1 Views of this sort are defended by, for example, Mackie, J.L. The Cement of the Universe (Oxford: Oxford University Press 1974)Google Scholar; Davidson, Donald ‘Causal Relations,’ Journal of Philosophy 64 (1967), 691–703CrossRefGoogle Scholar; and Tom Beauchamp and Rosenberg, Alexander Hume and the Problem of Causation (Oxford: Oxford University Press 1981).Google Scholar

2 The intuition that sometimes what is explained by a singular causal explanation is the occurrence of an individual event, where this is something different from explaining why that event has some property or is an event of a certain kind, is nicely captured in Nickles, Thomas ’essay ‘Davidson on Explanation,’ Philosophical Studies 31 (1977), 141–5.CrossRefGoogle Scholar Nickles writes ‘It certainly is intuitively appealing to think that we can explan event occurrences period and that if we have explained the occurrence of e and know that e = f, then we can also explain why f occurred.’ Nickles goes on to express doubts that this idea of explaining the occurrence of an individual event can be made clear, despite its intuitive attractiveness. In contrast to Nickles, I think that this idea can be made clear once we reject the claim that singular causal explanation is a variety of implicit covering law explanation.

3 For the purpose of this paper, I leave open the possibility that other varieties of singular causal explanation – e.g. those that seem to explain why particular events posses certain properties such as (Ex. 6) ‘The presence of potassium salts caused the fire to be purple and odd-shaped’ – are susceptible of a covering-law analysis. In fact, however, I do not think that (Ex. 6) is plausibly viewed as a covering-law explanation, either.

4 See Woodward, James ‘Scientific Explanation,’ The British Journal for the Philosophy of Science 30 (1979), 41–67CrossRefGoogle Scholar and ‘A Theory of Singular Causal Explanation,’ Erkenntnis 21 (1984), 231-62.

5 Hempel, Carl ‘Aspects of Scientific Explanation’ in Aspects of Scientific Explanation (New York: The Free Press 1965) 365.Google Scholar

6 Ibid., 348-9

7 Ibid., 423

8 The point that a covering-law explanation of why a particular object or event possesses a property is not an explanation of why that particular exists or occurs is made very clearly by Thomas Nickles in his ‘Davidson on Explanation.’

9 This is implicitly conceded by Donald Davidson in his analysis of singular causal explanation (see below).

10 I shall speak interchangeably of an event existing or an event occurring.

11 This suggestion is made by Donald Davidson, for example, in his ‘Causal Relations.’

12 cf. Hempel's discussion in ‘Aspects of Scientific Explanation’ 421-3.

13 All of this is perhaps even more obvious when we consider that the laws which, on Hempel's account, will figure in the explanation of a particular solar eclipse must specify the conditions under which one and only one eclipse will occur. Intuitively, it seems fairly obvious that it is not the function of laws of nature to specify directly the conditions under which only one eclipse will occur in a given spatio-temporal interval, any more than distinct laws of nature are required to explain why, say, one, two, or three solar eclipses will occur within a given spatio-temporal interval. One explains why two solar eclipses will occur within a spatio-temporal interval by applying the laws of nature which might be used to explain why one solar eclipse occurred within a spatio-temporal interval (i.e., the laws described above) to two distinct (although lawfully related) sets of initial conditions. An explanation of why three solar eclipses occurred within a given spatio-temporal interval proceeds by applying the relevant laws of nature to three distinct sets of initial conditions. In just the same way, one explains why one and only one solar eclipse occurred by explaining, using these laws, why an eclipse occurred and then showing that no other set of initial conditions was such that, given these laws, an additional eclipse would occur within the spatiotemporal interval in question. Unique eclipses are explained by applying the same laws which can be used to explain a non-unique eclipse to a distinctive set of initial conditions. There is no distinctive law which enumerates the conditions under which a unique eclipse will occur.

14 A promising place to begin such an account is with the idea that in a properly formulated scientific theory existential claims should not be introduced (even conditionally) by laws but rather by claims about initial conditions. This idea, in turn, gets its motivation from the idea that we can best specify what a law is by contrasting the different roles played by laws and initial conditions in scientific theorizing and the different requirements we employ in assessing claims about laws and claims about initial conditions. (For example, we expect that laws will obey certain symmetry and invariance requirements which claims about initial conditions need not obey.)

15 Davidson, ‘Causal Relations’ 699Google Scholar

16 Ibid., 700

17 Ibid., 700

18 I write ‘something like’ because (5) and (N) both quantify over events – the idea is that the events referred to in the original singular causal claim literally instantiate the underlying laws. If my argument on pages 261-5 is right, this analysis cannot be literally correct, for laws of nature do not involve quantification over events.

19 Nor, of course, does it follow that one must know the appropriate underlying laws of form (5) and (N) if one is to be warranted in asserting a singular causal claim. (The general point that the ordinarily assertability conditions for a sentence may be distinct from its truth conditions ought to be familiar enough from recent philosophy of language – consider, for example, natural kind terms.) A fuller discussion of the condition for warranted assertability for singular causal claims and an argument that these will be quite distinct from the truth conditions for such claims is presented in my ‘Causal Explanations in History,’ in process.

