HOW SKEPTICAL IS QUINE’S “MODAL SKEPTICISM”?

I. THE LOGISTICAL EXPLICATION OF MODAL METAPHYSICS

There is an ingenuous and sanguine conception of metaphysics that has regained the status of orthodoxy in our time even though it looks back, beyond Kant, to Aristotle and his precursors. In the old and new orthodoxy of classical, unreconstructed metaphysics, we aim to characterize systematically how things fundamentally are (independently of our conception of them): or, giving even free(r) reign to metaphor, to carve reality at the joints. Skepticism about this classical metaphysical project is a prominent element of the empiricist tradition. In the most extreme examples of empiricist skepticism, the positivists (logical or otherwise) aim to bury such metaphysics. But moderate empiricist skepticism regards the burial as premature, while also refusing to praise classical metaphysics in its unreconstructed form. It is on this moderate, and revisionary, part of the spectrum of empiricist skepticism about classical metaphysics that we find Quine, as I understand him.

To borrow a term from his mentor Carnap (1947, 8), Quine’s aim is to explicate “metaphysics” rather than to eliminate it. As sense-making animals and charitable interpreters of our ancestors- we understand classical metaphysics, and its practitioners, best by treating it as a protoscientific project. Explication requires that we establish continuity between that which is prescientific (old and inferior) and that which is scientific (new and superior).

The crucial dimension of continuity is not that of doctrine, nor that of method but, rather, that of aim. And the continuing aim that enables the explication of the classical metaphysical project, by affording a reconstruction entirely within science, is (broadly) this: to offer the optimal general and systematic characterization of all that there is and how it is. As such an explicator of classical metaphysics,

Quine finds himself able to share with the classical metaphysician a common language in which one can meaningfully speak of, and properly dispute, for example: the existence of natural numbers, the nature of attributes and-to bring us to our present topic-the presence of modal features in reality.The Quinean method of explicating classical metaphysics is logistical: we proceed by transforming questions in the material mode (those about numbers, attributes, modal features of reality, etc.) into questions in the formal mode (those about symbols).

When the logistical method is applied to explicate that part of metaphysics that is ontology, wherein Quine is anticipated by Frege (1884), questions about the existence of given objects such as numbers are, of course, not to be taken as equivalent to questions about the existence of the relevant symbols (numerals). The logistical approach is methodological rather than reductive: it guides us towards the appropriate symbols and the kinds of question about them that we have to answer in order to establish knowledge of what there is, and of how things are, beyond the symbols. A logistical explication of ontology has two elements.

The first element is the identification of (what I shall call) the telling discourse. This tells us where to look to settle questions of ontology. The second element is the proposal of a particular syntactic criterion that is to be applied to the telling discourse. This tells us how to find there that which is ontologically significant.Quine’s earlier and lesser-known paper on the explication of ontology, originally published in 1939, has a title that makes explicit his endorsement of this method- that is: “A Logistical Approach to the Ontology Problem” (1976a).1 The application of the method comes to maturity in its more famous successor, originally published in 1948, “On What There Is” (1953a). Quine takes the ontologically telling discourse to be the optimal formulation of best total science. Optimality is primarily a matter of simplicity in various respects, and the deployment of a canonical notation whose terms are primitive (not further definable).