Neutral Realism: A New Metaphysical Approach to Representation

Abstract

Metaphysics seeks an account of fundamental reality as it is independent of any observer or point of view. As such, one problem it faces is that any such account is necessarily created by some observer from some point of view. Does this mean that metaphysics is thereby inherently impossible? Or inherently incomplete? I argue that it is possible and it can aim at completeness, but it must acknowledge the contributions made by the human perspective on reality, human cognition, and features of the conceptual and linguistic representations in which it is couched. The idea that we can discover metaphysical insights by investigating concepts and language has had a remarkably tenacious grip on the field of metaphysics. I offer a diagnosis of how this grip took hold and an argument that it should be loosened. I also propose a means of pursuing metaphysical investigation that does not rely on an enquiry into language and that can yield fruitful results.

1. Introduction

In this paper, I investigate the relationship between our conceptual and linguistic representations of reality, and the reality they represent, in order to examine the metaphysical project. A methodology sometimes pursued in metaphysics is to interrogate representations of reality and infer conclusions from those representations about the fundamental, metaphysical nature of reality itself. I argue that this is a fallacious methodology. Furthermore, I argue that the underlying assumption that drives this methodology, concerning the relationship between representations and reality, has been responsible for establishing a particular view of the range of available positions that one can adopt with respect to the metaphysical underpinnings of any domain of discourse. That view conceals a further position, which has significant advantages over the alternatives, and which I call neutral realism.

According to neutral realism about some domain of discourse, it is possible for that domain to contain true, irreducible sentences, which are nonetheless neutral as to the nature of their truth makers. True sentences represent reality, but we cannot reliably draw conclusions about the fundamental ontological nature of their truth makers from observations about the nature of those true sentences. Neutral realism acknowledges that linguistic representations are very often context-dependent, incorporating features due to our perceptual and cognitive systems, as well as our perspective on reality, and the information we intend to convey about it by employing those linguistic representations. Any complete and correct account of reality ought to acknowledge these contributions. Finally, I make some suggestions about how metaphysics might develop if neutral realism is true. I argue that it ought to be naturalistic and a collaborative enterprise with all of the sciences, bringing together what the sciences tell us about reality and our perceptions, experience, and representations of it.

2. Representations and Reality in Metaphysical Enquiry

The aim of metaphysical enquiry is to provide an account of the nature of fundamental reality, as it is in itself, independent of any observer or point of view. We want to know what reality is like in itself, not just what reality is like round here, or what it is like to beings like us. For that reason we abstract away from particular perspectives, and particular cognizers, aiming for an account that achieves the status of ‘observer-independent’, a ‘God’s eye view’, sub specie aeternitatis. There are a number of things to say about such an aim. First, it’s important to remember that such an account will be just that: an account, a representation, something capable of being true or false. As such, it is necessarily an account of reality given by some observer from some perspective. If it is aiming for completeness then it should also include reference to the observer and the representation itself, but it is usual in these contexts to exclude any such self-reference, meaning that the representation is necessarily incomplete. As such, it should perhaps be thought of as a representation of an aspect of reality, or even a model of reality. Of course, such representations can make features of their perspective, such as spatial and temporal location, explicit, thereby progressing towards a non-perspectival account whose truth-value does not depend on any contextual features of the representation. The aim here is to give an objective representation, whose content contains only information about that which is represented, and whose truth value does not depend on any features of the perspective from which it is represented. Regardless, given that this is a representation of reality, an important question is how are we to understand the relation that it stands in to the reality it represents? I will return to this question.

Finally, there is a danger of mistaking, or conflating, a representation with what it represents. That is, metaphysicians should be careful not to draw conclusions about the nature of reality—that which is represented—from premises about the nature of the representation itself [1]. To do so would be to commit a fallacy known as the representational fallacy [2]. To commit the representational fallacy is to cross an inference barrier between two different types of proposition [3,4]. Such an inference barrier exists if and only if it is illegitimate to derive conclusions of one type of proposition from premises of the other type. Perhaps the most famous inference barrier is Hume’s First Law, according to which it is not possible to derive a conclusion containing an ‘ought’, a normative conclusion, from merely descriptive premises; no ‘ought’ from ‘is’ [5] (pp. 469–470). Others include the thesis that it is not possible to derive propositions of necessary truth from propositions of contingent truth (Kant’s Law), and that one cannot derive propositions of unrestricted scope from propositions of restricted scope [3] (p. 604), [4]. The representational fallacy is committed when one attempts to derive a proposition about the world from premises about representations of the world.

