INTRODUCTION [56]
which something can be simple—or by contrast, complex— to be usefully characterized by a single definition. There is simplicity with respect to the number of different types of entities postulated by a theory; the complexity of these entities (are they basic or do they have parts?); the number of different types of causes that produce the behavior of these entities; the number of different laws postulated that govern those causes; the complexity of these laws and the calculations required to apply them; and so on.
Perhaps a general definition of simplicity can be found that will cover all these and other types of simplicity or ways in which things can be simple. Following the example set by Newton and Einstein, I will not attempt such a task. I seriously doubt that a useful definition can be given.My question is different: What role or roles is simplicity supposed to play according to scientists who invoke it, and can it do so? In describing such roles, scientists and philosophers make various claims about simplicity, including that nature itself is simple; that simplicity is an epistemic virtue that provides a basis for believing that a theory is true; that it is the aim of science to produce simple theories; that in order to do science one must presuppose that nature is simple; and that scientific theories are always underdetermined by empirical evidence, so that simplicity must be invoked in choosing a scientific theory. In this chapter and the next, I propose to consider these claims and the reasons that scientists and philosophers offer or might offer for them.
I will argue that, despite what scientists and philosophers say, these claims about simplicity are (truth-relevant) speculations. '1 hey are introduced in the course of theorizing and believed to be true by their champions, without knowing that there is evidence for them—at least not evidence that provides a good reason for believing them.
But even considered as speculations, they should not receive high marks. In some cases no reasons, evidential or otherwise, are given for them, while in others the reasons given, both evidential and non-evidential, are very dubious. The speculations I have mentioned have serious problems, and should not be, and need not be, made to do or understand science. The virtue of simplicity, I will maintain, is not that it provides an ontological or epistemic guide to the way the world is or a necessary presupposition for science. It doesn't. Its virtue is mainly pragmatic—one that has to do with practical aims that can vary from one context to another. Moreover, simplicity frequently conflicts with virtues such as truth or empirical adequacy. When it does, whether to choose simplicity on the one hand or truth (or empirical adequacy) on the other is decided on pragmatic grounds. These ideas will be explained, expanded, and defended in the present chapter and the one that follows. It will turn out that simplicity is no simple matter.Two preliminary points should be made. First, scientists and philosophers who invoke simplicity usually apply the concept to abstract objects such as theories, laws, hypotheses, and explanations. But some of them apply it also to the world itself, or parts of it. I will do so as well, when they do. Second, simplicity enthusiasts, especially contemporary ones, invoke simplicity as a criterion of what they call “theorychoice.” But “theory-choice” is a vague expression: choice for what? One might choose a theory for a variety of different purposes including working on it, promoting it, using it to make calculations, and explaining its historical importance.
Whether simplicity is relevant in any of these “choices” depends on the purpose for which it is “chosen” and the context in question.
Defenders of simplicity, I think, have in mind something else, viz. “choosing to believe,” or if one rejects the idea that one can choose to believe, just “believe,” neither of which is to be understood as choosing for some purpose or other. The claim would then be this: the simplicity of a theory is a good reason, or among the good reasons, for believing a theory to be true, or empirically adequate (in one or more of the senses of the latter expression I will note in section 2). This is just one of the claims about simplicity that I examine in what follows.
2.