SORITES , ISSN 1135-1349

Issue #07. November 1996. Pp. 21-27.
«Van Inwagen and Gunk: A Response to Sider»
Copyright © by SORITES and Kelly J. Salsbery

Van Inwagen and Gunk: A Response to Sider

Kelly J. Salsbery

In a recent article (Sider [1993]), Theodore Sider raises an interesting objection to the ontological views of Peter van Inwagen (van Inwagen [1990]). He attributes to van Inwagen the following two theses:

(1) For any material objects X, the Xs compose something iff the activity of the Xs constitutes a life, or there is only one of the Xs.

(2) Every material object is either a mereological atom or a living thing.

Sider notes that here `mereological atom' means «an object lacking proper parts» (p. 285). He also notes that (2) seems to follow from (1).

Sider's objection to van Inwagen's approach is based on the claim that, «there are (or rather might have been) situations in which `objects' like tables and chairs are not composed of fundamental particles» (p. 286). He suggests that an alternative would be to posit material objects composed of `atomless gunk'. Sider borrows this term from David Lewis (See Lewis [1991], p. 20). He notes that such atomless gunk would have no mereological atoms (or simples) as parts and would be infinitely divisible (p. 286).

Sider asks us to imagine possible worlds at which there is only gunk. What he calls a «gunk world» would be a possible world at which there exist no mereological atoms (and for the sake of simplicity, no living things). If (2) is true, however, then the only things which exist are mereological atoms or living things. Thus, no material objects would exist at gunk worlds.

Sider finds such a claim implausible. Concerning this he writes:

Surely there is a gunk world in which some gunk is shaped into a giant sphere, and another where some gunk has the shape of a cube. Surely, there are gunk worlds that most of us would describe as containing objects much like objects from our world: tables and chairs, mountains and molehills, etc. (p. 286).

Sider thinks that plausibility of the existence of gunk worlds undermines van Inwagen's position because he believes that van Inwagen holds (2) to be necessarily true (p. 287). Sider's argument goes as follows:

(p1) If (2) is necessarily true, then gunk worlds at which material objects exist are impossible.

(p2) Gunk worlds at which material objects exist are not impossible.

Thus, it is not the case that (2) is necessarily true.

Sider further claims that such a conclusion undermines van Inwagen's approach because this approach fails to account for the existence of gunk worlds.

Sider's argument is clearly valid, but it is not so clear that it is sound. Premise p1 seems obviously true. If it is necessarily true that every material object is either a mereological atom or a living thing, then there cannot be world at which a material object is composed of gunk.

Premise p2, however, is somewhat problematic. Sider supports p2 mainly on the basis of imaginability or conceivability (as noted above). That is, Sider thinks that the fact that we seem to be able to imagine a world at which there exist objects composed of gunk supports the claim that such a possible world exists. Consequently, he thinks it is implausible to deny the existence of material objects at such worlds.

Such a response, however, very nearly begs the question against van Inwagen's approach. One might equally well claim that one can imagine lifeless possible worlds at which there seem to be objects such as rocks, mountains, etc. composed of mereological atoms. On van Inwagen's view, this would be a mistake. Surely there seem to be rocks and so forth at such a world, but really there are only mereological atoms arranged in certain ways. Van Inwagen describes how mereological atoms may come to compose a living thing, but he denies that there are any non-living composite objects. This is the crux of the controversy surrounding his view. Claiming that the picture it presents seems implausible is not enough.

Even if we admit the existence of gunk worlds, it does not follow that we must admit the existence of material objects at these worlds. There may be worlds at which there is gunk, but we need not admit that this gunk composes anything. It seems to compose some objects, but we are simply mistaken if we think this is the case. Further, since the gunk is infinitely divisible, there are no mereological atoms at such worlds. Thus, no material objects exist at such worlds. We need more than Sider's insistence here in order to claim that they do.

We can raise some other questions for Sider. For instance, what exactly is it for an object to be composed of gunk? I cannot reproduce it here, but van Inwagen gives us a rather detailed account of what it is for mereological atoms to compose living things (and why it is that composition fails for other putative objects). Roughly, it involves objects acting in such a way that they constitute the «life» of some living thing (van Inwagen [1990], pp. 81-82). Sider does not give such an account for gunk. We are left to imagine that gunk comes to compose an object in a way similar to mereological atoms. But why should we suppose that this is the case?

