The Aristotelian proof

Aristotle developed an argument from motion to the existence of a prime unmoved mover in book 8 of his Physics and book 12 of his Metaphysics. Later Aristotelians developed their own variations on this argument, the most famous being Aquinas's First Way. While the details of my own version of the argument are different from those of Aristotle's and Aquinas's versions, it is very much in the spirit of those earlier arguments. For that reason I call it an ‘Aristotelian proof’ of the existence of God (Feser (Reference Feser2017), ch. 1).

The argument proceeds in two stages, the first of which aims to establish the existence of what I call a purely actual actualizer, and the second of which aims to show that this purely actual actualizer must have the key divine attributes. I first expound the argument in an informal and discursive way, and then summarize the main thrust of the reasoning in a more formal step-by-step restatement. Oppy focuses on stage one of this restatement, which I will quote here to provide context:

1. 1. Change is a real feature of the world.

2. 2. But change is the actualization of a potential.

3. 3. So, the actualization of potential is a real feature of the world.

4. 4. No potential can be actualized unless something already actual actualizes it (the principle of causality).

5. 5. So any change is caused by something already actual.

6. 6. The occurrence of any change C presupposes some thing or substance S which changes.

7. 7. The existence of S at any given moment itself presupposes the concurrent actualization of S's potential for existence.

8. 8. So, any substance S has at any moment some actualizer A of its existence.

9. 9. A's own existence at the moment it actualizes S itself presupposes either (a) the concurrent actualization of its own potential for existence or (b) A's being purely actual.

10. 10. If A's existence at the moment it actualizes S presupposes the concurrent actualization of its own potential for existence, then there exists a regress of concurrent actualizers that is either infinite or terminates in a purely actual actualizer.

11. 11. But such a regress of concurrent actualizers would constitute a hierarchical causal series, and such a series cannot regress infinitely.

12. 12. So, either A itself is a purely actual actualizer or there is a purely actual actualizer which terminates the regress that begins with the actualization of A.

13. 13. So, the occurrence of C and thus the existence of S at any given moment presupposes the existence of a purely actual actualizer.

14. 14. So, there is a purely actual actualizer. (Feser (Reference Feser2017), 35–36)

The principle of causality

Oppy's criticisms are directed at steps 4 and 7. He tells us that he rejects step 4 (which is my preferred formulation of the principle of causality) for reasons illustrated by the following examples (Oppy (Reference Oppy2021), 494). Consider first a red chair which, at time t, has the potential to exist and to be red at t + Ɛ (where Ɛ is some short time interval such as a millionth of a second). There is in Oppy's view no reason to suppose that something must actualize this potential in order for it to be actualized at t + Ɛ. Rather, all that is required for the actualization to occur is that there be nothing that keeps it from happening. He writes: ‘Potentials to remain unchanged do not require distinct actualizers; all they require is the absence of any preventers of the actualization of those potentials.’ Oppy also claims that ‘there are even cases where potentials to change need not require distinct actualizers’. His example involves the universe expanding at some constant rate R. Suppose that at t, the universe has the potential to be expanding at rate R at t + Ɛ, and thus to have a larger volume at that later time. In order for this potential to be actualized, all that is necessary is that nothing intervenes to change the rate of expansion.

There are several problems with this objection. The first is that Oppy begs the question insofar as his objection amounts merely to asserting, without argument, that premise 4 is false. In particular, the scenarios involving the red chair and the expansion of the universe do not amount to falsifying counterexamples to premise 4. Rather, Oppy merely describes the scenarios in a way that presupposes the falsity of premise 4. Someone who accepts premise 4 would describe these situations differently, as we will see presently.

That would be bad enough if I had myself merely asserted, without argument, that premise 4 is true. Oppy would in that case have given us at most a stalemate between two competing undefended claims. However, I did not merely assert that premise 4 is true. I defended it at some length, both by responding to various objections and by offering positive considerations in support of it (Feser (Reference Feser2017), 19, 39–57, and 262–265). For example, I rebut objections to the principle of causality raised by Hume and the early Russell and objections that appeal to scientific ideas such as Newton's law of inertia and quantum mechanics. I point out that while a potential is not nothing, since it is also not an actuality it is difficult to see how it could do anything, including actualize itself. I point out that if potentials could be actualized without something to actualize them, we should expect this to be occurring constantly, and yet it is not. We should also not expect to be able very often to identify something which actualized a given potential, and yet we are in fact typically able to identify some actualizer. I have also argued that for a potential to be actualized without something to actualize it would violate the principle of sufficient reason (PSR), and I have defended that principle at length.Footnote ²

So, in order even to stalemate my premise 4, Oppy would need to rebut my arguments for it; and positively to undermine premise 4, he would, in addition, have to provide some argument against it. To describe the red chair and expanding universe examples in a way that simply presupposes the falsity of premise 4 is not sufficient to accomplish either task.Footnote ³

A second problem concerns a muddle in Oppy's characterization of his first example. Again, Oppy takes that example to illustrate the thesis that ‘potentials to remain unchanged do not require distinct actualizers; all they require is the absence of any preventers of the actualization of those potentials’. Oppy's phrase ‘potential to remain unchanged’ is a curious one, at least given the Aristotelian analysis of change as entailing the actualization of a potential – an analysis that Oppy does not challenge. For if change involves the actualization of a potential, then a ‘potential to remain unchanged’ amounts to a potential not to have a potential actualized. It is a kind of second-order potential.

