Naturalism and material objects MICHAEL REA
Regarding the way they are internal to the subject on internalist theories, it is interesting to note that the most frequently employed candidates for the characteristic mark of (at least most) mental states not only serve as ontological criteria, they also capture crucial elements for internalist epistemic theories. While it is difficult to give a characteristic mark of the mental (or the physical!) in terms that command universal assent, nevertheless, most agree that the most likely candidate is this: a property (or event) is mental just in case it is such that it is self-presenting or that to which the subject has private or first-person direct access or that which could not exist unless there were sentient subjects. However, it has been hard to sustain the claim that this characteristic mark could be true of a physical particular, property, or relation. Internalist properties and their relationship to epistemic subjects, as these actually function in internalist epistemological theories, are just not reducible to natural, physical, causal/functional entities. Thus, if the best version of naturalism is strict physicalism, then the best naturalist theory of knowledge or justification (or warrant) will be externalist.
We have looked at a number of philosophers who express different aspects of the epistemic attitude that constitutes naturalism. Let us now turn to an overview of the naturalist’s view of how things came to be.
2. The naturalist grand story
The naturalist has an account of how all things whatever came to be. Let us call this account the Grand Story. The details of the Grand Story need not concern us here.
Suffice it to say that some version of the Big Bang is the most reasonable view currently available. On this view, all of reality – space, time, and matter – came from the original
“creation” event and the various galaxies, stars, and other heavenly bodies eventually
developed as the expanding universe went through various stages. On at least one of those heavenly bodies – earth – some sort of pre-biotic soup scenario explains how living things came into being from non-living chemicals. And the processes of evolution, understood in either neo-Darwinian or punctuated equilibrium terms, gave rise to all the life forms we see including human beings. Thus, all organisms and their parts exist and are what they are because they contributed to (or at least did not hinder) the struggle for reproductive advantage, more specifically, because they contributed to the tasks of feeding, fighting, fleeing, and reproducing.
There are four important things to note about the Grand Story. First, the core of the Grand Story is the atomic theory of matter and evolutionary theory. If we take John Searle to be representative of naturalists here, this means that causal explanations, specifically, bottom-up but not top-down causal explanations, are central to the (alleged) explanatory superiority of the Grand Story.30 Second, it is an expression of a naturalist version of philosophical monism according to which everything that exists or happens in the world is susceptible to explanations by natural scientific methods.
Whatever exists or happens in the world is natural in this sense. Prima facie, the most consistent way to understand naturalism in this regard is to see it as entailing some version of strict physicalism: everything that exists is fundamentally matter, most likely, elementary particles (whether taken as points of potentiality, centers of mass/energy, or units of spatially extended stuff/waves), organized in various ways according to the laws of nature. By keeping track of these particles and their physical traits we are keeping track of everything that exists. No non-physical entities exist, including emergent ones. This constitutes a strict sense of physicalism. When naturalists venture away from strict physicalism, however, they still argue that additions to a strict physicalist ontology must be depicted as rooted in, emergent from, dependent upon the physical states and events of the Grand Story.
Third, a naturalist depicts living organisms as property-things or ordered aggregates. This follows from the two key elements of the Grand Story: the atomic theory of matter (and the rejection of any form of vitalism) and evolutionary biology (where macro-features – phenotypes – are explained by causal mechanisms at the micro-level – genotypes – and in terms of so-called functional explanations in terms of the constrains of reproductive advantage). The conjunction of these two elements implies two things about living creatures. To begin with, they are not genuine substances understood, in the classic sense, as essentially characterized particulars (1) whose parts stand in internal relations to the substantial whole; (2) which sustain absolute sameness through change; and (3) which have a nature possessed by all members of their natural kinds. Darwin’s theory of evolution has made belief in, e.g.
human substances with human natures, though logically possible, nevertheless, quite implausible. As E. Mayr has said:
The ontological status of properties 77
The concepts of unchanging essences and of complete discontinuities between every eidos (type) and all others make genuine evolutionary thinking impossible.
I agree with those who claim that the essentialist philosophies of Aristotle and Plato are incompatible with evolutionary thinking.31
This belief has, in turn, lead thinkers like David Hull to make the following observation:
The implications of moving species from the metaphysical category that can appropriately be characterized in terms of ‘natures’ to a category for which such characterizations are inappropriate are extensive and fundamental. If species evolve in anything like the way that Darwin thought they did, then they cannot possibly have the sort of natures that traditional philosophers claimed they did.
If species in general lack natures, then so does Homo sapiens as a biological species.
If Homo sapiens lacks a nature, then no reference to biology can be made to support one’s claims about ‘human nature.’ Perhaps all people are ‘persons,’
share the same ‘personhood,’ etc., but such claims must be explicated and defended with no reference to biology. Because so many moral, ethical, and political theories depend on some notion or other of human nature, Darwin’s theory brought into question all these theories. The implications are not entailments.
One can always dissociate ‘Homo sapiens’ from ‘human being,’ but the result is a much less plausible position.32
Thus, living organisms are not genuine substances for the naturalist. Rather, they are property-things. I mean by a “property-thing” an ordered aggregate, a set of parts put into external relations by some ordering principle (e.g. a law of nature) to form a whole. For our purposes, it is important to observe that given the naturalist etiology in general and evolutionary theory in particular, naturalists like Hull and Mayr conclude that certain abstract objects, namely, natural kind essences, either do not exist or they are not exemplified.
Finally, while some naturalists eschew questions about the nature of existence itself, others have formulated a definition of existence based on a naturalist epistemology and consistent with the Grand Story. Thus, Bruce Aune defines a exists as “a belongs to the space-time-causal system that is our world. Our world is, again, that system of (roughly) causally related objects. . .”.33 Along similar lines, D. M. Armstrong says that for any entities, the following question settles the issue of whether or not those entities can be said to exist: “Are these entities, or are they not, capable of action upon the spatio-temporal system? Do these entities, or do they not, act in nature?”34
Daniel Dennett claims that when we are trying to find out whether or not some entity
like the self exists, what we must do is locate the entity within the causal fabric.35 Keith Campbell applies the same reasoning to the question of the existence of emergent entities like social characteristics by claiming that the test of their existence turns on their ability to exhibit independent causality because “power has been recognized as the mark of being.”36 Finally, Jaegwon Kim says “Causal powers and reality go hand in hand. To render mental events causally impotent is as good as banishing them from our ontology.”37
3. The naturalist ontology
In order to characterize a naturalist ontology, we may distinguish global vs. local naturalism and weak vs. strong naturalism. Roughly, global naturalism is the view that the spatio-temporal universe of natural entities studied by science is all there is. Global naturalists (e.g. Wilfrid Sellars) reject abstract objects of any kind, including traditional realist properties. Local naturalists (e.g. Jeffrey Poland) either accept or are indifferent towards abstract objects but they insist that the spatio-temporal universe consists only of entities studied by the natural sciences. Local naturalists reject Cartesian souls, Aristotelian entelechies, and so forth. A distinction also exists between strong and weak naturalists. Strong naturalists (e.g., David Papineau) accept a strict physicalism for the natural world, while weak naturalists (e.g. John Searle) embrace various emergent entities.
For someone to be a consistent naturalist, he or she must hold that the entity which emerges or comes-to-be (as naturalists understand that notion) (1) fits solely within a naturalist ontology (i.e. it bears a relevant resemblance to paradigm case natural entities of physics in its being and behavior); (2) can, at least in principle, be given a naturalist etiological explanation in keeping with the naturalist epistemic attitude and Grand Story if it is to be explained at all. The naturalist ontology must be selected so it will cohere well with the naturalist epistemology and Grand Story in order for the naturalist to be justified in asserting the scientific explanatory superiority of naturalism vs. rival world-views.
Elsewhere, I have argued that these aspects of naturalism imply that a consistent naturalist should be a strong naturalist.38 I can only gesture at some salient points here. More than anyone else, D. M. Armstrong had clearly reflected on this topic and he concludes that all entities in the naturalist ontology must be (1) spatially located; (2) entities knowledge of which can be located within physical, causal processes; (3) capable of entering into causal relations and whose existence can be given a natural scientific causal explanation. To illustrate these points in Armstrong, the following statement is an example where he uses a naturalist externalist epistemology to settle issues in ontology:
The ontological status of properties 79
If any entities outside this [spatio-temporal] realm are postulated, but it is stipulated further that they have no manner of causal action upon the particulars in this realm, then there is no compelling reason to postulate them.39
In this context, Armstrong is claiming that the only way something can interact with natural entities – including cognitive processes so as to be objects of knowledge – is by way of causation.
Elsewhere, Armstrong employs the Grand Story as a criterion for an acceptable naturalist ontology: “I suppose that if the principles involved [in analyzing and explaining the origin of or processes of change in things within the single all-embracing spatio-temporal system which is reality] were completely different from the current principles of physics, in particular if they involved appeal to mental entities, such as purposes, we might then count the analysis as a falsification of Naturalism.”40
Finally, as a naturalist, Armstrong explicitly rejects internal relations on the grounds that they cannot be spatio-temporally located.41
Strong naturalists eschew things like genuinely emergent mental properties/
events for the same reason they set aside Cartesian dualism. Regarding Cartesianism, it is often claimed that dualist interaction is outside the constraints of naturalism because it requires an odd causal relation (i.e., one that relates entities from such different orders of being that the causal relation is no longer clearly a physical relation) and it postulates two radically different entities that enter into that relation one of which is not at home in the naturalist epistemology and etiology. Now these same points can and have been raised against a supervenience/emergence relation between physical and mental entities.
Perhaps enough has been said about contemporary naturalism to put us in a position to list without much development some of the main reasons why naturalists such as Sellars have embraced global and not local naturalism.
Naturalism and traditional realist properties
At least four reasons have been offered for the claim that global naturalism is preferable to local naturalism:
1.
Traditional properties and spatio-temporal location: Some have offered a sort of
“argument from queerness” against traditional realist properties to the effect that they are entities of a very strange sort, utterly different from anything else in the universe as depicted by the contemporary naturalist. Some have developed this argument by focusing on a special class of universals (e.g.
propositions, axiological properties), others have claimed that no physicalist or naturalist non-reductive account of deep, metaphysical modality is possible and, thus, realist properties and their relations to each other are utterly
different from the rest of the naturalist ontology.
2.
Traditional properties and the naturalist epistemology: As Laurence BonJour has recently noted, it is hard to accommodate traditional realist properties to physicalist depictions of human cognizers and of the causal interactions that constitute cognitive processes such as perception.42 Given this depiction of humans and their various epistemic relations to things outside them, it is hard to see how we could have cognitive access to traditional realist properties. On a naturalist view, a necessary condition for knowledge is the ability of the knower to interact causally with the object of knowledge in standard physical ways. If properties are as traditional realists say they are, then they are causally inert abstract objects, yet we do seem to have such knowledge (or at least to know things that can be expressed using property-talk). Moreover, for those naturalists who eschew first philosophy and the conclusions reached therein, the entities associated with traditional realist properties (e.g. the properties themselves, the predication relation, the nature of an individuator) must be set aside. These entities are not “required” by our best scientific theories; they are allegedly justified if at all by considerations that are independent of and, arguably, epistemically prior to scientific concerns.
3.
Traditional properties and being: Some argue that traditional realist properties violate the causal criterion of existence since they are non-spatio-temporal entities and remain that way even when exemplified. Moreover, given the non-spatio-temporal nature of properties, it becomes hard to see why their existence is dependent upon being exemplified. Thus, it is argued, uninstantiated universals exist and they clearly violate the causal criterion of being.
4.
Traditional properties and predication: Traditional realism is a classic example of a “two world” ontology and, as such, it becomes difficult to explain in naturalist terms how there could be any kind of connection between abstract objects on the one hand and the spatio-temporal world of particulars and events on the other. Moreover, because the predication relation (i.e. nexus) (1) is non-spatio-temporal and (2) connects entities from different “worlds”, it is hard to see how the relation itself bears a relevant similarity to strictly physical entities.
In this regard, the predication relation presents the same sorts of problems to global naturalists that Cartesian interaction and emergent/supervenient relations do to weak naturalists (e.g. Where does Cartesian interaction take place and where is the predication relation exemplified?). The so-called explanatory gap between physical and mental events/properties has a parallel between spatio-temporal entities and those completely outside space and time.
Finally, since at least in some sense, property-instances are ontologically dependent upon their properties, and since those properties are not themselves physical, if we take events as temporal states of affairs, the dependence of everything that exists or occurs upon physical factors in the physical domain becomes hard to maintain.
Rather than elaborate on these points, I want to turn to a consideration of naturalist strategies that accept these arguments and try to offer a naturalistically acceptable analysis of properties.
The ontological status of properties 81
Two naturalist attempts to reconcile naturalism with
acceptance of properties
Many naturalists, especially those sensitive to issues in first philosophy, have realized that (1) properties exist and (2) the traditional realist construal of properties is an uneasy fit with scientific naturalism. These naturalists also realize that traditional realism has a long and distinguished history and the onus is on them to develop an adequate theory of properties consistent with scientific naturalism. In my view, the two most powerful naturalist attempts at this project are those developed by Keith Campbell and D. M. Armstrong. In this section, we shall evaluate the adequacy of their efforts at such a rapprochement.
Keith Campbell
The most articulate version of nominalism currently available is the “trope”
nominalism of Keith Campbell.43 In this section, I shall describe the evolution of C ampbell’s nominalism over the last several years and criticize his current, refurbished version of nominalism. It will be instructive to analyze precisely the trajectory of Campbell’s thought because such an analysis will reveal what is essential to a nominalist position. Part of the impetus for Campbell to change his views were criticisms that others and I raised against his earlier claims.44 I hope to show that his refurbished nominalism is no more successful in avoiding certain problems than his earlier version. In what follows, I shall begin by investigating Campbell’s earlier version of nominalism followed by an analysis of his most recent position.
1. CAMPBELL’S EARLY VERSION OF NOMINALISM
According to the early Campbell, the proper way to assay “Socrates is red” is this: the simple trope (Campbell’s term for an abstract particular), red1, which is a member of the similarity set “redness,” is a part of the whole, Socrates, and Socrates is a bundle of compresent tropes.45 This assay contains at least four key elements that made up Campbell’s earlier position: the nature of a trope, the nature of a “universal,” the relationship between a trope and its “universal,” and the relation between a trope and a concrete particular that has it.
