Setting the story straight: fictionalism about grounding

Abstract

This paper explores a middle way between realism and eliminativism about grounding. Grounding-talk is intelligible and useful, but it fails to pick out grounding relations that exist or obtain in reality. Instead, grounding-talk allows us to convey facts about what metaphysically explains what, and about the worldly dependence relations that give rise to those explanations.

There has been an explosion of recent work on grounding. The term ‘grounding’ is semi-technical; it has a technical meaning in academic philosophy, and is also used in a related but non-technical way in ordinary language. Philosophers most often take ‘grounding’ to refer to a worldly relation that obtains between facts or propositions, though some (e.g. Schaffer, 2009, 2010) think that grounding relations obtain between entities of various other ontological categories. In all its guises, the term ‘grounding’ conveys some kind of explanatory relationship between the ground and what is grounded, such that the latter exists or obtains because or in virtue of the former. In this paper, I explore the possibility that there is no worldly grounding relation picked out by expressions generally thought to denote such a relation. I argue that utterances of sentences about grounding can instead to be understood as a kind of fiction or pretence. (My aim here is not to argue for the view as such, but to ‘put it on the map’.)

I begin in section one with a very brief account of the current state of play of debates about grounding. In section two I outline my preferred account of fictionalism about grounding, and section three offers some further detail about the nature of the fiction.

1 The state of the debate

Philosophers interested in grounding fall (for the most part at least) into one of two camps: realists and eliminativists. Realists about grounding think that grounding is a feature of objective, mind-independent reality. There are grounding relations, and those relations exist or obtain independently of anything anybody might think or say about them. Here are some paradigm examples of grounding claims:

• (i) The ball is coloured because it is green

• (ii) Moral facts depend on natural facts

• (iii) The mental is grounded in the physical

• (iv) The fact that P grounds the fact that P ∨ Q

Different locutions (‘because’, ‘grounds’, ‘depends on’)¹ might all be taken to express grounding relations.² As many prefer to think of grounding as a relation between facts, they would prefer to formulate (i) as (i*), where square brackets designate facts:

(i*) [the ball is green] grounds [the ball is coloured].

Orthodoxy has it that the formal properties of grounding include asymmetry, irreflexivity, transitivity, non-monotonicity, and hyperintensionality, though at least the first three have been challenged in the literature (see e.g. Bliss, 2018; Jenkins, 2011; Rodreguez-Pereyra, 2015; Schaffer, 2012; Thompson, 2016b). One key motivation for positing grounding is to cash out the idea that reality is ordered or structured; some facts or things metaphysically depend on other facts or things. Most commonly, philosophers maintain that reality has a well-founded structure; a subset of facts or things explain or account for (in this metaphysical sense) all of the other facts or things, and don’t themselves depend on anything further. This subset of entities are fundamental, and derivative entities are related to the fundamental in linear, non-repeating chains of ground. For example, biological facts might be grounded in chemical facts, themselves grounded in physical facts, which are grounded in microphysical facts. If nothing further grounds the microphysical facts, then those facts are fundamental. Again, this foundationalist orthodoxy has been challenged (see e.g. Bliss, 2013, 2014; Morganti, 2009; Thompson, 2016b), as has the idea that fundamentality should be accounted for in terms of grounding (see e.g. Barnes, 2012; Bennett, 2017; Sider, 2011; Wilson, 2014), but those details need not concern us here.

Eliminativists about grounding deny that there are any grounding relations, and advocate abandoning talk of grounding altogether.³ They might claim that grounding talk is incoherent or esoteric (e.g. Daly 2012; Hofweber, 2009, 2016) or that it has no distinctive role to play (Wilson, 2014; Koslicki, 2015; Miller & Norton, 2017). Very roughly, those who think grounding talk is incoherent claim not to understand purported examples of grounding such as those given above, or to find them lacking in (grounding-related) metaphysical substance. They might also argue that other notions that grounding might be thought to be associated with (such as explanation or fundamentality) are either too close to the notion of grounding to be helpful in illuminating it, or too far away. Those who think grounding-talk has no useful role to play might think that the work earmarked for grounding is all done by finer-grained dependence relations (such as mereological relations, determinate-determinable relations, truthmaking, identity, and so on), or that the work earmarked for grounding is simply not useful or interesting in the first place; nothing need play the grounding role. Miller and Norton (2017) argue that we can explain the observations generally thought to provide evidence of ground without appeal to the presence of such a relation.⁴

