysician, and so would the rest of the canonical examples cited above. The advantage, however, is that we are no longer limited to those who are ‘officially’ philosophers. Anyone engaged in the relevant search counts, regardless of where they find themselves inside (or outside) the academy. Identifying the metaphysicians with the set of agents who engage in the search for fundamental and general truths does not seem all that controversial. What is controversial is the idea that there might be unifying methodological themes among the various different sub-sets of agents engaged in this broader practice. For it seems that metaphysicians diverge quite significantly in their views on how the search is to be done.8 For Ladyman and Ross (2007) the project is that of unifying our best scientific theories. For others, it is the Ramsay-Lewis-Carnap project of conceptual analysis (see also Kant (1943), Dummett (1991), Strawson (2011) amd Jackson (1998). And on it goes.9 Now, if the goal of this paper were to provide a complete account of what metaphysics is, then I would have to sort this divergence out. More detail would be needed about what metaphysics is in the business of doing, and the stark differences between the sorts of things that different groups of metaphysicians do would have to be addressed. However, these issues are beyond the scope of the 7

The model outlined below is not beholden to these assumptions. The model may be run with any set of participants selected for their common epistemic aims. Still, these simplifying assumptions will be necessary in §4 when the model is applied. 8 If metaphysics is pseudo-artistic, then the metaphysicians won’t be definable in terms of what they are trying to know. I do not, however, have the space to discuss non-epistemic versions of the metaphysical original position here. 9 While I find it plausible that the Ladyman and Ross project and the Ramsay-Lewis-Carnap project both fall under the broader disciplinary practice of searching for general and fundamental truths, this could be denied. Denying this would force one toward the idea that metaphysics is really a group of sub-disciplines connected by a family resemblance relation, as discussed briefly in fn. 10.

8

current discussion and so can be reasonably set to one side. A fairly coarse-grained understanding of what metaphysics is – along the lines sketched by Paul – will do.10

3.2

The Model

I make two further assumptions about the metaphysical original position. First, I assume a veil of ignorance. The veil of ignorance in this case does not involve ignorance about social status. Rather, it factors out information about an agent’s beliefs and opinions on metaphysical issues. In addition, agents in the metaphysical original position do not know what theories they will end up defending once the veil of ignorance is lifted. Nor do they know what philosophical intuitions they might have. They do not know who has defended which theories in the past. Finally, because the line between scientific knowledge and metaphysical speculation can be quite thin in some places, it is also necessary to abstract away from any knowledge that agents have concerning their own views on scientific disputes (assuming that this knowledge, and knowledge like it, fosters bias over which principles of metaphysics are offered in the metaphysical original position, and accepted).11 As before, agents in the original position are not completely ignorant. They know general facts about what metaphysical theories there are; what scientific theories there are; what scientific evidence exists; what philosophical methodologies there are and what method10

It could be argued that these more fine-grained issues about the nature of metaphysics cannot be so easily set aside. Apparent disagreement among metaphysicians about the best way to search for the fundamental and general truths may, in the end, reveal that metaphysics is a cluster of sub-disciplines, each of which is both distinct from each other and distinct from other forms of rational investigation, and each of which possesses its own unique set of aims. If metaphysics is a cluster of sub-disciplines, then it is doubtful that there is a single class of individuals who are the metaphysicians. That said, we can still make sharp distinctions between individuals based on the sub-discipline of metaphysics in which they are involved, and based on each peculiar motivational set. We can, that is, identify the metaphysicians1 , the metaphysicians2 ... the metaphysiciansn . Each group will then attend its own metaphysical original position in order to determine the rules for metaphysics1 , metaphysics2 ... metaphysicsn respectively. As we shall see below, that’s fine. The model I outline is flexible enough to allow for its use across a range of sub-disciplinary instances of metaphysics just as easily as if metaphysics were a single discipline, though the results in each instance of the model may vary substantially from one another. 11 One might worry that too much information has been occluded. Perhaps it is necessary to know one’s opinions on some metaphysical disputes in order to be in a position to make decisions about the norms of metaphysics. If so, then the model can be made more specific, by focusing in on a particular metaphysical dispute: D. The norms of D are to be determined by abstracting away from the knowledge agents have of their opinions on that dispute. Opinions on other disputes are, however, allowed. The trouble with this way of proceeding, however, is that many metaphysical disputes are strongly connected to one another (at least at some level). So information about disputes other than D might be inadmissible.

9

ologies there are available to agents engaged in rational inquiry more generally.12 In short, they know whatever general facts might be relevant to the development of principles of metaphysics.13 The veil of ignorance in the metaphysical original position is required for similar reasons to the veil of ignorance that Rawls introduces. The aim of the veil here is to rule out those principles of metaphysics that it might be rational to propose for acceptance, however little the chance of success, only if one knows certain things that are irrelevant from the standpoint of general metaphysical methodology. To put the point another way, in both the Rawlsian original position and the metaphysical original position we’re looking to identify and agree upon rules with which to conduct ourselves when we come across predictable (and predictably unresolvable) disagreement. The central thought, then, is that if we metaphysicians can predict the details of the roles that we play in these disagreements, there will be a strong temptation to develop ground rules in a biased way. For example, if an agent knows that she is defending the A-theory of time (the view according to which, roughly, temporal passage is an objective feature of reality), she might find it rational to advance the anti-naturalistic principle that metaphysics should not seek compatibility with empirical science. This is for the simple reason that the A-theory is at odds with some of our best available scientific theories (namely, the special and general theories of relativity) and so ruling out appeals to science serves the A-theorist’s interests. Conversely, if a person knows that she is defending the B-theory of time (the view according to which, again roughly, temporal passage is not an objective feature of reality), she might find it rational to advance the principle that one ought not appeal to common sense in support of a metaphysical theory, since common sense intuitions are thought, by some, to provide evidence for the A-theory (see e.g. Zimmerman (2008)). Principles of metaphysical methodology that are developed in this way, in the service of a particular theory, are unlikely to be fair. They ‘stack the deck’ against certain metaphysical positions, by ruling out 12