For clarity's sake, let us give the label P to some particular first-order potential (such as the red chair's potential to become green), and P* to the second-order potential not to have P actualized (which is what the chair's ‘potential to remain unchanged’ with respect to its colour really amounts to). Now, Oppy goes on to speak of a ‘preventer’ of the actualization of P*. But to prevent P* is just to make it the case that P is actualized after all. So, a ‘preventer’ of the actualization of P* is really just the same thing as an actualizer of P. In which case, Oppy's claim that ‘potentials to remain unchanged . . . require [only] the absence of any preventers of [their] actualization’ amounts to a convoluted way of saying that P will not be actualized unless something actualizes it. And that is hardly incompatible with my premise 4!

Now, if that is what the thesis amounts to, then the red chair example Oppy uses in order to illustrate the thesis really just amounts to a scenario in which the chair will not at t + Ɛ turn green (or any other colour different from red) unless something actualizes its potential to be green (or some other colour different from red). (Oppy himself gives as an example a case where ‘the chair is sprayed with green paint’.) But that supports, rather than conflicts with, my premise 4! So, Oppy's claim about the colour of the chair at t + Ɛ seems to me to be incoherent as well as question-begging. Not only does it not do its intended job of undermining premise 4, on closer inspection it actually does precisely the opposite.

It is not surprising that the example would turn out to be problematic in this way. The Aristotelian proof is, after all, an argument from change, not from the absence of change. The case of a chair staying red from t to t + Ɛ is, accordingly, not even prima facie a promising candidate for a counterexample to premise 4. Spelling out the example in a way that would make it even relevant to the argument was bound to expose it as a misfire.Footnote ⁴

Oppy's remarks about the existence of the chair at t + Ɛ (as opposed to its colour) have a different problem, which is that they are doubly question-begging. They are question-begging, first, in the respect I have already described. But they are question-begging in a further respect insofar as Oppy ignores an argument I gave for why the sheer existence of a thing at a time requires an actualizing cause at that time. In other words, I not only gave reasons in support of premise 4 as a general principle, but I also gave reasons to believe that the specific case of the actualization of a thing's existence at any moment really does fall under that general principle; and just as Oppy fails to respond to the former, he fails to respond to the latter. He assumes the arguments are wrong, but does not show that they are.

The example I used to illustrate my claims was the existence of a cup of water, though the same points would apply to a chair or any other material object (Feser (Reference Feser2017), 26–29). I noted that there are several ways in which we might think of the water's existence at some moment t + Ɛ (to stick with Oppy's timestamp) as dependent on the actualization of potentials possessed by its parts at t + Ɛ. For example, we could think of the water as an aggregate of particles which potentially could constitute any number of other things at t + Ɛ, such as distinct quantities of hydrogen and oxygen, but instead constitute water. Or we could adopt an Aristotelian hylemorphist analysis of the water as a compound of substantial form and prime matter. I argued that, for the specific purposes of the Aristotelian proof, it doesn't matter how the details are worked out. What matters is just that there are parts which considered by themselves at t + Ɛ could potentially make up something other than water at t + Ɛ but actually make up water at t + Ɛ, and that this requires some actualizing cause. I also argued that in identifying this cause, it would not suffice to appeal merely to things that happened at some time t prior to t + Ɛ. It would have to be a cause operative at t + Ɛ.

I refer the interested reader to my book for the details of this line of argument. It suffices for present purposes to note that Oppy's objection ignores, rather than answers, the argument. Now, his failure to engage with it undermines the force of his example of the expansion of the universe no less than the force of the chair example. It does so in two ways. First, as I develop the Aristotelian proof, although the notion of the actualization of a potential is introduced by way of examples involving change over time, its crucial application is to the existence of a thing at a time. The idea is that while the change that occurs from t to t + Ɛ requires an actualizer, the metaphysically deeper point is that the sheer existence of the things that change, at t or at t + Ɛ, also requires an actualizer. This is a common theme in Thomism, though it is usually associated with what in a later chapter of my book I call ‘the Thomistic proof’ (an argument which begins with an appeal to the Thomistic distinction between essence and existence) rather than with the Aristotelian proof. My claim is that the Aristotelian proof affords a distinct avenue to the same conclusion.

One problem with Oppy's example of the universe expanding at rate R from t to t + Ɛ, then, is that even if this could occur without an actualizer (which he has not in fact shown, but merely asserted), this would not answer the metaphysically deeper question about what makes it the case that the universe exists at all at t or at t + Ɛ. But that sort of question is the focus of the Aristotelian proof.