Keeping these points about representations in mind, let us return to considering the aim of a metaphysical enquiry. It aims to give an account of the nature of fundamental reality that is objectively true, independent of any observer or point of view. How should metaphysicians go about pursuing this aim? Metaphysicians are not scientists, although scientists too are interested in understanding the nature of reality. Scientific methods involve observation, experiment, measurement, and prediction; they are primarily empirical. Metaphysics of a broadly Quinean character, which arguably most mainstream metaphysics is, sees itself as continuous with science, in the business of developing generalisations from our best-confirmed scientific theories. Such a view aims to develop an ontological inventory of the most general kinds of existing entity by considering what kinds of entity must exist to account for the truth of our best-confirmed scientific theories. Already we can see that even the most naturalistic view of the nature of metaphysics, that sees it as continuous with science, is concerned with representations of reality and the relationship between representations and the reality they represent. Even these views focus on interrogating our representations of reality in order to draw conclusions about reality itself. That is not to say that they thereby commit the representational fallacy. Rather, it is to say that any metaphysical enquiry will inevitably pay some attention to representations of reality but, I argue, beyond that recognition there are both legitimate and illegitimate ways of proceeding.

3. The Way of Ideas and the Way of Reference1 [6]

Clearly, the language we use to represent reality is relevant to the metaphysical project, if only because our theories about reality are couched in language. We seek to understand reality by forming beliefs about it. Theories and beliefs are two ways of representing the world. A common view of the methodology of metaphysics is that it takes as its starting point our ordinary thought and talk about the world, our common-sense beliefs and our conceptual representations, and it asks what general kinds of entity must exist in order for those conceptual representations to be true [7,8,9]. However, there are at least two things metaphysicians might be doing when investigating these representations. They might be investigating our conceptual and linguistic representations of reality and drawing conclusions about the reality represented by them on the basis of features of the representations themselves. Taylor, in his recent monograph exploring the relationship between semantic analysis and metaphysical enquiry, calls this “the way of ideas” [1] (p. 38). A metaphysician pursuing the way of ideas claims to gain metaphysical insight into the fundamental nature of reality by interrogating the ideational or conceptual contents of our thought and talk. If we want to understand freedom, material objects, causation, or time, the way of ideas recommends that we analyse and interrogate our concepts of freedom, material objects, causation, and time.

Alternatively, the attention metaphysicians pay to language may be focused on the worldly objects referred to by our thought and talk about the world. This approach involves using language to say what reality is like, and seeking to establish what the truth makers are for these truths. Taylor calls this “the way of reference” [1] (p. 38). According to the way of reference, we gain metaphysical insight into freedom, material objects, causation, and time by employing the largely a posteriori means of interrogating concrete instances of free agency, material objects, causation, and time. This will usually include an appeal to scientific enquiry. In addition, and importantly, it is defeasible, our “antecedent, linguistically encoded concepts of free agency [material objects, causation or time] may fail to be a very reliable guide to the ultimate metaphysical nature of that real world phenomenon” [1] (p. 39). Furthermore, proponents of the way of reference can and should concede that our representational abilities, ideas, concepts, and language are what grant us cognitive access to reality in the first place. Where they differ from proponents of the way of ideas, however, is in denying that that which grants us this initial cognitive access to some aspect of reality will suffice to reveal the deep metaphysical structure of that aspect of reality [1] (p. 56). Our representations may be a starting point for metaphysical enquiry, but they are not its final destination.

Much, but far from all, metaphysical enquiry over recent decades has pursued the way of ideas. Taylor cites examples including Kripke’s arguments for natural kinds [10], which rely heavily on a semantic analysis of natural kind terms. Davidson’s metaphysics of events relies on an analysis of the logical form of action sentences [11]. Ludlow argues for a tensed metaphysics of time by way of a semantic analysis of temporal language [12]. More recent examples include Chalmers’ “conceptual metaphysics” [13], Thomasson’s “easy ontology” [14], and Hale and Wright’s approach to the ontology of mathematics [15]. Taylor offers a series of arguments against the way of ideas. First, for any given concept, it is at the very least possible that there is nothing in reality answering to it. At some point a priori interrogation of our concepts must “give way to a degree of a posteriori empirical inquiry into what there really is” [1] (p. 51). It took empirical investigation, for example, to discover that there was nothing answering to our concept of phlogiston. Merely attending to our concepts and representations of reality “in which we meet with nothing extra-representational [could not] possibly suffice to give us access to a realm of free-standing objects” [1] (p. 86). The point here is that the way of ideas fails to forge a link between our representations and the reality they represent, and so leaves us stranded within the realm of representations. As such, it cannot assist with the primary aim of the metaphysical project: to investigate the fundamental nature of reality itself. By investigating representations, and not the reality they represent, the way of ideas lacks a bridging principle that would warrant conclusions about reality.

It could be objected here that our linguistic and conceptual representations cannot be completely divorced from the reality they represent, as it is hard to see how we could achieve successful communication and navigation through the world if there were quite literally nothing tying our representations to reality. Our representations must, in some sense, reveal rather than obscure the reality they represent. In response, it is possible to accept that our representations do stand in some significant relation to reality, but still maintain that any investigation into those representations is not guaranteed to yield insights into the nature of reality so represented. The way of ideas presupposes that there is some antecedent guarantee that the world will answer to some privileged description of it; that our representations of the world stand in a one-to-one relation with the features of the world they represent. The way of reference denies that there is any such antecedent guarantee. It accepts that any true description of some part of reality has a truth maker, but maintains that interrogating the truth will not necessarily reveal much about the nature of the truth maker. At some point, empirical investigation will have to be brought to bear on the enquiry.

