In the comments of my post “Indeterminacy, infinity, and participation,” an anonymous reader asks whether Aquinas’s First Way establishes an immovable first mover, or simply a mover who is not moved by the motion of the series they cause. Apparently, the question is motivated by the discussion of Joe Schmid and Daniel Linford in their book Existential Inertia and Classical Theistic Proofs. I have not read this book yet, so I will just answer the question directly rather than commenting on the discussion in the book.

Let me begin by clarifying the question for those unfamiliar. The First Way is the first of the five “Ways” for demonstrating God’s existence in his Summa Theologica. Each of the Ways proceeds from a different phenomenon, each in keeping with Aquinas’s commitment to the claim that we can know God as the first cause of effects that we can experience. As with many things in Thomism, there is debate over how exactly to understand each of the five ways, as well as their relation to one another. Regarding the First Way, the most important debate is around whether it should be understood as a natural philosophical argument or a metaphysical argument. This might seem like an arcane and unnecessarily nit-picky question, but in fact it has wide-ranging implications for how we understand the scope of the argument and its relation to the other Ways.

Natural philosophy has to do with the motion and principles discussed by Aristotle in the Physics. If the argument were of this kind, it would mean that Aquinas has a technical and limited concept of change in mind, consisting of local motion, alteration (change in quality), increase and decrease (change in magnitude), and generation and corruption. Underlying all these sorts of motion is matter. In particular, Aquinas would not have in mind the more general metaphysical concept of motion, which includes things like the motion of angels, and the motion of animal souls. Aquinas calls these acts of perfection, whereas physical motion is an act of imperfection (the imperfection comes from matter). A straightforward implication of this is that if the argument is natural philosophical rather than metaphysical, then the “unmoved mover” it concludes to could be something as mundane as an animal soul, or perhaps an angel. Neither of these are the absolutely immovable mover that God is supposed to be! Those who defend this interpretation of the argument agree: in both the parallel argument in Summa Contra Gentiles (SCG) I.13, as well as Aristotle’s original argument in Physics VIII, this is only the first stage of the argument, which is followed up with a second stage for an eternal immovable mover. In the context of the Five Ways, the Second, Third, or Fifth Ways are understood as completing the First Way, taking us all the way to God.

I have recently been reading Daniel Shield’s book Nature & Nature’s God, in which he defends a natural philosophical interpretation of the First Way. The first stage of the argument, represented by the First Way, concludes that any motion can be traced back, via an essentially ordered series of movers, to an unmoved mover. This unmoved mover may well be God, or, as we said, a mundane unmoved mover. The second stage proceeds as follows. First, we note that either the overall process of change, which is an accidentally ordered causal series extending back into the past, is perpetual (extending infinitely into the past) or had a beginning. For the purpose of this post we will assume that it is perpetual. Aristotle argued for this based on the premise that every change must be preceded by another, and while Aquinas believed that scripture taught otherwise he did not think you could establish this by reason alone.

Second, then, we need to show that even if the process of change is perpetual, it must have a cause. Here we run into the Hume-Edwards objection against “lumping strategies” used in some cosmological arguments. A lumping strategy argues that even if each member in some collection is explained by another member of that collection, still the collection as a whole needs an explanation, and since the whole includes each of the members this explanation must be outside of the collection. This strategy is a central move in contingency cosmological arguments, although it can play a supporting role in other cosmological arguments. Thomistic cosmological arguments tend not to rely on such lumping strategies, but instead focus on the nature of essentially ordered series (as in the De Ente argument, and the First and Second Ways) or leverage some causal principle that more directly establishes a first cause of the relevant kind (as in the supplementary arguments in SCG I.13, and the Fourth and Fifth Ways). Nevertheless, Shields draws attention to the fact that in the primary argument of SCG I.13, as well as Aristotle’s argument in the Physics VIII upon which it is based, the second stage does in fact make use of a lumping argument. Thus, while the Hume-Edwards objection is irrelevant to the first stage (which is covered in the First Way), it must be addressed in the second stage (which is related to the Third Way).

So, what is the objection? It is the recognition that lumping strategies don’t work in general, and so additional reasons need to be given for why we should believe they work in this or that particular case. Edwards gives a now-famous example of five Eskimos standing on a street corner in New York—if we have an explanation of why each is there, then we have an explanation of why all five are there. In our case, if every change in the accidentally ordered series of changes can be traced back to some earlier cause, why should we think that there must still be some overarching cause of the series as a whole?

