Chapter 2: Kim’s causal exclusion argument against non-reductivism
2.5 A hidden premise in the exclusion argument
2.5.1 Nomological necessitation and sufficiency interpretations of causation
The first question on our path to understand the causal side of the exclusion problem is this: what is a sufficient cause? The first interpretation of a sufficient cause that can be found in Kim’s early writings is based on the concept of nomological necessitation (Kim 1992), although he rejected this approach in later works (Kim 2005, 2007) it is interesting to see how that worked. First, it convinced a lot a people and interestingly enough it suffers from similar problems as those theories Kim started to favour in his later works.
According to the nomological necessitation interpretation C causes E in case there is a law of nature such that C necessitates the occurrence of E if C occurs. This is considered to be an old-fashioned concept, even in the late Kim’s own view, that dates back to the heydays of the covering law model of explanation. On the one hand it requires too much, as law-like necessitation is not a characteristic of causation or at least causation at non-fundamental levels, not even at the level of chemical reactions. Any “law” that describes the behaviour of mereologically complex objects is hedged with ceteris paribus clauses as there are background conditions under which the relation would not apply. So, it can be argued that this concept is too strong: causes do not necessitate their effects nomologically, at most they necessitate their effects if certain background conditions hold. Before turning to further problems this interpretation faces it’s better to go through a different possible interpretation because there is a common problem these interpretations share.
Even though Kim never refers to this theory, it seems natural to try to interpret the concept of a sufficient cause based on Mackie’s INUS condition analysis of the concept of a cause. Mackie (1974) formulated a theory according to which C causes E if and only if C is an Insufficient but Necessary condition as part of an Unnecessary but Sufficient system of conditions for E. The whole analysis is based on the notions of necessary and sufficient conditions. Let see how this could provide us with an interpretation of the concept of a sufficient cause. According to Mackie a system of conditions (A & B & C) can be sufficient to bring an effect E about. C or any other member of this system of conditions is insufficient in itself, nut necessary in the sense that without it the remaining conditions are insufficient to bring E about. (A & B & C) as a system of conditions is unnecessary for E as a different system of conditions (X & Y & Z) could do the same work.
Under this interpretation a sufficient cause is the presence of a set of causal factors that together are sufficient to bring the effect about. But this is not exactly how Kim talks about a sufficient cause, as whenever he provides examples he talks about a single causal factor, e.g. the presence of the physical realizer of a mental property. But I think, charitably read, he simply highlights one factor from the set of conditions that is jointly sufficient. It is common practice to talk about “the” cause of an outcome bringing it into the foreground while pushing the other factors into the background. Mackie, Lewis and many other philosophers agreed that this practice is understandable as a pragmatic feature of causal talk. This practice is a product of our human need to highlight what is relevant, informative in the particular context of communication. If Kim’s term of a “sufficient cause” is interpreted broadly, as an elliptic reference to a set of conditions together sufficient for an effect, it starts to make good sense.
The main problem with both the INUS-based and the nomological necessitation interpretation of a sufficient cause derives from the fact that we are talking about regularity theories of causation. In this respect the INUS theory is exactly like the concept based on nomological necessity: the criticisms put forward below apply to both theories. Regularity theories not only fail to handle problems concerning the direction of causation, but they also have a hard time accounting for causal relevance. The problem of relevance can be formulated in the following way. It is possible that the effect E is always present when C, the cause is present, but it happens because E and C have a common cause, not because C causes E (see: Figure 2.5-1). When that is true, the presence of
A E C C E C causes E A is the common cause of C and E Figure 2.5-1
C can still be interpreted as an INUS condition for E, also as being nomologically necessitated by C, but no one would say that it is a cause of E.
There are some other well-known cases in which there is a constant conjunction between C and E, but no causal connection between them. The classic example comes from Salmon (1984). A man takes birth control pills (C) to prevent pregnancy (E). The example is trivially non-causal, as a man is incapable of becoming pregnant, it is biologically not possible. Nonetheless, whenever C is present E is present, so, according to the regularity-based approach C should be a sufficient means to avoid pregnancy. The general problem is that an analysis based on regularities, cannot avoid counting non-causes as causes. And that is a problem for both the nomological necessitation view and the INUS analysis.