Preliminary remarks

I begin by considering two preliminary issues: One of interpretation, and one of technical clarification.

First, the interpretive issue. Just how to understand the ECA at it appears in Aquinas and Scotus has been a matter of some controversy. In this article I shall follow closely a line of interpretation that can be found in Gilby (Reference Gilby and McDermott1964), Brown (Reference Brown1966), Cross (Reference Cross2005), Frank and Wolter (Reference Frank and Wolter1995), Cartwright (Reference Cartwright1996), King (Reference King and Williams2003), Klima (Reference Klima, Klima and Hall2013), Feser (Reference Feser2013), and Kerr (Reference Kerr2015). There are three points common to this tradition of interpretation that I shall assume without argument in what follows. First, Aquinas and Scotus were not concerned to demonstrate a first cause in time, and indeed held it to be a logical possibility that the cosmos had no beginning. Second, they had no general antipathy for infinite series, admitting both the possibility of some infinite causal as well as mathematical series. And third, by ‘first cause’ in their arguments they did not primarily mean a numerical first (although they meant this as well), but rather a first in power or causal efficacy – that is, a ‘principal’ or ‘primary’ cause. This last point is important for appreciating that Aquinas and Scotus were not begging the question against a per se infinite series when deploying premises in their arguments like ‘if there were no first cause, then there would be no later causes’. This last point will be further clarified in the following section.

Second, I shall use the term ‘infinite causal series’ throughout this article as shorthand for ‘infinite and non-well-founded’. That is, I shall mean a causal series that more than being infinite in number, also fails to terminate in a first member that serves as a cause for all other members in the series. As is familiar from mathematics, not every series with infinite cardinality is non-well-founded. For example, the series of natural numbers under the less-than relation is infinite, but nevertheless is well-founded in the sense that it has a member, the number 1, that is less than all other members but for which no other member is less than it.

ECA

Aquinas and Scotus distinguished between two fundamentally different kinds of causal series: what they called ‘accidentally’ (or per accidens) and ‘essentially’ (or per se) ordered series.Footnote ¹ In a per accidens series, the members possess the causal power definitive of the series in themselves, or intrinsically. The medieval paradigm for this was a series of biological generators: a person generates a child, who goes on to generate another, who generates another, and so on. Although each person in the series exists only because of the prior generative activity of their parents, nevertheless they each have, in themselves, a generative causal power that can be exercised without dependence upon, or participation in, the ongoing activity of their parents. So Frank and Wolter (Reference Frank and Wolter1995), unpacking this distinction in Scotus, write that ‘when accidentally ordered causes are involved, although A may be dependent upon B for its original existence, B subsequently acts independently of A in its relationship to E’ (83). In other words, in a per accidens series, each member has the power to initiate the next causal act in itself, without depending upon or needing to participate in the power of earlier members of the series for that activity. Thus, in the generation case, each caused generator, qua generator, is related only ‘accidentally’ to its causal predecessors. As Aquinas put it, ‘it is accidental to this particular man as generator to be generated by another man; for he generates as a man, and not as the son of another man’ (ST I, Q. 46, a. 2). In this sense, each member of the series possesses the causal power definitive of the series intrinsically.

By contrast, in an essentially ordered series, each member with a prior cause does not in itself possess the causal power definitive of the series. Instead, it only possesses this power by depending upon or participating in the ongoing causal activity of prior members in the series. The paradigm here (drawn from Aristotle) was a series involving a person, stick, and stone, where the person moves the stone by way of the stick. In such a series the stick and stone, each of which has a prior cause, has no intrinsic power to move. Instead, it must be caused to move by the motion of prior members of the series, in this case ultimately terminating in something that possesses the causal power for motion in itself: the person or soul. Thus, each caused member of this series of movers, qua mover, is indeed related ‘essentially’ to its causal predecessor(s), for it is precisely in the act of moving that it depends upon being moved by its predecessor(s). In this sense, the caused members of a per se series possess the causal power definitive of the series in a merely contingent, or extrinsic, way.

