1. Aquinas’s Second Way: problems and method of analysis

Aquinas’s Second Way for proving the existence of God argues for the existence of an uncaused cause of all effects and is presented by him as giving a demonstration – a valid deductive argument from premises known with certainty – of its conclusion (cf. 1a q. 2 a. 2, S.C.G. book 1 chs. 12–13Footnote ¹). The object of this discussion is to answer the question; why did Aquinas hold that the Second Way is a demonstration?

The Second Way runs as follows:

1. 1) We know through experience that there is an order of efficient causes of effects.

2. 2) It is not possible for anything to cause itself; for if it did, it would be prior to itself, which is impossible.

3. 3) It is not possible to advance from efficient cause to efficient cause ad infinitum. In all ordered efficient causes, the first is the cause of the intermediate cause, and the intermediate cause is the cause of the final effect; and this is the case whether the intermediate causes are one or many. If a cause if removed, its effect is removed. Therefore, if there is no first efficient cause, there is no intermediate cause or final effect, which is patently false.

4. 4) Therefore, it is necessary to posit a first efficient cause, which all call God.

The claim that assertions 1) and 2) are known with certainty can be accepted as philosophically defensible. But the reasons given for the impossibility of infinite causal regresses in 3) are open to question, and do not seem more certain than the conclusion they seek to establish. Aquinas offers no explanation of why the argument is supposed to establish that there is only one uncaused cause of all effects. The same observations apply to Aquinas’s brief exposition of the argument for an uncaused cause in the Summa contra gentiles (S.C.G. 1, ch. 13 n.33). On the face of it, the argument is therefore insufficient as a demonstration. Hence, the question that is addressed in this discussion.

A number of different answers have been proposed for this question, and scholarly discussion of it has not converged on agreement. The approach to the question that will be taken here takes its cue from Aquinas’s views on the way in which we can attain a philosophical understanding of God. Aquinas thought that philosophical knowledge of God was the goal and summit of the philosophical enterprise and that the whole of philosophy leads up to it and is presupposed by it. It requires a long and difficult course of study which very few people have the ability and opportunity to undertake (logic, mathematics, natural sciences, ethics, then metaphysics; cf. In Eth. lib. 6 lect. 7, nn.1209–1212; S.C.G. I ch. 1, ch. 4 nn.3-5; In liber de Causis prol.). Aquinas’s Five Ways assume knowledge of the philosophy that is presupposed to them, and this philosophy must be understood if the Five Ways are to be understood. Such an understanding will enable us to identify the philosophical positions that led Aquinas to hold that the Second Way was based on known philosophical truths, and to discern the nature of Aquinas’s inference from these positions to the conclusion of the Second Way.

The way to understand the Second Way is thus to examine the metaphysics that it is based on. This does not mean recapitulating the substance of Aquinas’s philosophy, or retracing the pedagogical stages that Aquinas thought necessary for a mastery of natural theology. The Second Way is concerned only with efficient cause and effect and does not invoke the difficult and controverted notions of goodness, perfection, necessity, and teleology that appear in the other Ways. These pedagogical stages are presented by him as being required not only to grasp his philosophical arguments for the existence of God, but also to know with certainty that the premises upon which they rest are true. This is not being attempted here; we limit ourselves to inquiring why Aquinas thought that the Second Way provides a demonstration of its conclusion.

2. Aquinas on causes

Aquinas understands a cause in a general sense to be a feature of reality upon which another thing depends for its being or becoming (In Met. book 1 lectio 1). It is his understanding of efficient causation that matters for understanding the Second Way. Aquinas considers an efficient cause to be a thing, rather than an event. He holds that efficient causes produce effects through the exercise of active causal powers (‘potentia activa’; cf. Quaest. disp. de anima a. 13 co.). He considers substances, the basic components of created reality in his metaphysics, to be causes. Being a substance involves having causal powers (cf. 1a q. 80 a. 1 co.). The essence of a substance is identified by the causal powers that it has, and determines those causal powers (1a q. 79 a. 5 ad 3, 1a2ae q. 109 a. 1 co., De anima art. 12 co.). Aquinas does not limit the category of causes to substances, as this passage indicates:

