Here is an example. Suppose that we have a substance, S, that is both red and spherical. According to Classical Relational Realists, this fact consists in the fact that S stands in the instantiation relation to the universals REDNESS and SPHERICALITY. But since these universals aren't parts of S, S's being both red and spherical is not intrinsic to it. The objection can be put in the form of an argument based on grounding or truthmaking:
Assume that a's being F is intrinsic to a.
Assume, for reductio ad absurdum, that a's being F is grounded in a's having some relation R to F-ness, but this R-relation is compatible with F-ness's not being part of a. (Relational Realism) Assume, then, that in this case F-ness is not part of a.
The grounding-in relation is asymmetric and transitive. (There are no cycles or loops of grounding.)
If a's being R-related to F-ness is intrinsic to a, then a's being R-related to F-ness is grounded in a's being F.
If a's being R-related to F-ness is intrinsic to a, then a's being F is grounded in itself. (2, 3—transitivity, 4)
So, a's being R-related to F-ness is not intrinsic to a. (5, 3—asymmetry)
If a's being F is grounded in something that is not intrinsic to a, then a's being F is not intrinsic to a. (Definition D2.3, transitivity of grounding)
So, if a's being F is grounded in a's having some relation R to F-ness, then this R-relation entails F-ness's being part of a. (Discharging assumption of 2)
Premise 4 seems plausible. If a's being R-related to F-ness is intrinsic to a, it must be grounded in (made true by) some intrinsic feature of a's, and what other intrinsic feature of a's could that be besides a's being F?
This seems like a solid argument. What reply could Classical Relational Realists make?
Classical Relational Realists can reply that there are no truly fundamental properties of substances (except perhaps that of being a particular and not a universal). All properties are reducible to the instantiation relation plus the class of universals. Consequently, REDNESS and SPHERICALITY are not really fundamental, and so not really intrinsic, to red things and spheres. This is a surprising result of Classical Relational Realism! It is contrary perhaps to common sense, to the impression that we have of many apparently intrinsic properties.
There is another way for Classical Relational Realists to try to blunt the force of the Extrinsicality Objection. They could suppose that universals are really present in space and time, by being co-located with the substances that instantiate them. If a substance instantiates REDNESS, then REDNESS is present wherever that substance is. But this move doesn't make REDNESS a fundamental property of the particular, and so it doesn't really defuse the Extrinsicality Objection. However, it might make the extrinsicality of a thing's redness more palatable, since it enables us to suppose that REDNESS and the particular that instantiates it share a location.
If one doesn't think that sharing a location is enough to account for the impression of the intrinsicality of these other properties, then one may want to give Constituent Ontology a go.
9.3.2 Constituent Ontology
Here again is the definition of Constituent Ontology:
9.1T Constituent Ontology. When a substance instantiates a property, the instantiation relation between the two consists in the fact that the property is a part of the substance.
Since substances are thickly charactered, the Constituent Ontologist thinks that among the parts of a substance are each of the properties it instantiates. Properties are not, of course, parts in the way that ordinary physical parts are. Properties aren't like branches or pen caps or hands or tails. We might say that properties are metaphysical parts, or constituents. Just as some things have physical structure, defined by the nature of and relations among its physical parts, substances have metaphysical structure, defined by the nature of and relations among its metaphysical parts, its constituents. (Incidentally, we will use constituent-talk here, to avoid any confusion between physical and metaphysical parts.) One of the ways this metaphysical structure manifests itself is in the character of a thing. A thing's character seems to be a result of the way it is, the way it is structured metaphysically. Sometimes this thought is expressed by saying that a thing's properties are immanent in it, rather than separate from it as in Relational Ontology.
Many Constituent Ontologists accept what Michael Loux has called the ‘Principle of Constituent Identity’:
PMeta 4.1 Principle of Constituent Identity (PCI) for Substances. If A and B are substances, and every constituent of A is a constituent of B and vice versa, then A is identical to B.