What about logic? When the agents are arguing from within the original position to whatever principles of metaphysics they ultimately accept, they must be deploying some rules of inference. But which rules? It is tempting to say: the one correct logic, whatever that might be! But that won’t fly with anti-realists about the one correct logic, and they are sometimes metaphysicians too. So, better: agents in the original position have knowledge of all the logics, and use whichever logic they all agree is most apt for their current argument about the normative principles of metaphysics. 13 I have focused here on general scientific and metaphysical knowledge. It may be that, in order to make informed decisions about the methodology of metaphysics, all of human knowledge must be included, assuming that there is knowledge outside of these two domains (e.g. moral knowledge, political knowledge and so on). Widening the knowledge that agents have within the metaphysical original position in this way will make it more difficult to get methodological naturalism out as a principle of metaphysics (see §4.1 below), since it may be that we have to unify with more than just science. Still, something like the weak naturalism I consider should be recoverable even in this situation.

10

particular methodologies. As such, in the interests of developing those metaphysical principles that are justifiable to all we must abstract away from this information. In addition to a veil of ignorance, I assume that agents engaged in metaphysics are rational and that they have particular motivations. For Rawls, agents in the original position are rational and mutually disinterested. As such, they are motivated to maximise their own share in the primary social goods. This motivation is relevant to the contractarian approach to society in which Rawls is interested. It is not, however, the appropriate kind of motivation for metaphysics. We are not trying to build a society in the same sense. We are, rather, trying to determine how a field of inquiry can be profitably developed, and thus to determine the rules of association that bind parties within that field. The motivations in this case must therefore be ones that are relevant to the practice of metaphysics as a going concern. We must be careful, however, to abstract away from those motivations that one might have in virtue of knowing information occluded by the veil of ignorance. One cannot, for example, assume that all agents are motivated to defend a particular metaphysical theory. What we can assume, however, is that the agents in the metaphysical original position are ideally motivated to do the best metaphysics they possibly can. With this in mind, it is possible to identify three motivations that agents in the metaphysical original position are likely to share. First, it is reasonable to suppose that such agents are motivated by the ideal of progress. Metaphysicians want to maximise the progress that metaphysics as a discipline makes, though exactly what counts as progress in metaphysics will depend in large part upon what one thinks metaphysics is in the business of doing. But, for now at least, I will continue to assume that metaphysics aims to discover the fundamental and general truths and that progress in metaphysics is progress toward this aim. Second, agents in the metaphysical original position are motivated by the ideal of unification. Metaphysicians see the work that they do as contributing in some way to the development of a grand, unified philosophical theory. This is not to say that all metaphysicians are motivated to develop their own unified theory. It is, rather, to say that all metaphysicians seek to contribute something to the project of developing such a theory, whatever that theory might be. More precisely, agents in the metaphysical original position seek to maximise the contribution that metaphysics makes to this ideal. The third motivation is difficult to articulate, but it is something like the following. Agents in the metaphysical original position seek to maximise the ‘accessibility’ or ‘transparency’ of their theories. They want metaphysics (as far as possible) to be such that one need not be inducted into metaphysics in order to be able to take a metaphysical theory seriously. More precisely, metaphysicians in the original position do not want it to be the case that one must be a metaphysician in order to accept the evidence offered in support of any given metaphysical theory as evidence. Or that

11

one has to be a metaphysician in order to find the assumptions in a metaphysical argument compelling, assuming that one has understood the argument and is aware of the putative basis for the relevant assumptions. The reason for this motivation is that, without transparency, metaphysics runs the risk of becoming irrelevant. If metaphysics is convincing only to metaphysicians because it trades on implicit beliefs that only they hold in virtue of being metaphysicians, then it is hard to see what place metaphysics has in rational investigation more generally. For instance, to return to the A-theory/B-theory example used above, suppose that only metaphysicians take common sense intuition seriously as a guide to theory choice. Folks outside of metaphysics – e.g. scientists – are going to look askance at metaphysics as practiced. The onus is therefore on the metaphysician to show that common sense intuition is transparent in the above sense: that it has a positive evidential status both inside and outside of metaphysics. As in the Rawlsian original position, agents in the metaphysical original position must, from behind the veil of ignorance, agree upon a principle or set of principles about how metaphysics should be carried out. In particular, they must identify those principles that best satisfy the three motivations mentioned above. Because these principles are based on general motivations that all such metaphysicians have, and because they are not based on contingent knowledge of the position within the metaphysical ‘game’ that any one metaphysician occupies, such principles will be justifiable to all and, as such, will be fair. As before, it is possible to construe this as an optimisation problem: what sorts of principles ought we to adopt in order to maximise the ideals of progress, unification and transparency? The optimal solution to this problem is the one that everyone from within the metaphysical original position will be willing to agree upon.14 As before, the focus on optimality is a heuristic. The metaphysical original position can be distilled into three core features: Motivation

Agents in the metaphysical original position are motivated to maximise the ideals of progress, unification and transparency in metaphysics.

Ignorance

Agents in the metaphysical original position are ignorant of their beliefs and opinions about metaphysical issues; the theories they will end up defending; what philosophical intuitions they might have and exactly who has defended which

14

There’s a general epistemic problem here. The basic principles of metaphysics are the ones we would all agree to in the metaphysical original position. But you might think that such a state is epistemically inaccessible: how do we know what we would all agree to? Perhaps our judgements about this are being filtered through the same biases that the original position is supposed to remove. So our very judgements about the original position cannot be trusted. That’s a problem, but it’s a problem for Rawls too and so I set it aside for now.