In Campbell’s earlier writings, a trope was taken to be an abstract particular. By
“particular” he meant an entity that is exhausted in one embodiment; by “abstract”
Campbell meant an item which is got before the mind by an act of abstraction, that is, by concentrating attention on some, but not all, of what is presented. Examples of tropes are the specific, numerically singular taste of a lollipop, or its specific color, shape, etc.
Further, tropes are basic, primitive entities. Campbell says of a trope: “It embraces
no variety at all.”46 Again, “They are infimae species, taken as particulars.”47 In another place he says “They are in Hume’s sense substances, and indeed resemble his impressions, conceived realistically rather than idealistically.”48
On this view, a trope, construed as a basic particular, is simple, fundamental, and independent. By “simple” is meant that they have no further real parts (ones that can be separated off from the whole in reality and not merely in thought). The notion of being fundamental is explained this way. Entity A is fundamental relative to entity B if B’s existence depends on A but not vice versa. Finally, “independence” means that an entity is truly independent if it could be the only entity in the universe. Tropes (at least non-emergent ones), then, are basic particulars. Concrete particulars like Socrates are derivative and it is a contingent fact that tropes usually come grouped together as concrete particulars.
An additional, important feature of tropes on Campbell’s earlier view was the central role that space (or space-time) played in spelling out what a trope is. First of all, tropes do not exclude one another from being at the same place. In fact, a concrete particular is just a “clump” of compresent tropes at a place.
Second, a trope exists at one definite location; it is a quality-at-a-place. Campbell claimed that this should not be taken to imply that a trope has two different constituents, namely, its qualitative nature (which is a universal) and its place (which is particular). A quality-at-a-place is a single, particular reality. This is clarified by Campbell’s view of the special status of the geometric figures of an entity, i.e. its form and volume. Form and volume are special tropes not like any others. Their presence in a particular compresent sum of tropes is not contingent; tropes cannot be present except by being in a formed volume. Tropes are, therefore, essentially regional.
Wherever a trope is, there is formed volume. On the other hand, Campbell asserted that shape and size are not found except in company with characteristics. So formed volume is both essential to ordinary tropes and in itself insufficient to count as a proper being. Since the shape and size of a trope are inseparable from its color, then the former are not really different entities from the latter. Inseparable entities appear to be identical for Campbell; they differ only by a distinction of reason (when A and B
differ by a mere distinction of reason, then A is identical to B). Put differently, the place (located, formed volume) and qualitative nature of a trope are identical and differ only as different ways of thinking of or speaking about the trope.
So much, then, for tropes. For Campbell, the relationship between Socrates and all the tropes “in” Socrates is a part/whole relation. Each trope is part of a bundle of compresent tropes which is Socrates. Thus the predication relation is reduced to a type of part/whole relation in Campbell’s nominalism. The “universal” redness is merely a set of exactly resembling tropes, e.g. red1 (in Socrates), red2 (in Plato), etc.
In Campbell’s earlier writings the relationship between each trope and its corresponding “universal” is clear and straightforward: it is the ε of set membership.
The ontological status of properties 83
However, when we turn to an analysis of the exact similarity relation that obtains between or among tropes in a set, things are not so clear. In fact, the early Campbell’s treatment of similarity is easily the weakest part of his version of nominalism. I have subjected Campbell’s views on exact similarity to detailed analysis and critique elsewhere, and will content myself here with some general observations relevant to our current discussion.49 One thing that is clear is that the unity of a set of tropes is grounded in the exact similarity relation. And the exact similarity relation is like the relation of identity in that it is symmetrical, transitive, and, arguably, reflexive.
On the other hand, it is not clear whether in Campbell’s earlier views of exact similarity he saw it as an internal or external relation and he appears to have equivocated between the two views. To see this, we need to recall two things true of internal relations as they are usually construed, first, if the R of a to b is internal to a, then anything that does not stand in R to b is not identical to a. Second, internal relations are not primitive but, rather, are derived subsistences grounded in the natures of the entities they connect. As Gustav Bergmann put it, The ontological ground of an internal connection lies wholly ‘in’ the two or more entities it connects. More precisely, it lies in their natures. The notion is so crucial that I reword it. The ontological ground of an internal connection is the natures of the entities it connects and nothing else. Still differently, an internal connection has no ontological ground of its own . . .50
More recently, D. M. Armstrong has defined an internal relation as a relation which is logically determined by the nature of the related terms.51 Armstrong goes on to point out that we can explain why internal relations are such that given two internally related entities a and b, there is no possible world in which the objects remain unaltered but in which the internal relation fails to obtain by recognizing that internal relations are derived from and grounded in the natures of the entities so related.52 Since internal relations are ontologically grounded in the more basic metaphysical natures of the related entities, internal relations are derived subsistents, not primitive entities.
Now Campbell explicitly patterns his version of nominalism after that of D. C.
Williams and Williams clearly held that exact similarity between two tropes is an internal relation.53 Moreover, Campbell says,
Can’t we ask about two white tropes, no less than about two white things, what it is that gives them their resemblance? Perhaps we can answer legitimately just that they resemble each other because of what they are like, or because of their character or nature. Perhaps if we quantify over tropes we will have a cause to quantify over universals as well. But I for one have no confidence either way.54
In other words, in this passage Campbell does not clearly reject the internal relation view. But for three reasons, I think we should reject this interpretation. For one thing, in the passage just cited, Campbell grasps the fact that an admission that exact similarity is an internal relation commits one to universals and his entire system is meant to be an alternative to realism about properties. Second, elsewhere Campbell explicitly states that the relation of exact similarity is an unanalyzable, primitive, brute fact and this can only mean that the relation is an external one.55 Finally, though Campbell claims that an anti-naturalist could embrace nominalism (e.g. by accepting the existence of mental tropes), nevertheless, he clearly wishes to advocate nominalism as a way of allowing the existence of qualities within the confines of a naturalist stance. And as D. M. Armstrong has pointed out, naturalists, if they are self reflective and consistent, should eschew the existence of internal relations.56 I do not wish to defend Armstrong on this point, though I think he is right. Suffice it to say that if internal relations should be eschewed by a naturalist, then since Campbell wishes his ontology to be an expression of naturalism, he should avoid treating exact similarity as an internal relation. And it is charitable to interpret him in this way.
However, before I leave the topic of exact similarity, it should be pointed out that, at present, Campbell sees exact similarity as an internal relation.57 It is important to note that Campbell’s earlier confusion here, his change of position, and problems in his current treatment of exact similarity are rooted in a tension endemic to nominalism: (basic) tropes are simple entities – particularized natures – and there is a mere distinction of reason between either a trope’s nature and location (or formed volume) or its nature and particularity. Thus, when one thinks about the fact that two red tropes are red, it is easy to see how their resemblance could be grounded in their color. But such a temptation does not arise if we think of two resembling tropes as merely located entities (or particulars). The relation of similarity between the
“natures” of two red tropes is an internal one, but the relation between the location or particularity of those tropes is external. But how can this be if the nature and particularity of a trope differ by a distinction of reason and thus, are identical? I shall not probe this tension further here except simply to point out that Campbell’s use of a distinction of reason here obscures his views and causes him to be inconsistent.
Recall that if A and B differ by a distinction of reason, then A is identical to B. With this in mind, I want to raise three puzzles that are problematic for Campbell’s earlier views. Indeed, Campbell saw these puzzles in an earlier critique of mine and admitted that they were decisive against his position.58 Consider a concrete particular, say, an apple. The taste trope and color trope of the apple, indeed all the apple’s tropes, are at the same place since the apple is merely a bundle of compresent tropes. Now the nature of the taste trope differs from its location/formed volume by a distinction of reason only. Likewise, with the color trope and all the others. But then, the taste of the apple is identical to its place as are all the other tropes in the bundle. Now by the
The ontological status of properties 85
transitivity of identity, all the tropes of the apple are identical to each other and, indeed, the apple is reduced to a bare location. Concrete and abstract particulars turn out to be bare simples and this is incoherent. This is puzzle one.
To understand puzzle two consider two compresent bundles of tropes, A and B.
Bundle A includes a red and a taste trope at the same location. Bundle B includes a blue trope and a taste trope from the same exact similarity set as the taste trope in bundle A. In other words, the concrete particulars A and B differ in color but have the
“same” taste. Now, the red trope of A is in a set of other red tropes exactly similar to it.
Since the red trope’s “redness” is identical to its place/formed volume (they differ merely by a distinction of reason) and since the nature of A’s taste trope is also identical to the same place, A’s red trope (by the transitivity of identity) is identical to A’s taste trope. The same is true of B’s taste and blue tropes. Furthermore, A’s taste trope is in the same exact similarity set as B’s taste trope. But this means that A’s red trope and B’s blue trope are in the same similarity set of taste tropes. And since exact similarity is transitive, B’s blue trope will be in the similarity set of red tropes exactly like A’s redness. If this is true, it is hard to make sense out of the idea of a similarity set based on a trope’s nature. If an exact similarity set of red tropes includes a blue trope, the very idea of an exact similarity set of red tropes becomes vacuous.
The third puzzle is this. According to the early trope view, tropes are basic, independent existents and concrete particulars are derivative bundles of tropes. It is a purely contingent matter that a given trope is in a given bundle. The trope could have existed in another bundle of tropes or without being in any bundle at all. Now, a trope cannot exist if it is not individuated. But since place is the principle of individuation, this means that a given trope could not have existed at another place since that would make it a different trope. This same point holds for all the tropes at a given location which together form a compresent bundle. But then, it is not a contingent fact that these particular tropes combine to form a specific concrete particular, because in order that these tropes be these tropes, they must be at a given location. It is a matter of necessity that a concrete particular have just the tropes it does.
In fact, the concrete particular is identical to the bundle of tropes, and, therefore, to each trope. And since the concrete particular is identical to each trope, it is identical to a mere place.
It is important to see just what is going on in these puzzles. The trope view must assay a basic trope as a simple in order to avoid assigning the individuating and qualitative roles to non-identical constituents in the quality-instance, for this is what realists do (e.g. red1 has an individuator, say a bare particular expressed by 1, the universal redness, and a tie of predication). Since tropes are simples and since place/
formed volume is identical to a trope’s nature (differing by a distinction of reason only), then the trope nominalist must either remove the 1 and make the identity reflect the entity redness or else remove redness and make the identity reduce to bare location. The former collapses into realism and the latter is incoherent as the puzzles
indicate.59
2. CAMPBELL’S REVISED VERSION OF NOMINALISM
In light of these and other criticisms, Campbell has modified his trope nominalism.60
Two aspects of his current thinking are especially relevant to our present concerns: his assay of a trope, especially the role of location in that assay, and his clarification of the relation of exact similarity. He continues to hold that a basic trope like red1 is a simple entity – a particularized nature that as a matter of brute, unanalyzable fact sustains two roles: to be a particularized nature that is exactly similar in nature to other tropes in its similarity set and that is individuated from all other tropes. A basic trope is not a union of a particularizing entity and a particularizer. It is a single, simple item – a particularized nature. Campbell is not entirely clear about the distinction between the nature and particularity of a specific trope.61 On the one hand, he seems to continue to hold that the difference is just a distinction of reason since the nature and particularity of a trope do not constitute a genuine duality of entities which would turn tropes into complex entities. On the other hand, he says that the distinction between the nature and particularity of a basic trope is, perhaps, the formal distinction of Duns Scotus and, for Campbell, this amounts to the level of abstraction at which an entity is being considered (either as a particular or as a qualitative nature).
For two reasons I take it that Campbell’s mention of the formal distinction really amounts to a reassertion of a distinction of reason in different terms. First, Suarez summarizes the medieval discussion of Scotus’ formal distinction, correctly in my view, as being somewhat confused and hard to understand. But, nevertheless, based on the remarks of Scotus himself and his disciples, Suarez claims that the formal distinction turns out the be the distinction of reasoned reason.62 Second, C ampbell’s own clarification of his use of the formal distinction reduces it to a distinction of reason since, for him, the distinction is an epistemological one only and identity obtains between the entity singled out differently.
Campbell goes on to say that a trope is not individuated by place. As a matter of a posteriori Kripkean necessity, tropes are located, but the location of a trope is a quasi-trope (roughly, a mere appearance or pseudo-entity that is part of the manifest world; more on this latter) and as a matter of necessity is always connected to a trope (it is contingent at which place a trope is located but not that it be located): “. . . let us abandon the view that a colour trope is individuated by its place. Take compresence as a more abstract, more formal matter, recognize that a colour trope and a spatial quasi-trope are distinct entities and assay the presence of green at a place as the compresence of a green trope with a place one.”63 I have not been able to tell what distinction C ampbell sees between a trope’s formed volume and its place. But regarding formed volume, he says that it is also a quasi-trope, a sub region of space whose boundaries are fixed by the presence of the trope’s nature (e.g. color), that is
The ontological status of properties 87
not identical to the qualitative content within that formed volume.
So much, then, for Campbell’s new view of tropes. I have already mentioned that he has tried to clarify his view of exact similarity. In most places he says that this is an internal relation grounded in and supervenient upon the natures of the resembling tropes.64 However, in other places he says that the exact similarity between two tropes is an ultimate, basic, unanalyzable, primitive fact.65 There are at least two problems with these claims. First, in keeping with his naturalism, Campbell is committed to the idea that causal power is the mark of being.66 I, for one, do not know what the ontological status is for supervenient entities in Campbell’s ontology since they do not appear to have causal power. As we saw earlier, Jaegwon Kim has pointed out that for those with this metaphysical commitment, “To render mental events causally impotent is as good as banishing them from our ontology.”67 Kim’s remark applies to Campbell’s assay of exact similarity if that assay takes the relation to be a supervenient one. And Campbell himself admits that supervenient facts are mere pseudo-additions to our ontology in which no new being is involved.68
Second, if exact similarity is primitive, then no further analysis can be given either of the relation itself or of the fact that it obtains. But if the relation is internal, it is not primitive, not in the sense that the relation must now be reduced to identity – a reduction I accept but that Campbell is free to deny – but in the sense that exact similarity is grounded in the natures of the resembling tropes. As I hope to show below, once the need for a ground is admitted, difficulties follow that do not arise if such a need is denied. There are other interesting features of Campbell’s nominalism, but rather than examine those features, it will suffice if we can see what is wrong with these two fundamental aspects of his, or for that matter, any version of nominalism.
