Science and Ontological Commitment

The section has assumed a Quinean default since the criterion of commitment: ontology is continuous with science, so we regiment our best current theory and read off what its bound variables must range over. That assumption has been doing a great deal of work — it is what makes the indispensability argument for mathematical objects go through, and what makes it legitimate to consult physics about simples and levels.

It is time to examine it. The assumption has two parts, and only the first is uncontroversial: that we should take our best theories seriously, and that taking them seriously means believing what they say exists.

The realist argument

Scientific realism holds that our best theories are approximately true and that their theoretical terms refer — that there really are electrons, not merely that "electron" is a useful device for organising observations.

The principal argument is Putnam's no-miracles argument, following Smart: realism is the only philosophy that does not make the success of science a miracle. Theories that posit unobservable entities make novel, precise, risky predictions that come true. If the entities did not exist and the theories were not approximately true, this success would be an extraordinary coincidence — as if a map that corresponded to nothing nevertheless got one to the destination every time.

The argument is an inference to the best explanation applied to science as a whole, and it inherits IBE's difficulties. Van Fraassen's objection: the argument assumes that the true explanation is among the candidates we have thought of, and we have no reason to believe it. IBE licenses at most the conclusion that a theory is the best of a bad lot. A second objection is circularity — using IBE to justify belief in the entities IBE-based science posits is question-begging against someone who doubts IBE. The realist's reply is that no non-circular defence of any ampliative method is available, so the demand is unreasonable.

Constructive empiricism

The most disciplined antirealism is van Fraassen's. Its central distinction:

Science aims to give us theories which are empirically adequate; and acceptance of a theory involves as belief only that it is empirically adequate.

A theory is empirically adequate if what it says about observable things is true. Accepting a theory is not believing it: it is believing it empirically adequate and committing oneself to using it — to speaking its language, exploring its problems, and betting on its future success. One can be a fully engaged scientist while withholding belief about what the theory says exists beyond the observable.

The consequence for this section is worth stating precisely, because it is the page's main payload. Quine's criterion tells us what a theory says exists. It does not tell us what we should believe exists. These coincide only if accepting a theory is believing it, and constructive empiricism is precisely the denial of that. So the criterion is not, by itself, a route from science to ontology; it needs the realist premise, which is a further and contested commitment.

Is the observable/unobservable line ontologically significant?

Van Fraassen's position makes everything turn on a distinction most metaphysicians regard as carrying no ontological weight whatever. Two standard objections:

Maxwell's continuum argument. There is a smooth series from seeing an object directly, through a window, through spectacles, a magnifying glass, a low-powered microscope, an electron microscope. No principled line divides observing from detecting, so the distinction cannot bear the weight placed on it.

It is anthropocentric. "Observable" means observable by us, with our sense organs. Had we evolved differently, the line would fall elsewhere — but what exists does not depend on our physiology. Making a fundamental epistemic distinction turn on a contingent biological fact looks arbitrary.

Van Fraassen's reply concedes more than it seems to. He agrees the line is vague and anthropocentric, and denies that this matters: the distinction is not meant to mark a difference in what exists but a difference in what we can check, and our epistemic situation is an anthropocentric matter. Vagueness is no objection provided there are clear cases, and there are.

Note what the exchange reveals. The realist and the constructive empiricist do not disagree about the inventory of what there is; they disagree about what we are entitled to believe about it. That makes their dispute epistemological rather than ontological — and yet it determines the ontology, because ontology proceeds from what we believe.

Instrumentalism and fictionalism

More radical positions deny that theoretical claims are even candidates for truth.

Instrumentalism treats theoretical sentences as calculating devices — inference tickets from observation to observation, neither true nor false. This has largely been abandoned, because it cannot make sense of the fact that theoretical claims are asserted, denied, embedded in conditionals, and reasoned about like any others.

Fictionalism is its sophisticated descendant: theoretical statements have truth conditions and are systematically false, but are useful. This is exactly the position Field takes about mathematics, and the parallel is instructive. Mathematical fictionalism must explain why a false theory is indispensable to true science; scientific fictionalism must explain why a false theory generates true predictions. In both cases the explanation offered is that the fiction represents structure that is really there — which concedes a good deal to the structural realist.

Where this sits

This page belongs to the status question, and it identifies an assumption the section has been making without argument. The Quinean method is not a self-standing route from theory to ontology; it delivers commitments only for those who believe the theory, and whether one should is a separate question with a live literature.

That does not undermine the method. It locates it: ontological commitment is conditional on theory acceptance, and what the criterion offers is clarity about what one takes on if one accepts. That is a genuine service even to an antirealist, who needs to know what he is declining to believe.

The next page turns from whether we should believe our current theories to what happens to an ontology when a theory is replaced.