Consider Larry Laudan's (1984) recent statement of the role that consensus plays in scientific validation:

Cosmologists often claim that our universe is “fine tuned” for life. A change by 1% in the strong nuclear force would have meant little carbon would exist, and carbon can seem biologically essential. Again, the riches of chemistry and biochemistry depend on the neutron’s being heavier than the proton by no more than 0.1%. The early cosmic expansion rate may have needed fine tuning to one part in l055 to prevent speedy recollapse and speedy disintegration. To prevent excess turbulence the “smoothness” perhaps needed fine tuning to one part in 10 raised to the power of 10123 (a number far greater than 101230). Etcetera (Leslie 1982, 1983a, 1985, 1986).

The list of such claims is very long. No doubt some of its items are mistakes. Others may be dictated by basic physics so that they are not fine-tunable.

The so-called “orthodox” interpretation of quantum physics attributed to Niels Bohr is commonly regarded as abandoning realism. I have already opposed this view elsewhere (Folse 1985) but partially in response to criticism of my position (Shimony 1985), here I propose to relate Bohr’s realism to recent contributions to the realism debate given by Hacking (1983), Cartwright (1983), and Ellis (1985). Specifically, I argue that Bohr’s complementarity viewpoint requires a causal entity realism. Furthermore, labeling Bohr an anti-realist with respect to theories is misleading because it assumes the correspondence theory of truth that Bohr holds quantum theory forces us to reject.

Although there are many realist views, I take the root of realism to be the quest for knowledge about the reality producing the phenomena we experience. It is realism in this sense that Bohr defended and that makes science significant to metaphysics.

The bulk of this paper will be divided into three sections. I will first sketch what I take to be the main problem in the epistemology of science and explain how the issue of consensus is related to this problem. I will then try to solve the problem by using a social conception of scientific rationality based on evolutionary ideas. And I will finally elaborate that social conception by responding to several objections.

For over two decades now, many philosophers have talked about the existence of a crisis in the epistemology of science. This crisis presumably came about as the result of the dispute between two schools of thought: the logical and the socio-historical. The logical school is associated with the analytic and empiricist tradition in the field and is supposed to defend the rationality of science. The socio-historical, whose most notorious exponents are Kuhn (1970) and Feyerabend (1975), has been accused of irrationalism.

It is certainly not excessively original to maintain that an empirically oriented theory on the structure and on the history of Physics—that is a theory that can and should be put to the test of what physicists, qua physicists, actually do and have been doing and of what they actually come up with—is still badly needed. And it is equally unoriginal to hold that the implementation of such a project cannot but somehow relate to The Structure of Scientific Revolutions (Kuhn, 1962), a work which, to its author's mind at least, was meant to provide some kind of a starting point for such a project. But the fact is that The Structure (as well as all the other work of Kuhn of the same inspiration) has not, in the main, been read that way.

Leibniz never tired of stressing the fundamental importance of the concept of continuity for philosophy, nor was he shy of attributing major importance to his own struggle through “the labyrinth of the continuum” for the subsequent development of his whole system of thought. Unfortunately, however, his own thought on the subject is something of a labyrinth itself, and from a modern point of view many of his pronouncements are apt to seem blatantly contradictory.

Certain quotations seem to commit him unambiguously to atomism. Thus to de Voider he writes: “Matter is not continuous, but discrete…. The same holds for changes, which are not truly continuous.” (To de Voider, 11th October 1705: G.II.279).2

The question of how and why the application of mathematics to physical reality is possible has occupied philosophers for many centuries. In contemporary discussions, Philip Kitcher’s attack on a priorist approaches to the question is particularly interesting, for it suggests that there is no global answer (Kitcher 1983, Chapters 1-4). In this essay, I would like to develop his insight by arguing, first, that the problem of how mathematics relates to physical reality should be addressed by an appeal to the history of mathematics and the sciences in the form of case studies which analyze the peculiar and changing ways in which these domains interact. Generalizations about such interaction should be made carefully, in a kind of local and piecemeal retrospective; I believe that the historical record will show that the grounds of justification and the form of applied mathematics changes dramatically from one historical era to the next.

Baltas calls for an empirical theory of the structure and development of Physics—one which gives a unified ac count of both the internal and external aspects of the science. In this paper he proposes a tripartite structure for physical theory and describes the role of socially determined ideological “assumptions” upon that structure.