12

theories in the past (mutatis mutandis for scientific disputes). Justifiability

Agents in the metaphysical original position must agree upon those principles of metaphysics that are justifiable to everyone. This is represented through unanimous agreement on the principles in question.

A great deal hangs on the motivations that I have attributed to rational agents in the metaphysical original position. It is these motivations that, ultimately, inform the decisions that are made regarding the principles of metaphysics. I have attempted to identify a set of uncontroversial motivations that jibe with metaphysics as it is actually practiced, and with the sorts of things that metaphysicians are likely to say if questioned on the topic. Still, one might worry about the list as I have compiled it. Consider, for instance, the ideal of unification. As has been pointed out to me, metaphysicians as a matter of fact differ greatly in the extent to which they value consistency across metaphysical theories, or indeed across metaphysics and science. Accordingly, it may well be that unification is just not a motivation that all metaphysicians in the original position are likely to share. There are two things to say here. First, it may be useful to draw a distinction between ideal and non-ideal metaphysicians. If the participants to the metaphysical original position are idealised then while it may be true that many actual metaphysicians are not motivated by unification, it may still turn out that the ideal metaphysician will be. Of course, this idealising move is just as slippery here as it is in Rawls’s theory of justice and so I won’t make too much of it. Second, it is important to keep sight of the two assumptions made in §3.1. I am assuming that the goal of metaphysics is to discover the fundamental and general truths about reality. Given this way of understanding metaphysics and thus who the metaphysicians are, unification seems quite important. For it seems plausible that the full set of fundamental and general truths is to be found only in the most unified philosophical theory and so metaphysicians of this ilk are likely to have a unificationist bent. This is especially so if the search for general and fundamental truths is to be carried out by unifying our best scientific theories. That said, there may well be accounts of what metaphysics is according to which metaphysicians so defined are not motivated to unify. So I admit that I may be wrong about the importance of unification, and about the list of motivations more generally. It might be that some competing list of motivations is a better contender for characterising the motivations of metaphysicians behind the veil of ignorance. But that’s fine. I’m not wedded to the list I have compiled. What matters is that there are some general motivations shared by all metaphysicians in the original position that are relevant to the way in which metaphysics is practiced, and that these motivations

13

can be ultimately used to constrain the norms of metaphysical methodology. Of course, it may be the that there are no such universal motivations for every metaphysician simply because metaphysicians do not share a common set of epistemic aims. In that situation, the first assumption made in §3.1 is false: if different aims really do dictate completely different motivations with no overlap at all, then what seems to be a difference in aim is really a difference in practice. In such a situation, the reasonable thing to do is to split metaphysics into a number of sub-domains, each interested in different things and each governed by its own rules. If that is how things must go, then I am inclined to say: lesson learned. Such a lesson does not make a difference to my goal here: whether we have one practice with several aims, or many practices with distinct aims, so long as these aims can be used to define the motivations of agents in the metaphysical original position, it will be possible to use a Rawlsian method for determining the principles of metaphysics. It is worth emphasising, however, two things. First, some accounts of what metaphysics is may ultimately pick out something much broader than metaphysics and so the rules that are determined may bleed off into other domains. Second, candidate sub-disciplines for metaphysics are potentially quite disparate and so may yield quite different results when run through the Rawlsian model.15 At any rate, what I am stressing is really the flexibility of the method that I am proposing, and it is the method that matters. The flexibility of the method is due 15

It could turn out that there is no single, unified discipline that is metaphysics and no set of subdisciplines that are crisply delineable either. One manner in which metaphysics may fail to be a distinct discipline or a distinct set of sub-disciplines is if there turns out to be no strong way to split the difference between scientific and metaphysical investigation. For instance, it may be that there is no residue for metaphysics over and above a certain clutch of debates in the special sciences. In this situation, the model developed here could either be applied to science, or it could be applied to philosophy more generally, if the relevant debates within science are ones that philosophers, and not scientists, should take on. Note that for the model to be applied to philosophy, an account of what philosophy is must be given, at which point the same options become available: philosophy is a single discipline, a set of sub-disciplines, or no distinct discipline at all. If pressed to take the last option, then we may be forced to fall back again and again, until we reach a point where the goal is to determine the rules of all rational investigation. Presumably, in this scenario, the rule-makers will be (something like) the class of individuals who are motivated to gain knowledge about the world. Being pushed all the way back to the class of individuals whose goal it is to learn about the world may be no bad thing. For I am confident that with work (much more work than I can carry out in a single paper), the model outlined below will apply even here, at the limit and that once it is applied, rules will flow on down to whatever is left of metaphysical investigation. Indeed, applying the model at these giddy heights of abstract generalisation opens up quite an exciting prospect. Once we’ve pushed the boundaries of the model all the way out to the class of individuals who are motivated to gain knowledge about the world we may find a deep unification with Rawls’s theory of justice. In both cases the driving force behind the original position may turn out to be rational self-interest. Rational self-interest, for Rawls, dictactes the decisions made about distribution of wealth. It may also dictate decisions about the best way to generate knowledge, since that is a commodity like any other to be desired and used; justice and epistemology may enjoy justificatory integration in the abstract.