Second, we have to ask why the universe expands at rate R, specifically. Now, one answer might be that this reflects some capacity or causal power of the universe, considered as a single entity. Another answer might be that it reflects the aggregate effect of the operations of the capacities or causal powers of all the things that are in the universe, considered as discrete entities. A third might be that it reflects the laws that govern the universe. No doubt there are other possible answers. But from the point of view of the Aristotelian proof, they will all ultimately entail the actualization of potential at a time, in just the way that (as I noted above) the argument holds that the sheer existence of a physical object like a cup of water at a time entails the actualization of potential at that time.

Hence, suppose, for example, we say that the reason the universe expands at rate R from t to t + Ɛ is that it is governed at t by some law L. I would say that, just as the particles that make up water have at t the potential either to constitute water at t or to constitute discrete quantities of hydrogen and oxygen at t, so too the universe at t has the potential either to be governed by L at t or to be governed instead at t by alternative laws M, N, O, etc. And just as there needs to be an actualizer at t to account for why the particles in fact constitute water at t rather than something else, so too there needs to be an actualizer at t to account for why the universe is governed by L at t rather than by some other law. Or suppose we say that the reason the universe expands at rate R from t to t + Ɛ is that this is a consequence of the operation at t of some capacity or causal power C. I would say that in that case the universe has at t the potential either to possess C or some alternative capacity or causal power D, and that there needs to be an actualizer at t in order to account for why what it actually possesses then is C rather than D.

Oppy would, no doubt, disagree with all of this. The point, however, is that he does not consider, much less answer, this potential response to his example, even though I made similar points in my book.Footnote ⁵

Existential inertia

Let me turn to an objection that may by now have occurred to some readers. Aquinas's way of formulating the argument from motion in his First Way appeals to the premise that nothing can be both potential and actual in the same respect at the same time. But doesn't what I have been saying conflict with that? For example, am I not saying that the particles that make up the water at t are both potentially water and actually water at the same time?

Oppy seems to be raising an objection similar to this in his criticism of my premise 7. Recall that that premise states that the existence of a substance S at any given moment presupposes the concurrent actualization of S's potential for existence. Oppy responds:

So, Oppy supposes that I am assuming that S both actually exists at a time t and potentially exists at that same time t, and objects to such an assumption in the way that Aquinas would.

But I am making no such assumption, and there is no conflict between what Aquinas says and what I am saying. Consider the protons, neutrons, and electrons within a particular cup of water. Considered qua constituents of the water, they are of course actually water. But just considered qua protons, neutrons, and electrons, and in abstraction from the water they make up, they are potentially either water or any number of other things. So, they are not both potentially water and actually water in the same respect. So, there is no conflict with Aquinas's premise. Neither am I claiming that the water, or any other substance, both actually exists at t and potentially exists at t. What I am saying is rather that the protons, neutrons, and electrons considered just as such are potentially water, but also happen to constitute water rather than some other kind of thing, so that there must be something additional to what is true of them just qua protons, neutrons, and electrons that accounts for the water's existence.Footnote ⁶

But couldn't there be some further factor F that is intrinsic to the water, rather than an extrinsic cause, that accounts for the protons, neutrons, and electrons constituting water at t rather than something else? (For example, F might be the configuration of these particles.) The problem with this suggestion, as I point out in the book, is that it just kicks the problem back a stage (Feser (Reference Feser2017), 26 and 83–84). For now we need to ask what it is that makes it the case that the protons, neutrons, and elections exist together with F at t rather than with some alternative factor G, H, or whatever.

Oppy's position is that a substance S's continued existence from t to t + Ɛ requires only the absence of anything that would prevent it from continuing to exist. He endorses the thesis that material things have what is sometimes called ‘existential inertia’. Now, in my book I explicitly criticize this thesis (ibid., 233; cf. also Feser (Reference Feser2011) ). One problem with it, I maintain, is that it is ungrounded. Its proponents do not tell us exactly what it is about material things, and contingent things in general, that would give them such a remarkable property. It is merely asserted that they have it. Another problem is that there is positive reason to judge that they cannot have it, namely that they are composite in various respects. For example, material objects are composed of physical parts and, Aristotelians claim, at a more fundamental level they are composites of substantial form and prime matter. Contingent things in general are, Thomists maintain, compounds of essence and existence. But anything that is composite in any way requires a cause to maintain it in existence, or so Thomists and others argue. Indeed, a further argument I defend in my book, the Neo-Platonic proof, is devoted precisely to showing that whatever is composed of parts requires, at any moment that it exists, a sustaining cause. Needless to say, what requires such a cause does not have existential inertia.

Oppy's only response is to assert, of the chair in his example, that ‘not only does the chair have “existential inertia”, its parts have “configurational inertia” ’ (Oppy (Reference Oppy2021), 501). That is to say, the parts will remain together in the absence of anything that prevents them from doing so, without the need of something positively to cause them to stay together. But this response merely pushes the problem back a stage, because it is subject to exactly the same objections raised against the original ‘existential inertia’ thesis. For one thing, the ‘configurational inertia’ thesis too is simply asserted, without argument. For another, Oppy fails to see that objections like those I raise against the existential inertia thesis apply to the configurational inertia thesis as well.

