The Atlas of Reality
Page 12
Is conceptual grounding the same thing as conceptual analysis? If the truth of proposition p is grounded in the truth of q, does this mean that q is the correct analysis of the meaning of p? We don't think so, since we've said nothing to suggest that conceptual grounding is always discoverable a priori, simply be reflecting on the meanings of our concepts “from the inside,” so to speak. It could turn out that the facts about what grounds the application of a concept might be discoverable a posteriori, that is, by some scientific investigation of the human mind and its environment, while conceptual analysis is supposed to be something that can be carried out entirely in the philosophical armchair. Where conceptual analysis is possible, we would have excellent grounds for a claim about conceptual grounding, but there might be cases of conceptual grounding without conceptual analysis. Conceptual grounding could be fleshed out in terms of truthmaking rather than conceptual analysis, as in the Grounding to Truthmaker Link discussed above (Section 3.2.1).
3.4.1 The ontological import of grounding
What is the ontological import of grounding? Does grounding show that the fundatum doesn't “really” exist? Let's call the view that the fundatum doesn't really exist the ‘eliminativist’ option.
The eliminativist option certainly breaks down in the case of extra-conceptual grounding, where the grounding depends in part on the essence of the fundatum. If the fundatum doesn't really exist, it can't have an essence. This is true even if the fundans contains a complete explanation of that nature. If x is grounded in y in part because of x's essence, then it seems that x's derivative status does not count against its really existing.
Thus, once again it is crucial that we distinguish merely conceptual from extra-conceptual grounding:
That p is grounded in the fact that q because of the essence of some worldly component of the facts corresponding to p and q.
That p is true is grounded in the fact that q is true because of the essence of some conceptual constituents of the propositions p and q.
Case (ii), conceptual grounding, would seem to have very strong deflationary consequences. We can still believe in the conceptual constituents of p, but we would no longer have to believe in any of the apparent worldly commitments of p at all (e.g., things that terms in p seem to refer to, properties that predicates of p seem to express). However, case (i), extra-conceptual grounding, would seem to have much weaker deflationary implications. For example, if we believe in the conceptual grounding of the truth of ethical propositions in the truth of non-ethical or natural propositions, then we could simply deny the real existence of ethical properties and facts. In contrast, if we believe that ethical facts are non-conceptually grounding in the natural facts, in part because of the very essence of ethical properties, then we are fully committed to the real existence of ethical facts and ethical properties. Nonetheless, there would still be some gain, since we will have reduced the number of fundamental or ungrounded properties.
3.4.2 Support for the One Truthmaker per Fundamental Property Principle
In Chapter 2, we argued that defenders of truthmaker theory should accept the principle that two truths that involve different fundamental properties must have different truthmakers (PTruth 1). We are now in a position to define fundamental property:
Def D3.9 Fundamental Property. F-ness is fundamental if and only if, necessarily and for all x, if x is F, then that x is F is ungrounded (or zero-grounded).
Attributions of fundamental properties are always ungrounded. If we combine this definition with a grounding-to-truthmaking link, then we can provide a criterion for individuating fundamental properties, that is, for saying when two fundamental properties are identical or distinct.
If two propositions have of necessity the same truthmaker, then either the truth of one is grounded in the other or the truth of both is grounded in the truth of some third proposition. So, if x is F and x is G are both necessarily made true (when either is true) by the same truthmaker, then at least one of the two propositions must have its truth grounded in the truth of some other proposition. Consequently, it is impossible, given these assumptions, for F-ness and G-ness to be distinct fundamental properties.
3.7T Distinct Fundamental Properties. F-ness and G-ness are distinct fundamental properties if and only if it is necessarily the case that, for any x, a truthmaker for the proposition x is F is distinct from any truthmaker for the proposition x is G.
From a slightly more general version of this thesis, we can derive One Truthmaker Per Fundamental Property (PTruth 1):
PTruth 1 One Truthmaker per Fundamental Property. If p is the true predication of a fundamental property P to x1 through xn, and q is the true predication of a different fundamental property Q to the same things x1 through xn, then p and q have distinct truthmakers.
3.5 Alternatives to Grounding?
We've shown that grounding is a useful notion, distinct from reduction, supervenience, truthmaking, and conceptual analysis. In this section, we will consider three competitors to grounding theory. These competitors attempt to do the same work without treating grounds as an undefined primitive. The first follows the method of philosophical paraphrase defended by Willard van Orman Quine. The second relies on Kit Fine's notion of reality. The third seeks to take fundamentality as primitive, defining grounding in terms of fundamentality, instead of the other way around.
3.5.1 Quine's method of paraphrase
As Quine (1953/1980) pointed out, even if we have no notion of truthmaking or grounding or fundamentality, we can still ask the basic ontological question, What is there? As Quine also noted, there is the obvious and simple answer, Everything. However, we can ask what exactly this “everything” includes. Does it include minds, organisms, artifacts, or composite entities of any kind? Events or processes? Numbers, sets, or propositions? Quineans deny that there is any ontological order to be placed on what exists. If a supposed “something” is really distinct from but dependent on one or more other objects, then “it” doesn't exist at all.