The way of reference denies that there is a privileged description of the world that stands in any such one-to-one relation with the world. Instead, it is possible for multiple non-synonymous truths to have the same truth maker [2] (p. 99). If this is possible, it follows that there is indeed an inference barrier between premises about representations and conclusions about reality. I argue below that there are many instances of multiple, non-synonymous sentences having the same truth maker. That being the case, it is not possible to infer conclusions about the reality represented by those sentences from premises about any particular one of those sentences.

Taylor develops this idea in terms of the notion of “direction of fit” [1] (p. 73), [2] (pp. 63–66). The way of ideas assumes that the world is guaranteed to fit our representations of it; it sees the primary direction of fit as being from the representation to the reality [2] (p. 64). By contrast, the way of reference sees our representations as attempting to fit the world, but defeasibly so; the primary direction of fit is from reality to our representations of it. To illustrate this idea, consider the relation between a map and the terrain it represents. We can certainly learn some things about the terrain by studying the map. However, all maps are selective about the information they convey, and what information they convey depends on the interests of the cartographers in making the map. Any map will leave undetermined certain elements of the terrain represented. Therefore, it would be a mistake to think that there is only one way the terrain can be given the way the map is. Similarly, we can learn some things about reality from investigating our true sentences about it. However, just like maps, true sentences are selective about the information they convey, and what information they convey depends partly on the interests of those uttering them [2] (p. 63). As Taylor notes, however, once we see the difference between the way of ideas and the way of reference as a difference in direction of fit, the possibility opens up that eventually there may be some convergence between them, although they will approach any point of convergence from opposite directions [1] (pp. 73–74).

A second argument against the way of ideas arises out of a consideration of the notion of conceptual change over time. Our concepts of water, solidity, and colour, for example, differ in various ways from those of the Ancient Greeks.2 The Ancient Greeks’ concept of water would primarily have involved perceptually grounded notions, such as being a clear liquid, and being thirst quenching. Our present-day concept of water is constituted by a much richer network that also involves scientific notions such as chemical structures, freezing and boiling points, and so on. Nevertheless, our concept and the Ancient Greek concept both refer to the very same stuff in the world. Conceptual schemes are not fixed and static, but evolving, in response to scientific and empirical discoveries. As a result, the attempt to gain metaphysical insight by investigating just our present-day conceptual scheme is likely to be fruitless, as that conceptual scheme will no doubt undergo further change in response to new scientific and empirical discoveries.

A final argument against the way of ideas involves what Taylor refers to as “problematically related domains of entities” [1] (p. 64). He cites examples: “the normative and the natural; qualitative states of consciousness and neurophysiological states of the brain” [1]. However, there are others: the domain of ordinary objects and the matter that constitutes them, the tensed and the tenseless, and the modal and the non-modal. What makes these pairs of domains problematic, according to Taylor, is that it is not obvious how the concepts in one domain are related to the concepts in the other. Attempting to explain the relationship between two such domains by deploying the way of ideas would lead one to focus primarily on the concepts in the domains, and not what they refer to in the world. One might then ask whether, for example, the concepts in one domain are reducible to the concepts in the other domain, such that the domain to be reduced is somehow less fundamental than, or parasitic on, the more fundamental domain?

One might ask whether, for example, moral concepts are reducible to non-moral, natural concepts. If one thinks they are, one might conclude that there are no moral properties answering to the moral concepts, only natural, non-moral properties answering to the non-moral concepts. One might further conclude that the non-moral domain is more fundamental, ontologically speaking, than the moral domain, and that the moral domain somehow supervenes on the non-moral domain. However, if reduction is not possible, proponents of the way of ideas might be tempted to conclude that there exist properties out there in the world, moral properties, that are distinct from any natural, non-moral properties. What leads them to this conclusion is the fact that moral concepts are irreducible to non-moral concepts. That is, considerations purely concerning our conceptual representations lead them to a conclusion concerning the nature of extra-linguistic reality.

Alternatively, proponents of the way of ideas might antecedently think that there are no moral properties, and conclude that all propositions involving moral predicates are therefore, contrary to common sense belief, false. Such a view would constitute an eliminativist or error-theoretic approach to moral discourse. Conversely, they might think that despite appearing to be fact-stating, in reality, moral propositions perform some other linguistic function, in which case they might opt for expressivism or emotivism about moral language.