Indeed, in my book I essentially responded to precisely the configurational inertia thesis (albeit not under that name) in the course of defending the Neo-Platonic proof (Feser (Reference Feser2017), 83–85). I noted that a critic of the proof might propose several ways in which the parts of a composite thing could remain together without an external cause, and I argued that none of these proposals works. First, the critic might hold that the parts of a thing A and B are held together by some further part C. The problem with this suggestion is that it just kicks the problem back a stage, since we now need to know what keeps A, B, and C together. And to posit some further part D to account for this would lead to a vicious regress.

Alternatively, the critic might say that it is just an irreducible fact about A and B that they stay together. But what does such a claim amount to? One interpretation would be that the thing made up of A and B is itself the cause of A and B staying together. The problem with this interpretation is that it leaves us with an explanatory vicious circle. It makes the whole thing out to be both the cause of its parts A and B being unified and the effect of its parts A and B being unified. An alternative interpretation would be that it is just a brute fact that A and B stay together. The problem with this interpretation is that the critic was supposed to be giving us an alternative explanation to the claim that a composite thing requires an external sustaining cause. And to say that it is simply a brute fact that A and B stay together is not to offer an alternative explanation, but rather to offer no explanation at all.

A third proposal would be to posit some law of nature that accounts for A and B staying together. But this too simply pushes the problem back a stage, for A, B, and the purported law would themselves amount to a kind of composite. The proposal just raises the question why A and B are governed by that law rather than by some alternative law or no law at all.

Similar problems would afflict any attempt to apply analogous proposals to a more direct defence of the existential inertia thesis, i.e. a defence that does not appeal to a second ‘configurational inertia’ thesis. For example, if it were proposed that it is simply a law of nature that things have existential inertia, the response would be that this just leaves us with the question of why things are governed by that law rather than some other law or no law at all. If it were proposed that it is simply a brute fact that things have existential inertia, the reply would be that this leaves us with no explanation at all, rather than with an alternative explanation to the claim that composite things have a sustaining cause.

Vicious circularity also threatens the existential inertia thesis. Existential inertia would be an attribute of the things that have it. But attributes are metaphysically dependent on the substances that have them. Hence we would have a scenario in which a substance depends for its continued existence on its attribute of existential inertia, and its attribute of existential inertia in turn depends for its continued existence on the substance which has it.Footnote ⁷

Oppy's position, then, once again begs the question. And once again, this is not merely a matter of his meeting one unsupported assertion with another, which would be bad enough. For my own assertions were not unsupported. It's not that Oppy merely asserts that material things have existential inertia and I merely assert that they do not have it, which would leave us with a stalemate. It is that Oppy merely asserts that material things have existential inertia even though I have given arguments that purport to show that they cannot have it, and Oppy has not rebutted those arguments.

In general, Oppy seems to me to be much too quick to brush aside concerns about begging the question. When replying to the referee referred to earlier, he writes:

Surely Oppy would not deny that begging the question is still a fallacy that logic teachers do and ought to teach their students to avoid. So, I Imagine that his point is simply that the charge of begging the question is often flung too quickly. He is probably right about that. But of course, that doesn't mean that the charge is never justified. When one philosopher holds that p and the other holds that not-p, but neither has offered arguments for his thesis and each is concerned merely to clarify it and work out its implications, it is fair to suppose that neither one is necessarily guilty of begging the question against the other. However, when one philosopher offers arguments for p and criticisms of the arguments against p, and the other philosopher, responding to the first, merely keeps asserting that not-p without substantively addressing the arguments and criticisms, it is reasonable to accuse him of begging the question.Footnote ⁸

Hierarchical causal series

Though his focus is on premises 4 and 7 of my argument, Oppy also makes some brief remarks about step 11, which claims that a regress of concurrently operating actualizers of potentials would constitute a hierarchical causal series, and that such a series cannot regress infinitely.

As I explain in the book, a hierarchical series of causes is one in which the members of the series other than the first have their causal power in only a derivative or borrowed way. A stock example would be a man using a stick to push a stone. The stick has the power to move the stone, but only insofar as the man uses it as an instrument for doing so. Left to itself, the stick would just lie there and move nothing. The man, by contrast, has causal power in a built-in or underived way. No one has to use him as an instrument in order for him to be able to move the stick. The first cause in such a series is ‘first’ in the sense of having causal power in this primary or underived way, as contrasted with the secondary or derivative causal power of the other members of the series. Accordingly, ‘first’ in this context does not entail temporal priority. In fact, in a hierarchical series the secondary causes operate simultaneously with the primary cause. For example, the stick's moving of the stone is simultaneous with the hand's moving of the stick.

By contrast, in a linear series of causes, the members are temporally separated and have independent causal power. A stock example would be a father begetting a son, who in turn has a son of his own. Whereas the stick cannot move the stone in the absence of the man who uses it, the son can beget a son of his own in the absence of his father. Father and son are not related as primary and secondary cause, the way the man and the stick are. Because the later members of a linear series do not depend on the earlier members in the way the secondary members of a hierarchical series depend on its primary member, a linear series need not trace back to a ‘first’ cause in the relevant sense. (Cf. Feser (Reference Feser2014), 146–154.)

