1 Introduction

The grounding relation is supposed to be a distinctive relation of non-causal determination that supports metaphysical explanation. Realists argue that positing it allows us to, among other things: appreciate the commonality among various more specific non-causal determination relations; understand what is involved in metaphysical explanation; define important notions such as fundamentality; and state philosophical theses at a fortuitous level of grain. Realists hold that positing grounding helps us to understand better the structure of reality and our cognition of it via explanation (see e.g. Raven 2015, Rosen 2010, and Schaffer 2009).

Eliminativists dispute these benefits and prescribe that we demur from grounding-talk altogether. For example, Wilson (2014) highlights the importance of the various specific non-causal relations of determination (e.g. mereological parthood, type/token-identity, functional realization, and the determinate/determinable relation) and argues that positing grounding in addition gets in the way of good metaphysics. Other eliminativists claim that the notion of grounding is unintelligible (Daly 2012).

The dialectic here is familiar. In many ontological disputes, realists lean on the theoretical benefits of positing a certain kind of entity, while antirealists dispute those benefits or highlight the costs of the posit. In situations like this, a popular antirealist move is to adopt fictionalism, on which sentences concerning the relevant kind of entity are strictly untrue, but treating them as though they are true brings all the benefits the realist could want. Fictionalism offers a halfway house between realism and eliminativism.

Thanks to Thompson (2022), grounding fictionalism is now officially on the table. Thompson defends hermeneutic fictionalism on which sentences about grounding are actually used non-literally to convey claims about what metaphysically explains what. But there is also a nearby revolutionary proposal, on which sentences about grounding are actually used literally to communicate about grounding, but should be used non-literally to communicate about metaphysical explanation.

Despite the apparent dialectical kinship with other debates where fictionalism has a healthy tradition, I claim that the situation with respect to grounding talk is different in two key respects. Firstly, grounding talk is not indispensable, nor even particularly convenient as a means of communicating about metaphysical explanation (§2). Secondly, talk of grounding primarily occurs within metaphysics, which means the usual options for motivating a non-literal interpretation are ineffective (§3). These differences remove motivation for the revolutionary and hermeneutic proposals, respectively.

A cautionary note before continuing. Thompson’s proposal (and my criticisms of it) are premised on the assumption that grounding is an ontological posit. This assumption is not universally shared, even within the mainstream of grounding theory. One prominent theory declines to posit a grounding relation, but retains the notion of grounding as what is expressed by sentential connectives like ‘because’ and ‘in virtue of’, or some formal-language counterpart thereof (see e.g. Correia 2010; Fine 2012; and Schnieder 2011). Such connective theorists are neither realists nor eliminativists. There may be a further problem for Thompson here: the availability of non-revisionary, ontologically innocent approaches to grounding may lessen the appeal of fictionalism. However, I set this complication aside in what follows.

2 Against Revolutionary Fictionalism

The central challenge for fictionalism is to answer the following question: Why engage in talk that is systematically untrue? The fictionalist must show how the relevant sentences can be used for something valuable enough to be worth it. The answer to this challenge can be relied upon to help explain why people use the sentences of the discourse in this way (for a hermeneutic proposal), or to urge that they should use them in this way (for a revolutionary proposal). I focus on revolutionary fictionalism in this section.

For brevity and concreteness, let the proper relata of the grounding relation be facts (according to the story); let the predicate ‘x grounds y’ mean that x partially grounds y (where y may have other grounds as well); and, if A is a declarative sentence, let [A] be the fact that A and < A > be the proposition that A.

Drawing on Yablo (2001), Thompson (2022: 350) argues that grounding sentences can be used non-literally to communicate about what metaphysically explains what. On this view, ‘[Trump exists] grounds [{Trump} exists]’ literally expresses the false proposition <[Trump exists] grounds [{Trump} exists]>. However, it can be used non-literally to convey the true proposition <[Trump exists] metaphysically explains [{Trump} exists]>. The latter is true iff (i) the contents of [Trump exists] determine the contents of [{Trump} exists] via one or more specific metaphysical determination relations, and (ii) this suffices for [Trump exists] to be explanatory of [{Trump} exists]. We know that Trump bears set membership to {Trump}, and we know that sets (iteratively conceived) depend for their existence on their members, so (i) and (ii) are plausibly satisfied.

More generally, where we are tempted to say ‘[A] grounds [B]’, there will usually be a specific relation of metaphysical determination relating the contents of [A] and [B] in such a way that is sufficient for [A] to explain [B]. So there is a systematic mapping of literally false grounding sentences to truths about what metaphysically explains what that those sentences can be used non-literally to convey. This goes some way towards answering the central challenge: the grounding fiction is useful because it can be used to convey truths about what metaphysically explains what.