The position I develop and advocate in this paper treads a path between these opposing realist and eliminativist views. Grounding talk is neither unintelligible nor uninteresting, but expressions taken by realists to pick out objective, mind-independent grounding relations do not so refer. There are a number of interesting ways in which this non-eliminative irrealism about grounding might be developed,⁵ but I take the fictionalist proposal developed here to be one of the more attractive alternatives.⁶

2 Fictionalism about grounding

Many philosophers think that when I say ‘Harry Potter is a wizard’, I don’t really express the proposition < Harry Potter is a wizard > ⁷(or that if I do, what I express is strictly and literally false). It seems sensible to understand the purpose of my utterance as conveying a true proposition about a fiction: < according to Harry Potter, Harry Potter is a wizard > . Fictionalists think that something similar might be the case in other domains of discourse. For example, fictionalists about morality hold that moral claims (e.g. ‘murder is wrong’) express propositions that are strictly and literally false (because there aren’t any moral properties). Nevertheless, some moral claims are acceptable because they say something correct according to some kind of fiction or game of make believe, which serves some useful purpose.⁸

Fictionalists about grounding can all agree that grounding talk serves some non-literal purpose (i.e. some purpose other than that of accurately reporting facts about grounding relations in mind-independent reality). They can also agree (i) that when a competent speaker makes a typical utterance of some grounding sentence S, she doesn’t assert (i.e. ‘get behind’) the proposition normally associated with S, and (ii) that acceptance of S does not involve belief in that proposition.⁹ So, to utter ‘moral facts are grounded in natural facts’ is not to assert the proposition < moral facts are grounded in natural facts> , and accepting S doesn’t imply belief in the proposition <moral facts are grounded in natural facts> . That said, fictionalism comes in many different flavours. The differences between different fictionalist positions are substantial, and so there is little to be gained for our purposes here by discussing fictionalism about grounding at this coarse level of grain. Instead, I’ll focus on outlining my preferred account of fictionalism about grounding. To lay the groundwork, I start with a discussion of dependence and explanation.

2.1 Dependence relations and explanation

Reality is furnished with dependence relations. There might be mereological relations between parts and wholes, set-membership relations between sets and their members, determinate-determinable relations, and type or token identity relations or functional realization relations between the mental and the physical. Setting aside the question of whether there are also grounding relations, these myriad non-causal dependence relations abound.¹⁰ Jessica Wilson has argued at length that attention to metaphysical dependence is not new: ‘many, perhaps most, contemporary metaphysicians have spent their careers investigating forms of such dependence’ (2014, 539). Debates about the relationship between the mental and the physical, about the nature of properties, laws, objects, and persons, and about reductionist accounts of realism all serve to illustrate Wilson’s point.

With dependence comes structure, and we can therefore think of reality as replete with pockets of structure: part-whole structure; mathematical structure; logical structure; modal structure; causal structure; and so on. Dependence is not just a preoccupation of philosophers. The following sorts of conversations are common: “are you coming for a drink tonight?”, “I’m not sure—it depends whether Dave can pick up the kids.”; “Why didn’t Sam apply for that London job?”, “I don’t think she could convince Katie to move.”; “Riz! You mustn’t hurt animals!” “But it’s a gerbil!” “Yes, and gerbils are animals!” Counterfactual dependence relations generate counterfactual structure, causal relations generate causal structure, and conceptual dependence relations generate conceptual structure.¹¹ Thinking and talking in terms of these kinds of dependence is familiar and ordinary, and they play an important role in everyday reasoning (Hofweber, 2009, 268).

Structures generated by the obtaining of different dependence relations are not commensurable with one another. When we’re thinking about the structure generated by set membership relations, my desk is fundamental. It doesn’t depend, in set-membership terms, on anything else (though other things—the set with my desk as its sole member, the set of that set, and so on—depend on it in the relevant way). But it doesn’t follow that, mereologically speaking, my desk is fundamental. Quite the opposite. My desk is composed of four legs and a top, all themselves composed of molecules, composed of atoms, composed of sub-atomic particles. As such, my desk should be considered a derivative entity. It’s not possible to map the set-membership structure onto the mereological structure (just as it’s not possible to map causal structure onto set-membership structure).¹² These various dependence relations don’t on their own give rise to generic metaphysical structure of the kind grounding relations are taken to generate. They do, however, figure in explanations.

