When I reported to Peter Lupu over Sunday breakfast that Hugh McCann denies that natural causation is existence-conferring, he demanded to know McCann's reasons. He has three. I'll discuss one of them in this post, the third one McCann mentions. (Creation and the Sovereignty of God, p. 18)
The reason is essentially Humean. Rather than quote McCann, I'll put the matter in my own rather more detailed way.
But first I should limn the broader context. McCann's God is not a mere cosmic starter-upper. He keeps the universe in existence moment to moment after its beginning to exist -- assuming it has a beginning -- such that, were God to cease his creative sustenance, the universe would vanish. On such a scheme, God is needed to explain the universe and its continuance in existence even if it always existed. But now suppose natural causation is existence-conferring and the universe always existed. Then the naturalist might argue as follows: (i) the universe is just the sum-total of its states; (ii) each state is caused to exist by earlier states; (iii) there is no first state; ergo (iv) every state has an immanent causal explanation in terms of earlier states; (v) if every state has an explanation of its existence in terms of earlier states, then the universe has an immanent, naturalistic explanation of its existence; ergo, (vi) there is no need for a God to explain why the universe exists, and (vii) if there were a God of McCann's stripe, then the existence of the universe would be causally overdetermined.
The above reasoning rests on the assumption that natural causation is existence-conferring. This is why McCann needs to show that natural causation is not existence-conferring. Here is one reason, a Humean reason.
One monsoon season I observed a lightning bolt hit a palm tree which then exploded into flame. A paradigm case of event causation. Call the one event token Strike and the other Ignition. One would naturally say that Strike caused Ignition. To say such a thing is to refer to the salient cause without denyng the contribution of such necessary causal conditions as the presence of atmospheric oxygen.
But what exactly did I observe? Did I observe, literally observe, an instance of causation? Not clear! What is clear is that that I observed two spatiotemporally contiguous events. I also observed that Strike occurred slightly earlier than Ignition. Thus I observed the temporal precedence of the cause over the effect. But I did not observe the production (the bringing-into-existence) of the effect by the cause. Thus I did not observe the cause conferring existence on the effect. Strike and Ignition were nearby in space and time and Ignition followed Strike. That I literally saw. But I did not literally see any producing or causing-to-exist. What I actually saw was consistent with there being no causal production of the effect by the cause. Admittedly, it was also consistent with there being unobservable causal production.
The point is that conferral of existence by natural causation is not empirically detectable. One cannot see it, or hear it, etc. Nor is there any such instrument as a causation-detector that one could use to detect what one's gross outer senses cannot detect.
Nothing changes if we add the third Humean condition, constant conjunction. Some event sequences are causal and some are not. How do we distinguish the causal from the noncausal? Since we cannot empirically detect existence-conferral, we cannot say that causal event sequences are those that involve existence-conferral. So the Humean invokes constant conjunction: in terms of our example, whenever an event of the Strike-type occurs it is spatiotemporally contiguously followed by an event of the Ignition type. Accordingly, there is nothing more to causation on this empiricist approach than regular succession. A causal event sequence is one that instantiates a regularity. What makes a causal sequence causal is just its instantiation of a regularity. But then, causation is not the bringing into existence of one event by another. The two events are what Hume calls "distinct existences." The events are out there in the world. But the causal link is not out there in the world, or rather, it is not empirically detectable.I hope my friend Peter will agree to at least the following: if we adopt a regularity theory of causation, then natural causation is not existence-conferring. The regularity theory can be stated as follows:
RT. x (directly) causes y =df (i) x and y are spatiotemporally contiguous; (ii) x
occurs earlier than y; (iii) x and y are subsumed under event types X and Y that
are related by the de facto empirical generalization that all events of type X are followed by events of type Y.
If this is what causation is, it is is not existentially productive: the cause does not produce, bring about, bring into existence the effect. On the contrary, the holding of the causal relation presupposes the existence of the cause-event and the effect-event. It follows that causation as understood on (RT) merely orders already existent events and cannot account for the very existence of these events. Since Peter is a B-theorist about time, he should be comfortable with the notion that the universe is a four-dimensional space-time manifold the states or events of which are all tenselessly existent logically in advance of any ordering by whatever the exact relation is that is the causal relation.
Peter should tell me whether he accepts this much.
Of course, the naturalist needn't be a Humean about causation. But then the naturalist ought to tell us what theory of causation he accepts and how it can be pressed into service to explain the very existence of events. My challenge to Peter: describe a theory of natural causation on which the cause event confers existence on the effect event, as opposed to merely ordering already existent events. Nomological and counterfactual theories won't fill the bill (or satisfy the Bill.)
Here is another little puzzle for Peter to ruminate over. Causation is presumably a relation. But a relation cannot obtain unless all its relata exist. So if x directly causes y, and causation is a relation, then both x and y exist. But then x in causing y does not confer existence on y. To the contrary, the obtaining of the causal relation presupposes the logically antecedent existence of y.
This little conundrum works with any theory of causation (regularity, nomological, counterfactual, etc.) so long as it is assumed that causation is a relation and that no relation can hold or obtain unless all its relata exist. For example, suppose you say that x causes y iff had x not occurred, then y would not have occurred. That presupposes the existence of both relata, ergo, etc.
For details and a much more rigorous development, see my article "The Hume-Edwards Objection to the Cosmological Argument," Journal of Philosophical Research, vol. XXII, 1997, pp. 425-443, and the second article below.