This approach to metaphysical enquiry, via consideration of problematically related domains of entities, commits the error of conflating features of representations with features of what they represent. From the fact that one concept is not reducible to another, it does not follow that there must exist in reality some particular entity or property answering to that concept, that is itself ‘irreducible’ to any other entity or property. Irreducibility is a feature of representations, conceptual or linguistic, not entities out there in the world.3 As we saw above, concepts might have nothing at all answering to them. To conclude from the irreducibility of one kind of concept to another kind of concept that reality must therefore be thus and so is to cross the inference barrier discussed above. It is to draw conclusions about the nature of reality from premises about the nature of our representations of reality.

Taylor makes a similar point but expresses it in terms of conflating possibilities for our representations with possibilities for objects themselves. A possibility for a representation is “a matter of what that representation might have represented, had the extra-representational facts gone differently” [1] (p. 67), while a possibility for an object concerns “how things might have been for the object itself, had things been otherwise” [1] (p. 67). Taylor illustrates this with Putnam’s example of H₂O and its Twin Earth counterpart XYZ [16]. It is a possibility for our water representations that they might have represented XYZ, but being XYZ rather than H₂O is not a possibility for water itself. Consider also the riddle attributed to Abraham Lincoln: if you call a dog’s tail a leg, how many legs does it have? The answer, as Honest Abe was quick to point out, is four. Calling a tail a leg does not make it a leg. Therefore, possibilities for representations do not automatically translate into possibilities for the objects represented. Conceptual distances between the concepts in problematically related domains do not necessarily imply ontological distances between the entities answering to those concepts.

Yet, as Taylor and others have noted, metaphysical enquiry is too often pursued deploying the way of ideas, blithely crossing the inference barrier between conceptual content and world, and with an implicit conflation of possibilities for representations with possibilities for objects in the background, driving the enquiry along. In the next section I will outline how this can happen, and show that this approach leads to a general agreement about what constitutes the range of acceptable metaphysical positions in such a way that one viable metaphysical position is structurally concealed. I will illustrate this using as a concrete example the debate over the existence of tensed facts in the philosophy of time.

4. A Viable Metaphysical Position Concealed by the Way of Ideas

There is often widespread disagreement about the metaphysical underpinnings of any given domain of discourse. Proposed views range from full-blown realism, to varieties of anti-realism, or eliminativism. For any kind of discourse, M-discourse,4 [17] metaphysicians ask: What must reality be like in order for M-sentences to be true? Despite differences between the answers given to this question, there is often broad agreement about what constitutes the range of acceptable metaphysical positions. Take mathematical discourse, for example. Suppose the sentence ‘There is a prime number between 11 and 14′ is true. A realist about mathematical discourse will say that the sentence is true in virtue of the existence of a mathematical fact, a fact involving mathematical entities; objects such as numbers, and properties such as ‘being prime’. For the realist, such facts are denoted, or referred to, by such sentences; the mathematical facts act as truth makers for the mathematical sentences.

Others, who reject the existence of mathematical facts, entities, and properties, will disagree, but there are different ways in which this disagreement can be developed. For reductionists, since there are no mathematical facts, any sentence the truth of which appears to require their existence must be reducible to some other sentence which lacks that metaphysical import. They will offer, perhaps, a naturalistic sentence that is committed only to the existence of natural properties and facts, and show how the mathematical sentence can be reduced to the naturalistic sentence. Others may argue that since there are no mathematical facts, any sentence which appears to require their existence must be false. This approach characterises the error theorist, the eliminativist, instrumentalist, and fictionalist, who may then argue in their various ways that even though all (atomic) mathematical sentences are false, mathematical discourse serves a particular purpose so it is useful for us to retain it and treat it as if it is true. The non-cognitivist, or expressivist, might argue that, despite appearances, mathematical discourse is not really fact-stating at all, so mathematical sentences are not even candidates for truth or falsity. A similar range of positions can be taken with respect to other domains of discourse. I illustrated above a parallel treatment of moral discourse, generating realist, reductionist, error-theoretic, and non-cognitivist positions. Consider also modal, mental, tensed, and religious discourses.

This understanding of the metaphysical landscape conceals an important position that is available for any kind of discourse [2]. This position holds that, from the fact that there are truths of some domain of discourse, M-truths, it does not follow that there are particular facts or properties associated with just that discourse: M-facts or M-properties. The M-truths are representational entities that do not determine the ontological nature of what they represent, but rather are consistent with a range of possible ontologies. Furthermore, maintaining this position does not require one to find an alternative sentence—an ontologically more palatable sentence—to which the original sentence can be reduced. The reductionist, error theorist, and non-cognitivist approaches are all unnecessary.