14

to its modularity. By substituting in one’s preferred motivations, aims and even epistemic restrictions, one can produce a model that may then be used to elucidate and justify the rules that govern the practice of metaphysics. This flexibility makes the method quite powerful. It also highlights the way in which the method feeds back into a much larger research project, one that seeks to illuminate the aims and motivations behind metaphysics as practiced.16 Of course, it may be that the method ultimately fails to deliver up a general set of rules, upon which everyone can agree. This is a concern also for the Rawlsian original position: there may be disagreement behind the veil of ignorance and, as such, no general principles of justice that are justifiable to everybody. If that were right in the metaphysical case, it would be an important result. For if no principles of metaphysics are justifiable to everybody from within the metaphysical original position then it follows that one cannot do metaphysics without begging the question against somebody, methodologically speaking. That would pose a substantial problem for metaphysics as a practice and would constitute a novel, and quite significant meta-philosophical objection. For my part, I am optimistic that there are general rules of metaphysics that we can all get behind, and that the Rawlsian model will deliver them (or, at least, point us in the right direction). But I doubt that everyone will share this optimism.17 To the pessimist, I can only say that minimising methodological question begging, or at least minimising the harm done by question begging of this kind, is still worthwhile and that a useful way to minimise this harm is to abstract away as far as possible from any biases we might have regarding metaphysics, and aim for principles that can be justified to every metaphysician, even if 16

Alongside the social-justice-like structure of determining the principle of metaphysics, there are also the actual social justice issues of metaphysical methodology. Who do we take seriously as metaphysicians, and how do we come to take them seriously in this way? Which topics fall under the rubric of ‘general metaphysics’? And which topics count as ‘specialised metaphysics’? Which books get reviewed, and by whom? Which articles get responded to and why? Who is treated how at workshops and who is invited to which selective conferences and why? A method for determining the ground rules for metaphysical methodology ought to also get a handle on these issues. The Rawlsian paradigm is well-placed to do so. 17 Here’s a reason to be pessimistic: suppose that I develop a metaphysical theory using classical logic. My opponent claims that I am being unfair, because they endorse a non-classical logic. I shift my ground: all I need is consistency. But it turns out my opponent is a dialethist. So I am still being unfair. I shift my ground again: I just require the one-true logic! But my opponent is also an antirealist about the one true logic. So on it goes. On the one hand, that the model can be applied even to a case such as this is impressive. On the other hand, I worry that there will be no way to settle on the most basic of methodological principles: namely, which logic to use. But here is the beginning of a solution: for most disputes we can, for the purposes of argumentation, settle on a logic acceptable to both sides. So that is what we ought to do. It is only when what’s at stake is the very logic being used that matters become difficult. Even here, though, we can perhaps agree on a logic to use for the purposes of argumentation; a ladder that we can ultimately kick away if we need to, when all is said and done.

15

one cannot, ultimately, get hold of any such principles.

4

Application

That, then, is the proposal: in order to determine and justify the principles of metaphysics, we place ourselves into the metaphysical original position. Those principles that we as metaphysicians would all agree upon from behind a veil of ignorance (one that, roughly, occludes our theoretical and pre-theoretical commitments) are the principles that are justifiable to everyone working in metaphysics and ipso facto the principles that ought to govern metaphysical practice. I recognise, however, that the proof here is in the pudding: the model must be applicable. Accordingly, I now demonstrate the practical utility of the model by using it to identify three principles of metaphysical methodology that I believe are likely to be agreed upon by everyone from within the metaphysical original position. Before proceeding, however, two points of clarification are in order. First, my goal in applying the Rawlsian model to metaphysics is primarily to show how the method works, and how it can be used to gain traction on existing metaphysical disputes. So the focus is really on the method; I am less concerned to defend the particular principles of metaphysics that I outline below. Second, I will continue to assume a particular view about the aims and motivations of metaphysicians, namely that it aims at uncovering fundamental and general truths, and that the motivations are as described above. What I say is therefore provisory, and subject to revision should these further issues be decided differently. That said I would very much like to get the principles of metaphysics discussed below out of the model. As such, the relevant principles can be thought of as constituting a wish list. No doubt others will have their own wish lists, and so I encourage everybody to consider how it is that the principles on their lists can be justified using the method outlined.

4.1

Methodological Naturalism

The first principle of metaphysics might be called ‘methodological naturalism’. By methodological naturalism I have something quite particular in mind, and something that is perhaps a bit weaker than the view most often referred to by this phrase. Methodological naturalism in my sense is the idea that metaphysicians ought to develop metaphysical theories in a manner that is informed by our best science; that is sensitive to empirical developments and evidence; and that seeks wherever possible to integrate our metaphysical theories with our best scientific theories. It is also the idea that metaphysicians ought to deploy scientific methods where appropriate for any aspects of their theories that have empirical content. This means, among other things, that practicing metaphysicians ought to be up to speed with contemporary 16

scientific theories and methods. Of course, not every metaphysical theory is going to be informed by science, and so not every theory will be one in which methodological naturalism will come into play. But many theories are so informed and methodological naturalism as a norm of metaphysics exerts pressure on metaphysicians to look for connections with science and develop them wherever possible.18 Methodological naturalism competes with a number of other principles, the most extreme of which is that metaphysicians are required to develop their theories without paying attention to science. Call this principle: methodological anti-naturalism. In the space between methodological naturalism and methodological anti-naturalism we find a range of principles that permit varying degrees of contact with science; however none of these principles require a metaphysician to develop her theory in a scientifically informed manner other than methodological naturalism. Methodological naturalism is (or is at least supposed to be) the weakest principle of metaphysics that exerts this requirement. Methodological naturalism, as a principle of metaphysical methodology, is more likely than competing principles to maximise the motivational ideals identified in §3. Consider, first, the ideal of unification. As noted, this is the idea that metaphysicians are motivated by the ideal of maximising the contribution that metaphysics makes to the development of a unified philosophical theory, in this case: a theory of fundamental reality. Since such a theory is very likely to include all of our current science, the best way to contribute to this goal is to proceed in a scientifically informed manner. Indeed, if one does not proceed in this way, then one runs the risk of developing one’s metaphysical theories in a way that is ultimately incompatible or otherwise difficult to integrate with our best science. That would ultimately cut against the ideal of unification and so if we are to maximise that ideal, then we should adopt a naturalistic approach to metaphysics. One might demur: unification will not speak in favour of methodological naturalism if science is compatible with many metaphysical systems. Even here, however, unification gains traction. For presumably science is not compatible with all possible metaphysical systems. Accordingly, metaphysicians who aim to develop a metaphysical system should ensure that their favoured system falls within the class of systems compatible with science. To do that, however, they must defend their theories in a scientifically informed way, which is all 18