My first set of remarks targets Campbell’s new assay of a basic trope, starting with his claim that the new role that location or formed volume plays in his refurbished nominalism avoids problems that plagued his earlier position. Note first, that since a realist assay of a quality-instance treats it as a complex entity with its nature being a universal predicatively tied to an individuator, to avoid realism Campbell must hold that a basic trope is simple and this is where the main problems with nominalism lie.
Apparently, Campbell fails to recognize that these problems turn on the simplicity of a trope and not on the role of location/formed volume in a trope’s assay.
Moreover, Campbell’s new view of the role of location/formed volume makes his nominalism less available to the naturalist than was his former position and Campbell is clearly concerned to present his trope nominalism as an attractive option for a naturalist ontology. Why is his new position more problematic for a naturalist? As we have seen, a widely accepted understanding of a naturalist ontological commitment is that the spatio-temporal universe of strictly physical entities that constitute the appropriate objects of natural scientific study is all there is. Campbell accepts this understanding of naturalism at least in the sense that he understands naturalism to
require all entities to be spatio-temporal. In his earlier view, it was easy to see why tropes could be viewed as natural entities that are entirely within the spatio-temporal world of natural entities. But the current picture of the role of location/formed volume actually comes perilously close to turning tropes into traditional abstract (non-spatiotemporal) entities, specifically, Platonic abstract perfect particulars with the proviso that they must be exemplified by a spatio-temporal particular to exist.
To see this, recall that for the traditional view of properties as abstract entities, properties are “in” the particulars that exemplify them, but they are not themselves spatio-temporally located at the place of those particulars. Nor is the way they are “in”
those particulars a spatial relation; rather, it is the primitive relation of predication.
Redness is (non-spatio-temporally, predicatively) in a ball and the ball is on the table, but neither redness nor predication are natural, spatio-temporal entities. It is widely recognized that when a universal is exemplified by a particular, the resulting state of affairs (the having of the universal by the particular) is itself a particular. This has been called the victory of particularity. Now in the same way there is a victory of spatiotemporality. When a ball is red, the ball is spatio-temporally located as is the state of affair consisting in the-having-of-redness-by-the-ball. But neither redness nor predication is spatio-temporal. If one were to add the Aristotelian constraint that properties must be exemplified in order to exist, the result would rule out transcendental (unexemplified) universals, but it would not turn predication or exemplified properties into spatio-temporal, natural entities. They would still be abstract objects.
Now Campbell’s current views seem to parallel this picture precisely. A trope’s nature is a metaphysically complete entity ontologically prior to its tie to the quasi-trope of location/formed volume since the nature is not identical to the location and neither enters into the being of the other in a part/whole way. This means that the trope’s nature itself is not spatial but receives spatiality from its tie to the quasi-trope of location/formed volume. Nor is it clear that this tie is spatial since it connects a nature to a location. Now how is this different from the victory of spatio-temporality in the traditional realist view? The only difference is that a trope can be embodied only in one place. But not only is it difficult now to see why this is the case – if natures are identical to locations it is easy to see this, but since spatio-temporality is not inherent in their being, but rather, tied to them, it is less clear and more ad hoc to say such a tie can occur only once – it is also hard to see how a naturalist could appropriate such an analytic ontology into his or her world view. And even if we grant that a nature can be tied to space in one embodiment only, tropes turn out to be perfect particulars on one interpretation of Plato, and such entities were still non-natural abstract entities as was the tie that Plato claimed connects them to particulars in the space-time world.
Further, the simplicity of a basic trope in Campbell’s refurbished nominalism generates the same metaphysical difficulties that he admits were telling against his earlier position. Consider two red tropes. Qua red entities, they stand to each other in
The ontological status of properties 89
an internal exact similarity relation grounded in their natures. But qua particulars, they are spatially related to each other in external primitive spatial relation. But the particularity and nature of a trope are identical to each other (they differ by a mere distinction of reason). It follows from this that the two simple entities stand to each other in internal and external relations due to the same metaphysical features of the relata. But how can this be? Further, by the transitivity of identity, the particularities of the two tropes stand in an internal exact similarity relation to each other and the two red natures are externally related to each other. But this is just incoherent. Moreover, since the nature and particularity of a trope are identical to each other, then the same dilemma of simplicity presents itself to Campbell: either tropes are reduced to properties as universals or to bare particulars. This is a matter of ontology, not epistemology. And the distinction of reason, along with the epistemological use of abstraction, different modes of counting, and so forth obscure the real metaphysical problems involved here. Campbell’s shift from using the distinction of reason between a trope’s nature and location to using it between that nature and particularity does not solve the problem which centers primarily on the simplicity of tropes and not the special role of place in their assay. And as I have already pointed out, his new view of place is less, not more compatible with the naturalist stance.
Someone may respond that a similar problem can be raised against universals. A universal has a nature (what it is, e.g. redness for the universal red) and universality. If a trope’s nature and particularity are identical for a trope, then the nature and universality of a universal are identical. Thus, red and green would be different by nature but identical by universality, which is incoherent. This response against the realist does not work. The realist need not take redness and greenness as simples.
Therefore, if a realist grounds the universality of redness and greenness in a property possessed by both – being universal – he can distinguish this property from the natures of red and green. But the same option is not available to the trope nominalist, given the fact that a trope’s nature and particularity differ only by a distinction of reason.
Here is an additional problem with the simplicity of a basic trope. A quality-instance, in this case a trope, must be extended in space or else it is not clear how a color nature can be present. A bare mathematical point of space cannot have a color nature at it. Unfortunately, something which is spread-out, i.e. extended, has to be a complex and not a simple entity because it has non-identical (perhaps, inseparable) parts at different spatial locations. In fact, it would seem that the very same nature (redness) occupies all the sub-regions throughout the extended formed volume. If an entity is to be extended (as an instance of color must be), it must be a complex entity (e.g. it contains non-identical inseparable parts) and a color trope, therefore, cannot be simple. G. Dawes Hicks raised this point against nominalist G. F. Stout when nominalism was just coming on the contemporary metaphysical scene earlier in this century:
What, for example, in his view, is really meant by saying that ‘qualities are in the same place as the things they qualify?’ In what sense can it be asserted that a visual sense-datum has a specific colour? Each infinitesimal portion of such a sense-datum is ex hypothesi no less numerically distinct from every other portion than one sense-datum is numerically distinct from another. If, then, the colour of one sense-datum A cannot be identical with the colour of another sense-datum B, how can the colour of the part x of the sense-datum A be said to be identical with the colour of the part y of the sense-datum A? In other words, where, in such a case, do we ever reach a concrete particular thing which can be rightly said to have a particular character?69
Campbell is sensitive to these difficulties, especially to problems with stating identity conditions for spatially or temporally extended tropes that seem to constitute the primary and secondary qualities of ordinary macro-objects. Put briefly, his solution is this.70 Campbell claims that basic tropes must be found that are partless, changeless, and unambiguous in their boundaries and he identifies five or six of them: space-time (construed as an absolute, non-relational trope), each of the four fundamental forces of physics taken as different space-filling fields that distribute some quantity across space-time to varying degrees, and, perhaps, consciousness understood as a pan-psychic field. By contrast, the manifest world is composed of the things and qualities, both primary and secondary, that are given to us in everyday experience. This world is composed of quasi-tropes: supervenient entities that are mere appearances. Such quasi-tropes are not truly real, and while they are not human inventions, nevertheless, there is a air of arbitrariness and conventionality about them regarding both their existence and identity conditions.71 A manifest trope is a quasi-trope, a pseudo-entity, a mere appearance. Such quasi-tropes are taken in thought via abstraction to be separable entities. They are chunks of field tropes treated as if they were distinct and independent realities. For Campbell, an ontologist does not need to specify identity conditions for such multiply-located quasi-entities so my objections raised against them are wide of the mark. By contrast, the five or six omni-located basic tropes have clear identity conditions and that is all that should be required of Campbell’s trope ontology.
Does Campbell’s refurbished trope ontology provide an adequate response to the types of criticisms I have been raising? Is it an adequate analytic ontology and speculative cosmology? For at least three reasons, I think not. First, it is not clear that his response allows room for the types of considerations that justify the acceptance of tropist qualities in the first place. As far as ontology is concerned, the existence of such entities is justified in terms of various metaphysical or linguistic arguments from the manifest world, especially the phenomena of predication, resemblance, and abstract reference. To the degree that Campbell’s basic tropes still exhibit these phenomena
The ontological status of properties 91
(e.g. most of his tropes have the property of being a field), then to that degree he has not solved difficulties in regard to them but merely re-located those difficulties. If basic tropes do not exhibit these phenomena, then why accept the existence of basic tropes? A decision to do so seems to me to be an arbitrary choice. For example, Campbell calls space (or space-time) a trope. But he also tells us that it can change and expand. Space also has various properties, e.g. elasticity, extension, shape, volume, various other geometrical and topological properties, and the trait (whatever it is) that distinguishes empty space from the material stuff or fields that fill it. Given that Campbell is an absolutist regarding space, what is it that justifies treating it as a trope as opposed to a substance with various properties? He does not tell us and I suspect that, given the above characterization, most philosophers would take it in substantival terms.
Second, Campbell admits that the basic forces of nature are normally thought of as being exerted by various bodies with properties in virtue of which the fields of nature are generated.72 On this view, the existence and properties of bodies are basic and fields are derivative. Campbell’s ontology reverses this order. But so far as I can tell, he does not offer a single argument in favor of his preference. Perhaps this omission could be excused if his claims about these matters were merely speculative explorations of an ontology already justified on other grounds. But that is not the case.
Campbell’s field ontology is foundational to his trope nominalism precisely because his exposition of fields is meant as a response to serious criticisms that are both damaging to the entire trope ontology and used to justify a realist view of qualities. In light of this fact, Campbell’s failure to justify his field ontology limits the acceptability of his view of tropes to those who agree with him about fields and, more importantly, leaves his analytic ontology resting on what amount to mere assertions in speculative cosmology. I doubt that this failure will convince many realists to adopt trope nominalism.
Third, Campbell’s exposition of manifest tropes is so ambiguous that it is hard to tell what their ontological status is. On the one hand he says that the meaning of statements about supervenient entities cannot be captured by reductive paraphrases using only terms referring to subvenient entities, but for two reasons this seems to have no ontological implications for Campbell.73 First, he explicitly says that supervenient entities are “pseudo-additions” and not real additions to our ontology; they provide cases that expand our descriptive resources without involving a commitment to the realities described. Second, for Campbell causal power is the mark of being and supervenient entities lack this power on his view.74 The same ambiguity characterizes his statements about dependent entities like relations, manifest tropes, and subregions of space or fields. On the one hand, he says that relational facts exist but he also claims that they are non-existent supervenient entities. Such entities are not real because they are dependent entities. He also excepts foundationism, the view that all
relational facts can be accounted for in terms of non-relational foundational facts with no further ontology beyond the latter. So while talk about relations cannot be paraphrased away, relations do not exist.75 Manifest tropes and other subregions of space or fields are quasi- or pseudo-entities that are not mere inventions, yet they are not distinct items and do not give us additional entities in our ontology.76 They are not illusions, yet they are mere appearances that are treated as if they are independent entities by an act of abstraction and they do not really exist.
In my view, this confusion is the result of an inadequate set of distinctions. In the language of Francis Suarez, Campbell has no modal distinction that allows him to affirm the existence of dependent entities, e.g. Husserlian moments.77 For Suarez, a modal distinction intervenes between an entity and its mode. A mode is a dependent, inseparable, genuinely distinct entity from that of which it is a mode. If a modal distinction obtains between two entities A and B (where B is a mode), there is a non-identity between A and B as well as inseparability in this sense: A can exist without B
but not vice versa. For Campbell, only fundamental, separable entities exist. Talk about dependent entities may be irreducible but this has no ontological implications.
Two inseparable “entities” differ by a mere distinction of reason. If I am right about this, than Campbell’s ontology amounts to a denial of the existence of macro-objects and their qualities which I and, I suspect, others will find implausible. It may be that he accepts some sort of modes-of-being view in which dependent entities like manifest tropes are appearances in the sense that they have some sort of existence between non-being and full being. Now the difficulties with modes of being views of existence are widely recognized and I cannot discuss them here. Suffice it to say that if this is Campbell’s position, he has not made that clear. But in any case, whether he denies or accepts a diminished sort of reality for manifest tropes, he would still face the difficulties I have been raising in the last few paragraphs. If manifest tropes don’t exist, his basic ones still exhibit predication, resemblance, and abstract reference and the debate about nominalism will be located there. If they have enough existence to be colored, extended, and so forth, then problems of identity conditions for manifest tropes will still be relevant.
The simplicity of Campbell’s revised tropes presents another difficulty for him: Campbell is inconsistent and, in fact, equivocal in his description of tropes. The fact that a trope is a simple entity that must simultaneously play two roles (have a qualitative nature that stands in the internal relation of exact similarity with other tropes like it and individuate) in a basic, primitive way is not a clear or natural way of thinking about quality-instances, and one must try hard to keep the view in one’s mind without unwittingly treating the quality-instance as would a realist. This is why I think Campbell’s own description of his position moves back and forth between what are apparently two inconsistent positions. I will mention two areas where this happens.
First, he repeatedly tells us that tropes have their natures.78 On the other hand, he
The ontological status of properties 93
also says that tropes are identical to their natures.79 Which is it? The having relation is not the same as the identity relation (the latter is reflexive and symmetrical, the former is not) and it is the simplicity tension that causes this lack of clarity. Second, he says that the exact similarity between two tropes is an internal one grounded in the trope’s natures and he also claims that this relation is primitive and unanalyzable.80 The former view tends to treat tropes as complex entities that have natures as constituents and the latter tends to view tropes as simples. Of course, the realist will accept the former horn and claim that quality-instances have natures that ground the derived relation of similarity between two instances with the same nature. Campbell clearly needs the second horn – two tropes of the “same” kind are simples which stand in a primitive similarity relation – and it is this option that suffers from the problems I have been raising.
Campbell is aware of these problems about simplicity and so far as I can tell, he has two basic responses to them. First, he offers a sort of tu quoque argument against me and others who hold to bare particulars as individuators. He says that even though tropes qua simples must simultaneously sustain two different roles, the same problem attaches to bare particulars: they are simple and must simultaneously individuate and particularize.81 According to Campbell, a particularizer gives the object in which it occurs a particular reality, as a definite object. But a specific object is not just an object, but this very object. And therefore, an object needs an individuator in addition to a particularizer, and bare particulars must sustain both roles.