The reason many Constituent Ontologists accept PCI is quite simple: Constituent Ontologists tend to think of a substance as a complex object composed out of its constituents and their interrelations. Thus, given that the interrelations among a substance's constituents are like the interrelations among other substances' constituents—which we will see is a common commitment of Constituent Ontology as well—it is hard to see how things with identical constituents could fail to be identical. If substances are made out of their constituents, and if this substance and that one have the same constituents, this and that must be the same substance.
PCI is very much like a corresponding principle about physical parts. Suppose there was a thing A that had just the same physical parts as a thing B. That is, every part of A is a part of B and vice versa. It seems to follow from this that A just is B, that the “two” are really just one. So the intuition underlying PCI is very much like the intuition underlying this claim about the connection between physical parts and identity. Whether we accept PCI should depend on exactly how we think of parts and wholes.2 If a whole is “nothing over and above” its parts, that is, if it represents no “addition to being,” then PCI seems undeniable. In contrast, if the basic relation is the relation of composition, the relation by which simples x1, x2,…, xn compose a new, composite entity y, then composite things are just as fundamental as their parts. If so, it is hard to see why two different composite things might not share all of the same proper parts, in contradiction to PCI (see Hochberg 1965 and Rodriguez-Pereyra 2004).
Nonetheless, even if PCI were not true of all composite objects, there might be a good reason to think that PCI should apply to bundles of properties. The first issue to consider is whether bundles have their parts essentially or can gain or lose parts. If B is a mere bundle of various parts, then it seems reasonable to assume that B could not survive the addition or subtraction of parts from the bundling relation. Thus, it does seem that all bundles of properties have those property-parts essentially because the bundle is nothing more than the combination of those properties. If this is true, then it also seems plausible to think that sameness of constituents is sufficient for the identity of property-bundles. Consider the following, weaker version of PCI:
PMeta 4.2 Weak PCI. If x and y are necessarily composite and necessarily have the same proper parts, then x = y.
Weak PCI is plausible for bundles. Bundles of properties are intrinsically featureless—all of the features of the bundle are derived from the features of its constituents. So, bundles have no holistic character, and so they couldn't have any essence above and beyond the properties they contain. Weak PCI then applies to property-bundles, since each bundle is essentially composite and essentially has the components it does. If Weak PCI applies to bundles, then it is reasonable to infer that PCI also applies to them, since bundles have their parts essentially.
Further, suppose contrary to Weak PCI that two distinct bundles had the same constituents. If the constituents are essential to each, neither can gain or lose constituents. They are not only indistinguishable in fact—they are essentially indistinguishable. What then could possibly make the two bundles two?
Let's briefly consider some of the cases involving material objects where it seems that PCI might be false. One classic example involves a statue, Goliath, and the lump of clay of which the statue is composed, Lump. The two, Goliath and Lump, seem to be com
posed of exactly the same material parts (the same clay granules, the same atoms and molecules), and yet there seem to be two things rather than one, since Lump can survive if it is squashed into a ball, but Goliath cannot survive such squashing.
If we reflect on a case like this, we see that the statue is not a mere bundle of clay granules. It has, in addition, a certain essential structure or shape. If this shape is destroyed, the statue ceases to exist, even if the clay granules remain bundled together. However, bundles of properties have no internal structure or internal feature of any kind except insofar as that structure or feature is represented or constituted by the presence of an appropriate property in the bundle. Bundles of properties are mere bundles, and so their identity is constituted by the particular set of properties that are bundled together. Same properties, same bundle, same substance.
Suppose, though, that a property-bundle had some internal structure. Since we are working with Constituent Ontology, the bundle's having that structure must consist in its containing some appropriate property or properties. This is just an application of the general Constituent Ontology strategy of explaining a thing's attributes in terms of its constituent properties. Consequently, any bundle having exactly the same constituents would have exactly the same internal structure. Unlike the case of Goliath and Lump, there would be nothing to distinguish one bundle from the other.
