This paper discusses "Russell's hypothesis" that the world sprang into existence five minutes ago. The three most widely accepted "solutions" to the Russell's hypothesis problem are shown to be unsatisfactory. Two main points of interest are involved in the paper's discussion. First, I show all the widely accepted "solutions" to be unacceptable by using the same device--alternatives to Russell's hypothesis. The device, which has never previously been applied to this problem, is a familiar one in discussions of the problem of (...) induction. (Parallels between these two problems are sketched at the paper's end.) Second, the discussion introduces the new concept of crucial consequences for hypotheses that are otherwise different in only trivial linguistic ways. (shrink)

This target article presents a new computational theory of explanatory coherence that applies to the acceptance and rejection of scientific hypotheses as well as to reasoning in everyday life, The theory consists of seven principles that establish relations of local coherence between a hypothesis and other propositions. A hypothesis coheres with propositions that it explains, or that explain it, or that participate with it in explaining other propositions, or that offer analogous explanations. Propositions are incoherent with each other if they (...) are contradictory, Propositions that describe the results of observation have a degree of acceptability on their own. An explanatory hypothesis is acccpted if it coheres better overall than its competitors. The power of the seven principles is shown by their implementation in a connectionist program called ECHO, which treats hypothesis evaluation as a constraint satisfaction problem. Inputs about the explanatory relations are used to create a network of units representing propositions, while coherende and incoherence relations are encoded by excitatory and inbihitory links. ECHO provides an algorithm for smoothly integrating theory evaluation based on considerations of explanatory breadth, simplicity, and analogy. It has been applied to such important scientific cases as Lovoisier's argument for oxygen against the phlogiston theory and Darwin's argument for evolution against creationism, and also to cases of legal reasoning. The theory of explanatory coherence has implications for artificial intelligence, psychology, and philosophy. (shrink)

According to principles of probability coordination, such as Miller's Principle or Lewis's Principal Principle, you ought to set your subjective probability for an event equal to what you take to be the objective probability of the event. For example, you should expect events with a very high probability to occur and those with a very low probability not to occur. This paper examines the grounds of such principles. It is argued that any attempt to justify a principle of probability coordination (...) encounters the same difficulties as attempts to justify induction. As a result, no justification can be found. (shrink)

The dilemma objection charges that ‘weak’ versions of the precautionary principle (PP) are vacuous while ‘strong’ ones are incoherent. I respond that the ‘weak’ versus ‘strong’ distinction is misleading and should be replaced with a contrast between PP as a meta-rule and PP proper. Meta versions of PP require that the decision-making procedures used for environmental policy not be susceptible to paralysis by scientific uncertainty. Such claims are substantive because they often recommend against basing environmental policy decisions on cost–benefit analysis. (...) I argue that the second horn of the dilemma fails as a result of disregarding the role of proportionality in applications of PP. (shrink)

The paper is based on ranking theory, a theory of degrees of disbelief (and hence belief). On this basis, it explains enumerative induction, the confirmation of a law by its positive instances, which may indeed take various schemes. It gives a ranking theoretic explication of a possible law or a nomological hypothesis. It proves, then, that such schemes of enumerative induction uniquely correspond to mixtures of such nomological hypotheses. Thus, it shows that de Finetti's probabilistic representation theorems may be transformed (...) into an account of confirmation of possible laws and that enumerative induction is equivalent to such an account. The paper concludes with some remarks about the apriority of lawfulness or the uniformity of nature. (shrink)