Unfortunately, there is no genuine metaphysical distinction between particularity and individuation in this context. In fact, Campbell’s treatment of them closely parallels Thomas Aquinas’ discussion of undesignated and designated matter which, as Aquinas pointed out, differ by a mere distinction of reason with undesignated matter (e.g. flesh and bone in general) simply being a more general, abstract way of attending to designated matter (e.g. this flesh and bone). In Campbell’s sense, particularity is not a different metaphysical entity from individuation. The former differs from the latter only in the level of epistemic abstraction one uses in attending to an object. Thus, there are not two metaphysical facts that need grounding as there are with so-called particularized natures (e.g. red and 1). If Campbell thinks these are, indeed, different metaphysical facts, then he has not given sufficient arguments for believing this to be the case. As far as I can tell, Campbell merely asserts his views in this regard without arguing for them. By contrast, I have offered reasons for thinking that a trope’s nature and particularity are different metaphysical facts. Campbell admits that my arguments were successful against his identification of a trope’s nature with its location/formed volume, but as I have shown, these same arguments apply to the particularity and nature of a trope.
Second, Campbell uses an illustration to argue that there is no problem with cases where a simple entity plays two roles: “A point is as simple as you can get. Yet points are
both different from all other points and also, for example, 10 mm distant from some others. That x is involved in more than one sort of fact does not show that x is complex.”82 But this example reveals a deep misunderstanding of the problem with tropes. A point is an individuated entity and this is so either because it has an individuator in it or else because it is identical to a bare individual. Either way, its particularity is a feature intrinsic to its being. But a point’s spatial relation to another point is an external relation and is not intrinsic to the point. If that point were the only entity to exist, it would be an individual point but it would not be spatially related to anything else. Moreover, if a point is to the left of x but to the right of y, these two facts are not grounded in two different, intrinsic features of the point itself, but in two relations external to the point. By contrast, red1 is intrinsically individual and red. If it were the only entity in existence it would be both red and particular. Thus, the ground for these two facts about red1 must be sought in its intrinsic being and not in external relations it sustains to other things. It is because red1 is red that it exactly resembles other red tropes and it is because it is a particular that it is individuated from other entities. So Campbell’s analogy breaks down.
We have seen that the role of location or formed volume in Campbell’s nominalism, along with the simplicity of tropes, are serious difficulties in his understanding of qualities. A third problem area for Campbell involves abstract reference. Consider again statements like “red is a color” or “necessarily, red is a color.” Now it is generally recognized that arguments based on inadequate paraphrases of such statements have been decisive against extreme nominalism. To see why, consider the scattered location, L, of all red things. An extreme nominalist might paraphrase the statements just mentioned as follows: “red things are colored things” or “necessarily, red things are colored things.” Unfortunately, this paraphrase will not work, because red things are extended things or L located things (and this is necessarily the case for naturalist advocates of extreme nominalism), but redness is not extension or L location. The problem is that concrete particulars are complex entities in a way not true of first and second order universals (redness, coloredness).
In my view, the phenomenon of abstract reference is equally telling against nominalism. A nominalist would paraphrase the relevant sentences as “reds are colors” or “necessarily, reds are colors.” But this will not work. In Campbell’s earlier view, the counter example could read (with or without the modal operator) “reds are formed volumes” or “reds are L located” (where L is the scattered location of all red tropes). But redness is not formed volume or L location. In Campbell’s current position, the counter example could read “reds are particulars,” but redness is not particularity. Again, the problem here is not essentially one about location but complexity: in spite of what Campbell tells us, tropes exhibit a complexity (a nature and individuation) that is not like what is the case for first and second order universals.
Campbell’s response to a closely related problem provides an insight as to how he
The ontological status of properties 95
would handle this difficulty.83 Campbell’s strategy involves what D. C. Williams called painless realism. Consider a case where an interior decorator says he or she used four colors to decorate a house. Obviously, four tropes is not what is meant by this assertion and, in fact, the statement seems to view colors as universals. Nominalists, says Campbell, must make sense of cases where we appear to assert that matching tropes have something in common when, according to nominalism that cannot be the case.
Following Williams, Campbell makes a distinction between inherent and adherent properties. Inherent properties are the qualitative, monadic characteristics (e.g.
being red) intrinsic to a specific item, i.e. that a complex concrete item can have in its own right. Adherent or “external” characteristics are relational properties of an object (e.g. being the heaviest man in the room). Now particulars can be “alike” in inherent (matching) characteristics and they can be distinguished by different adherent features, e.g. their compresent relations. The difference between a case (trope) and a kind (a “universal”) is not a difference in ontological category, but rather, a difference in the rule used for counting. If we look at six red tropes which match in inherent but not adherent features, if you count one for each case, you treat the tropes as particulars. When, without regard to adherent characters, you only count more than one where there is a difference in inherent traits you take note of, you identify all occurrences of matching tropes, i.e. you treat them in a way that likens them to a realist universal. Applied to abstract reference, reds can be viewed as colors or as particulars, depending on what rule you use for counting.
I think painless realism is smoke and mirrors. For one thing, it is simply a label for the problem to be solved and is not a solution to the problem. The issue is not that we can notice a set of entities in a natural class and either count them as one entity on the basis of their common nature or as many entities according to their individuality. That is just what at least one version of the problem of the one and the many has always been about. The real issue is what grounds this activity and Campbell has no answer to this question. Moreover, his appeal to different rules of counting is just another example of the inevitable strategy of nominalists of reducing metaphysical issues to epistemological ones (e.g. try as they may, nominalist attempts to treat “abstract” as a metaphysical notion turn out, on closer inspection, to treat it an epistemic one).
Indeed, they have no alternative since, for them, tropes are simples that sustain two functions, and these two functions will differ by the epistemic distinction of reason. In fact, Campbell’s (and William’s) appropriation of epistemology to solve a metaphysical problem obfuscates an obvious difficulty with the appeal to adherent features. If the problem of individuation is basically an epistemological one, namely, the task of counting, then it may well be that external relations are what we use to single out distinct entities in order to count them. But if the problem is metaphysical, adherent, external relations like compresence will not do because such relations presuppose and therefore cannot constitute the individuation of their relata.
Elsewhere, I have raised other arguments against Campbell’s earlier views and
some of my points are equally relevant to his current position.84 Rather than survey those problems here, I will content myself with the conclusion that nominalism, as exemplified by its most articulate paradigm case, has been shown to have enough serious difficulties to warrant a rejection of it. In spite of its subtle nuances, Campbell’s refurbished nominalism is no more preferable than the position he himself has abandoned. Moreover, it is a less plausible option for a naturalist than his earlier position.
D. M. Armstrong
Armstrong is clear that a naturalist must find unacceptable uninstantiated universals, those classes of universals that are not the natural properties and relations characterizing the causal, physical objects of physics, and a priori knowledge. I shall not examine his case for these claims here. On the other hand, he believes that all versions of nominalism are unacceptable and that a robust version of realism about universals can be integrated with naturalism and physicalism. His own form of
“realism” is just such an attempted integration and his model centers on explicating a view of (1) the nature of universals themselves; (2) the predication “relation”; and (3) individuation consistent with naturalism.
Regarding universals, the only way Armstrong can sustain realism and deny that universals are outside space and time is to reject the axiom of localization: no entity whatsoever can exist at different spatial locations at once or at interrupted time intervals. Armstrong rejects this axiom and claims that universals have multiple location and exist wherever the particulars having them are located. Thus, universals are entities that are spatially located at the place of the particulars that have them and they are fully present at each exemplified location.
Armstrong’s treatment of predication and individuation are closely related.
Universality and particularity are, he says, inseparable aspects of all existence, they are neither reducible nor related to each other and, though distinct, their union is closer than a relation. The distinction between universality and particularity is similar to Duns Scotus’ formal distinction and is like the way that shape and size are united in a particular.85
This leads him to distinguish a thick from a thin particular. A thick particular is a state of affairs, e.g. A’s being F, and as such, it is a particular along with its properties.
The particular “enfolds” its properties in the sense that they are spatially located where the thick particular is. In the statement “this is hot,” the word “this” refers to a thick particular and says that hotness is among its properties. The thin particular is the particular considered in abstraction from all its properties. It is not a thing per se, but amounts to bare numerical difference or thisness, the individuating factor that makes the thick particular more than just a bundle of universals.
The ontological status of properties 97
Armstrong’s view of individuation needs to be elaborated a bit more before it can be clear just what he means by it. He rejects transcendental realism in favor of immanent realism. We will see shortly that he is confused about the real nature of this distinction, but as he understands it, transcendental realism has universals standing entirely apart from their instances and immanent realism depicts them as being in the being of their instances, in Armstrong’s position, being at the very place of their instances and being “enfolded” by those instances. Now Armstrong wants to reject what he calls relational immanent realism, roughly, the view that a thick particular is a complex state of affairs composed of a universal, a bare particular or substratum, and a relational tie of predication. Armstrong’s basic argument against relational immanent realism is the claim that it falls prey to Bradley’s famous regress argument against relations: if a substratum requires a relation to be bound to a property, then the substratum will require a relation to be bound to the first relation, and so on.
Because of this, Armstrong claims that what is needed is a non-relational immanent realism. And his account of thick and thin particulars is just such an account. Says Armstrong, “. . . this version of Immanent Realism [is one] which distinguishes the particularity from the properties of a particular, while denying that the two aspects are related.”86 In sum, particularity and universality differ by a formal distinction, they are indistinguishable though inseparable, and “they” are too intimately conjoined to be related.87
Is Armstrong’s version of realism a viable way to embrace properties within the constraints of naturalism? I do not think so and to see why, I want to focus on certain issues in Armstrong’s view of properties as universals and on issues in particularity and individuation. Let us begin with properties.
First, is it not at all clear that his understanding of properties is, even if acceptable on other grounds, compatible with naturalism. In my view, naturalists should accept the axiom of localization and I think most naturalists would agree with me in this regard. Paradigm case naturalist entities would appear to obey what I shall call the Principle of In toto Location (PIL):
PIL:
If entity e exists in toto at location L, then both (1) if x has proper parts, they overlap with and only with sub-regions of L, and (2) neither e nor any of its proper parts overlap with some other location P such that P is not identical to L or a sub-region of L.
If a necessary condition for being an entity in a naturalist ontology is conformity to PIL, then Armstrong’s universals are not natural entities. It is just not clear how entities that are fully, completely present at each embodiment are relevantly similar to other entities embraced by naturalism to yield a consistent, economical ontology.
Besides considerations of PIL, most naturalists would find it difficult to believe that an entity could simultaneously be in motion and at rest if such a view is even coherent.
But Armstrong’s properties are precisely such entities. If stationary particular x has F
and moving particular y has F, then since F itself is at the location of the particulars, F
is both moving and at rest. In my opinion, his universals are just as unacceptable for naturalism as is their construal as abstract objects.
It may be objected that only particulars must obey the laws of motion and properties are under no such constraint. However, to his credit, Armstrong realizes that a consistent naturalist is most reasonably taken to require that universals resemble particulars in having spatio-temporal location. But, then, why not make the same requirement regarding the prohibition of being simultaneously in motion and at rest? If Armstrong relaxes the second requirement, why not relax the first one? At the very least, Armstrong owes us an account of this problem and to my knowledge, he has not offered one.
Second, it is not clear that properties are located at the place of their instances.88
For one thing, many properties (being even, triangularity) and relations (the musical relations between sounds in a music chord, being the father of) are such that it is just not clear what it would mean to take them as spatial, natural properties or relations.
Armstrong could reject such entities on the grounds that they do not have causal powers and, thus, they have no being. But I suspect that many philosophers would find it more reasonable to reject causality as the mark of being than to eschew these entities.
Moreover, if a red ball is on the table and Armstrong claims that redness is spatially located on the table where the ball is, a reductive paraphrase is available to an advocate of properties as abstract objects for such a situation: the ball is on the table and redness is “in” the ball so the ball is located but the redness is not. The realist paraphrase allows one to say everything that needs to be said as far as spatial location is concerned, and the paraphrase includes a spatial particular (the ball) and two non-spatial entities (redness, and the way it is “in” the ball). Further, the instance of redness (see below) may be spatial, but the redness “in” that instance need not be. Put differently, every spatial predicate relevant to the state of affairs in our purview can be ascribed to either the ball or the instance of red (Husserl called it a moment)89. But redness itself is not spatial. When one attends to something spatial, one is attending to an instance of redness (or the ball itself), and not to redness. Redness is a color, it has hue, intensity, saturation, it resembles orange more than it resembles blue, and so forth. All the facts we know about red itself are not spatial facts. Spatial predicates are true of instances of redness, not redness itself.
Third, is spite of what Armstrong claims, statements like (1) “red is a color” or (2)
“red resembles orange more than it resembles blue” express necessary truths.
Traditionally, these have been understood as truths of reason, synthetic a priori truths that are true throughout possible worlds, including those with no red particulars. In general, for some universal F, there is a possible world w1 such that F exists in w1 but
The ontological status of properties 99
no objects exemplify F in w1.90 Propositions like (1) and (2) are true throughout possible worlds and they are grounded in a view of universals as necessary beings whose existence is independent of the existence of particulars exemplifying them, and in genus/species or internal relations among universals. For example, (1) expresses a genus/species relation between a determinable (being colored) and a determinate (being red). And (2) expresses internal relations that obtain in a quality order consisting of the various determinates of a higher order, determinable universal. And arguably, both the truth and modality of these propositions are knowable a priori.
As far as I can tell, Armstrong has two main arguments against such universals. As we have already seen, the first one turns on his causal criterion of being. I find this criterion both inadequate and question-begging in this context. The second one represents a serious misunderstanding of what it means to say that some universal F is a transcendental universal. Armstrong seems to think that an advocate of transcendental universals must take such entities as “standing apart” from the spatiotemporal world, i.e., properties stand apart and exist separated from their instances in a different realm. The properties do not enter into the being of their instances.91
This leads Armstrong to embrace what he calls the Principle of Instantiation: for each property P, there exists (not necessarily now) a particular, x, such that Px.
