Thursday, September 6, 2012

Role of causes and effects in scientific explanation

To: Yasuko S.
From: Geoffrey Klempner
Subject: Role of causes and effects in scientific explanation
Date: 21 April 2008 11:24

Dear Yasuko,

Thank you for your email of 11 April with the second version of your University of London Methodology essay, 'A scientific explanation is the explanation of effects by causes.' Discuss.

This is a very clear and well structured answer which shows that you have a good grasp of the issues. I have nothing to say about sections 1-3. However, section 4 was a bit brief and 'breathless', and I am not sure that we understand the role of human agency in the same way.

You are right to say that Hume analyses causation on two levels. This is how I would explain it:

1. On the objective level, a causal statement of the form 'A causes B' is true by virtue of a universal law. Hume avoids the question of how one states ceteris paribus conditions. This is a big problem in itself because A and B are particular events, whereas 'an event of type A' and 'an event of type B' are general terms which necessarily lack all the properties of actual events A and B. That's why one can't simply say, 'All cases of A are followed by B'. Arguably, precise ceteris paribus conditions can never be adequately stated. However, Hume can still assert that there 'exists' a law, albeit unstatable, by virtue of which any given causal statement is true, if indeed it is true. There is nothing else in reality, no 'influence' or 'necessary connection' or whatever.

We should observe that the idea of universal law is an extremely powerful notion. We are talking about a 'truth' which applies at all times and in all places. That does not go without saying. No-one can ever know that such a law is true, because human beings are finite. At best, we can only have inductive evidence for its truth.

2. Human beings are interested in causes. That also does not go without saying. We form our empirical beliefs by means of a process whereby an impression (e.g. a stone thrown at a window) calls up a 'lively idea' of the expected consequence (e.g. the window breaking), according to certain principles which Hume explains. This psychological account of how beliefs are formed is indeed a causal explanation (a point on which Hume has been wrongly criticized; in fact, he is being perfectly consistent in talking of causes and effects in the mental realm of ideas, then analysing these in terms of 1.).

If one imagined a race of intelligent trees, then it seems that they could do this too. For example, they learn from experience to form the lively idea of rain from the impression of the sky becoming dark with clouds.

However, the point I was trying to make about agency is more contentious than anything Hume claims. It is that in order to form the full-blooded notion of a 'cause' it is necessary to be a physical agent, capable of intervening in the world, and not just a passive observer thinking (= 'mental agency') about what it observes.

It would seem to follow that if we (per impossibile) were not physical agents, then the only kind of 'explanation' we could be interested in is one which was perfectly symmetrical with prediction. Do you agree with that? I'm not sure I do.

How is it that we are so good at identifying 'the' cause of something, when we are dealing with complex conditions? Consider the role of causation in the law: what was it that 'caused' the car crash, the stormy weather, the road works, the lazy mechanic who didn't adjust the brakes properly, the inattentive driver, the reckless pedestrian, etc. etc.? All these conditions were required for the car crash to occur, but we don't say that it was 'caused' by the weather or the road works.

One might try to apply J.L. Mackie's definition of 'cause' in terms of an 'INUS condition', an 'insufficient but necessary part of an unnecessary but sufficient condition'. But exactly *which* part? that's the question. (For Mackie, see the Stanford Encyclopedia article on 'Probabilistic Causation'.) The law is concerned to punish and deter, so we are led to identify the bad action or decision by a human agent as 'the' cause, although as in this case, we may have to accept that there are several 'causes', which played a lesser or greater part.

Salmon's account seems to have a lot going for it because we are interested in 'mechanisms', and this goes deeper than simply the ability to predict. Wouldn't intelligent trees be interested in mechanisms too? Yes, up to a point. Mechanism gives detail, in terms of which one can distinguish deeper, fuller explanations from more superficial ones. Maybe this is enough and one doesn't need to factor in the extra variable of human physical agency.

So, you see, this is not such an easy issue to resolve. I think that the thing to say is that the fact that we are physical agents is significant in analysing our concept of a 'cause', but it is not totally clear whether, if counterfactually one removed the element of physical agency, there might still be a role for a notion of cause which goes beyond the simple D-N model as Salmon claims.

In analysing causation we are not dealing with a 'cut and dried' issue but a deep philosophical problem. In an exam, it is perfectly OK to recognize this, to admit that your view is not settled and that you are aware of conflicting considerations. If I had to answer this question, I guess that is what I would say.

All the best,

Geoffrey