We show there is only one consistent way to update a probability assignment, that given by Bayes's rule. The price of inconsistent updating is a loss of efficiency. The implications of this for the problem of induction are discussed.

Much contemporary epistemology is informed by a kind of confirmational holism, and a consequent rejection of the assumption that all confirmation rests on experiential certainties. Another prominent theme is that belief comes in degrees, and that rationality requires apportioning one's degrees of belief reasonably. Bayesian confirmation models based on Jeffrey Conditionalization attempt to bring together these two appealing strands. I argue, however, that these models cannot account for a certain aspect of confirmation that would be accounted for in any adequate holistic confirmation theory. I then survey the prospects for constructing a formal epistemology that better accommodates holistic insights.

Philip Kitcher has proposed a theory of explanation based on the notion of unification. Despite the genuine interest and power of the theory, I argue here that the theory suffers from a fatal deficiency: It is intrinsically unable to account for the asymmetric structure of explanation, and thus ultimately falls prey to a problem similar to the one which beset Hempel's D-N model. I conclude that Kitcher is wrong to claim that one can settle the issue of an argument's explanatory force merely on the basis of considerations about the unifying power of the argument pattern the argument instantiates.

In a number of publications, John Earman has advocated a tertium quid to the usual dichotomy between substantivalism and relationism concerning the nature of spacetime. The idea is that the structure common to the members of an equivalence class of substantival models is captured by a Leibniz algebra which can then be taken to directly characterize the intrinsic reality only indirectly represented by the substantival models. An alleged virtue of this is that, while a substantival interpretation of spacetime theories falls prey to radical local indeterminism, the Leibniz algebras do not.

I argue that the program of Leibniz algebras is subject to radical local indeterminism to the same extent as substantivalism. In fact, for the category of topological spaces of interest in spacetime physics, the program is equivalent to the original spacetime approach. Moreover, the motivation for the program—that isomorphic substantival models should be regarded as representing the same physical situation—is misguided.

Standard philosophical discussions of theory-ladeness assume that observational evidence consists of perceptual outputs (or reports of such outputs) that are sentential or propositional in structure. Theory-ladeness is conceptualized as having to do with logical or semantical relationships between such outputs or reports and background theories held by observers. Using the recent debate between Fodor and Churchland as a point of departure, we propose an alternative picture in which much of what serves as evidence in science is not perceptual outputs or reports of such outputs and is not sentential in structure.

Figuring prominently in their decisions regarding which theories to pursue are scientists' appeals to the promise or lack of promise of those theories. Yet philosophy of science has had little to say about how one is to assess theory promise. This essay identifies several indices that might be consulted to determine whether or not a theory is promising and worthy of pursuit. Various historical examples of appeals to such indices are introduced.

We examine two approaches to functions: etiological and forward-looking. In the context of functions, we raise the question, familiar to philosophers of mind, about the explanatory role of properties that are not supervenient on the mere dispositional features of a system. We first argue that the question has no easy answer in either of the two approaches. We then draw a parallel between functions and goal directedness. We conclude by proposing an answer to the question: The explanatory importance of nonsupervenient properties (like having the function of doing something, or like being goal-directed) does not lie in any special causal mechanism through which these properties bring about their effects; it lies rather in the different classification of the explananda types that these properties generate.

In his important paper “1953 and All That: A Tale of Two Sciences” (1984), Philip Kitcher defends biological antireductionism, arguing that the division of biology into subfields such as classical and molecular genetics is “not simply … a temporary feature of our science stemming from our cognitive imperfections but [is] the reflection of levels of organization in nature” (p. 371). In a recent discussion of Kitcher's views, Alexander Rosenberg has argued, first, that Kitcher has shown that the reduction of classical to molecular genetics is impossible only because of our intellectual limitations and, second, that this kind of antireductionism supports an instrumentalist approach to biological theory. I argue that both of Rosenberg's claims should be rejected despite the fact that Kitcher misdiagnoses the central reason for the failure of reduction.

This paper takes up the cause of species pluralism. An argument for species pluralism is provided and standard monist objections to pluralism are answered. A new form of species pluralism is developed and shown to be an improvement over previous forms. This paper also offers a general foundation on which to base a pluralistic approach to biological classification.

Roger Jones asks what Newtonian realists should be realists about, given that there are four empirically equivalent formulations of Newtonian mechanics which have different ontological commitments and explanatory mechanisms. A realist answer is sketched: Newtonians should be realists about what the best metaphysical considerations dictate, where the best metaphysical considerations are those which have yielded the best physics. Metaphysical considerations are required within physics, just as they are required to eliminate idealist and surrealist theories which are empirically equivalent to realist ones. Realists must reject the positivist assumption that empirically equivalent theories are explanatory and evidential equivalents, too.