Unfortunately, Armstrong’s statements in this regard are confused. For one thing, the debate between Platonists and Aristotelians about the relationship between properties and their instances is not a debate about the “location” of those properties (e.g. either in Plato’s heaven or spatially contained in their instances), but rather, about the ontological independence of properties from their instances. Second, I have already shown how someone who sees universals as transcendental, abstract entities can still take it to be the case that universals enter into the very being of their constituents. Armstrong seems to think that the reading of Plato which turns properties into perfect particulars that are merely copied in their instances, or something of the sort, is the only view available for the transcendental realist, but this is just mistaken. The issue here turns on the way a property is “in” its instance.
Third, Armstrong’s statement of the principle of instantiation reveals what I believe is a problem with his position. He claims that a property can exist when is not currently being instantiated as long as it will be instantiated at some time. Presumably, this could allow him to embrace properties ontologically (not temporally) prior to the Big Bang or properties (e.g. certain colors) that have only recently been exemplified.
But how can the current existence of some entity depend upon the fact that in the future it will be exemplified in space and time? And how are we to conceive of the current mode of existence of such an entity? How does it currently fit into a naturalist ontology? Moreover, even if we grant the principle of instantiation, it does not follow that properties are spatially located at the place of their instances, but only that their
existence depends on that of the instances themselves. It could still be the case that properties are abstract entities “in” their instances in a non-spatial way. I conclude, therefore, that Armstrong’s view of transcendental universals is historically inaccurate, a caricature of the position itself, and involved in problems in its own right.
For these reasons, I do not think he has made his case against the existence of transcendental universals.
When we turn from properties to his account of particularity and individuation, things get no better. I, for one, cannot understand what Armstrong means here. If he takes the distinction between universality and particularity to be a formal distinction between “entities” that do not need to be related, then this amounts to the claim that they are identical, as we saw in our discussion of Campbell’s tropes. If, on the other hand, he says that the thin particular (entity a) is a constituent in a thick particular (entity A) that is non-identical to a property/properties of A, then his view seems to reduce to a form of immanent realism with bare particulars. In this case, a thin particular turns out to be a bare particular that individuates, and a thick particular is a particular in virtue of its possession of the thin particular. As we have seen, this is the victory of particularity. But we have also seen that the victory of particularity is paralleled by the victory of spatiality, so on this latter reading of Armstrong, it is easy to see the universal in A and the predication relation between that universal and the thin particular as non-spatio-temporal entities. If this is the most satisfactory analysis of a thick particular, then given Armstrong’s own statements about what is required of a naturalist analysis of properties, it counts as a refutation of naturalism.
On the other hand, it may be that Armstrong’s earlier views on particularity and individuation collapse into nominalism, and it is no accident that in his later writings, he has opened up considerably to nominalism compared to his earlier writings.92 In one place he says, “So provided you abandon uninstantiated universals (good riddance, I say), and provided Universals theorists and Trope theorists coordinate their views on just what properties and relations the world contains, it is easy to pass back and forth between the theories.”93 Indeed. Armstrong’s claims that universals are spatio-temporally located at the place of their instances and his inadequate view of individuation are virtually indistinguishable from the trope nominalism of Campbell which, as we have already seen, is riddled with problems.
However, the various problems with nominalism have pushed Armstrong towards bare substrata though a failure to make a distinction in two senses of “being bare”
obscures Armstrong’s views on this subject. Advocates of bare particulars distinguish two different senses of being bare along with two different ways something can have a property. In one sense, an entity is bare if and only if it has no properties in any sense.
Now bare particulars are not bare in this sense. They do not exist unless they possess properties.
There is another sense of bare, however, that is true of bare particulars. To understand this, consider the way a classic Aristotelian substance has a property, say,
The ontological status of properties 101
some dog Fido’s being brown. On this view, Fido is a substance constituted by an essence which contains a diversity of capacities internal to, within the being of Fido as a substance. These capacities are potentialities to exemplify properties or to have parts that exemplify properties. The capacities are grounds for the properties like brownness that Fido comes to have. When a substance has a property, that property is
“seated within” and, thus, an expression of the “inner nature” of the substance itself.
Thus, Richard Connell is correct to distinguish the way substances and bare substrata have properties when he claims that properties are not simply tied to substances, but rather “rooted in…and caused by the substance.”94
By contrast, bare particulars are simple and properties are linked or tied to them.
This tie is asymmetrical in that some bare particular x has a property F and F is had by x. A bare particular is called “bare,” not because it comes without properties, but in order to distinguish it from other particulars like substances and to distinguish the way it has a property (F is tied to x) from the way, say, a substance has a property (F is rooted within x). Since bare particulars are simples, there is no internal differentiation within them. When a property is exemplified by a bare particular, it is modified by being tied to that particular. Thus, bare particulars have a number of properties, e.g. being red, and they have some properties necessarily, e.g. particularity, in the sense that a bare particular can exist only if it has certain properties tied to it. Now, this fact about bare particulars neither makes them identical to their properties nor does it entail that properties are constituents within a bare particular. Just because a man never comes out of his house naked, it does not follow that he is his clothes or that they compose him as constituents.
Armstrong’s failure to distinguish these two senses of “being bare” have lead to confusion in attempts to interpret his account of individuation because he explicitly rejects yet implicitly seems to accept bare particulars. He rejects bare particulars for the following reason: “A particular that existed outside states of affairs would not be clothed in any properties or relations. It may be called a bare particular. If the world is to be a world of states of affairs we must [reject] Bare Particulars.”95 Yet as we have seen earlier, Armstrong makes a distinction between a thick particular (a’s being F) and a thin particular (a). He calls the latter a Lockean substratum and accepts it as an individuator. For Armstrong, the particularity of a normal (thick) particular is an irreducible, unanalyzable feature of normal particulars. For him, particulars qua particulars do not differ in nature, but in virtue of bare, numerical difference, bare particularity, and he explicitly identifies this individuator (the thin particular) with a Lockean substratum.96 Armstrong rejects bare particulars in the first sense but not the second, and given the fact that advocates of bare particulars embrace the second sense, his view is a bare particular position.97
There is a lesson to be learned from Armstrong’s developing position. If one accepts a realist construal of properties, then one must also embrace some type of
individuator that is not a normal property (e.g. an impure property) or is not a property at all, or else the position collapses into nominalism. In his most recent treatment of the subject, Armstrong has acknowledged this point.98 He continues to embrace universals, he explicitly rejects the individuative adequacy of spatio-temporal location, tropes and impure properties, and he takes it that “bare numerical difference” is what grounds the particularity of thick particulars. But on this reading of Armstrong, it is hard to see how his account is superior to a traditional realist view of individuation and particularity. In fact, it is hard to see how his position avoids collapsing into the traditional realist account. If I am right about this, then, in the absence of a better realist attempt to coordinate a joint commitment to universals and naturalism, and in view of the difficulties with both extreme nominalism and nominalism, the traditional historical position on the matter seems to be the correct one: the existence and nature of properties refutes naturalism as a world-view.
At this juncture, let us recall Johanna Seibt’s assertion that “since the basic relation of a Platonist theory of predication, i.e. exemplification, cannot be defined in naturalist terms, a nominalist theory of predication proves to be ‘. . . the very foundation of a naturalist ontology’.”99 Among other things, my critique of Campbell and Armstrong has focused on their analyses of predication. I have tried to show that their accounts fail to be adequate as naturalist theories. I have also claimed that they are intrinsically problematic in their own right. To reinforce this second claim, it may be helpful to compare my traditional realist account of properties, predication, and property-instances mentioned earlier with the views of Armstrong and Campbell. In my view, there are at least four advantages to my traditional realist assay of an property-instance when it is compared to views like those of Armstrong or Campbell.
First, a universal F-ness must be brought together with its instances in such a way that they can be called F. Redness must be “tied to” its instances so as to make them red and not bare entities (or green for that matter). My traditional realist assay of property-instances explains why red1 is red and not bare (or green). Red1 is red because the universal, redness, is in it as one of its constituents and redness is the color of red1.
Second, my assay of red1 clarifies the relation between redness and red1 in a way that is not found in views that liken this relationship to a set membership relation (Campbell) or a “being wholly spatially located at the same place as” relation (Armstrong). These treatments of this relation leave unclear precisely how the
“universal” is really related to its instances. By contrast, the traditional realist sketch presents the universal as a constituent in its instances and, thus, treats this relation along the lines of part/whole relations. This part/whole relation involves a more deeply grounded relation between redness and the individuator in red1. Admittedly, this relation is basic and undefined. But it can be viewed as a tie or a link between two entities. Moreover, the response to the claim that this version of “immanent relational realism” falls prey to Bradley’s regress argument is one that Armstrong himself admits
The ontological status of properties 103
is hard to refute: it is a unique feature of relations (including the ones included in our assay of qualities and quality instances) that they can relate relata without needing other entities to relate them to those relata.
Third, my traditional realist account presents red1, as a complex entity with two constituents that serve to ground both the universality and the particularity of red1. By contrast, instances for Campbell and, possibly, Armstrong (if his formal distinction is merely a distinction of reason) are simple entities without further entities in them. In a way that should now be familiar, this move leaves the abstract particular as a universal (red), a bare particular (1), or some unclarified entity that somehow sustains both functions without grounding either.
Finally, my account explains how it is that the universal transcends its instances.
The universal is indifferent to any particular instance (and as a Platonist, I would add all its instances) since the universal can be a constituent in many instances through the non-spatio-temporal, inhomogeneous nexus of exemplification. When redness has red1, as one of its instances, this is due to the fact that some entity (a bare particular) outside the nature of redness, has entered into an exemplification relation with redness. Something happens to redness, namely, it is modified and becomes exemplified. Thus, redness transcends its instances. If a universal is just a set of tropes (Campbell) then the universal cannot transcend its instances because the identity conditions for sets are constituted by the members of those sets. Change the membership and you change the set. Armstrong’s views are more clear than Campbell’s in this regard because he can claim that a universal transcends a specific instance in that it is wholly present at the same location as its other instances. However, this solution goes against the axiom of localization and for those who accept that axiom, Armstrong’s model of transcendence will be problematic.
I have tried to show that if properties exist and are what traditional realists take them to be, then a widely accepted contemporary version of naturalism is false. It would seem, then, that one must either accept the traditional realist view and abandon naturalism, try to reconcile the two, offer a more adequate revised view of properties consistent with naturalism, or accept naturalism and extreme nominalism. I have attempted to offer at least some reasons why I think the first alternative is the best one to take. Whether or not my case has been persuasive, one thing seems clear: the ontological status of properties will continue to be a central issue of debate for those who have taken the naturalistic turn.
Notes
1 It is, of course, always dangerous to claim that there is a single, traditional view of some perennial issue in philosophy and I am making no such claim. Clearly, there have been different nuances given to realist views of properties as universals in the history of
philosophy. Still, I think it is safe to say that most would identify the traditional realist view of properties in the minimal way I have characterized it.
2 Howard Robinson, Matter and Sense (Cambridge: Cambridge University Press, 1982), 50. For more on this, see J P. Moreland, “How To Be a Nominalist in Realist Clothing,” Grazer Philosophische Studien 39 (1991): 75–101.
3 See Reinhardt Grossmann, The Existence of the World (London: Routledge, 1992), 1–45; cf. my review of Grossmann in Mind 102 (July 1993): 407–10.
4 Wilfrid Sellars, Naturalism and Ontology (Atascadero, Calif.: Ridgeview Pub. Co., 1979), 109.
5 Ibid., 47. Cf. Wilfrid Sellars, “Towards a Theory of Predication,” in How Things Are, ed. by James Bogen, James E. McGuire (Dordrecht, Holland: D. Reidel, 1985), 285–322.
6 Colin McGinn, Mental Content (Oxford: Basil Blackwell, 1989), 13. C f. Jeffrey Poland, Physicalism (Oxford: Clarendon, 1994), especially pp. 10–44, 226–32, 307–12.
7 For a helpful discussion of this dialectic with a good list of resources, including the contributions by Field and Quine, see Penelope Maddy, Realism in Mathematics (Oxford: Clarendon, 1990).
8 See Maddy, Realism in Mathematics.
9 These are nicely offered in D. M. Armstrong, Universals & Scientific Realism Vol. I: Nominalism
& Realism (Cambridge: Cambridge University Press, 1978), chapters 2–6; Michael J. Loux, Substance and Attribute (Dordrecht, Holland: D. Reidel, 1978), chapters 2–4.
10 D. C. Williams, “On the Elements of Being: I,” Review of Metaphysics 7 (September 1953): 3–
18; Keith C ampbell, Metaphysics: An Introduction (Encino, Calif: Dickenson Publishing Company, Inc., 1976), 206–19.
11 Charles Landesman, “Abstract Particulars,” Philosophy and Phenomenological Research 33
(March 1976): 323–37.
12 Gustav Bergmann, Realism (Madison: University of Wisconsin Press, 1964), 25.
13 Nicholas Wolterstorff, On Universals (Chicago: The University of Chicago Press, 1970), 130–
9.
14 Woltersdorff, On Universals, 133–4.
15 Gareth B. Matthews and S. Marc Cohen, “The One and the Many,” The Review of Metaphysics 21 (June 1968): 630–55.
16 J. R. Jones, “What Do We Mean by an ‘Instance’?,” Analysis 11 (October 1950): 1–9.
17 Edmund Husserl, Logical Investigations, 2 vols., trans. J. N. Findlay (London: Routledge & Kegan Paul, 1970), especially volume I, 408–10; II, 427, 436, 440, 453, 467–8.
18 See Grossmann, The Existence of the World, 1–45, 51–62.
19 See J. P. Moreland, “Was Husserl a Nominalist?,” Philosophy and Phenomenological Research 49
(June 1989): 661–74.
20 Johanna Seibt, Properties as Processes: A Synoptic Study of Wilfrid Sellars’ Nominalism (Atascadero, Calif.: Ridgeview, 1990), 184.
21 Cf. Alex Rosenberg, “A Field Guide to Recent Species of Naturalism,” British Journal for the Philosophy of Science 47 (1996): 1–29.
22 Wilfrid Sellars, Science, Perception, and Reality (London: Routledge & Kegan Paul, 1963), 173.
23 Steven J. Wagner, Richard Warner, Naturalism: A Critical Appraisal (Notre Dame: University of Notre Dame Press, 1993), 1.