Dependence relations allow us to account for something in terms of some other thing or things (this comes out clearly in the above examples), and they thereby license explanations of the former thing in terms of the latter. Multiple non-causal relations all seem to give rise to one particular kind of explanation: an explanation of what makes something the case. This type of explanation is to be distinguished from the kind of explanation that causal relations give rise to: explanations for why something happens. When we ask why the window is broken, we expect an answer like ‘I threw a brick at it’. But this doesn’t metaphysically explain the window’s being broken. For that we need to know what makes it the case that the window is broken (perhaps that its parts are scattered).

Causal explanation is our explanatory paradigm, and given that causal explanations all appeal to a single generic causal relation,¹³ it is natural to assume that metaphysical explanations too are unified by involving a single relation. That relation is generally thought to be grounding. By appeal to grounding we can think of the world as exhibiting a generic metaphysical structure over and above the disparate local pockets of structure generated by the obtaining of patterns of finer-grained dependence relations.

But we are exploring a view according to which there aren’t (really) any grounding relations; a view according to which metaphysical explanations couldn’t be backed by (or identical to) worldly grounding relations. I want to suggest that there might be no single relation that backs metaphysical explanation, but that there is a pervasive fiction according to which explanations of a given type are all backed by a generic grounding relation. The presence of the grounding fiction allows us to project the generic metaphysical structure that is characteristic of ‘big-G’ grounding (see Wilson, 2014) onto reality. Grounding is a notion with which we can do useful work, but which is not, strictly speaking, part of metaphysical reality. The grounding fiction provides a useful way to convey information about metaphysical explanation and about genuine (fine-grained) dependence.

2.2 Two fictionalist accounts

When a speaker makes a grounding claim S, she expresses a proposition which is strictly and literally false, because it contains an expression (‘grounds’) which has no referent.¹⁴ Nevertheless, fictionalists might maintain that the speaker conveys something with her utterance that might be correct or otherwise, rendering the utterance itself either appropriate or inappropriate. Let’s call this the real content of S. The (false) proposition usually associated with S is the literal content of S, but this is not, according to the fictionalist, typically asserted when a speaker utters S. (We might say instead that the speaker quasi-asserts S, where quasi-assertion has the outward appearance of assertion, but commits the speaker merely to accepting rather than to believing what is said). A fairly straightforward fictionalist position one might take in the debate about grounding is a metalinguistic fictionalism, such that the real content of a speaker’s utterance P: ‘moral facts are grounded in natural facts’ is P*: < according to the fiction of grounding, moral facts are grounded in natural facts> . (The literal content of P is simply < moral facts are grounded in natural facts> .)

This metalinguistic fictionalism demands answers to some questions that can also guide the development of more sophisticated fictionalist accounts. Initially, we might wonder whether fictionalism is a thesis about the actual nature of the grounding discourse, or rather a normative position about how the discourse ought to be. This is the distinction between hermeneutic and revolutionary fictionalism (see Stanley, 2001, 36). The fictionalist positions I consider here are versions of hermeneutic fictionalism,¹⁵ and so an account is owed about how and why utterances about grounding have come to convey something other than their literal content.

The fictionalist first owes an explanation for why, if propositions about grounding are strictly and literally false, we don’t (and shouldn’t) simply avoid talking in terms of grounding altogether. In response, the fictionalist can highlight the benefits of engaging in grounding talk. These benefits are wide-ranging. Grounding-talk is simplifying and systematic; it allows us to formulate and to relate various established metaphysical principles (Rosen, 2010); and it enables fruitful discussion of issues of philosophical interest (Dasgupta, 2017). The recent burgeoning literature making use of grounding-talk is at least a prima facie case for thinking that the introduction of grounding-talk has enabled philosophical progress.¹⁶

Moreover, it is clear that we do in fact engage in grounding-talk in philosophy. If we grant to the fictionalist that propositions about grounding are strictly and literally false, then we have two options. We can revise our practice and stop talking in terms of grounding, or we can revise our understanding of the nature of the discourse. Given the explosion of recent work on grounding, the latter seems the less revisionary option. Grounding-talk has been used to further debates in a number of areas, and by reinterpreting the discourse we can continue to reap those benefits.

Take, for example, the dispute between physicalist and non-physicalist philosophers of mind. The former claim whilst the latter deny that the physical (fully) grounds the mental. The grounding eliminativist must reject both grounding-based claims, where the fictionalist can embrace them. The eliminativist can plausibly claim that the real work is done in explaining how the physical grounds the mental, but it seems nevertheless true that there is an important division in logical space between those who think that there is some metaphysical explanation of the mental in terms of the physical, and those who deny this (see Dasgupta, 2017, 86). Continued engagement with the grounding discourse allows us neatly to express this division.¹⁷

It is my contention here that the benefits of talking in terms of grounding are not lost if it turns out that there is in fact no single relation playing the role that grounding is thought to play, because the fiction of grounding is sufficient to perform the work of simplification and systemisation, to facilitate discussion, and to allow for helpful reformulation of various metaphysical principles. Nevertheless, the metalinguistic fictionalism introduced above doesn’t provide any explanation as to why this would be the case. Grounding-talk has instrumental benefits, but it’s not clear why the fiction developed in the first place. (And without any such account, it’s tempting to think that the realist about grounding must be right after all).