A position of this kind emerged with respect to tensed discourse in the philosophy of time in the 1980s. Previously, the debate had been carried out between tense-realists and tense-reductionists. According to tense-realists, tensed discourse was ineliminable from language. They concluded from this that tensed facts exist (for example, [18]). Their opponents, tense-reductionists, antecedently denied the existence of tensed facts, but felt compelled by this denial to explain the existence of tensed truths. They did so by arguing that tense could be eliminated from language; that is, that tensed discourse could be replaced by, or reduced without loss of meaning to, tenseless discourse (for example, [19]). The new position that emerged in the 1980s—the new B-theory of time—took issue with this dialectic. New B-theorists agreed with tense-realists that tensed truths are indeed ineliminable from language. However, they also agreed with tense-reductionists in rejecting the existence of tensed facts (for example, [20,21]). In order to maintain this position, they must have rejected the underlying assumption that facts about representations somehow determine, or reveal, the precise ontological nature of what they represent. In other words, adopting such a position involves rejecting the way of ideas. In general, opposition towards the way of ideas has been growing in recent years among both metaphysicians [2,22,23] and philosophers of language [1].

In my view, the implicit commitment to this understanding of the relationship between our linguistic and conceptual representations, and the reality they represent, causes a significant problem for metaphysical theorising. It is responsible for the widespread agreement about the available range of positions with respect to a given domain of discourse, in such a way that the important option adopted in the philosophy of time is concealed. To see this, consider the position of the realist who argues that since there are irreducible M-truths, there must be M-facts to which those truths refer or which make those truths true. This inference is only valid if one accepts the following assumption:

The truth-to-fact conditional: If an M-sentence is true, it designates an M-fact.

Realists accept this conditional. The only explanation, according to them, for the truth of an irreducible M-sentence, is the existence of an M-fact.

However, it is important to note that commitment to the truth-to-fact conditional also explains the reductionist, error theorist, and non-cognitivist strategies. Reductionists reject the existence of M-facts, but since they accept the truth-to-fact conditional they feel compelled to explain away the apparent commitment to M-facts by M-truths. They do this by seeking ontologically more palatable alternatives to the M-truths, which carry no commitment to M-facts, and argue that the M-truths can be reduced to these other truths. This is metaphysical theorising via the way of ideas. The metaphysical conclusion is reached via a consideration of the relationship between two conceptual domains. For the reductionist, the only facts that M-truths commit one to are ontologically acceptable facts. Error theorists, eliminativists, and non-cognitivists also reject the existence of M-facts, but they do so by denying the truth of any M-sentences. For the error theorist and eliminativist, there are no M-facts, so all the (atomic) M-sentences must be false. For the non-cognitivist, there are no M-facts, so M-sentences, despite appearances, must be doing something other than stating facts.

As well as explaining the rationale behind each of these realist and anti-realist approaches to M-discourse, acceptance of the truth-to-fact conditional also explains why this range of metaphysical approaches is typically seen as exhaustive. Accepting it, while rejecting the existence of M-facts, leaves a limited range of options, all requiring attention to be focused on the nature of the M-truths. What distinguishes the concealed metaphysical approach is its rejection of that conditional. It takes the antecedent of the conditional to be true: there are (irreducible) M-truths. But it rejects the consequent of the conditional, maintaining that there are no M-facts. Furthermore, it sees no requirement to explain away the M-truths in order to maintain that position. It rejects the reductionist approach as unnecessary. I call this approach neutral realism [24]. Like the realist, the neutral realist accepts that there are irreducible M-truths, but unlike the realist she does not take this acceptance to commit her to the existence of any M-facts. She is neutral as to the nature of the truth makers of the M-truths. She may of course have other, non-linguistic arguments for a particular view of the nature of the truth makers of M-truths. This is true of Mellor [20] and other proponents of the new B-theory [25,26,27]. According to these philosophers there are good reasons for thinking that time is not tensed, that is, that there is no distinction between past, present and future, and no flow of time. In particular, they appeal to scientific thinking about time that supports this view, as well as, in their view, the untenability of the view that time is tensed. What is important, however, is that these reasons are not based on any analysis of tensed language or concepts.

In the next section I will expand on, and argue for, neutral realism, and explain how it sees the relation between conceptual and linguistic representations and the reality they represent.

5. Neutral Realism

Neutral realism about some domain of discourse is the view that it is possible for that domain to contain true, irreducible sentences, while those sentences are neutral as to the nature of their truth makers. It thus holds that while true sentences represent reality, it is not possible to draw conclusions about the fundamental, ontological nature of reality merely from observations about the nature of those true sentences. The true sentences from some domain of discourse, M-truths, might be irreducibly tensed, modal, moral, aesthetic, and so on, but those are features of the representations and not necessarily of their truth makers.

Neutral realism can be shown to be true if it is possible for two or more non-synonymous sentences, that are irreducible to one another, to nevertheless have the same truth maker. The existence of such a scenario would establish that there is no direct route from any particular true sentence to the nature of its truth maker. Suppose S1 and S2 are non-synonymous, true, and that there is no way to reduce S1 to S2 or vice versa. Suppose also that there is one fact in the world, F, that makes both S1 and S2 true. In that case it will not be possible to determine the precise nature of F by examining the nature of S1 or of S2. That is, there is no direct route from either S1 or S2 to the nature of F. Is such a scenario possible? I will give three examples.

