The old quantum theory of black body radiation was manifestly logically inconsistent. It required the energies of electric resonators to be both quantized and continuous. To show that this manifest inconsistency was inessential to the theory's recovery of the Planck distribution law, I extract a subtheory free of this manifest inconsistency but from which Planck's law still follows.

Causal modeling methods such as path analysis, used in the social and natural sciences, are also highly relevant to philosophical problems of probabilistic causation and statistical explanation. We show how these methods can be effectively used (1) to improve and extend Salmon's S-R basis for statistical explanation, and (2) to repair Cartwright's resolution of Simpson's paradox, clarifying the relationship between statistical and causal claims.

The question of coherence of rules for changing degrees of belief in the light of new evidence is studied, with special attention being given to cases in which evidence is uncertain. Belief change by the rule of conditionalization on an appropriate proposition and belief change by “probability kinematics“ on an appropriate partition are shown to have like status.

There are several interpretations of the argument structure of Darwin's Origin of Species, representing Covering-Law, Inference-to-the-Best-Explanation, and (more recently) Semantic models. I argue that while all three types of interpretation enjoy some textual support, none succeeds in capturing the overall strategy of the Origin, consistent with Darwin's claim that it is ‘one long argument’. I provide detailed criticisms of all three current models, and then offer an alternative interpretation based on the view that there are three main argument strategies in the Origin, all supporting the ‘causal efficacy’ of Darwin's theory. This interpretation provides both a more unified treatment of the text, and some important implications concerning the relation between general philosophical models of scientific theory support and specific historical cases.

The key problem in the controversy over group selection is that of defining a criterion of group selection that identifies a distinct causal process that is irreducible to the causal process of individual selection. We aim to clarify this problem and to formulate an adequate model of irreducible group selection. We distinguish two types of group selection models, labeling them type I and type II models. Type I models are invoked to explain differences among groups in their respective rates of production of contained individuals. Type II models are invoked to explain differences among groups in their respective rates of production of distinct new groups. Taking Elliott Sober's model as an exemplar, we argue that although type I models have some biological importance—they force biologists to consider the role of group properties in influencing the fitness of organisms—they fail to identify a distinct group-level causal selection process. Type II models if properly framed, however, do identify a group-level causal selection process that is not reducible to individual selection. We propose such a type II model and apply it to some of the major candidates for group selection.

The controversy regarding the unit of selection is fundamentally a dispute about what is the correct causal structure of the process of evolution by natural selection and its ontological commitments. By characterizing the process as consisting of two essential steps—interaction and transmission—a singular answer to the unit question becomes ambiguous. With such an account on hand, two recent defenses of competing units of selection are considered. Richard Dawkins maintains that the gene is the appropriate unit of selection and Robert Brandon, in response, argues that the individual organism is better suited to the role. This paper argues that by making explicit the underlying questions that each of these views addresses, the apparent conflict can be resolved. Furthermore, such a resolution allows for a more complete and realistic understanding of the process of evolution by natural selection.

Entropy and information are both emerging as currencies of interdisciplinary dialogue, most recently in evolutionary theory. If this dialogue is to be fruitful, there must be general agreement about the meaning of these terms. That this is not presently the case owes principally to the supposition of many information theorists that information theory has succeeded in generalizing the entropy concept. The present paper will consider the merits of the generalization thesis, and make some suggestions for restricting both entropy and information to specific arenas of discourse.

This paper is an assessment of an attempt, by James Greeno, to measure the explanatory power of statistical theories by means of the notion of transmitted information (It). It is argued that It has certain features that are inappropriate in a measure of explanatory power. In particular, given a statistical theory T with explanans variables Si and explanandum variables Mj, it is argued that no plausible measure of explanatory power should depend on the probability P(Si) of occurrence of initial conditions in the systems to which T applies or the magnitudes of the conditional probabilities P(Mj/Si), in the manner in which IT does.

Horwich argues that we should reject metaphysical realism, but that we can preserve semantic realism by adhering to a redundancy theory of truth and a confirmationist account of linguistic understanding. But the latter will give us semantic realism only if it allows that the truth-values of sentences may transcend our recognitional capacities, and this is possible only insofar as we covertly reintroduce metaphysical realism. In spite of its intuitive appeal, we should not endorse semantic realism, but this need not bear upon the tenability of scientific realism.

This paper utilizes Scott domains (continuous lattices) to provide a mathematical model for the use of idealized and approximately true data in the testing of scientific theories. Key episodes from the history of science can be understood in terms of this model as attempts to demonstrate that theories are monotonic, that is, yield better predictions when fed better or more realistic data. However, as we show, monotonicity and truth of theories are independent notions. A formal description is given of the pragmatic virtues of theories which are monotonic. We also introduce the stronger concept of continuity and show how it relates to the finite nature of scientific computations. Finally, we show that the space of theories also has the structure of a Scott domain. This result provides an analysis of how one theory can be said to approximate another.

In this paper I wish to argue that counterfactual analyses of causation are inadequate. I believe the counterfactuals that are involved in counterfactual analyses of causation are often false, and thus the theories do not provide an adequate account of causation. This is demonstrated by the presentation of a counterexample to the counterfactual analyses of causation. I then present a unified theory of causation that is based upon probability and counterfactuals. This theory accounts for both deterministic and indeterministic causation, and is not subject to many of the traditional problems facing theories of causation.