We can supplement the fictionalist story by appealing to our earlier observation about causal explanation: our familiarity with why-questions and with causal explanations as generated by an apparently unified causal relation might lead us to expect that metaphysical explanations elicited by (variants of) what-makes-it-the-case-that questions are similarly generated by a single generic, unified relation.¹⁸ Grounding seems like an excellent candidate for such a relation. We are interested in metaphysical explanations of how things fit together, and we expect a single relation to be involved in those explanations. There is no such relation, but we have good reason to think there might have been, and speaking as though there is serves us well in our theorising.

But without further supplementation, this metalinguistic fictionalist account requires a happy accident: that we erroneously come to believe that there is a grounding relation, and that this false belief in fact serves us well. A more satisfying and complete explanation would explain why this false belief is useful; what connects the nature and purpose of the grounding fiction. Perhaps the metalinguistic fictionalist can provide such an account, but I wish now to suggest an alternative variety of fictionalism which meets this challenge.

According to what (following Yablo (2001)) I’ll call figuralism, the real content of a grounding claim is not a claim about some kind of fiction. Grounding-talk is a figure of speech used to convey (or perhaps even assert—I won’t take a stand on this here) some other kind of information. What is conveyed when a competent speaker utters some sentence S about grounding (‘[A] grounds [B]’) is that the world is in a condition such that [A] is metaphysically explanatory with respect to [B]. [A] is rendered metaphysically explanatory with respect to [B] by the presence (or assumed presence) of some kind of dependence relation that obtains between (the contents of) [A] and (the contents of) [B].¹⁹ This might be a determinate-determinable relation, a set membership relation, a mereological relation, a supervenience relation, or any other non-causal dependence relation.²⁰ The speaker’s grounding claim thus conveys something about the world (that some such dependence relation holds) and something about explanation (that [A] is metaphysically explanatory with respect to [B]). It is fictional that there is any inflated relation of ‘big-G’ Grounding holding between [A] and [B]. The grounding fiction is useful not just because of the theoretical benefits of talking in terms of grounding, but also because doing so allows the speaker to convey this information about dependence and explanation.

2.3 The purpose of the fiction

An obvious objection to make against the fictionalist view described above is that it is undermotivated because there is little reason to participate in the fiction as opposed to telling the literal truth about the finer-grained metaphysical dependence relations. It is part of a majority of fictionalist views that what can be expressed or conveyed using the fiction could also be stated literally, but it is an aspect of our communicative practice that we often appeal to metaphor or to pretence in order to make a point. There might be various different reasons for this, including simplicity or ease of expression, a lack of precision, the history of the discourse, and artistic licence.

In the case of the figuralism described here, it seems likely that there are a number of contributing factors. First, it is simpler to talk in terms of grounding than it would be to talk in terms of a disjunction of metaphysical dependence relations and the system of metaphysical explanation we impose on them (just as it is simpler to talk about ‘the even numbers below 11’ than it is to talk about ‘two or four or six or…’). Second, the grounding discourse has an unusual history in that it is a semi-technical notion that codifies a folk concept of explanation. As noted above, it is plausible that theorising about causal explanation has led us to expect that a single, unified relation is involved in metaphysical explanations, and that we now have an inflated conception of the role of that relation (see Dasgupta, 2017). Third, engagement in the grounding fiction imposes a kind of regimentation on discourse about metaphysical explanation.

Our familiar understanding of explanation is of a phenomenon rich in epistemic overtones. Explanations are often communicated, and even when we think specifically about what it is that is communicated (as opposed to about the communicative act) it is tempting to think that a condition on something counting as an explanation (or, at least, a successful explanation) is that it increases our understanding. It’s not clear whether we should think of the formal features of explanation as inherited from the worldly dependence or determination relations with which they are associated, or as a consequence of epistemic constraints (e.g. circular explanations are bad because they are uninformative).²¹ Friends of metaphysical explanation often insist that what they have in mind is a purely ontic affair, free from such epistemic considerations. But it’s not really clear how such a notion of explanation is to be understood. To avoid these difficulties, it is easier to talk directly about the determination relations involved, and to insist that they are explanatory relations. The fewer relations we need do this for, the better. We talk in terms of grounding because it presents a simple way of talking about explanation free from epistemic connotations.

