Recently I've been interested in reading the philosopher John Duns Scotus. I am in general very much sympathetic to the classical philosophical tradition represented by medievals such as Scotus and Aquinas, as well as by Aristotle, Plotinus, and others in the ancient world. In specific I've been considering Scotus' argument for the existence of God, at least the version to which I have access in the collection Philosophical Writings, tr. by Allan Wolter, O.F.M. (Indianapolis: Hackett, 1986).
John Duns Scotus in this work is concerned to prove the existence of a being which exhibits primacy as regards efficient causality, final causality, as well as pre-eminence. In other words, this being he is trying to prove is the cause of the existence of everything else, that for the sake of which everything else exists, and exceeds everything in perfection and value.
Here I want to address Scotus' argument for the existence of a first efficient cause, which he defines as something which exercises efficient causality (i.e., it brings things into existence) entirely of itself, with no assistance from anything else, and could not itself be caused to exist by anything else. We might classify his argument for this conclusion as a modal-cosmological argument, insofar as it has to do both with modality (possibility, necessity, contingency, actuality) as well as cosmology (origins of being, causality, etc.).
Before I begin, let me note that this post will make use of some of the technical terms of Scholastic philosophy (e.g., "substance", "accident", etc.), which do not have the kind of "street" meaning that an ordinary speaker of English would understand upon hearing or reading them. At times I may try to define them in context, but apart from a more sophisticated grasp of ancient and medieval philosophy in general, there is a real chance that you may not understand the argument as offered. That being said, the argument is roughly like this:
(1) Some being can be produced.
I take it here that Scotus is neutral between substantial and accidental being; that is to say, he is affirming the possibility of producing both individual existents (e.g., some horse, some human) as well as aspects of being in a thing (e.g., color of hair, weight, height).
(2) Therefore it can be produced either (i) by nothing at all, (ii) by itself, or else (iii) by something else.
Scotus thinks the first two options are non-starters: nothing can come from nothing whatsoever, and equally obviously a thing could not bring itself into existence; in that case it would preexist its own existence, which is absurd. So (iii) must be affirmed. Let's call this other thing which can produce being A.
(3) A is either a first cause in the sense defined, or it is not.
If it is, then the desired conclusion is granted. If it isn't, however, then A either (i) is not actual, but could be actualized by something else, B, or else (ii) is actual, but can only actualize being through the assistance of another actual entity, B. But now the same questions can be asked of B: either it is first or it is not. If it is not, then it depends on something else, C, in order to produce being.
But Scotus posit that
(4) An infinite regress in this case is not possible.
(5) There is a first efficient cause, which ultimately grounds the possibility of the production of being because it is capable of producing being of its own, and it is not nor can it be actualized itself by anything else.
The critical objection at this juncture is to (4): why not allow an infinite regress? Scotus offers very elaborate argumentation at this point to prove that this is not possible. He first posits a distinction between two sorts of causal sequences or chains.
On the one hand, we have per se or essentially ordered causal series, in which a 'later' member of the series only acts by virtue of the continued activity upon it by a 'previous' member. One example would be a train pulling a long chain of cabooses: the various train cars only move forward because of the continued activity of the first car upon them (namely, by pulling). Another example would be this: a stone leaving an impression in sand as you push it along with a stick. In this case, the stone only leaves the impression in the sand insofar as the stick acts upon it, but the stick only exhorts force on the stone insofar as your hand pushes it, and your hand only insofar as you will to move it, etc. In such causal series, to summarize, the later members of the series only act in the relevant ways insofar as they are given the power to do so by previous members.
On the other hand, there are per accidens or accidentally ordered causal series. In these cases, a later member of the series exhibits the salient causal power independently of the continued activity of a previous member. For example, consider Paul, who has a son Peter, who in turn has his own son, James. Peter can exercise the causality involved in bringing James into existence independently of Paul, who by the time Peter had a son may have been long dead. In these causal chains, the activity of a later member does not depend on the activity of a former member.
Now Scotus argues that however we understand the causal series linking possible being to A, and A to B, etc., in the above argument, it cannot be an infinite chain.
No essentially ordered causal chain can itself be infinite, insofar as it involves derived power, whereas the derivation of power presupposes the ultimately independent existence of power. Consider this from another angle. Suppose James borrows a book from Peter, who borrowed it from Paul. Clearly the borrowing of a book presupposes the independent existence of a book, since if the book did not exist on its own independently of the chain of borrowing-and-lending, it could not be borrowed or lent in the first place. In other words, there is a critical presupposition of the infinite chain of lending and borrowing -- namely the existence of the book in the first place -- which the chain itself cannot account for; consequently the chain cannot be the whole story, it cannot be all that there is. Now suppose there were an infinitely long essentially ordered causal series. Insofar as each member of the chain does not exhibit the relevant causal powers on its own, the chain itself would need to be actualized by something outside of it, which cannot itself be a part of the chain and has the power to actualize and confer power to other beings of its own.
(Scotus gives further arguments as well. For instance, in per se causal series, as I've noted, all the relevant causal activity is simultaneous. Thus if there were an infinite causal series of this sort, there would be an infinity of beings simultaneously producing some effect -- something which no one posits or accepts.)
Furthermore, Scotus argues that even if there were an infinitely long accidentally ordered causal series, it would in turn depend on an ontologically prior essentially ordered causal series, which as shown above could not be infinite. Consider the example of Paul fathering Peter. Paul only can exhibit this causal power by virtue of his form as a healthy human male: in other words, it is only because of the continued proper-arrangement of the material stuff from which he is made that he can bring a child into the world; take away that arrangement (say, by cutting up his body, removing some critical parts, or aging and deforming it sufficiently) and he can no longer father any child. Therefore this accidentally ordered causal series, even if it did exist, would be ontologically posterior to an essentially ordered causal series, only by virtue of which the former is even possible, and this could not be infinite. An accidentally ordered causal series is only possible by virtue of a prior essentially ordered causal series, by which the members of the per accidens series exist in the first place.
Thus however you understand the causal chain in which some possible being, A, and B are involved, there must be a first efficient cause, which can bring something else into being entirely by itself and which is not actualized by anything else.