Thursday, August 30, 2012

Deductive-nomological model of scientific explanation

To: Yasuko S.
From: Geoffrey Klempner
Subject: Deductive-nomological model of scientific explanation
Date: 25th March 2008 13:44

Dear Yasuko,

Thank you for your email of 16 March, with your University of London Methodology essay, in response to the question, 'Scientific explanation is the explanation of effects by causes.' Discuss.

This is a solid essay which addresses the question relevantly and clearly, and would be very adequate as an exam answer. Although you say that you find the topic difficult, what you have written is relatively clear and straightforward.

I agree that the main issue is whether the D-N model is adequate, and, if not, whether adding conditions relating to our intuitions about causal relations would be sufficient to make good the deficiency.

However, one thing that this simple picture ignores the fact that one of the motivations for proposing the D-N model was to avoid the need to refer to the 'primitive' notion of cause and effect.

You do mention Hume's critique of causation in your essay, but only in the context of the question whether Hume's revisionary analysis of causation is itself adequate.

Why do we think we 'know' what a cause is? Could it be that we are under some kind of illusion or primitive superstition that there are such things as 'causes' and 'effects'?

Bertrand Russell is one example of a very prominent philosopher who argues that the notion of causation ought to be entirely removed from science. Here is an actual quote:

'In the following paper I wish... to maintain that the word "cause" is so inextricably bound up with misleading associations as to make its complete extrusion from the philosophical vocabulary desirable...' ('On the Notion of a Cause' in 'Mysticism and Logic' Unwin 1917).

What I am suggesting is that a better structure for the essay might be to start off briefly with objections to causation, then a discussion of Hempel, followed by a reconsideration of causation which avoids, or tries to avoid, the original objections.

Who needs causation when you have laws which explain and predict? This is not, however, a rhetorical question as your examples show. Some alleged 'explanations' which follow the D-N model are not explanations, and the reason they are not is because they 'explain' a cause in terms of its effect.

This still leaves us in the position of relying on ungrounded 'intuitions'. Why are these intuitions so strong? Why is it so difficult to reduce causation to other notions?

One possible line that I would consider if I had to answer this question would be to note the important relationship between the concept of causation and human agency. Identifying causes and effects is what we do all the time in our daily lives, whenever we seek to intervene in the natural course of nature in order to bring about a result which we are aiming for.

For example, I am building a giant sun-clock using a flag pole. How long does the pole have to be in order to cast a shadow which will reach the hour markings which I have set out on the lawn? Well, obviously it depends on the time of year! But the point of the example is that if I want to change the length of the shadow, I have to lengthen or shorten the pole. I can't alter the pole by altering the shadow.

In general, we identify 'causes' as the things we can alter through our agency, and 'effects' as those which result from the interventions which we make, using the apparatus at our disposal.

Against Russell, it could be argued that we simply *can't* get rid of the notion of cause and effect because it is an essential part of understanding ourselves as agents in a world in which we do things and in which things are done to us.

Rather than simply re-introducing 'primitive' causation, however, this suggests a possible way in which one might meet the objections to the D-N model by viewing D-N explanation within the context of human technology. Say, if you like, that every example of a D-N pattern is an 'explanation'. Nevertheless, some 'explanations' in this broad or weak sense are ineffectual, while others are not, and the difference lies in their practical utility. An example of a D-N pattern is an 'explanation' in a strong sense when it represents the structure of a procedure which one might ideally employ to bring about the explanandum through human intervention in the course of nature. (Of course, such intervention is not always practically possible, but we can still imagine what we would do if our powers were suitably extended.)

Although we started off criticizing Hempel, this looks like a possible defence of Hempel rather than a criticism. Yes, the D-N model produces some funny results, but we have an explanation for these funny results which doesn't undermine but rather supports the basic idea of D-N explanation.

As far as your essay is concerned, if you were answering this question in an exam, then a short section explaining how the D-N model was intended as a better alternative to cause-and-effect explanations would help to underline the point that the D-N model, useful though it may be, doesn't succeed, after all, in removing the need for a concept of a 'cause'.

This is close to the eclectic view which you support at the end of your essay, that what we should do is 'not to make models of explanation compete with each other but to keep integrating them and expanding the model of explanation in order to make it possible to cover its diversity.'

All the best,

Geoffrey