The theory of explanatory unification was first proposed by Michael Friedman (1974) and developed in more detail by Philip Kitcher (1981,1989). The primary motivation for this theory is the case that the theory adequately accounts for the genesis of scientific understanding. Standard models of explanation are, moreover, allegedly constructed on the basis of a misconception of the nature of understanding. My thesis is that the unificationist account of understanding proposed by Friedman and Kitcher is false, and hence, that the theory of explanatory unification is fundamentally misconceived.

How does an explanation of some explanandum render the explanandum as ‘understood’? Carl Hempel wrote of the D-N covering law model that it shows that “given the particular circumstances and the laws in question, the occurrence of the phenomenon was to be expected; and it is in this sense that the explanation allows us to understand why the phenomenon occurred.”

The debate about values in science in the last decade or so reflects an important shift in what many acknowledge as the “post-positivist” era in philosophy of science. It reflects the erosion of the fact-value distinction in at least one of its more simplistic forms (facts belong in science, values outside), and it marks the path to a more enhanced understanding of the roles of both facts and values in scientific inquiry. This discussion has, above all, contributed to what we might call a revaluation of value: from the point of view of the epistemologist or philosopher of science values are neither uniform nor uniformly “bad”.

According to a recent argument, models play two essential roles in the argumentative structure of solid state physics and chemistry (Hofmann 1990). On the one hand, models are the culmination of phenomenological description. That is, models are idealized representations of the molecular structures thought to be causally responsible for the processes experimentally monitored and measured. Secondly, dieoretical physicists and chemists require that models ultimately be cast in a mathematical form appropriate for the application of the Schroedinger equation. In this respect models become the means through which the Schroedinger equation gives a theoretical unity to what would otherwise be a disparate set of empirical phenomenological laws and descriptions with limited scope. That is, it is an important theoretical goal to show that experimentally generated phenomenological laws can be approximately derived through an application of the Schroedinger equation to a necessarily idealized and simplified mathematical description of the relevant system. The two functions of models are not incompatible, but they do reflect two distinct theoretical orientations toward the interpretation of data.

For a number of years now, we, as philosophers of science, have been enjoined by more and more of our colleagues to understand the task of developing a philosophy of science to be itself a scientific task. We are told that if we want to understand science we have no better (and perhaps indeed no other) path to such an understanding than the path of science itself. We should view ourselves as ultimately attempting to arrive at a relatively complete theoretical understanding of how science proceeds. This is a call to naturalize philosophy of science.

Somewhat more recently some philosophers of science have become impatient even with those who have taken up the naturalist banner. It isn’t enough, they tell us, to arrive at an understanding of the nature of philosophy of science that proclaims it to be one science among many.

Traditional accounts of the relationship between experimental evidence and hypotheses have been challenged in the recent work of Hacking (1983), Galison (1987), and Franklin (1986,1990). These authors have pointed to the independent aspects of the experimental tradition and forced us to rethink the view that experiment is simply the handmaid of theory. As a result we have come to recognize the tremendous intricacies involved in theory testing and experimentation and have come to appreciate the various ways experiment shapes scientific practice.

In what follows I want to extend this discussion by taking a closer look at the way in which theories are supported by empirical evidence. In addition to the traditional testing model the method of deduction from phenomena, made famous by Newton, is also thought to furnish experimental evidence insofar as the theory itself is supposedly deduced from empirical data.

Many philosophers of science tend to take for granted the proposition that, as a rule, scientific experiments are reproducible (or, in other words, repeatable, replicable). Very few of them, however, have examined the issue in any detail (a recent exception is Hones 1990). Yet, by now an important body of literature exists, mainly written by sociologists and historians of science, in which the above proposition is discussed and analyzed on the basis of elaborate studies of experimental practice. These studies claim to offer not only a descriptively adequate account of experimentation, but also a number of fundamental philosophical conclusions concerning (experimental) science as a whole.

In the present paper I analyze and evaluate the views of H.M. Collins, an early and influential thinker concerning the issue in question. In particular I discuss what he has called the ‘experimenters’ regress’.

How is the deep disagreement between what Kuhn characterized as paradigms ultimately resolved and how do we interpret such debates epistemically? A close analysis of the Ox-Phos Controversy in bioenergetics from the 1960s and 70s (§§2-3 below) suggests that one justifies a set of questions through an ensemble of empirical demonstrations. This contrasts to decisions between theoretical alternatives through ‘crucial experiments'. When viewed along with other historical episodes, this case suggests a philosophical category of ‘demonstrations', distinguished from crucial tests and complementary in justificatory status to falsifying instances. The distinction also suggests specific strategies for scientists (§4).

The Ox-Phos Controversy is an especially valuable case for studying theory development and scientific change, and for investigating the problems of disagreement, originally highlighted by Kuhn, where two incompatible conceptual or experimental gestalts converge on the same empirical domain (see also Gilbert and Mulkay 1984a; 1984b; Robinson 1984; 1986; Rowen 1986; Weber 1986; 1991; Allchin 1990; 1991).

My topics are two theses about quantum mechanics that are widely doubted but curiously tempting: value-definiteness and contextualism. Value-definiteness holds that every observable has a definite value; contextualism follows from value-definiteness, and claims that one Hermitian operator can represent many observables. In fact, contextualism is my deeper interest Although it is not a view with an army of defenders, it has a long history of discussion in the literature. (See, for example, Belinfante 1973, van Fraassen 1973, Shimony 1984, Redhead 1987). The bulk of this paper will be concerned with a puzzle about degenerate operators that arises out of the standard accounts of contextualism. But before we get to the details, let us begin with a lurid example.