The above does not commit us to the view that metaphysical explanation is free from epistemic overtones. The view we have presented so far is consistent with a variety of different accounts of the nature of metaphysical explanation. These range from robustly realist accounts whereby one fact metaphysically explains another in virtue of the dependence relations between the contents of the facts, to (with a little bit of re-interpreting) pragmatic accounts whereby metaphysical explanations are responses to contextually-determined what-makes-it-the-case-that questions (where those answers will mention dependence relations).²²

2.4 Structure and grounding-talk

On the realist’s conception of grounding, grounding relations furnish reality with its structure. According to fictionalist about grounding, there is, strictly speaking, no such generic structuring relation. One might argue then that fictionalist ground cannot serve the purpose of limning reality’s structure. Here again, the figuralist has a response unavailable to the metalinguistic fictionalist. The grounding fiction allows us not only to project a notion of generic metaphysical structure onto reality which we can use to guide our investigation in the ways described above, but also enables us to convey truths about what structure reality does have. Through the grounding fiction the speaker conveys information about pockets of structure generated by myriad non-causal dependence relations and assimilated in a single system of metaphysical explanation.

For the hermeneutic fictionalist, this further helps explain why the fiction developed in the way that it did: using the fiction speakers convey what truths there are about structure. But once we have recognised that grounding talk is mere useful fiction, we might wonder again whether we couldn’t merely state directly the facts about the relevant notions of dependence and about how we take things to fit together in an explanatory system.

In response, note that in speaking literally in this way, something would be lost. There are multiple benefits of talking in terms of grounding, and the ability to convey truths about structure was just one of them. In using the grounding fiction a speaker is able to make a claim about structure (e.g. ‘mental facts are grounded in physical facts’) without having any specific beliefs about the particular species of dependence relevant to this case. In making her grounding claim the speaker conveys that the physical facts are explanatory with respect to the mental facts (whatever relation it is that licences this explanation). Her grounding claim can thus be appropriate even when the speaker herself doesn’t have entirely clear beliefs about the real content of her utterance. This explains some of the benefits of grounding-talk: it is useful for limning issues of metaphysical interest and for guiding our investigations because it allows us to convey information about dependence even when we are in doubt about the specifics. If we assume that the explanatory intuitions that lead us to make the grounding claims we do are tracking worldly dependence relations, then we should expect our grounding claims to convey information that this correct.²³

The question of what makes a grounding claim appropriate deserves some further attention. What is conveyed through the grounding fiction is that the world is such that some fact or entity is metaphysically explanatory with respect to some other. There are thus two different sorts of things that combine to make what is conveyed correct (and thus to make a grounding claim appropriate): (i) a worldly contribution and a (ii) speaker contribution. (i) is the condition the world is in with respect to some metaphysical dependence relation between the relevant entities, and (ii) is the contribution the speaker makes in constructing a generic system of metaphysical explanation out of those worldly dependencies. A speaker might convey something which is incorrect and thus make a grounding claim which is inappropriate by making a mistake about (i); by misrepresenting the condition that the world is in (e.g. by claiming that ‘mental facts are grounded in physical facts’ when there is no dependence relation running in that direction between the mental and the physical). This sort of mistake is akin to claiming ‘Rome is on the heel of the Italian boot’; the claim is inappropriate because it conveys incorrect information about the world. But contribution (ii) has to do with something which might be far more subjective, and could be subject to some variation between speakers even when both are right about whatever condition the world is in. Different people with access to the same information might maintain different accounts of what metaphysically explains what, if our account of metaphysical explanation allows us to build in some agent-sensitivity. This gives rise to a potential objection to our version of grounding fictionalism.

Suppose Nathan is a moral naturalist; he says that moral fact Φ is grounded in natural fact Ψ. Norman the non-naturalist denies this grounding claim. Grounding fictionalism has it that when Nathan says M: ‘Φ is grounded in Ψ’, he conveys that the natural facts are metaphysically explanatory with respect to the moral facts. (Perhaps he thinks the natural properties constitute moral properties, and this is what gives rise to the relevant explanation;²⁴ perhaps he’s not sure). Norman denies this grounding claim; he does not take the natural facts to explain the moral facts in the relevant sense. The worry is that when Nathan says ‘M’ and Norman says ‘it is not the case that M’, both speak correctly. But we are presumably unwilling to accept that both M and not-M are correct in the grounding fiction.

