van Fraassen's defense of a modified instrumentalism

Maxwell's realism amounts, I think, to the view that we have the same kinds of reasons for thinking electrons are real as we do for thinking trees and rocks are real.

For a while, a new realism was becoming quite dominant in the philosophy of science, until van Fraassen came along and single-handedly revived a new instrumentalism.

van Fraassen makes an important concession: The new realists are right in thinking that the theoretical and observational parts of our language are intertwined, that is, in particular, there is no such thing as a purely observational claim, all language has a theoretical component.

van Fraassen makes an important clarification: Why does he think the contrast between observational and theoretical is misleading? Because theory is linguistic in nature, while observables are things in the world.

Thus, there are really two distinctions:

• The distinction between theoretical language and language used to describe observable objects.
• The distinction between observable objects and unobservable one.

The "observational/theoretical distinction" muddles that by presuming that observable objects will be described using special language, namely, observation terms or sentences and that unobservable objects will be described using theoretical language.

van Fraassen accepts that the first distinction is untenable and should therefore be abandoned, but he thinks a failure to appreciate that there are two distinctions has led philosophers to abandon the second mostly because they abandoned the first.

van Fraassen then calls to mind the distinction between observing and observing that: the first is about what is perceived, the second about what is assertible on the basis of the perception.

If you fail to distinguish seeing from seeing that, then the second distinction is hopelessly infected by the first: seeing that will always involve theory.

So, is there a usable distinction between the observable and the unobservable? Maxwell gives two kinds of examples:

• looking through a vacuum, ..., looking through a high-power microscope
• looking at an atom, ..., looking at a diamond

The first sequence is about various ways of detecting (neutral between observing and not) something. But our concern is not whether something is observed on a particular occasion, but whether it can be observed: even if you detect something using a microscope, if you can see it, it is observ able even if it hasn't been observed. (He uses the moons of Jupiter as an example.)

What about the second sequence? Diamonds are observable, atoms, let us assume for the sake of the example, are not. The example forms a sorites sequence. That is a symptom, and many would say the main test of, vagueness. The observable-unobservable distinction is vague. Like other vague terms, "observable" may be used with the line drawn in different places for different purposes (like the standard example, flat for a soccer field vs flat for a pool table). It follows that there can be no distinction in principle between observable and unobservable things.

That, which van Fraassen grants, is a central part of one of Maxwell's arguments against taking rocks to be real but not atoms. Maxwell is criticizing theories that give different semantics for theoretical and observational terms. The simplest example is the instrumentalists, for whom the observation terms refer, but the theoretical ones are "meaningless," mere tools. You can't get a sorites sequence between those, and so, Maxwell concludes, you can't make an observable/"unobservable" distinction suitable for those theories.

To devise a tenable theory of the antirealist sort, you, or van Fraassen, needs one that does not require a sharp distinction. Since observation terms may refer, van Fraassen allows that theoretical terms may refer. That is, he does not dispute that there might be such things as electrons, and that the term "electron" might be used to refer to them. What he does dispute is that we have the same kinds of reasons for thinking that there are electrons as we do for thinking that there are rocks, or, in fact, that we have any good reasons for thinking that there are electrons. Everything we can say about rocks or electrons is based on seeing that some statement or other is true. Such statements are, as van Fraassen grants, theoretical, and hence not distinguishable in the two cases.

van Fraassen proposed that all we have good reason to believe about any theory, however well verified, is that it is empirically adequate: that is that the predictions it make will hold good about what we can observe. Realists agree that a good theory must be empirically adequate. They, however, also think that we have reason to think that a good theory is true, that is, in particular that predictions it makes even about unobservables will hold.

-- ShaughanLavine - 29 Jan 2009