According to Quine, ontological commitment is a matter of theory selection and quantification. We identify our best scientific theories of the world, look to what they quantify over, and these are the things that we take to exist.

Quine rejects wholeheartedly the idea that ontological commitment can be “read off” of ordinary subject-predicate sentences; that from “The table is red,” one can conclude that redness exists or from “Four is an even number” that numbers exist. This is because surface-grammar notwithstanding, one can interpret these sentences in ways that neither the subject nor the predicate refer to an actual entity. Only explicitly quantificational sentences straightforwardly entail ontological commitments, according to Quine, and which quantificational sentences we should accept depends upon whether or not our best scientific theories must include them. 

Hofweber counters that it certainly looks as if ontological commitments can be read off of ordinary subject-predicate sentences and that it is not at all clear that explicitly quantificational statements are always indicative of ontological commitments. He makes these points by way of an analysis of some very interesting ways in which we use certain kinds of sentences.

Take the following, which Hofweber refers to as “metaphysically innocent,” insofar as it suggests no particular metaphysics:

A.  Fido is a dog.

Now, consider this sentence:

B.  Fido has the property of being a dog.

Hofweber describes this sentence as being “metaphysically loaded,” as it would seem to suggest a metaphysics and specifically, a metaphysics of properties. Notice that this sentence is not only entailed by the first but is materially equivalent to it. That is, it has precisely the same truth-conditions: both (A) and (B) are made true and false by the same fact, namely, whether or not Fido is a dog.

The question of how a metaphysically innocent sentence can be so easily and smoothly translated into a metaphysically loaded sentence strikes me as interesting in itself, especially considering their material equivalence. Certainly, there is disagreement as to whether truth-conditions tell us all there is to know about a sentence’s content, but there are many highly regarded thinkers who say that they do, including Quine, Davidson, and every other proponent of so-called “extensional” semantics, and regardless, it’s not at all obvious how two statements that express exactly the same fact can have such radically different implications.

But things get even worse, for from (B), one can deduce:

C.  There is a property that Fido has.

Which is a straightforward quantification over properties that would seem impossible to interpret away, as we did with (A). The mystery is compounded, for (C) is straightforwardly entailed by (B) (and via (B), from (A)), and is also materially equivalent to both (A) and (B). It too is made true or false by whether or not Fido is a dog. So, from a metaphysically innocent statement, one can deduce an explicitly, unavoidably metaphysical one, with a very specific ontological commitment, one that is materially equivalent to its metaphysically innocent predecessor. Curioser and Curioser.

Logic aside, what Hofweber wants to know is what people actually use constructions like (B) and (C) for and whether, typically, they intend to signal ontological commitment when they do so, and here his analysis gets really interesting. Not only because of what it tells us about ontological commitment, but because of what it reveals about the many and fascinating ways in which we use language.  

Let’s look at another set of statements:

(D)  I had two bagels.

(E)  The number of bagels I had is two.

(F)  The person who had two bagels is me.

Like our first group of statements, (D), (E), and (F) are materially equivalent. Their truth or falsity is a matter of whether or not I ate two bagels. But notice that while (D) states this fact neutrally, (E) and (F) state it with a different focus. (E) Stresses how many bagels I had and (F) stresses that it is I who had them.  

In speech, this sort of focus is typically effected phonetically: by way of inflection, tone, volume, cadence, etc. In writing, the only way to do it is either through markings whose purpose is to simulate phonetic effects – italics, boldface, underlining and the like – or syntactically, which is what we see here.

Focus has a substantial effect on communication, with the effect of these sentences varying widely, despite their common truth-conditions. One important effect is to bring a particular element into a position of special prominence. Consequently, one statement may be appropriate in a given context, while another may be completely inappropriate, even if, as a matter of propositional content, the two statements state exactly the same fact. If, for example, I asked you:

(G)  Who had two bagels?

(E) would be an inappropriate answer, even though, truth-conditionally speaking, it is identical to (D) and (F), both of which would be appropriate. This is because of the relevant syntactical focus of (E) on the how many, rather than on the who, which is what (G) is asking about.

Notice something else: That I aim to focus on one element or another has entirely to do with the sort of conversation I am having, and it is not at all clear that in doing so, I am also intending to commit to entities. I don’t make the focus that I do in (E) in order to ontologically commit to numbers or in (F) to ontologically commit to persons. I do so in order to stress certain aspects of the fact that I am describing, whether to appropriately answer certain questions, perhaps forestall certain questions, make certain things clear that perhaps were unclear, etc., but certainly not because I want to advance a Platonic view of numbers or a Cartesian view of personhood. Hofweber’s analysis, then, would seem to explain the relationship between metaphysically innocent and metaphysically loaded statements, without lending any support to the suggestion that beyond intended shifts in focus, the latter represent any sort of desire to commit ourselves ontologically.  

The problem still nags at us, however, because of the quantificational statements that also seem to follow:

(H) There is a number, which is the number of bagels I had, namely two.

Now, this statement is straightforwardly metaphysical. It is unambiguously ontological. And it follows, logically, from (D)and is materially equivalent to it (and to (E) and (F)). From the standpoint of just what is involved in – and what involves us in — ontological commitments, it represents a tremendous problem, insofar as such sentences can be generated out of every ordinary subject predicate sentence, with all the ontologically inflationary implications that follow.