"How do we go about weighing evidence, testing hypotheses and making inferences? According to the model of 'inference to the Best explanation', we work out what to inter from the evidence by thinking about what would actually explain that evidence, and we take the ability of a hypothesis to explain the evidence as a sign that the hypothesis is correct. In inference to the Best Explanation, Peter Lipton gives this important and influential idea the development and assessment it deserves." "The (...) second edition has been substantially enlarged and reworked, with a new chapter on the relationship between explanation and Bayesianism, and an extension and defence of the account of contractive explanation. It also includes an expanded defence of the claims that our inferences really are guided by diverse explanatory considerations, and that this pattern of inference can take us towards the truth. This edition of Inference to the Best Explanation has also been updated throughout and incudes a new bibliography."--BOOK JACKET. (shrink)

There is considerable disagreement about the epistemic value of novel predictive success, i.e. when a scientist predicts an unexpected phenomenon, experiments are conducted, and the prediction proves to be accurate. We survey the field on this question, noting both fully articulated views such as weak and strong predictivism, and more nascent views, such as pluralist reasons for the instrumental value of prediction. By examining the various reasons offered for the value of prediction across a range of inferential contexts , we (...) can see that neither weak nor strong predictivism captures all of the reasons for valuing prediction available. A third path is presented, Pluralist Instrumental Predictivism; PIP for short. (shrink)

Predictivists use the no miracle argument to argue that “novel” predictions are decisive evidence for theories, while mere accommodation of “old” data cannot confirm to a significant degree. But deductivists claim that since confirmation is a logical theory-data relationship, predicted data cannot confirm more than merely deduced data, and cite historical cases in which known data confirmed theories quite strongly. On the other hand, the advantage of prediction over accommodation is needed by scientific realists to resist Laudan’s criticisms of the (...) no miracle argument. So, if the deductivists are right, the most powerful argument for realism collapses. There seems to be an inescapable contradiction between these prima facie plausible arguments of predictivists and deductivists; but this puzzle can be solved by understanding what exactly counts as novelty, if novel predictions must support the no miracle argument, i.e., if they must be explainable only by the truth of theories. Taking my cues from the use-novelty tradition, I argue that (1) the predicted data must not be used essentially in building the theory or choosing the auxiliary assumptions. This is possible if the theory and its auxiliary assumptions are plausible independently of the predicted data, and I analyze the consequences of this requirement in terms of best explanation of diverse bodies of data. Moreover, the predicted data must be (2) a priori improbable, and (3) heterogeneous to the essentially used data. My proposed notion of novelty, therefore, is not historical, but functional. Hence, deductivists are right that confirmation is independent of time and of historical contingencies such as if the theorist knew a datum, used it, or intended to accommodate it. Predictivists, however, are right that not all consequences confirm equally, and confirmation is not purely a logical theory-data relation, as it crucially involves background epistemic conditions and the notion of best explanation. Conditions (1)–(3) make the difference between prediction and accommodation, and account for the confirming power of theoretical virtues such as non ad-hocness, non-fudging, non-overfitting, independence and consilience. I thus show that functional novelty (a) avoids the deductivist objections to predictivism, (b) is a gradual notion, in accordance with the common intuition that confirmation comes in degrees, and (c) supports the no miracle argument, so vindicating scientific realism. (shrink)

an observation to formulate a theory, it is no surprise that the resulting theory accurately captures that observation. However, when the theory makes a novel prediction—when it predicts an observation that was not used in its formulation—this seems to provide more substantial confirmation of the theory. This paper presents a new approach to the vexed problem of understanding the epistemic difference between prediction and accommodation. In fact, there are several problems that need to be disentangled; in all of them, the (...) key is the concept of overfitting. We float the hypothesis that accommodation is a defective methodology only when the methods used to accommodate the data fail to guard against the risk of overfitting. We connect our analysis with the proposals that other philosophers have made. We also discuss its bearing on the conflict between instrumentalism and scientific realism. Introduction Predictivisms—a taxonomy Observations Formulating the problem What might Annie be doing wrong? Solutions Observations explained Mayo on severe tests The miracle argument and scientific realism Concluding comments. (shrink)

Traditional analyses of the curve fitting problem maintain that the data do not indicate what form the fitted curve should take. Rather, this issue is said to be settled by prior probabilities, by simplicity, or by a background theory. In this paper, we describe a result due to Akaike [1973], which shows how the data can underwrite an inference concerning the curve's form based on an estimate of how predictively accurate it will be. We argue that this approach throws light (...) on the theoretical virtues of parsimoniousness, unification, and non ad hocness, on the dispute about Bayesianism, and on empiricism and scientific realism. * Both of us gratefully acknowledge support from the Graduate School at the University of Wisconsin-Madison, and NSF grant DIR-8822278 (M.F.) and NSF grant SBE-9212294 (E.S.). Special thanks go to A. W. F. Edwards.William Harper. Martin Leckey. Brian Skyrms, and especially Peter Turney for helpful comments on an earlier draft. (shrink)

