Thursday, May 20, 2010

Successful Theistic Explanations (Part 4) - Informativeness

This post is part of my series on Chapter 7 of Gregory Dawes's book Theism and Explanation. In this chapter, Dawes considers the merits of theistic explanations in light of six explanatory virtues.

Part one introduced the topic and looked at the explanatory virtue of testability. Part two focused on the explanatory virtues of consistency with background knowledge and past explanatory success. And part three dealt with the virtues of simplicity and ontological economy.

In this final entry in the series, we will consider theism in light of the virtue of informativeness, and we will also summarise the conclusions reached in this chapter.

1. What is Informativeness?
In defining informativeness, Dawes relies on Peter Lipton's definition of a "lovely" explanation. This definition has two aspects to it. First, the explanation must specify some causal mechanism. Second, it must allow us to deduce precise effects that would arise from this mechanism.

Earlier in the book, Dawes argued that theistic explanations are unlikely to be able to specify a causal mechanism, but that this is not necessarily a fatal objection to them. Still, it is worth seeing how well theism does in terms of the second aspect of informativeness, i.e. our ability to deduce precise effects from it.

It will no doubt shock you to learn that theism is not very informative in this regard. Let's see why that is.

2. Quantifiable Predictions and Intentional Explanations
The first, somewhat spurious, objection to theism is that we cannot deduce precise quantifiable predictions from it. This is something that we can do in the case of the most successful scientific explanation: we can specify a causal mechanism and specify what the measurable effects of this mechanism would be (see my posts on neuroscientific explanations for more).

But the fact that theism fails in this regard is not a serious charge. Theistic explanations are a species of intentional explanation. As such they explain events and states of affairs in terms of the goals and intentions of an agent. We employ such explanations all the time, e.g. when trying to make sense of the behaviour of our family and friends.

None of these everyday intentional explanations yield precise quantifiable predictions, so we shouldn't expect theism to do so either.

3. The Problem of Mysteriousness
There is, however, a problem with theistic explanations that is not shared by typical intentional explanations. In dealing with human behaviour and social interaction, we know what to expect.

For instance, we know the type of conduct to expect from someone if they are feeling hungry, or if they have just learned that their husband/wife are cheating on them, or if they have a desire to become a doctor and so on. In this sense, everyday intentional explanations are informative.

Theistic explanations are not. We are not able to generate similar expectations when applying mental or action predicates to God because the nature of his agency is "wholly other". The mental and actions predicates that are applied to human beings are tailored to their finite, temporal and physical properties. What would it mean to apply the same predicates to an infinite, eternal and non-physical being.

Theists have ways out of this problem. They can argue that there is a core meaning to the predicates that does depend on the properties of human beings. For example, they could argue that the action predicate "creates" has a core meaning of "uses imagination to bring into being". This core meaning does not rely on temporal, finite or physical properties.

This may be true. But in extracting the core meaning, we take away all the things that make these predicates informative (in the sense defined above). We do not really know what to expect from a creation that emanates from an infinite, eternal and non-physical being.

4. The Problem of Accommodation
This leads to a final problem for theistic explanations. Because they have a tendency to be uninformative, they also have a tendency for accommodationism. That is: they tend to make the evidence fit with the hypothesis, rather than predicting what the evidence should be given the hypothesis.

The example that Dawes uses to illustrate this point is quite interesting. He imagines a believer who has a powerful experience of confidence and joy when reading passages of scripture. The believer is told that he has experienced what is known as the "internal testimony of the Holy Spirit".

This explains his experience and also serves to corroborate religious claims concerning the authority of the scripture.

But there is a huge problem with this explanation of the experience. Since theism is uninformative, we have no idea what it would mean to receive testimony from an incorporeal being. And it is because theism is uninformative that it is all too easy for the believer to offer spurious explanations of this sort.

Dawes concludes by noting that uninformativeness is a serious problem for theists. Indeed, it is so serious that it means non-theistic, natural explanations that yield precise predictions will almost always be preferable.

5. Summary
In this series we have investigated the explanatory merits of theism in light of six explanatory virtues. The results of this investigation are summarised in the image below (click to embiggen).


  1. I don't understand how informativeness is different from testability. It seems to be just rephrasing the latter.

  2. Yeah they are pretty much equivalent. I think the distinction Dawes tries to make is that testability is about passing tests, whereas informativeness is about being specific enough to make testable predictions.

    It is like the difference between the logical strength of an explanation and its truthfulness. Informativeness is about logical strength and testability is about truthfulness.