1.1. Intelligibility

Chris Daly's arguments for the first kind of eliminativism consist mostly in rebutting realist arguments for the intelligibility of grounding talk. Since friends of grounding generally assume that grounding is a theoretical primitive, the onus is on them to clarify the nature of the relation. Daly argues that each of the strategies employed by friends of grounding to explicate their notion is unsuccessful.

First, Daly argues that the formal properties of grounding don't fix the content of the term ‘grounding’ (because those properties are shared with ‘explanation’).Footnote ¹⁰ Second, he claims that tracing analytic connections between grounding and other notions won't help,Footnote ¹¹ because those other notions are either too close to grounding not to be themselves tainted by its obscurity, or far enough away that their connection to grounding is questionable. Daly's final claim is that appeal to purported examples of grounding to elucidate the notion will fail because anyone who fails to understand ‘grounding’ will consequently fail to understand any examples using that notion.

Though Daly's arguments go some way to towards motivating irrealism about grounding, we ought not to exaggerate their efficacy.Footnote ¹² That the formal properties of grounding don't serve to fix its content is not by itself reason to resist realism about grounding. By taking those properties into account we might intend only to restrict the notion sufficiently to get a fairly good idea of what is at stake, even if doing so does not distinguish grounding from all other notions in the vicinity. Similar responses also limit the scope of Daly's second argument. Although, if successful, the argument robs the grounding-advocate of an attractive way to elucidate grounding talk, the friend of grounding can still endorse Rosen's plea that we ‘relax our antiseptic scruples for a moment and admit the idioms of metaphysical dependence into our official lexicon’, in the understanding that ‘if this only muddies the waters, nothing is lost; we can always retrench’, but that ‘if something is gained…we may find ourselves in a position to make some progress'.Footnote ¹³ In fact, the case Rosen makes for making use of grounding locutions is one that might appeal to an irrealist about grounding (see e.g. section 2.4).

Most parties to the grounding debate agree that the most effective way to argue for the intelligibility of grounding talk is by appealing to purported examples of grounding, but Daly's sceptical response elicits a kind of dialectical stalemate. It is true that the sceptic can always claim not to understand the examples, and such a claim might sometimes be appropriate. The worry is that one can always deny understanding, whether doing so is really appropriate or not (one is reminded of Lewis' quip: ‘any competent philosopher who does not understand something will take care not to understand anything else whereby it might be explained’).Footnote ¹⁴ If the majority of people think they do have a good enough grip on the notion, the fault may be with the eliminativist rather than with the proponent of grounding. If we have a notion that enough people understand enough for it to do useful, recognisable metaphysical work, we at least ought not to dismiss it out of hand. Other irrealist strategies discussed in the sections below allow for grounding talk to do that work without incurring the problematic commitments of full-blown realism about grounding.

Hofweber defines ‘esoteric’ metaphysics as metaphysics that is focused on questions involving distinctly metaphysical terms, and takes idioms of dependence meant in a metaphysical sense to belong to esoteric metaphysics. So far as Hofweber is concerned, grounding talk is unintelligible to the uninitiated. Moreover such talk is redundant because purported instances of grounding are really just examples of logical entailment, or conceptual priority, or mathematical priority.Footnote ¹⁵ Whether or not we buy into Hofweber's characterisation of esoteric metaphysics, one particularly interesting suggestion he makes is that the idioms of dependence he attacks conflate an understanding of priority in the sense in which it is familiar from natural language and from more ‘egalitarian’ metaphysics (i.e. metaphysics where questions are expressed in ordinary, everyday, accessible terms) with a distinctively metaphysical conception of priority. Examples given to elucidate the notion of grounding are of the former understanding of priority, where the notion they are employed to encourage understanding of is of the latter. The possibility of this sort of conflation motivates some of the positions discussed below.