His example of an army can be developed in order to explain his understanding of causes that are not substances. An army is not a substance, but it acts, and its activity depends not only on the individual contributions of the parts of it that are substances, but also on the structure and activity of the army as a whole. All the soldiers of the army might be defeated attacking in accordance with a bad plan when they would win if they attacked following a good plan. The organization and command of the army and its conformity to a given plan of attack make an essential causal contribution that is not reducible to the causal contribution of its parts. Such non-substantial causes must nonetheless ultimately have substances as their basic parts, although they can have other parts that are themselves made up of substances whose character is itself of causal significance (e.g., the organization and discipline of a subunit in an army will make a difference to the army’s fighting power).

3. Aquinas on effects

This account of Aquinas’s conception of cause presents a view that has a familiar place in past and present philosophy. His account of the nature of effects is less familiar. Aquinas’s account of causes is basically Aristotelian, but his account of effects is influenced by the Neoplatonic sources of his thought. He provides a general account of what it is to be an effect in the course of explaining and arguing for divine simplicity: ‘Every composite thing has a cause, for things that are different in themselves cannot be joined together in some unity except through the action of something that causes them to unite’ (1a q. 3 a. 7; see also In liber de Causis lectio 21, De potentia q. 7 a. 1 co.) ‘Whatever belongs to a thing and is yet not of its essence belongs to it from some cause; for, if things that are not through themselves one are joined, they must be joined through some cause’.Footnote ³ (S.C.G. I, ch. 22 n.6; see also De ente et essentia ch. 3 n.80). The idea seems to be that a real unity resulting from the joining together of things that are really different is an addition to the real things that exist, and as such cannot come into being of itself without a cause.

This general account needs to be fleshed out by an account of what it is to be joined together in the requisite way, and an account of what it is to be different in the requisite way. Aquinas provides this fleshing out by giving an account of what it is to be an effect for the different kinds of beings in his ontology. All causal activity falls into one of six categories in Aquinas’s view: (i) causing accidents in one or more substances, (ii) causing the existence of a substance, (iii) causing the beginning of existence of a substance, (iv) causing the essence of a substance, (v) causing the beginning of the essence of a substance, or (vi) causing some real entity that is composed of more than one substance, or causing a real property in such an entity. These categories correspond to the entities in Aquinas’s ontology that are brought into substantial or accidental being through causal action.

Accidents exist as attributes of substances, rather than existing independently as substances do. Aquinas divides accidents into three kinds: proper accidents, such as risibility in humans, which follow from the essence of a thing and are possessed by it necessarily; inseparable accidents, such as masculinity and femininity, which have a permanent cause in the individual that possesses them rather than in the essence of that individual; and extraneous accidents, which their bearer can lose or acquire (Quaest. disp. de anima q. 12 ad 7).

The difference between an extraneous accident and its bearer is a real difference, since both accident and bearer are real entities that are distinct from one another, and the possession of the accident by its bearer is a real unity. This is a straightforward case of Aquinas’s general definition of what it is to be an effect (cf. De potentia q. 7 a. 4 co., S.C.G. I, ch. 23 n.3). Hence, all extraneous accidents will be effects (1a q. 77 a. 6 co.).

Aquinas describes proper accidents as being caused by the essence of a thing (1a q. 77 a. 1 ad 5, De spiritualibus creaturis art. 11 resp., De ente et essentia ch. nn.106–107), and inseparable accidents as being caused by the individual subject they exist in (1a q. 54 a. 3 ad 2). Evidently, he considered essences to cause accidents in the general sense of being something that accidents depend on for their being or becoming, but the specific kind of causation that he had in mind is not clear. He cannot consistently hold that essences are efficient causes, since he asserts that it is things rather than forms that are efficient causes (IIaIIae, q. 1 a. 2 co.). Regardless of what Aquinas had in mind when he said that proper accidents are caused by the essence of a thing, it seems that he would agree that the cause of the essence of a thing is also the cause of its proper accidents (cf. In Phys. book 2 lectio 1, n.144).

