Predicating and Necessity

Predicating and Necessity

(Bear in mind Kripke’s Naming and Necessity when reading the following.) When a speaker uses a predicate (“man”, “cat”, “rose”, “square”, “red”, “runs”, “clever” etc.) he or she refers to a property or attribute: but how is this done? One answer is by means of ostension: the speaker points to an instance of the property, being momentarily acquainted with it (as Russell would say). Another answer is that the speaker has in mind a description of the property or attribute: “large-brained speaking biped”, “furry feline with sharp claws”, “nice smelling pretty flower”, “four-sided figure with equal angles”, “the color of British mail boxes”, and so on.[1] The first theory could be called the direct reference theory; the second theory could be called the description theory. The meaning of the predicate is given by the property it directly refers to, or by the descriptive concepts the speaker uses to identify that property.[2] There is descriptive mediation, or there is not. The proposition expressed either contains the reference directly and intrinsically, or it contains only the concepts used by the speaker to latch onto that reference. The description theory of predicate meaning looks plausible and powerful—it is hard to see how it could fail to be the case (we can use predicates and not be in a position to point to instances of the properties they denote). Yet the theory seems demonstrably false; and the arguments against it are quite obvious. First, speakers can make mistakes about the properties they refer to—for example, British mail boxes might not be red (this is an illusion they give off). Second, speakers may not have sufficient information to identify the property uniquely—many types of flowers can be described as pretty and nice smelling. Third, it is not generally analytic to couple a predicate with a description of its denotation—it may just be a contingent empirical fact. Fourth, a predicate “rigidly expresses” its associated property but a speaker’s descriptive beliefs about it are generally not thus rigid—in some possible worlds, cats don’t have sharp claws, though another species may. Fifth, at some point we will reach descriptions that have no definition in terms of other descriptions, so that the predicates used in the descriptions will not be explicable in terms of the description theory. Sixth, syntactically speaking simple predicates are not complex in the way envisaged by the description theory—they are not pieces of shorthand, not disguised descriptions. Seventh, nothing like this actually passes through the mind of the average speaker: he or she just doesn’t bring to bear this sort of descriptive knowledge. Eighth, the whole model of descriptively mediated reference smacks of overgeneralization: definite descriptions work by singling something out by means of uniquely identifying description, but not every device of language works like a definite description—take proper names or demonstratives or pronouns or sentence connectives. Thus, the description theory of predicating runs into decisive refutation, despite its apparent attractions.

I have just run through Kripke’s arguments (and some others) against description theories of names for the case of predicates (common nouns, adjectives, verbs). I said nothing about ordinary proper names. So, those arguments have nothing specifically to do with names; they apply also to predicates. It isn’t that namesconstitute a special problem for a description theory of reference; predicates do too. The substance of Naming and Necessity could have been presented under the title Predicating and Necessity. Kripke himself extends the doctrines he puts forward to the case of “common names”, so he implicitly acknowledges (indeed asserts) that his critique applies also to this category of expression. But he could have proceeded by first mounting his critique for the case of predicates and then extending it to proper names. He could have begun with natural kind terms and moved on to the case of names of individuals. And, if the description theory doesn’t work for descriptive predicates, it is hardly going to work for non-descriptive names. Neither category of expression is to be understood as abbreviations of descriptive definitions. On the positive front, there is nothing to prevent Kripke from offering his theory of initial baptism and chain of communication in the first instance to predicates, then extending it to proper names. Long ago the English language baptized squareness “square”, then the word was passed along from person to person in a reference-preserving chain—with no descriptive reference-fixing knowledge necessary. First, a causal theory of predicates; then, a causal theory of names. There is nothing distinctively name-oriented about this conception. The focus on names is entirely adventitious in these debates. In fact, names are a relatively marginal feature of natural languages; predicates are where the real work gets done. For some reason, proper names became the focus of discussion, going back to Mill and Russell (not so much Frege), and Kripke is simply following in this tradition; but it gives a skewed impression of what the real issues are. If Kripke had not extended his discussion to the case of common nouns, he would clearly have distorted the import of his arguments—as if they concerned only the very limited part of language comprised of proper names (of people and places). In fact, they apply to a wide region of language—not only common nouns but also predicates in general. The title should have been Naming, Predicating, and Necessity(or better Names, Predicates, and Necessity, since Kripke’s book is not so much about naming as an action as about names as a semantic category).[3]

Finally, names and necessity: does the former have anything particularly to do with the latter? It does not. This is by now old hat: names have no more to do with necessity (epistemic or metaphysical or analytic) than other classes of expression—definite descriptions, indexical words, predicates, connectives, quantifiers. In particular, definite descriptions can be as rigid as names (“the successor of 3”). De re necessity has nothing to do with names as such, being quite independent of language. We can express the necessity of water being H2O either by using “Water is H2O”, where the terms are used as names (singular terms), or by using “Anything that is water is H2O”, where the terms are used predicatively. We can either use “Heat is molecular motion” or “If something is hot, it has high molecular motion”. Both formulations express the fact that a certain de renecessity obtains. When Kripke begins his lectures by saying that he hopes people see some connection between the two topics of the title, he is being misleading (and has misled many readers). I suppose there is “some connection” (everything is connected to everything else somehow), but there is no special proprietary connection between names and necessity—whether the necessity is analytic or synthetic, de dicto or de re. It is not even true that names form necessarily true identity statements but descriptions never do, since descriptions can be rigid (as well as occur with wide scope). When I say “Nothing can be red and green at the same time” there is no name in sight, only general color predicates, yet my statement is as necessary as can be; so yes, there is “some connection” between predicates and necessity! As there is “some connection” between connectives and necessity (look at the theorems of propositional logic). If we read Naming and Necessity as claiming that names are uniquely not open to a description theory and uniquely connected to necessity, then we read a lot of error into that classic text; but I don’t think anything important hangs on those claims, mistaken as they are. It can all be rephrased to avoid such mistakes. Surely Kripke would agree.[4]

[1] A variant of this approach, Russellian in spirit, would opt for descriptions that refer only to sense-data, as in “the animal species that causes such-and-such sense-data” or “the property that seems thus-and-so”. There is something empiricist about the traditional enthusiasm for description theories.

[2] The famous deeds version of the description theory of names of people finds a parallel in the “conspicuous instance” description theory of predicates, as in “the shape of the earth” for “spherical” or “the kind of animal that my pet Tabby belongs to” for “cat”. This is the kind of knowledge that ordinary speakers may be expected to possess.

[3] We could also choose to treat names as predicates: the name “John Smith” is parsed as “a John Smith” or “he John Smiths”. Then we would have assimilated names to predicates, treating predicates as basic in logical form.

[4] It is puzzling to me why Kripke falls into these incautious formulations, or fails to warn against natural but incorrect interpretations of his words and procedure. It is difficult to believe he didn’t see the points I am making, which are hardly earth-shattering. Nor was he much of a respecter of intellectual tradition. Let me also add that his classic discussion of the identity theory of mind and brain does not depend on the assumption that “pain” is a name as opposed to a predicate.