Other conditions were sometimes added for a series to count as per se or per accidens (for example, Scotus thought per se causal series must also be simultaneous).Footnote ² For my purposes, however, it is only the aforementioned power-related conditions that are relevant. We may summarize the distinction between causes ordered per accidens and causes ordered per se, then, by way of the following contrasting necessary conditions (where ‘Φ’ denotes some causal activity (moving, heating, pulling, etc.)):

Causes that satisfy the first condition the scholastics called ‘primary’ or ‘principal’ causes of Φ. By contrast, causes that satisfy the second condition they called ‘secondary’, ‘instrumental’, or ‘intermediate’ causes of Φ.Footnote ³ Thus, in a per accidens series, each member is a primary cause of its effect, whereas in a per se series, each member with a causal predecessor is merely a secondary, or instrumental, cause of its effect. It is important to note, however, that to be a primary cause of Φing is not necessarily to be a primary cause simpliciter. Something may be a primary cause of Φing, while itself being caused to do this by some distinct causal activity Ψ. So, for example, a train's railway engine is a primary cause of the locomotion of all of the train cars to which it is attached, because it possesses intrinsically a power of locomotion (unlike the train cars). But it is not a primary cause simpliciter, because it must be caused to move the others by electrical energy and so on. Similarly, a fire is a primary cause of heat, because it possesses intrinsically a power to heat (unlike, say, a metal pan), but it is not a primary cause simpliciter, because it must be caused to heat by the ongoing action of oxygen, fuel, and so on. A primary cause simpliciter would be something that is not caused in any sense to engage in the activity it is engaged in.

We can then define the scholastic conception of a ‘first cause’ as follows:

We can now consider the style of cosmological argument that is the subject of this article: the Essential Cosmological Argument (ECA). The ECA traditionally has two stages: First, it is argued that every essentially ordered causal series must terminate in a first cause, and second, that there is a single primary cause of every essentially ordered series, one which must have some of the classical divine attributes – necessary existence, simplicity, and so on.Footnote ⁴ I am not here concerned with this second stage of the argument, but only the first. However, for simplicity, I will simply refer to this first stage as ‘the ECA’.

The ECA has as its conclusion that every essentially ordered causal series must have a first cause, and can be summarized as follows:

1. (1) For every essentially ordered causal series S, either (a) S is infinite, or (b) S terminates with a first cause.

2. (2) (a) is impossible.

   Therefore,

3. (3) For every essentially ordered causal series S, S terminates with a first cause.

Once we exclude the possibility of symmetric causation (as the medievals did), the truth of (1) is evident: Every essentially ordered series is either infinite, or terminates in a first cause.Footnote ⁵ The substance of the argument, then, lies in (2). But is a per se infinite series of causes impossible?

Aquinas and Scotus argued in the affirmative. Their argument for (2) can be reconstructed as follows:

1. (2a) In an infinite essentially ordered series, each cause would exert purely secondary causal power with no source in primary causal power.

2. (2b) It's not possible for anything to exert purely secondary causal power with no source in primary causal power.

   Therefore,

3. (2c) An infinite regress in essentially ordered causes is impossible (i.e. (2) is true).

The truth of (2a) is obvious, for it simply follows from the definition of an infinite essentially ordered series. An essentially ordered series is defined as one in which each caused member with a successor has no power in itself to cause its successor to Φ. Instead, it must be caused to Φ, and be causally sustained in its Φing, by something else. But, as pointed out above, to be a cause of this sort just is to be a secondary cause. Thus, in an essentially ordered series every caused member with a successor exerts purely secondary causal power. And this secondary causal power can have no source in primary causal power in the case of an infinite series. For no matter how far back you trace the causal ancestry of a given effect, you're met with only further secondary causes.

(2b), then, is the substantive claim. Why think that it is true? Aquinas justifies it in the following way (keeping in mind that the main sense of ‘first’ here is ‘primary’ or ‘principal’, rather than numerical): ‘a second cause does not act save through the influence of the first [causae primae]: so that every action of a second cause presupposes that of the [first] active cause’ (Aquinas (Reference Aquinas and Selner-Wright2011), 29), and ‘all second causes derive their action from the first cause [primo agente]’ (ibid., 31). Along similar lines, Scotus defends the exercise of secondary causal power necessitating a source in primary causal power with these words: ‘in essentially ordered causes the second, in so far as it causes, depends upon the first’ (Scotus (Reference Scotus and Roche1949), 43). But what, precisely, do these claims amount to? Here recent interpretative work by Edward Feser, Caleb Cohoe, and Gaven Kerr has been most helpful.