Methodological naturalism in my sense is not the view that science trumps all. Nor does it rule out non-natural metaphysics tout court. It only does so when the relevant metaphysics is scientifically naive. Endorsing methodological naturalism does not thereby immediately commit one to a Ladyman and Ross-type vision of naturalized metaphysics, according to which metaphysics is in the business of unifying science. Rather, the naturalized metaphysics imagined here is weaker than that: it is merely required to render itself compatible with science and to integrate scientific and metaphysical theories wherever possible. There is plenty of scope within this weaker conception of naturalized metaphysics for metaphysics that is independent of science; not so for Ladyman and Ross. If Ladyman and Ross’s account of metaphysics is used to set up the metaphysical original position something much stronger than the methodological naturalism proposed here would likely fall out.

17

that methodological naturalism of the kind discussed here demands. Consider, second, the ideal of progress. If we do not develop our metaphysical theories in line with methodological naturalism, then because we run the risk of developing theories that are at odds with our best science, we also run the risk of developing theories that must be abandoned for reasons of empirical inadequacy. Which is to say that metaphysics developed in this way may result in a great deal of work being poured into metaphysical theories that must eventually be abandoned. Now, of course, this might happen regardless of how closely we attend to science, but the point is that methodological naturalism will minimise this possibility and thus minimise the possibility that metaphysicians will proceed up blind alleys. This, in turn, maximises the chances of genuine progress in metaphysics. Finally, methodological naturalism is likely to contribute something toward the ideal of transparency. If metaphysicians are motivated to produce arguments that are convincing to those outside of metaphysics, then they should not produce arguments that are empirically naive. Such arguments are likely to contain hidden empirical assumptions about which science has an opinion, and one that has not been properly taken into account. Such arguments are therefore unlikely to be convincing to anyone but other, similarly naive metaphysicians. The most transparent theories, then, are ones that are developed in a scientifically informed way. Methodological naturalism helps to maximise the ideals of progress, unification and transparency, when compared to various anti-naturalistic principles that do not exert a requirement to engage with science. As a principle of metaphysics, then, it is a principle that is likely to be justifiable to all, within the metaphysical original position. The Rawlsian framework developed here thus furnishes methodological naturalism with added normative force. Of course, as before, whether the Rawlsian model delivers the normative force at issue depends very much on the inputs. If one is in the business of attempting to discover the fundamental and general truths about reality by unifying science, then it is relatively clear that progress toward that end will be maximised by adopting a naturalistic approach. If, however, one takes metaphysics to be in the business of discovering fundamental and general truths by conducting conceptual analysis, then the link to naturalism is less clear. Still, even in the case of conceptual analysis, if we really want to know what the structure of our concepts is, it seems reasonable to suppose that an empirical understanding of those concepts is required and thus that some dialogue with empirical psychology is in the best interests of the project. Perhaps this result is at odds with what those engaged in conceptual analysis are actually doing, but such is the divide between is and ought. At any rate, let us briefly consider two cases where the added normative force of methodological naturalism comes into play, to gain a flavour for the kind of work that notion can do once properly justified. First, consider the debate over the existence of God. In this debate there is a tendency by some metaphysicians to defend

18

theism in a manner that requires rejecting a great deal of science. I have in mind here those who defend intelligent design in such a way that they end up rejecting evolutionary theory. Many believe that defending the existence of God in this way is troubling. We can now see why: metaphysicians who defend the existence of God by appealing to intelligent design of the kind that is incompatible with our best science are cheating. They are cheating because they shirk the rules of metaphysics that they themselves would agree upon, from within the metaphysical original position. The case is therefore analogous to slavery in Rawls’s theory of justice. While slavery might benefit the slaveholders, it is a practice that they themselves would rule out from behind the veil of ignorance. They would rule it out because it is not a practice that is justifiable to everybody. Analogously, while ignoring science may further the aims and interests of metaphysicians who are theists, it is a practice that they themselves would rule out from behind the veil of ignorance. Similar considerations apply to the second example. As briefly noted above, there is a debate in the philosophy of time over the compatibility between the A-theory and the special and general theories of relativity. Some A-theorists have attempted to avoid this incompatibility by recommending a return to pre-Einsteinian notions of relativity. Craig (2000), for instance, argues that we should reject special relativity – one of our most well-confirmed scientific theories – because of this incompatibility, and embrace an aether theory of the kind defended by Lorentz. The debate on this topic is rather technical, but, for now at least, the details are not important. What matters is that scientists consider Lorentz’s theory to be a failed physical theory. It is therefore difficult to see how the recommendation to return to pre-Einsteinian mechanics properly accords with methodological naturalism. Recommending a failed scientific theory on metaphysical grounds does not satisfy the requirement to develop one’s metaphysical theories in a scientifically informed way. So Craig is cheating: he is failing to obey the rules of metaphysics that he himself would agree to be bound by from within the metaphysical original position. Now, I have no doubt that anti-naturalists will find it outrageous to suppose that flouting methodogical naturalism constitutes a form of cheating. The worry, as it ever is with any Rawlsian approach such as this, is that it is implausible to get so much out without implicitly putting it in. However, consider an analogous issue in the case of distributive justice. Consider, in particular, a conservative who rejects Rawls’s methodology of the original position because the two principles outlined in §2 are too ideologically substantive to follow from a mere methodological suggestion. To be convincing, that is not how the conservative should argue against those principles and the associated theory of justice. Rather, she must produce arguments to the effect that the rule-makers behind the veil of ignorance would not reach consensus, and thus there would be deep disagreement within the original position, or that they would reach consensus but would not converge on the difference principle. In the