Thomistic cosmological arguments, such as the arguments I defend in my book and Aquinas's Five Ways, typically appeal to hierarchical causal series rather than linear causal series. Hence, unlike the kalām cosmological argument, they are not concerned with tracing the history of the universe back to a temporally first cause operating at the Big Bang. Rather, they aim to show that the secondary causes operating here and now presuppose a divine primary cause operating here and now. They take God to be cause of the world in the sense that he conserves the world in existence at every moment.

Now, it is important to understand that the examples commonly used in introducing the distinction between hierarchical and linear causal series have no significance other than serving as rough but easily understood illustrations of the general principles. When Thomists appeal to the example of a man using a stick to move a stone, their aim is simply to introduce the general idea of a secondary cause which is dependent for its efficacy on a simultaneously operating primary cause. They are not interested in sticks or stones per se, and nothing at all rides on the physics of the situation. When a scientist who is trying to elucidate the concept of a black hole asks you to consider what you would see if you fell into one, it would be pedantic, and entirely miss the point, to object that someone would be ripped apart by gravity before making it into the black hole and thus would not be around to see anything. Similarly, when a Thomist appeals to examples like the stick moving the stone, it would be pedantic, and miss the point, to raise questions about how the example would be described by a scientist interested in the physics of motion.

In my book, I make a point of warning against such pedantic misunderstanding (Feser (Reference Feser2017), 57–60). Unfortunately, Oppy nevertheless falls into it. In my exposition of the Aristotelian proof, I introduced the idea of a hierarchical causal series by way of the example of a cup which is held up by a table, which is in turn held up by the earth. The earth, I noted, is not itself held up by anything else in the same way, and thus illustrates the idea of a primary cause operating through secondary causes. Oppy points out that from the point of view of physics, the earth and the cup simply differ in comparative mass, and that in interstellar space the cup would attract smaller objects to itself in just the way that the earth attracts the cup to itself (Oppy (Reference Oppy2021), 497). From this he concludes that the earth does not after all have more in the way of underived causal power than the cup does. Oppy is of course correct about this, but only in the way that the critic who points out that you would be ripped apart by gravity before making it into a black hole is correct. The point is pedantic and simply irrelevant to the substance of the Aristotelian proof or of any other Thomistic cosmological argument.

Here is the reason it is irrelevant. In all of the versions of the cosmological argument I defend in my book, what is at issue is the question of what explains the sheer existence of the things of our experience at any moment. In each case, the reason this is said to be something in need of explanation has to do with some completely general feature of these things. In the Aristotelian proof, it has to do with the idea that their existence requires the actualization of potential; in the Neo-Platonic proof, it has to do with the fact that they are composed of parts; in the Thomistic proof, it has to do with each having an essence distinct from its existence; and in the rationalist proof, it has to do with the fact that each is contingent. The claim is that anything that has potentiality, or is composed of parts, or has an essence distinct from its existence, or is contingent, has the specific kind of dependency on a cause that is characteristic of a hierarchical causal series.

Whatever one might say about such arguments, the empirical and scientific details about the physics of sticks and stones, the difference in mass between a cup and the earth, and so on, fall away as entirely irrelevant. The purpose of such examples is merely to serve as simple ways of getting across the basic idea behind some unfamiliar general metaphysical principles. It is the metaphysical principles themselves, and not the illustrative examples, that are doing the actual work in the arguments. Again, this is the sort of thing that is done all the time in expositions of scientific ideas, as when we are asked to think of an atom as like a miniature solar system or a molecule as like balls held together with sticks. Even though these models are not quite correct and in some respects even misleading, they are adequate to the purposes for which they are typically deployed, and it would be pedantic to quibble about them.Footnote ⁹

In my book, I also argue that hierarchical causal series are metaphysically more fundamental than linear causal series, and perhaps the reason is obvious from what I have just said. In order for something to serve as a member of a linear series that extends through time, it has to exist at a time. For example, for a father to act at time t in a way that will make it the case that his son exists at some later time t + Ɛ, the father has to exist at t. So if, as I argue, his existence at t in turn depends on a hierarchical causal series at t, and the son's existence at t + Ɛ depends on a hierarchical causal series at t + Ɛ, and so on, then this linear causal series will presuppose the hierarchical causal series in question.

Oppy tells us that he rejects the claim that linear causal series presuppose hierarchical series (Oppy (Reference Oppy2021), 497). But there are two problems with the reasons he gives. First, he repeats his claim that a thing will continue to exist from t to t + Ɛ as long as nothing prevents it from doing so, and I have already explained how that claim begs the question. Second, in order to illustrate the thesis that linear series needn't depend on hierarchical series, he suggests a way in which the example of a cup resting on a table which rests on the earth might be analysed into a linear series of causes rather than a hierarchical series. But in doing so he fails to see that the fundamental way in which linear series depend on hierarchical series has to do with the sheer existence of a thing at any moment, and details about the causal relations between tables and cups over time are irrelevant to this.

