Experimental research is commonly held up as the paradigm of “good” science. Although experiment plays many roles in science, its classical role is testing hypotheses in controlled laboratory settings. Historical science (which includes work in geology, biology, and astronomy, as well as paleontology and archaeology) is sometimes held to be inferior on the grounds that its hypothesis cannot be tested by controlled laboratory experiments. Using contemporary examples from diverse scientific disciplines, this paper explores differences in practice between historical and experimental research vis-à-vis the testing of hypotheses. It rejects the claim that historical research is epistemically inferior. For as I argue, scientists engage in two very different patterns of evidential reasoning and, although there is overlap, one pattern predominates in historical research and the other pattern predominates in classical experimental research. I show that these different patterns of reasoning are grounded in an objective and remarkably pervasive time asymmetry of nature.

Recently, Rueger and Sharp (1996) and Koperski (1998) have been concerned to show that certain procedural accounts of model confirmation are compromised by non-linear dynamics. We suggest that the issues raised are better approached by considering whether chaotic data analysis methods allow for reliable inference from data. We provide a framework and an example of this approach.

In this paper I examine Salmon's response to two counterexamples to his conserved quantity (CQ) theory of causation. The first counterexample that I examine involves a time-wise gerrymandered world line of a series of patches of wall that is absorbing energy as a result of being illuminated in an astrodome. Salmon says that since the gerrymandered world line does not fulfill his “no-interaction requirement,” his CQ theory does not suffer from the counterexample. But I will argue that his response fails both at a theoretical level and at a practical level. In so doing I point out a problem for CQ theorists’ definition of a causal interaction. The second counterexample is concerned with a time-wise gerrymandered world line of a series of patches that are in shadow, in Hitchcock's well-known example. Salmon's response is based on a principle that Salmon thinks is derivable from the concept of a conserved quantity. However, I argue that the principle has a counterexample.

This essay examines logical empiricism and American pragmatism, arguing that American philosophy's embrace of logical empiricism in the 1930s was not a turning away from Dewey's pragmatism. It places both movements within scientific philosophy and finds two key points on which they agreed: their revolutionary ambitions and their social engineering sensibility. The essay suggests that the disagreement over emotivism in ethics should be placed within the context of a larger issue on which the movements disagreed: demarcationism and imperialism.

Sober and Lewontin's critique of genic selectionism is based upon the principle that a unit of selection should make a context-independent contribution to fitness. Critics have effectively shown that this principle is flawed. In this paper I show that the context independence principle is an instance of a more general principle for characterizing causes,called the contextual unanimity principle. I argue that this latter principle, while widely accepted, is erroneous. What is needed is to replace the approach to causality characterized by the contextual unanimity criterion with an approach based on the concept of causal mechanism. After sketching such an approach, I show how it can be used to shed light on the units of selection problem.

We examine, from the partial structures perspective, two forms of applicability of mathematics: at the “bottom” level, the applicability of theoretical structures to the “appearances”, and at the “top” level, the applicability of mathematical to physical theories. We argue that, to accommodate these two forms of applicability, the partial structures approach needs to be extended to include a notion of “partial homomorphism”. As a case study, we present London's analysis of the superfluid behavior of liquid helium in terms of Bose-Einstein statistics. This involved both the introduction of group theory at the top level, and some modeling at the “phenomenological” level, and thus provides a nice example of the relationships we are interested in. We conclude with a discussion of the “autonomy” of London's model.

The need to align multiple experimental procedures and produce converging results so as to demonstrate that the phenomenon under investigation is real and not an artifact is a commonplace both in scientific practice and discussions of scientific methodology (Campbell and Stanley 1963; Wimsatt 1981). Although sometimes this is the purpose of aligning techniques, often there is a different purpose—multiple techniques are sought to supply different perspectives on the phenomena under investigation that need to be integrated to answer the questions scientists are asking. After introducing this function, I will illustrate it by considering three of the major techniques in cognitive neuroscience for linking cognitive function with neural structure.

The semantic concept of information is one of the most important, and one of the most problematical concepts in biology. I suggest a broad definition of biological information: a source becomes an informational input when an interpreting receiver can react to the form of the source (and variations in this form) in a functional manner. The definition accommodates information stemming from environmental cues as well as from evolved signals, and calls for a comparison between information-transmission in different types of inheritance systems—the genetic, the epigenetic, the behavioral, and the cultural-symbolic. This comparative perspective highlights the different ways in which information is acquired and transmitted, and the role that such information plays in heredity and evolution. Focusing on the special properties of the transfer of information, which are very different from those associated with the transfer of materials or energy, also helps to uncover interesting evolutionary effects and suggests better explanations for some aspects of the evolution of communication.