A complication enters when we consider Quineans who wish to deny the existence of things whose existence seems to be taken for granted in much of our everyday speech. Peter van Inwagen (1990a), for example, denies that there are any artifacts, including tables, forks, and automobiles. But even van Inwagen, in his ordinary life, will speak in a way that seems to imply that such things exist, like ‘I own an automobile’ or ‘Please pass a fork’. Grounding theorists can make sense of this by supposing that the truth of the propositions expressed by these ordinary sentences are grounded in facts that don't include anything corresponding to the apparent reference to artifacts—that is, they are grounded in or made true by facts that don't include any actual automobiles or forks. Quineans, in contrast, must deny that these statements are, strictly speaking, true at all. They must suppose that, when we seem to refer to such things, we are speaking in a “loose and popular way,” with an insouciant disregard for the actual truth of the matter. Within the ontological seminar room, in contrast, we must be careful to speak only the strict and literal truth.
One difficulty with this paraphrase theory is that it seems to ascribe very sophisticated communicative intentions to ordinary people who have never considered sophisticated philosophical or scientific arguments purporting to show that many ordinary objects don't really exist. It is probably better for Quineans to adopt a stance of agnosticism toward the psychological question of what ordinary people really believe or really intend to convey by sentences of natural language. Perhaps ordinary people really do believe the propositions that some ontologists reject. It may be, as Theodore Sider (1999) argues, that people are relatively indifferent to the question of the real, strict truth of what they and others say. People are content if what is said is quasi-true, the sort of thing that would be true if the general ontological framework embedded in common sense really corresponded to the way the world is. In everyday contexts, although we speak with the intention of being understood strictly and literally, we are in
fact content if we achieve the expression of things that are quasi-true, close enough to the truth for practical purposes. If so, we might be tolerant of van Inwagen's speaking, in ordinary contexts, as if artifacts existed, even though he believes that they don't, so long as van Inwagen believes that what he says would be true, if (counterfactually) artifacts did exist.
In many cases, when we are considering what truths and entities are fundamental, it won't matter whether we understand that issue in terms of grounding theory or in terms of a Quinean theory. As we shall see, grounding theorists believe that we should be economical with respect to the fundamental entities of our ontology, and Quineans believe that we should be economical with respect to the entities we accept as existing in the context of doing serious ontology. In both cases, metaphysicians will inquire about what theoretical and explanatory work can be done by the entities in question.
3.5.2 Fine's really operator
Fine (2001) argues that philosophers should recognize the usefulness of a really operator, a linguistic device that would enable them to say that certain ordinary objects (like forks or automobiles) exist, even though they do not really exist. The adverb ‘really’ signals that we intend to speak of entities that make up what Fine calls “the intrinsic structure of reality.” Other things exist, in a sense, but only in a qualified and attenuated sense.
Fine did not offer the really operator as an alternative or competitor to the notion of grounding. He thought that we needed both notions as primitive terms of metaphysical theory. Nonetheless, we can still ask: would really make grounding redundant? Can we define grounding in terms of the really operator? Here is a possible definition:
Grounding-Really Link (Fine). The fact that p grounds the fact that q if and only if the truth of p entails the truth of q, Really(p), and not Really(q).
The idea is that only real facts can ground anything, and only unreal facts can be grounded. A real fact grounds an unreal fact if and only if it metaphysically entails that unreal fact. Fine notes a number of difficulties with this proposal. First, it does not respect the idea that grounds should be as minimal as possible. For example, suppose that the fact that p is grounded in the fact that (q & r), and that p is unreal, while q and r are real, with r a real truth that is completely irrelevant to the truth of p. The Grounding-Really Link wrongly entails that p is grounded in the conjunction (q & r). This problem could be solved by suitable repairs to the Grounding-Really Link. We could require, for example, that p is one of the logically weakest propositions satisfying the Link. This would rule out conjunctions with irrelevant conjuncts.
There is, however, a more fundamental problem with the Grounding-Really Link: it entails that no grounded fact can be a part of reality. Fine argues that we shouldn't assume this:
We may grant that some things are more explanatorily basic than others. But why should that make them more real?… We cannot read off what is nonreal from what is nonbasic. Indeed, it is possible to imagine metaphysical scenarios in which the nonbasic, or grounded, is plausibly taken to be real. Suppose [for illustration] compound events enter into both basic and grounded causal relations (suppose their being caused is always grounded but their causing some effect is sometimes basic). Then we should count the grounded causal facts as real. Source: (Fine 2001: 25, 27)
Fine suggest that there is a general presumption in favor of the grounded not being real. However, this presumption need not hold in every case, as the Grounding-Really Link assumes.
