Theory-Iadenness

Instrumentalists believe only in the existence of observable things and their observable properties. But the distinction between what we can and cannot observe is relative.

There is, in principle, a continuous series beginning with looking through a vacuum and containing these as members: looking through a windowpane, looking through glasses, looking through binoculars, looking through a low- power microscope, looking through a high-power microscope, etc., in the order given. The important consequence is that, so far, we are left without criteria that would enable us to draw a non-arbitrary line between “observation” and “theory.”

This continuum is not very worrying in itself. But it is rather trou­bling to suppose, as the instrumentalists do, that whether we should say that something exists depends on the apparently arbitrary ques­tion of where in this continuum we draw the line.

This objection to instrumentalism is one of the reasons that many philosophers have given it up. But a much more basic objec­tion than this has been developed in recent years, one that grows out of the work of the American philosopher Russ Hanson. His objection, put at its simplest, is that there is no such thing as an observation language! If Hanson is right, then the idea that we should regard only the sentences of the observation language as true would have the consequence that we would have to regard all theories as untrue. And that would, surely, be a reductio of the instrumentalist position.

To understand Hanson's view, we must remember how we defined the distinction between observational and theoretical terms.

To see why Hanson thought this, it helps to begin by noticing that whenever we see something we also see that something. When I see a ripe apple, I see that there is a ripe apple before me. You cannot observe something without observing that a certain state of affairs obtains. But when I see that something is an apple, this commits me to believing something beyond what I have actually observed. It commits me to believing that, if I stretch out my hand, I will be able to touch it, for example; it also commits me to believing that it grew on an apple tree. (You might like to consider how this fact is con­nected with Frege's discovery of the primacy of the sentence; see 3.4. We can't use names except in sentences; we can't experience the referents of names except in the context of facts.)

Now, though we would not normally say that

Things that look like this apple are ripe apples and grow on trees

is a theory, it is a theory in the philosopher's sense. It says something about the world, something that might or might not be true. To make the observation statement “This is a ripe apple,” on the basis of this experience, you have to suppose that this little theory is cor­rect.

The instrumentalists might argue, at this point, that I have cheated.

Sellars attacked the myth of the given, arguing that belief in this myth results from a confusion between having a sense experience and making a judgment on the basis of it. (This is a distinction that goes back at least to Immanuel Kant.) When I see something red, I have a certain experience, and the experience might indeed be inde­pendent of any other experiences. But I also make the judgment that I am seeing something red. It is that judgment on the basis of which I make further judgments (the stoplight is shining, say). To make that judgment, however, I must be able to apply the concept red not just on this occasion but on others. (Otherwise I am not using the concept correctly.) That capacity is not independent of other experiences, and the connection between different experi­ences it presupposes requires some theory.

These arguments are difficult but, I think, persuasive. Still, even if they were not, there is an overwhelming reason not to require observation statements untainted by theory as the basis for your phi­losophy of science. For even if there were things we could know on the basis of no theory at all, they would not be the sorts of things that science is concerned about. To see why this is, consider a sen­tence, S, which is supposed to be one that we can make on the basis of observation without any theory at all. Suppose that you are hav­ing the experience that justifies you in believing S is true.

Hanson's view that every observation statement depends on some theory, however simple it is and however convinced we are that it is true, is called the view that observation is theory-laden. (Hanson actually used the term “theory-loaded,” but it didn't stick!) Observation is theory-laden, because whenever we make a judg­ment on the basis of our sensory experience, the judgment commits us to the existence of objects, events or properties that go beyond that evidence. This fact, that evidence always leads us to make claims beyond the evidence, is called the “underdetermination of empirical theory.” The contents of our empirical beliefs are not fully determined by the evidence we have for them. There is an obvious connection between the underdetermination of empirical theory and the defeasibility that we noticed (in 2.3) as a characteris­tic of our judgments about the world. Just because our empirical claims always go beyond the evidence, they could always turn out later to be wrong. The sight of an (illusory) apple could fail to be fol­lowed by the feel of an apple when you stretch out a hand.

The theory-ladenness of observation threatens the received view because the received view depends on making a distinction between the observation language, on one hand, and the theoretical lan­guage, on the other.

But Hanson made a more radical suggestion than this one. He suggested that even those sentences whose truth value we can decide easily by using our senses change their meaning when we use them in connection with new theories. I suggested that terms such as “gene” got their meaning from something like a Ramsey- sentence—in other words, that their meaning is fixed by their rela­tionships with terms for things that we can observe. Hanson sug­gested that the converse holds: what observational terms mean depends on their connections with theoretical terms also. Whenever there is a change of theory, all terms, including relatively observa­tional ones, change their meaning.

The meaning-variance hypothesis, if true, would threaten the DN model of theory reduction. For example, when we came to use the chromosome theory to derive, say, an observational law of MG, we would be trying to derive a sentence that used “pea” to mean one thing from a theory that used “pea” to mean something different! And, obviously, in a valid deduction you have to keep the meanings of words constant throughout the argument. (Not to do so is a mis­take that has a name: it's called a “fallacy of ambiguity.”) It would follow that we could not give the rather natural explanation of how science progresses that went with the received view.

Fortunately, there are serious problems with the meaning-vari­ance hypothesis. The main objection to the meaning-variance view is that Hanson offers no grounds for thinking that every term must change its meaning with a change in our theories about that thing. If this were right, then every time we changed our beliefs about any­thing, that would involve changing the meanings of all the sentences about that thing. Someone who came to believe that water is H₂O would have to mean something different by “Fill the bath with water, please” from someone who didn't believe it. But this is a reductio of Hanson's position. For it follows from his view that you and I would mean different things by most of the words we use, since we certainly differ in some of our views on almost every sub­ject.

Nevertheless, Hanson's position does make us conscious of the possibility that as our theories change, some of our words do change their meanings. Mendel may have meant the laws of segregation and independent assortment to be part of the definition of a gene. If that is so, then his theory is not true of what we call “genes.” For, on our meaning of the word “gene,” some genes do not obey both these laws. We would have to say that Mendel's views were about genes on different chromosomes. But we could still say both that the chromosome theory was an addition to the knowledge acquired by Mendel and that the chromosome theory explained his theory's suc­cesses. For, if we said that his word “gene” referred to what we call “genes on different chromosomes,” we would be able to derive his laws from our theory.

4.9