Remarks on Inference to the most probable cause
Cartwright’s concerns cross cut those of Van Fraassen.
Like him, she takes there to be no link between explanation by citing a law and truth. So she rejects the use of Inference to the Best Explanation against his position (Constructive Empiricism). Her reasons for this are diverse. They include:
The observation that scientists do not act as though the high level are unique and true. They pick and choose versions of high level theories, and explanations from them of lower level laws and theories, for convenience. That suggests that they are noty treating the high level laws as strictly true.
High level laws typically have to say ‘all other things being equal’ or ceteris paribus. But all other things are typically not equal. So such laws do not actually cover, are not true of, real situations.
There is no a priori reason to think that the world is generally determined by high level laws. God may have written a few and then got tired.
But perhaps the most important claim is this.
When you derive a lower level description of an actual system (a phenomenological law) from a higher level law, even if the derivation works smoothly, still, the higher level law does not make the lower level law true. There’s no connection between the organising role (summarised in what you can derive from a higher level law) and its truth.
By contrast, when you explain an event by citing the cause of the event, that explanation only works if the cause really exists. Such an explanation has ‘existential import’. So there is a fundamental difference between Inference to the Best Explanation and Inference to the Probable Cause. The truth of the higher level theory is an optional extra for the former. The existence of the cause is not an optional extra for the latter.
So from this, she agrees with Van Fraassen in rejecting Inference to the Best Explanation. There is no inference from explanation to the truth of theory. But she disagrees when it comes to causal explanation. A good causal explanation gives reason to believe in the cause whether observable or unobservable.
But this leaves a worry. Whilst Cartwright *may* have drawn good distinction between two sorts of explanation (one with existential import and one not), how *exactly* does that distinction interact with the general under-determination argument against belief in unobservables? If an observable effect can be explained by an unobservable cause then it can at least seem to be equally well explained by an infinite number of other unobservable causes.
Of course, if one thinks that to be a real explanation something has to be true, one will say that the real explanation must mention the real cause whilst an apparent explanation which mentions a cause that, actually, does not exist is not a real explanation. But the argument from under-determination is still a threat because it challenges us how we can know the real explanation from the merely apparent. Why is Cartwright’s position actually any better when it comes to using things that look like – and may be but may not be – causal explanations to pick out real causes than the same sort of move when it comes to finding true theories by using what look like – and may be but may not be – true explanations?
She will say that something can be an explanation by high level theory without being true. She thinks that truth is an optional extra for such explanations. By contrast causal explanation needs there to be a cause. But suppose we are interested just in explanations which are also true: so either true causal explanations or true theory explanations. How has she helped distinguish our right to believe in particular causes over particular high level theories?