But this benefit must be good enough to be worth using grounding talk over more direct ways of speaking. In other debates where fictionalism has a healthy tradition, it is often claimed that the fiction rewards speakers with expressive benefits. For example, Yablo (2001) argues that number talk can be used non-literally to express physical content that would otherwise be difficult to express.

For example, suppose I’m sharing some crisps with someone, and (due to an overactive sense of justice) I eat one crisp for every crisp my friend eats, and no more. I eventually lose count of how many crisps we have eaten, but I know that neither of us has eaten more than the other. I might try to express this by saying ‘My friend has eaten no crisps and I have eaten no crisps, or my friend has eaten one crisp and I have eaten one crisp, or …’ And so on ad infinitum. This has the benefit of directness: I mention only those things I intended to communicate about. The drawback is that I will never finish what I am trying to say. Number talk to the rescue: ‘My friend and I have eaten the same number of crisps’. With this crisp (ahem) formulation, I have communicated everything I wanted to without having to go on and on about it. In this way, number talk regiments and enhances our communication about physical things throughout ordinary and scientific language, and that seems like a good reason to engage in it, even if it is literally untrue. (See Yablo 2002 for detailed development of these ideas.)

Thompson clearly has this kind of story in mind: ‘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’ (2022: 351); ‘engagement in the grounding fiction imposes a kind of regimentation on discourse about metaphysical explanation’ (2022: 351).

Talking in terms of grounding is indeed simpler than talking in terms of a disjunction of metaphysical dependence relations. This is most obvious in situations where the specific relations involved elude us. For example, one may be convinced that [I am in physical state X] is metaphysically explanatory of [I am in pain], but have no idea how contents of the former determine the contents of the latter. It is tempting to think that we can either painstakingly say ‘My tokening of the type X is identical to my tokening of the type pain, or the kind X is identical to the kind pain, or my being in X is one of many ways of realizing the functional type being in pain, …’ and then go on to say that the truth of any one of these disjuncts is sufficient for [I am in X] to be explanatory of [I am in pain], or just utter ‘[I am in physical state X] grounds [I am in pain]’ non-literally.

If this is our choice, it would indeed be fortuitous to engage in the grounding fiction. But this is a false dilemma. There is another, direct, literal locution that is exactly appropriate for the situation. We can just say ‘[I am in physical state X] metaphysically explains [I am in pain]’. More generally, I offer the following schematic advice:

Instead of ‘[A] grounds [B]’, say ‘[A] metaphysically explains [B]’.

The initial draw of grounding fictionalism stems from the fact that, in cases where ‘[A] grounds [B]’ is fictional, there will be one or more specific metaphysical determination relations connecting the contents of [A] and [B] that is sufficient for [A] to metaphysically explain [B]. But the fiction is superfluous. We can just say ‘[A] metaphysically explains [B]’, remaining neutral about the kind of metaphysical determination at work without having to talk in long disjunctions. Further, we do not need to talk about the system of metaphysical explanation we impose on these relations; we need only apply it.

This last point is important because the other advantage Thompson attributes to the grounding fiction is that it allows us to communicate about determination relations while remaining neutral on whether explanation is a purely objective ontic affair (2022: 351). Directly asserting ‘[A] metaphysically explains [B]’ no more takes a stance on this issue than does indirectly communicating <[A] metaphysically explains [B] > via the grounding fiction. Both ultimately communicate that there is an objective metaphysical determination relation in place, and that this is sufficient for [A] to explain [B]. If explanation is fully objective, then what makes the presence of this relation sufficient for explanation has nothing to do with our perspective on the matter. If explanation is less than fully objective, then our perspective plays a role. But we need not settle this to literally and truly assert ‘[A] metaphysically explains [B]’ (assuming it does), just as we need not settle the matter of whether colour is an objective feature of reality to literally and truly assert ‘the sky is blue’. Besides, if one wishes expunge all possibility of implicating matters less than fully objective, then one can say ‘[A] metaphysically determines [B]’ in place of ‘[A] metaphysically explains [B]’. The grounding fiction has no claim to neutrality that cannot be achieved easily without it.

The very fact that ‘[A] metaphysically explains [B]’ is literal and direct is a prima facie reason to prefer using it in place of less direct ways of communicating, at least when doing metaphysics. Moreover, engaging in the grounding fiction raises difficulties that can be avoided by talking directly and literally. Here are the two I consider most pressing.

First, the fiction introduces unwanted ambiguity. Suppose I use ‘[A] grounds [B]’ non-literally to convey <[A] metaphysically explains [B]>. My interlocutor replies ‘[A] does not ground [B]’. What do they mean? They could be speaking figuratively to convey <[A] is not metaphysically explanatory of [B]>, in which case we disagree. Or they could be speaking literally to deny that [A] and [B] instantiate the grounding relation, in which case we agree. Such ambiguities could be avoided by making explicit the spirit with which one is using grounding talk, but this is at best an unnecessary inconvenience.