3.5.3 Primitive fundamentality, naturalness, and primitive structure
Jessica Wilson (2014) argues that we can replace grounding with a primitive notion of fundamentality. Once we have fundamentality in place, we can define ‘x grounds y’ in terms of x's being fundamental, y's being non-fundamental, and x and y's standing together in one of a small number of possible relations, like part/whole, member/set, determinate/determinable, realizer/realized. Call these latter relations the ‘small-g grounding relations’. Wilson proposes the following definitional link:
Fundamentality-Grounding Link (Wilson). The fact that p grounds the fact that q if and only if p is fundamental, q is non-fundamental, and p and q (or the entities involved in p and q) stand in some small-g grounding relation.
All of Wilson's small-g grounding relations seem to be relations between worldly entities and not explanatory relations among facts or true propositions, except perhaps for the realization relation (one fact might realize another). It is not obvious that every explanatory relation involving essences can be fit into one of the categories that Wilson provides.
Wilson's proposal also raises the question of why just these relations are cases of grounding. What is it that they have in common? The most plausible answer to this question is that, at least in certain cases, these small-g grounding relations are instances of grounding (simpliciter).
In addition, it is doubtful that x grounds y can really be defined as (i) x is fundamental, (ii) y is not fundamental, and (iii) x and y stand in one of a list of relations. For example, we can imagine cases in which x is fundamental, y is non-fundamental, x is a part of y, and yet y is not even partly grounded in x. Imagine, for example, a theory in which non-fundamental entities are grounded in the fundamental wholes which contain them, and not in their fundamental parts. In a view like van Inwagen's (1990a), the only fundamental things are simple particles and living organisms. Biological facts concerning an organic, functional part of an organism, like an eye, are grounded in the structure of the whole organism of which it is a part, and not in facts about the subatomic parts of which the eye is composed. Wilson's linkage is unable to make the relevant distinction between those facts that are grounded in facts about a non-fundamental thing's parts, and those that are grounded in facts about the fundamental whole that includes it. In addition, there is the worry that we might discover new ways for facts to be grounded in other facts, ways that do not correspond to any of the relations on Wilson's official list of small-g grounding relations.
Finally, as Litland has suggested,2 we can have grounding without a difference in fundamentality. Suppose the world is homogeneous and infinitely divisible. It could still be the case that wholes are wholly grounded in their parts, even though there is no fundamental level (or all levels are equally fundamental).
A number of other philosophers have developed accounts of a kind of primitive or indefinable fundamentality. For example, David K. Lewis introduced the idea of perfectly natural properties (Lewis 1986a: 60–69). A perfectly natural property is one that corresponds with a maximal degree of intrinsic similarity. In addition, a property has to be perfectly natural in order to figure in our most fundamental laws of nature. A natural property is to be contrasted with an unnatural property, one that has been constructed in a highly artificial, gerrymandered way. For example, being spherical or having negative charge are fairly natural properties, while being two-feet from an avocado or being green or oblong or well-liked are quite unnatural. Lewis's notion could be extended to define notions of fundamental entity and fundamental fact. A fundamental entity might be defined as something whose essence or defining characteristic is a natural property, and a fundamental fact could be one that consists in attributing a perfectly natural property to one or more fundamental things.
Lewis's theory could be seen as a version of grounding theory, rather than as a competitor to it, especially since Lewis suggested that the naturalness is degreed, that some properties are more natural than others. We could combine Lewis's notion of degree of naturalness with Wilson's list of small-g grounding relations, and define grounding as follows:
Grounding-Naturalness Link (Lewis). The fact that p grounds the fact that q if and only if p is more natural than q, and p and q (or the entities involved in p and q) stand in some small-g grounding relation.
This will have some of the defects of Wilson's Grounding-Fundamentality Link, but it will be able to handle cases in which there is no fundamental level (no perfectly natural properties), and it can
make sense of grounding relations between non-fundamental facts and entities.
Sider (2011) recently defended another notion of primitive fundamentality, which he labels ‘structural’. For Sider, the term ‘structural’ can be applied to entities of many different categories. There are structural properties, entities, and propositions. Even logical elements, like ‘and’, ‘some’, ‘not’, or ‘all’ can count as structural. In fact, Sider allows ‘structure’ to be applied as a linguistic operator to any meaningful component of an assertion.
In place of grounding, Sider recommends that we proceed by means of metaphysical semantics (Sider 2011: 118–125). Sider's notion of metaphysical semantics corresponds very closely to what we have called conceptual grounding. Sider in effect denies that there is any such thing as ontological or non-conceptual grounding. Therefore, Sider's approach to metaphysics is largely Quinean in spirit. We should acknowledge as really existing only those entities and kinds that are part of the fundamental structure of the world. What is most innovative about Sider's approach is his inclusion of logical elements in the world's fundamental structure.