Aquinas thinks that the composition that makes substances effects with a cause results from the real distinction between their essence and their existence. There are long-standing disputes over the nature of Aquinas’s distinction between essence and existence,Footnote ⁴ but we can identify substances that are caused to exist according to Aquinas’s metaphysics without resolving these disputes. If a substance possesses an essence that can be completely understood without knowing whether or not any substance with that essence exists, then its essence is distinct from its existence. Aquinas borrows Avicenna’s example of the phoenix to illustrate this assertion; one can fully understand what a phoenix is, without knowing whether or not a phoenix exists in reality (De ente et essentia ch. 3 n.77, and see also In 2 Sent. d. 3 q. 1 a.1). This is a criterion for identifying a real distinction between essence and existence, not a statement about what such a difference consists in. A real distinction exists when the existence of a thing is not a component of, or identical with, or a necessary consequence of its essence. The essence of a thing that is distinct from its existence cannot exist independently of that existence, or vice versa. Nonetheless, when the conditions just specified obtain, the essence and existence of such a thing are really distinct components of that thing, and their union – which is necessary and sufficient for the thing to exist – is a real union that must have a cause (cf. 1a q. 4 a. 4). A real distinction between essence and existence in a substance is only absent if the essence and the existence of that substance are one and the same thing. While a real distinction between essence and existence means that a substance is caused to exist, the absence of such a distinction means that a substance exists necessarily and cannot be caused to exist.

Aquinas considers that the cause of the existence of substances must be God acting directly and without any intermediary. This causation cannot be identified as happening at a particular time during the existence of the thing (De potentia q. 3 a. 4 ad 11; 1a q. 104. a. 1 ad 4). Aquinas describes God as conserving things in existence, but he does not understand this conservation as a process occurring over time that can be divided into temporal stages. God’s single undivided action of causing the existence of a thing is responsible for its continuing to exist.

Aquinas thinks that the causing of the existence of a substance and the causing of the beginning of existence of a substance are distinct actions (a distinction that does not obtain with the causing of accidents). This follows from his belief that only God can cause the existence of a substance, and from the observable fact that created substances sometimes cause the beginning of existence of other substances, as in the generation of living things (cf. 1a q. 104 a. 1 co.). When created things cause the beginning of existence of a substance, however, they must do this by acting on already existing things. God alone is capable of causing the beginning of a substance’s existence without acting on any preexisting thing. The beginning of the essence of a thing can also be caused by something other than God (cf. De pot. q. 5 a. 1 co.; the example is a fire being lit by another fire). Some interpreters have considered that Aquinas thought that the essence of a creature, as well as the beginning of its essence, can in certain cases be caused by another creature, but it is not entirely clear that he is personally committed to this idea.Footnote ⁵

Aquinas develops his account of what it is to be an effect in the context of his discussion of divine simplicity, with the object of identifying what characteristics must be excluded from the nature of the first, uncaused cause. This context leads him to focus on characterizing effects that are individual substances or attributes of individual substances, since it is an individual being – God – who is in question in the investigation of divine simplicity. In this, he follows a Neoplatonic tradition of describing the nature of effects in order to accurately characterize the supreme divinity by excluding anything caused from it.Footnote ⁶

Aquinas is less specific about the properties of caused groups of substances that make these groups effects. His work does however permit us to draw some conclusions about his views on this subject. The difference between distinct substances is a real difference of the kind postulated by Aquinas’s general account of effects. A causal property of a collection of substances whose actualization is an operation that is proper to the whole, and is not proper to its component substances considered as individuals or as a mere collection of individuals, would be a property that would constitute a real union of these substances. It would thus be an effect.