Second, when engaging with the grounding fiction, certain non-sequiturs may seem attractive. Suppose that the transitive metaphysical determination relation R and the non-transitive determination relation R* are each always sufficient for metaphysical explanation. Suppose further that metaphysical explanation is not generally transitive. When [A] bears R to [B] and [B] bears R to [C], we can infer that [A] metaphysically explains [C]. When [A] bears R* to [B] and [B] bears R* to [C], we cannot infer that [A] metaphysically explains [C]. Engaging with the grounding fiction means seeing R and R* as instances of the same determination relation, which may invite mistakes. If one is used to dealing with explanations backed by R, where it is fictional that [A] grounds [C] if it is fictional that [A] grounds [B] and [B] grounds [C], one may find analogous but invalid inferences backed by R* tempting. The problem is not that the inferences look alike. It is rather that reasoning well about metaphysical explanation may require sensitivity to the differences between specific metaphysical determination relations, and pretending that they are instances of the same relation may make this more difficult.

Some realists about grounding believe that there are metaphysical explanations not backed by grounding. Grounding is typically thought to be the distinctive manner in which more fundamental portions of reality give rise to less fundamental portions of reality, yet examples such as ‘Socrates is the very individual he is at least in part because he has Sophroniscus as a father’ are arguably metaphysical explanations where their explananda are no less fundamental than their explanans (see Brenner et al., 2021). If this is right, then my proposal will need to be finessed slightly. Instead of ‘[A] grounds [B]’, one would need to say ‘[A] metaphysically explains [B] and [A] is more fundamental than [B]’. This is admittedly more verbose, but verbosity is a price worth paying for a straightforward and literal manner of speaking that avoids the above problems. Besides, if the verbosity proves truly bothersome, one is free to introduce some technical vocabulary to avoid it. One could even recycle the term ‘grounds’ for this purpose.

On the assumption that there is no grounding relation, there is no positive reason to engage with the fiction that there is, and good reasons to avoid doing so. Revolutionary fictionalism about grounding is therefore unmotivated.

3 Against hermeneutic fictionalism

Hermeneutic grounding fictionalism is the view that philosophers already speak non-literally when they utter grounding sentences. We have seen that there are no obvious benefits to doing so, and some potential harms. But people may engage in unhelpful fictions. So the lack of utility of the fiction does not rule out hermeneutic fictionalism, though it does speak against it.

How might one motivate hermeneutic fictionalism about grounding without appealing to the utility of the fiction? Direct linguistic evidence is not an option. There are no instances in the literature of theorists explicitly saying that they are speaking non-literally, and those who explicitly reject the existence of the grounding relation do not endorse the continued use of the term (see e.g. Wilson 2014).

There may be less direct evidence available, however. One is to argue that a non-literal interpretation provides a good explanation of otherwise puzzling features of the discourse. For example, Yablo (2001, 2002) points to features of our ordinary number talk that (according to him) are puzzling if we assume a literal interpretation, such as:

Impatience. People making statements purporting to be about numbers are strangely indifferent to the question of their existence. Suppose that you as a math teacher tell Fred that what 2 and 3 add up to is 5. And suppose some meddler points out that according to the Oracle (which let us assume we all trust), everything is concrete and so not a number. Instead of calling Fred in to confess your mistake, you tell the meddler to bug off. (2001: 195)

Such impatience is supposed to emerge in discourse we all agree is non-literal. For instance, if I claim to have a chip on my shoulder, I would have little time for any ensuing debate about the existence of the chip.

Do philosophers engaging in grounding talk exhibit this kind of impatience when they are challenged by meddling eliminativists? They do not. For example, Michael Raven (2015: 330) takes Wilson’s (2014) aforementioned challenge to grounding seriously and provides reasons to think we should be optimistic about the existence of grounding in spite of it. Examples like this are hard to reconcile with the view that philosophers are doing anything other than speaking literally.

Of course, mathematical realists have taken the claims of mathematical antirealists seriously as well, at least enough to offer replies not charitably paraphrased as ‘bug off!’; but this does not undermine hermeneutic fictionalism about mathematics. The key difference here is the target of the hermeneutic proposal. Hermeneutic mathematical fictionalism is a proposal about what ordinary speakers (i.e. those currently not engaged in the philosophy of mathematics) are doing when they assert sentences that appear to commit them to the existence of abstract mathematical objects. As such, realists who are explicitly arguing for the existence of mathematical objects, or theorising about their nature, are not taken to be speaking figuratively by the proposal.

In contrast, hermeneutic grounding fictionalism targets philosophers who appear to be arguing explicitly for the existence of the grounding relation and theorising about its nature. The fact that these philosophers take meddling eliminativists seriously is therefore damning evidence against a non-literal interpretation.