Aquinas's Five Ways

In his book Naturalism and Religion, Oppy makes some cursory critical remarks about Aquinas's Five Ways (Oppy (Reference Oppy2018), 109–113 and 117–121). The remarks are so cursory that it seems to me that Oppy cannot reasonably be said to have raised any serious difficulties for the arguments, much less refuted them. To be fair to Oppy, he does not claim to have done so, and has only the more modest aim of rebutting any suggestion that Aquinas's arguments refute naturalism. But in my estimation, his treatment is too superficial even to do that much.

To understand the objection I want to raise against Oppy's approach, consider the following analogy. Suppose a critic of Wittgenstein's Private Language Argument pulled some passages out of the Philosophical Investigations beginning at §243, paraphrased key sentences from these passages, numbered the paraphrased sentences and arranged them as premises and conclusion, and then noted that, so arranged, we weren't left with an argument that was valid or even complete. Suppose this critic said nothing about the way that our interpretation of certain key terms (such as ‘private’ and ‘inner’) might affect our understanding and estimation of the argument, or about the larger philosophical context of the argument, such as the views against which Wittgenstein was reacting. Suppose the critic also completely ignored what the many commentators on Wittgenstein's argument have had to say about it, including the ways they would propose reconstructing the argument so that it could be seen to be interesting and powerful. Suppose instead that he merely noted that he couldn't find other passages from Wittgenstein's corpus from which he could pull sentences that he could paraphrase, number, and add to the list so as to yield a valid or complete argument. Suppose he concluded from this procedure that Wittgenstein has failed to produce a decisive refutation of the possibility of a private language.

I think it fair to say that any philosopher would regard this as an obviously flat-footed and unhelpful way to approach Wittgenstein's argument, or any other philosophical argument. Yet it is precisely the way Oppy approaches Aquinas's Five Ways in Naturalism and Religion. Key sentences are pulled from the relevant passages in the Summa Theologiae, paraphrased, numbered, and arranged as premises and conclusion. The proposed arrangements are declared either invalid or at least incomplete as they stand. We are told that giving the same treatment to related arguments from Aquinas's other works yields similar results. And that is pretty much it. There is no consideration of how a Scholastic Aristotelian like Aquinas understood notions crucial to a proper interpretation and evaluation of the arguments (motion, cause, essentially ordered series, contingency, etc.) in a way that differs from how contemporary philosophers and other modern readers are liable to understand them. Indeed, there is no discussion of these key concepts at all. There is no attempt to do what analytic philosophers routinely do when evaluating arguments, viz. to consider ways the arguments could be interpreted or reconstructed, so as to see whether there might be more to them than a superficial reading would indicate.

Most oddly, there is no consideration at all of the work of contemporary philosophers who have in fact done that. For example, several contemporary analytical Thomists, such as John Haldane (Smart & Haldane (Reference Smart and Haldane2003) ), Christopher Martin (Reference Martin1997), and David Oderberg (Reference Oderberg, Cottingham and Hacker2010), have proposed ways of reading one or more of the Five Ways that show them to be challenging arguments. I have done the same (Feser (Reference Feser2009), ch. 3). Oppy writes as if this work does not exist. Yet he claims to have shown that ‘naturalism is not threatened by any of the arguments found in the works of Aquinas’ (Oppy (Reference Oppy2018), 125). This seems to me comparable to a defence attorney claiming to have acquitted a client, while the trial is still going on and the prosecutor's case remains unanswered.

Now, in other writings, Oppy does have more to say about some of the Five Ways. There is a substantive discussion of the first three of them in Arguing about Gods (Oppy (Reference Oppy2006), 98–107). He raises the following objections. None of the arguments shows that an infinite regress of causes is impossible. Even if they had shown this, none of them establishes that there is only one first cause rather than several. Even if they had shown that there is only one, none of them justifies identifying this cause with God. The First Way does not prove that there couldn't be changes that are not brought about by anything else, such as chancy processes and acts of libertarian free will. The Second Way fails to rule out the possibility of circles of causes. The Third Way fails to establish that whatever is contingent must at some time fail to exist, or that whatever comes into existence requires a cause, or that the physical universe is not a necessary being.

Most of these are common objections. I have answered all of them at length elsewhere, and the other Thomists I've referred to have responded to most of them as well. Here I will just briefly summarize the main points, and direct the reader interested in a complete response to other works.Footnote ¹⁰ First, when Aquinas judges that an infinite regress of causes is impossible, he is talking about causal series of the kind that I characterized above as hierarchical rather than linear, and the reasons for his judgement are the same as those I summarized when discussing this distinction.Footnote ¹¹ This is a point often emphasized in Thomistic discussions of the Five Ways, but Oppy appears not to be aware of it. At a couple of places in his discussion he remarks that Big Bang cosmology need not be interpreted in a way that rules out an infinite regress of causes, which indicates that he is making the exegetical mistake of supposing that Aquinas is concerned with linear causal series extending backward in time. Oppy thus fails to perceive, much less rebut, Aquinas's rationale for ruling out an infinite regress.