Aquinas provides a criterion that can be used to identify effects in his discussion of fortune and chance. He asserts that things that happen all the time, such as the sun rising, or most of the time, such as humans being born with eyes, must have causes (In Phys. book 2 lectio 8 n.208). The connection that he makes between regularity and causation is reminiscent of Hume, but the differences between the role of regularity in Hume’s understanding of causation and its role in Aquinas’s understanding are worth noting. For Hume, on a standard interpretation of his thought, causation in the objects just is constant conjunction. For Aquinas, the criterion of regular occurrence is sufficient rather than necessary for identifying effects. There is nothing in Aquinas’s theory of causation that rules out unique effects. For Hume, the regularity of cause and effect is instantiated by pairs of cause and effect; both cause and effect are open to observation; and knowledge of a causal relation comes from observing cause and effect together. For Aquinas, the regularity that indicates the existence of a causal connection is a regularity of the effects. Knowledge of the existence and character of a cause can result from knowledge of the regularity of the effects alone; and the cause of the effects is often not observed, and may not be observable, but is frequently inferred from the effects.

Hume considers that the regularity that is constitutive of causation in the objects is a purely relational property. The relata of the causal relation have to have certain intrinsic properties that permit them to enter into this relation – for example, an infinite series of events with no beginning or end could not be a cause or an effect, on Hume’s view, since nothing could come before or after it – but these properties are just necessary conditions for entering into the causal relation. They do not constitute being a cause or an effect. Provided such necessary conditions are satisfied, there are no intrinsic, non-relational properties of causes and effects that are essential to their being causes and effects. For Aquinas, there are intrinsic properties involved in being a cause or an effect. Aquinas would have held Hume’s account of causation in the objects to be no more than a confidence trick that presents an effect of causal action – universal regular succession – as being causation itself.

4. The causal activity of the uncaused cause

This exposition of Aquinas’s metaphysics of causation enables us to understand his view of the causal links between the first cause and its effects.

On some accounts of causation, transitivity is an inevitable and straightforward property of causation. Consider a simple model where causes and effects are events, and causation is the instantiation of a universal generalization of the form ‘All events of type x cause events of type y’. A serious philosophical account of causation will elaborate on this model by making further claims about the nature of the universal generalizations that feature in causation, but these refinements are not relevant for our purposes. In such a model, if A causes B and B causes C, A also causes C, since the relevant causal generalizations will mean that C must be caused to occur if B occurs. The causal connection between an uncaused cause and its effects will be straightforward in this model.

This is not the case with Aquinas’s metaphysics of causation. Causal series will be more complex in this metaphysics. The members of these series can include all the types of cause and effect enumerated above; causal action will not necessarily be transitive; causes can produce different results depending on the circumstances of their action; and causal activity can typically be changed or prevented by interference from other causes.

The nature of the causal series that connect effects to the uncaused cause in Aquinas’s metaphysics is determined by his conception of a first cause. The first cause is not first purely on account of its being uncaused. Aquinas understands the first cause to be the sole ultimate causal principle of the existence of the effect that it causes. The title ‘first’ refers to the status of the first cause as a first principle, not just to its numerical position in the causal series (see In Met. book 2 lectio 2). Its character as a first principle means both that the first cause is uncaused – otherwise it would not be first – and that all other causes for its effects are themselves caused by it to exist and to act in the production of these effects.

Given Aquinas’s metaphysics of causation, this entails that if the first cause is not the proximate and sole cause of an effect, it must cause the existence, nature, power, and causal activity of the proximate cause of that effect – and it must do the same for any cause of these features of the proximate cause that is itself caused, for any further caused cause of these causes, and so on. If a caused cause of an effect is an entity composed of several substances, the first cause must cause the component substances of this entity and the property or properties that joins it together as a single thing. If the first cause is not the cause of all these factors, it is not the sole ultimate principle of the causing of the effect. It is not even a sufficient cause for the effect, since if any of these factors is lacking the effect will not occur.

There will usually be more than one cause of the existence, nature, power, and activity of an intermediate cause, and these causes will often produce their effects at different times.Footnote ⁷ The sole cause of the existence of an intermediate cause will always be God, when that intermediate cause is a single substance. The causes of the power and activity of the intermediate cause can themselves be effects, and can be different entities each of whose power and activity are also caused by more than one thing. This means that the causal series leading from an effect to the first cause will typically branch out into many causal chains as it ascends from effect to cause. The nodes from which branches divide in each causal chain will be causes whose existence, natures, powers, and activation are themselves caused. When more than one cause produces these features of an intermediate cause, there will be different causal chains leading from the intermediate cause to the causes of these features.

