15.05.2013 Views

Theism and Explanation - Appeared-to-Blogly

Theism and Explanation - Appeared-to-Blogly

Theism and Explanation - Appeared-to-Blogly

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

106 <strong>Theism</strong> <strong>and</strong> <strong>Explanation</strong><br />

In terms of confi rmation theory, the gremlin hypothesis has a low posterior<br />

probability because it has a low prior probability. What is it about the<br />

gremlin hypothesis that gives it a low prior probability? Well, one reason<br />

is that it enjoys little support from our background knowledge: it bears<br />

little relationship <strong>to</strong> those other things which we know <strong>to</strong> be true. So, as<br />

Swinburne is well aware, a key issue is the prior probability of the theistic<br />

hypothesis, <strong>and</strong> a key issue in determining this is the role of background<br />

knowledge. I shall come back <strong>to</strong> these questions shortly (7.2).<br />

We can approach the same question more formally, by way of the Bayesian<br />

reasoning that forms so important a part of Swinburne’s argumentation. As<br />

we have seen (2.1.3.1), the relationship I have been discussing—that between<br />

likelihood <strong>and</strong> posterior probability—is formalised by Bayes’s theorem:<br />

Pr(H|E) =<br />

Pr(E|H) × Pr(H)<br />

Pr(E).<br />

If, with Swinburne, we assume that a hypothesis is confi rmed <strong>to</strong> the degree<br />

that it is rendered probable—<strong>and</strong> I have already hinted at an alternative<br />

account (2.1.3.2)—then Bayes’s theorem also gives us the degree <strong>to</strong> which<br />

a hypothesis is confi rmed. Let’s see how this works by assigning some fi gures,<br />

more or less at r<strong>and</strong>om. Let’s say that the prior probability of E is 0.2.<br />

(It is a surprising fact that we’re trying <strong>to</strong> explain.) And let’s say the likelihood<br />

of E, given H, is 0.9. (H has a high degree of explana<strong>to</strong>ry force.) Can<br />

we say that E confi rms H? Yes, in the weak sense of “confi rm” we can. It<br />

renders it more likely than it would be otherwise. But by how much? Well,<br />

it depends on the prior probability of H. If the prior probability of H is low,<br />

let’s say 0.1, then its probability is raised <strong>to</strong> 0.45. This is impressive, but it<br />

falls short of demonstrating that H is probably true (if that is what you are<br />

interested in doing). 23<br />

6.1.2.2 The Probability of <strong>Theism</strong><br />

So the second step in Swinburne’s reasoning is <strong>to</strong> turn these various<br />

C-inductive arguments, as he calls them, in<strong>to</strong> a P-inductive argument. 24<br />

A P-inductive argument is one that makes the conclusion more probable<br />

than not. Swinburne argues that the prior (or “intrinsic”) probability of<br />

theism is high, at least “relative <strong>to</strong> other hypotheses about what there<br />

is.” 25 For prior probability, he claims, will be directly proportional <strong>to</strong> the<br />

simplicity of a hypothesis, its narrowness of scope, <strong>and</strong> its conformity <strong>to</strong><br />

background knowledge. 26 Swinburne argues that background knowledge<br />

need not be taken in<strong>to</strong> account when assessing the theistic hypothesis, since<br />

theism “purports <strong>to</strong> explain everything logically contingent (apart from<br />

itself).” 27 But this means that the theistic hypothesis leaves nothing in the

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!