Husserl argues in the Crisis that the prevalent tradition of positive science in his time had a philosophical core, called by him “Galilean science”, that mistook the quest for objective theory with the quest for truth. Husserl is here referring to Göttingen science of the Golden Years. For Husserl, theory “grows“ out of the “soil” of the prescientific, that is, pretheoretical, life-world. Scientific truth finally is to be sought not in theory but rather in the pragmatic-perceptual praxes of measurement. Husserl is faulted for taking measuring processes to be “infinitely perfectible“. The dependence of new scientific phenomena on the existence of prior “prescientific” inductive praxis is analyzed, also Husserl's residual objectivism and failure to appreciate the hermeneutic character of measurement. Though not a scientific (theory-)realist, neither was he an instrumentalist, but he was a scientific (phenomena-)realist.

Two EPR arguments are reviewed, for their own sake, and for the purpose of clarifying the status of “stochastic“ hidden variables. The first is a streamlined version of the EPR argument for the incompleteness of quantum mechanics. The role of an anti-instrumentalist (“realist”) interpretation of certain probability statements is emphasized. The second traces out one horn of a central foundational dilemma, the collapse dilemma; complex modal reasoning, similar to the original EPR, is used to derive determinateness (of all spin components of two spin-½ particles in the singlet state) from just (a form of) weak locality, result definiteness, and an assumption on propensities based on conservation. Theories meeting these conditions are therefore constrained by the Bell inequalities. Neither controversial assumptions of “strong locality” (“factorability”) nor of determinism are employed in the derivation. The categories of “stochastic hidden variables” are then analyzed; one can focus on “quasi-definite” theories, without loss of generality. A means of excluding these is proposed, based on a demand that certain ideal cases be accurately treated. Theorems from quantum measurement theory, sometimes cited as showing that such cases are not physically possible, are found inapplicable.

Several contemporary philosophers, like G. J. Whitrow, argue that it is logically impossible for the past to be infinite, and offer several arguments in support of this thesis. I believe their arguments are unsuccessful and aim to refute six of them in the six sections of the paper. One of my main criticisms concerns their supposition that an infinite series of past events must contain some events separated from the present event by an infinite number of intermediate events, and consequently that from one of these infinitely distant past events the present could never have been reached. I introduce several considerations to show that an infinite series of past events need not contain any events separated from the present event by an infinite number of intermediate events.

Husserl's (1970) discussion of “Galilean science“ is often dismissed as naïvely instrumentalist and hostile to science. He has been explicitly criticized for misunderstanding idealization in science, for treating the life world as a privileged conceptual framework, and for denying that science can in principle completely describe the world (because ordinary prescientific concepts are irreplaceable). I clarify Husserl's position concerning realism, and use this to show that the first two criticisms depend upon misinterpretations. The third criticism is well taken. Nevertheless, this is consistent with Husserl's fundamental claim that the manifestations of things are important to discuss, but are inaccessible to empirical science.

Introduction. During the past two decades philosophers of psychology have considered a large variety of computational models for philosophy of mind and more recently for cognitive science. Among the suggested models are computer programs, Turing machines, pushdown automata, linear bounded automata, finite state automata and sequential machines. Many philosophers have found finite state automata models to be the most appealing, for various reasons, although there has been no shortage of defenders of programs and Turing machines. A paper by Arthur Burks (1973) convinced me long ago that “all natural human functions” are, or can be fruitfully modeled to be, finite state automata with output. Further work in the field has reinforced this conviction. There is room, however, for the use of any of the above models in philosophy of mind and in the ongoing development of cognitive science.

Richard Otte (1985) has recently criticized the resolution of Simpson's paradox given by Nancy Cartwright (1979). He argues that there are difficulties with the version of the theory of probabilistic causality that Cartwright has developed, and that there is a way in which Simpson's paradox can arise that Cartwright's theory cannot handle. And Otte develops his own theory of probabilistic causality. I defend Cartwright's solution, and I argue that there are difficulties with the theory of probabilistic causality that Otte proposes.

A Quantum-Mechanical automaton, equipped with mechanisms for the measurement and the recording and the prediction of certain physical properties of the world, is described. It is inquired what sort of empirical description such an automaton would produce of itself. It turns out that this description would be a very novel one, one such as was never imagined in the conventional discussions of measurement.

L. Jonathan Cohen has written a number of important books and articles in which he argues that mathematical probability provides a poor model of much of what paradigmatically passes for sound reasoning, whether this be in the sciences, in common discourse, or in the law. In his book, The Probable and the Provable, Cohen elaborates six paradoxes faced by advocates of mathematical probability (PM) when treating issues of evidence as they would arise in a court of law. He argues that his system of inductive probability (PI) satisfactorily handles the issues that proved paradoxical for mathematical probability, and consequently PI deserves to be thought of as an important standard of rational thinking. I argue that a careful look at each of the alleged paradoxes shows that there is no conflict between mathematical probability and the law, except when for reasons of policy we opt for values in addition to accuracy maximization. Recognizing the role of such policies provides no basis for questioning the adequacy of PM. The significance of this critical treatment of Cohen's work is that those interested in revising the laws of evidence to allow for more explicitly mathematical approaches ought to feel that such revisions will not violate the spirit of forensic rationality.