Introduction
The essays in this volume represent three major areas of philosophy of science that I have explored over a period of more than forty years: evidence, explanation, and realism. They have appeared in journals or books; some have been reprinted in collections; three were selected for inclusion in The Philosopher's Annual, the editors' choice of the best ten essays in philosophy that year.
Philosophy of science, like many other disciplines, has become a field with several fairly narrow subspecialties, where questions are raised about how to understand the concepts and foundations of some particular scientific theory and where practitioners in one area often do not engage those in another. This is in stark contrast with what I take to be a glorious past, when Isaac Newton could formulate a very general scientific methodology under the title “Rules for the Study of Natural Philosophy”; William Whewell, who explicitly rejected Newtonian methodology, could formulate his own methodology in a chapter called “Of Certain Characteristics of Scientific Induction”; John Stuart Mill, who in turn rejected Whewell's ideas, could champion a very general methodology in chapters such as “Preliminary Observations on Induction in General” and “Of the Four Methods of Experimental Inquiry”; Pierre Duhem could title one of the sections of a classic work “What Is the True Nature of a Physical Theory and the Operations Constituting It?”; and in the twentieth century, Carl G. Hempel could write influential articles such as “Studies in the Logic of Explanation” and “Studies in the Logic of Confirmation.”
Broadly speaking, my essays are in this tradition. They are based on the idea that one can say something very general, informative, and even true about the topics selected. Admittedly, in pursuing generality, there is always a danger of the sort noted by Niels Bohr when he was asked to say what the difference is between a physicist and a philosopher.
“The physicist,” he said, “learns more and more about less and less until he knows everything about nothing. The philosopher learns less and less about more and more until he knows nothing about everything.” Whether the essays in this volume succeed in avoiding this fate, the reader must decide.In addition to the generality of its approach, the tradition has another feature of importance. It is based on the idea that although particular causal factors, including historical, psychological, and sociological ones, can be invoked in understanding both how scientists came to develop and defend their ideas in the way they did and how these ideas influenced other scientists and traditions, these factors can be separated from the “logic” of the arguments they used and the methodologies they championed. One can examine and criticize Newton's four rules of reasoning, as Whewell did, from a general philosophical or logical viewpoint, while recognizing the historical and perhaps psychological fact that Newton promulgated these rules, among other reasons, because he sought to convince Cartesian physicists of the truth of his law of universal gravitation, and while recognizing that these rules had a profound effect on how later scientists argued for their own scientific theories.
By contrast with some members of this tradition, the approach I take is historical in this sense. The best way to proceed to understand a concept such as evidence, explanation, or realism in the sciences involves not only producing an “abstract” definition but also, very importantly, showing how that definition applies to actual scientific cases. The essays I have selected include ones that are historically oriented in this sense. For example, in exploring the meaning and truth of scientific realism, I proceed by asking whether Jean Perrin's argument in 1908 for the existence of molecules establishes the truth of scientific realism, as he himself thought.
Although the essays focus on three different topics, they form a natural set because, on my view, the topics are importantly related. The definition I offer for evidence invokes a concept of explanation, and that for scientific realism relies on a suitable concept of evidence.
Briefly, to take the third concept first, scientific realism is a doctrine committed to the idea that “unobservables” (e.g., molecules, electrons) exist and can be known to exist and that it is one of the aims of science to provide true theories about them, as well as about the observable parts of the world. I argue that scientific antirealists, such as nineteenth-century positivists and some twentieth-century writers, rejected these claims and that, contrary to their views, scientific realism can be established by invoking suitable empirical evidence. So an important question arises: what is evidence? On my account, evidence for a hypothesis or theory is supposed to provide a good reason for believing that the hypothesis or theory is true. In accordance with the definition I arrive at, which reflects this idea, some fact is evidence that a hypothesis is true only if, given that fact, the probability is high that there is an explanatory connection between the hypothesis and the fact. I explain this definition and show its advantages over alternative accounts, which, unlike the present account, are subject to various types of counterexamples.This will prompt the question of how to understand “explanatory connection” and, more generally, the idea of explanation. My general account of explanation begins by first considering acts of explaining—what it means for someone to explain something to someone. An explanation, in the sense of the product of such an act, is then defined as an ordered pair, one of whose members is an act-type “explaining q,” where q is a certain type of indirect question, and whose other member is a proposition that provides an answer to the question q. A correct explanation is one in which the propositional member of the ordered pair is true. And there is an “explanatory connection” between propositions A and B if and only if either the fact that A is true correctly explains why B is true, or the reverse is the case, or the fact that some third proposition C is true correctly explains why both A and B are true.
Although the concept of explanation is defined by reference to a type of explaining act, the resulting concepts, as well as that of a correct explanation, are, I argue, objective concepts—in the sense that whether something is an explanation, and whether it is a correct explanation, does not depend on, or vary with, anyone's beliefs or knowledge or on whether anyone is explaining anything to anyone. This objective concept of explanation will yield objective concepts of evidence and scientific realism. I also argue that there is another important way of evaluating an explanation—resulting in what I call a “good explanation”—which is not objective in this way. Whether an explanation is a good one does depend on particular epistemic situations, which can vary from one context to another. The “objective” and “contextual” ways of evaluating explanations are different, and both are used to generate important types of evaluations, but the most basic concept of evidence I introduce, as well as the ideas underlying scientific realism, require a noncontextual concept of “correct explanation.”Of the fifteen essays that follow, one is from 1965, and the rest are from the late 1970s to the present. (Excluded are essays written mostly in the 1980s and published in my Particles and Waves [1991] that deal with wave-particle debates in physics in the nineteenth century.) I have made a few changes in some of the essays included in the present book to conform to later formulations of the views expressed. In some cases, I have added an addendum for this purpose.