The answer lies in the distinction between per se and per accidens effects. In Edward’s example, the presence of the five Eskimos at the one spot is a chance, that is, per accidens event. Per accidens effects do not need their own cause above and beyond that of their components. If someone digs a grave and finds a treasure, the discovery is an accidental, that is, chance effect. It needs no cause above and beyond the cause of its components: the finder’s act of digging and the owner’s act of burying. Each of these has a cause: the finder’s desire to make a grave, and the owner’s desire to keep his money safe. If the perpetuity of the accidentally ordered series of causes discussed above were an accidental effect, then it would not need its own cause, as Hume and Edwards and Ockham argue. But per accidens effect cannot be perpetual. What is change or accidental is something that occurs on certain occasions, not all the time. If the continuation of the temporal series of causes and effects was by chance, then at each causal juncture it would be possible for the series to fail. In that case, given enough time it eventually would fail. Once it failed, there could never be any new effects again. Hence if the temporal series of causes and effects is infinite, its perpetuity must be a per se, not a per accidens effect. But per se effects require a per se cause. There must, then, be a cause outside the whole ensemble, keeping the process going. Since the process goes on forever, its cause must be sempiternal. (Daniel Shields, Nature & Nature’s God, p. 78, emphasis original)

Here we see the Thomistic distinction between accidental (per accidens) and essential (per se) being deployed in service of this question, rather than the more familiar question of infinite regress. The argument is that even if we have an accidental series extending infinitely into the past, we would still need a cause of the perpetuity of the series, since this is a distinct effect over and above any accidentally ordered act within it (an argument similar to this is made by Aristotle in Physics VIII.6). Shields goes on to note, along with Aquinas and Aristotle, that the perpetuity could not be caused by any one of the members in the series, for they do not last the full duration of it. Nor could it be caused by all members together, since then they would form an essentially ordered series, which cannot be infinite.

Let me unpack this answer in more detail, in the hopes of making it clearer and more compelling. A per accidens effect is merely the sum of smaller effects, each of which is to some extent caused independently by the members of some collection, while a per se effect is a distinct effect in its own right. This is why a per accidens effect can be explained by simply explaining its parts and how they coincide, while a per se effect requires an explanation that accounts for it specifically rather than merely as an aggregate of other effects.

Now, part of what makes a series accidentally ordered is that later members have some measure of independence from earlier members. I have explained this in terms of act and potency before: the members in an accidentally ordered series contribute the actuality whereby they cause the next. A man begets their son independently of his father, even though he depended on his father in being begotten; Alice depends on Bob to be lifted above the ground, but she independently reaches for and grasps the item on the top shelf. This independence means that even if  a member is successfully caused by its predecessor, it may be interrupted or fail to cause its own effect. The man may choose not to have children, and Alice may be too weak to lift the item. In fact, the same thing may happen in a non-simultaneous essentially ordered series. When Charlie throws a stone that breaks a window, the stone only contributes the potential for being accelerated in the direction of the window: its nature will keep it moving in whatever velocity it has, but cannot determine it to this or that velocity. Thus, this series is essentially ordered, even though the throwing and the breaking are separated by time. And this separation by time introduces once again a sort of independence: something may interrupt the stone before it reaches the window, or the stone might not be moving fast enough and cannot make itself faster.

Thus, in an accidentally ordered series or a non-simultaneous essentially ordered series, the success of later causes has some measure of independence with respect to the success of earlier causes: even if the earlier causes succeed, the later causes may be interrupted or fail. Thus, as such a series continues, it becomes increasingly likely that it is a per se effect. Consider, for example, a series in which each person passes a ball to the next if they so choose. With each step of passing the ball, it becomes increasingly unlikely that each person had made their decision independently. Even if the likelihood of each person passing on the ball is high, say 0.9, the probability of n people making such a decision independently is still 0.9n, which tends to 0 as n increases. In the limit, then, it is far more reasonable to think that there was some plan behind the series that the people agreed to beforehand, making it a per se effect rather than per accidens. The same consideration applies to any accidentally ordered or non-simultaneous essentially ordered series, precisely because of the measure of independence that obtains between its members. Since the series of changes extending back into the past is both accidental and non-simultaneous, we should likewise conclude that its perpetuity is a per se effect. Not only this, but its cause must be simultaneous with it, lest we simply move the problem up a level.

Now, even after this stage we have not necessarily reached the conclusion of a being of pure act, or an immovable mover. We can only infer things about this cause based on the effect we used in concluding its existence. So, for instance, this cause must be everlasting, since it is simultaneous with the perpetual change of the world. It must be immaterial, because material things are part of the perpetual change of the world. It must therefore be incorruptible, because corruption occurs by the indifference of matter to the continued existence of a thing. Where we go from here will determine how far we can get. We may, for instance, go the route of the Third Way, and construct an essentially ordered series of incorruptible beings each causing the incorruptibility of the next, leading to a being which is incorruptible of its very nature. This is more of a metaphysical route, and in taking it we have shifted our focus from motion to something else. If we wish to remain in the realm of natural philosophy, we will need to consider questions about this mover being a self-mover, whether it is accidentally moved, and how it is ordered to its ends. This is the route taken by Shields, following Aquinas (SCG I.13.20–32), who himself was following Aristotle (Physics VIII).

A full explication of the natural philosophical route requires far more space than a short blog post allows, so I will simply recommend Shield’s book for further reading to anyone interested. The aim of this post was simply to draw attention to the existence of these two different interpretations, and illustrate that implications. Although both start out fairly similarly, defending some version of the mover principle (whatever is moving is moved by another), they very quickly diverge. The second stage we discussed here is unnecessary on most (or perhaps all) metaphysical interpretations, and at each stage we must decide once again whether we wish to stick to natural philosophy or switch to metaphysics. Either could work, and will produce philosophical fruit, but the resulting arguments grow ever different.