24 Roy Bhaskar, The Possibility of Naturalism (New Jersey: Humanities Press, 1979), 3.
25 John Searle, The Rediscovery of the Mind (Cambridge, Mass.: MIT Press, 1992), 11.
26 David Papineau, Philosophical Naturalism (Oxford: Blackwell, 1993), 3.
The ontological status of properties 105
27 Robert C. Stalnaker, Inquiry (Cambridge, Mass.: MIT Press, 1984), 6.
28 John F. Post, Metaphysics (New York.: Paragon House, 1991), 121.
29 We should make a distinction between strong and weak externalism here. Focusing on justification, according to strong externalism, no factors that contribute to justification are internal to the agent. According to weak externalism, at least some factors that contribute to justification must be external to the agent. I am using the term externalism to stand for strong externalism because the weak version is not an option for consistent naturalists for the reasons just mentioned.
30 Searle, The Rediscovery of the Mind, 83–93.
31 E. Mayr, Populations, Species, and Evolution (Cambridge, Mass.: Harvard University Press, 1970), 4.
32 David Hull, The Metaphysics of Evolution (Albany, N.Y.: State University of New York, 1989), 74–
5.
33 Bruce Aune, Metaphysics: The Elements (Minneapolis: University of Minnesota Press, 1985), 35.
34 D. M. Armstrong, “Naturalism, Materialism, and First Philosophy,” Philosophia 8 (1978): 263.
Cf. Universals and Scientific Realism Volume I: Nominalism & Realism (Cambridge: Cambridge University Press, 1978), 126–35. Subsequently, Armstrong has modified and weakened this formulation of his criterion of being: Everything that exists makes a difference to the causal powers of something. See A World of States of Affairs (Cambridge: Cambridge University Press, 1997), 41–3.
35 Daniel Dennett, Elbow Room (Cambridge, Mass.: MIT Press, 1984), 76.
36 Keith Campbell, Abstract Particulars (Oxford: Basil Blackwell, 1990), 172.
37 Jaegwon Kim, “Mental Causation and Two Conceptions of Mental Properties,” unpublished paper delivered at the American Philosophical Association Eastern Division Meeting, Atlanta, Georgia, December 27–30, 1993, p. 23.
38 See J. P. Moreland, “Searle’s Biological Naturalism and the Argument from Consciousness,”
Faith and Philosophy 15 (January, 1998): 68–91; and Moreland, “Should a Naturalist Be a Supervenient Physicalist?,” Metaphilosophy 29 (January/April 1998): 35–57.
39 D. M. Armstrong, Universals and Scientific Realism Volume I: Nominalism & Realism (Cambridge: Cambridge University Press, 1978), 130.
40 D. M. Armstrong, “Naturalism: Materialism and First Philosophy,” Philosophia 8 (1978): 262.
41 D. M. Armstrong, “Can A Naturalist Believe in Universals?,” in Science in Reflection, ed. by Edna Ullmann-Margalit (Boston: Kluwer Academic Publishers, 1988), 111–12; Universals & Scientific Realism Volume II: A Theory of Universals (Cambridge: Cambridge University Press, 1978), 84–8.
42 Laurence BonJour, In Defense of Pure Reason (Cambridge: Cambridge University Press, 1998), 153–86, especially 156–61.
43 Cf. Keith Campbell, Abstract Particulars.
44 See J. P. Moreland, “Keith Campbell and the Trope View of Predication,” Australasian Journal of Philosophy 67 (December 1989): 379–93; cf. J. P. Moreland, “A C ritique of C ampbell’s Refurbished Nominalism,” The Southern Journal of Philosophy 35 (Summer 1997): 225–46.
45 Cf. Keith Campbell, “Abstract Particulars and the Philosophy of Mind,” Australasian Journal of Philosophy 61 (June 1983): 129–41; “The Metaphysics of Abstract Particulars” in Midwest Studies in Philosophy Volume VI: The Foundations of Analytic Philosophy ed. by Peter A. French, Theodore E. Uehling, and Howard K. Wettstein (Minneapolis: The University of Minnesota Press, 1981), 477–88; and Campbell, Metaphysics: An Introduction (Encino, Ca.: Dickenson
Publishing Co., 1976), 206–19.
46 Campbell, Metaphysics, 213.
47 Campbell, “Abstract Particulars,” 141.
48 Ibid., 130.
49 J. P. Moreland, Universals, Qualities, and Quality-Instances (Lanham, Md.: University Press of America, 1985), 8–11, 41–5, 109–33. It is possible to identify exact similarity as identical to the entire set of exactly resembling tropes, or to identify different “kinds” of exact similarity relations with different sets of exactly resembling tropes (e.g. red tropes, sweet tropes), or to view exact similarity as a genuinely relational entity that is either a dyadic or an n-adic relation holding among n tropes. I will only consider the dyadic view because it is the most likely one to ascribe to Campbell and, in any case, it is the most plausible.
50 Bergmann, Realism, 54.
51 Armstrong, Universals and Scientific Realism, 172.
52 D. M. Armstrong, Universals: An Opinionated Introduction (Boulder: Westview Press, 1989), 43–
4, 55, 100.
53 See D. C. Williams, “On the Elements of Being: I,” Review of Metaphysics 7 (September 1953): 3–18; “On the Elements of Being: II,” Review of Metaphysics 7 (December 1953): 171–92;
“Necessary Facts,” Review of Metaphysics 16 (June 1963): 601–26.
54 Campbell, Metaphysics, 216–17.
55 Campbell, “The Metaphysics of Abstract Particulars,” 484.
56 D. M. Armstrong is one of the few naturalists to see and note this point. See, “Can A Naturalist Believe in Universals?” in Science in Reflection, ed. by Edna Ullmann-Margalit (Boston: Kluwer Academic Publishers, 1988), 111–12; Universals & Scientific Realism Volume II: A Theory of Universals (Cambridge: Cambridge University Press, 1978), 84–8. There are two main reasons naturalists must eliminate or reduce internal relations: (1) Such relations presuppose entities (e.g. essences, non-spatio-temporal quality orders like the colors, logical syllogisms taken as abstract objects) or a type of holism (e.g. living things are Aristotelian substances and not property-things) not consistent with naturalism. (2) It is hard to locate them spatially and temporally.
57 Campbell, Abstract Particulars, 37, 38, 59–60.
58 Cf. J. P. Moreland, Universals, Qualities, and Quality-Instances, 71–4; “Keith Campbell and the Trope View of Predication,” Australasian Journal of Philosophy 67 (December 1989): 379–93.
For Campbell’s admission of the force of these puzzles, see Keith Campbell, Abstract Particulars, 65–6.
59 It is important to note that the problems revealed by these puzzles have nothing essentially to do with the fact that Campbell held to a bundle theory of substance. If one held a trope version of a traditional, Aristotelian-type view of substance, with minor adjustments, these same puzzles would apply. Dogs as much as reds exactly resemble each other and are individuated. And on Campbell’s earlier view, a dog’s nature would differ from its location by a distinction of reason. Elsewhere, I have criticized what amounts to a trope ontology combined with a traditional substance view. See J. P. Moreland, “How to be a Realist in Nominalist Clothing.”
60 The primary place where Campbell has expressed his new position is Abstract Particulars.
61 Ibid., 56–8; 89–90.
62 Francis Suarez, On the Various Kind’s of Distinctions, disputation 7, 1:13–15 (pages 24–7 of the 1947 Marquette edition).
The ontological status of properties 107
63 Campbell, Abstract Particulars, 68–9
64 Ibid., 37, 59–60
65 Ibid., 31, 37–40.
66 Ibid., 67, 150–1, 172.
67 Kim, “Mental Causation and Two Conceptions of Mental Properties,” 22–3.
68 Campbell, Abstract Particulars, 37. See Armstrong, Universals: An Opinionated Introduction, 56, 100. For Armstrong, the supervenience relation is the “nothing over and above” relation such that supervenient entities do not add any real, new entities to one’s ontology.
69 G. Dawes-Hicks, “Are the Characteristics of Particular things Universal or Particular?,” (in symposium with G. E. Moore and G. F. Stout), Aristotelian Society Supplementary Volume III (1923): 125.
70 Campbell, Abstract Particulars, 135–55.
71 Campbell says that if an entity is conventional, then it isn’t real or natural, yet he also says that manifest tropes are not figments even though their identity conditions are arbitrary and conventional. These and other confusions permeate Campbell’s discussion of these issues.
See Abstract Particulars, 137, 152.
72 Ibid., 146.
73 Ibid., 37, 100–1, 103, 110, 121–2.
74 Ibid., 67, 150–1, 172.
75 Ibid., 98, 100, 101, 103–4, 110, 121–2.
76 Ibid., 125, 132, 145, 148, 150, 152–5.
77 See J. P. Moreland, “Was Husserl a Nominalist?.”
78 Campbell, Abstract Particulars, 20, 22, 27, 56, 60.
79 Ibid., 27, 89.
80 Ibid., 37–8, 59–60.
81 Ibid., 9–10, 68–71.
82 Ibid., 70.
83 Ibid., 43–5.
84 Cf. J. P. Moreland, “Was Husserl a Nominalist?”; “Keith Campbell and the Trope View of Predication”; “Nominalism and Abstract Reference,” American Philosophical Quarterly 27
(October 1990): 325–34; “Issues and Options in Exemplification,” American Philosophical Quarterly 33 (April, 1996): 133–47.
85 In one place, Armstrong expresses a mild inclination to reject the literal multiple location of properties in favor of the view that all universals are “in” states of affairs that are themselves spatio-temporally located in the cosmos and, in this sense, are part of what constitutes the cosmos. See Universals: An Opinionated Introduction, 75–7, 81–2, 98–9. For three reasons, I shall not discuss this further. First, it runs contrary to almost everything Armstrong says about spatio-temporal location as a criterion of ontological commitment, e.g. that internal relations should be rejected because they cannot be located. This may be why he expresses himself here with the qualification that he is simply inclined to the view, not that it actually represents his considered judgment about the matter. Subsequently, he seems to have reaffirmed the spatio-temporal location of every existent. See A World of States of Affairs, 8, 39–
41. Second, Reinhardt Grossman has pointed out that when Armstrong claims that all relations must be located in space and time except spatio-temporal ones, this concedes what the anti-naturalist argues, namely, that they are abstract objects, and Armstrong is clearly against the existence of the latter. See Reinhardt Grossmann, The Existence of the World, 27–8.
But this runs contrary to Armstrong’s entire ontology. Finally, one of Armstrong’s main concerns with uninstantiated universals is that if they exist, they exist “in another realm”
outside space and time. Now, if the concern is to avoid such entities, it is not sufficient to claim that universals must be instantiated, because the traditional realist will say that properties remain abstract objects even while they are non-spatio-temporally “in” their instances. Thus, Armstrong needs to locate properties to avoid treating them as abstract; it is not sufficient merely to require of them that they be exemplified.
86 Armstrong, Nominalism and Realism, 109.
87 In his more recent thinking, Armstrong has followed Frege and sought to incorporate exemplication into the universal itself. See A World of States of Affairs, 28–31. He claims that properties are unsaturated entities that are filled by particulars, rather than being complete entities in their own right that are exemplified by particulars. According to Armstrong, this move allows him to avoid a Bradley-type regress that comes from admitting that exemplification is a relation and it makes sense of the fact that universals are dependent upon states of affairs and that there are no uninstantiated universals. Unfortunately, this move seems merely to be a verbal shift, for now universals must sustain two functions – having an intrinsic qualitative nature and being had by something – instead of having two entities each of which sustain one of these functions. It is hard to see how the former is preferable to the latter. Moreover, being unsaturated must now be taken as a real feature of universals different from their qualitative natures (e.g. since redness and greeness differ regarding their qualitative natures but agree in being saturateable). If so, it is hard to see how being unsaturated is a spatially located feature of either the universal or the state of affairs into which it enters. It is important to remember that for Frege, neither concepts nor their being unsaturated were spatio-temporally located. Third, there are other ways to avoid Bradley-type regresses without appealing to the notion of saturation and these strategies are well known. Finally, no gain is made for Armstrong regarding the dependence of universals on states of affairs or in providing a rationale for no uninstantiated universals. Frege himself believed that concepts could exist without entering a state of affairs (Armstrong’s language) or without being saturated. It is just not clear why something could not exist while unsaturated. Without further argumentation which Armstrong fails to provide, a mere appeal to saturation and a particular’s filling a universal provides no justification whatever for what Armstrong is after. Indeed, Frege’s ontology, including his views about saturation, have long been seen as contrary to the naturalist stance.
88 Reinhardt Grossmann has shown that Armstrong’s view of relations as spatial is even more problematic than is his position on properties. See Reinhardt Grossmann, The Existence of the World, 26–8; cf. 51–7.
89 See J. P. Moreland, “Was Husserl a Nominalist?”
90 See Michael Loux, Substance and Attribute, 92–6.
91 Armstrong, Nominalism & Realism, 64–6, 68–9, 102; A Theory of Universals, 3, 76.
92 See D. M. Armstrong, Universals: An Opinionated Introduction, 113–33.
93 Ibid., 122.
94 Richard Connell, Substance and Modern Science (Notre Dame: University of Notre Dame Press, 1988), 90.
95 Armstrong, Universals: An Opinionated Introduction, 94.
96 Armstrong, A World of States of Affairs, 124. Cf. 95–112; 123–7. Armstrong’s only problem with Locke’s view of a substratum, is that it was an unknown postulate and Armstrong insists that
The ontological status of properties 109
the particularity of a thick particular is presented in acts of perception. Unfortunately, he is not as clear on this as one might wish because he seems to go back and forth between identifying what is so presented as (1) the fact that thick or propertied particulars are not exhausted by their properties and relations and (2) the thin particular itself. But as we have seen, the victory of particularity (which Armstrong accepts) does not clearly allow one to assert (2) just because one is acquainted in acts of perception with the thick particular qua particular.
97 For Armstrong’s rejection of bare particulars in sense one, see A World of States of Affairs, 86, 153, 267–8. For a defense of bare particulars as individuators, see J. P. Moreland, “Theories of Individuation: A Reconsideration of Bare Particulars,” Pacific Philosophical Quarterly 79
(September 1998): 251–63.
98 Armstrong, A World of States of Affairs, 107–11, 123–6.
99 Seibt, Properties as Processes: A Synoptic Study of Wilfrid Sellars’ Nominalism, 184.
objects1
Michael Rea
In recent decades, philosophers have devoted a great deal of attention to questions about how philosophical naturalists should understand mental objects, such as beliefs and desires, and apparently abstract objects, such as properties, numbers and sets.
