Saturday, 19 September 2009

Swinburne: Cosmological arguments

In Chapter 7, Swinburne finally gets round to looking at the evidence for or against God. He starts with the Cosmological argument, in essence looking for answers to the question “Why is there anything rather than nothing?”

It is of course a very good question, and one we don’t know the answer to. We know that the universe does exist if only because we are part of it and observing it. At this point, Swinburne is concerning himself solely with the existence of the universe, not with any of its properties.

Swinburne is very keen on quoting 18th-century philosophers. He frequently quotes Hume (whom he doesn’t like) and Liebnitz (whom he does like), but broadly speaking he doesn’t draw on anybody more recent except in as far as they comment on Hume and Liebnitz, or on Aquinas. There is one exception to this. He likes to quote himself, extensively.

But this gives his arguments are rather mechanistic feel to them, as if his understanding of atoms is that they are Newtonian billiard balls. He is aware of the Big Bang theory but if he knows anything more about it than the name of the theory and the approximate age of the universe (which he puts at 15 billion years) then he doesn’t let on.

Swinburne discusses the regression to the first cause. He spends about 10 pages on this, but after all the quotations and arguments, it boils down a fairly simple point. Either the universe is infinitely old or it isn’t. If it is infinitely old, then it is ultimately uncaused. If it isn’t then it had a first cause, which either comes from God, or is ultimately uncaused. So we ultimately have a choice of explanation, either the universe exists uncaused, or the universe was created by God, who exists uncaused.

And not surprisingly, he prefers the latter explanation. He conveniently summarises his reasons in the last paragraph of the chapter.

There is quite a chance that, if there is a God, he will make something of the finitude and complexity of the universe. It is very unlikely that a universe would exist uncaused, but rather more likely that God would exist uncaused. Hence the argument from the existence of the universe to the existence of God is a good C-inductive argument.

His assertion that the universe existing uncaused is less likely than God existing uncaused is based on his belief that God is simpler than the universe, and that simpler things are more probable than complex ones. At the end of Chapter 5, Swinburne said that the intrinsic probability of theism was low, but because he regarded theism as a simple concept, it was more probable than any one of many other individual hypotheses for the existence of things. He reached that conclusion without examining or even naming any alternative hypothesis, so if we are to agree with him, we have to take his assertion on trust, without evidence.

But now, in Chapter 7, this has suddenly transformed itself into a 2-way choice between the universe existing uncaused and the universe being caused by God, with the latter being judged more probable. It is almost as it he expects us not to have read (or at least not to have remembered) what he wrote in the previous chapters.

Note that he characterises this as a C-inductive argument. Remember that back in chapter 1 he gave an example of a C-inductive argument, where the premise that 100 ravens have all been observed to be black is a C-inductive argument towards the conclusion that all ravens are black.

A C-inductive argument is characterised by the existence of a hypothesis where all the relevant evidence confirms the hypothesis and none of it disproves it, but where the possibility remains that some evidence at present unknown might someday disprove the hypothesis. Irrelevant matters do not contribute to the C-inductive argument, so for instance the colour of seagulls is not relevant to the hypothesis that all ravens are black. Seagulls of any colour are equally consistent both with the hypothesis that all ravens are black and with its converse, that there exists at least one non-black raven.

At the moment, we don’t know enough about the universe to know its first cause. Therefore, so far as we can tell, the existence of the universe is consistent both with God’s existence and with God’s nonexistence. It is therefore a piece of irrelevant evidence just like the colour of seagulls. To call the existence of the universe a C-inductive argument for God is incorrect.

However, Swinburne’s definition of a C-inductive argument has morphed a bit. It now isn’t anything to do with the example he gave. In terms of how he is actually using the phrase, a good C-inductive argument is one which results in a conditional probability greater than the prior probability you had before you evaluated the new evidence. In other words, a C-inductive argument is one where you feed the numbers into Bayes’ theorem and come out with a probability greater than you started with.

In this, he is demonstrating again this fundamental misunderstanding of what Bayes’ theorem does. As I described in the previous chapter, Bayes allows us to refine our understanding of the population of a certain class of entity, by allowing us to eliminate on the basis of evidence certain proportions of a larger population whose characteristics we are aware of. In the example I gave in the previous article, we were able to refine probabilities by eliminating from consideration skirt-wearers and children with short hair from an overall population of schoolchildren, in order to assess how many long-haired trouser-wearers were girls.

We don’t have a population here. We don’t have a bunch of universes, some of which exist uncaused and some of which are made by God. We don’t have a way of distinguishing caused universes from uncaused. All we have is Swinburne making numbers up again.


  1. As I consider a godless universe simpler, it seems I should prefer atheism even according to Swinburne. Of course, I agree with all your critiques (except that Bayes' theorem is applicable in principle - but we've been there). Your point about C-Inductive reasoning is particularly poignant. Swinburne's epistemology of preferring the simpler of two hypothesis without any evidence whatsoever is especially galling.

    I take it that Swinburne doesn't consider other causal possibilities? I assume he rejects circular causation as a matter of axioms, i.e. positing that nothing can cause itself. Does he talk about a-temporal causation? A modern example would be the creation of the universe from a mutliverse (which doesn't happen in "time", which is a concept that only applies within our universe); an ancient example would be Augustine's arguments for the atemporal nature of god's actions and nature, and specifically for creation/genesis.

  2. I take it that Swinburne doesn't consider other causal possibilities?

    I believe so. At least, in an earlier chapter he says that is it surely impossible for something to explain itself, so I suspect he takes a similar view towards something causing itself.

    He also in an earlier chapter briefly described the idea of God existing outside time, but says that while it might be true, it is unnecessary for a layman's understanding of an eternal God, and so he won't addres it in any detail.