Since there is very little social consensus amongst philosophers today on many elements of his project, before discussing his proposal in detail it might be helpful to sort out some of the widely disparate enterprises which are associated with terms such as external or ideological when applied to the study of science. Baltas does not always use these terms in quite the ways I would expect.

One of the enduring problems in history and philosophy of science is the interaction between theory and experiment. Hacking (1983) has argued persuasively that experiment often has a life of its own, independent of theory. Lakatos (1970), on the other hand, has argued that theory often proceeds independently of experimental results. More typically, the interaction seems to be rather complex. Experiment may confirm or refute a theory or call for a new theory. Theory may suggest new experiments, offer a new interpretation of known results, or help to validate an experimental result. These are, of course, not the only possibilities (Hacking 1983 and Franklin 1986, Chapter 4).

In dealing with the complex interaction between theory and experiment the old cliche, “Man proposes, Nature disposes,” has been shown to be far from adequate. Pickering's study of quark models and theory (1984a) has shown that it is not always clear at the time what is being proposed.

Concerning evidence there are two important questions: (1) what is going to count as evidence? and (2) what are the appropriate means for employing evidence? These two problems pervade the analysis of the scientific process. They are with us as much today as in Galileo’s time. For example, with respect to contemporary arguments between Evolutionists and Scientific Creationists, if the issue is taken in its cognitive rather than its political dimension, the entire discussion turns on the criteria for acceptable, evidence, i.e, the heart of the problem is disagreement over the status of the data one wants to invoke as evidence.

Many of the disputes between Galileo and his philosophical, interlocutors in his Dialogue on the Two Chief World Systems turn on some of the same issues that bedevil the debate between evolutionists and creationists, i.e., what is going to count as evidence?

“An experiment,” the Oxford Dictionary of the History of Science records, “unlike an experience, is a designed practical intervention in Nature; its upshot is a socially contrived set of observations, carried out under artificially produced and deliberately controlled, reproducible conditions. At the experiment's core is the notion that the conditions for producing a given effect can be separated into independently variable factors, in such a way as to demonstrate how the factors behave in their natural (i.e. the non-experimental) state.” (Dictionary 1981, p. 136). Around 1900, some biologists would have acceded to such a definition, yet many would have offered alternatives. Some would have denied that the conditions in question must be “designed” or artificially produced; nature may also provide experiments. Others would have emphasized the use of experimentally-derived observations for hypothesis testing; experimentation goes beyond the mere experiment itself, that is. A variety of additional definitions would have been advanced as well, illustrating the lack of one such orthodox interpretation as the Dictionary now offers.

Kant’s philosophy of mathematics presents two fundamental problems of interpretation: (1) Kant claims that mathematical truths or “judgments” are synthetic a priori; and (2) Kant maintains that intuition is required for generating and/or understanding mathematical statements. Both of these problems arise for us because of developments in mathematics since Kant. In particular, the axiomatization of geometry--Kant’s paradigm of mathematical thinking--has made it seem to some commentators as, for example, Russell, that both (1) and (2) are false (Russell 1919, p. 145).2 If virtually all of mathematics, including geometry, is axiomatizable, it would seem that mathematics results in analytic judgments that are totally independent of sensibility, the source, according to Kant, of intuition. In this paper I will address both of these difficulties. I shall argue that Kant’s understanding of both “synthetic” and “intuition” make his position immune to these criticisms.

Since the revival of historicist philosophy of science in the 1960s many philosophers have acknowledged a debt to Duhem. But Duhem’s opinions are imperfectly understood and, as McMullin has shown in his (1970) and (1979), there are many strands in the current revival of historicism. We consider here Duhem’s views on the role of history in the appraisal of scientific theories. However, there is no single text offering Duhem’s views on the subject; rather, they are revealed during their application to various historical and contemporary cases. Duhem’s most sustained examination of a contemporary case is his critique of Maxwell’s science and scientific methodology.

Duhem’s critique of Maxwell is not a set of isolated or incidental dicta, but the expression of Duhem’s mature thought over the last quarter-century of his life, ranging from (at least) 1893 to his death in 1916.

This paper is about realism. The ideas presented were developed as part of a response to a persistent criticism levelled against the relativist/constructivist programme in the sociology of scientific knowledge (SSK), namely that this programme fails to allow the real any role in the articulation of scientific thought. This criticism I believe to be mistaken, and justified, if at all, only by the relative silence of the SSK tradition on the role of the real. To indicate that this silence is a contingent, rather than a necessary attribute of the tradition, my aim here is to explore what can be said about the relation between the real and the articulated on the basis of empirical studies of scientific practice.