19

same vein, an anti-naturalist who wants to attack methodological naturalism as I have defended it here should not do so by denying that anything quite so substantive can follow from mere methodology. Rather, she should provide an argument to the effect that metaphysicians in the original position would not converge on anything, or would converge on a rule for metaphysics that permits at least sometimes ignoring best (relevant) science. I have no doubt that such arguments could be developed, and so I do not view methodological naturalism to be unassailable. But developing arguments of this kind would be a substantial undertaking, and that’s all I aim to show. Here’s a sketch of one such argument for anti-naturalism. Above I mentioned the idea that science might be compatible with a number of metaphysical systems and dismissed the idea that it is compatible with all. Perhaps this was too quick however. For suppose that all scientific theories were compatible with all possible metaphysical systems and that this could be shown. If that is information that metaphysicians in the original position have available to them, then it is doubtful that they would legislate in favour of any kind of naturalism. Indeed, a vehement anti-naturalism, whereby metaphysics is developed independently of science, would seem to be the order of the day. Now, to be clear, I am not defending the idea that science is compatible with all possible metaphysical systems. I am merely trying to demonstrate the kind of argument that must be produced by the anti-naturalist as a response to the proposed Rawlsian derivation of naturalism. I leave it to the anti-naturalist to take the argument further.

4.2

Reasons For All

The second and third principles of metaphysics are less general than methodological naturalism, in the following sense. These principles are required to maximise individual motivational ideals only. The second principle is targeted, in particular, at the ideal of transparency, and can be stated in maxim form as follows: the reasons that are provided in favour of a metaphysical theory ought to be reasons for all suitably empirically informed rational agents. That is, the reasons provided in favour of a metaphysical theory ought not to be reasons that trade on implicit beliefs that only some rational agents hold. They should, rather, be compelling to everybody as reasons and, as such, in so far as they rely on implicit beliefs at all, they ought to rely on beliefs that pretty much every empirically informed rational agent is likely to have. This principle is more likely to maximise the ideal of transparency than the alternatives. Again, to see this, we require some sense of the alternatives. The alternatives in this case correspond to a class of principles, each of which allows some scope to offer reasons in support of one’s view, where the status of those reasons as reasons is something that not every suitable empirically informed rational agent would accept. At one end of this spectrum sits a principle which requires metaphysicians to offer 20

highly idiosyncratic reasons in support of their views, reasons that very few such agents, if any, would accept as reasons. As we move away from this position, each alternative principle renders metaphysics more transparent than the principle identified above. This is clearest for the extremal case. In the extremal case, metaphysics as practiced appears to be irrelevant to anyone except for a very select few. Each weakening of this extremal principle will render metaphysics more transparent to some degree, making it more relevant to other empirically informed rational agents. However, it is not until we come to the other end of the spectrum, and thus to the principle outlined above, that we reach a principle that maximises this ideal. As with methodological naturalism, this principle can be used to catch cheats. Here’s a toy example. Consider, again, the debate over the existence of God. In arguing for the existence of God one may cite, as evidence, testimony in favour of the existence of miracles. Many cite claims by persons to have special experiences of awareness of the divine. But not every empirically informed rational agent will agree on the evidential status of such testimony. Rather, many will maintain that testimony in favour of miracles is not to be trusted as evidence; and nor is testimony of special experiences of the divine. Similarly, suppose one takes the bible to be evidence in favour of the existence of God. Again, the status of the bible as evidence is very much up for debate. Now, the claim that there is something untoward about the use of these items of evidence as a basis for belief in the existence of God is nothing new. Nor, even, is the idea that this strategy of arguing for the existence of God amounts to a kind of cheating. What is new is that we now have a further sense of why this is cheating. It is cheating because it flouts a principle that even those metaphysicians who are theists would accept from behind the veil of ignorance. There is therefore a sense in which such metaphysicians are cheating by their own (ideal) lights. Similar considerations apply to various versions of supernaturalism in meta-ethics. Suppose Alice and Bob are two metaphysicians disagreeing about whether there are objective moral facts. Suppose that Bob offers, as evidence for the existence of objective moral facts, God’s existence. Alice can legitimately criticise Bob’s dialectical move as unfair. Bob has provided reasons in favour of his view that rely on holding metaphysical beliefs that not every empirically informed rational agent is likely to share. Alice can therefore demand of Bob that he provide further evidence in favour of the existence of moral facts, evidence that does not require Alice to become a theist in order to take it seriously. Of course, if Alice is already a theist, then perhaps it is legitimate for Bob to make the move that he does. As such, perhaps a caveat is in order. The principle offered above, namely that the reasons provided in favour of a metaphysical theory ought to be reasons for all, is sensitive to the context of a dispute. If every agent engaged in a particular metaphysical dispute accepts p as a basic assumption of the debate, then perhaps it is legitimate to provide reasons for or against a theory that

21

presupposes p, even if p is not universally accepted. However, ultimately p itself must be argued for, and in a way that is accessible to all (or if p is not accessible to all and relies on q, then it is q that must receive this treatment, and so on). In short, at some level, all metaphysical claims must be supported by reasons that rely only on widely held beliefs, in order to maximise the ideal of transparency. Metaphysical debates cannot bottom out, as it were, in evidence, the status of which as evidence relies on controversial assumptions.