A token of the sentence ‘the meeting starts now’ uttered at time t, and a token of the sentence ‘the meeting starts at t’, do not have the same meaning [28,29,30] and they cannot be reduced to each other. According to the new B-theory, however, they are both made true by the same fact, namely, that the meeting starts at t. It is important to note, however, that neutral realism itself makes no substantive claims about the nature of the truth maker for any two such sentences. It simply opens up a space for metaphysical theorising about the nature of their common truth maker. The new B-theory brings other considerations to bear in arguing for its view of the common truth maker of these two sentences.

To adapt an example given by de Laguna [31],5 consider the event constituting Joe Biden’s victory in the 2020 presidential election. We can describe this event as a Democratic victory using the sentence ‘The 2020 presidential election was a Democratic victory for Joe Biden’. Alternatively, as de Laguna remarks, that event,

is resolvable into a vast mass of occurrences, such as the going to the respective polls of the voters all over the country, the marking of ballots, the subsequent fall of the ballots into the boxes, etc. And each of these occurrences may be similarly broken up, until, as an ideal limit, we may conceive that whole group of events which constituted the election and the Democratic victory as a multitude of redistributions of mass and transformations of energy.

[32] (p. 179)

The same event described by the sentence ‘the 2020 presidential election was a Democratic victory for Joe Biden’, can thus also be given a very long and complex physical description. Suppose we use the placeholder predicate ‘is P’ to represent the multitude of physical properties possessed by this complex event. Then, the two sentences ‘the 2020 presidential election was a Democratic victory for Joe Biden’ and ‘the 2020 presidential election is P’ are both made true by the same fact in the world. One suggestion for the nature of their common truth maker is that it is the fact that the 2020 presidential election is P. The two sentences, however, clearly do not have the same meaning.

Idiomatic language provides a third example. The sentence ‘Helen is barking up the wrong tree’ is true if and only if Helen is looking in the wrong place for a solution to her problem. It does not require Helen to actually bark up a tree in order to be true. More generally, the figurative meaning of an idiom is different from the literal meaning of the composition of the individual words that make it up. It follows that the sentence ‘Helen is barking up the wrong tree’ and the sentence ‘Helen is looking in the wrong place for a solution to her problem’ are both made true by the same fact in the world, even though they do not have the same meaning. Again, one suggestion for the nature of their common truth maker is that it is the fact that Helen is looking in the wrong place for a solution to her problem.

It follows from these three examples that it is possible for a pair of non-synonymous sentences to have the same truth maker, which is all that is required for neutral realism to be true. Accepting neutral realism allows us to differentiate between the implications of our language for the world and its implications for us, the language users. In each of these examples, the two sentences convey different information from each other, even though they are both made true by the same fact. In the temporal case, the tensed sentence conveys additional, perspectival information about the utterer’s temporal location. Our use of the political sentence reveals something about our political context and concerns, while our use of its physical counterpart reveals our interest in achieving a correct physical description of the event. Our use of the idiomatic sentence is a more engaging, vivid, and perhaps a more memorable way of expressing information about Helen and her project, than our use of its non-idiomatic counterpart.

Consideration of examples such as these shows that our use of language is very often context-dependent to some extent. It often reveals something about the language-users and their interests, as well as about the facts referred to. Neutral realism accommodates this by allowing for multiple, non-synonymous sentences to have the same truth maker, even when those sentences are not semantically reducible to each other. It is not the case that one such sentence is more fundamental or more directly gets to the truth about some event, object, or part of reality. The same portion of reality can truly be described in many different ways, and how we choose to describe it is as much to do with us, our interest in it, and what information we wish to convey about it, as it is to do with the nature of the bit of reality being described.

Contextual features such as these affect or alter our linguistic representations. That being so, we should be particularly wary of drawing metaphysical conclusions from linguistic representations. Consider the object located on my desk. It is a framed photograph. It is a photograph of my children; a photograph taken by my husband ten years ago in England. It is a coloured photograph; an object made of wood, paper, and glass; and a physical object. It is a collection of molecules. It is sometimes used as a paperweight, and so on. When we describe some portion of reality we do so from a particular perspective, and with our own interests in that portion of reality in place, influencing our choice of description. The context of our descriptive language use can make just as important a contribution to the meaning of what we say as does the feature of reality that it describes.

6. The Future of Metaphysical Enquiry: Naturalistic Metaphysics

I have not argued that all of contemporary metaphysics is based on the fallacious way of ideas. On the contrary, there is much metaphysical enquiry that proceeds via the way of reference, and yet other metaphysical enquiries that make no inferences about reality from premises about language. However, I have argued that where metaphysical enquiry does proceed via the way of ideas, it should be treated with caution. I have further argued that adopting the way of ideas as a legitimate metaphysical methodology conceals from view the position I have been defending, neutral realism, which is a potentially fruitful metaphysical program. However, all this leaves one important question unanswered: if neutral realism is true, how are we to do metaphysics? In this section, I will make some tentative suggestions in answer to this question, using metaphysical enquiry into the nature of time, once more, as an example.