According to Feser, the underlying reason that secondary causes must depend upon primary causes in order to act, and so there cannot be an infinite regress in such causes, is precisely because they have no causal power in themselves to engage in the activity that they are engaged in. Because of this, they require something to impart to them this contingently possessed causal power. Otherwise, there would be no explanation of why something is engaged in a causal activity that it simply has no intrinsic power to engage in. Feser, unpacking this idea, writes:

To illustrate this point, Feser uses the example of a cup sitting on a desk, which is keeping it three feet off the floor (ibid., 22). The cup, in itself, has no power to be three feet off the floor. As such, the table on which it sits is required in order to keep it there. But then, the desk also does not, in itself, have a power to keep the cup three feet off the floor, and thus must be held up by the house's foundation. Ultimately, this series terminates at the earth, which does have in itself the power to sustain the whole cup-desk-foundation series, and confers on everything it is holding up a power to be such-and-such feet off the ground. Without the table, the cup would fall; without the house's foundation, the desk would fall; and without the support of the earth underneath the foundation, the entire series would fall. Feser concludes: ‘Since the desk, the floor, and the foundation have no power of their own to hold the cup aloft, the series could not exist in the first place unless there were something that did have the power to hold up these intermediaries, and the cup through them, without having to be held up itself’ (ibid., 23; emphasis in original). That which possesses this primary, or as I am calling it ‘intrinsic’, power, which it imparts to everything else in the series, is thus the ‘first cause’ of the series: ‘what it means for such a series to have a first member is that there is something which can impart causal power to the other members of the series without having to have that power imparted to it – something that has its causal power in a “built-in” or nonderivative way’ (ibid., 26–27).Footnote ⁶

Feser's analysis accords with that of Cohoe, who writes:

Like Feser, Cohoe sees the source of a per se series requiring a first cause in the purely contingent, or ‘derivative’, way in which the intermediate members possess their causal powers. Because the intermediate causes ‘depend on the first for their causal powers’ (ibid., 848), if the series had no first member with a primary power, as a per se series would not if it were constituted by an infinite regress in secondary causes, ‘there will be something that is an effect but that lacks a cause capable of producing this effect, and this is impossible’ (ibid.).

Kerr argues similarly, though instead of speaking in terms of causal power, he speaks of causal ‘efficacy’ – which, I take it, amounts to much the same thing. He writes as follows:

Thus, agreeing with Feser and Cohoe, Kerr asserts that in a per se ordered series the intermediate causes must have their causal efficacy ‘granted’ to them (‘imparted’, ‘conferred’, etc.). And this because the intermediate causes ‘do not possess such causality in themselves; it is distinct from what they are, distinct from their essences. Hence the effects of such series, i.e. things dependent on the primary cause, do not possess the causality of the series essentially, whereas the primary cause does’ (Kerr (Reference Kerr2019), 104). Elsewhere, he puts this in terms of the causal activity exercised by intermediate causes in a per se series ‘exceeding their capacity’ (ibid., 107), and ‘thereby requiring the causal influence of the primary cause’ (ibid.).Footnote ⁷ And this, again, is why there cannot be a per se infinite regress in causes:

We might summarize Feser, Cohoe, and Kerr's reasoning here as a defence of premise 2b above as follows:

1. (1) If it were possible for something to exert purely secondary causal power with no source in primary causal power, then something could exercise a causal power that must be conferred without anything having conferred it.

2. (2) It's impossible for something to exercise a causal power that must be conferred without anything having conferred it.

   Therefore,

3. (3) It's not possible for anything to exert purely secondary causal power with no source in primary causal power. That is, (2b) is true.

Before turning to consider how this defence of the ECA can be complemented by a modal approach that utilizes the powers theory of possibility, a word about a potential confusion is in order. The potential confusion is this: we may be misled into conflating the mere triggering of a previously possessed power with the conferral of a power not previously possessed. Failing to appreciate this distinction may render the force of the reasoning behind the ECA, either in its standard formulation or in my modal formulation to be considered shortly, opaque.