Here is a further argument for PCI's applicability to all property-bundles. In every case, the distinctness of two things should be intrinsic to the pair. Nothing outside of the two things could be responsible for their being distinct from each other. But what could be intrinsic to a pair of bundles except their constituents? Hence, two bundles cannot be distinct except by containing distinct constituents. In contrast, two simple things (like two universals or two bare particulars) can be distinct despite the fact that neither contains any proper part. Simple things can carry their own distinct identity within them, while mere bundles must derive their distinct identities from the distinctness of their constituents.
In conclusion, we have good reason to accept PCI if we limit its scope to property-bundles:
PMeta 4.3 PCI for Property-Bundles. If x and y are bundles of properties, and x and y have the same proper parts, then x = y.
Since Constituent Ontologists maintain that a substance's properties are among its constituents, an obvious question is whether substances have constituents other than those properties. If one says that the only constituents a thing has are its properties, then one is a Bundle Theorist. If one says that a substance has a constituent or constituents other than its properties, then one is a Substrate Theorist.
9.1T.1T Bundle Theory: The only constituents of each substance are its characterizing properties.
9.1T.1A Substrate Theory: Each substance has a constituent other than its characterizing properties, a substrate.
We will consider these views in turn.
9.3.2.1 Bundle Theories.
The idea behind Bundle Theory is that substances are nothing but bundles of properties. As we will see in detail below, this contrasts with Substrate Theory since substrates are in a different fundamental category from properties. Bundle Theorists thus have a putative qualitative advantage over Substrate Theorists because they maintain that the category of substance can be wholly reduced to the category of property. Bundle Theory has need of just one fundamental category in their account of substance, while Substrate Theory requires two. Bundle Theory thus constitutes a natural, simple starting point for working out one's Constituent Ontology. The question is whether Bundle Theory can pull off its promised reduction.
Since properties come in two varieties, tropes and universals, so does Bundle Theory:
9.1T.1T.1T Trope Bundle Theory: Bundle Theory is true, and tropes ground character.
9.1T.1T.1A Classical Bundle Theory: Bundle Theory is true, and universals ground character.
Trope Bundle Theory thinks of substances as bundles of tropes, whereas Classical Bundle Theory thinks of substances as bundles of universals. Therefore, Trope Bundle Theory reduces one category of particular, substance, to another category of particular, trope. Classical Bundle Theory, on the other hand, reduces the category of substance to a category of non-particular, universal. Each of these Bundle Theories comes in many varieties, distinguished by the ways in which they variously construct bundles. We will, however, only discuss those varieties with respect to Classical Bundle Theory, since the troubles for Trope Bundle Theory have more to do with the proposed reduction to tropes and less to do with the way the bundles of tropes are constructed.
Consider first Trope Bundle Theory. Tropes are either modifying or modular. There are serious problems for Trope Bundle Theory on both views.
Suppose tropes are modular and, therefore, that each trope has the character it grounds. As we saw with the Modular Trope variety of Relational Ontology above, there is a danger of character duplication if one thinks that substances themselves are charactered. The Trope Bundle Theorist, though, has a strategy unavailable to the Relational Ontologist: Trope Theorists can deny that modular tropes ground character by making something else, distinct from the trope itself, a possessor of that characteristic. The idea is to emphasize the bundle theoretic reduction of substances to tropes and to therefore deny that the substance itself has character. The sense in which a substance has some dimension of character, on this way of thinking, is just by having a constituent with that character. But this kind of character-having should be thought of in an attenuated sense. The substance doesn't really have character; only modular tropes do that. (Otherwise you get duplication!) This move is problematic, though, for it runs afoul of the following, plausible principle:
Existence of Thick-Characteredness. There are thickly charactered things.
If modular tropes are the only truly charactered things, then the Existence of Thick-Characteredness is false. It certainly seems, though, that there are things that are thickly charactered.