In this scenario, it is likely that Nathan and Norman’s disagreement is not merely over the facts about metaphysical explanation. If Norman were to agree with Nathan that the natural properties constitute the moral properties, he might also take the natural facts to be metaphysically explanatory with respect to the moral facts. One way we might therefore expect Nathan to convince Norman then is to try to persuade him that he, Nathan, has the correct view about the constitution of the moral facts by the natural facts. Disagreements about grounding will often be conducted in this way; through scrutiny of the objective, worldly conditions that give rise to the grounding claim. This is not distinctive to grounding fictionalism, but is the way disagreements about grounding ordinarily proceed according to the realist.²⁵

But where the fictionalist might depart from the realist is in maintaining that there might be cases where we have agreement on all the relevant dependence facts, but take them to give rise to different systems of metaphysical explanation.²⁶ In such cases, where there is disagreement it would seem as though both participants in the fiction speak appropriately, and we have the seemingly unpalatable result that both a statement and the denial of that statement can be appropriate or correct in the fiction. This is a prima facie challenge to the usefulness of the grounding fiction since presumably if the fiction is to play a useful role in metaphysics, it should play by the established rules; inconsistencies are not to be tolerated.

The fictionalist can respond to this challenge by agreeing that inconsistencies in the grounding fiction are not to be tolerated. When we have a case where one participant affirms P and the other not-P, they will attempt to resolve the apparent conflict with further argument, and those arguments will appeal to the broader philosophical views of the participants. This is precisely what we should expect, given what the fiction conveys. Though there might be a degree of subjectivity surrounding judgements about metaphysical explanation, we have a strong sense that others (once we agree on the objective facts) should bring their beliefs about explanation into line with ours. Judgements about explanation are more like judgements about morality than they are like judgements about taste. The fiction thus lends an air of objectivity to our judgements about metaphysical explanation and so to our account of structure, providing further reasons to conduct our discussions about structure in terms of grounding.

Nevertheless, it is strictly speaking the case that the grounding fiction could be inconsistent, keen as participants in the fiction are (and, for practical reasons, should be) to resolve them when they notice them. This is to be expected: there is no ban on inconsistencies in fiction, and there are various examples of them in the literature.²⁷ That the grounding fiction might be inconsistent is in line with our understanding of fiction in general. That we are keen to iron out the inconsistencies is no surprise given what the fiction conveys, and helps maintain the utility of the fiction.

3 The nature of the fiction

We have explored the development, content, and purpose of the grounding fiction. In this section we address further questions about the nature of the fiction.

3.1 Representational aids and things represented

Here’s a worry about the view as it stands. There is a burgeoning literature which deals with the nature and properties of the grounding relation. As I briefly mentioned above, there are questions about, for example, whether grounding relations are transitive, asymmetric, and irreflexive, and whether they form a well-founded partial order. When a philosopher makes a claim like T: ‘grounding is transitive’, the purpose of her utterance seems to be to convey something about the fiction itself.²⁸ This is to be contrasted with the more standard case of making a grounding claim when the purpose of the utterance is (according to the fictionalist) to convey something about the condition the world is in such that it gives rise to a particular system of metaphysical explanation. The fictionalist account should make room for this difference.

The situation here is similar to that with a sentence like ‘the Italian Boot is sexy’. The sentence is literally false (there is no Italian Boot in the relevant sense), and the speaker is not using the fiction to convey anything about Italy. When she utters this sentence, the speaker seems to be focussing on the fiction directly (whereas usually she merely uses the fiction but the focus of her attention is the world). Borrowing terminology from Yablo (2001), we can say that in these cases the Italian Boot functions as thing represented (rather than as representational aid).

Discussions about the nature and properties of grounding are discussions about grounding itself, as the thing represented. Disagreement is genuine disagreement about the content of the grounding fiction, and not about the real-world conditions conveyed using the fiction (though of course, both parties’ views will be informed by their beliefs about the non-fictional world). When philosophers argue about the properties of grounding, they really are arguing about the properties of grounding: they turn the fiction into an object of study in its own right. Where the fictionalist departs from the realist is that she takes this argument to be more like arguing about whether the stupefied relation is transitive than it is like arguing about whether causation is transitive.²⁹

Theorising about grounding as thing represented is not guaranteed to be particularly useful for finding out about the world. In turning the fiction into an object of study in its own right, its window onto the world is distorted. Insights into the fiction gleaned by reflecting on grounding as thing represented might be misleading if we try to project them back onto the world. For example, we might convey our beliefs about the way in which a series of things are explained by other things by uttering a series of grounding claims. Here, grounding functions as representational aid. But we might then consider a number of these series of grounding claims, and draw some conclusions about the nature of grounding itself. Now grounding functions as thing represented. Suppose we then take that conclusion (e.g. T: grounding is transitive) and use it to support a judgement that (e.g.) metaphysical explanation is transitive. This would be a mistake. To avoid such mistakes we must be mindful of when grounding functions as representational aid and when as thing represented.

