Phi 272 F06, reading guide

This last week of the first part of the course brings us to what was probably the key concern of the philosophy of science for most of the twentieth century, the nature of theories.

Ch. 23. Carnap frames his introduction to the distinction between theories and other parts of science in ch. 23 as a distinction between theoretical laws and empirical laws. The various issues surrounding this distinction between the theoretical and the observable will probably form the heart of our discussion of this chapter.

However, not everyone would accept Carnap’s assumption that a theory consists of laws. The most common alternative view sees a theory as given by a model or collection of models. Such models are often very abstract, so they must be presented by descriptions that could be seen as laws; but, on this alternative to Carnap’s view, the theory is to be found in the model, not the description of it.

Another respect in which this view differs from Carnap is that while, for him, laws provide the only grasp we have of the theory, someone who holds that theories are models may hold that one route to the model is by way of analogy. The classical example is the kinetic theory of gases, which Carnap describes in the second paragraph of ch. 24 (pp. 232f). The analogy here is with ordinary physical objects, and it is only an analogy because certain features of these objects (for example, the material composing them) may be irrelevant for the model. One of classic presentations of the view of theories as models, which focuses on the idea of analogy, is quoted at the end of this guide.

This disagreement about the nature of scientific theories is related to the difference between positivists and realists. A positivist thinks the first question we should always ask is how an aspect of science is connected to our sense experience, and the connection between nonobservable entities and sense experience comes by way of laws that enable us to predict some sense experiences on the basis of others (or enable us to derive further laws that do this). On the other hand, a realist thinks that the first question we should ask about science is the sort of underlying reality that science is uncovering, and a model provides a way of pointing to that reality by way of analogy.

So, in addition to thinking through the difference between the theoretical and observable as Carnap discusses it, ask yourself which of these two ways of understanding the theoretical seems preferable. Should we think of laws as primary, models serving merely as a way of suggesting or illustrating laws, or should we think of models as primary, with laws serving merely as a way of describing models?

Ch. 24. Although Carnap agrees with someone who thinks of theories as models in holding that theories have a substantial degree of autonomy in relation to the laws that may be derived from them, he also insists that there must be some connection between the nonobservable vocabulary of the theory and the observable vocabulary used to state empirical laws. He describes this views on this connection in ch. 24. Here he returns to his differences from Bridgman from a different point of view, saying that he does not believe we should seek to define all the vocabulary in observable terms.

Positivists have sought to define all scientific terms observationally as a way of distinguishing science from metaphysical speculation. Do you think the partial interpretations by way of correspondence rules that Carnap recommends does enough to ground science in sense experience? In the period after Carnap wrote this book, philosophers began to ask increasingly whether an attempt to distinguish science and metaphysics was even worthwhile. While few would see no difference between natural science and metaphysical speculation, many held that metaphysical assumptions played an important role in scientific theories and that it was impossible to distinguish this aspect of theories from their empirical significance. Where would you stand on this issue? Do you think the attempt to distinguish science from other sorts of theoretical activity is feasible and valuable?

… I said just now that it was a commonplace that analogies were important in the framing of hypotheses, and that the name “hypotheses” was usually given in this connection to the propositions (or sets of propositions) which are here termed theories. This statement is perfectly true, but it is not generally recognised by such writers that the “hypotheses” of which they speak are a distinct class of propositions, and especially that they are wholly different from the class of laws; there is a tendency to regard an “hypothesis” merely as a law of which full proof is not yet forthcoming.

If this view were correct, it might be true that the analogy was a mere auxiliary to the discovery of laws and of little further use when the law was discovered. For once the law had been proposed the method of ascertaining whether or no it were true would depend in no way on the analogy; if the “hypothesis” were a law, its truth would be tested like that of any other law by examining whether the observations asserted to be connected by the relation of uniformity were or were not so connected. According as the test succeeded or failed, the law would be judged true or false; the analogy would have nothing to do with the matter. If the test succeeded, the law would remain true, even if it subsequently appeared that the analogy which suggested it was false; and if the test failed, it would remain untrue, however complete and satisfactory the analogy appeared to be.

… But a theory is not a law; it cannot be proved, as a law can, by direct experiment; and the method by which it was suggested is not unimportant. For a theory may often be accepted without the performance of any additional experiments at all; so far as it is based on experiments, those experiments are often made and known before the theory is suggested. Boyle’s Law and Gay-Lussac’s Law were known before the dynamical theory of gases was framed; and the theory was accepted, or partially accepted, before any other experimental laws which can be deduced from it were known. The theory was an addition to scientific knowledge which followed on no increase of experimental knowledge and on the establishment of no new laws; it cannot therefore have required for its proof new experimental knowledge. The reasons why it was accepted as providing something valuable which was not contained in Boyle’s and Gay-Lussac’s Laws were not experimental. The reason for which it was accepted was based directly on the analogy by which it was suggested; with a failure of the analogy, all reason for accepting it would have disappeared.

…

… A theory is valuable, and is a theory in any sense important for science, only if it evokes ideas which are not contained in the laws which it explains. The evocation of these ideas is even more valuable than the logical equivalence to the laws. Theories are often accepted and valued greatly, by part of the scientific world at least, even if it is known that they are not quite true and are not strictly equivalent to any experimental laws, simply because the ideas which they bring to mind are intrinsically valuable.…

Norman Robert Campbell, Foundations of Science (New York: Dover Publications, 1957), pp. 130-132. (This work was originally published in 1920 as Physics: The Elements.)