1.2. Level of grain

A related argument for eliminativism about grounding is given by Jessica Wilson, who claims that philosophers almost never make general ‘big-G’ Grounding claims without a more specific relation in mind.Footnote ¹⁶ For example, when naturalists say that the mental is grounded in the physical, they might be a type-identity theorist, or a token-identity theorist, or a functionalist. When people say that the dispositions of a thing are grounded in its categorical features, they again have in mind either a token-identity theory, or a functionalist theory, and so on. Wilson claims that grounding is ‘metaphysically underdetermined’ because further more highly specified accounts of the dependence in question are always available. She argues that it cannot then be the case that Grounding is needed in specific investigations into metaphysical dependence, because we can always work with the more specific account we have in mind.Footnote ¹⁷

The best response to this argument is one Wilson herself considers – that (big-G) Grounding marks an appropriate level of grain for investigations into metaphysical dependence. Grounding is a useful addition to our toolkit alongside the more specific ‘small-g’ grounding relations we already admit because it allows as to make appropriately general claims (e.g. that grounded entities cannot come apart modally from their grounding entities).Footnote ¹⁸ We might add that this is cause to reject Wilson's characterisation of things – it is not the case that philosophers always have a more specific relation in mind when they make grounding claims, because sometimes those claims are claims about big-G grounding.

Wilson's reply to this response is that it motivates adopting grounding as a merely pragmatic, and not as a metaphysical notion. This she takes to rob grounding of any interesting metaphysical substance, and thus to make it into a very different notion to that which friends of grounding are keen to discuss. The irrealist about grounding can hold that Wilson's arguments provide excellent motivation for irrealism about grounding, but not for the eliminativism we have been discussing. The idea that grounding talk might have some pragmatic benefit independently of the metaphysical status of grounding relations is itself a strong argument for non-eliminative irrealism. The point of departure between Wilson and the non-eliminative irrealist concerns how interesting an irrealist account of grounding might be.

2.1. Explanation

Grounding claims are thought to be explanatory. In the minds of at least some prominent friends of grounding the connection between ground and explanation is one of identity; grounding is a relation of metaphysical explanation.Footnote ¹⁹ According to Fine, this means that in addition to familiar causal explanation there is a distinctive kind of metaphysical explanation in which explanans and explanandum are connected through a constitutive determination relation – grounding.Footnote ²⁰

On the realist picture, there seems to be a tension between the metaphysical and the explanatory aspects of ground.Footnote ²¹ On the one hand, ground is an objective, mind-independent, worldly relation which describes reality's fundamental structure. On the other, it is a relation of explanation, and explanations are sensitive to explanatory interests and the background beliefs and commitments of enquirers. Friends of ground have work to do in order to explain how this apparent tension is to be reconciled.Footnote ²²

Should the friend of grounding move towards taking metaphysical explanation to be somehow more robust or ‘objective’ than more familiar forms of explanation, she risks losing the benefits that are supposed to come from thinking about grounding claims as explanatory. The most important of these is that thinking of grounding as a form of explanation helps to shed light on an otherwise opaque, primitive notion, open to sceptical attacks on its intelligibility. Our understanding of and intuitions about explanation can only help elucidate grounding if the relevant sort of explanation is one we understand and have intuitions about. If metaphysical explanation is a distinct form of explanation, the friend of grounding must either demonstrate that we are already familiar with it or provide us with a stand-alone account.

Should this prove too difficult a challenge, the friend of grounding might instead widen the gap between explanation and ground. Perhaps the connection between grounding and explanation can be preserved in a weaker form by taking metaphysical explanations to ‘track’ grounding relations (in much the same way as causal explanations might be said to track causal relations). One worry about such a picture concerns the mechanism for this tracking; how is it that metaphysical explanations are able to ‘latch on’ to worldly grounding relations? Proponents of grounding talk might be tempted to describe such tracking relations by appeal to grounding; grounding relations ground metaphysical explanations.Footnote ²³ But such an account would be viciously circular. We can't expect to shed light on the connection between ground and explanation by appeal to grounding.

Non-eliminative irrealism about grounding allows for reconciliation of the apparent tension between the metaphysical and the explanatory aspects of grounding. Talk of grounding is talk of metaphysical explanation; metaphysical in the sense that the relata of the grounding relation are worldly facts, and explanatory in the sense that when such a relation obtains, we come to expect or to understand the explanandum on the basis of the explanans. The threat of elimination on the basis of unintelligibility diminishes because grounding is an explanatory relation, and familiarity with the notion of explanation can help elucidate that of grounding. But this isn't a realist view of grounding, because grounding relations are not out there in the world for us to discover.