The so-called “non-commutativity” of probability kinematics has caused much unjustified concern. When identical learning is properly represented, namely, by identical Bayes factors rather than identical posterior probabilities, then sequential probability-kinematical revisions behave just as they should. Our analysis is based on a variant of Field's reformulation of probability kinematics, divested of its (inessential) physicalist gloss.

Experimenters sometimes insist that it is unwise to examine data before determining how to analyze them, as it creates the potential for biased results. I explore the rationale behind this methodological guideline from the standpoint of an error statistical theory of evidence, and I discuss a method of evaluating evidence in some contexts when this predesignation rule has been violated. I illustrate the problem of potential bias, and the method by which it may be addressed, with an example from the search for the top quark. A point in favor of the error statistical theory is its ability, demonstrated here, to explicate such methodological problems and suggest solutions, within the framework of an objective theory of evidence.

In their recent book, Elliott Sober and David Wilson (1998) argue that evolutionary biologists have wrongly regarded kinship as the exclusive means by which altruistic behavior can evolve, at the expense of other mechanisms. I argue that Sober and Wilson overlook certain genetical considerations which suggest that kinship is likely to be a more powerful means for generating complex altruistic adaptations than the alternative mechanisms they propose.

Bechtel and Mundale (1999) argue that multiple realizability is not plausible. They point out that neuroscientists assume that psychological traits are realized similarly in homologous brain structures and contend that a biological aspect of the brain that is relevant to neuropsychological state individuation provides evidence against multiple realizability. I argue that Bechtel and Mundale adduce the wrong sort of evidence against multiple realizability. Homologous traits do not provide relevant evidence. It is homoplasious traits of brains that can provide evidence for or against multiple realizability.

This paper discusses features of an epistemically valuable form of evidence that raise troubles for received and new epistemological treatments of experimental evidence.

There is a bias in neuroscience toward localizing and modularizing brain functions. Single cell recording, imaging studies, and the study of neurological deficits all feed into the Gallian view that different brain areas do different things and the things being done are confined to particular processing streams. At the same time, there is a growing sentiment that brains probably don't work like that after all; it is better to conceive of them as fundamentally distributed units, multi-tasking at every level. This sentiment, however, is much less congenial to the tried-and-true experimental protocols available today and to theorizing about the brain in general. This essay examines the tension between current experimental methods and large-scale views of the brain. We argue that this disconnection between experiment and what really are guiding theoretical metaphors seriously impedes progress in neuroscience.

I examine the epistemic import of collaborative research in science. I develop and defend a functional explanation for its growing importance. Collaborative research is becoming more popular in the natural sciences, and to a lesser degree in the social sciences, because contemporary research in these fields frequently requires access to abundant resources, for which there is great competition. Scientists involved in collaborative research have been very successful in accessing these resources, which has in turn enabled them to realize the epistemic goals of science more effectively than other scientists, thus creating a research environment in which collaboration is now the norm.

Naive deductive accounts of confirmation have the undesirable consequence that if E confirms H, then E also confirms the conjunction H & X, for any X—even if X is utterly irrelevant to H (and E). Bayesian accounts of confirmation also have this property (in the case of deductive evidence). Several Bayesians have attempted to soften the impact of this fact by arguing that—according to Bayesian accounts of confirmation— E will confirm the conjunction H & X less strongly than E confirms H (again, in the case of deductive evidence). I argue that existing Bayesian “resolutions” of this problem are inadequate in several important respects. In the end, I suggest a new-and-improved Bayesian account (and understanding) of the problem of irrelevant conjunction.

Methodological norms are seen as rules defining a competitive game, and it is argued that rational recognition-seeking scientists can reach a collective agreement about which specific norms serve better their individual interests, especially if the choice is made ‘under a veil of ignorance’, i.e., before knowing what theory will be proposed by each scientist. Norms for theory assessment are distinguished from norms for theory choice (or inference rules), and it is argued that pursuit of recognition only affects this second type of rule. An inference rule similar to ‘eliminative induction’ is defended on the basis of such a possible agreement. According to this contractarian approach, both the explanation and the justification of scientific norms only need to refer to the preferences of individual scientists, without assuming the existence of ‘collective’ points of view.