The difference between conferring a power not previously possessed, and triggering a power already possessed, is this. In the former case, the activity being manifest cannot be maintained without the ongoing activity of that which conferred the power. By contrast, in the latter case, the activity being manifest can be maintained without the ongoing activity of that which triggered the power, and this is indicative of the fact that the given power belongs to the affected object intrinsically. So, for example, when I roll a ball down a hill, I have not thereby conferred on the ball a power to move down an incline that it previously did not possess. Why? Because the ball can go on manifesting the activity of moving down the incline without my ongoing activity of moving it. So, in this case, I have simply triggered the ball's previously possessed power of moving down inclines. By contrast, when I move the ball up an incline, I am indeed conferring upon it a power that it did not previously possess. Why? Because the ball cannot go on moving up the incline without my continuously maintaining it in this activity, pushing it up the hill. The moment I stop conferring this power that it lacks in itself, that it has only contingently, it will immediately cease to move up the incline and roll back down.

Or consider a chandelier hanging from a fixture on the ceiling (to borrow an example from Feser (Reference Feser2017), 21). In manifesting its activity of being in the air six feet off the floor (say), it is not manifesting a power that it previously possessed, a power merely ‘triggered’ by the fixture holding it up. Instead, the fixture has conferred on it a power that it did not previously possess, since the manifestation of its activity of being in the air six feet off the floor must be maintained by that which has conferred it. By contrast, if I release a balloon from my hand, and it proceeds to float up to the ceiling six feet off the floor, I have not thereby conferred upon it a power that it did not previously possess to be in the air six feet off the floor. Instead, I have simply triggered the manifestation of its previously possessed power to do so. Once again this is because, unlike the chandelier, the balloon does not need my ongoing support to keep it in the air six feet off the floor.

Consider one final example. When I hold my hand over an open fire, I do not thereby confer on the flame a power to burn. Instead, I have simply (and unfortunately) triggered its pre-existing burning power. By contrast, when a fire is lit underneath a piece of metal, the fire indeed thereby confers on the metal a power to burn. It has no power, in itself, to burn anything, but insofar as it is being heated by the fire, it acquires this power. And this, again, is because the presence of the fire is required to sustain the metal's capacity to burn. Upon removing the fire, the metal's burning power will quickly disappear.

Note that this distinction between the mere triggering and the conferring of a power maps onto the distinction between per se and per accidens causation: a per se causal relation is generated when a cause confers a power on one or more effects. By contrast, a per accidens causal relation is generated when a cause merely triggers the pre-existing powers of one or more effects. Thus, essentially ordered series are generated by the conferring of powers, and accidentally ordered series by the triggering of powers.

Before unpacking my modal approach to defending the ECA, which complements the standard approach discussed in this section, let us first consider just what the modal theory in question – the powers theory of possibility – amounts to.

The Powers Theory of Possibility

Recently, a powers theory of possibility (hereafter ‘PTP’) has received notable defence in Pruss (Reference Pruss and Gale2001a), Borghini and Williams (Reference Borghini and Williams2008), Jacobs (Reference Jacobs2010), and Vetter (Reference Vetter2015).Footnote ⁸ According to PTP, possibilities are grounded in the powers of actually existing substances. Roughly, p is possible just in case there exist(s) something(s) with a power to bring it about that p.

According to Pruss (Reference Pruss and Gale2001a), PTP is the thesis that ‘[a] non-actual state of affairs is possible if there actually was a substance capable of initiating a causal chain, perhaps non-deterministic, that could lead to the state of affairs that we claim is possible’. Cashing out the idea more fully elsewhere, he writes:

Similarly, according to Vetter (Reference Vetter2015), who uses the term ‘potentiality’ as a catch-all for powers, dispositions, capacities, etc., ‘It is possible that p = _df Something has an iterated potentiality for it to be the case that p’ (ibid., 197). Vetter includes the ‘iterated’ qualification here in order to capture the idea that some proposition p may be possible not only because something currently possesses a power for it to be the case that p, but also because something may have a power to acquire further powers, which would be powers for it to be the case that that p.

