This paper investigates the revolutionary conceptual changes that took place when the phlogiston theory of Stahl was replaced by the oxygen theory of Lavoisier. Using techniques drawn from artificial intelligence, it represents the crucial stages in Lavoisier's conceptual development from 1772 to 1789. It then sketches a computational theory of conceptual change to account for Lavoisier's discovery of the oxygen theory and for the replacement of the phlogiston theory.

In recent years philosophers have attempted to clarify the units of selection controversy in evolutionary biology by offering conceptual analyses of the term ‘unit of selection’. A common feature of many of these analyses is an emphasis on the claim that units of selection are entities exhibiting heritable variation in fitness. In this paper I argue that the demand that units of selection be characterized in terms of heritability is unnecessary, as well as undesirable, on historical, theoretical, and philosophical grounds. I propose a positive account of the proper referent of the term ‘unit of selection’, distinguishing between the processes of evolution and phenotypic selection. The main result of this analysis is greater clarity about the conceptual structure of evolutionary theory.

Paul Churchland argues that the continuity of human intellectual development provides evidence against folk psychology and traditional epistemology, since these latter find purchase only at the later stages of intellectual development. He supports this contention with an analogy from the history of thermodynamics. Careful attention to the thermodynamics analogy shows that the argument from continuity does not provide independent support for eliminative materialism. The argument also rests upon claims about continuity which do not support the claim that the continuity of intellectual development is evidence for the elimination of folk psychology. Traditional epistemology and folk psychology should not yet be abandoned.

An introduction to the model-theoretic approach in the philosophy of science is given and it is argued that this program is further enhanced by the introduction of partial structures. It is then shown that this leads to a natural and intuitive account of both “iconic” and mathematical models and of the role of the former in science itself.

We consider drawbacks to scientific methods that prefer simple hypotheses to complex ones that cover the same data. The discussion proceeds in the context of a precise model of scientific inquiry.

This note makes a contribution to the issue raised in a paper by Popper and Miller (1983) in which it was claimed that probabilistic support is purely deductive. Developing R. C. Jeffrey's remarks, a new general approach to the crucial concept of “going beyond” is here proposed. By means of it a quantitative measure of the inductive component of a probabilistic inference is reached. This proposal leads to vindicating the view that typical predictive probabilistic inferences by enumeration and analogy are purely inductive.

The theory of evolution is supported by the theory of genetics, which provides a single causal mechanism to explain the activities of replicators and interactors. A common misrepresentation of the theory of evolution, however, is that interaction (involving interactors), and transmission (involving replicators), are distinct causal processes. Sandra Mitchell (1987) is misled by this. I discuss why only a single causal mechanism is working in evolution and why it is sufficient. Further, I argue that Mitchell's mistaken view of the causal mechanism in evolution prevents her from resolving the conflict between Dawkins and Brandon. I conclude that the unit-of-selection question remains very much alive.

This paper is concerned with a recent argument of Jerry Fodor's to the effect that the frame problem in artificial intelligence is in principle insoluble. Fodor's argument is based on his contention that the mind is divided between encapsulated modular systems for information processing and ‘central systems’ for non-demonstrative inference. I argue that positing central systems is methodologically unsound, and in fact involves a muddle that bears a strong family resemblance to the basic error in dualism. I therefore conclude that Fodor's position on the frame problem should be rejected.

The kinematical principle of Equal Passage Times (EPT) was introduced by Winnie in his 1970 derivation of the relativistic coordinate transformations compatible with arbitrary synchrony conventions in one-dimensional space. In this paper, the claim by Winnie and later Giannoni that EPT is a direct consequence of the relativity principle is questioned. It is shown that EPT, given Einstein's 1905 postulates, is equivalent to the relativistic (synchrony independent) clock retardation principle, and that for standard synchrony it reduces to an isotropy condition for contraction (and dilation) effects.