In this paper I attempt to tie together a longstanding dispute between Henry Kyburg and Isaac Levi concerning statistical inferences. The debate, which centers around the example of Petersen the Swede, concerns Kyburg's and Levi's accounts of randomness and choosing reference classes. I argue that both Kyburg and Levi have missed the real significance of their dispute, that Levi's claim that Kyburg violates Confirmational Conditionalization is insufficient, and that Kyburg has failed to show that Levi's criteria for choosing reference class are problematic. Rather, the significance of the Petersen case is to show that other aspects of their respective systems are defective: for Levi his account of credal judgments other than direct inference, and for Kyburg his explanation of how indexes are associated with a body of knowledge.

No one has a well developed solution to Duhem's problem, the problem of how experimental evidence warrants revision of our theories. Deborah Mayo proposes a solution to Duhem's problem in route to her more ambitious program of providing a philosophical account of inductive inference and experimental knowledge. This paper is a response to Mayo's Error Statistics (ES) program, paying particular attention to her response to Duhem's problem. It turns out that Mayo's purported solution to Duhem's problem is very significant to her project, for the epistemic license claimed by ES and the philosophical underpinnings to her account of experimental knowledge depend on this solution. By introducing the partition problem, I argue that ES fails to solve Duhem's problem and therefore fails to provide an adequate account of experimental knowledge.

Anyone, today, with even a slight interest in the methodology of science will be aware of the heated debate which has raged in regard to the thesis of the logical symmetry between explanation and prediction, which is entailed by the hypotheticodeductive account of scientific theory. The symmetry thesis, which received its classical exposition in a well-known article by Hempel and Oppenheim [2], has been subject to a steadily growing criticism by several eminent thinkers. My intention, in this article, is to argue that the current reaction to the Hempel-Oppenheim position, as exemplified by Professors Hanson, Scriven, et al., represents a retrograde movement in the philosophy of science, which, as the number of its converts grows, could undermine the progress, which has been accomplished in the broad tradition that extends from Mach and Poincaré through the Vienna Circle to the present day. Specifically, it is my contention that the aforementioned reaction is symptomatic of a contemporary trend in the direction of irrationalism which, of all branches of philosophy, the philosophy of science should seek to dispel rather than to foster. For the sake of brevity, I propose to focus my remarks wholly on the original widely read article of Hanson on this subject [1]. That I single out this particular source should not be taken to belie the admiration I hold for the masterful scholarship of its author.

Certain suggestions recently made by Brunswik (2) concerning the design of experiments in psychology seem to have far reaching implications. Indeed, Brunswik's suggestions appear to the writer to be congruent with Einstein's General Theory of Relativity. Congruences between such diverse disciplines as psychology and physics bear watching if for no other reason than the fact that psychologists frequently point to the physicist as the ideal scientist. Unfortunately, in the writer's opinion, the ideal which the psychologist still admires is the classical, or Newtonian, physicist rather than the modern, or Einsteinian, physicist—the psychologist's outlook being concomitantly distorted. Brunswik's theories of experimental design in psychology, however, are appealing for the very reason that they do not fit the mold of the classical physical experiment, but do seem to be congruent with Einstein's physical theories—with concomitant expansion of scope.

The term coherence is used in many connections. We often speak of coherent reasoning, coherent activities, coherent bodies, and coherent phenomena. In science this term is often used to indicate continuity of a structure in space and time. If many individuals can simultaneously observe similar sets of coherent phenomena, we infer with some justification, first, that the phenomena observed have some kind of correspondence with activities in an external world, that is, a world independent of the existence of the individual observers, and second, that the coherence itself is an attribute of the external world. In a recently published article, to which I shall make several references in this paper, I have called attention to the fact that the science of physics is based on the form and the changes in form of an immediately observed phenomenological world, called the physical world, which can be completely characterized as a structure in space and time. In his analysis of the foundations of the science of physics Eddington has clearly shown that theoretical physics deals exclusively with structures of the type described in the mathematical theory of groups, and that our knowledge of the physical world could have been reached by our sense of vision alone. Coherence in the physical world is that characteristic of the structure which describes the more or less permanent relationship between phenomena in different parts of space at approximately the same time, or a regular sequence in the changing phenomena at approximately the same region of space.

The old quarrel between the logical empiricists or pragmatists and the defenders of metaphysics is still far from being settled. This situation is at least partly due to the fact that neither of the two antagonists has a clear insight in the historical background, the systematical validity, and the practical purposes of the concepts and methods used in traditional philosophy. Therefore further clarification is needed.

The following paper does not claim to give final definitions and decisions. It will rather try to draw the attention to certain ways of reasoning still very widespread in metaphysics and moral theory. Their careful analysis may throw light on some of the so-called eternal philosophical problems.

One of the obstacles facing massive modularity is how a pervasively modular mind might generate non-domain-specific thoughts by integrating the content produced by various domain-specific modules. Peter Carruthers has recently argued that the operations of the language faculty are constitutive of the process by which the human mind is able to integrate content from heterogeneous conceptual domains. In this article, I first argue that Carruthers's data do not provide support for either of two possible interpretations of his thesis. In addition, I provide empirical and theoretical reasons for thinking content integration is performed external to the language faculty.

