Michael Kremer:

Comments on Klima, Contemporary "Essentialism" vs. Aristotelian Essentialism

Gyula begins with a contrast between contemporary scare-quotes essentialism and Aristotelian full-blooded essentialism. The former is a semantic thesis couched in the vocabulary of possible-worlds semantics, holding that some terms are rigid designators, while the latter is a metaphysical thesis, couched in a more ancient vocabulary, holding that things have essences. Gyula argues that the more traditional metaphysical framework deserves reconsideration, both because it can help us with problems arising from the contemporary approach, and because it possesses greater expressive power than the contemporary approach. He presents a fragmentary formal semantics for the traditional approach, and argues that this semantics enables us to see how the problems of contemporary essentialism can be avoided while at the same time other properly metaphysical issues, which are unapproachable from within the contemporary model due to its expressive weakness, become available for investigation. The semantics that he presents seems intended to establish the intelligibility of the traditional vocabulary, and so of the thesis of Aristotelian essentialism which is couched in it. Gyula then argues that this thesis is in fact true, using the vocabulary which the semantics has rendered legitimate.

The primary problem which Gyula identifies with the contemporary approach is that it gives too weak an analysis of the notion of essence. A typical analysis within this approach will look something like this: (See (1) on your handout)

(1) x is essentially F = necessarily, if x exists then x is F = x is F in every world in which x exists.

Such an analysis has the defect, among others, that everything is said to exist essentially. Yet intuitively few beings are so fortunate as to count existence as part of their essence. The traditional view sketched out semantically by Gyula allows us to distinguish two senses of "essence." The first of these corresponds to the contemporary sense. But the second sense of essence is much stronger, and in this sense it need not at all be the case that existence is essential to a thing. This second sense of essence is defined in terms of the idea of a "substantial predicate." (See (2), (3) and (4) on your handout.) As I understand it, the idea is this: "human," say, is substantial just in case u's existence = u's humanity's existence, for any u. Then for humanity to be part of the essence of Socrates, say, it must be the case that Socrates is human and humanity is a substantial property. In that case we can say that for Socrates to be is for Socrates' humanity to be, or in other words for Socrates to be is for Socrates to be human. Now apparently there is no need that "exists" should be a substantial predicate, and so in this sense of "essence" the existence of u need not be essential to u.

Let me pause here to register a slight worry. The worry is that it might seem arguable that "exists" is a substantial predicate after all. Gyula appears to admit this in his first argument on p. 15. By Gyula's definition: (See (5) and (6) on your handout)

(5) "exists" is a substantial predicate iff SGT("exists")(SGT("exists")(u)(t))(t) = SGT("exists")(u)(t)

This I take to mean something like the following:

(6) u's existence's existence = u's existence

which to me at least has something of the ring of truth. In any case if "exists" is not to turn out to be a substantial predicate (and so essential to all things that exist) Gyula will have to deny this. Perhaps this is no problem however; just as it seems we can deny that u = u's existence, so we can deny that u's existence's existence = u's existence, since these are the existences of distinct though interdependent things -- I have perhaps just confused a necessary interdependence for identity.

What I've just read was based on the version of the paper I had in preparing these comments. In the final version Gyula avoids this by excluding "exists" from the definition of substantial predicates. However this move seems an ad hoc solution. This thought seems to me to be supported by the following bit of history: in the original version of the paper Gyula calls `exists' a substantial predicate on p. 15. In that version, his first argument for essentialism began "For suppose there is a substance the only substantial predicate of which is `exists'..." This is modified to "a substance that has no substantial predicates" in the final version. This suggests that it was not his intention all along to exclude `exists' from the definition of `substantial predicate.'

Passing on from this worry let me address Gyula's claim that this second sense of "essence" and the notion of substantial predicate on which it is based, is beyond the expressive reaches of the contemporary possible worlds framework. I do not see that this has been established. In 1980, in fact, Anil Gupta (following up ideas of Hintikka, Thomason and especially Aldo Bressan) published The Logic of Common Nouns, a in which he introduces a distinction between common nouns and other predicates. According to Gupta common nouns express "sorts," which are distinguished from the "properties" expressed by ordinary predicates, by involving a principle of identity as well as a principle of application. For example we can meaningfully ask not only whether x is a dog but also whether x is the same dog as y. Among sorts he distinguishes "substance sorts" which he suggests pick out essential properties of things. These have to be "modally constant" -- roughly speaking they are "rigid" in the strong sense that both their principle of application and their principle of identity is invariant from world to world. Importantly, on Gupta's view it is quite possible that existence is not a "substance sort." This is because it is plausible that existence is not a sort at all; for existence to be a sort we would require a principle of identity for existing things, which would determine for any existents x and y whether they were the same existent. If such a principle is not available existence fails to be a sort, and so fails to be a substance sort.

Gupta shows how to model these distinctions using apparatus available within the possible worlds framework. (There is not time here to give details though I'd be happy to pursue this in the discussion.) Gupta then goes on to characterize Aristotelian essentialism in the following terms: the use of common nouns commits us to a weak grade of essentialism, according to which we can say of a thing that it is essentially F qua K; essential predications are relativized to a principle of identity. It is the principle of identity supplied by "human" that makes it the case that I, qua human, am essentially rational -- nothing lacking rationality can be the same human as me. Aristotelian essentialism for Gupta is the thesis that "for each object there is a sort that answers the question, What is it?" This allows us to make sense of the claim that x is essentially F, simpliciter, as the claim that x is essentially F qua K, where K is the privileged sort which tells us what x is.

The point of all this in the present context is that Gupta's logical apparatus, developed entirely within the possible-worlds framework, provides that framework with expressive resources which Gyula ignores when he claims that "due to the inherent non-metaphysical origins of this framework, lacking the required expressive devices ... it is just not the "natural playground" for such metaphysical inquiries." (pp. 14-15)

Gyula concludes his main argument by presenting four metaphysical arguments in favor of Aristotelian essentialism. If correct these arguments should also lead us to accept the thesis of contemporary "essentialism." I will conclude by briefly considering these arguments. It seems to me that argument 1 is fallacious and the other three systematically beg the question against the nominalist at best. Consider first argument 1.

1. If x has no substantial predicates other than "exists" then x has no essence and "so" could have no true predicates other than "exists" at all. This involves an obvious howler. From the assumption that for any F, x's existence is distinct from x's F-ness it may follow that: (See (7) and (8) on your handout) (7) for any F, possibly x is not F. But Gyula's conclusion is that (8) possibly, for any F, x is not F. And this does not follow from (7). The fallacy here is much like that of inferring from "x is possibly not F" and "x is possibly not G" to " x is possibly neither F nor G."

Arguments 2-4 on the other hand all seem to me to beg the question. Consider argument 2, for example:

2. For a living thing to be is for it to live, etc. This argument assumes that for something to be is for it to be "what it is" in a sense answering to Aristotelian essence, so that for a thing to change without ceasing to be "what it is" in this sense is for it to continue to exist. But what is assumed here is precisely the point at issue. I would contend that the same criticism applies, with slight modification to arguments 3 and 4 as well. Consequently Gyula may have succeeded in making the thesis of Aristotelian essentialism intelligible (but this may be possible also using the tools of possible-worlds semantics); but he has not proven the thesis to be true.

To sum up, I have raised three issues: (1) is existence really not essential, by Gyula's definition? (2) is contemporary essentialism necessarily as impoverished as Gyula claims, or is it only his representation of it which is impoverished? (3) Do Gyula's arguments for Aristotelian essentialism work? It is a measure of the richness of this paper that I could have raised a half a dozen other issues instead. But my time is up!