Simplifying these formulations, in this article I shall assume the following definition of PTP (where the term ‘proposition’ is meant to be neutral as to the nature of propositions, and where something has a power to bring it about that p if it either (i) currently has a power to do so, or (ii) has a power to acquire a power (an iterated power) to do so):

The theoretical motivations for PTP are manifold. They include that of providing an account of possibility that avoids the pitfalls both of possibilism and of ersatzism about modality, and that accords with common sense (cf. Jacobs (Reference Jacobs2010)). To unpack such justifications for PTP in detail here would take us too far afield. Suffice it to say that this approach to the metaphysics of possibility has received sustained defence in recent literature, and has become a contender for a compelling account of modality, along with traditional approaches like the actualism of Plantinga (Reference Plantinga1978) and Adams (Reference Adams1981), and the possibilism of Lewis (Reference Lewis2001). I shall hereafter take for granted that PTP is a well-motivated theory of possibility, independent of any consideration of cosmological arguments in particular.

PTP and the ECA

Using PTP, we can formulate an argument to demonstrate that any possible per se ordered causal series must terminate in a first cause, and so (by implication) it is impossible that any regress infinitely. The argument infers the necessity for a first cause of any per se ordered series from the facts that (1) possibilities are grounded in powers, and (2) secondary causes are not equipped with the powers necessary to ground corresponding possibilities. In what follows I lay out each premise, defending or unpacking them as necessary as the argument proceeds.

Our intended conclusion is that every possible per se ordered causal series terminates in a first cause. As such, we begin by assuming an arbitrary instance of the antecedent of this claim:

By PTP we can then infer,

But what is it for S to exist? Simply for each member of S to be engaged, in order, in the relevant causal activity: x being caused to φ by y φing, y being caused to φ by z φing, and so on. So, for example, if we have an essentially ordered series constituted by the causal activity of motion, then for that series to be actualized is for each member to be moving, and by its moving causing its successor (if it has one) also to move. Or, if the series is constituted by the causal activity of suspension (say, a chain suspended from a ceiling), for the series to be actualized is for each member to be suspended, and by its being suspended causing its successor (if it has one) to also be suspended. To unfold more clearly the reasoning of our argument, then, we infer,

Next we consider whether p is a conferred power or is rather intrinsic:

Premise 5 is the core of the argument, so let us pause to consider its justification.

The reasoning behind (p5) is this. If a power is conferred, then it has to be conferred (as we saw in the previous section) by something that possesses the relevant power intrinsically, and so can confer it on others. But if so, then anything with a purely conferred power for some activity cannot exercise that power without the ongoing activity of that which confers it. As such, it is insufficient to bring about the relevant effect. Instead, it can only do so in conjunction with that which is conferring the power upon it, and so maintaining its causal activity. But then it is clear that anything with a merely conferred power cannot constitute the grounds for the possibility of it, or anything else, engaging in the relevant activity. Instead, that which confers the power – that which has it in itself, intrinsically – must be the grounds. Thus, (p5) is justified by the very nature of what it is to have a conferred power.

To clarify this point further, let us consider again two of the examples of conferred powers given in the previous section: the ball's power to move uphill, and the metal's power to burn. First, the possibility that the ball move uphill surely cannot be grounded in the ball's merely conferred power, for it is causally impotent to exercise, and indeed entirely lacks, this power in itself. Instead, the grounds of the possibility must involve my power (or some other ‘self-mover’) to move it uphill. The possibility, in this sense, more fully resides in me, not the ball, because the ball has no power in itself to move up the hill, though I can confer upon it this power, and so realize the possibility.

Second, consider the metal heated by fire. Insofar as it is being heated by the fire, it acquires the power to burn something: a power that the fire has conferred upon it. Now consider the possibility that the metal burn something (say, my hand). The metal in itself, and the powers native to it, cannot ground this possibility. It has no intrinsic capacity at all to burn anything. As such, the possibility that the metal may burn must be grounded in the intrinsic power of the fire to burn. Without the existence of fire, and its natural burning power, the possibility that the metal may burn could not exist, for the metal in itself is impotent to engage in this activity. Instead, it must participate in the power of fire.