3.2 How many fictions?

The discussions in Sects. 2.4 and 3.1 above makes salient another question the fictionalist must have an answer to; that of how many grounding fictions the fictionalist is willing to endorse. In the face of apparent disagreement, one option open to the fictionalist is to claim that more than one fiction is in operation. One might even go so far as to say that each participant is the author of his or her own grounding fiction. If we think of explanatoriness as an agent-relative notion, this might seem an attractive thing to say; grounding talk conveys information about metaphysical explanation, but this information about what metaphysically explains what depends in part on the interests, abilities, and situation of the explanation seeker.

One might think that to take this kind of line would greatly weaken the claim that the grounding fiction serves a useful purpose. If engaging in the grounding fiction is to allow us to make progress in framing, directing, and even resolving philosophical disputes, it will be important that all parties to the dispute are operating with the same conception of grounding in mind. If grounding talk is mere useful fiction, the fiction they are engaged in must be the same one. Arguments about what grounds what would be too easily resolved if grounding is relativized to an individual, and so disputants shown to be talking past one another.³⁰

This line of argument is a little quick. We could make sense of the conveyed content of a grounding utterance along non-cognitvist lines, taking such utterances not to assert or convey some non-literal propositional content (such as a proposition about metaphysical explanation) but rather to express a certain state of mind. When a speaker says ‘[A] grounds [B]’ on this view, she conveys her understanding of [B] in terms of [A]; when she considers [A], she feels a sort of [B] directed ‘ah-ha!’ If understanding varies with context by depending, for example, on the speaker’s other beliefs and cognitive abilities, then we might be tempted to deny that there is genuine disagreement about grounding. What disagreement there is is about whether or not to proceed in investigating [B]; a disagreement in a plan (see e.g. Gibbard, 2003, 69).³¹

This conception of what is going on when we make a grounding utterance explains away the appearance of genuine disagreement about grounding rather than explaining how grounding talk might still be useful despite such disagreement. We might supplement the position by insisting that on this conception grounding talk might still helpfully inform and direct further enquiry, but it is hard to shake the thought that grounding talk is far more useful if participants are engaged in the same fiction, with no threat of faultless disagreement or mere talking past one-another.

The alternative is to say that there is only one grounding fiction, or at least that grounding fictions are few in number. I think there are good reasons to think that there is a single grounding fiction insofar as the fiction is used to represent the world. However, there are (at least) two fictions of grounding as thing represented. An analogy should help here. A fairytale or a parable might be told in a number of different ways, though the basic story is the same. There might be differences in the language used, in which character makes certain remarks, in the complexity of the telling, and so on. In each case however, not just the story but also what the story conveys about the world—the ‘moral’ of the story, is consistent.

Similarly, different grounding frameworks might be used to convey the same worldly content, to express the same facts about explanation and dependence. It is common (in conversation if not in print) to hear people distinguish between ‘Fine-grounding’ and ‘Schaffer-grounding’. Where Fine-grounding relates only sentences and propositions, Schaffer-grounding relates entities of various ontological categories. Fine-grounding has consequences for debates about realism and antirealism that Schaffer-grounding doesn’t have. Schaffer-grounding is a worldly relation in a way that Fine-grounding isn’t. But importantly, Fine-grounding and Schaffer-grounding are different ways to express the same kind of story about a way in which things are related, providing reality with a unified structure. And this story is itself taken by the grounding fictionalist to convey further facts about metaphysical explanation and dependence.

3.3 Competing views

To finish, I’d like to suggest how fictionalism compares with a couple of competing views in order to help clarify the fictionalist’s distinctive position. I can’t present comprehensive arguments against these alternatives here, but I hope that in drawing attention to them, the fictionalist position I introduce here will become clearer.³²

The first is a view I call ‘explanation-first unionism’, according to which grounding claims are to be identified with claims about metaphysical explanation. This view differs from the perhaps more standard ‘grounding-first unionism’ by taking metaphysical explanation to be the primary notion. We understand grounding in the light of our understanding of a notion of metaphysical explanation, rather than the other way around. This sort of view is found in Dasgupta (2017), and can be thought of as a deflationary approach to grounding; the grounding facts are trivially settled once we settle the facts about metaphysical explanation (or ‘constitutive explanation’ to use Dasgupta’s terminology).