To situate what follows, let me note that the standard philosophical debate on realism concerns the reality, or lack of it, of unobservable entities appearing in well confirmed theories (Leplin 1984).

While some philosophers have assumed that there are only two options for characterizing the ontological status of mental models in cognitive information processing psychology--treating them as nearly autonomous from theories of brain activity (Putnam 1975 and Fodor 1974) or eliminating them in favor of neuroscience accounts (Churchland 1979)-- cognitive scientists have often tacitly assumed a third option. This involves treating the mental models as systems of rules and representations that are instantiated in the nervous system much in the way computer programs are instantiated in computers. While this seems to be the position of those endorsing the autonomy of psychology, it, in fact is consistent with a much weaker interpretation if one recognizes that the vehicle in which mental processes or computer programs are instantiated may limit and constrain what kinds of mental processes or computer programs can be instantiated.

It is impossible to overrate the important contribution which statistical mechanics makes to our contemporary understanding of the physical world. Cutting across the hierarchy of theories which describe the constitution of things, and related in subtle and not yet fully understood ways to the fundamental dynamical theories, it provides the essential framework for describing the dynamical evolution of systems where large domains of initial conditions lead to a wide variety of possible outcomes distributed in a regular and predictable way. For the special case of the description of systems in equilibrium, the theory provides a systematic formalism which can be applied in any appropriate situation to derive the macroscopic equation of state. Here the usual Gibbsian ensembles, especially microcanonical and canonical, function as a general schematism into which each particular case can be fit. In the more general case of nonequilibrium the situation is less clear.

In 1941 Konrad Lorenz published a paper with the title “Kant’s Doctrine of the A Priori in the light of Contemporary Biology”. This essay stands as the foundation of the Austro-German School of Evolutionary Epistemology. As indicated by the title of the paper, the Lorenzians attempt to interpret Kantian transcendentalism along biological lines.

Lorenz was, however, by no means the first who attempted to biologise Kant. Philosophers and scientists such as Ernst Mach, Henri Poincare, Ludwig Boltzmann and Jean Piaget had made similar points previously. Nevertheless, it remains Lorenz’s achievement to have formulated his thesis explicitly and provided ample empirical evidence in favour of it.

Lorenz’s seminal paper remained largely unknown, and even his book on evolutionary epistemology published in 1973 raised little interest among philosophers, particularly in the English-speaking world. Meanwhile several other biologists and philosophers have joined Lorenz in his endeavour to elucidate epistemological issues through biological research.

The nature and basis of irreversibility has been intensively investigated over the last 100 years by physicists, mathematicians, and philosophers. But if progress is measured by convergence of opinion among the “experts” then it is hard to detect; in fact the field is a virtual Tower of Babel. J.L. Lebowitz, one of the leading mathematical physicists in the field put it this way:

With the appearance of J.H. Conway's On Numbers and Games (1976), the mathematical and philosophical communities have much to celebrate. It is Conway's important discovery that the familiar Dedekind cut and von Neumann ordinal constructions are part of a more general construction which leads to a proper class of numbers embracing the reals and the ordinals as well as many less familiar numbers including -ω, ω/2, l/ω, √ω and ω-π, where ω is the least infinite ordinal. Conway further shows that the arithmetic of the reals may be extended to the entire class yielding a real-closed ordered field (1976, pp. 40-42); that is, an ordered field where every positive element is a square, and every polynomial of odd degree with coefficients in the field has a solution in the field.

Could anything provide a philosophically convincing mark of the mental in simple organisms (Lloyd 1984)? Individual organisms’ capacities to modify behavior adaptively as a result of past encounters with the environment might mark the first step in the phylogeny of minds. The simplest examples of mental representation are likely to be found in the simplest forms of animal learning.

The most scientifically rigorous test case of “bottom- up” strategies in cognitive neuroscience is provided by current studies of the cellular and molecular biology of associative learning in higher invertebrates, but particularly gastropod molluscs, snails or slugs (Quinn 1984). For the relevant neurobiological research community (Farley and Alkon 1985; Hawkins, Clark, and Kandel 1986), “Bottom-up” does not just mean letting one’s study of the properties of the nervous system control one’s analyses of animal behavior or cognition.