The multiplicity of these causal chains does not mean that the effect must be ultimately produced by more than one uncaused cause. Since the same cause can act many times, every causal chain involved in the production of an effect can ultimately begin with different actions by one and the same being. The complete causal series that produces a given effect will be composed of all the causal chains leading to the production of that effect. The outermost branches of every tree of causal relations will all end in the one uncaused cause. When Aquinas argues for the existence of a first cause of all effects, he is arguing for an uncaused cause that produces every causal series containing every caused cause involved in the proximate or remote causation of every effect.

Aquinas’s argument for an uncaused cause is often read as describing a causal series where every member of the series is a single cause that gives rise to a single effect and is itself caused by a single cause, as in his example of the hand moving the stick moving the stone. He does not describe the causal series connecting the first cause and its effects in this way. Aquinas describes a causal series where an effect is produced through intermediate causes as follows:

The first cause is the superior cause of all other agents and will thus cause all these aspects of the action of these agents.

This description of the sort of causal series that Aquinas has in mind in the Second Way might be rejected, on the grounds that the Second Way is concerned only with causal series made up of the activations of causal powers. On this interpretation, the claim of the Second Way is that if A causes effect E, and A’s power to cause E is activated by B, B’s activation of A’s power to cause E is caused by C, and so on, this series of activations of causal powers cannot continue without end, but must begin with a cause that causes all of these activations of causal powers, and is not itself caused to cause these activations. Causation of the existence, natures, and causal powers of intermediate causes would not be included in the causal series that are the subject of the Second Way.

This interpretation is not a plausible one. So interpreted, the Second Way does not establish that the existence of the first cause is uncaused, only that the action of the first cause in producing an effect is uncaused. This permits a causal series to begin with an uncaused action of a thing that is itself caused to exist, and not with a first cause as Aquinas understands it. It is possible that Aquinas overlooked this obvious objection, but it is not likely. The interpretation does not have textual support in Aquinas’s works, which nowhere limit the causal series in arguments for an uncaused cause to the activations of causal powers. The texts in which Aquinas discusses the argument for an uncaused cause all refer to efficient causality in general, with no restriction on the type of effect produced. Aquinas contradicts this interpretation in his description of the first cause’s action:

The interpretation is incompatible with the passages in his works where he asserts that the cause of the powers that are activated to produce an effect must be considered as a cause of that effect (S.C.G. book 3 ch. 67 n.2, 1a q. 105 a. 5 co.). It makes the Second Way superfluous, since the activation of a causal power is a change (unless the activation in question is one that has always been going on with no beginning, a possibility that Aquinas does not think is ever realized). Such an activation, being a change, would fall under the First Way, which argues for an unchanging cause of all change. The Second Way would thus be no more than a restatement of the First Way. But Aquinas describes the Second Way as a different argument – ‘alia via’ – from the First Way; and it would be pointless to repeat the First Way in the Second.

There seems to have been a development in Aquinas’s thought on the argument for an efficient cause of all effects. In ch. 4 of the De ente et essentia, an early work composed ca. 1252–1256, he appealed to the impossibility of an endless regress of causes of existence in order to argue for a first cause of all existence. Near the beginning of his career, Aquinas held that a created being could cause the existence of another created being. In his commentary on the Sentences of Peter Lombard, another early work (ca. 1252-–57), he accepts Lombard’s position that although no creature has in fact created anything, it is possible for a creature to create another thing through God’s using it as an instrumental cause in creating (In II Sent., d. 1, q. 1, a. 3) – although he did not think it possible for a creature to create anything through its own power. On this assumption, it makes sense to allow for the possibility of a series of caused instrumental causes of existence, and to argue that such a series must end in a first cause of existence.