Consider one final example not discussed in the previous section: the possibility that a mirror may illuminate something. No mirror, in itself, has a power to illuminate. As such, it must have this power conferred upon it by a light-source: lightbulb, fire, lightning, etc. Once light is shone upon it, it then acquires a derivative illuminating power. Because of this, the possibility that the mirror may illuminate something simply cannot have its grounds in the purely conferred illuminating power of another mirror. Rather, the ground must be in the intrinsic illuminating power of a light-source, which power can be conferred upon the mirror.

So much for a defence of (p5). But the consequent of (p5) is inconsistent with (p3), which states that x ₁ … x_n having power p indeed grounds the possibility that each member of S is φing. So, by modus tollens, we infer,

Which along with (p4) entails,

Now, it might be objected at this point that I have created a false dichotomy: that either the grounds for a possibility only involve that which has an intrinsic power to bring the possibility about, or they only involve that which has a merely conferred power to do so. To frame the objection concretely: do not the grounds of the possibility that the ball move uphill need to involve both the ball with its purely conferred power to move uphill and that which can confer the power?

This, however, fails to grasp the distinction between merely participating in a given causal activity, and actively bringing about or initiating that activity. And it is the power to actively bring about a certain activity, a certain state of affairs, that is relevant to grounding the possibility that such a state of affairs may obtain, not the power to merely participate, or be involved in it, once it obtains. Conferred powers merely participate in a causal act, are merely involved in a state of affairs, once it is brought about, but by definition they cannot themselves bring it about. By contrast, intrinsic powers, by their very nature, are powers to bring about or initiate the relevant act or state of affairs.

This distinction between merely being involved in, and bringing about, a state of affairs can be illustrated as follows. I have the power to bring it about that an apple is cut with a knife. Once this event is brought about, it is of course the case that, necessarily, the apple, knife, and the knife's conferred cutting power are constitutively involved in the event. Nevertheless, neither the apple, nor the knife with its conferred cutting power, are what actually has brought about the cutting of the apple with the knife. Instead, it is I who have done this, by exercising my intrinsic power to cut apples with knives. We must not conflate, then, that which has a power to bring a certain state of affairs about, and that which would essentially be involved in that state of affairs once brought about. And again, it is the former, not the latter, that is relevant to grounding a possibility on PTP.

Now, if that which grounds the possibility that every member of S is φing has an intrinsic power to φ, and so to confer the power to φ on the members of S (exempting itself), it then follows that,

This premise is a straightforward consequence of PTP itself. If some proposition P is possible because some x ₁ … x_n have a power to bring it about, then it follows that, if P were to come about, it must be by way of x ₁ … x_n exercising their power. If P could be brought about without the causal activity of that which grounds the possibility that P comes about, then it is difficult to see why x ₁ … x_n should be relevant to P's possibility in the first place. Note, however, that there need not be any unique xs whose power grounds a particular possibility. There may be a variety of powerful ‘witnesses’ to a given possibility. So, for example, the possibility that I am suspended 5,000 feet in the air is witnessed to by the existence of planes, helicopters, jetpacks, and so on (note that all such witnesses have an intrinsic power to suspend something in air, for they can do this without themselves being suspended in air by something else – unlike, say, a chain). Any one of these has the power to actualize this possibility. However, the principle behind (p8) still holds: For all it says is that at least some of the witnesses to a possibility must be exercising their power to actualize that possibility in order for the proposition to become true. My argument is neutral regarding, and is independent of, whether there are one, several, or many such witnesses to the possibility that S exists. For no matter how many or how few witnesses there are, the argument has so far established that any such witnesses must have an intrinsic, rather than conferred, power to bring it about that S exists. This is because x ₁ … x_n are simply a stand-in for any of the empowered witnesses to the possibility that S exists.

