Davidsonian Event Causation and the Thomistic First Cause: What It Clarifies and What It Cannot Rescue — Epoche C1
A hand moves a stick, the stick moves a stone. The example is Aristotle's, from the Physics (VIII.5, 256a), and it is the standard illustration of the kind of causal series that Thomas Aquinas's Second Way — the argument from efficient causation in the Summa Theologiae (Ia, q.2, a.3) — claims cannot run backwards without end. Everything in that series happens at once: the stone is moving now because the stick is moving now because the hand is moving now, and if the hand stops, the whole arrangement stops together. The proposal examined here is that Donald Davidson's account of causation — on which causes and effects are events , dated unrepeatable particulars, and causal relations hold between them — supplies the modern vocabulary in which this ancient argument can be stated and defended. The proposal is partly right. Davidson does supply one clarification that removes a common objection, and it is a clarification the original version of this essay was reaching for without quite getting hold of. But Davidson's framework also carries two commitments that work directly against the Thomistic conclusion, and one move the original essay made is a category mistake within the very framework it invokes. The verdict is that Davidson clarifies the argument without rescuing it, and the title of this piece has been altered to say so. What the Second Way argues The argument in Summa Theologiae Ia, q.2, a.3 has four steps, and each is worth having in front of us because the objections attach to different ones. In the observable world we find an order of efficient causes — causes that bring about an effect by acting, as distinct from the material something is made of or the end for which it is done. Nothing is the efficient cause of itself, since it would then have to exist prior to itself. In an ordered series of efficient causes, the first is the cause of the intermediate and the intermediate of the last; remove the cause and you remove the effect. So if there is no first cause, there is no intermediate and no last. But there manifestly are last effects. Therefore there is a first efficient cause. Step 3 carries the whole weight, and it is not the innocuous claim it appears to be. Read as a claim about any chain whatever, it is plainly false: my existence does not require that my great-grandfather still exist. Read as a claim about a particular sort of chain, it becomes defensible. The distinction is the one the original essay named without unpacking, and unpacking it is the first task. Two kinds of series, and why Aquinas permits an infinite regress in one of them A series ordered per accidens — incidentally — is one in which each member's causal power is its own, and prior members are needed only to have produced it. Fathers and sons are the standard case: a man begets a son by a power he possesses in himself, and his own father may be long dead. Remove the grandfather now and nothing follows about whether a grandson is conceived tomorrow. A series ordered per se — in its own right — is one in which the intermediate members have no causal power of their own at that moment, but exercise a power that is being conferred on them concurrently. The stick moves the stone only in so far as the hand is moving the stick at that very instant; take the hand away and the stick does not carry on moving the stone with a diminished force, it does nothing at all. The members of such a series are instrumental in a strict sense: they transmit power without originating any. Aquinas's argument concerns the second sort only, and he says so about the first with a directness that is regularly overlooked. He holds that an infinite regress of efficient causes ordered incidentally is not impossible, and, more strikingly, that reason cannot demonstrate that the universe had a temporal beginning at all. His short treatise De aeternitate mundi (c. 1271) argues precisely this: that a created universe without a first moment involves no contradiction, and that its temporal finitude is known by revelation rather than proved by argument. So Aquinas is committed to the view that his own First Cause argument must go through even if the past is infinite. Whatever the Second Way is, it is not an argument that the universe must have started. This distinguishes it sharply from the kalām cosmological argument, revived by William Lane Craig in The Kalām Cosmological Argument (1979), which does turn on the impossibility of an actually infinite past and therefore does depend on temporal firstness. Edward Feser's Five Proofs of the Existence of God (2017) presses the same separation from the other side, insisting that the Aristotelian and Thomistic arguments concern what sustains things here and now rather than what started them. The objections that a beginningless universe or a probabilistic physics would defeat the argument are aimed at the kalām target and pass over this one. A probabilistic transition, on the Thomistic scheme, is still the actualising of a potential and still requires something already actual to do the actualising; indeterminism changes what the actualiser makes likely, not whether one is needed. What Davidson actually holds Three theses from Donald Davidson's papers, collected in Essays on Actions and Events (1980), bear on this. The first, from "The Logical Form of Action Sentences" (1967), is that action and event sentences should be analysed as quantifying over events. "The hand moved the stick violently" becomes $$\exists e \,\bigl(\mathrm{Moving}(e) \wedge \mathrm{Agent}(e, h) \wedge \mathrm{Patient}(e, s) \wedge \mathrm{Violent}(e)\bigr).$$ Davidson's argument for this is not aesthetic. It is that the inference from "moved it violently" to "moved it" is valid, and that on the event analysis its validity is a matter of logical form — dropping a conjunct — whereas on an analysis treating "moved violently" as an unanalysed two-place predicate the inference has to be underwritten by a separate meaning postulate for ever