The use of null hypothesis significance testing (NHST) in psychology has been under sustained attack, despite its reliable use in the notably successful, so-called “hard” areas of psychology, such as perception and cognition. I argue that, in contrast to merely methodological analyses of hypothesis testing (in terms of “test severity,” or other confirmation-theoretic notions), only a patently metaphysical position can adequately capture the uneven but undeniable successes of theories in “hard psychology.” I contend that Measured Realism satisfies this description, and characterizes the role of NHST in hard psychology.

I analyze the two main theses of Helmholtz's “The Applicability of the Axioms to the Physical World,” in which he argued that the axioms of Euclidean geometry are not, as his neo-Kantian opponents had argued, binding on any experience of the external world. This required two argumentative steps: 1) a new account of the structure of our representations which was consistent both with the experience of our (for him) Euclidean world and with experience of a non-Euclidean one, and 2) a demonstration of why geometric propositions are essentially connected to material and temporal aspects of experience. The effect of Helmholtz's discussion is to throw into relief an intermediate category of metrological objects—objects which are required for the properly theoretical activity of doing physical science (in this sense, a priori requirements for doing science), all while being recognizably contingent aspects of experience.

Credibility in a scientific community (sensu Shapin) is a vicarious selector (sensu Campbell) for the reliability of reports by individual scientists or institutions. Similarly, images from a microscope (sensu Hacking) are vicarious selectors for studying specimens. Working at different levels, the process of indirect reasoning and checking indicates a unity to experimentalist and sociological perspectives, along with a resonance of strategies for assessing reliability. The perspective sketched here can open dialogue between philosophical and sociological interpretations of science and resolves at least one tension regarding the “primary” factors by which scientists evaluate claims.

Functional explanation in the social sciences is the focal point for conflict between individualistic and social modes of explanation. While the agent thought she was acting for reasons, the functional explanation seems to reveal the hidden strings of the puppet master. This essay argues that the conflict is merely apparent. The erotetic model of explanation is used to analyze the forms of intentional action and functional explanations. Two explanations conflict if either the presuppositions of their respective why-questions conflict or the typical answers identified by their relevance criteria conflict. While a functional explanation may have the same topic and foil as an intentional action explanation, both the why-questions and their typical answers are compatible.

This paper defends a social functionalist interpretation, modeled on psychological functionalism, of the meanings of social facts. Social functionalism provides a better explanation of the possibility of interpreting other cultures than approaches that identify the meanings of social facts with either mental states or behavior. I support this claim through a functionalist reinterpretation of sociological accounts of the categories that identify them with their collective representations. Taking the category of causality as my example, I show that if we define it instead in terms of its functional relations to moral rules, it becomes easier to recognize in other cultures.

I shall introduce the units of decision problem in the theory of decision, which as I shall explain is a sibling to the units of selection problem in evolutionary theory. And I shall present an argument to the effect that, contrary to Bayesian wisdom on the subject, undertaking decision in group settings (in multi-individual units) violates no precepts of rationality.

Carter and Leslie (1996) have argued, using Bayes's theorem, that our being alive now supports the hypothesis of an early ‘Doomsday’. Unlike some critics (Eckhardt 1997), we accept their argument in part: given that we exist, our existence now indeed favors ‘Doom sooner’ over ‘Doom later’. The very fact of our existence, however, favors ‘Doom later’. In simple cases, a hypothetical approach to the problem of ‘old evidence’ shows that these two effects cancel out: our existence now yields no information about the coming of Doom. More complex cases suggest a move from countably additive to non-standard probability measures.

The likelihood principle of Bayesian statistics implies that information about the stopping rule used to collect evidence does not enter into the statistical analysis. This consequence confers an apparent advantage on Bayesian statistics over frequentist statistics. In the present paper, I argue that information about the stopping rule is nevertheless of value for an assessment of the reliability of the experiment, which is a pre-experimental measure of how well a contemplated procedure is expected to discriminate between hypotheses. I show that, when reliability assessments enter into inquiries, some stopping rules prescribing optional stopping are unacceptable to both Bayesians and frequentists.