On the contrary, we seem to have good reasons for doubting that this is the case. As van Inwagen notes:

I assume that every material thing is composed of things that have no proper parts: «elementary particles» or «mereological atoms» or «metaphysical simples.» I suppose that questions about whether two objects are composed of or constituted by the same «quantity» or «parcel» of matter -- or «the same matter» tout court -- make sense only in the case of composite objects, and that in that case these questions must be understood as asking whether the composite objects are composed of the same ultimate parts. Thus, in my view, there is no notion of sameness of matter that is prior to or independent of the notion of sameness of objects (van Inwagen [1990], p. 5).

If van Inwagen is right about this, it is not clear how Sider can address the notion of sameness of matter in the case of gunk. Moreover, even if we grant that gunk could come to compose objects in a way similar to that of mereological atoms, this would not yield the result that tables and chairs exist at some gunk world. If matter in the form of mereological atoms cannot come to compose any non-living thing (as claimed by van Inwagen) why should we suppose that gunk can? Again, Sider needs to give an account of gunk that answers what van Inwagen calls the «Special Composition Question» (see especially van Inwagen [1990], pp. 20-22 and pp. 30-31). That is (roughly), what conditions must an object or objects satisfy in order to compose something? Sider does nothing either directly to undermine van Inwagen's treatment of composition or to give an alternative account of his own.

Sider finds it particularly problematic that van Inwagen's approach provides for so-called «virtual objects» at worlds where there are mereological atoms, but not at gunk worlds. That is, instead of claiming that the sentence:

(s1) There is a table here,

is false, van Inwagen comes up with a paraphrase that preserves the truth of the sentence. He would paraphrase the sentence s1 by a sentence such as:

(s2) There are Xs arranged tablewise here.

Strictly speaking then, there are no tables. In van Inwagen's terms, the furniture of the world has simply been rearranged without any addition (van Inwagen [1990], p. 124). Sider claims that van Inwagen's approach is unable to supply a paraphrase and hence, is unable to supply a virtual object in the case of objects at a gunk world. Thus, he concludes that there is no way to eliminate commitment to composite objects at gunk worlds because van Inwagen's paraphrase in terms of simples fails at such worlds (p. 287).

It is not entirely clear, however, what exactly Sider finds so problematic about this feature of van Inwagen's approach. Perhaps it is because sentences such as s1 would turn out to be false (while at a world where some atoms are arranged tablewise here, the sentence would be true). Thus, at some gunk world, all our utterances concerning objects such as tables, chairs, rocks and so forth would be false. Surely this is an unintuitive consequence of van Inwagen's approach, but we need not view it as undermining the approach.

Sider also claims that van Inwagen must accept (2) as a necessary truth because «his arguments seem to be based on non-contingent considerations» (p. 287). First, it is not so clear that his arguments are based on purely non-contingent considerations. Van Inwagen does cite current physical theory (see van Inwagen [1990], p. 99 for one example). Second, even if his arguments were based on non-contingent considerations, it is not clear why it should follow that he must accept (2) as a necessary truth.

On the other hand, it seems that van Inwagen's account might well be modified in order to address the case of gunk. Let us tentatively admit that there could exist gunk worlds where there seem to be material objects such as tables and chairs. Even if we admit that such worlds are possible, we might still be able to apply an analysis which eliminates putative commitment to objects such as tables and chairs. As I note above, Sider claims that we can imagine gunk being shaped into geometrical forms or even into tables and chairs. It does not follow that we must accept the existence of tables and chairs at such worlds. Suppose we take the sentence:

s3) There is a table at R,

(where R is a spacial region) and paraphrase it (in a way parallel to van Inwagen's approach) by the sentence:

s4a) Gunk is shaped tablewise at R,

or by the sentence:

s4b) Some gunk is shaped tablewise at R.

Such sentences do not either implicitly or explicitly quantify over tables. That is, it does not follow from either s4a or s4b that:

s5) There exists an x such that x is a table and x is made of gunk.

It may be that sentences such as s4a or s4b do quantify over masses of gunk, but we need not accept that such masses compose anything. Thus, we might well have a world filled with gunk, but this gunk need not be said to compose objects such as tables, chairs, or rocks.