In a recent article Itamar Pitowsky (1991) has analyzed some measurement possibilities presented by an entangled state of three spin ½ particles. From that analysis he has concluded that the valid predictions of results of possible future measurements, made by observers who actually carry out mutually space-like separated measurements, can be incompatible with one another in the sense that their conjunction contradicts the quantum theory. From this result Pitowsky infers the inherently observer specific, or relative, character of quantum predictions as opposed to the results of actual observations or measurements, which, he claims, are covariant; i.e. (I take it) compatibly translatable from one observer to another.

The Einstein-Podolsky-Rosen argument for the incompleteness of quantum mechanics involves two assumptions: one about locality and the other about when it is legitimate to infer the existence of an element-of-reality. Using one simple thought experiment, we argue that quantum predictions and the relativity of simultaneity require that both these assumptions fail, whether or not quantum mechanics is complete.

EPR’s (1935) argument for the incompleteness of QM turns upon the following sufficient condition:

“If, without in any way disturbing a system, we can predict with certainty (i.e., with probability equal to unity) the value of a physical quantity, then there exists an element of physical reality corresponding to this physical quantity.“

EPR apply this condition in a case where knowledge gained from measuring one quantum system allows one to predict with certainty a value for a physical quantity on another.

In Theory and Evidence, Clark Glymour asserts that many philosophers are convinced that “evidence can only bear on the entire body of our beliefs, and cannot be parceled out here and there“(1980,p.5). Attributing this view to Quine, Glymour says:

Some philosophers might counter that Glymour’s criticisms apply properly to a view Quine had significantly modified two decades before the publication of Theory and Evidence.

It is commonly acknowledged in science that model construction is one of the most important components of theorizing. Philosophers of science are gradually coming to acknowledge this situation, spurred on by holders of the semantic view of theories. In this paper I wish to defend a very deflationary version of the semantic view of theories, which is more or less a re-statement of the above commonplace. I reject the view encapsulated in the identity statement “scientific theories are families of models,” although acknowledging the useful insights into science that holders of this strong position have given us. My position derives from a critique of various of the semantic views of theories, and further from a guiding presupposition that rather than providing necessary and sufficient conditions for what a theory is, philosophers should focus on the nature of scientific theorizing. Theorizing is carried out by practicing scientists, and we cannot say what scientific theories are unless we appreciate the myriad ways they are used and developed in all of the sciences.

Since “one can seldom directly deduce from a speculation consequences that are even in principle testable,” a scientist must “articulate” a theory, mathematically altering a given speculation to bring it “into greater resonance with the world” (Hacking 1983, p. 214). When confronted with computational difficulties caused by analytically intractable equations, imprecise specifications of initial conditions, or the absence of required auxiliary theories, scientists commonly articulate a theory by applying approximations and idealizations to the theoretical equations they have at hand. Recognizing this, the following question immediately arises for both the scientist and the philosopher of science, “How does one judge the quality of a particular approximation? I.e., when is a given approximation valid?“

For Hacking, the word “real” is one of J.L. Austin's trouser-words, taking its meaning from its negative uses in much the same way as the admittedly sexist expression “wear the trousers”. In Representing and Intervening, Hacking proposes that at least some scientific realists can be interpreted as using the word “real” in this way. The word “real” is also substantive-hungry, he adds. Thus when a philosopher states that a type of entity is not real we need to know just what is being denied. The causal entity realist, for example, holds that the entities scientists postulate in their theories can be regarded as real if they have causal powers that can be manipulated to create observable and repeatable effects. Thus the causalist offers sociology a possible way to command respect: if social science entities could be manipulated, the social sciences would then be on a par with physics.

On the surface and to a first approximation, economic theory and evolutionary theory share salient features: a commitment to optimality and maximizing, a consequent similarity in mathematical formalism, an appeal to equilibrium as the preferred explanatory strategy. Moreover, both have at various times been stigmatized as empirically empty. However, the theory of natural selection has been widely accepted as the scientific cornerstone of modern biology, while microeconomics has remained open to repeated challenges as an explanation of human action and its aggregate consequences.

Because of their similarity, evolutionary theory looks like part of a powerful defense of the status quo in economic theory. I think that Darwinian theory is a remarkably inappropriate model, metaphor, inspiration, or theoretical framework for economic theory.

In the Preface to his recent philosophy of biology (1988) and in his earlier Growth of Biological Thought (1982), Ernst Mayr emphasized that recent progress in evolutionary biology is a result of conceptual clarification, not a consequence of improved measurements or better scientific laws. Although complete agreement on the meaning of key concepts is not essential for all communication in science (see Hull 1988, pp. 6-7,513), we likewise believe that conceptual clarification is perhaps the most important key to progress in community ecology. In order to investigate some of the reasons that might explain why community ecology has been unable to arrive at a widely accepted general theory, in this essay we analyze “stability,” one of its most important foundational concepts. After reviewing the stability concept and sketching its associated problems, we assess the epistemological status of four difficulties with the concept.

I relate this homily to several puzzles. In one, three cards, of which I know one is a king, lie face-down. After I select, without inspecting, a card and bet that it is a king, you reveal that a certain other card is not a king. I wonder what this does to my chances on that bet. In another puzzle I am one of three prisoners and learn that one of us will be released. Then I learn that a certain other prisoner will not be released. Again I wonder what this does to my chances.