2.2. Epistemology of grounding claims

This section raises concerns about the epistemology of grounding claims; that grounding relations conceived of in a realist spirit are not the sorts of things we can reliably come to know about.

One species of worry is that we are not in possession of adequate resources for forming reliable beliefs about grounding, and so knowledge of grounding claims ought to be considered impossible. Note, however, that grounding facts are generally assumed to be metaphysically necessary,Footnote ²⁴ and so care must be taken to present such epistemic worries in a way that doesn't rely on our being able to evaluate counterfactuals which the friend of grounding will take to be metaphysically impossible (i.e. counterfactuals of the form ‘if A didn't ground B, then…’). For example, a sensitivity constraint on knowledge of grounding claims (for an agent S to know some grounding claim G, it must be the case that had G been false, S would not have known G) does not provide a legitimate basis for an argument that we cannot have knowledge of grounding claims. The friend of grounding can simply deny that G could have been false, and the argument cannot get off the ground.

Like most debates in metaphysics, discussions about what grounds what are insensitive to empirical investigation. Instead, judgements about grounding are generally made by appeal to intuitions about cases. The debate about the kind of justification that can be afforded by intuitions rages on, and this is not the place to get into it. It seems fair to assume though that the irrealist about grounding is at least as justified as the realist in taking intuitions about grounding to provide support for the truth of grounding claims. The difference is that it is fairly easy to see why we might take intuitions about grounding to justify grounding-talk if the truth of grounding claims depends, somehow or other, on our mental lives.Footnote ²⁵ The realist about grounding must demonstrate that intuitions about grounding are somehow capable of providing evidence for the truth of claims about an objective, mind-independent grounding relation. This seems at least to be a harder task.

What we can reliably expect to learn from reflecting on our intuitions about purported examples of local grounding relations (such as the relation between Socrates and his singleton set) is how the entities concerned are related within our conceptual scheme, and we do not have good reasons to think that our conceptual scheme (which is partly dependent on our theoretical commitments) provides a perfect reflection of reality. As David Wallace quips, ‘our intuitions…were designed to aid our ancestors on the savannahs of Africa, and the universe is not obliged to conform to them’.Footnote ²⁶ It is certainly conceivable that the structure of the world could have been the same, and our beliefs about it have been very different.

2.3. Metaphysical queerness

A different form of scepticism about realism about grounding bears some similarity to Mackie's argument from queerness.Footnote ²⁷ Mackie thought that if moral properties existed, they would be both metaphysically and epistemically queer; metaphysically queer because of their unusual motivational force, and epistemically queer because of the perceptual faculty we would seem to require in order to track these strange properties. There are at least two ways in which grounding seems metaphysically queer.

First, primitive grounding relations are ‘spooky’ in much the same way as primitive causal relations are often considered spooky. For those who take grounds to necessitate what they ground, the analogy is particularly strong; grounding relations are necessary connections in nature, and to the extent that we are suspicious of such connections, we should be suspicious of the grounding relation so understood. But even those who don't find this species of spookiness disquieting might find grounding relations metaphysically ‘queer’.

In section 2.1 above we reviewed the options concerning the connection between ground and metaphysical explanation. Grounding relations are such that they either are themselves relations of metaphysical explanation, or they back metaphysical explanations. Suppose first that ground is a relation of metaphysical explanation. Not only does this objective, worldly explanation itself seem metaphysically queer but it must have an unusual motivational force not unlike that of Mackie's moral properties.