Later on, however, Aquinas firmly rejected the possibility of anything aside from God causing the existence of a substance in any way, instrumental or otherwise (1a q. 45 a. 5; De pot. q.3 aa. 4, 5; S.C.G. book 2 ch. 15). This change of mind led to his abandonment of the argument in the De ente et essentia, which does not recur in his later works. There is no point in arguing that a series of causes of existence must have a first member, if such a series is impossible. The materials of the argument in the De ente et essentia are divided into two different arguments in his later works. One is the argument of the Second Way, which retains the objection to endless causal regresses but extends the scope of the argument to all effects whatsoever. The other kind of argument drops the reference to causal regresses and argues for an uncaused proximate cause of the existence of substances whose essence is distinct from their existence (see 1a q. 45, S.C.G. book 2 ch. 15, De pot. q. 3 a. 5). Some interpreters have identified the Second Way with the argument for an uncaused cause of existence.Footnote ¹⁰ However, the considerations given above show that the Second Way is not limited to the causation of existence.

5. The impossibility of endless causal regresses

Having clarified the nature of the causal series that Aquinas is talking about in the Second Way, we can proceed to address the question; why does Aquinas think that an endless causal series of this kind is impossible?

We can note at the outset that Aquinas does not maintain that every kind of causal series must have a first cause (cf. 1a q. 46 a. 2 ad 7). He accepts that a series of causes that are required per accidens for the production of an effect can be endless. As an example of causes ordered per accidens to the production of an effect, he instances a father begetting a son who begets a grandson. The father does not cause the begetting of the grandson; causation is not transitive in a series of per accidens causes. The causal connection between the uncaused cause and its effects, as described above, will however be transitive. This is the kind of causal series that necessarily has a first cause; Aquinas describes it as a per se causal series.

In order to argue that a causal series must have a first cause, the possibilities of circular causation and of an endless causal regress must be excluded. This is pointed out by Proclus in proposition XI of his Elements of Theology, which Aquinas had read, but Aquinas does not mention circular causation or repeat Proclus’s objections to it. He may have considered that he had covered the question of circular causation by his rejection in the Second Way of the possibility of a cause being the cause of itself, something which would occur in the case of circular causation.

Aquinas gives this argument for the necessity of a first cause for any effect:

Aquinas takes the argument from Aristotle (Metaphysics book 2 ch. 2 994a), where it is stated in much the same terms. He follows Aristotle in asserting that a first member of a causal series would be necessary even if there were an infinite number of intermediate causes between the first cause and the final effect (In Met. book 2 lectio 3).

This argument is correct in stating that for every causal series that has a first cause, the suppression of this first cause would result in the suppression of all the subsequent members of the series and the suppression of the final effect (assuming that there is no causal overdetermination of any of the members of the series). But as an argument for the conclusion that all causal series must have a first cause, it seems to beg the question. From the fact that a causal series with a numerically first member requires this first member in order to exist, and would not exist if that cause were removed, it does not follow that every causal series must have a numerically first cause in order to exist – unless we assume that all causal series have a first member; which is the point at issue.

Peter Geach has presented Aquinas's argument against an endless causal regress as follows:

This is wrong, but interesting and suggestive. Geach asserts that Aquinas does not hold that God could be reached by following to its end a causal chain starting from any random object. Other interpreters have not denied that Aquinas considers that God can be reached by following such a causal chain to its end, but they have asserted that Aquinas does not base his argument in the Second Way on this claim. Neither of these positions can be reconciled with Aquinas’s thought. The claim that if we follow a transitive causal chain starting with a randomly chosen effect all the way to its end we must arrive at God is precisely the argument of the Second Way. It is plainly stated by Aquinas in a number of places (see, e.g., In Met. book 2 lectio 3). There is no room in Aquinas’s metaphysics of causation for a cause of an effect that does not appear in the causal series that begins with the proximate cause of the effect, continues with the proximate cause or causes of the proximate cause, and so on. Causes and effects of the types that Aquinas believes to exist cannot be arranged in an ordering that would include a cause outside this series.