Contemporary Bayesian confirmation theorists measure degree of (incremental) confirmation using a variety of non-equivalent relevance measures. As a result, a great many of the arguments surrounding quantitative Bayesian confirmation theory are implicitly sensitive to choice of measure of confirmation. Such arguments are enthymematic, since they tacitly presuppose that certain relevance measures should be used (for various purposes) rather than other relevance measures that have been proposed and defended in the philosophical literature. I present a survey of this pervasive class of Bayesian confirmation-theoretic enthymemes, and a brief analysis of some recent attempts to resolve the problem of measure sensitivity.

Bayesian psychology, in what is perhaps its most familiar version, is incomplete: Jeffrey conditionalization is not a complete account of rational belief change. Jeffrey conditionalization is sensitive to the order in which the evidence arrives. This order effect can be so pronounced as to call for a belief adjustment that cannot be understood as an assimilation of incoming evidence by Jeffrey's rule. Hartry Field's reparameterization of Jeffrey's rule avoids the order effect but fails as an account of how new evidence should be assimilated.

In the curve fitting problem two conflicting desiderata, simplicity and goodness-of-fit pull in opposite directions. To solve this problem, two proposals, the first one based on Bayes's theorem criterion (BTC) and the second one advocated by Forster and Sober based on Akaike's Information Criterion (AIC) are discussed. We show that AIC, which is frequentist in spirit, is logically equivalent to BTC, provided that a suitable choice of priors is made. We evaluate the charges against Bayesianism and contend that AIC approach has shortcomings. We also discuss the relationship between Schwarz's Bayesian Information Criterion and BTC.

This paper analyzes Deborah Mayo's recent criticism of use-novelty requirement. She claims that her severity criterion captures actual scientific practice better than use-novelty, and that use-novelty is not a necessary condition for severity. Even though certain cases in which evidence used for the construction of the hypothesis can test the hypothesis severely, I do not think that her severity criterion fits better with our intuition about good tests than use-novelty. I argue for this by showing a parallelism in terms of severity between the confidence interval case and what she calls ‘gellerization’. To account for the difference between these cases, we need to take into account certain additional considerations like a systematic neglect of relevant alternatives.

A (normalized) interpolant I in Craig's theorem is a kind of explanation why the consequence relation (from F to G) holds. This is because I is a summary of the interaction of the configurations specified by F and G, respectively, that shows how G follows from F. If explaining E means deriving it from a background theory T plus situational information A and if among the concepts of E we can separate those occurring only in T or only in A, then the interpolation theorem applies in two different ways yielding two different explanations and two different covering laws.

Van Fraassen (1989) argues that Inference to the Best Explanation is incoherent in the sense that adopting it as a rule for belief change will make one susceptible to a dynamic Dutch book. The present paper argues against this. A strategy is described that allows us to infer to the best explanation free of charge.

Charles S. Peirce argued that, besides deduction and induction, there is a third mode of inference which he called “hypothesis” or “abduction.” He characterized abduction as reasoning “from effect to cause,” and as “the operation of adopting an explanatory hypothesis.” Peirce's ideas about abduction, which are related also to historically earlier accounts of heuristic reasoning (the method of analysis), have been seen as providing a logic of scientific discovery. Alternatively, abduction is interpreted as giving reasons for pursuing a hypothesis. Inference to the best explanation (IBE) has also been regarded as an important mode of justification, both in everyday life, detective stories, and science. In particular, scientific realism has been defended by an abductive nomiracle argument (Smart, Putnam, Boyd), while the critics of realism have attempted to show that this appeal to abduction is question-begging, circular, or incoherent (Fine, Laudan, van Fraassen). This paper approaches these issues by distinguishing weaker and stronger forms of abduction, and by showing how these types of inferences can be given Peircean and Bayesian probabilistic reconstructions.