20 If it is a mistake to regard claims (Ex. 1) as implicit deductive-nomological explanations, in virtue of what features do such claims qualify as explanations? While a full treatment of this question must be beyond the scope of this essay, my general suggestion is that (Ex. 1) qualifies as an explanation in virtue of specifying conditions under which an alternative to the explanandum-phenomenon would have occurred – (Ex. 1) tells roughly (and putting aside complications having to do with over-determination), that if the short circuit had not occurred, the fire wouldn't have occurred. I have argued elsewhere that the central unifying strand in all why-explanations is not that such explanations show that the explanandum-phenomenon ‘was to be expected’ (as DN theorists like Hempel have maintained) but rather that all such explanations will provide something like the counterfactural information described above. That is, serious scientific explanation, as in quantum mechanics or general relativity, will always proceed by specifying the conditions under which one or more alternative to the phenomena being explained will occur – just as (Ex. 1) does. It is this common element that warrants one in taking claims like (Ex. 1) as explanatory. Provision of this counterfactual information is facilitated by, but does not require, information about general laws. This is why we can legitimately regard (Ex. 1) and (Ex. 2) as explanatory even though they do not (even ‘tacitly’ or implicitly’ provide information about general nomological corrections). For more detailed discussion see my ‘Scientific Explanation,’ The British Journal for the Philosophy of Science 30 (1979), 41-67, and my ‘A Theory of Singular Causal Explanation,’ forthcoming in Erkenntnis 21 (1984), 231-62.

21 Cf. ‘Aspects of Scientific Explanation,’ 348-9. Exactly how much information the user of singular causal explanation must know (if he is to be warranted in asserting the singular causal explanation) is never made very clear in Hempel's discussion. But I think it fair to say that Hempel assumes that the user must know enough to enable him to get started in constructing a rough approximation to a covering-law explanation.

22 I claim little orginality for the following argument, which relies very heavily on Thomas Nickles’ papers ‘Davidson on Explanation’ and ‘On the Independence of Singular Causal Explanation in Social Science: Archaeology,’ Philosophy of Social Sciences 7 (1977) 163-87.

23 Davidson, Donald ‘Mental Events’ in Experience and Theory, Foster, L. and Swanson, J.W. eds. (Amherst: The University of Massachusetts Press 1970)Google Scholar; reprinted in Davidson, Donald Actions and Events, (Oxford: Oxford University Press 1981) 207–25Google Scholar; page references are to reprinted version of essay.

24 Davidson, ‘Causal Relations,’ 700; this claim appears to be endorsed as well in ‘Mental Events,’ 223: Davidson writes that ‘ … mental events as a class cannot be explained by physical science; particular mental events can when we know particular identities.’ My own view has (at least) affinities with this. I would hold that on Davidson's assumptions about the non-existence of psychophysical laws, it follows that we cannot give a covering-law explanation of why any event possesses mental properties, but that we can, by using particular event identities, give a singular causal explanation of why particular mental events occur. What is puzzling, as I note below, is how Davidson can combine his claim about the role of individual event identities in singular explanation with the claim that causal explanation is always implicitly deductive in structure.

25 Causey, Robert The Unity of Science (Dordrecht: Reidel Publishing Co. 1977)CrossRefGoogle Scholar

26 It might seem that there are obvious counterexamples to this claim. If one has a singular causal explanation for the occurrence of the fire, and the fire was a purple, odd-shaped fire, does one also have materials for constructing a singular causal explanation of the occurrence of the purple, odd-shaped fire? I would argue ‘yes’ and that the temptation to suppose otherwise derives from a failure to distinguish providing a singular causal explanation of why the purple, odd shaped fire occurred from providing a singular causal explanation of why the fire was purple and odd-shaped (d. my remarks on the importance of distinguishing what is explained from what is presupposed in singular causal explanation on 258-61 above). This whole issue is treated in more detail in my ‘A Theory of Singular Causal Explanation.’

27 See Woodward, ‘A Theory of Singular Causal Explanation.’ The fact that Davidson advocates (i) within a general framework which does not fit with (i) perhaps also indicates what a natural assumption (i) is. Compare also the remarks by Nickles which I quoted in footnote 2 above. Those who, like Nickles, reject (i) generally do so for theoretical reasons – because they assume that all causal explanation basically has a covering-law structure and because they think, correctly, that it is impossible to reconcile (i) with such an assumption.

28 In the passages from Davidson under discussion he speaks of explaining the occurrence of an individual event (or the fact that an individual event has occurred) simpliciter. Elsewhere (particularly in ‘Mental Events’) Davidson speaks instead of explaining events ‘under a description’ where the suggestion is that a given explanans may serve to explain an event under one description, but not under some different co-referential description. It might be thought that the distinction between explaining the occurrence of an event and explaining that occurrence under a description can be used to provide a way out of the difficulty described above. While I cannot argue the point in detail here, I think that this is a mistake and that talk of explaining an event under a description is simply a confused way of expressing the claim that what is explained is why an event satisfies a certain description or (what comes to the same thing) possesses a certain property.