The key distinction between the deflationist and the fictionalist is that whilst the deflationist would usually be thought of as a realist about grounding, the fictionalist is an antirealist. The deflationist takes many claims about grounding to be literally true, where the fictionalist takes them to be systematically false. Even so, the views may seem similar in that in both cases, it is judgements about metaphysical explanation that underwrite claims about grounding. However, whilst the deflationist identifies facts about metaphysical explanation with facts about grounding, for the fictionalist the connection is looser. Grounding claims are used to convey judgements about systems of metaphysical explanation and the dependence relations that give rise to them, but a grounding claim is not the same thing as a judgement about metaphysical explanation.

One of the advantages that the fictionalist has over the deflationist is the ability to recognise a distinction between grounding and metaphysical explanation. Unionists of any kind must either insist that metaphysical explanation is entirely objective, or deny that grounding is (see Thompson, 2016a). Neither option is generally considered attractive. The fictionalist about grounding earns (or perhaps steals) the right to talk about grounding as an objective relation, whilst maintaining tht metaphysical explanation is (or at least might be) subject to various different pragmatic constraints.

A second competing view is grounding pluralism. The kind of pluralism most helpfully contrasted with the fictionalism I have outlined here would be a view according to which there are multiple different grounding relations, but no unifying notion of ‘big-G’ Grounding in addition to those relations. This denial that there is (really) any such notion along with the recognition of the role of local (or small-g) dependence relations is common to both the fictionalist and the pluralist, though the fictionalist claims (and the pluralist denies) that utterances about ‘big-G’ Grounding are useful and that such talk should be maintained despite its literal falsity.

One of the difficulties for the grounding pluralist is that they must maintain on the one hand that these local dependence relations are similar enough all to count as grounding relations, but not similar enough to merit thinking that they are determinates of a common determinable, or that they form a genuinely unified class. It’s instructive to note that I am not aware of the kind of pluralism just introduced having any proponents in the literature. Instead, philosophers with a view in the vicinity tend either to be grounding sceptics (as with Wilson, 2014) or proponents of a weaker kind of pluralism according to which there is a role for a unified notion of grounding, but that grounding is always (or at least usually) accompanied by a finer-grained relation (see Trogdon, 2018 and Griffith, 2018 for different versions of this view).³³

Whilst the fictionalist thinks of the fine-grained, local dependence relations as playing an important role, they are not (as Wilson would have it) the whole story. Perhaps then the greater challenge to fictionalism comes from the weak pluralist, who agrees with the fictionalist that coarse-grained ‘big-G’ Grounding talk plays various useful roles.

One advantage that the fictionalist has over the weak pluralist is that the fictionalist can explain how local dependence relations might differ in their formal properties without undermining the claim that grounding is unified. For the fictionalist, the relationship between local dependence relations and grounding is somewhat distant, mediated as it is through judgements about metaphysical explanation. The fictionalist can impose onto the local dependence relations a way of fitting things together that means that sometimes the presence of a particular relation is ignored, or overridden by the presence of another relation. It also allows that certain dependence relations might sometimes but not always contribute to a system of metaphysical explanation, and so that metaphysical explanations might sometimes but not always run in the same direction as a particular local dependence relation. It is therefore compatible with local dependence relations having different formal properties from one another and from those usually associated with grounding (as e.g. Wilson, 2014, 568–570 argues that they do).

There is a sense then in which the grounding fictionalist expects both more and less from grounding than the pluralist. She expects more in the sense that she doesn’t think that we can understand the role or even the mechanism for grounding purely in terms of local dependence relations. She expects less in the sense that she doesn’t require there really to be any grounding relations. The fictionalist thinks that we should appeal to grounding insofar as it plays a useful role in our theorising, and agrees with the monist and with the weak pluralist that it does indeed play such a role. However, like the strong pluralist, she doesn’t think that we have sufficient reason to believe in the existence of an extra relation over and above the local dependence relations. In my view, this is the best of both worlds.

There is much more to be said in fleshing out a complete fictionalist picture, but the challenges facing the realist and the eliminativist and the story that the fictionalist can tell about the development and the uses of the grounding fiction suggest that fictionalism about grounding is a promising avenue for further investigation.