First, the idea of objective explanation is in itself somewhat jarring. The idea that reality comes furnished with an explanatory structure conflicts with our understanding of explanation as an epistemic phenomenon. Explanations (unlike information that might figure in an explanation) aren't ‘out there’ in the world for us to discover. Explanations are constructed to improve the epistemic position of an agent given her explanatory interests, background beliefs, and theoretical commitments. It is an assumption in the literature on explanation that explanation is intimately connected to understanding, such that in being provided with an appropriate explanation one comes to understand or to expect the explanandum on the basis of the explanans. As Kim remarks, ‘the idea of explaining something is inseparable from the idea of making it intelligible; to seek an explanation of something is to seek to understand it’.Footnote ²⁸ This Kim takes to be ‘untendentious and uncontroversial’, and yet we must reject it if we are to maintain that explanations obtain in the absence of explanation-seekers.

Even supposing we can reconcile our understanding of explanation with the requisite objectivity of metaphysical explanation, some mystery remains. When presented with a good explanation, we come to understand the explanandum on the basis of the explanans and thus are motivated to accept the explanation. In the normal case, we can account for this motivation by pointing out that part of what constitutes a good explanation is that it increases our understanding. Metaphysical explanations are supposed to be exemplary, qua explanations. Thus, Fine says ‘if there is a gap between the grounds and what is grounded, then it is not an explanatory gap’.Footnote ²⁹

But because what makes metaphysical explanations good explanations must (if they are to remain objective) be divorced from their effect on our understanding, it is perfectly conceivable that we might be presented with a good metaphysical explanation which we are not motivated to accept. Caveats usually introduced to ensure that explanations are proportionate, informative, and not overly complex can't get any traction. This is the sense in which the friend of grounding must (when ground is identified with metaphysical explanation) countenance a relation with an unusual motivational force. An agent may find herself in the unusual position of believing an explanation without understanding it.

We already noted that there is another position available to the friend of grounding; perhaps metaphysical explanations merely track grounding relations. The above might be considered reason to adopt a tracking view of the connection between grounding and explanation, maintaining that grounding relations are objective and mind-independent, but the explanations that track them need not be. But recall the problems introduced in section 2.2; the weaker the connection between grounding and explanation, the harder it is to explain how it is that we can come to know about grounding relations.

Here we should be mindful of a disanalogy between the case of grounding and that of causation. In the causal case we generally distinguish between the network of causal relations, and explanatory information about that network. The former is objective and interest-insensitive, and the latter is not. But it is not the case that our only knowledge of causation is based on our understanding of and intuitions about causal explanation. The availability of independent accounts of causation allows us to distinguish between causation and casual explanation, and there is no analogue of these independent accounts in the case of grounding.Footnote ³⁰

Here's a different way to put the argument of this section. Knowledge of grounding requires a hyperintensional epistemology; a way of knowing that is sensitive to different epistemic intensions. The kind of knowledge we get from explanations gives us precisely this kind of sensitivity, and so it is attractive to think of grounding as a form of explanation (rather than merely as a relation that backs explanations). But if the relevant form of explanation is the objective, Finean kind, grounding starts to seem metaphysically queer. This problem can be resolved by rejecting the existence dimension of realism; by denying that there is any relation of grounding to be the bearer of this metaphysical queerness.

2.4. Against eliminativism

Arguments based on the epistemology of grounding claims provide reasons to be sceptical about the existence of any grounding relation whatsoever. One might be tempted then to think of such arguments as motivating eliminitivism rather than some form of non-eliminative irrealism. The non-eliminative irrealist must show both that an objective, mind-independent grounding relation would be metaphysically dubious, and that eliminativism is not a credible option. Plausible versions of irrealism about grounding must therefore demonstrate that grounding talk has an important role to play in philosophical theorising. There have been a number of attempts by realists about grounding to make such arguments by way of motivating realism about grounding. Those arguments can also be used to motivate irrealism, so long as the irrealist can tell a plausible story about how those roles can be fulfilled within the framework of her irrealist approach. Here is one example.

Rosen offers a clarification of notions such as grounding and metaphysical dependence and a plea that we might ‘relax our antiseptic scruples for a moment and admit the idioms…into our official lexicon’. This is to be done in an experimental spirit, under the understanding that ‘if this only muddies the waters, nothing is lost; we can always retrench. If something is gained, however…we may find ourselves in a position to make some progress'.Footnote ³¹ Rosen proceeds to offer examples of grounding, to formulate the logical and structural properties of the notion, and to undertake an extensive survey of metaphysical principles that might be framed in terms of grounding, and to demonstrate how those principles might interact with other accepted principles. His discussion covers interactions with (amongst others) logic, universal facts, modal truths, reduction, and the determinable-determinate connection.