What has not been so often discussed, however, is the question of how naturalists should think about the wide variety of ordinary material objects that populate the ontology of common sense. This is the question that I shall address in the present essay.
At the outset, we must observe that our question is vexed – vexed by the fact that the house of naturalism is a house divided. There is little agreement about what naturalism is, or about what sort of ontology it requires. Some philosophers cast naturalism as a metaphysical view.2 Others cast it as an epistemological or methodological view.3 Still others deny that it is a single unified position at all, holding instead that there are several different, loosely connected versions of naturalism (metaphysical, epistemological, methodological, and so on).4 Some naturalists allow abstract objects into their ontology;5 and some say that naturalism is even compatible with belief in God, ghosts, souls, objective values, and other sorts of things widely held to be non-natural or supernatural.6 Some say that naturalism, or some version thereof, implies or is even to be identified with either materialism or physicalism;7 others deny that it is to be identified with any specific ontology at all.8 Most naturalists would affirm Wilfrid Sellars’s slogan that “science is the measure of all things: of what is that it is and of what is not that it is not”;9 and many no doubt would say that this slogan captures the heart and soul of naturalism. But a precise and even modestly uncontroversial explanation of what that slogan means has yet to appear in the literature. Hence, in one sense there really is no clear answer to the question of how naturalists should think about ordinary material objects because (as yet) there is no clear answer to the question of what it means to be a naturalist.
My own view is that naturalism is not a philosophical position at all, but rather a research program – a plan to conduct inquiry in a particular way. The naturalistic
Naturalism and material objects 111
program consists primarily of a plan to use the methods of the natural sciences, and those methods alone, in the development of philosophical theories. The consequences or implications of naturalism are just those views to which one is committed by virtue of following this program. Elsewhere I have argued at length that this way of characterizing naturalism is more faithful to the tradition than any of the other characterizations mentioned above, and it has the further advantage of avoiding such nasty problems as self-defeat and incompatibility with naturalistic methodological preferences.10 Furthermore, the research program that I have identified with naturalism is one that seems to be universally endorsed by those who call themselves naturalists (one good reason for making the identification!), even if they disagree with the idea that acceptance of the research program is constitutive of their naturalism. Thus, by taking naturalism in this way, I am fairly well assured of defending conclusions that will apply to all naturalists, whatever their particular disagreements with one another might happen to be.
My thesis is that material objects as they are ordinarily conceived have no place whatsoever in the strict ontology of a consistent naturalist. On the ordinary conception of such things, every material object has the following property: facts about what sorts of changes it can and cannot survive do not depend upon human beliefs, attitudes, concepts, or conventions. More succinctly, but less precisely: material objects are mind-independent.11 What I shall argue is that the methods of the natural sciences alone do not justify belief in mind-independent material objects; hence, naturalists have no business adopting an ontology that includes such things. I believe that in showing this, I am also showing that commitment to naturalism precludes belief in material objects entirely. This is because I believe it is impossible for there to be material objects that are not mind-independent in the sense just specified. But this claim is contentious, and I do not have the space to defend it here.
Thus, I am officially defending only the more restrained conclusion.
The chapter has four parts. In the first, I argue that we can be justified in believing that there are mind-independent material objects only if we can be justified in believing that modal properties are exemplified in at least some of the regions of space-time that we take to be occupied by material objects. In the second, I argue that we can be justified in believing that modal properties are exemplified in a region only if we can be justified in classificatory judgments – judgments like ‘this region contains an F’, where ‘F’ is a name for a natural kind and furthermore constitutes a metaphysically better answer to the question, “What kind of object is in that region?”
than any name for any other kind. In the third part, I dismiss as failures three views about how naturalists might be able to be justified in classificatory judgments. I consider a fourth which presupposes that, if there are material objects, then we are justified in believing that science reveals some of them to have proper functions; and I note that if this fourth proposal fails, there is good reason to think that any other proposal will as well. In the final section I argue that this fourth proposal does indeed
fail by showing that if there are material objects, science does not reveal any of them to have proper functions. If I am right, and if (as seems plausible) we are justified in believing that there exist mind-independent material objects (e.g. people), then we ought to reject naturalism.
I
Material objects have persistence conditions. By this I mean that for every material thing, there are facts about what changes it can and cannot survive. A man over six feet tall cannot survive being crushed down to the size and shape of a sugar cube, and a sugar cube cannot survive its own dissolution in a cup of tea. On the other hand, a man usually can survive the sorts of changes normally undergone in the course of an afternoon stroll, and a sugar cube usually can survive the sorts of changes normally involved in the daily existence of a sugar cube (being in a bowl, being jostled as nearby sugar cubes are removed, and so on). Even for instantaneous objects (if there are such things) which do not last over time at all, there are nevertheless facts about whether and in what circumstances it is possible for them to last over time. If there are no facts at all about what sorts of changes a putative thing X can and cannot survive, then there is no such thing as X.
Those who doubt this are invited to consider the claim that there is such a thing as Socrates but literally no fact at all about whether Socrates could survive a trip to Macedonia or a trip through a meat grinder. The claim is patently absurd; and the same can be said for similar claims about other material objects. Many are willing to accept some indeterminacy in persistence conditions. Thus, for example, one might think that if we were to annihilate Socrates’ constituent atoms one by one, there would be no fact about which atom was the one whose annihilation finally did him in. But accepting this kind of indeterminacy is not at all the same as accepting the view that Socrates simply lacks persistence conditions. I know of no one who would say that there is no fact about whether Socrates could survive the simultaneous annihilation of all of his constituent atoms; and, again, similar remarks could be made with respect to any persisting material object. Where there are no persistence conditions, there is no object.
Having persistence conditions is part of our concept of a material object – just as having boundaries is part of our concept of a geometrical figure. But notice that there is a substantive ontological commitment lurking in the neighborhood. I have already indicated that persistence conditions are facts about what an object can and cannot survive; thus, they are modal facts (facts about what is necessary or possible for an object). Alternatively, they might just as well be construed as modal properties –
properties like being essentially such as to undergo mental activity or being essentially such as to have the chemical structure H2O (where X is essentially φ just in case X cannot exist
Naturalism and material objects 113
without exemplifying φ). Either way, acceptance of material objects carries with it a commitment to modality.
Once this is seen, it is perhaps easier to motivate the question: What justifies us in believing in material objects? At first blush, modal facts and modal properties seem wholly inaccessible to the natural sciences. Scientific methods can tell us, for example, that individual water molecules have the chemical structure H2O; but it is hard to see how those same methods could tell us that water molecules have their chemical structure essentially, or that it is a necessary truth that water molecules have the chemical structure they do. Surely no experiment could confirm any of this.
Moreover, the proposition that water molecules necessarily have the chemical structure they have doesn’t obviously play any integral role in a total comprehensive scientific theory. We do not have to believe that water-molecule phenomena are caused by material objects in order to predict and explain those phenomena. Instead of an ontology of objects, we might simply adopt an ontology of stuff.12 That is, we might suppose that our sensory experiences are caused not by discrete material objects but rather by variously propertied stuff, to which the notions of survival, generation, and destruction do not apply. There is nothing incoherent about supposing that nothing is destroyed when the stuff of which a water molecule is made is rearranged so as no longer to constitute a water molecule; and the same goes for other putative objects. In general, we do not need to believe in objects at all in order to make predictions and explain the various empirical phenomena associated with the (apparent) generation, existence and destruction of such things. Believing only in the stuff that constitutes them, along with its various qualitative properties, will do just as well. Thus, there is at least prima facie reason for worrying that, from a naturalistic perspective, belief in modal facts and modal properties – and so belief in material objects as well – will be unjustified.
One way of addressing this worry is to deny that there are modal properties and to maintain that modal facts are grounded somehow in human mental activity. To accept this sort of view is to embrace modal anti-realism. Modal anti-realism comes in two main varieties: conceptualism and conventionalism. According to conceptualism, modal facts are grounded in human concepts; according to conventionalism, they are grounded in human linguistic conventions.13 The details of the two views are not important for our purposes. What is important is simply the fact that, with regard to saving the ontology of material objects, modal antirealism has one clear advantage and one fatal disadvantage. The advantage is that it makes it easy to see how we might acquire justified beliefs about persistence conditions: we simply reflect upon our concepts, or upon the empirically discovered qualitative features of the world and the semantic conventions governing our language. The disadvantage is that it is incompatible with the claim that material objects are mind-independent. Well, not quite: there are other possible versions of modal antirealism. For example, one might
think that modal facts are grounded in divine concepts or conventions or, less plausibly, that they are grounded in the concepts or conventions of other non-human beings (say, dolphins, apes, or Martians). Strictly, such versions of modal antirealism would be compatible with the mind-independence of material objects as I have defined it. But they are incompatible with naturalism, at least if we assume (as most naturalists do) that the methods of science alone provide no warrant for believing in other non-human beings with sufficiently rich concepts or languages to ground all of the modal facts in the world.14
If the ontology of mind-independent material objects is to be saved, then, refuge must be sought in modal realism.15 For some, this fact will be sufficient to establish the conclusion of this chapter. Many philosophers, naturalists and non-naturalists alike, agree that naturalism is incompatible with modal realism. But many do not; and I do not need to establish anything so strong as the incompatibility of naturalism with modal realism in order to secure the conclusion of this chapter. The reason is that more than just modal realism is required to justify belief in material objects. Modal realism, but not belief in material objects, would be justified even if we were only justified in believing that sets or propositions have modal properties. To be justified in believing in material objects, we need to be justified in believing not only that there are modal properties but that such properties are exemplified in at least some of the regions that we take to be occupied by material objects. In the following sections, I will argue that commitment to naturalism precludes our being justified in believing that modal properties are so instantiated.
II
It is well known that essentialism about material objects – that is, the view that material objects have at least some essential properties – is plagued by the appearance of paradox. One particularly troublesome problem is what I have elsewhere called the problem of material constitution. 16 The problem is raised by a variety of apparently different puzzles, all of which reveal conflicting intuitions about what modal properties are instantiated in various regions of space-time. Consider, for example, a region of space-time occupied by a leafless oak tree.17 Many people, at least initially, have the intuition that there is something in the region (namely, the tree) that cannot survive being chopped up into firewood, as well as something (namely, the aggregate of cellulose molecules that constitutes the tree) that can survive such chopping. The problem is that if the tree and the aggregate have different essential properties, it seems that they must be distinct; but if they are distinct, then there are two objects in exactly the same region of space-time, which seems impossible.
Reflection upon this puzzle reveals something about what is involved in our being justified in believing that particular modal properties are instantiated in particular
Naturalism and material objects 115
regions of spacetime. The puzzle arises because, in the region allegedly occupied by our oak tree, there exists matter that is arranged in at least two different ways – a way that is qualitatively distinctive of mere aggregates of cellulose molecules, and a way that is qualitatively distinctive of oak trees. The fact that the matter is arranged in these different ways is what allows us to set up the puzzle by saying that the region is both occupied by a tree and occupied by an aggregate. The puzzle is solved only when we figure out what the relationship is between having matter that is arranged in the distinctive qualitative way exemplified by members of a particular kind and having something that exemplifies the essential properties normally associated with that kind. But notice: we also need to be able to figure out this relationship in order to justifiably believe (apart from testimony) that particular modal properties are exemplified in a particular region. Solving the puzzle about the tree and the aggregate is precisely a matter of figuring out whether we have something that is essentially a tree, something that is essentially an aggregate, both, or neither, in the region allegedly occupied by the oak tree. In other words, it is a matter of figuring out which (if any) of a set of essential properties are instantiated in the region; and to do that, we must know something about the link between essential properties and ways of arranging matter.
Kind concepts are the tools by which we forge the link between essential properties and ways of arranging matter. It is traditionally assumed that there is both a metaphysical and an epistemological connection between kind-membership and persistence conditions. If this is right, then given the further assumption (which we shall continue to make throughout the remainder of this discussion) that persistence conditions are or are grounded in essential properties, there are also such connections between kind-membership and essential properties. But there are two ways of belonging to a kind. Something belongs to a kind in the classificatory way just in case terms referring to that kind are metaphysically better answers to the “What is it?” question for the object than terms referring to any other kind. On the other hand, something belongs to a kind in the nominal way just in case it exemplifies the distinctive qualitative features of those things that belong to the kind in the classificatory way.18
On the traditional view, it is only by belonging to a kind in the classificatory way that an object is guaranteed to have the essential properties associated with that kind.
Having one’s matter arranged in such a way as to exemplify the distinctive qualitative features of the members of a kind is not sufficient. Thus, in order to justifiably believe that certain modal properties are instantiated in a region – indeed, to justifiably believe that any modal properties at all are instantiated in a region – we must somehow be able to distinguish between cases where there is merely matter arranged so that the distinctive qualitative features of the members of some kind are exemplified and cases where the matter really composes something that is classifiable as a member of that kind. In other words: we need to be able to distinguish regions that are nominally K-containing (for some kind K) from regions that contain something classifiable as a K.
(I should note here that I am assuming that every way of arranging matter is kind-constituting. But one might deny this, and in so doing one might come to the view that belonging to a kind in the nominal way entails belonging to it in the classificatory way.
Thus, for example, Peter van Inwagen believes that the only genuine kinds are living-organism kinds;19 and one might readily add to his view the claim that nothing can be nominally some kind of living organism without also being essentially that kind of organism. If this is right, then the problem we face in acquiring justified beliefs about the exemplification of modal properties is not one of distinguishing between regions that are nominally K-containing and regions that are essentially K-containing, for some kind K. Rather, it is a problem of distinguishing between ways of arranging matter that are kind-constituting and ways of arranging matter that are not. The two problems are equivalent, and they admit of equivalent solutions. Thus, I take it that my assumption that every way of arranging matter is kind-constituting is innocent.) Are there other ways of acquiring justified beliefs about the essential properties exemplified in a region apart from testimony and apart from acquiring justified classificatory beliefs? Apparently not. Suppose you travel to a foreign planet and encounter a region containing matter arranged in a wholly unfamiliar way; and suppose a fellow traveler asks you whether there’s anything in the region that could survive a blast from your laser gun. How could you even begin to answer? You could take a shot at it with your gun, but how would you know whether your shot destroyed anything? Even if you produced a gaping hole in the putative object, apart from testimony (e.g. squeals of pain from the object, or a booming voice imploring you not to take another shot) and apart from some justified view about what the object is, there would be no way of knowing whether you had damaged anything or simply rearranged some matter in an innocuous way. Thus, it seems that if we have justified beliefs at all about the exemplification of modal properties in regions of space-time, our having such justified beliefs depends either on testimony or on our having justified classificatory beliefs.