Geach’s presentation of the argument of the Second Way as starting with a ‘lumping-together’ argument is, however, on the right track. Arguments that reason from the parts of a whole all having a certain property to the whole itself having that property are termed composition arguments. Such arguments do not have the form of the fallacy of composition or combination that Aristotle described in his Sophistical Refutations (166a20-31). Some composition arguments are deductively valid and others are invalid. The validity of composition arguments depends on the property that features in them.Footnote ¹² ‘Every tile that makes up the roof is red; therefore, the roof is red’ is valid; ‘every tile that makes up the roof is square; therefore, the roof is square’ is invalid. There are no known general rules for distinguishing valid from invalid composition arguments, but properties for which it is obvious that a composition argument is valid, and properties for which it is obvious that a composition argument is invalid, are both fairly common. Aquinas was familiar with the idea of a property that will support a valid composition argument. In 1a q. 11 a. 3 ad 2, he distinguishes between homogeneous wholes that are composed of similar parts and heterogeneous wholes composed of dissimilar parts. At the beginning of his De generatione et corruptione (314a25-314b1), Aristotle mentions Anaxagoras’s idea of a homoeomerous element; this is a body in which the parts and the whole are the same in name and nature. Aquinas wrote a commentary on this text. An attribute that is the same in name and nature for a whole and for the parts of that whole is an attribute that can be used to construct a deductively valid composition argument.

The claim that Aquinas bases the Second Way on a composition argument points us in the direction of the assertion that Aquinas uses to rule out the possibility of an endless causal regress. This assertion states that a causal series that is made up entirely of intermediate causes will itself be an intermediate cause. We should not understand ‘intermediate cause’ in this context in a question-begging way as ‘a cause that occurs between the final effect and the first cause in a causal series’; we should understand ‘intermediate cause’ as simply meaning ‘a cause that produces an effect and that itself has a cause’. It is the double character of being a cause and being caused that makes something an intermediate cause.

Aquinas does not argue for this premise by presenting it as a valid instance of the composition argument form. He simply presents it as being obviously true. This is a common practice for composition arguments, since it is the particular property being used that makes such an argument a valid one, rather than the mere fact of its instantiating the argument form; no-one would argue that since all the tiles of a roof are red, the roof is red.

Aquinas’s characterization of intermediate causal series as themselves being intermediate causes is likely to have been based on more than the fact that all the parts of such a series are themselves intermediate causes. The causal relations between the members of such a series, as well as the attributes of its individual members, would have formed part of his reasons for saying this. Thus he does not reason exactly as a composition argument does. It remains the case that his premise about intermediate causal series is a ‘lumping-together’ resembling the inference involved in composition arguments. The premise lumps together all the intermediate causes – all the causes that both cause and are caused – that produce a given effect, and asserts that the whole described by this ‘lumping-together’ is itself an intermediate cause.

Aquinas makes this claim more clearly in his commentary on the Metaphysics of Aristotle than he does in the Second Way. When presenting Aristotle’s argument against the possibility of an infinite regress of causes of motion or of efficient causes, he describes Aristotle as asserting that the totality of all the intermediate causes in a causal series are to be understood as one intermediate cause regardless of their number, insofar as they are considered in their character of being intermediate causes (‘quia omnia plura media accipiuntur loco unius, inquantum conveniunt in ratione medii’; In Met. book 2 lectio 3 n.303).

Aquinas performs a ‘lumping-together’ in what he says about intermediate causes as causes, as well as in what he says about intermediate causes as effects. He holds that if an effect is caused by a series of intermediate causes, then the series as a whole is itself a cause of the effect. This inference can be illustrated by an example. Consider a cart that is being pulled by a chain. Every link of the chain will be pulling the cart; from this, we can infer that the chain as a whole is pulling the cart. This will be the case even if the number of links in the chain is endless. Aquinas might ask why such an argument would succeed in establishing that an endless series of intermediate causes is itself a cause, as in the case of the endless chain pulling the cart, but would be invalid if used to argue that an endless series of intermediate causes is itself caused.

When this premise about intermediate causal series is granted, the argument against the possibility of an endless causal regress is straightforward. Parts of this proof are stated by Aquinas, and parts of it are assumed by him as being obvious enough to take for granted.