Bernoulli’s 1713 golden theorem is viewed retrospectively in the context of modern model-based frequentist inference that revolves around the concept of a prespecified statistical model Mθx\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathcal{M}}_{{{\varvec{\uptheta}}}} \left( {\mathbf{x}} \right)$$\end{document}, defining the inductive premises of inference. It is argued that several widely-accepted claims relating to the golden theorem and frequentist inference are either misleading or erroneous: (a) Bernoulli solved the problem of inference ‘from probability to frequency’, and thus (b) the golden theorem (...) cannot justify an approximate Confidence Interval (CI) for the unknown parameter θ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta$$\end{document}, (c) Bernoulli identified the probability PA\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$P\left( A \right)$$\end{document} with the relative frequency 1n∑k=1nxk\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{1}{n}\sum\nolimits_{k = 1}^{n} {x_{k} }$$\end{document} of event A as a result of conflating f(x0|θ)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f({\mathbf{x}}_{0} |\theta )$$\end{document} with f(θ|x0),\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f(\theta |{\mathbf{x}}_{0} ),$$\end{document} where x0\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathbf{x}}_{0}$$\end{document} denotes the observed data, and (d) the same ‘swindle’ is currently perpetrated by the p value testers. In interrogating the claims (a)–(d), the paper raises several foundational issues that are particularly relevant for statistical induction as it relates to the current discussions on the replication crises and the trustworthiness of empirical evidence, arguing that: [i] The alleged Bernoulli swindle is grounded in the unwarranted claim θ^nx0≃θ∗,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hat{\theta }_{n} \left( {{\mathbf{x}}_{0} } \right) \simeq \theta^{*},$$\end{document} for a large enough n, where θ^nX\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hat{\theta }_{n} \left( {\mathbf{X}} \right)$$\end{document} is an optimal estimator of the true value θ∗\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta^{*}$$\end{document} of θ. [ii] Frequentist error probabilities are not conditional on hypotheses (H0 and H1) framed in terms of an unknown parameter θ since θ is neither a random variable nor an event. [iii] The direct versus inverse inference problem is a contrived and misplaced charge since neither conditional distribution f(x0|θ)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f({\mathbf{x}}_{0} |\theta )$$\end{document} and f(θ|x0)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f(\theta |{\mathbf{x}}_{0} )$$\end{document} exists (formally or logically) in model-based (Mθx\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathcal{M}}_{{{\varvec{\uptheta}}}} \left( {\mathbf{x}} \right)$$\end{document}) frequentist inference. (shrink)

Identifying the necessary and sufficient conditions for individuating and classifying diseases is a matter of great importance in the fields of law, ethics, epidemiology, and of course, medicine. In this paper, I first propose a means of achieving this goal, ensuring that no two distinct disease-types could correctly be ascribed to the same disease-token. I then posit a metaphysical ontology of diseases—that is, I give an account of what a disease is. This is essential to providing the most effective means (...) of interfering with disease processes. Following existing work in the philosophy of medicine and epidemiology (primarily Christopher Boorse; Caroline Whitbeck; Alexander Broadbent), philosophy of biology (Joseph LaPorte; D.L. Hull), conditional analyses of causation (J.L. Mackie; David Lewis), and recent literature on dispositional essentialism (Stephen Mumford and Rani Anjum; Alexander Bird), I endorse a dispositional conception of disease. Following discussion of various conceptions of disease-identity, their relations to the clinical and pathological effects of the diseases in question, and how diseases are treated, I conclude (i) that diseases should be individuated by their causes, and (ii) that diseases are causal processes best seen as simultaneously acting sequences of mutually manifesting dispositions. (shrink)

Reichenbach's well-known distinction between the context of discovery and the context of justification has recently come under attack from several quarters. In this paper I attempt to reconsider the distinction and evaluate various recent criticisms of it. These criticisms fall into two main groups: those which directly challenge Reichenbach's distinction; and those which (I argue) indirectly but no less seriously challenge that distinction by rejecting the related distinction between psychology and epistemology, and defending the "naturalizing" of epistemology. I argue that (...) these recent criticisms fail, and that the distinction remains an important conceptual tool necessary for an adequate understanding of the way in which scientific claims purport to appropriately portray our natural environment. (shrink)

The paper claims that analytic philosophy has failed within the philosophy of science due to the way the dynamic aspect of scientific theories is traditionally treated. On the formal side this failure manifests itself in the first-order logical and the model-theoretic analyses of scientific theories. An amendment of the treatment is sketched. It is based on using model generation, of the kind used in proving the Completeness Theorem for first-order logic, in such a way that some dynamic quantities in the (...) dynamic theory are formally represented as functions relating closed terms and sentences to their interpretations in a generated model. (shrink)

I argue that none of the usual interpretations of probability provide an adequate interpretation of probabilistic theories in science. Assuming that the aim of such theories is to capture noisy relationships in the world, I suggest that we do not have to give them classical truth-conditional content at all: their probabilities can remain uninterpreted. Indirectly, this account turns out to explain what is right about the frequency interpretation, the best-systems interpretation, and the epistemic interpretation.

