This paper examines a recent attempt by Evan Jobe to account for the asymmetric character of many scientific explanations. It is argued that a purported counterexample to Jobe's account, from Clark Glymour, is inconclusive, but that the account faces independent objections. It is also suggested, contrary to Jobe, that the explanatory relation is not always asymmetric. Sometimes a singular sentence C can figure in a DN derivation of another singular sentence E and E can also figure in a DN derivation (...) of C. Yet while we are inclined to regard the first derivation as an explanation of E, we are not inclined to regard the second derivation as an explanation of C. As Sylvain Bromberger pointed out in a now classic article (1966), one can explain the period of a pendulum by reference to its length and yet, although one can give a DN derivation of the length of a pendulum by reference to its period, this derivation does not seem to represent an explanation. Evan Jobe has recently offered an interesting account of such explanatory asymmetries and Clark Glymour has in turn proposed a counterexample which seems to show that Jobe's account is defective. The aim of this paper is two-fold. I shall attempt to show that (a) Glymour's proposed counterexample can be rejected on the grounds that it violates an independently plausible restriction on the role that equalities may play in DN explanation, and that (b) although Glymour's counterexample can be avoided in this way, Jobe's account is defective in several other respects. (shrink)
This paper explores, in a rather schematic way, some issues having to do with the conception of causation and explanation implicit in regression analysis. I argue that (a) regression analysis does not yield lawlike generalizations but rather claims about causal connections in particular populations and that (b) regression analyses are not plausibly viewed as part of a neo-Humean program of analyzing causal claims in terms of claims about patterns of statistical association. I also argue that (c) the conception of explanation (...) implicit in regression analysis is deductive and involves the exhibition of a pattern of counterfactual dependence between mean values of the independent and dependent variables. (shrink)
Causal explanation proceeds by citing the causes of the explanandum. Any model of causal explanation requires a specification of the relation between cause and effect in virtue of which citing the cause explains the effect. In particular, it requires a specification of what it is for the explanandum to be causally dependent on the explanans and what types of things (broadly understood) the explanans are. There have been a number of such models. For the benefit of the unfamiliar reader, here (...) is a brief statement of some major views. On David Lewis’s account, c causally explains e if c is connected to e with a network of causal chains. For him, causal explanation consists in presenting portions of explanatory information captured by the causal network. On Wesley Salmon’s reading, c causally explains e if c is connected with e by a suitable continuous causal (i.e., capable of transmitting a mark) process. On the standard deductive-nomological reading of causal explanation, for c to causally explain e, c must be a nomologically sufficient condition for e. And for John Mackie, for c to causally explain e there must be event-types C and E such that C is an inus-condition for E.53 In a series of papers and a book, James Woodward (1997, 2000, 2002, 2003a, 2003b) has put forward a ‘manipulationist’ account of causal explanation. Briefly put, c causally explains e if e causally depends on c, where the notion of causal dependence is understood in terms of relevant (interventionist) counterfactual, that is counterfactuals that describe the outcomes of interventions. A bit more accurately, c causally explains e if, were c to be (actually or counterfactually) manipulated, e would change too. This model ties causal explanation to actual and counterfactual experiments that show how manipulation of factors mentioned in the explanans would alter the explanandum. It also stresses the role of invariant relationships, as opposed to strict laws, in causal explanation. Explanation in this model consists in answering a network of “what-if-things-had-been-different questions”, thereby placing the explanandum within a pattern of counterfactual dependencies (cf. Woodward 2003a, p.. (shrink)
The paper tries to provide an alternative to Hempel’s approach to scientific laws and scientific explanation as given in his D-N model. It starts with a brief exposition of the main characteristics of Hempel’s approach to deductive explanations based on universal scientific laws and analyzes the problems and paradoxes inherent in this approach. By way of solution, it analyzes the scientific laws and explanations in classical mechanics and then reconstructs the corresponding models of explanation, as well as the types of (...) scientific laws appearing in it. Finally, it compares this reconstruction with the approaches of J. Woodward and C. Hitchcock, C. Liu and with the views of M. Thalos on analytic mechanics. (shrink)
The aim of this series is to bring together important recent writings in major areas of philosophical inquiry, selected from a variety of sources, mostly periodicals, which may not be conveniently available to the university student or the general reader. The editor of each volume contributes an introductory essay on the items chosen and on the questions with which they deal. A selective bibliography is appended as a guide to further reading. This volume presents a selection of the most important (...) recent writings on the nature of explanation. It covers a broad range of topics from the philosophy of science to the central philosophical terrain of the theory of knowledge. The distinguished contributors include Peter Achinstein, Wesley C. Salmon, Carl G. Hempel, Philip Kitcher, Bas C. van Fraassen, Jaegwon Kim, B. Brody, Timothy McCarthy, Peter Railton, David Lewis, Peter Lipton, James Woodward, and Robert J. Matthews. (shrink)
According to James Woodward’s influential interventionist account of causation, X is a cause of Y iff, roughly, there is a possible intervention on X that changes Y. Woodward requires that interventions be merely logically possible. I will argue for two claims against this modal character of interventions: First, merely logically possible interventions are dispensable for the semantic project of providing an account of the meaning of causal statements. If interventions are indeed dispensable, the interventionist theory collapses into (some (...) sort of) a counterfactual theory of causation. Thus, the interventionist theory is not tenable as a theory of causation in its own right. Second, if one maintains that merely logically possible interventions are indispensable, then interventions with this modal character lead to the fatal result that interventionist counterfactuals are evaluated inadequately. Consequently, interventionists offer an inadequate theory of causation. I suggest that if we are concerned with explicating causal concepts and stating the truth-conditions of causal claims we best get rid of Woodwardian interventions. (shrink)
I clarify the difference between pluralist and monist interpretations of levels of selection disputes. Lloyd has challenged my claim that a plurality of models correctly accounts for situations such as maintenance of the sickle-cell trait, and I revisit this example to show that competing theories don't disagree about the existence of `high-level' or `low-level' causes; rather, they parse these causes differently. Applying Woodward's theory of causation, I analyze Sober's distinction between `selection of' versus `selection for'. My analysis shows that (...) this distinction separates true causes from pseudocauses, but it also reveals that the distinction is irrelevant to the levels debate; it makes no sense to say true causes are at higher levels and not lower levels. The levels debate is not about separating real causes from pseudocauses; it's about finding useful ways to parse and disentangle causes. (shrink)
