The Facticity of Explanation and Its Consequences

Abstract

This paper argues that, contrary to the views of Nancy Cartwright and Brian Ellis, explanations are factive: if a statement is taken to be an explanation, it also has to be accepted as true. Taking explanations to be true, in turn, seems to imply that all the entities posited in explanations are real. But this is precisely what some philosophers, such as Cartwright and Ellis, want to deny. What these philosophers do not want to deny, however, is that such statements do explain. As a result, they see themselves forced to reject the facticity of explanation, a strategy that is unacceptable in my view. In order to avoid the further conclusion that all of the entities explanations posit exist, I propose to separate truth from ontology: explanations can be true, but the truth of these explanations alone does not commit us to the entities that these explanations posit. To determine where the ontological commitments of our statements lie, we must instead start with a criterion (a necessary and sufficient condition) for what exists. For instance, if all and only entities with causal powers are real, then this tells us that the only entities our true statements are committed to are entities with causal powers.

Acknowledgements

I would like to thank Jody Azzouni, Arnold Koslow, and two anonymous referees for helpful comments.

Notes

[1] Although Cartwright and Ellis deny the existence of forces and mathematical entities, respectively, they are not generally anti‐realists. Rather, both are entity realists about some of the posits of our scientific theories: they believe, for example, that theoretical entities (e.g., electrons) are real. (For a more detailed discussion of entity realism and how it is distinguished from theory realism, see, for instance, Hacking 1983 1983, 27.) In this paper, I am only concerned with the attitudes they take about forces and mathematical entities.

[2] Cartwright also discusses the point that fundamental laws are true only ceteris paribus in her The Dappled World (see, for instance, Cartwright 1999, 4, as well as 28, 29). While one of Cartwright’s objectives in The Dappled World is to develop further the views initially presented in her 1983 (Cartwright 1999, 10), she does not further elucidate her views on the nature of explanation. Therefore, in what follows, I will stick to her (1983) work.

[3] And here, she (explicitly) opposes the views of Duhem and van Fraassen. As she summarizes it, Duhem and van Fraassen believe that even causal explanations do not have to be true (Cartwright 1983, 4). Van Fraassen’s discussion of this can be found in his (1980), 97–101.

[4] This issue is further complicated by the fact that Ellis, in part III of his book, proposes what he calls a ‘pragmatic’ theory of truth. In his view, truth ‘is what it is right to believe’ (Ellis 1990, 2). If one held such an internal realist conception of truth, then perhaps the truth of a statement would not even commit one to the existence of the objects it posits. Thus, perhaps the truth of statements about triangles would be no more and no less ontologically committing than the truth of statements about apples. But if Ellis thinks truth can thus be separated from ontology, then why would he, when he presents his ontology in part I of his book, labour so very hard to draw a distinction between explanations that are literally true (and therefore ontologically committing) and explanations that aren’t literally true (and therefore aren’t ontologically committing)? After all, Ellis could simply allow for all explanations to be true in this internalist sense, which would make much of the work in part I unnecessary.

[5] Hempel draws a distinction between potential and actual explanations, which has led some philosophers (e.g., Ruben 1993, 139) to believe that Hempel also endorsed the facticity of explanation, but the evidence for this is ultimately inconclusive. In his 1948, Hempel says that for an explanation to be sound, ‘the sentences constituting the explanans must be true’ (Hempel 1965a [1948], 248). But this implies that there can be unsound explanations (just as there can be unsound, but valid, arguments). In his 1965 work (Hempel 1965b [1965]), he makes the distinction between a true explanation, a strongly supported or confirmed explanation, and a potential explanation. In a true explanation, the ‘conjunction of its constituent sentences is true’, in a strongly supported or confirmed explanation, the explanans ‘is more or less strongly confirmed by the given evidence’, and a potential explanation is one which has ‘the character of an … explanation except that the sentences constituting its explanans need not be true’ (Hempel 1965b [1965], 338). Azzouni would say that a potential explanation wouldn’t be an explanation at all because it isn’t factive, but an explanation form (just as an argument with an unsound premise would be an argument form).

[6] Incidentally, van Fraassen’s view that explanations are not factive is also based on an analysis of how ‘explanation’ is ordinarily used: ‘There are many examples, taken from actual usage, which show that truth is not presupposed by the assertion that a theory explains something’ (van Fraassen, 1980, 98). Azzouni, who confronts precisely this discussion in his 2000 when arguing that ‘explanation’ is factive, replies that van Fraassen’s examples do not provide conclusive evidence for his view (Azzouni 2000, 61). Azzouni then offers the passage I quote in the text as support. My leanings, obviously, are with Azzouni.

[7] What about a sentence like ‘Ptolemy’s explanation of the retrograde motion of the planets is that they move on epicycles’? Wouldn’t that be a good example of an unsatisfactory explanation? No. Planets do not move in epicycles, and thus Ptolemy did not explain their motion with this statement. Ptolemy only thought he did. He took what he offered to be an explanation, when in fact, it was not. It would, indeed, be equally natural to say that ‘Ptolemy thought that the explanation of the retrograde motion of the planets is that they move on epicycles’, instead of ‘Ptolemy’s explanation of the retrograde motion of the planets is that they move on epicycles’. This reveals the intentional nature of that sentence. Perhaps another way to put this is to say that Ptolemy offered what a possible candidate for an explanation. But because his claims about the motion of the planets were false, the candidate does not, in the end, qualify. It is not an unsatisfactory explanation, but no explanation at all.

[8] I’m indebted to Daniel Leafe for this example.

[9] It should also be pointed out that all these explanations are causal. And this is important: it means that the problem that Ellis and Cartwright have generalizes to causal explanations as well; causal explanations, too, posit objects some philosophers do not want to be committed to. We can also be sure that Ellis, whose ontology is basically causal, would not want to be committed to unwritten treatises. So for Ellis to say that only model theoretic explanations are not ontologically committing to everything they refer to is not sufficient in any case. He also has to show that some causal explanations are not ontologically committing.

[10] Of course, this brief treatment of ontological commitment does not address all the issues about the relationship between the truth of a sentence and what that statement is ontologically committed to. A much more detailed analysis is called for. This analysis has been provided elsewhere: the reader is referred to Raley (2005; 2007), as well to Azzouni (1998; 2004).

Additional information

Notes on contributors

Yvonne Raley

Yvonne Raley is at the Department of Philosophy, Felician College.