4.3

Don’t Go Primitive

We come now to the third principle of metaphysics, which is the final principle I will discuss. The third principle speaks primarily to the ideal of progress. I have already discussed a principle that is required to maximise progress in metaphysics: methodological naturalism. But there is a further, less general principle needed to achieve this goal as well. Sometimes in the course of a metaphysical debate, someone might take something as primitive. While we want to allow the use of primitives within the context of a metaphysical theory – everyone is allowed some primitivity in their theories, to spread around wherever they like – at times the use of primitives seems questionable. I have in mind here the introduction of some new primitive in order to solve a problem for one’s theory, or the taking of something as primitive to the same end, where (i) the primitive is tailored specifically to solve the problem and (ii) there is no other reason for adopting the primitive than to solve the problem at hand. I will outline an example of this in a moment but first we need to get clear on the relationship between this kind of manoeuvre and the ideal of progress in metaphysics. The trouble with introducing new primitives to solve a problem for a metaphysical theory where (i) and (ii) apply is that pretty much any problem with any theory can be addressed using a variant of this strategy. Suppose a theory has some kind of metaphysical shortfall: there is something that we need the theory to do that it cannot, and for many that is a reason to doubt the view. In response to this problem one can, in principle at least, introduce some new primitive and simply stipulate that it does the relevant work. Or suppose that a theory maintains, on pain of inconsistency, that x and y are distinct, but cannot give a good account of the distinctness of these two things. One can simply stipulate that x and y are primitively distinct. In short, the move of taking something as primitive or introducing some new primitive into one’s theory is extremely powerful. If one is allowed to ‘go primitive’ in this fashion, then it will be all too easy to make our metaphysical theories impervious to criticism. Moreover, the temptation to do so will be significant, since going primitive in this way is an easy solution to one’s problems. It is also a lasting solution, since by its very nature it prevents further questions being asked (because one is invoking primitives), and thus further objections from being raised. 22

If going primitive is easy to do then progress in metaphysics will be difficult. The situation won’t be quite as bad as one in which all metaphysical theories are on a par, and so there is no way to choose between them, but it will be close. It won’t be quite this bad, because one can always call on parsimony considerations as a way of choosing between theories. Those theories that require going primitive less frequently than others will be the stand out theories, since it is better for a theory to reduce the number of primitives it has wherever possible. Even then, parsimony will only take us so far because it can only rule out a handful of theories. In the situation where going primitive is allowed, most theories will have a range of primitives tailormade to render each theory viable, and so most of these theories will be on a par from the standpoint of parsimony. Once we have identified those theories, there will be nothing left for metaphysics to do and so the entire enterprise will grind to a halt well short of its stated aims. In order to maximise the progress that can be made in metaphysics, we must minimise the use of primitives. Now, one might want to rule out the use of primitives completely to solve problems with a metaphysical theory, but that strikes me as a bit extreme. At any rate, it is the more moderate position that I will focus on here. This moderate position speaks in favour of a principle of metaphysics whereby some condition is placed on the use of primitives. It is not entirely obvious what the condition needs to be, but at the very least it should be strong enough to avert the scenario discussed above. Here’s one stab at formulating such a principle. One ought not appeal to primitives in order to solve a problem for one’s view unless one has very strong independent reasons for doing so, where ‘independent’ here means: independent of the problem at hand. Exactly what counts as a very strong independent reason is going to vary on a case by case basis, but the basic idea is that one must show that the primitives at issue do more work than merely solving a particular problem for one’s theory. This sanction against the use of primitives is a corollary of a more general principle, one that restricts the use of ad hoc assumptions in general. I suspect that the more general principle can also be justified using the Rawlsian model. But, for now at least, I will focus only on the invocation of primitives since I believe the notion of ad hocness hides complexities that I cannot adequately engage with here. As before, it is useful to consider an example. Presentism, a popular view in the philosophy of time, is the view according to which only present entities exist. Presentists find it difficult to satisfy the demands of truthmaker theory. Truthmaking requires, roughly, that for any proposition P , there exists some entity E such that E’s existence makes P true. But now consider some proposition about the past, such as . For the presentist, past entities do not exist. So cannot be made true by the existence of any dinosaurs. But it is far from obvious what else might make this proposition true, given presentism. If nothing

23

can be found to make the proposition true, however, then presentists are forced to accept that (and, indeed, all claims about the past) is false, an outcome that is generally thought to be unattractive. In order to satisfy the demands of truthmaker theory, then, presentists must find some further truthmakers for propositions about the past. In response to this demand, Bigelow (1996) invokes a certain kind of property, possessed by the world itself: a tensed property. Tensed properties are primitive, unanalysable properties of the world that, in some primitive sense, ‘point’ toward the past and, in virtue of so pointing, serve as the truthmakers for past-directed propositions. So, for example, the proposition is made true by the presently instantiated world-property having been such that there were dinosaurs, the existence of which guarantees that dinosaurs existed. Appealing to primitive tensed properties to solve the truthmaking problem for presentism is thought by many to constitute a kind of cheating. The manner in which this charge of cheating is developed, however, is usually metaphysical and not methodological. The move of introducing primitive tensed properties qua move is not itself questioned. Sider (2001), for instance, argues that tensed properties are hypothetical properties – properties that point beyond their instances – and so should be disallowed for this reason. Similarly, Cameron (2011) argues that tensed properties do not make an intrinsic difference to their bearers and, as such, are metaphysically contentious posits. In so far as a methodological worry has been pressed at all, the worry has been one of parsimony: invoking new primitives in this way is metaphysically profligate. There is, however, something troubling about the move that Bigelow makes that is unrelated to worries about parsimony. Tensed properties are introduced for the sole purpose of addressing a certain kind of objection against presentism. Moreover, this is a job that the relevant tensed properties are tailor-made to do. Their use as truthmakers does not receive independent motivation by Bigelow, and nor does the idea that these properties exist. One cannot help but feel, then, that Bigelow’s solution to the truthmaking problem has been achieved far too easily. The principle of metaphysics outlined above, concerning primitives, does justice to this intuition and, moreover, tells us what’s wrong with the move. Because no independent motivation for invoking these primitives has been given; because we are given no reason to believe in the existence of the relevant primitives, other than that they can be used to solve a particular metaphysical problem for presentism, the appeal to primitives in this context has failed to meet the right normative standards.