The purpose of Rosen's project is to demonstrate that framing metaphysical principles in terms of grounding doesn't lead to confusion or incoherence. If such notions can be put to use in making sense of the puzzling domain of metaphysics, then ‘the strategy of acquiescing in these ways of speaking will be vindicated’.Footnote ³² Grounding is established as a legitimate resource for metaphysics. Rosen's target is the grounding eliminativist who claims that we do not understand the relevant notions. But demonstrating that grounding talk has instrumental value in metaphysics does not serve to vindicate realism. Non-eliminative irrealists can appeal to Rosen-style arguments to justify their continued engagement in the grounding discourse, but doing so is consistent with adopting various forms of irrealism. To see this, we need to know more about some plausible irrealist proposals. Outlining some such proposals is the project for the remainder of the paper.

3.1. Metalinguistic fictionalism

A simple form of metalinguistic fictionalism employs operators to give an account of the real content of a speaker's grounding claim. Following Lewis,Footnote ³⁵ it is standard to understand utterances of a sentence like ‘Smaug is a dragon’ as elliptical for ‘according to The Hobbit, Smaug is a dragon’. One version of metalinguistic fictionalism about grounding similarly appeals to fiction operators, but in an account of grounding-talk. When a speaker utters a grounding claim S, this metalinguistic fictionalist takes the real content of her claim to be <according to the grounding fiction, S> . Other kinds of metalinguistic views maintain that the real content of a quasi-assertion about grounding concerns some other non truth-involving property of the fiction. For example, that the fictional content of the relevant claim is pragmatically advantageous in simplifying and systematising metaphysical theories and disputes.Footnote ³⁶

3.2. Objectual fictionalism

Fictionalists need not accept that the real content of a target sentence is about the content of a fiction. Instead, they might maintain that the real content of a sentence S about grounding is the real-world conditions that make it fictionally true that S. The champion of this approach to fiction is Kendall Walton, who takes fictions (in all their forms) to be games of make-believe.Footnote ³⁷ Imagine a group of children playing a game of Cops and Robbers. If one of the children playing a robber starts to run away, and a child playing a cop shouts ‘Quick, a robber is getting away!’ then she asserts something that is true in the pretence (although it is, of course, literally false that a thief is running away from the children). What makes the cop's assertion appropriate is the real-world event of a child, designated ‘robber’ starting to run away. Real-world conditions generate fictional truths.

So long as a speaker is engaged in a pretence, a quasi-assertion of S does not commit her to the truth of its fictional content.Footnote ³⁸ When a speaker makes a grounding claim such as ‘singleton sets are grounded in their sole members' she quasi-asserts that certain real-world conditions obtain. We can give various accounts of what real-world conditions a speaker quasi-asserts obtain when she utters such a sentence. The view I think most plausible is that the speaker conveys that the world is in a condition such that she is apt to find the member of a singleton set metaphysically explanatory with respect to that set.Footnote ³⁹

The fictionalist can tell a further story about the origin of the pretence. Some such stories make plausible hermeneutic fictionalism about grounding. Causal explanations are backed by causal relations, and so we assume that metaphysical explanations are also backed by some metaphysical dependence relation. Metaphysical explanations are backed by (small-g) dependence relations, but not by any generic notion of dependence. Some are backed by set-theoretic relations, others by composition relations, identity relations, determinate-determinable relations, and so on. The assumption that there must be some generic relation of metaphysical dependence led to the development of the grounding fiction.Footnote ⁴⁰

3.3. Non-assertion fictionalism

A final option for the fictionalist is to deny that any proposition is associated with a quasi-assertion of a target sentence. The sentence thus has no real content at all, but is to be used merely as a device for simplifying or systematising the relevant discourse.Footnote ⁴¹ The non-assertion fictionalist about grounding could argue that grounding talk plays a useful role in metaphysics, but refrain from commenting on what (if anything) sentences in the domain could be used to assert.