Notice that the trouble is created by the threat of character duplication, and this threat is in turn created by the fact that modular tropes have the character they ground. Modifier tropes, though, do not have the character they ground. Modifier Trope Bundle Theory does not, therefore, have to deny that substances really do have thick character, at least not for the reason Modular Trope Bundle Theory does.
Modifying Trope Bundle Theory, however, faces a different problem with character grounding. Modifying tropes do not have the character they ground. So consider the following question: What are the modifying tropes grounding the character of? What, that is, are modifying tropes modifying if Trope Bundle Theory is true? Given that Bundle Theorists reduce substances to a collection of properties (which are the only constituents of substances), the only answer Modifying Trope Bundle Theorists can give is that tropes are characterizing other tropes. There is just nothing else there to be characterized, on this view. But modifying tropes are usually supposed to have only formal character. If modifying tropes are characterizing other modifying tropes, then some modifying tropes will have to have non-formal character. This is now a second reason for defenders of modifying tropes to take seriously the idea that some modifying tropes modify other modifying tropes. Now we have a reason for thinking that modifying tropes might contain other modifying tropes as parts. However, we will still face the problem of saying what the biggest (most inclusive) modifying tropes modify.
The best solution for the Trope Bundle Theorist might be to adopt a hybrid theory—a theory according to which there are both modular and modifying tropes. We could now identify a substance with a single modular trope that contains many modifying tropes as parts. Now, however, the result looks less and less like a version of a bundle theory. The substance isn't merely a bundle—it is a modular trope (perhaps a spatial quantity module) filled with modifying tropes of various kinds.
What we have run up against is the need for an additional constituent, if Modifying Trope Bundle Theory is true, a non-pr
operty (or anyway a non-modifying trope) constituent. The only plausible way to work this out is by including some type of substrate. This, of course, simply abandons Modifying Trope Bundle Theory in favor of a Trope Substrate Theory. We will take up the latter view in Section 9.3.2.2.
We turn, therefore, to Classical Bundle Theory. Classical Bundle Theory purports to reduce substances to bundles of universals. Historically, Classical Bundle Theories have gone in for a more or less sparse theory of properties. This is not surprising, since if one has an abundant theory of properties, then each substance will have innumerable universals as constituents. Thus, substances would have a surprising degree of metaphysical complexity. We will assume a sparse theory here according to which the only universals there are are natural properties (in the sense discussed in Section 7.2.1.1).
The most obvious way to work out one's Classical Bundle Theory is to identify bundles with sets of universals. This is Constructive Bundle Theory:
9.1T.1T.1A.1 Constructive Bundle Theory. Substances correspond one-to-one to sets of universals.
Do all sets of universals correspond to particulars? It would seem not. There seem to be in fact no particulars that are green elephants. If that were so, then the set consisting of the universals GREEN and ELEPHANT would not be a particular. At the same time, there are many gray elephants. Consequently the set consisting of the universals GRAY and ELEPHANT does not constitute a single particular but a whole host of particulars.
Let's start with some ordinary particular, Jumbo the elephant, for example. Let's say that Jumbo is some actual elephant. We would like to say that there are some universals (like GRAY and ELEPHANT) that Jumbo exemplifies, and others (like GREEN or ANT) that it does not. A natural move for Classical Bundle Theorists to make would be to identify Jumbo with the set containing exactly those universals that Jumbo intuitively exemplifies. Since Classical Bundle Theorists deny that particulars are fundamental entities, it will also follow that the exemplification of universals by particulars is not a fundamental relation. Instead, Classical Bundle Realists will have to introduce a primitive relation of co-instantiation among universals. Co-instantiation is a property of sets of universals. Intuitively, a set of universals has the property of co-instantiation just in case the set corresponds exactly to the universals that are exemplified by some one particular. Jumbo would then correspond to a co-instantiated set of universals.
The Atlas of Reality Page 30