We saw at the beginning of this paper that metaphysics is concerned with giving an objective account of the nature of fundamental reality, a goal it shares with scientific enquiry. Since science and metaphysics both have the same subject matter, viz., the nature of reality, an initial constraint on metaphysics is that it ought to be naturalistic. That is, it ought to be consistent and continuous with our best current scientific theories [33]. Naturalistic metaphysics pursues the goal of metaphysical enquiry in a manner that is continuous with science which, broadly speaking, means that it is consistent with science and that it is capable of informing and being informed by scientific and empirical discoveries. However, there are different positions one can take with respect to the question of what it means to be continuous with science and the relationship between science and metaphysics, some more stringent than others. At one end of the spectrum, it is thought that metaphysics can only be of use in the service of science, because science is the primary means by which we can develop a theory of the world [34]. At the other end of the spectrum is the view that philosophy takes precedence in our enquiries into the nature of reality [35]. A more moderate position is that if metaphysics is to be taken as having something direct to say about reality, then it must take seriously the implications of science and, in particular, fundamental physics [36].

Following the moderate position of French [36], if the philosophy of time is to be taken as having something direct to say about the nature of time, then it must “properly appreciate” [36] (p. 212) the implications of our best scientific theories about the nature of time, in particular, the Special Theory of Relativity (STR). According to STR, the simultaneity relation between events does not obtain absolutely, but only relative to a frame of reference. It is therefore difficult to see how STR can be reconciled with the claim that there is an absolute present and the associated claim that time passes or flows from future to present to past. The A-theory holds that there is an absolute present and there is temporal passage, while the B-theory holds that there is no absolute present and no passage of time; instead, time is constituted by the temporal relations earlier than, later than, and simultaneous with, obtaining between times and events. On the face of it, the B-theory looks better placed than the A-theory to be consistent and continuous with what our best current scientific theories tell us about the nature of time.

The B-theory, however, faces a significant challenge, which is that it departs quite radically from what our temporal phenomenology and experience appear to tell us about the nature of time and our ordinary, common-sense beliefs about it. This is where the A-theory appears to have an advantage, as it seems to comport well with what ordinary experience and common-sense beliefs tell us about the nature of time.

However, it is important to realise that being continuous with science means being continuous with all of the sciences, not just physics, and I argue that the B-theory is better placed to satisfy that demand. A naturalistic metaphysical theory of time must sit comfortably and consistently within the network of findings of all scientific investigations. It must be consistent with what fundamental physics tells us about time, but it must also be consistent with what cognitive science, neuroscience, psychology, and environmental biology tell us about the nature of human temporal experience.

The A-theory, I argue, meets neither requirement. On the one hand, it fails to properly appreciate the implications of physics. However, it is also insulated from, and therefore not continuous with, what the special sciences tell us about human temporal experience. The A-theory says that there is an objective present moment and time flows, and that this would be the case whether or not there were any observers. These features characterise time independently of human, or any other, experience. Time would be this way no matter what our temporal experience was like.

The A-theory claims that our temporal experience is straightforwardly veridical. Our perceptual and cognitive apparatus acts like a transparent window on the world, showing us what time is like independently of us. But what we’re learning from the special sciences about human temporal experience is that this is not the case. Our perceptual and cognitive capacities make their own contribution to the character and phenomenological feel of our temporal experiences [37]. Our temporal experience is affected by myriad external factors, such as how bored or engaged we are, how novel are our experiences, and whether we’re in a life-threatening situation.6 [38] By taking our temporal experience to be straightforwardly veridical, the A-theory effectively rules out any further illumination on either time or temporal experience from work in the special sciences.

The foregoing is not intended to be a knock-down argument against the A-theory. Instead, it is intended to show that when we take seriously the requirement to properly appreciate the implications of our best scientific theories, that should include not just physics, but all of the sciences. In the philosophy of time, this approach opens a pathway to developing a metaphysical theory of time that is both consistent with our best physics of time, and seeks to explain why our temporal experience is the way that it is. In my view, the B-theory is best placed to pursue this goal. Furthermore, the considerations that count against the A-theory in this regard bring us neatly back to our earlier discussion of the opposing approaches to metaphysics: the way of ideas and the way of reference.

The A-theory builds its case for an objective present moment and objective temporal passage on the nature of our temporal experience, temporal intuitions, temporal language, and our common-sense beliefs about time. Consider, for example, this quote from prominent A-theorist, William Lane Craig:

I should say that belief in the reality of tense and temporal becoming enjoys such powerful positive epistemic status for us that not only can we be said to know that tense and temporal becoming are real, but also that this belief constitutes an intrinsic defeater-defeater which overwhelms the objections brought against it.

