Mechanisms and Laws: Clarifying the Debate - Mechanism and Causality in Biology and Economics - page 133

their interaction); nobody else matters, and so no p-regularity (or any regularity)
matters, to whether they interact.
A second kind of argument for the separability of regularity and causation turns
on the possibility of causal relations that have no echo in the correlational structure
of the world. For example, one might have a mutation that reduces the overall
chance that one will get lung cancer (i.e., the mutation has negative statistical
relevance for cancer) but that, in a few unfortunate individuals, is, in fact, the
trigger for lung cancer. And one might get lung cancer in virtue of having that
mutation. One might smoke three packs a day (raising the chance of getting lung
cancer) and in fact get lung cancer because of the mutation. The actual causal
structure in such cases would appear to run counter to the regularities. To borrow a
kind of example first described by Jonathan Schaffer ( 2000 ), we might imagine two
neurons, A and B, synapsing on a third neuron, C. Suppose we know from experi-
mental investigation that the probability of C's firing given A's firing alone is 0.5,
that the probability of C's firing given B's firing alone is 0.5, and that the probabil-
ity of spontaneous firing in C is 0. Now suppose that A, B, and C all fire. These facts
leave the causal facts under-determined. For in this situation, it might be that A
caused C to fire, that B caused C to fire, or that both A and B caused C to fire. The
difference between these possibilities cannot, ex hypothesi, depend on the
regularities involved. It would seem that there is a further fact about the actual
causal structure of the situation. Regularities, it might be thought, provide evidence
about the causal structure of a mechanism. But the causal structure of the mecha-
nism is something over and above the regularities by which that structure can be
detected.
We do not insist on the view that causation is intrinsic, actual, and singular. We
simply note that Leuridan does not address the heart of the debate about whether
regularities are more fundamental than causation and mechanisms. Some
philosophers, most explicitly Glennan ( 2002 ), Woodward ( 2002 ), and Craver
( 2007 ), appear to agree (to a first approximation) with the idea that the interactions
in a mechanism should be characterized in terms of invariant change relating
generalizations. They stress, for example, that knowledge of causes is practically
valuable precisely because it is general. And they emphasize the close connection
between the generality of causation and the methods used to test causal relations
(see Woodward 2004 ). Bogen, we have seen, disagrees. The merits of and relations
among these approaches have been discussed at some length by Craver ( 2007 ),
Glennan ( 2002 , 2010 ), Psillos ( 2004 ), Tabery ( 2004 ), and Woodward ( 2002 , 2010 ).
Leuridan again does not address this discussion.
One last point deserves mention before almost leaving Leuridan's putatively
ontological discussion. Leuridan distinguishes between p-laws and (c)p-laws. This
distinction is required for Leuridan to distinguish p-laws that are merely useful for
prediction from those that, in addition, allow one to explain and control events. One
might predict that one is about to run out of gas by looking at one's gas gauge, but
the reading on the gas gauge does not explain the emptiness of the tank. Nor could
one make it further down the road by breaking the gauge. For this reason, Leuridan
(like the mechanists Glennan ( 2002 ) and Craver ( 2007 )) appeals to Woodward's