The difficult task for our non-assertion fictionalist is that of justifying our continued engagement in grounding talk. For the metalinguistic fictionalist and the objectual fictionalist there is some kind of link between the propositions expressed by the target sentences in the domain, and the quasi-asserted real content associated with utterances of the target sentences. Where there is no such real content and merely a false proposition expressed by utterances of the target sentences, the fictionalist has a harder task justifying the continued use of the relevant sentences. Field justifies our continued engagement in mathematical discourse by claiming both that mathematical theories are conservative over nominalistic ones (that nothing that can be proven using mathematics cannot be proven without it), and by making a strong case for the instrumental benefits of continued engagement in mathematical discourse.Footnote ⁴²

Like the non-assertion fictionalist about mathematics, this kind of fictionalist about grounding can point to various benefits of continued engagement in the grounding discourse. Alongside the aforementioned role grounding might play in simplifying and systematising debates in metaphysics, reference to ‘big-G’ grounding is beneficial because it ranges schematically and neutrally over more specific ‘small-g’ grounding relations (composition, set membership, type identity, functional realization, etc.).Footnote ⁴³ It is often beneficial to talk in terms of features common to all of these relations, perhaps because we want to convey some sort of significant dependence (its nature and its direction) without getting clear on the details, or because it's not yet clear to us which of these small-g relations obtains (though it is obvious that at least one of them does) or because our metaphysical theorising is guided by a distinctive epistemic feature of these small-g relations, such as a direction of explanatory dependence, or an understanding that a grounded entity is ‘nothing over and above’ the entity that grounds it.Footnote ⁴⁴

One might object here that if we shouldn't be realists about grounding, continued engagement in grounding-talk just serves to make metaphysics more murky. The goal of metaphysics is to get to ultimate categories and explanations, and irrealists deny that grounding is among these. But this view of metaphysics is overly concerned with which entities and notions belong in a description of fundamental reality. Notions of dependence which we might use to frame metaphysical principles might have no place themselves among the fundamental, but they are nevertheless useful for the practice of metaphysics (compare how we might use inaccurate models to teach scientific concepts).

I'll mention one further pragmatic advantage of appeal to grounding-talk. Fine appeals to grounding-talk in order to mark a distinction between realist and irrealist about a given domain of discourse.Footnote ⁴⁵ The motivation for antirealists to accommodate the way in which language is used by ordinary speakers, combined with the rise in popularity of minimalist theories of facts and truth (such that all there is to truth is something like collected instances of the schema ⌜S⌝ is true iff S) has meant that realists and antirealists alike are willing to utter the same sentences. Fine suggests that we can distinguish proponents of each position by asking what grounds a relevant proposition (e.g. <murder is wrong>). Fine argues that whilst the realist's answer will involve reference to moral properties, the irrealist takes the proposition to be grounded in something like speaker-attitudes towards child-torture.Footnote ⁴⁶ The key move is that in asking a grounding question, we can adopt a ‘metaphysically neutral’ stance concerning the reality of the proposition in question – we can consider grounding questions whether we are realists or irrealists about the relevant discourse, and the language used to frame the question is neutral on the issue of realism.

This benefit of engagement in grounding talk does not require realism about grounding. Grounding talk merely brings out a distinction that is already present in the commitments of the realist and of the irrealist, but it is a distinction that is hard to get at in other terms (here the analogy with Field's project is fairly close). The collected benefits of engaging in grounding discourse provide justification for our continuing to talk in terms of grounding in spite of the systematic falsity of propositions about grounding.

4.1. Prescriptivism

Prescriptivists about grounding emphasise the familiar claim made by grounding theorists that grounding locutions are explanatory locutions, and that the relevant explanatory connection (between explanans and explanandum) is very tight. The prescriptivist about grounding takes statements of (full) ground to be prescriptions to understand or to cease explanatory enquiry. For example, when we say then that the fact that P grounds the fact that P∨Q, we prescribe the end of enquiry concerning P∨Q; we dictate that there is no further explanatory work to be done in accounting for P∨Q, once we have understood that P.

