Wednesday, 26 August 2009

Swinburne: "Complete" explanations

In chapter 4, Swinburne looks at what makes a Complete Explanation. He starts out by looking at what he regards as possible limits to scientific explanations.

Phenomena of two kinds can be shown not to be explicable scientifically. First there are the phenomena which are too odd to be fitted into the established pattern of scientific explanation, and, secondly, there are phenomena that are too big to be fitted into any pattern of scientific explanation.

This seems to be a variety of the Argument from Personal Incredulity. If Swinburne thinks some phenomenon is too strange or big to have occurred naturally, then God possibly/probably did it. (The phenomena he is thinking of are both too strange and too big to fit into the category of personal explanations in the form of intentional actions by humans.)

The most obvious problem with this approach can be easily shown by a brief look at the history of science, which shows a progressive widening of the scope of things which were previously thought to be the work of God but which have since been shown to be the result of the operation of unchanging natural laws. It seems that our perception of what is too odd to have a natural explanation is constantly changing, a fact which Swinburne seems not to take into account.

Swinburne goes on further about odd phenomena:

To show phenomena too odd to be explicable scientifically, the theist needs to show that there is good evidence for a scientific system h covering a certain range of phenomena, but that it is not a consequence of h (within the general range of h) occur, and that any attempt to amend or expand h that would allow it to predict e would make h so complex that it would be very improbable that it is true.

This is again a gross misuse of the term “improbable”.

Swinburne describes two categories of odd phenomena. One is miracles, and the other is general classes of phenomena for which a scientific explanation is too complex to be plausible.

If a phenomenon can be accurately explained and predicted by means of a theory, I see no reason to be concerned with how complex the equations are. I’m sure I would rapidly get lost in the mathematics of quantum electrodynamics for which Richard Feynmann received his Nobel Prize. Is the fact that the equations for explaining something supposedly very simple (i.e. the behaviour of subatomic particles such as electrons) are very complex a reason for thinking the theory isn’t true? Do we have instead to assume that Feymann’s quantum electrodynamics is wrong (even though its predictions are very accurate) and that it is really God making all those electrons behave in that particular way? Swinburne (at this stage) offers no criteria for deciding whether a scientific explanation for a phenomenon is too complex to be plausible.

Swinburne also makes a description of big phenomena

The other phenomena that cannot be explained scientifically are phenomena that are too big for science, and too big not merely for some particular well-established scientific system, but for any scientific system. … But what, as I shall show more precisely in chapter 7, science cannot explain is why there are any states of affairs at all; it can only explain why, given that there are such states, this state is followed by that state. Nor could it explain, as I shall show more precisely in chapter 8, why the most fundamental natural laws of all hold. Either these are brute facts about the world, or they have an explanation of a different kind.

Quite clearly, we don’t have those scientific explanations yet. But Swinburne’s assertion is that it is fundamentally impossible for science to elucidate these matters at any point in the future.

When making such statements regarding ultimate limits to knowledge (as opposed to the limits of what we have learned so far), it is wise to couch them in something more than a statement to the effect “I don’t see how this could ever be found out”.

Let’s give Swinburne the benefit of the doubt here - all he is so far saying is that he thinks there are ultimate limits - he isn’t (yet) saying where they are or why. We can wait for those later chapters to see him expand on this theme.

Next Swinburne goes on to make several definitions of varieties of complete explanation. I suspect these will get used extensively further on in the book, so we had better be familiar with them.

A full explanation of a phenomenon exists if it includes a cause and a reason, which together necessitate its occurrence. (The cause is the initial condition and the reason is the operation of some law of behaviour.) If a phenomenon has a full explanation, nothing more needs to be described about how the initial condition resulted in the final state.

A partial explanation is where the suggested cause and reason, while necessary to the understanding of the final state, leave unexplained or unstated the operation of some underlying law. (In Dennett’s nomenclature as described previously, the use of the design stance would be a partial explanation of exactly this kind.)

A complete explanation is a full explanation where the details of the causes - the initial conditions, and the manner of operation of the reasons is itself explained.

Quite why this deserves a separate definition is not clear, but there it is. It seems to me that there are varying degrees of partial explanation which are described at varying levels of abstraction, and the varying levels don’t really need separate definitions - you merely need to know what level of abstraction you are working with and therefore what assumptions are being made about the underlying physical laws whose operation has been abstracted. For instance, laws of chemistry assume that the laws of physics are in operation, laws of cell biology in turn assume the operation of the laws of chemistry, and the laws of botany or zoology assume that the laws of cell biology are working. When explaining something, you choose the level of abstraction that is appropriate to the situation. Of course, there is a need to ensure that your abstractions are correct, and that nothing that you postulate within your chemical theories actually violates any known laws of physics.

Then an ultimate explanation is a complete explanation which traces itself back to original causes - ultimate brute facts for which there is no further explanation.

Finally there is an absolute explanation which is an ultimate explanation in which all the factors cited are either self-explanatory or logically necessary.