Obviously not everyone’s beliefs about essential properties can come from testimony. Thus, our being justified in believing that particular essential properties are exemplified in particular regions of spacetime must ultimately depend upon someone’s being justified in their classificatory beliefs, and this too apart from testimony. But, of course, commitment to naturalism implies that someone can be so justified only if the methods of science alone allow us to distinguish between nominally K-containing regions (for some kind K) and regions that contain something classifiable as a K. In the next section, I shall argue that the methods of science alone do not allow us to make such determinations.
Naturalism and material objects 117
III
I will begin by dismissing two suggestions about how we might acquire justified classificatory beliefs that are initially tempting but that are readily seen, upon brief reflection, to be non-starters. I will then consider what I take to be the two most promising explanations and show why they fail. I should say up front that, in order for my argument in this chapter to be strictly valid, I need the further premise that if the alternatives I am here considering fail to provide a naturalistically acceptable explanation for how we might acquire justified classificatory beliefs, then every other alternative explanation will fail as well. I cannot prove that this premise is true, since (as the reader can easily verify) the alternatives I am considering are not logically exhaustive. However, I see no reason to think that there are other even remotely promising alternatives; and so I think it is reasonable to assert on this basis that if these alternatives fail, any others will as well. Those who disagree may simply take my argument as a challenge to provide a successful alternative.
One might think that we can acquire at least one justified classificatory belief in the following way: we introspect and find experiential evidence for our own existence. We then infer from this that essential properties must be exemplified in at least one region of spacetime – namely, the region we ourselves occupy. Strictly speaking, this procedure saves belief in material objects. The trouble, however, is that it saves it only by presupposing it. We have experiential evidence of our own existence, but not of the fact that we are material objects. The belief that we are our bodies depends for its justification upon the belief that we have bodies – a belief which is justified only if we are justified in believing that the matter in the region putatively occupied by our bodies is more than nominally a human-body-containing region. The question is why we should think that it is, and so far we have no answer.
On the other hand, one might think that the belief that we are material objects is a Moorean fact – a datum of common sense, automatically justified and sufficiently so to constitute an argument against any set of premises that imply its denial. Though I think it is far from obviously true that materialism about the mind has the weight of common sense behind it in the same way that realism about the external world does, I need not press that point here. The main problem with this suggestion is that following it is not compatible with commitment to naturalism. If the belief that we are material objects is established through Moorean common sense rather than the empirical methods of science, then it is not justified from a naturalistic point of view.
So much, then, for non-starters. As I see it, there are only two even remotely promising views about how we might acquire justified beliefs about classification through the methods of science alone. The first may be treated rather briefly.
Undermining the second will occupy the whole of Section IV.
According to the first of the remotely promising suggestions, our classificatory
beliefs are justified by their role in the formulation of natural laws. The belief that a particular region contains a tree rather than an aggregate of cellulose molecules, for example, might be justified on the grounds that similar classifications have provided a basis for the articulation of natural laws governing the behavior of trees which have then been confirmed through empirical testing. Other classificatory beliefs would be justified in similar ways. As a general rule, then, the contents of regions are to be classified in ways that allow for the simplest, most economical formulation of an overall set of natural laws; and the justification for the total classificatory scheme is precisely the same as the justification for the total theory of which it is a part.
This view suffers from two weaknesses. The first is that it makes the justification for all of our classificatory beliefs pragmatic rather than epistemic. The fact that our task of theory-construction is simplified by adopting one set of classificatory beliefs rather than another is not at all obviously a reason for thinking that they are true. The classificatory scheme we adopt, as well as the whole ontology of material objects itself, might well be a convenient fiction; but why should convenience warrant belief? I realize, of course, that many naturalists (though not all!) think that pragmatic justification of this sort is sufficient for epistemic justification (or – what might come to the same thing – that pragmatic justification is the best we could ever hope for in the way of justification). But even if I were to concede that pragmatic justification for our classificatory beliefs is sufficient enough for whatever justification we might desire for belief in material objects, there would still remain a further weakness with the view under consideration. That weakness is just that there seems to be no good reason to believe that our theories can be significantly simplified by adopting one set of classificatory beliefs rather than another. In biology, for example, there seems to be good reason to think that any of a variety of classificatory systems will do equally well; and in physics it seems that we can get along just fine by restricting our ontology to include only space-time, regions of space-time, and the various qualitative properties that might be exemplified at those regions.20 Thus, even if pragmatic justification were sufficient to meet our needs, it is not at all clear that we could get it.
A more promising view about how we can acquire justified classificatory beliefs is (roughly) that we can acquire such beliefs by acquiring, through the methods of science alone, justified beliefs about proper function.21 But here we must be careful.
Obviously we cannot say that our classificatory beliefs depend for their justification upon our ability to acquire justified beliefs to the effect that some mind-independent material object has a proper function; for I have already argued that belief in mind-independent material objects depends for its justification upon our ability to acquire justified classificatory beliefs. Nevertheless, one might think that whether or not there really are mind-independent material objects in the world, still there are phenomena in nature revealed by scientific methods such that, on the assumption that there are mind-independent material objects, there is strong evidence for believing that science
Naturalism and material objects 119
reveals some of those objects to have proper functions. And one might think that it is these proper-function-phenomena that somehow ground our classificatory beliefs.
The view now under consideration may be expressed more precisely as follows. We can acquire justified classificatory beliefs through the methods of science alone if, and only if, (3.1) is true:
(3.1)
If we were to assume that there are material objects, then there would be strong evidence for believing that science reveals some of those objects to have proper functions.
As I said before, the ‘only if’ is justified by the fact that no other alternative seems available. Those who doubt this are again invited to supply another alternative. In the remainder of this section, I will briefly explain why I think that (3.1) implies that we can acquire justified classificatory beliefs through the methods of science alone. In Section IV, I will argue that 3.1 is false by assuming that there are material objects and showing that there is no reason to believe that science reveals anything to have a proper function.
Before explaining why I think that (3.1) implies that we can acquire naturalistically respectable classificatory beliefs, I must make some preliminary remarks about function ascriptions. Almost everyone writing on functions agrees that function ascriptions are system-relative. In other words, statements of the form “x has the function of doing φ” are always elliptical for statements of the form “In system s, x has the function of doing φ.” The same is true for ascriptions of proper functions.
Nevertheless, we do sometimes speak of whole systems as functioning properly or improperly even when it is doubtful that the system as a whole has a proper function.
For example, we speak of cars as functioning properly or improperly. We might even speak of ourselves, or our pets as functioning properly or improperly. (We might do this; but in many contexts it would sound strange. One who pointed at his rabid dog and said “Boy, it looks like my dog is malfunctioning!” would probably appear to be making a joke.) So proper functioning does not necessarily involve performing a proper function. In one legitimate sense of the term, a system as a whole is functioning properly just in case it is healthy (or, for artifacts, in a state relevantly analogous to health); and this is so whether or not the system itself has a proper function. Toward preserving these distinctions, then, I will say that something has a proper function or a proper part-function just in case it has a function it is supposed to perform within some larger system, and I will say that something has a proper system-function just in case there is some way of functioning that counts as proper functioning (or health or a relevant analog of health) for that system. That said, I turn now to my defense of the claim that (3.1) implies that we can acquire naturalistically acceptable classificatory beliefs.
In short, the idea is that the existence of proper-function-phenomena in nature might reasonably be construed as evidence to the effect that, though the matter in a
region is arranged in multiple ways at once, one of the ways in which it is arranged is privileged over the rest; and such evidence provides a basis for classification. Consider a region of space-time that contains matter organized in such a way that (a) there seems to be a system S in that region, and (b) S seems to have parts that have proper functions relative to S.22 Suppose further that, on the assumption that there really are mind-independent material objects, the evidence is that the methods of science alone have revealed the proper part-functions of some of the parts of S. Then it seems clear that the methods of science alone have revealed that among the various ways in which the matter in the region allegedly occupied by S is arranged, one is metaphysically privileged over the rest as the way that matter is supposed to be arranged. But now we have a basis for a classificatory judgment, for, given that the matter in the region allegedly occupied by S is supposed to be arranged in a particular way, it is quite plausible to think that the region is occupied by an object classifiable as a member of whatever kind is associated with the particular way in which the matter in that region is supposed to be arranged. So if (3.1) is true then classificatory judgments (and hence modal attributions and belief in material objects generally) can be justified through the methods of science alone.
There is one very obvious objection that might be raised against the view just described. Even if the view does explain how we might be justified in our classificatory beliefs involving macroscopic objects like organisms and perhaps artifacts, it doesn’t at all seem to explain how we might be justified in classificatory beliefs involving non-living natural objects like rocks, planets, water molecules, atoms, and the like. After all, neither planets nor water molecules nor any of their respective parts seem to have proper functions. Thus, at best, the view seems to provide the naturalist only with a way of admitting some material objects into her ontology. It does not offer the resources for countenancing the wide variety of non-living natural objects in which many of us believe.
There is a genuine concern here, but I do not think it is decisive. Suppose one thinks (with Aristotle, for example) that living organisms are paradigmatic examples of material objects, and that our beliefs about proper function do indeed ground our beliefs in those paradigmatic items. One might then go on to argue that our beliefs in other things (like rocks and atoms) are justified either by analogical extension, or by their role in scientific theories. I have already argued that the role of talk about such objects in scientific theories is not sufficient on its own to justify belief in their existence or in material objects generally. But it is quite plausible to think that once we have already become justified in believing in some material objects by way of our beliefs about proper function, there is no longer any obstacle to believing in the many other sorts of objects that are talked about in our scientific theories. Our beliefs about proper function give us reason to abandon a stuff-ontology in general, and once that is abandoned, it becomes legitimate to take all of our apparent talk about material objects at its face value.
Naturalism and material objects 121
Incidentally, my own view is that something like this story is the correct story about how we acquire justified classificatory beliefs (and hence justified beliefs in modal properties and material objects). That is, I am inclined to think that we in fact acquire such beliefs by way of our beliefs about proper function. However, as I shall argue in the next section, this sort of view about how we acquire justified classificatory beliefs is not available to the naturalist. The reason is that (3.1) is false: even if we assume that there are material objects, there is no good evidence for believing that the natural sciences reveal some of them to have proper functions.
IV
Let us begin the argument against (3.1) by observing that science can reveal the proper system-function of an object only if it can also reveal the proper part-functions of some of its parts. This is not necessarily to say that we learn facts about proper system-function by inferring them from facts about proper part-function (though perhaps we do). My point is simply that there is no reason to suppose that science can tell us what it is for an organism to enjoy proper system-function but leave us totally in the dark about what its various parts are supposed to do. Proper system-function supervenes on proper part-function; so if the former is accessible to science we would expect the latter to be also.23 Furthermore, facts about proper system-function are not radically different in kind from facts about proper part-function; so, again, there is no reason to suppose that science can reveal the former but not the latter. (In the same way there is no reason to suppose that science can reveal facts about rights but not facts about obligations, or facts about mental phenomena but not facts about brain chemistry.) Thus, the naturalistic acceptability of beliefs about proper system-function stands or falls with the naturalistic acceptability of beliefs about proper part-function.
For my purposes, a part of a thing has a proper function just in case (a) it has a function, and (b) one of its functions is metaphysically privileged – i.e. the part is objectively supposed to have that function, or it absolutely must have the function, or its having that function contributes in some way to the Good (construed as objective and universal), etc. I think that these two conditions accord well with ordinary intuitions about what it is for something to have a proper function; but I should note that, whether they do or not, at any rate this is the sense of proper function that is needed in order for (3.1) to imply that we can acquire justified classificatory beliefs. If proper functions are not metaphysically privileged functions, then there is no reason to suppose that we can discover a metaphysically privileged arrangement of parts for a thing by discovering proper functions.
I will take it for granted that, on the assumption that there are material objects, there is good evidence for thinking that the methods of science alone can reveal that something has a function. One might doubt this. For example, some philosophers
think that an object has a function just in case it contributes to a goal of some system of which it is a part; and some philosophers think that a system can be goal-directed only if it has been designed. If both views are correct, then it follows that biological objects have functions only if there is a cosmic designer. But we are assuming that the methods of science alone cannot reveal the existence of a cosmic designer; hence, it is hard to see how (if this view were correct) the methods of science could reveal that anything has a function. Of course, if the methods of science cannot reveal that anything has a function, then a fortiori they cannot reveal that anything has a proper function. Hence, I take it that in assuming that science can reveal that some things have functions, I am simply giving the naturalist a running start.
There are various different views about what it is for something to have a function.
Though it would certainly be interesting to do so, there is no need to canvass that territory here. All that is important for present purposes is the fact that, no matter what view of functions ultimately turns out to be correct, in the literature on functions there are only two suggestions compatible with commitment to naturalism as to what might make it the case that a function of a thing is also its proper function. One suggestion is that statistical normality is what determines proper function. For example, one might think that X has F as its proper function just in case X belongs to a kind whose members (statistically) normally have F as their function. The other suggestion is that the etiology, or causal history, of a thing determines its proper function. So, for example, one might hold that X has F as its proper function just in case X exists at all or exists where it is because it tends (or, better, because things of its type tend) to do F.24 I will argue that, though science can indeed detect facts about statistical normality and etiology, neither statistical normality nor natural etiology is sufficient for proper function. Hence, lacking other naturalistically acceptable suggestions as to what properties might make it the case that a function of a thing is also its proper function, I will conclude that the methods of science alone cannot reveal that anything has a proper function.
Consider first accounts of proper function in terms of statistical normality (let’s call them ‘SN-accounts’ for short). No one that I am aware of has explicitly proposed an SN-account. However, I discuss it because, as Alvin Plantinga notes, this sort of analysis is “widely popular in the oral tradition.”25 Plantinga himself offers several counter-examples; but most of them seem directed against SN-accounts of proper system-function rather than against SN-accounts of proper part-function. So, for example, he writes:
Most 60–year old carpenters have lost a finger or thumb; it is not the case that those who have not have hands that are not normal and not capable of proper manual function. . . . Perhaps most male cats have been neutered; it hardly follows that those that haven’t are abnormal and can’t function properly. The vast