24

5

Conclusion

Here, then, are three putative principles of metaphysics: Methodological Naturalism

Metaphysicians ought to develop metaphysical theories in a manner that is informed by our best science.

Reasons for All

The reasons provided in favour of a metaphysical theory ought to be reasons for all empirically informed rational agents.

Don’t Go Primitive

One ought not to appeal to primitives in order to solve a problem for one’s view unless one has very strong independent reasons for doing so.

As noted, I offer these three principles in a provisory mood, as instances of how to apply a particular Rawlsian framework to the methodology of metaphysics. What really matters is the method used to justify these principles. The basic idea behind this method is to abstract away from the particular metaphysical disputes in which we are engaged and place ourselves in the metaphysical original position. From the metaphysical original position we attempt to agree on the principles of methodology by which metaphysics will be bound, once a veil of ignorance is lifted; a veil which, in this case, occludes pertinent information that might influence the sorts of decisions we make about which principles to adopt. This Rawlsian approach to determining and justifying the principles of metaphysics promises to deliver a set of principles that are justifiable to everybody engaged in metaphysical inquiry, and that are therefore fair. Moreover, because these principles are determined by the motivations of agents determined to do the best metaphysics they can, they are also likely to be in the best interests of metaphysics as a discipline. The refinement and application of the method developed here is therefore a worthwhile project in meta-metaphysics. At best, it will produce a rule-book for metaphysics as an end product, a guide that can be used to do real work within first-order debates, at worst it will tell us why no such guide is forthcoming. There is also the possibility that something like this method can be extended to other disciplines. One might, for instance, use this method to motivate and justify the norms of science, or of history. It is interesting indeed to consider just how far the Rawlsian paradigm can be pushed.

25

Acknowledgements I am very grateful to Jamin Asay, Christopher Daly, Thomas Dougherty, Raamy Majeed, David Ripley and three anonymous referees for this journal for comments on earlier drafts of the paper. I am also grateful to Miri Albahari, John Bigelow, David Braddon-Mitchell, Rachael Briggs, Mark Colyvan, Mauro Dorato, Peter Evans, Suzy Killmister, Kristie Miller, Bradley Monton, James Norton, Greg Restall, Michael Rubin, Caroline West, Robbie Williams and the audience at the 2012 Australasian Association of Philosophy Conference for extremely useful discussion on the ideas behind “Metaphysics as Fairness”.

6

Works Cited

Baker, A. (2003) Quantitative Parsimony and Explanatory Power. British Journal for the Philosophy of Science, 54(2): 245—259. Baker, A. (2012) Science-Driven Mathematical Explanation. Mind, 121(482): 243– 267. Bigelow, J. (1996) Presentism and Properties. Philosophical Perspectives (10): 35–52. Cameron, R. (2011). Truthmaking for Presentists. In Oxford Studies in Metaphysics Vol. 6. K. Bennett and D. W. Zimmerman (Eds.), 55—101. (Oxford: Oxford University Press). Craig, W. L. (2000). The Tensed Theory of Time: A Critical Examination. (Dordrecht: Kluwer Academic Publishers). Dummett, M. (1991). The Logical Basis of Metaphysics. (Harvard: Harvard University Press). Fine, K. (2012) What is Metaphysics? In Contemporary Aristotelian Metaphysics. T. E. Tahko (Ed.), 8–25. (Cambridge: Cambridge University Press). Jackson, F. (1998). From Metaphysics to Ethics. (Oxford: Oxford University Press). Kant, I. (1943). Critique of Pure Reason. (New York: Willey Book Co.). Ladyman, J., and Ross, D. (2007). Every Thing Must Go: Metaphysics Naturalised. (Oxford: Oxford University Press). Lowe, E. J. (2011) The Rationality of Metaphysics, Synthese, 178(1): 99–109. Lyon, A., and Colyvan, M. (2008) The Explanatory Power of Phase Spaces. Philosophia Mathematica, 16(2), 227-–243. Maclaurin, J., and Dyke, H. (2012) What Is Analytic Metaphysics For? Australasian Journal of Philosophy, 90(2): 290—306. Maclaurin, J., and Dyke, H. (2013) What Shall we do with Analytic Metaphysics?: A Response to McLeod and Parsons. Australasian Journal of Philosophy, 91(1), 179–182. Nolan, D. (forthcoming). Method in Analytic Metaphysics. In The Oxford Handbook of Philosophical Methodology. H. Capellen, T. S. Gendler, J. Hawthorne and D. Sgaravatti (Eds.). (Oxford: Oxford University Press). 26

Nolan, D. (1997) Quantitative Parsimony. British Journal for the Philosophy of Science, 48(3): 329–443. Paul, L. A. (2012) Metaphysics as Modeling: The Handmaiden’s Tale. Philosophical Studies, 160(1): 1–29. Rawls, J. (1999). A Theory of Justice. (Cambridge, MA: Harvard University Press). Schaffer, J. (2009). On What Grounds What. In Metametaphysics: New Essays on the Foundations of Ontology. D. Manley, D. J. Chalmers and R. Wasserman (Eds.), 347–383. (Oxford: Oxford University Press). Sider, T. (2001). Four-Dimensionalism: An Ontology of Persistence and Time. (Oxford: Oxford University Press). Sider, T. (2011). Writing the Book of the World. (Oxford: Oxford University Press). Strawson, P. F. (2011). Philosophical Writings. (Oxford: Oxford University Press). Wilson, J. (forthcoming). Three Dogmas of Metaphysical Methodology. In Philosophical Methodology: The Armchair or the Laboratory? M. Haug (Ed.) (Routledge) Zimmerman, D. W. (2008). The Privileged Present: Defending an “A-theory" of Time. In Contemporary Debates in Metaphysics. T. Sider, J. Hawthorne and D. W. Zimmerman (Eds.), 211–225. (Oxford: Blackwell Publishing Ltd).

27