This paper describes the corner-stones of a means-ends approach to the philosophy of inductive inference. I begin with a fallibilist ideal of convergence to the truth in the long run, or in the 'limit of inquiry'. I determine which methods are optimal for attaining additional epistemic aims (notably fast and steady convergence to the truth). Means-ends vindications of (a version of) Occam's Razor and the natural generalizations in a Goodmanian Riddle of Induction illustrate the power of this approach. The paper (...) establishes a hierarchy of means-ends notions of empirical success, and discusses a number of issues, results and applications of means-ends epistemology. (shrink)

Reichenbach sought to resolve Hume's problem of the justification of induction by means of a pragmatic vindication that relies heavily on the convergence properties of his rule of induction. His attempt to rule out all other asymptotic methods by an appeal to descriptive simplicity was unavailing. We found that important progress in that direction could be made by invoking normalizing conditions (consistency) and methodological simplicity (as a basis for invariance), but that they did not do the whole job. I am (...) proposing that, in the end, Reichenbach's own distinction between discovery and justification holds the key to the solution. (shrink)

: Some feminist epistemologists make the radical claim that there are varieties of epistemically valid warrant that agents access only through having lived particular types of contingent history, varieties of epistemic warrant to which, moreover, the confirmation-theoretic accounts of warrant favored by some traditional epistemologists are inapplicable. I offer Aristotelian virtue as a model for warrant of this sort, and use loosely Aristotelian vocabulary to express, and begin to evaluate, a range of feminist epistemological positions.

Some feminist epistemologists make the radical claim that there are varieties of epistemically valid warrant that agents access only through having lived particular types of contingent history, varieties of epistemic warrant to which, moreover, the confirmation-theoretic accounts of warrant favored by some traditional epistemologists are inapplicable. I offer Aristotelian virtue as a model for warrant of this sort, and use loosely Aristotelian vocabulary to express, and begin to evaluate, a range of feminist epistemological positions.

The debates between Bayesian, frequentist, and other methodologies of statistics have tended to focus on conceptual justifications, sociological arguments, or mathematical proofs of their long run properties. Both Bayesian statistics and frequentist (“classical”) statistics have strong cases on these grounds. In this article, we instead approach the debates in the “Statistics Wars” from a largely unexplored angle: simulations of different methodologies’ performance in the short to medium run. We conducted a large number of simulations using a straightforward decision problem based (...) around tossing a coin with unknown bias and then placing bets. In this simulation, we programmed four players, inspired by Bayesian statistics, frequentist statistics, Jon Williamson’s version of Objective Bayesianism, and a player who simply extrapolates from observed frequencies to general frequencies. The last player functions as a benchmark: a statistical methodology should at least outperform a crude form of induction. We focused on the performance of these methodologies in guiding the players towards good decisions. Unlike an earlier simulation study of this type, we found no systematic difference in performance between the Bayesian and frequentist players, provided the Bayesian used a flat prior and the frequentist used a low confidence level. Unlike that study, we were able to use Big Data methods to mitigate problems of random error in the simulation results. The Williamsonian player, who is a novel element of our study, also had no systematic differences in their performance, provided that they use a low confidence level. These players performed similarly even in the very short run, when players were making different decisions. Our study indicates that all three methodologies should be taken seriously by philosophers and practitioners of statistics. However, the frequentist and Williamsonian performed poorly when their confidence levels were high, and the Bayesian was surprisingly harmed by biased priors, providing some unexpected lessons for these methodologies when facing this type of decision problem. (shrink)

