Formulating the causal closure principle

In document The chances of higher-level causation: an investigation into causal exclusion arguments (Page 84-89)

Chapter 2: Kim’s causal exclusion argument against non-reductivism

2.2 A summary of Kim’s exclusion argument

2.3.1 Formulating the causal closure principle

The causal closure principle is at the heart of physicalism. The basic idea is as follows: “One way of stating the principle of physical causal closure is this: If you pick any physical event and trace out its causal ancestry or posterity, that will never take

you outside the physical domain.” (Kim 1998:40, my italics)

As Kim summarizes it in his earlier writings:

To deny it would mean that physical properties, either fundamental or broadly physical properties, are insufficient to provide a complete explanation even of physical effects. According to most physicalists the success of the physical sciences provides good inductive evidence to think otherwise (see: Kim 2005:70). Kim formulated this succinctly:

“If you reject this principle, you are ipso facto rejecting the in-principle completability of physics - that is, the possibility of a complete and comprehensive physical theory of all physical phenomena. For you would be saying that any complete explanatory theory of the physical domain must invoke nonphysical causal agents. Never mind a complete physical explanation of everything there is; there couldn't even be a complete physical explanation of everything physical. It is

safe to assume that no serious physicalist could accept such a prospect.” (Kim

1998:40, my italics)

Unfortunately for physicalists, this simple formulation is probably too weak to capture the whole idea. It does not rule out the possibility of a causal chain between two physical events, C and E, that includes a non-physical link D. But this would mean that the physical domain is not closed, as to account for the physical event E we would need to evoke something non-physical, e.g. a mental cause. In case causation is transitive, the above formulation of closure allows for non-physical causes without running into a contradiction. Even though E has a non-physical cause D, by transitivity it also has a physical cause C (see: Lowe 2000).

Let me draw attention to something that usually goes unnoticed with respect to this counterargument against the simple formulation of closure. In many popular theories of causation, the relation is assumed to be transitive, but at present this is considered to be a

vexed issue. There are serious, stubborn counterexamples to the transitivity of causation, therefore intransitivity seems to be a plausible option (see: Hitchcock 2001, Maslen 2004, Hall 2004a). In chapter 3, I argue for the plausibility of a version of the latter view34. However, if

causation is not transitive, there is no guarantee that when C causes D and D causes E then C is also a cause of E.

Applying this to Lowe’s criticism of simple closure, there is no guarantee that when a physical event has a non-physical cause it has a physical cause as well just because the non- physical cause itself was caused by a physical cause. If causation is intransitive any causal chain might fail transitivity and we are faced with the duty to check what follows from this to the acceptability of the simple formulation of the closure principle.

To make the project meaningful, we should accept that even though we might not be able to decide which particular causal chain is transitive and which is not, due to epistemic limitations, the transitivity of a particular causal chain is a metaphysical fact. On the one end of the spectrum, it is possible that in all those cases when a physical effect is caused by a mental cause and in turn the mental cause is caused by a physical cause, transitivity fails. In possible worlds where such scenarios are realized, there are at least some physical effects without a sufficient physical cause and therefore the simple closure principle excludes them

34 The transitivity of causation is complicated issue. In recent literature there is a debate concerning the

assumption because from the 1990s onwards philosophers up came with more and more persuasive counterexamples to transitivity. These counterexamples drove philosophers to develop more nuanced counterfactual theories like versions of the contrastive theory in the 2000s. Others working in the interventionist framework reacted to those problems in a different fashion. While contrastivists like Schaffer (2005, 2013) and Maslen (2004) argued that the counterexamples can be explained away, interventionists like Hitchcock (2001) argued for intransitivity. My own investigation in the context of contrastive theories arrived at a result close to Hitchcock’s. As I see the problem causation is intransitive, but there is a way to identify those special cases where causation fails to be transitive and one can provide sufficient conditions to secure transitivity for other kinds of cases.

from the set of physically closed worlds. On the other end of the spectrum, it is possible that in all cases when a physical effect is caused by a mental cause and in turn that mental cause is caused by a physical cause transitivity holds. In possible worlds where this kind of scenario is realized every physical effect has a sufficient physical cause via an intermediate mental cause and therefore the simple closure principle includes them into the set of physically closed worlds even though intuitively, they should be held outside.

So, even if causation is intransitive, there is still good reason to reject the simple closure principle. If we cannot decide what kind of word is our word in this respect, we should assume the worst and according to that option this is a world where every physical effect has a sufficient physical cause, but in many cases the connection is mediated via a non-physical cause.

This is good enough against the simple formulation of closure in itself, even if causation is intransitive. But it would be even better if we could identify and differentiate deviant causal chains, chains where transitivity fails, from normal causal chains where it holds. Fortunately, it is possible to do that. As my investigation in chapter 3 shows, one can provide sufficient conditions for the transitivity of causation to hold in causal chains. The condition is as follows: if the occurrence of an earlier event in a causal chain is not a precondition (or to use more familiar language, a background condition) for the existence of a later causal connection in the same chain then transitivity holds35. It fails in other cases, where earlier

events in a causal chain are preconditions for the existence of a later causal link.

35 In the causation literature such cases are called short-circuit scenarios (see: Hall 2004a, 2004b). The condition

I developed abstracts away from the content of concrete examples and formulates the problem at the level of theoretical language shared by most theories of causation.

Let us translate this to our present case. If mental cause (D) of the final physical effect (E) in a chain only causes the final effect if it was (C) that caused (D) providing the necessary background conditions for (D) to be able to cause (E) at the same time, then we are faced with an intransitive chain where (C) is not a cause of (E). In ordinary examples of mental causation this doesn’t apply. Suppose that my mental pain is caused by a dog bite and that pain causes me to swallow some painkillers. The difference-making ability of my pain to make me swallow painkillers is independent of the occurrence of the dog bite.

If ordinary causal scenarios involving mental causation, conform to my condition for transitivity, and they seemingly do, then most causal chains are transitive and the argument against the simple formulation of closure goes through. However, there is a way out for the defender of closure. The principle can be restricted to immediate causes of physical effects. It is helpful to use a time-based restriction on the definition as immediate causation is hard to define by other means:

Causal closure: Every physical effect at time t has a sufficient physical cause at time

t.

There are other formulations of the principle similar in spirit, but this slightly amended formulation is considered to be more or less canonical. Unless indicated otherwise in the next section on the empirical backing for closure I will assume this version of the principle. But as we will see, certain further modifications will be required in order to incorporate certain empirical facts concerning the physical world.

In document The chances of higher-level causation: an investigation into causal exclusion arguments (Page 84-89)