Perhaps we would be committed to the existence of one thing at gunk worlds; namely, the scattered object composed of all the gunk at a given world. In Word and Object, Quine suggests that we might allow a term such as `water' to denote the aggregate of all the scattered bits of water in the world, «a single scattered object» (Quine [1960], pp. 98-99. See also pp. 120-121). We need not go so far as to characterize the gunk itself (at a world) as constituting some sort of blob or «blobject» to which we can refer. (Horgan [1991] addresses the notion of a blobject in the context of a Parmenidean materialism.)

Doing so, however, raises an intriguing alternative. At one point van Inwagen claims that Aristotle's view of organisms being «entirely composed of absolutely continuous stuffs» entails «that living organisms are simples» (van Inwagen [1990], p. 98). This suggests that the blobject might itself be treated as a simple. Given this, we would be committed to the existence of at most one thing at any gunk world (the blobject) since we would have a thing composed of continuous stuff and having no proper parts. Thus, we would be committed to the existence of the blobject qua metaphysical simple, but we would not be committed to things like tables, chairs, or rocks.

Need we be committed to the existence of the blobject? Here it is crucial to recognize a deep connection between the notions of countability and quantification. This point is nicely addressed by Jose Benardete when he discusses the difficulties involved in the application of Quine's criterion to mass terms (Benardete [1989], p. 35). For instance, in Thales' view, all is water. Benardete notes that Nicholas White has suggested that Thales' claim be symbolized as:

(x) (x is water),

but since water is not a count noun, it would be difficult for Thales to talk about how many things there are in the world. That is, it would not make sense for him to say how many water(s) there are. Benardete concludes that this seems to be a case where ordinary quantification is inapplicable and Quine's criterion simply does not apply (Ibid.). This is perhaps the cause of Quine's frustration with mass terms in Word and Object when he disparagingly refers to the category of mass terms as «that archaic survival of the first phase of language learning» (Quine [1960], p. 121.).<42>Foot note 2_1

We can, however, characterize Thales' claim in a way more favorable to a Quinean approach. We might take Thales' claim to mean something like:

(x) (x is entirely made of water).

Such a thesis would be true of (say) the visible, non-biological parts of an Eskimo's environment.<43>Foot note 2_2

Benardete, however, explicitly denies that either of these alternatives adequately captures Thales' position. He claims that Thales might respond to such suggestions in the following way:

But there is no number of things, absolutely speaking, though pragmatic considerations allow us to parcel out the world's water as it may suit our convenience, as we notice that here it is more condensed, there the more rarified. If on my theory there is only water and nothing else, that is not to be understood as entailing that there is either only one thing in the universe, which `the world's water' might be supposed to denote, or that there are two or more things each of which consists solely of water (Benardete [1989], p. 35).

That is, the notion of quantification over a certain class of objects is connected with the notion that those objects are in some sense countable. Given Benardete's view, this would not be the case for something like gunk.

Sider's argument against van Inwagen's approach raises some important questions for any such approach. His argument is clearly valid, but the second premise is problematic. It is difficult for Sider to support the claim that gunk worlds containing material objects exist without begging the question against van Inwagen. Further, Sider's claims against van Inwagen's approach do not hold up under careful scrutiny.

First, Sider fails to give us an account of how gunk could come to compose an object, nor does he directly dispute van Inwagen's approach to composition. Second, the existence of gunk worlds does not cause the sort of trouble for van Inwagen that Sider claims. Van Inwagen need not accept the existence of objects such as tables, chairs, and rocks at gunk worlds. Moreover, van Inwagen's approach to paraphrase can easily be modified to handle the case of putative objects composed of gunk.<44>Foot note 2_3


Benardete, José [1989]: Metaphysics: The Logical Approach, Oxford University Press.

Horgan, Terence [1991]: «Metaphysical Realism and Psychologistic Semantics», Erkenntis, 34, pp. 297-322.

Lewis, David [1991]: Parts of Classes, Basil Blackwell.

Quine, W. V. O. [1960]: Word and Object, The MIT Press.

Sider, Theodore [1993]: «Van Inwagen and the Possibility of Gunk», Analysis, 53, 4, October, pp. 285-289.

van Inwagen, Peter [1990]: Material Beings, Cornell University Press.

Kelly J. Salsbery
Dept. of Philosophy
Syracuse University
Syracuse, New York, USA