There seem to be some very good reasons for a philosopher of science to be a deductivist about scientific reasoning. Deductivism is apparently connected with a demand for clarity and definiteness in the reconstruction of scientists' reasonings. And some philosophers even think that deductivism is the way around the problem of induction. But the deductivist image is challenged by cases of actual scientific reasoning, in which hard-to-state and thus discursively ill-defined elements of thought nonetheless significantly condition what practitioners accept as cogent argument. And arguably, these problem cases abound. For example, even geometry—for most of its history—was such a problem case, despite its exactness and rigor. It took a tremendous effort on the part of Hilbert and others, to make geometry fit the deductivist image. Looking to the empirical sciences, the problems seem worse. Even the most exact and rigorous of empirical sciences—mechanics—is still the kind of problem case which geometry once was. In order for the deductivist image to fit mechanics, Hilbert's sixth problem (for mechanics) would need to be solved. This is a difficult, and perhaps ultimately impossible task, in which the success so far achieved is very limited. I shall explore some consequences of this for realism as well as for deductivism. Through discussing links between non-monotonicity, skills, meaning, globality in cognition, models, scientific understanding, and the ideal of rational unification, I argue that deductivists can defend their image of scientific reasoning only by trivializing it, and that for the adequate illumination of science, insights from anti-deductivism are needed as much as those which come from deductivism.

A first step is taken towards articulating a constructive empiricist philosophy of mathematics, thus extending van Fraassen's account to this domain. In order to do so, I adapt Field's nominalization program, making it compatible with an empiricist stance. Two changes are introduced: (a) Instead of taking conservativeness as the norm of mathematics, the empiricist countenances the weaker notion of quasi-truth (as formulated by da Costa and French), from which the formal properties of conservativeness are derived; (b) Instead of quantifying over spacetime regions, the empiricist only admits quantification over occupied regions, since this is enough for his or her needs.

In this paper I offer an ‘integrating account’ of singular causation, where the term ‘integrating’ refers to the following program for analysing causation. There are two intuitions about causation, both of which face serious counterexamples when used as the basis for an analysis of causation. The ‘process’ intuition, which says that causes and effects are linked by concrete processes, runs into trouble with cases of ‘misconnections’, where an event which serves to prevent another fails to do so on a particular occasion and yet the two events are linked by causal processes. The chance raising intuition, according to which causes raise the chance of their effects, easily accounts for misconnections but faces the problem of chance lowering causes, a problem easily accounted for by the process approach. The integrating program attempts to provide an analysis of singular causation by synthesising the two insights, so as to solve both problems.

In this paper I show that extant versions of the integrating program due to Eells, Lewis, and Menzies fail to account for the chance-lowering counterexample. I offer a new diagnosis of the chance lowering case, and use that as a basis for an integrating account of causation which does solve both cases. In doing so, I accept various assumptions of the integrating program, in particular that there are no other problems with these two approaches. As an example of the process account, I focus on the recent CQ theory of Wesley Salmon (1997).

David Lewis's best-system analysis of laws of nature is perhaps the best known sophisticated regularity theory of laws. Its strengths are widely recognized, even by some of its ablest critics. Yet it suffers from what appears to be a glaring weakness: It seems to grant an arbitrary privilege to the standards of our own scientific culture. I argue that by reformulating, or reinterpreting, Lewis's exposition of the best-system analysis, we arrive at a view that is free of this weakness. The resulting theory of laws has the surprising consequence that the term “law of nature” is indexical.

I consider the problem of extending Reichenbach's principle of the common cause to more than two events, vis-à-vis an example posed by Bernstein. It is argued that the only reasonable extension of Reichenbach's principle stands in conflict with a recent proposal due to Horwich. I also discuss prospects of the principle of the common cause in the light of these and other difficulties known in the literature and argue that a more viable version of the principle is the one provided by Penrose and Percival (1962).