Are There Two Types of Necessity?

Are There Two Types of Necessity?

It used to be thought there was only one type of necessity, analytic necessity. All necessity is de dictonecessity, stemming from, and about, language. There is no necessity in the extralinguistic world; to suppose otherwise is a fallacy of projection. Necessity is in the head, a product of words not things. But all this changed with Kripke’s Naming and Necessity: now we see (supposedly) that there is another kind of necessity—metaphysical necessity. This kind is de re, not always a priori, and independent of language and meaning; sentences reporting it can be and often are synthetic. Such necessities involve identity, composition, kind, and origin. They hold in virtue of an object’s nature not in our ways of describing the object. We thus live in a world that contains two radically different types of necessity—meaning-dependent and meaning-independent. As knowledge falls into two natural kinds, a priori and a posteriori, so necessity falls into two natural kinds, analytic and metaphysical. Both are types of necessity, but of quite different categories—prior to language and posterior to language, as we might say. But this mixed position leaves something to be desired: shouldn’t we be able to unify the two categories somehow? What do they have in common in virtue of which they are both called “necessary”? Please don’t say “family resemblance”; we need some account of the relevant similarities not just a blank declaration of heterogeneity. As it happens, I don’t think this is terribly difficult; in fact, so-called de dicto necessity is a special case of de re necessity. There is really only one kind of necessity, de renecessity; de dicto necessity is simply a type of de re necessity.

What are the bearers of de dicto necessity? We might say propositions, or sentences, or statements, depending on predilection. Let’s sidestep these niceties and focus on propositions (the others will naturally follow). Then consider our old friend the proposition that bachelors are unmarried males and compare this with our other old friend Kripke’s table. About the table, we have been schooled to accept that it offers necessity of identity, necessity of kind, necessity of composition, and necessity of origin (Kripke actually uses different examples to illustrate his category of metaphysical necessity, but it will be convenient to press the table into multiple modal service). The table is necessarily identical to itself; it is necessarily a table (not a worm, say); it is necessarily made of wood and of this particular piece of wood; and it necessarily has a certain origin in a specific tree (possibly also a specific carpenter). Let’s accept all that for purposes of comparison (personally, I do). Then we can ask analogous questions about our favorite analytic proposition: does it exhibit the same kind of modal profile? Well, it is clearly identical to itself, necessarily so; it is necessarily a proposition and a marital status one; it is necessarily composed of particular concepts or properties (married, male); and it necessarily has a specific origin. The last claim may seem stretched, or even a category mistake, but let us not be too literal. First, we must ask what the constituents of a proposition are and how propositions come into existence. They might be composed of senses or references: if these have origins, then the proposition has origins. The origins could just be the origins of the things the proposition is about (say, Queen Elizabeth II), or they could be origins of the concepts we apply to those objects. Putting aside Platonism about senses and concepts, the origin of these entities will involve the origin of psychological states of a certain sort, so that we will be asserting that such entities necessarily derive from the causal antecedents that actually gave rise to them (or their underlying brain states). We need not quibble about how exactly this goes, accepting that somesort of origin story must hold of them, and that this will mirror what holds of states in general (the state of being frozen, say, could not have derived from heating the substance involved). If we are very picky about this, we could always drop the necessity of origin for propositional constituents and stick to references of such constituents, or else allow that origin doesn’t apply to propositions (like numbers or Platonic forms). The point is that modal questions of the usual four kinds can be raised about propositions (sentences, statements): for these entities have essential natures too–involving identity, kind, composition, and origin (or the lack thereof). Thus, propositions are subjects of de re metaphysical necessity (and contingency). They also have a grammatical or logical form as a matter of metaphysical necessity: the proposition that all men are mortal is necessarily a universally quantified proposition (it couldn’t have been an existentially quantified proposition, i.e., that proposition). Propositions are things with properties, so they raise the same kinds of modal questions as other things, with the same kinds of answers.

One of the properties that the proposition that bachelors are unmarried males has is the property of being true. Evidently, it is true in virtue of being the proposition that it is (and not, say, the proposition that bachelors are miserable). So, we can say that an essential property of this proposition is that it is true—as it is an essential property of the table that it is made of wood. It has a number of essential properties, as just discussed, and this is one of them. Being necessarily true is like being necessarily made of wood—something that follows from the nature of what instantiates it. There is no departure from standard metaphysical necessity here: metaphysically, being necessarily true is like being necessarily made of wood—an object necessarily having a property (a thing necessarily having an attribute). The apparatus that has been developed to talk about de remetaphysical necessity (and contingency) carries over smoothly to so-called analytic necessity, with propositions taking the role of res. The right thing to say is that metaphysical necessities can concern physical objects, artifacts, organisms, mental states, numbers, geometrical figures, and propositions (sentences, statements). Analytic propositions are distinguished by the fact that they are always true (in all possible worlds) as a matter of meaning, but this is not a new type of necessity: it is simply the old type of metaphysical necessity applied to a different class of entities. It is the same with contingency: propositions are contingently true in virtue of their nature (necessarily so). And just as every entity has both necessary and contingent properties, so too do propositions: it may be contingently true of a given proposition that it was uttered five times on a single day, or that its utterance incited a riot. Language and meaning are things like other things and therefore have contingent and essential properties. They are part of “the world” and share its modal proclivities. So far from excluding necessity from the world and confining it within language, analytic necessity is simply part of the world. Once the world was created its necessities were created, and once language was created its necessities were created. These are all de re metaphysical necessities. Language did not create necessity in the world, as the world did not create necessity in language; but both are necessities of nature, part of the modal structure of reality.[1]

[1] If we ask what it is that makes the proposition that bachelors are unmarried males necessarily true, the answer is that the concept bachelor is composed of the concept unmarried and the concept male; similarly, if we ask what makes the proposition that water is H2O necessarily true, the answer is that water is composed of hydrogen and oxygen. The cases are exactly analogous. Modally, reality is homogeneous, not dualistic. It isn’t, say, that analytic necessity is a feature of the use of words while metaphysical necessity concerns facts—all necessity concerns facts (linguistic and non-linguistic).