One might object that grounding talk in philosophy is not limited to grounding theory. For example, some physicalists express their view by saying that all mental facts are grounded in physical facts. Would such philosophers act with impatience if confronted by a meddling eliminativist? As far as I know, there is no textual evidence to decide this matter. But even if philosophers outside grounding theory were to exhibit impatience when challenged by grounding eliminativists, there are other more charitable ways to explain this phenomenon. Perhaps they don’t find eliminativism to be a tenable position, or perhaps they don’t buy into the assumption that grounding is a distinctive ontological posit in the first place (see §1). In light of these options, a non-literal interpretation seems perverse, particularly given the availability of the direct and literal locutions ‘metaphysically explains’ and ‘metaphysically determines’.

One could also object that non-philosophers engage in grounding talk when they use locutions such as ‘in virtue of’, ‘because’, ‘gives rise to’ and so on, and that hermeneutic fictionalism about grounding should instead target these utterances. But the resulting position wouldn’t be particularly interesting or plausible.

It wouldn’t be plausible because, even if these scattered uses of related expressions are systematic enough to warrant a common interpretation, it is unlikely that an adequate interpretation will be one on which ordinary speakers are exploiting a fiction resembling the received view about grounding. If ordinary speakers use locutions like ‘in virtue of’ to convey things about metaphysical explanation, why not take such locutions to literally and directly express things about metaphysical explanation? With this straightforward option available, a fictionalist interpretation seems perverse.

It wouldn’t be interesting because the case for realism about grounding doesn’t rest on the fruitfulness of folk locutions, but rather the fruitfulness of the application of the regimented philosophical notion of grounding to metaphysical theory. Even if hermeneutic fictionalism about the folk notion of grounding were plausible, this would bear little relevance to the debate between realism and eliminativism about the philosophical notion.

Another way to motivate hermeneutic fictionalism is to appeal to a principle of charity. One might think that, when interpreting the speech or behaviour of a population, we should avoid attributing error or irrationality to them. On this view, assuming there is no grounding relation, we should avoid interpreting philosophers as expressing belief in the grounding relation. The kind of charity at stake here involves interpreting someone’s linguistic behaviour in a way that maximises the number of their beliefs that are in agreement with our own (and so true by our account). Applied to grounding talk, such a principle could be metasemantic, concerning what makes it the case that a linguistic interpretation of a population is the correct one, or ethical, concerning what it takes to respect one’s interlocutor as a competent epistemic agent. (Perhaps this distinction merely reflects a difference of emphasis.)

It has already been said that, even if we interpret speakers so that they are right about as many things as possible, that does not guarantee that they will be right about any particular sphere (Daly & Liggins 2010: 212). I will further show that appealing to a principle of charity fails to support hermeneutic fictionalism in the particular sphere of grounding talk.

Understood as an ethical principle, charity is inappropriate when applied to grounding talk. When a metaphysician offers an argument for taking a portion of reality to be a certain way, they do so in perfect awareness that they may err. They offer reasons for their position and ask that we give them due consideration. If we find them unconvincing, and have reasons for adopting an opposing position, we should conclude that our interlocutor is in error and try to explain where they went wrong. To interpret our interlocutor instead so that they agree with us is patronising. It fails to give their reasons due consideration, and fails to countenance the possibility that we may be wrong and they may be right. The ethics of metaphysical debate therefore speak against hermeneutic fictionalism about grounding.

Understood as a metasemantic principle, charity fails to single out hermeneutic fictionalism about grounding talk as the correct interpretation, even if we grant that what metaphysicians often intend to convey when they utter grounding sentences are propositions about what metaphysically explains what. If propositions of the form <[A] grounds [B] > are false but often have true propositions about metaphysical explanation associated with them, then charity seems to favour an interpretation of grounding talk on which ‘[A] grounds [B]’ literally means <[A] metaphysically explains [B]>. This is because it is likely that many grounding theorists believe that they are speaking literally when they utter grounding claims. On hermeneutic fictionalism, these beliefs are false. On the alternative proposal just outlined, they are true. The metasemantics of the grounding debate therefore speak against hermeneutic fictionalism.

We have seen that there is no direct linguistic evidence in favour of a non-literal interpretation of grounding talk, and that more theoretical means of motivating such an interpretation speak against hermeneutic fictionalism rather than in favour of it. I conclude that hermeneutic grounding fictionalism is unmotivated.

4 Conclusions

I have focused on Thompson’s (2022) fictionalist proposal about grounding and shown that it is unmotivated. Perhaps there is another development of grounding fictionalism that will avoid my objections, but I cannot think of one. As things stand, I suggest eliminativists should stand by their abandonment of grounding talk.