This article argues that common intuitions regarding (a) the specialness of ‘use-novel’ data for confirmation and (b) that this specialness implies the ‘no-double-counting rule’, which says that data used in ‘constructing’ (calibrating) a model cannot also play a role in confirming the model’s predictions, are too crude. The intuitions in question are pertinent in all the sciences, but we appeal to a climate science case study to illustrate what is at stake. Our strategy is to analyse the intuitive claims in (...) light of prominent accounts of confirmation of model predictions. We show that on the Bayesian account of confirmation, and also on the standard classical hypothesis-testing account, claims (a) and (b) are not generally true; but for some select cases, it is possible to distinguish data used for calibration from use-novel data, where only the latter confirm. The more specialized classical model-selection methods, on the other hand, uphold a nuanced version of claim (a), but this comes apart from (b), which must be rejected in favour of a more refined account of the relationship between calibration and confirmation. Thus, depending on the framework of confirmation, either the scope or the simplicity of the intuitive position must be revised. (shrink)

The debate about the relative epistemic weights carried in favour of a theory by predictions of new phenomena as opposed to accommodations of already known phenomena has a long history. We readdress the issue through a detailed re-examination of a particular historical case that has often been discussed in connection with it—that of Mendeleev and the prediction by his periodic law of the three ‘new’ elements, gallium, scandium and germanium. We find little support for the standard story that these predictive (...) successes were outstandingly important in the success of Mendeleev's scheme. Accommodations played an equal role—notably that of argon, the first of the ‘noble gases’ to be discovered; and the methodological situation in this chemical example turns out to be in interesting ways different from that in other cases—invariably from physics—that have been discussed in this connection. The historical episode when accurately analysed provides support for a different account of the relative weight of prediction and accommodation—one that is further articulated here. (shrink)

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. (shrink)

There is a striking difference between the methodology of the young Einstein and that of the old. I argue that Einstein’s switch in the late 1910s from a moderate empiricism to an extreme rationalism should at least in part be understood against the background of his crushing personal and political experiences during the war years in Berlin. As a result of these experiences, Einstein started to put into practice what, drawing on Schopenhauer, he had preached for years, namely to use (...) science as his means of escaping from “the merely personal.” Whatever the exact sources of Einstein’s about-face, the older man has left us with a highly misleading picture of how the younger man achieved the successes that we still celebrate today. This has had a harmful influence on theoretical physics. If the young Einstein’s successes are any guide as to how successful theoretical physics is done, close adherence to general features of the empirical data is much more and mathematical elegance is much less important than the old Einstein wanted us to believe. (shrink)

According to the thesis of Strong Predictionism, we typically have stronger evidence for a theory if it was used to predict certain data, than if it was deliberately constructed to accommodate those same data, even if we fully grasp the theory and all the evidence on which it was based. This thesis faces powerful objections and the existing arguments in support of it are seriously flawed. I offer a new defence of Strong Predictionism which overcomes the objections and provides a (...) deeper understanding of the epistemic importance of prediction. I conclude by applying this account to strategies for defending scientific realism. (shrink)

Scientific theories are developed in response to a certain set of phenomena and subsequently evaluated, at least partially, in terms of the quality of fit between those same theories and appropriately distinctive phenomena. To differentiate between these two stages it is popular to describe the former as involving the accommodation of data and the latter as involving the prediction of data. Predictivism is the view that, ceteris paribus, correctly predicting data confers greater confirmation than successfully accommodating data. In this paper, (...) I take issue with a variety of predictivist theses, argue that their role for issues of confirmation is extremely limited, and attempt to account for the appeal that predictivism has enjoyed. Introduction Temporal Predictivism Heuristic Predictivism Weak Predictivism 4.1 Inference to better theories 4.2 Inference to better methods Arguments for Strong Heuristic Predictivism 5.1 Best explanations argument 5.2 Conditional support 5.3 Unique explanations Increased Explanatory Unification 6.1 Explaining what other theories can't 6.2 Contrived hypotheses 6.2 Strength and simplicity Conclusions CiteULike Connotea Del.icio.us What's this? (shrink)

The French physicist Jean Baptiste Perrin is widely credited with providing the conclusive argument for atomism. The most well-known part of Perrin’s argument is his description of thirteen different procedures for determining Avogadro’s number (N)–the number of atoms, ions, and molecules contained in a gram-atom, gram-ion, and gram-mole of a substance, respectively. Because of its success in ending the atomism debates Perrin’s argument has been the focus of much philosophical interest. The various philosophers, however, have reached different conclusions, not only (...) about the argument’s general rationale but also the role that the multiple determination of N played in it. This paper emphasizes the historical development of Perrin’s experimental work in order to understand the role that the multiple determination of molecular magnitudes played in his argument for molecular reality. It claims that Perrin used the multiple determination strategy to put forward an exceptionally strong no-coincidence argument to argue for both the correctness of the values for the molecular magnitudes determined and the validity of the auxiliary assumptions upon which the different determinations were based. The historicist approach also allows the identification of the elements responsible for the epistemic strength of Perrin’s no-coincidence argument. (shrink)