Our cognitive capabilities force us into a description of the world by levels. But theories on different levels result in descriptions that differ qualitatively. Therefore, the resulting incommensurability requires ontological bridges between such theories. These are obtained uniquely when the equations of the reduced theory are compared with a suitable limit of the equations of the reducing theory. Four case studies from theoretical physics and astronomy support this claim, two for theories of composites and two for non-composites (field theories). There results a coherent view of a single real world despite its ontological pluralism. The cumulativity of scientific knowledge is thus ensured and realism is supported.

It is generally not recognized that Whewell's conception of necessary truth evolved only gradually; his early statements are misleading. For this reason, and because of certain peculiarities in his expository style over his publishing history, he is commonly thought to have used the term “necessary” in the sense of “absolutely necessary”. I argue that, on the contrary, the term is essentially relational in his mature view. This conclusion leads, in turn, to a re-interpretation of his doctrine of “fundamental ideas”. Here I argue that the crux of his position is that cognitive principles which are initially regulative become, through invention, repeated employment, and stipulative definition, constitutive ones.

Heisenberg's (3) principle of indeterminacy or uncertainty has led most theoretical physicists and philosophers to two important steps: 1) the denunciation of the law of physical causality; 2) the decision of biological and psychological problems in favor of indeterminism.

The most recent resurgence of philosophical attention to the so-called ‘functional talk’ in the sciences can be summarized in terms of the following questions: (Q1) What kind of restrictions, and in particular, what kind of evolutionary restrictions as well as to what extent, is involved in functional ascriptions? (Q2) How can we account for the explanatory import of function-ascribing statements? This paper addresses these questions through a modified version of Cummins’ functional analysis. The modification in question is concerned with phylogenetical restrictions on causal role functions, and it stems from an analysis of some primary areas in molecular biology. I examine how evolutionary consideration affects the so-called ‘function-analytical explanatory strategy’ (Cummins [1975] 1998, 2002). Finally, I argue that the neo-functional analysis here proposed accounts for a certain convergence between the main rival theories of biological function.

Many examples of calibration in climate science raise no alarms regarding model reliability. We examine one example and show that, in employing classical hypothesis testing, it involves calibrating a base model against data that are also used to confirm the model. This is counter to the ‘intuitive position’ (in favor of use novelty and against double counting). We argue, however, that aspects of the intuitive position are upheld by some methods, in particular, the general cross-validation method. How cross-validation relates to other prominent classical methods such as the Akaike information criterion and Bayesian information criterion is also discussed.

In his article, “Is Essentialism Unscientific?” (1988), Jarrett Leplin claims that I do not have sufficient grounds for rejecting the customary “philosophical method of discovery” that allows for the direct transfer of theories developed in the philosophy of language to science. While admitting that all attempts at transfer thus far have failed, he still maintains that method is sound. I argue that the wholesale failure of these attempts is reason enough to suspect the method and to try to devise one more suitable to fathoming how “meaning”, “reference”, and “meaning change” are to be understood for scientific theories. The method I have proposed in Nersessian (1984b), and subsequent work, demands that we learn how to incorporate the actual practices of meaning construction in science into our analyses. Leplin distorts my analysis and, thus, fails to understand the insights that study provides.

Logicism Lite counts number-theoretical laws as logical for the same sort of reason for which physical laws are counted as as empirical: because of the character of the data they are responsible to. In the case of number theory these are the data verifying or falsifying the simplest equations, which Logicism Lite counts as true or false depending on the logical validity or invalidity of first-order argument forms in which no numbertheoretical notation appears.

An option available to an agent is stable if it maximizes expected utility on the hypothetical assumption that the agent is going to choose it. As is well known, some decision problems lack a stable solution.

Paul Weirich (1986 and 1988) has recently proposed a decision principle which prescribes that the option chosen should be at least weakly stable—or “weakly ratifiable”, to use his terminology. According to him, full stability is an excessively strong demand. I shall argue that Weirich's proposal conflicts with the familiar condition of dominance. But I shall also prove that this difficulty can be avoided if we replace weak stability by “moderate” stability—where the latter property is somewhat stronger than the former.

It will be seen, however, that this modification does not help against other ailments connected with stability. In particular, to heed the demand of stability (of any kind) is to engage in a form of “wishful acting”. Also, the different stability demands all conflict with a close relative of the dominance condition: the condition of “indifference”. According to this condition, two actions are equally choiceworthy if they would always lead to the same outcomes—whatever state the world is in. On the other hand, the conditions of dominance and indifference would both be satisfied if we replaced a demand for stability (of some kind) by a related but distinct principle of “retrievability”. Retrievability and (full) stability are mutually independent properties of options, but each of them entails moderate stability. The paper ends with a discussion of the relevance of retrievability to theories of choiceworthiness and practical reason.

Naturalism implies unity of method—an application of the methods of science to the methodology of science itself and to value theory. Epistemological naturalists have tried to find a privileged discipline to be the methodological model of philosophy of science and epistemology. However, since science itself is not unitary, the use of one science as a model amounts to a reduction and distorts the philosophy of science just as badly as traditional philosophy of science distorted science, despite the fact that the central theme of naturalized philosophy of science is that methodology should be true to science as practiced. I argue that naturalized philosophy of science must apply a plurality of methods to epistemological issues.

I introduce a distinction between general and local model-data symbiosis and offer three examples of local symbiosis in the fields of meteorology and climate science. Local model-data symbiosis refers to a beneficial, two-way interdependence between a particular model and data set or between closely related models and data sets. For each example presented here, I show how the symbiotic relationship works—what the interdependence consists in and how it is supposed to be beneficial—and I consider whether, alongside the benefits, there is some risk of problematic circularity.