What Is Realism?

With this article, for the first time, we are studying a topic that is going to be our main concern for the last part of the semester, namely, metametaphysics, or, more specifically, metaontology. We've studied various positions about whether or not certain things are real, but we've never raised a question that should have been raised first: What do we mean by claiming that something is real? That is, Dummett, instead of arguing that something or other is or is not real, tries to characterize what it means to be real, what all such arguments are about.

I'm of the opinion that no one knows what such arguments are about, and that as a result, at least half the time, people who think they are disagreeing with each other and trying to carry on a debate are, in fact, talking past each other.

First of all, there is no such philosophical position as realism. What there is, is realism about X.

• Realists about medium-sized dry goods.
• Realists about mental phenomena.
• Realists about the objects of one or another science.
• Realists about abstract entities of one or another type: mathematical objects, universals, and others.
• Moral realism.
• Aethetic realism.

In discussing such a position, the first thing is to be clear about what an "X" is. Dummett takes it to be characterized linguistically: to be a realist is to be a realist about some class of allied statements, about a type of linguistic usage.

Historically, and even now, most who have engaged in realist--antirealist debates have taken it to be more or less clear what realism (about X) is, and so they've rarely tried to present general theories about what constitutes a realism. Dummett is one of the few and one of the first and one of the most prolific on that topic, and so, today, many philosophers who just want to get on with their real business of metaphysics just pay lip service to Dummett's characterization.

There is rough and ready characterization of realism that many would agree on, and Dummett takes it as a starting point: To be realist about X is to take it that what is true about Xs is independent of our knowledge of Xs. Thus, one test that a position is a realist position is that it follows from the position that we could, despite the best possible evidence, still be wrong. After all, if evidence is always decisive, that means that the truth about whatever is in part constituted by our knowledge of it, and hence not independent of us. The test, as Dummett shows, doesn't always work, but it is often employed and almost viewed as a truism.

The idea of independence is a delicate one, and many philosophers have thought that it poses a problem for realism about the mental. If you think, for example, that ideas are really in our minds, and that our awareness of them is what shows that they are real, then they can't possibly, it seems be independent of our knowledge of them. The test fails. The test does fail, but the basic idea works without modification: that I experience a pain is conceptually independent of the process of acquiring knowledge of the pain. Presumably, for example, some primitive animals can have pain without acquiring knowledge of it, and in certain drugged states, I can do the same. The idea is not that for something to be real it must be independent of human experience, rather that it must be independent of the process by which we acquire knowledge of it. Truth must be distinct from knowledge.

Now, what is Dummett's definition of realism about a class of statements X? It has two parts:

• Bivalence
• Reference

A statement is bivalent if it is either true or false. The law of bivalence is that every statement is either true or false, and so, presumably, the law of bivalence about X is that every statement in the class X is either true or false.

A couple of examples:

• We are not realists about the stars that dance around our heads after a blow. It is not either true or false that the number of stars is even. (Similarly, dreams.)
• Some philosophers of mathematics and mathematicians are not realists about the infinite "collection" of natural numbers, and so they don't think that the usual idea that a statement of the form "Every number is such that P" is true if P is true of 0 and 1 and 2 and ..., that is, of all the numbers, makes sense. They (constructivists, specifically, intuitionists) therefore say that to say that "Every number is such that P" is true only if we have a proof of P(x) for arbitrary x. But then it is not true, or at least not obviously true, that every such statement is either true or false.
• In the Matrix, it may very well not be either true or false that the number of chairs in the room next to this one is even.

So, it is at least clear that a failure of bivalence is connected with antirealism.

Why isn't that enough? Can one hold that a certain class of statements is bivalent, yet still not be a realist about them? Dummett says yes, and he gives Frege's artificial example of directions. Frege says (and it just an example, according to Dummett, he never really thought it worked) that "A is in the same direction as B" is true if "A is a line and B is a line, and A is parallel to B." The direction North is just given by any line that runs to the North: something is north just if it is along a line parallel to a North line. Using this method of translating, every statement about directions is equivalent to one about lines and parallelism. Every such statement will be either true or false, and so bivalence holds. Nonetheless, Frege was not a realist about directions. Why not? Because what makes statements about a direction true or false doesn't involve reference to that direction: it only involves some lines.

Thus, to be a realist about X is to not only hold that statements about X are bivalent, but also to hold that what makes them true or false is the kinds of things the X language refers to. (In addition, those references must make the statements true or false in the right way.)

-- ShaughanLavine - 10 Mar 2008