In this timely study, Dawes defends the methodological naturalism of the sciences. Though religions offer what appear to be explanations of various facts about the world, the scientist, as scientist, will not take such proposed explanations seriously. Even if no natural explanation were available, she will assume that one exists. Is this merely a sign of atheistic prejudice, as some critics suggest? Or are there good reasons to exclude from science explanations that invoke a supernatural agent? On the one hand, Dawes concedes the bare possibility that talk of divine action could constitute a potential explanation of some state of affairs, while noting that the conditions under which this would be true are unlikely ever to be fulfilled. On the other hand, he argues that a proposed explanation of this kind would rate poorly, when measured against our usual standards of explanatory virtue.

Table of Contents

• • Acknowledgements     ix
  • 1     Against Religious Explanations     1
  • 2     On Explanations in General     19
  • 3     What are Theistic Explanations?     33
  • 4     What Would They Explain?     59
  • 5     Potential Theistic Explanations     77
  • 6     Inference to the Best Explanation      101
  • 7     Successful Theistic Explanations     115
  • 8     Conclusion     143
  • Appendix: Intentional Explanations     147
  • Notes     167
  • Bibliography     191
  • Index     203

Summary of the Argument

Excerpt from pages 143-146.

I examined a number of the objections that might be raised against proposed theistic explanations. … The first is that proposed theistic explanations exclude no possible state of affairs; the second is that the actions of an agent capable of miracles would be unpredictable; the third suggests that the very concept of God is incoherent; the fourth maintains that the will of God cannot be a cause. I have argued that while these objections raise some serious issues with which a theist philosopher ought to grapple, they do not, in themselves, rule out the possibility of a successful theistic explanation.

In Chapter 5, I set out the circumstances in which invoking a divine agent would constitute a potential explanation of some state of affairs. It would do so, I argued, only if we could conceive of no better way in which a divine agent could have brought about his posited intention. This optimality condition constitutes a powerful constraint upon any proposed theistic explanation. Given the existence of apparently pointless evils, the theist will have a difficult task showing that his proposed explanation meets this condition. But let’s assume that the theist could do so. Let’s say that we could be warranted in regarding an account of divine agency as a potential explanation of some state of affairs.

What would follow? Well, not very much. The theist would still need to show that his proposed explanation was a successful one, that we had sufficient reason to accept it. Chapter 7 set out the conditions that a potential theistic explanation would have to meet in order to be regarded as the actual explanation of some state of affairs. It has shown that measured against a list of accepted explanatory virtues, a theistic hypothesis is simply incapable of achieving a high score. It is not (as things stand) consistent with the rest of our knowledge, it comes from a tradition whose proposed explanations have previously scored poorly, it is ontologically extravagant, and it does not enable us to predict the precise details of the effect. In other words, it lacks many of the qualities we would normally demand of successful explanations.

…

My analysis does support the view that we ought to have a preference for natural explanations, since these have a better chance of exhibiting the features that characterise an adequate explanation. But this is a very modest position. After all, none of the features I have discussed — testability, consistency with background knowledge, simplicity, ontological economy, and informativeness — rule out proposed theistic explanation a priori. A proposed theistic explanation formulated with the requisite degree of intentional specificity would be testable. Whether it survies the test is another question, but it is not a question which we can decide in advance. Consistency with background knowledge is a contingent matter: if we already had a tradition of successful theistic explanations, then there would be no reason to reject yet another. Ontological economy is a ceteris paribus condition. It suggests that we should not posit new kind of entities unless these are required and it is at least conceivable that positing a divine agent might be required, to explain some phenomenon. And while a proposed theistic explanation may not be a “lovely explanation,” which would allow us to deduce the precise details of the effect, this does no seem (by itself) a fatal objection, if it were to score highly on our other criteria.

So yes, my arguments might give us reason to prefer natural explanations when these are available, and to seek natural explanations when they are not. It follows that a proposed theistic explanation should be, at best, an explanation of last resort. One might argue that this view — that we should abandon the search for natural explanations only in extremis — represents a kind of “presumption of naturalism.” And so it does.