The extent of harm and suffering caused by the coronavirus pandemic has prompted a debate about whether the epidemic could have been contained, had the gravity of the crisis been predicted earlier. In this paper, the philosophical debate on predictive reasoning is framed by Hume’s problem of induction. Hume argued that it is rationally unjustified to move from the finite observations of past incidences to the predictions of future events. Philosophy has offered two major responses to the problem of induction: (...) the pragmatic induction of Peirce and the critical rationalism of Popper. It is argued that of these two, Popper’s critical rationalism provides a more potent tool for preparing for unanticipated events such as the Covid-19 pandemic. Popper’s notion of risky predictions equips strategic foresight with clear hypotheticals regarding potential crisis scenarios. Peirce’s pragmatic induction, instead, leans on probabilities that are slower to be amended as unexpected events start unfolding. The difference between the two approaches is demonstrated through a case study of the patterns of reasoning within the World Health Organization in the early stages of the coronavirus pandemic. (shrink)

Many arguments that show p to be enthymematic (in an argument for q) rely on claims like “if one did not believe that p, one would not have a reason for believing that q.” Such arguments are susceptible to the neg-raising fallacy. We tend to interpret claims like “X does not believe that p” as statements of disbelief (X's belief that not-p) rather than as statements of withholding the belief that p. This article argues that there is a tendency to (...) equivocate in arguments for the enthymematicity of arguments (e.g., Lewis Carroll's paradox, Hume's problem) as well as in arguments for the enthymematicity of action explanations (e.g., arguments for psychologism and for explanatory individualism). The article concludes with a warning, because the equivocation is often helpful in teaching and because neg-raising verbs include philosophically vital verbs: desire, want, intend, think, suppose, imagine, expect, feel, seem, appear. (shrink)

Malgré les critiques de toutes sortes dont elle fut et dont elle est encore l'objet, la grammaire transformationnelle est devenue en vingt ans, sous l'impulsion de Noam Chomsky, le paradigme dominant de la linguistique contemporaine. Et, bien sûr, l'incroyable rapidité de cette ascension s'est accompagnée de nombreuses crises de croissance. Linguistes, philosophes et psychologues, ainsi amenés à s'interroger sur les fondements épistémologiques de cette étrange discipline, se sont heurtés à de redoutables énigmes métathéoriques: de quel domaine factuel la linguistique est-elle (...) la science? les grammaires sont-elles empiriquement falsifiables? quel rôle l'intuition du locuteur – et singulièrement celle du linguiste – joue-t-elle dans l'élaboration et la confirmation de la théorie? quelle relation y a-t-il entre règles et lois? quel est le rapport théorique de la linguistique à la psychologie, à la sociologie, à la neurophysiologie? … Ces questions et d'autres semblables ont donné lieu ces dernières années à une étonnante profusion de livres et d'articles, où, à propos d'une discipline particulière et d'exemples précis, sont discutés, souvent en profondeur, tous les principaux problemes de l'épistémologie contemporaine. (shrink)

Keynes introduces three different principles under the single title “the principle of indifference”. The first is Bernoulli's princple of non-sufficient reason.If there is no known reason for predicating of our subject one rather than another of several alternatives, then relatively to such knowledge the assertions of each of these alternatives have an equal probability. Thus equal probabilities must be assigned to each of several arguments, if there is an absence of positive ground for assigning unequal ones.

This article highlights a use of analogies in science that so far has received relatively little systematic discussion: providing reasons for pursuing a model or theory. Using the development of the liquid drop model as a test case, I critically assess two extant pursuit worthiness accounts: that analogies justify pursuit by supporting plausibility arguments and that analogies can serve as a guide to potential theoretical unification. Neither of these fit the liquid drop model case. Instead, I develop an alternative account, (...) based on the idea that analogies facilitate the transfer of a well-understood modelling strategy to a new domain. 1Introduction2Case Study: The Development of the Liquid Drop Model3Plausibility Accounts 3.1Bartha on plausibility and analogical inference3.2Plausibility and the drop analogy4Analogies as a Guide to Unification5Generative Accounts 5.1Analogy-based modelling strategies5.2Did analogies play a merely generative role?6A New Pursuit Worthiness Account of Analogies 6.1Transferring understanding-with through analogies6.2Understanding-with and the liquid drop model7Conclusion. (shrink)