Second, Oppy's discussion in Arguing about Gods evinces no awareness that Aquinas in fact gives many arguments for the claim that there can be only one first cause.Footnote ¹² To take just one example, Aquinas holds that a first cause (or a first mover, or an absolutely necessary being) would have to be purely actual rather than a mixture of actuality and potentiality. He then argues that what is purely actual must be simple or non-composite. The next step is to argue that what is simple or non-composite must be unique, because for there to be two or more things that share an essence, there must be a distinction within each of them between its essence and its existence. Yet there is no such distinction in what is simple or non-composite. Oppy might raise various objections against this and Aquinas's other arguments for the uniqueness of a first cause, but the point is that he speaks, at least in that book, as if the arguments did not exist.Footnote ¹³

Similarly, the unwary reader would have no idea from Oppy's discussion that Aquinas in fact devotes literally hundreds of pages to developing arguments to the effect that a first cause would not only be non-composite and unique, but would have attributes such as incorporeality, supreme goodness, infinity, omnipresence, immutability, eternity, omniscience, omnipotence, life, will, love, and so on.Footnote ¹⁴ In other words, he argues that a first cause would indeed be God. Evincing no awareness of these arguments, Oppy does not answer them.

Aquinas's view that changes require a cause is grounded in the Aristotelian analysis of change as involving the actualization of potential. I have already responded above to Oppy's more recent objections to the claim that such actualization requires a cause. Aquinas would reject Oppy's suggestion that chance events can lack causes, insofar as he analyses such events as the result of the convergence of non-chance causal processes.Footnote ¹⁵ He also explicitly denies that free actions are uncaused.Footnote ¹⁶ Of course, these are large topics, and Oppy might end up disagreeing with what Aquinas has to say about them. The point, though, is that he does not consider, much less rebut, what Aquinas has to say about them.

Aquinas would rule out circles of causes for the same reason he would rule out an infinite series of causes. Again, what he has in mind are hierarchical causal series, in which causes other than the first have their causal power in only a secondary or derivative way, and the first has its causal power in a primary or built-in way. Whether secondary causes form a finite series, or an infinite series, or a circular series is simply not to the point. The point is that secondary causes would lack efficacy without a primary cause. For example, a railroad boxcar cannot move on its own and without an engine, and neither can a finite series of boxcars, nor an infinite series of boxcars, nor a series of boxcars that loops around on itself to form a circle. Hence, even if there were such a series of secondary causes that proceeded to infinity or looped around in a circle, there would still have to be a primary cause outside the series which imparted efficacy to the whole.

Finally, Oppy's criticisms of the Third Way also rest on exegetical mistakes that evince a failure to take into account the Aristotelian metaphysical background of Aquinas's arguments. Oppy interprets Aquinas's claims about contingency and necessity in terms of possible worlds, but they have nothing to do with that. They are grounded instead in the Aristotelian analysis of physical objects as composites of substantial form and prime matter. It is because prime matter never decisively locks onto any one form, and is always ready to take on another, that a physical object of which it is a constituent will, given enough time, corrupt and give way to something else. It is because prime matter is by itself merely the pure potentiality to take on form that, even if it did in some sense exist of necessity, its necessity would have to derive from some actualizing cause outside it. It is precisely because a composite of substantial form and prime matter (including the universe as a whole if we thought of it as one big substance) is a composite that it would require some actualizing cause outside of it that accounts for how its parts come together. Hence even if the universe were a necessary being, it too would have only a derivative kind of necessity. Once again, though Oppy might object to such arguments if he were familiar with them, the point is that he does not seem to be familiar with them, and thus does not in fact offer any response.

In a more recent article (2012), Oppy (Reference Oppy and Marenbon2012) has responded to Oderberg's defence of the First Way. Oppy suggests that there are naturalistic models of the universe that are consistent with the premises of Aquinas's argument while avoiding its theistic conclusion. He develops an example involving particles which generate gravitational fields that in turn affect the positions of the particles, so that the particles are all changing and each change has a cause. He then suggests that this series of changes may not regress infinitely, but instead trace back to some initial distribution of the particles. But we would not need an initial unchanging changer, or if we did, then it could be identified with the mass of the particles rather than with God.

There are several problems with this response. First, once again we see that Oppy misinterprets Aquinas as concerned with linear series of causes tracing backward in time, whereas he is in fact concerned with hierarchical series the members of which all exist here and now. Hence, whether the series of changes Oppy describes traces back temporally to an initial distribution of particles is neither here nor there. What matters from Aquinas's point of view is the question of what actualizes the potential of the particles to exert gravitational pull on other particles here and now, and at any other particular moment. Second, even if we go along with framing the issue the way Oppy does, the transition from the initial distribution of the particles to their beginning to exert gravitational pull on each other involves the actualization of a potential. That, Aquinas argues, requires a cause. Oppy does not show otherwise, but simply assumes otherwise. Third, Aquinas also holds that the first member of the sort of causal series he is talking about would have to be purely actual or devoid of potentiality, and, for several reasons, that would entail that the mass of the particles in the initial distribution could not be the unchanging changer. For example, we have seen that Aquinas holds that there can only be one purely actual cause, but there are multiple instances of mass, as many as there are particles. Mass is a property of the particles, and thus ontologically dependent on them, whereas what is purely actual is ontologically dependent on nothing outside it. Things with mass are contingent, whereas what is purely actual is necessary. And so on.