We don’t yet have ultimate or absolute explanations for the universe, and might never discover them. Nevertheless, Swinburne is confident that it is possible to decide on the basis of balance of probability whether the ultimate explanation for the universe is God. He shies away from claiming that God is an absolute explanation, since he considers it impossible for anything to explain itself.

Swinburne then describes how Newton’s laws of motion were so delightfully simple and so powerful in their explanatory scope that the scientists of the day genuinely believed that science had reached a terminus in Newton’s laws. We aren’t in that position today, but Swinburne suggests that if and when a theory of that explanatory power is devised that explains all the phenomena we know of today, then we will have good reason to believe we are at a terminus. Quite why a mistaken past belief of the existence of a terminus in explanation should be regarded as a model for defining criteria for a future belief that a terminus has been found is left unexplained. It seems to me that something more than simplicity and explanatory power is going to be needed if we are to be sure we haven’t fooled ourselves into thinking that there are no new unexplained phenomena to be discovered.

Although simplicity is regarded by Swinburne as an indicator of “prior probability” (a term I dislike for reasons stated before) he gives no indication as to what level of complexity in his view rules out a natural law as being a plausible explanation. Trying to work out what he means in all this is a bit like trying to wrestle with fog. One always harbours the suspicion that these criteria will get refined at a later stage so that God just so happens to turn out to be really quite probable.

Swinburne now turns again to personal explanations. Time for another quote.

Scientific and personal explanations are on a level—that is, are rivals for the explanation of phenomena—it would seem to follow that a scientific explanation could explain a personal one, and conversely; and that the criteria that it does so are any gain of prior probability and explanatory power that would result from supposing that it does. By a scientific explanation explaining a personal explanation, I do not mean the one being analysed in terms of the other—we saw in Chapter 2 that a personal explanation cannot be explained in terms of a scientific explanation, and it is surely equally plausible to suppose that a scientific explanation cannot be analysed in terms of a personal explanation. What, rather, I do mean by a scientific explanation explaining a personal explanation is the existence and operation of the factors being involved in a personal explanation being explained by the existence of factors involved in a scientific explanation. A scientific explanation might be given of how people come to exist, and to have the intentions, beliefs and basic powers that they have.

Hang on! Back in chapter 2 he was saying that personal explanations aren’t scientifically analysable, and now he is saying that they are. Or perhaps he is saying that they both are and aren’t. If intentions are analysable in the terms he has just described (and which by the way agrees quite nicely with Dennett’s approach), then there is no need to posit the inescapable dualism he brought up in Chapter 2. It seems that this argument is in danger of disappearing up its own fundament.

But let us go on and see what more he has to say on this.

It is the programme of physicalism to effect a reduction of just this kind. The theist who tries to explain why the world is and works as it does is attempting the reverse programme—to give a personal explanation in terms of God, of the existence and operation of the factors involved in scientific explanation.

In what sense would such an explanation be an ultimate or absolute explanation which Swinburne is seeking? By his own definitions, complete explanations work by taking full explanations pitched at a high level of abstraction, and explaining them by means of the operation of physical laws at a lower level of abstraction. By all appearances, explanations expressed in terms of intentions are an extremely high level of abstraction, depending in turn on the laws of biology, of chemistry and physics for progressively more detailed levels of explanation.

So if we are looking for an ultimate explanation expressed in terms of theism, we would need to explain how God’s intentions (unlike human ones) operate at the lowest level of abstraction – below that of physics, instead of at a very high level of abstraction – above that of biology. Let’s see what Swinburne makes of that in future chapters.


  1. Very enjoyable; keep up the good work!

  2. Swinburne's "odd" exception example is sheer nonsense, a clear misunderstanding of epistemology - he actually advocates a theory that doesn't explain the data, amazing. But his two exceptions are valid, and won't be resolved with the progress of science. There are other exceptions too, other truths science cannot reveal - science, in its narrow sense, cannot for example determine what makes a good scientific theory.

    However, what Swinburne misses is that science is just the pursuit of natural philosophy - the love of wisdom related to nature - and as such is connected and continuous with the rest of philosophy. While scientific methods may not be able to determine that we've reached an ultimate explanation, philosophical ones (such as having an absolute explanation) can convince us that we have. Similarly, while science cannot explain extremely rare and hardly-observed phenomena (because these are not subject to scientific scrutiny) or the universe as a whole (because it is unique, so there is nothing to compare it by), or the origin of the fundamental laws of nature (because they are assumed, not proved, in physics) - all these can be at least addressed via philosophy, and cannot at all be addressed (reasonably) via religion.

    As for personal explanations - these are only ultimate if you accept that persons are ultimate elements of reality, which is entirely complex and unwarranted. The naturalistic hypothesis is that persons are emergent phenomena, and seems much better suited to explain reality by Swinburne's own criteria - it's simpler, and fits better with our current knowledge.