Bayesian confirmation theory, as traditionally interpreted, treats the temporal relationship between the formulation of a hypothesis and the confirmation (or recognition) of evidence entailed by that hypothesis merely as a component of the psychology of discovery and acceptance of a hypothesis. The temporal order of these events is irrelevant to the logic of rational theory choice. A few years ago Richmond Campbell and Thomas Vinci offered a reinterpretation of Bayes' Theorem in defense of the view that the temporal relationship between (...) hypothesis and evidence really does matter. More specifically, they advocated the thesis that successful predictions implied by scientific hypothesis H will increase the degree of confirmation of H only if they are novel predictions in the following sense; Evidence E is heuristically novel with respect to hypothesis H if and only if H was not deliberately designed to explain E (if E has already been corroborated) or in anticipation of E (if E is regarded as likely to be corroborated at some future date). Campbell and Vinci argue that the traditional interpretation of Bayes' Theorem misconstrues the significance of predictive novelty by ignoring heuristic novelty. In this paper I review the formal component of their preferred interpretation and demonstrate that it fails to establish that heuristic novelty has any special effect on theory confirmation. That is, even on their revisionist interpretation, regardless of whether H was deliberately designed to explain known evidence E (or in anticipation of suspected E), or whether H was designed without any awareness of the entailment relationship between H and E, Bayes' Theorem will generate the same epistemic probability for hypothesis H. Cases where new evidence was not foreseen at the time a hypothesis first emerged do not carry any more epistemic weight than they would have if such evidence had been foreseen. * Research on this paper was supported by an NEH summer grant. My thanks also to Richmond Campbell and Thomas Vinci for their reply to my initial evaluation of their position (in private correspondence), which compelled me to reconsider their arguments more carefully. (shrink)

In a formal theory of induction, inductive inferences are licensed by universal schemas. In a material theory of induction, inductive inferences are licensed by facts. With this change in the conception of the nature of induction, I argue that the celebrated “problem of induction” can no longer be set up and is thereby dissolved. Attempts to recreate the problem in the material theory of induction fail. They require relations of inductive support to conform to an unsustainable, hierarchical empiricism.

A standard way to challenge convergence-based accounts of inductive success is to claim that they are too weak to constrain inductive inferences in the short run. We respond to such a challenge by answering some questions raised by Juhl (1994). When it comes to predicting limiting relative frequencies in the framework of Reichenbach, we show that speed-optimal convergence—a long-run success condition—induces dynamic coherence in the short run.

Does the viability of the discovery program depend on showing either (1) that methods of generating new problem solutions, per se, have special probative weight (the per se thesis); or, (2) that the original conception of an idea is logically continuous with its justification (anti-divorce thesis)? Many writers have identified these as the key issues of the discovery debate. McLaughlin, Pera, and others recently have defended the discovery program by attacking the divorce thesis, while Laudan has attacked the discovery program (...) by rejecting the per se thesis. This disagreement over the central issue has led to communication breakdown. I contend that both friends and foes of discovery mistake the central issues. Recognizing a form of divorce helps rather than hurts the discovery program. However, the per se thesis is not essential to the program (nor is the related debate over novel prediction); hence, the status of the per se thesis is a side issue. With these clarifications in hand, we can proceed to the next stage of the discovery debate--the development (or revival) of a generative conception of justification which goes beyond consequentialism to forge a strong linkage of generation (or rather, generatability) with justification. (shrink)

Rolston (1988) argues that in order to act ethically in the environment, moral agents must assume that their actions are potentially harmful, and then strive to prove otherwise before implementing that action. In order to determine whether or not an action in the environment is harmful requires the tools of applied epistemology in order to act in accord with Rolston’s ethical prescription. This link between ethics and epistemology demands a closer look at the relationship between confirmation theory, particularly notions of (...) plausibility, in the philosophy of science and environmental ethics. Upon taking this look, I conclude that, at least logically, we are no better off assuming that actions are maximally risky (Rolston) than when we assume minimal risk. “Our vulnerability to error is greatest not from the things that we include in the model, but from the prophecies we leave out entirely.” Francis Bretherton “What is the main epistemic problem concerning science? I take it that it is the explication of how we compare and evaluate theories, as a basis either for theory acceptance or for practical action.” Bas van Fraassen “We face two kinds of ethical difficulties. One is where we know what ought to be done but not how to get the company to do it. The other is where we do not know what is right. We do not know the facts, or how to weigh the facts, especially statistical ones. We do not know the probabilities for development of future technologies. We do not know how to attach values to facts, or how to trade this good off against that one.” Holmes Rolston. (shrink)

Richmond Campbell and Thomas Vinci have recently presented a Bayesian theory of confirmation by novel evidence (Campbell and Vinci [1983]). I shall review their theory as briefly as possible and present three objections to it.