Support for this view might be extracted from the work of philosophers such as Kit Fine, who draw attention to the explanatory character of ground. Fine says that it is ‘properly implied by the statement of (metaphysical) ground that there is no stricter or fuller account of that in virtue of which the explanandum holds…if there is a gap between the grounds and what is grounded, then it is not an explanatory gap’;Footnote ⁴⁷ and that there is ‘no explanatory connection that stands to ground as grounding stands to…other forms of explanation…it is the ultimate form of explanation’.Footnote ⁴⁸ It is the view of such friends of grounding that ground provides the most illuminating explanation; the explanation which, when we are in possession of it, dictates that we have no need for further explanatory inquiry. It is a small step, the prescriptivist claims, from the view that grounding is a relation of metaphysical explanation to the idea that all there is to a statement of ground is a prescription that we end explanatory inquiry. In the face of concerns about the legitimacy of any notion of ground that goes beyond this claim about explanation, prescriptivism might look like an attractive alternative to realism about grounding.

Prescriptivism about grounding of the form described here requires that we think of the relevant sort of explanation as something objective enough that it will be the same in relevantly similar contexts – that similarly situated agents would make the same judgements of ground (it is this that guarantees that the prescription be universal). It is this fact that is responsible for grounding talk being subject to various constraints, including restrictions on the logical and structural features of ground. Friends of grounding might welcome this apparent legitimisation of the somewhat obscure notion of metaphysical explanation, and the independent role that the grounding prescriptivist takes metaphysical explanation to play. Nevertheless, one might worry that some reasons for suspicion about the notion of ground (particularly those based on concerns about the epistemology of grounding claims) will carry over to any notion of explanation we can think of as objective enough to play the relevant role. Those persuaded by such arguments are likely not to find this sort of prescriptivism about grounding attractive, and so might think that a different form of non-cognitivism is more plausible.

4.2. Non-cognitive expressivism

Non-cognitive expressivism can be characterised as the conjunction of two theses, one negative and the other positive. The negative thesis states that the grounding vocabulary is not ‘descriptive, not belief-expressing, not fact-stating, not truth-evaluable, or not cognitive’.Footnote ⁴⁹ The positive thesis says that the vocabulary expresses a non-cognitive attitude. The task for an expressivist about grounding is to give an account of the non-cognitive attitude expressed when competent speakers utter sentences involving grounding locutions. There are various accounts the grounding expressivist might choose to give of the relevant non-cognitive attitude. I'll mention one promising proposal.

The non-cognitive expressivist is can take grounding claims to express attitudes of acceptance towards particular systems of explanation; to say that x grounds y is to endorse a particular system of explanation in accordance with which x explains y, and thus to ‘plan’ to take relevantly similar explanations to be explanatory. The expressivist position here is subtle. The realist about grounding generally takes it to be the case that when x grounds y, x explains y, but the realist takes claims of the form ‘x grounds y’ to be truth-apt, and to be made true by mind-independent features of reality. The expressivist denies both of these realist commitments. The claim ‘x grounds y’ expresses an attitude (rather than a proposition) and its appropriateness depends on the attitudes of the speaker (i.e. on their endorsement of a system of explanation in accordance with which x explains y).

This form of expressivism shares some features with the plan-expressivism associated with Gibbard,Footnote ⁵⁰ and also with the prescriptivism about grounding described above. To judge that x grounds y is to judge that it is apt to find x explanatory with respect to y. Unlike with the prescriptivism introduced above, the expressivist need not think of the relevant sense of explanation as a particularly objective one (though there is room for positions on which the relevant sense of explanation is an objective one). This form of expressivism takes seriously the explanatory nature of grounding claims, but does so without stating any facts about what explains what. The position is motivated by the thought that agents are apt to find relevantly similar systems of explanation explanatory, and that interpreting their grounding claims as expressing this attitude makes sense of the apparent connection between grounding and explanation. Making a grounding claim functions as an invitation to others to endorse a similar system of explanation, and so the expressivist can account for disagreement about grounding: disagreements represent different attitudes towards a given system of explanation.