Once again we have claims that Oppy would no doubt disagree with, but the point is that they are all claims that form the larger context of Aquinas's arguments, and Oppy says nothing in reply to them. Thus he doesn't engage with Aquinas's position at sufficient metaphysical depth, at least not in a way that avoids begging the question.

The De Ente argument

In Naturalism and Religion, Oppy (Reference Oppy2018, 121–125) also briefly discusses the argument for a first cause that Aquinas develops in De Ente et Essentia. This argument begins with a defence of the Thomistic thesis that in each of the contingent things of our experience, there is a real distinction to be drawn between its essence and its existence. It then argues that anything whose essence and existence are distinct must be caused to exist by something whose essence and existence are identical, and that there can only be one such cause.

In response to this argument, Oppy makes points such as the following. Few if any naturalists would agree that there is a thing whose essence is its existence, a thing which has no potentiality, and so on. Naturalists also have many deep metaphysical disagreements with Aquinas. For example, few of them would agree with all of Aquinas's Aristotelian views about the metaphysics of substance, form and matter, genus and differentia, final causes, and the like. The De Ente also makes metaphysical claims about souls, angels, and God that seem hard to square with the Aristotelian claims, and with which naturalists would also disagree. These disagreements are not easy to resolve.

It is hard to know what to say about all this other than that it is as cursory as Oppy's treatment, in the same book, of the Five Ways. The fact that naturalists disagree with Thomists about various metaphysical issues is hardly news. Aquinas's argument for God's existence in the De Ente claims to settle one of these disputes in favour of the Thomists, so the thing to do is to evaluate that argument. But Oppy's remarks are too general and cursory to count as an adequate evaluation.

For one thing, some of the metaphysical disagreements cited by Oppy are not relevant to the evaluation of the argument for God's existence. For example, someone could accept that argument without necessarily agreeing with Aquinas about form and matter or souls and angels. So it is beside the point to note that naturalists would reject those particular views. For another thing, Aquinas and other Thomists do not simply assume, without argument, the background metaphysical claims that are relevant to the evaluation of the argument for God's existence. They give arguments for those claims too. So, it does not suffice merely to note that naturalists would disagree with those claims. The reasons why Thomists make the claims have to be evaluated as part of the evaluation of the argument for God's existence. Thomists are also well aware of claims to the effect that some of Aquinas's views do not sit well with the Aristotelian underpinnings of his philosophy. But they have answers to such claims, and Oppy ignores these.

Indeed, as with Oppy's discussion of the Five Ways, his discussion of the De Ente argument has the further deficiency that it ignores the work of contemporary analytical Thomists who have defended Aquinas. For example, Brian Davies (Reference Davies2006, ch. 2) and Gaven Kerr (Reference Kerr2015) have defended the De Ente argument, and I have done so as well (Feser (Reference Feser2009), 84–90).Footnote ¹⁷ Oppy says nothing about this work. It is no surprise, then, that Oppy misunderstands several crucial Thomistic ideas.

For example, Oppy misunderstands the notion of an angel as a ‘pure form’ (Oppy (Reference Oppy2018), 124). He seems to think that such a form would be a form that existed apart from a substance. But that is not correct. A pure form is a form that exists apart from matter. But it would still be a constituent of a substance, viz. an immaterial substance. Oppy also claims that Aquinas's identification of God's essence with his existence ‘entails the claim that pure existence – just as such – exists. But pure existence is not of the right ontological category for it to be true that – just as such – it exists’ (ibid., 125). He does not elaborate, but presumably he is endorsing Anthony Kenny's influential claim that to identify God's essence with his existence is to be making an unintelligible statement to the effect that ‘God's essence is there is an x such that’ (Kenny (Reference Kenny2002), 41 and 43–44). But this assumes the Fregean thesis that the notion of existence is captured by the existential quantifier. Since that is an assumption that Thomists deny, the objection simply begs the question (Klima (Reference Klima2004) ). On the other hand, Oppy also suggests that even ‘in the case of a familiar particular, it seems that existence is part of essence’ (Oppy (Reference Oppy2018), 124). For example, he claims that this is true of human beings. But as Thomists have pointed out, such a thesis would entail the falsehood that these familiar particulars exist of necessity rather than contingently (Oderberg (Reference Oderberg2001), 39).Footnote ¹⁸

As with his treatment of the Five Ways, Oppy neglects to engage with the underlying metaphysical premises of Thomistic natural theology in sufficient depth. Accordingly, though he raises important questions, he ultimately fails to show either that Aquinas's metaphysical premises are false, or that they are inadequately defended, or that the theological conclusions do not follow from them.Footnote ¹⁹