"To be an empiricist is to withhold belief in anything that goes beyond the actual, observable phenomena, and to recognize no objective modality in nature. To develop an empiricist account of science is to depict it as involving a search for truth only about the empirical world, about what is actual and observable. Since scientific activity is an enormously rich and complex cultural phenomenon, this account of science must be accompanied by auxiliary theories about scientific explanation, conceptual commitment, modal language, and much else. But it must involve throughout a resolute rejection of the demand for an explanation of the regularities in the observable course of nature, by means of truths concerning a reality beyond what is actual and observable, as a demand which plays no role in the scientific enterprise" (The Scientific Image, pp. 202-203).

A. The Big Picture

     I will try to state in a series of reflections how VF stands with respect to the others we have looked at.

     1. The first point to note is that the whole discussion of causality and explanation in chapter 5 of The Scientific Image is set wholly and explicitly within a Humean framework. That is, it is taken for granted that (i) earnest talk about causality takes us beyond what is observable and into the realm of the metaphysical, and that (ii) an Aristotelian conception of causality as action productive of esse is beyond the pale.

     2. VF's strategy is to demonstrate to his empiricist colleagues (including Lewis, Mackie, Salmon, Tooley, and others) that a basically Humean notion of causality is a metaphysically tame notion whose proper context is a theory of explanation rather than a theory of scientific description. That is, to say that A is a cause is mainly to say that 'Because  A' is an answer to a 'why' question, i.e., a request for explanation in a certain respect within a given context. Only then can what empiricists say about notions like counterfactual dependence, transmission of probabilities, and statistical relevance be seen as having no 'special' metaphysical implications, but to be aimed instead at providing criteria for good answers to 'why' questions.

     Another way to put the main point is this: When basically Humean accounts of causality pretend to be general theories about causality as it is in the objects, they invariably fall into difficulties with problem cases, with the context-dependent nature of counterfactual reasoning, and with the arbitrariness of singling out as causes just some of the things which satisfy the putative metaphysical analysis; but if we see these accounts of 'causality' as providing criteria for good answers to highly contextualized requests for explanation, then they can be genuinely useful and helpful.

     3. What's more, once we begin to think of the connective 'because' as not singling out metaphysically special entities like causes, causal processes, causal interactions, etc., we can come to see that explanation does not and need not, even in the sciences, involve any metaphysical commitments not already involved in accepting a theory as describing various structures. (Of course, VF will then plug in his distinction between accepting a theory and believing a theory.) In particular, we need not think of scientific explanation as invoking a special class of metaphysical notions such as 'cause', 'causal process', 'causal propogation', etc. The mistake of empiricists like Hempel and the early Salmon was not in thinking that descriptions explain but in failing to understand that explanation is highly contextualized description and not a direct relation between theory and fact. This mistake paved the way for creeping realism:

"Many attempts were made to account for such 'explanatory power' purely in terms of those features and resources of a theory that make it informative (that is, allow it to give better descriptions ... But these attempts ran into seemingly insuperable difficulties. The conviction grew that explanatory power is something quite irreducible, a special feature differing in kind from empirical adequacy and strength. An inspection of examples defeats any attempt to identify the ability to explain with any complex of those more familiar and down-to-earth virtues that are used to evaluate the theory qua description. Simultaneously, it was argued that what science is really after is understanding, that this consists in being in a position to explain, hence what science is really after goes well beyond empirical adequacy and strength. Finally, since the theory's ability to explain provides a clear reason for accepting it, it was argued that explanatory power is evidence for the truth of the theory, special evidence that goes beyond any evidence we have for the theory's empirical adequacy" (p. 154).

     In truth, however, the main difference between theoretical description and explanation is the difference between pure science and applied science, i.e., between purely theoretical description and the application of such pure science (or scientific description), within particular contexts and particular background assumptions, to particular 'why' questions:

"So to say that a given theory can be used to explain a certain fact is always elliptical for: there is a proposition which is a telling answer, relative to this theory, to the request for information about certain facts (those counted as relevant for this question) that bear on a comparison between this fact which is the case and certain (contextually specified) alternatives which are not the case" (p. 156)

     4. The mistake realists make is to think that scientific explanation is distinctive and special because it goes beyond theoretical description to talk about the causes of particular effects, or about causal processes that issue in certain effects, etc.:

"Once you decide that explanation is something irreducible and special, the door is opened to elaboration by means of further concepts pertaining thereto, all equally irreducible and special. The premisses of an explanation have to include lawlike statements; a statement is lawlike exactly if it implies some non-trivial counterfactual conditional statement; but it can do so only by asserting relationships of necessity in nature. Not all classes correspond to genuine properties; properties and propensities figure in explanation. Not everyone has joined this return to essentialism or neo-Aristotelian realism, but some eminent realists have publicly explored or advocated it" (p. 155).

     To the contrary, what is distinctive about scientific explanation is merely that it draws its answers from scientific theories or descriptions rather than from elsewhere.

B. An Aristotelian Reply

     In some respects the Aristotelian should be willing to stay on the sidelines and enjoy this dispute among the Humeans. After all, Aristotelians take as much pleasure as VF in pointing out the inadequacies of empiricist accounts of causality when the latter purport to capture our intuitive notion of a cause. And, as we saw earlier in the course, the Aristotelian need not disagree with VF's insistence that explanation is applied science and that it involves answering various sorts of 'why' questions posed within certain contexts. This is in large part what the doctrine of the four causes is about.

     The disagreements come at a much deeper level. First of all, the Aristotelian will deny that the invocation of real causes is irrelevant to scientific explanation--just because the Aristotelian sees all explanation, in both ordinary and scientific contexts, as invoking real causes of one sort or another. For this reason, the Aristotelian can see properly scientific explanation as an extension of ordinary explanatory practices and not as something "special and irreducible."

     A second and intimately related disagreement has to do with the question of whether causal talk is 'theoretical' as opposed to 'observational'. We have already investigated this question in some depth, though it's difficult to see any knock-down arguments for one side over the other.

C. Probabilities and Causality

     One of the lessons of recent discussions of objective probability and causality is that there is no simple relation between causes and prior probabilities (relative to a theory).

     For instance, it seems wrong to say that, in general, C is a cause of E only if Prob(E/C) > Prob(E/not-C). (See VF's example on p. 148)

     Again, it seems wrong to say that C is causally relevant to E only if Prob(E/C) is greater than or equal to Prob(E). (See 'Simpson's Paradox' on p. 149.)

     Also, it seems wrong to insist even that C is causally relevant to E if Prob(E/C) is not equal to Prob(E). (The plant/defoliant example--I will explain this. Also, let's take a brief look at the example on p. 150.)