Despite the widespread use of key concepts of the Neyman–Pearson (N–P) statistical paradigm—type I and II errors, significance levels, power, confidence levels—they have been the subject of philosophical controversy and debate for over 60 years. Both current and long-standing problems of N–P tests stem from unclarity and confusion, even among N–P adherents, as to how a test's (pre-data) error probabilities are to be used for (post-data) inductive inference as opposed to inductive behavior. We argue that the relevance of error probabilities (...) is to ensure that only statistical hypotheses that have passed severe or probative tests are inferred from the data. The severity criterion supplies a meta-statistical principle for evaluating proposed statistical inferences, avoiding classic fallacies from tests that are overly sensitive, as well as those not sensitive enough to particular errors and discrepancies. Introduction and overview 1.1 Behavioristic and inferential rationales for Neyman–Pearson (N–P) tests 1.2 Severity rationale: induction as severe testing 1.3 Severity as a meta-statistical concept: three required restrictions on the N–P paradigm Error statistical tests from the severity perspective 2.1 N–P test T(): type I, II error probabilities and power 2.2 Specifying test T() using p-values Neyman's post-data use of power 3.1 Neyman: does failure to reject H warrant confirming H? Severe testing as a basic concept for an adequate post-data inference 4.1 The severity interpretation of acceptance (SIA) for test T() 4.2 The fallacy of acceptance (i.e., an insignificant difference): Ms Rosy 4.3 Severity and power Fallacy of rejection: statistical vs. substantive significance 5.1 Taking a rejection of H0 as evidence for a substantive claim or theory 5.2 A statistically significant difference from H0 may fail to indicate a substantively important magnitude 5.3 Principle for the severity interpretation of a rejection (SIR) 5.4 Comparing significant results with different sample sizes in T(): large n problem 5.5 General testing rules for T(), using the severe testing concept The severe testing concept and confidence intervals 6.1 Dualities between one and two-sided intervals and tests 6.2 Avoiding shortcomings of confidence intervals Beyond the N–P paradigm: pure significance, and misspecification tests Concluding comments: have we shown severity to be a basic concept in a N–P philosophy of induction? (shrink)

I conceive of inductive logic as a project of explication. The explicandum is one of the meanings of the word `probability' in ordinary language; I call it inductive probability and argue that it is logical, in a certain sense. The explicatum is a conditional probability function that is specified by stipulative definition. This conception of inductive logic is close to Carnap's, but common objections to Carnapian inductive logic (the probabilities don't exist, are arbitrary, etc.) do not apply to this conception.

The paper offers a new understanding of induction in the empirical sciences, one which assimilates it to induction in geometry rather than to statistical inference. To make the point a system of notions, essential to logically sound induction, is defined. Notable among them are arbitrary object and particular property. A second aim of the paper is to bring to light a largely neglected set of assumptions shared by both induction and deduction in the empirical sciences. This is made possible by (...) appealing to the logic of common nouns and applying it to the logic of natural-kind terms.This strategy yields a new insight into the concept of natural kinds. While the strategy reveals deep affinity between empirical induction and deduction it also reveals two problems peculiar to induction. This helps to explain the intuition that induction is the more problematic of the two. The paper does not set out ‘to solve the problem of induction’. 1The paper has benefited from the support and critical comments of several friends: Steven Davis, Kevin Dunbar, Anil Gupta, Michael Hallett, Ray Jackendoff and Storrs McCall. Two friends, however, deserve special thanks, Michael Makkai and Gonzalo Reyes. Many of the points made in the paper, otherwise unacknowledged, are due to their generosity and patience. In fact, the paper can be seen as an application of the logic of kinds on which Gonzalo Reyes has been working for several years. (shrink)