It further follows from (p8) that, were x ₁ … x_n to exercise this power, they would ipso facto be a primary or first cause of the essentially ordered series S. For recall the definition given above of a first cause:

Something(s) with an intrinsic power to φ, and which grounds the possibility that every member of S is φing, straightforwardly satisfies this definition. It satisfies (1) because, by the definition of their grounding the possibility of S and the nature of S itself, they would cause every member of S to φ if they exercised their power to bring S about. And it satisfies (2) because everything that exercises an intrinsic power to cause something else to φ, is thereby a primary cause of φing in that thing, since it does not require something else to confer on it the power it has conferred on that which it causes. So we can infer,

By modus ponens we then have,

But S was just any arbitrary instance of a per se ordered causal series. So we can conclude by generalizing as follows:

Before proceeding, let us briefly summarize this modal proof for the necessity of a first cause in every per se ordered series. According to PTP, every possibility is grounded in the existence of something(s) with a power to bring that possibility about. Consequently, the possibility that an essentially ordered causal series may exist is grounded in the existence of something(s) with a power to bring that series about. This power to bring the series about must either be a conferred power, or an intrinsic power. It cannot be conferred, for conferred powers, precisely because they are conferred, are insufficient to bring about the relevant effect. Instead, they are only sufficient insofar as they participate in the intrinsic power of that which has conferred this very power upon them. Thus, the power to bring about a per se series S must be intrinsic, not conferred. But then the per se series will only exist if this intrinsic power is exercised, and if it is exercised, the series by definition has a first cause. This is because a first cause of any activity φ in something else is that which can cause φing without needing this power to be conferred upon it by something else's φing. And this is just what an intrinsic power to φ is. So, every essentially ordered causal series, if it is to exist, must have a first cause.

If this argument is sound, then an infinite per se ordered causal series is impossible, and every such series must terminate in a first cause. It might be wondered at this point, however, what this modal approach adds to the standard defence of the ECA that the latter does not already possess. After all, isn't the core of this modal proof simply identical to the standard defence, hinging as it does on the notion that conferred powers must be conferred by something that possesses the relevant power intrinsically?

I believe this modal approach adds to the standard defence in at least two ways. First, it is arguable that part of what creates the illusion of the possibility of an infinitely regressive per se ordered series is that we begin by thinking of the series as already actual, and then consider whether something would be explanatorily deficient about the series were it so. Those who endorse the possibility of such a regress answer ‘No’, because in such a case each member of the series would have a cause, and so what is there left to explain (cf. Hume (Reference Hume and Coleman2007), 65–66)? Those who reject such a possibility answer ‘Yes’, because each member of such a series would have nothing from which to derive their purely conferred, or derivative, causal power. The latter are in my view correct. However, my modal approach complements the standard one by forcing us to take a step back from this ‘moment’ in the dialectic, where we consider the series as already actual and then ask what might or might not be explanatorily deficient about it, and to a logically prior ‘moment’ in which we consider what could ground the very possibility of such a series in the first place.

This is important, because if we begin by thinking of the series as already actual, the distinction between conferred and intrinsic powers is liable to become opaque. Thinking of one engineless train car pulling another, which pulls another, which pulls another, and so on ad infinitum (for example), our attention is drawn not to the intrinsic (or lack thereof) potencies of the train cars, but rather to their actualities. And in terms of the act of pulling, a locomotive (which has an intrinsic pulling power) that is pulling a train car, and another train car (which lacks such an intrinsic power) pulling a train car, are identical. And because the act is identical, conceiving of the series as already actual naturally lends itself to the illusion that the motion of each train car can be completely accounted for, even without a locomotive ‘first puller’, because each train car has something that is acting on it in the right way: pulling it. The modal approach to the ECA breaks this spell, by forcing us to think of the series not as actual, but as possible, and then connecting its possibility to an ontology of powers. When we do so, it becomes transparent that only certain kinds of powers – intrinsic powers – can ground the relevant possibilities. And from this, as the argument shows, it logically follows that the series could not exist without a first cause.

Second, and relatedly, this approach helpfully ties the ECA to an independently plausible modal theory: the powers theory of possibility. In so doing, it both helps to motivate those who endorse such an approach to modality to seriously consider the soundness of the ECA, as well as clarifies the underlying modal presuppositions involved in debates over its soundness. Those who defend the ECA typically presuppose (rightly, in my view) an ontology of causal powers, and that certain kinds of powers are more or less relevant to the possibility that a certain effect may take place. Those who reject such an ontology (either knowingly or unknowingly) will, by implication, naturally fail to see the force of the standard defence of the ECA. Unpacking the argument in terms of that ontology, in particular the ontology as it bears on modality, thus helps to illuminate the metaphysical framework in which the ECA becomes intelligible.