[39] (p. 138)

Craig here exemplifies the A-theory’s tendency to adopt the way of ideas. He interrogates our conceptual, linguistic, and psychological representations of time, and draws conclusions about the nature of time from them. The problem with this approach in this particular domain is that there are explanations from psychology, cognitive science, and neuroscience that explain much about our temporal phenomenology and experience, which in turn influence our conceptual, linguistic, and psychological representations of time. Add to this the fact that there is significant evidence from physics that the A-theory is false, and it starts to look much more promising to seek to explain the nature of our temporal representations by appealing to cognitive science, neuroscience, and psychology. The evidence suggests that the highly intuitive character of the A-theory derives from facts about us and our psychology, and does not reflect the scientific or metaphysical facts about the world.

What this approach shows is that it is important to understand whether a set of intuitions is due to our developmental cognitive or psychological features that lead us to see the world in a particular way. If we discover that we would have developed these intuitions regardless of whether they reflect objective features of the world, it would not necessarily follow that they are misleading. After all, if we would have developed them whether or not they are veridical, there are at least two explanations for why we have them. One is that they are veridical, and another is that there is some evolutionary advantage to us in seeing the world that way [40]. However, it does suggest that these intuitions would be extremely compelling even if they are misleading. Therefore, it would be advisable for us to investigate their origins and understand the cognitive processes that lead to them, rather than simply assume that they are veridical.

Such an approach is a paradigm example of the way of reference as endorsed by Taylor [1]. The way of reference acknowledges that we have true representations of reality, but denies that we can learn much about the fundamental nature of reality from interrogating them. Instead, we should investigate the truth makers of our true representations to discover the nature of reality. This investigation involves empirical and scientific enquiry into the nature of reality, and it also acknowledges that our representations themselves constitute part of reality. Our investigation into the nature of time proceeds by paying due attention to what our best current scientific theories tell us about it and also seeking to understand why we have the temporal phenomenology, experience, and intuitions that we do. It recognises that we may have these temporal representations for reasons other than that they are veridical.

Goldman argues that “empirical findings in cognitive science can play a significant evidential role for an optimal methodology for metaphysics” [41] (p. 171). If we are interested in the nature of reality as it is independently of us, then we ought to be interested in the apparatus that we use to perceive and experience that reality, because features of our experience may be introduced by our perceptual apparatus rather than by the way things are independently of us. Furthermore, there is a growing view among cognitive scientists that perception is a generative, creative act. On this view, our brains are “prediction machines” [37] (p. 166), engaged in figuring out the most likely causes of the sensory signals they receive. Additionally, importantly, this function has come about as a means of enhancing our survival prospects. It has not evolved in order to be a transparent window on the world, so the extent of its transparency depends on its utility, not the other way around.

Bringing these ideas together suggests that the future of metaphysical enquiry is, at least in part, one of collaborative enterprise with the sciences and other areas of enquiry, rather than primarily an armchair-based, a priori enterprise. Seeing the task of metaphysics as part of a joint project with other special sciences is an approach that has been emerging in recent years. Callender, for example, argues that reconciling what science tells us about the nature of time with how time appears to us in experience is a project that requires a collaborative effort from science, physics in particular, but also psychology, physiology, neuroscience, and evolutionary biology [23]. Linguistics, semantics, and logical analysis have their role to play too. Philosophy brings particular strengths to this interdisciplinary project, in particular, the ability to see past disciplinary boundaries and offer an overarching view of the project.

Taylor diagnoses the urge towards the way of ideas as symptomatic of the fear that philosophy will have nothing to contribute to metaphysical inquiry if it cannot be done a priori [1] (p. 106). Callender takes a similar view. Their work, among others, demonstrates that metaphysics is undergoing an evolution and embracing its role as part of an interdisciplinary project to uncover “the vast and layered labyrinth of existence in its sprawling totality” [1] (p. 165).

7. Conclusions

I have argued that careful attention needs to be paid to the difference between our conceptual, linguistic, and psychological representations of reality, and the reality they represent. Not paying due attention to this difference risks fallacious inferences being drawn about the nature of reality from the nature of representations, and risks conflating representations with what they represent. In my view, some metaphysical investigation, either wittingly or unwittingly, has failed to pay due attention to this distinction, and this has led metaphysics astray. It has also been responsible for concealing a particular view about the relation between the truths of some domain of discourse and the truth makers of those truths, a view I call neutral realism.

According to neutral realism about some domain of discourse, it is possible for there to be true, irreducible sentences in that domain, but the sentences themselves are neutral as to the nature of their truth makers. In order to discover metaphysical insights into the nature of those truth makers, we must target the truth makers themselves, as well as our perceptual, cognitive, linguistic, and otherwise representational faculties. This is because neutral realism recognises that some features of our intuitions, common sense beliefs, and beliefs based on our experience may be introduced by our representational faculties, rather than by the way things are independently of us. The resulting view of the metaphysical project is that it is primarily a collaborative enterprise involving philosophy together with fundamental physics and the special sciences.