Suppose that an endless per se causal series produces a given effect. Every member of the series will be an intermediate cause, since every member will both cause something – either the final effect, or the cause that is subsequent to it in the series – and will itself be caused by the member prior to it in the series. Because every part of the endless series is an intermediate cause, the endless series itself – by the ‘lumping-together’ argument – will be an intermediate cause. Since the endless causal series is an intermediate cause, and intermediate causes are effects, an endless causal series must be an effect, and hence must have a cause.

But an endless per se causal regress, since it is endless, cannot have a cause. The cause of an endless per se series would also be the cause of the final effect in such a series. Every cause that produces the final effect must appear somewhere in the causal series that produces it. It must either precede the endless series or be included in it. If the cause of the final effect is not part of the per se series of intermediate causes that causes this final effect, but precedes it, the series would not be endless. If the cause of an endless per se causal series producing the final effect is included in this series, the series would include its own cause, and hence would cause itself – which is impossible. In consequence, an endless per se causal series cannot have a cause. Since an endless per se causal series must be an effect, and an effect cannot exist without a cause, an endless per se causal series cannot exist.

Caleb Cohoe assertsFootnote ¹³ that Aquinas’s argument against endless causal regresses is based on a conception of effects as ontologically dependent on their causes. This is true in a general sense, but Aquinas does not postulate an ontological dependence of effect on cause that goes beyond the dependence that an effect has on its cause simply in virtue of being its effect. This is characteristic of his metaphysical approach, which is down-to-earth and prefers to avoid appeal to abstract metaphysical generalities. He bases his argument against endless causal regresses on specific features of particular causal phenomena – the characteristics of intermediate causes and of per se causal series.

We should accept that Aquinas’s ‘lumping-together’ argument is valid for a series of intermediate per se causes that has a finite number of members. Since it is finite, such a series will have a first member which is produced by a prior cause that is not intermediate. Every individual subsequent member of the series, and the causal action of each subsequent member in producing its immediate effect, will be caused by the prior cause. Since the prior cause produces both the individual causes in the series of intermediate causes and the causal relations that exist between them, it causes the series of intermediate causes as a whole; the series as a whole is an effect. This is a consequence of the transitivity of causation in the kind of causal series that Aquinas is talking about.

In order to reject Aquinas’s lumping-together argument, one would have to argue that although it is valid for a series of intermediate causes with a finite number of members, it is not valid for an endless series of intermediate causes. One could motivate this objection by arguing that what makes the lumping-together argument valid for a finite series of intermediate causes is the fact that such a finite series must have a cause prior to it that causes it. With an endless series of intermediate causes, however, there will be no such prior cause; and therefore the factor that makes a finite series of effects itself an effect will not be present.

However, this argument will not work in Aquinas’s metaphysics. The objection specifies that it is the difference between having a prior cause and not having a prior cause that makes a finite series an effect, whereas an infinite series is not. But this difference is a purely relational property. It asserts that a finite causal series is an effect because of its relation to a prior cause, which is an entity outside of it; and that an endless causal regress will not be an effect because it lacks this relation to an external entity. Aquinas however asserts that a thing’s being an effect necessarily follows from its having certain intrinsic properties of a fairly general kind, and that all effects necessarily have those intrinsic properties. The characteristic of being an effect of course also implies a relation to something else, viz. a cause, but it is not a purely relational property. This means that we cannot distinguish between things that are effects and things that are not effects solely on the basis of possession or lack of possession of a relation to an external entity.

On Aquinas’s understanding of causation, the difference of finite vs. infinite number between a finite causal series and an infinite causal series is not something that can make finite series effects, while infinite series are not. It would be necessary to identify some real intrinsic property that a finite causal series has and that an endless causal series lacks in order to justify the claim that the lumping-together argument works for the former but not for the latter. No such property is available. If we move from a causal series with a finite number of members to an endless causal series, there remains a real difference between the members of the series, and no real intrinsic property of the series that unites their members is subtracted.

6. A single uncaused cause

When asking why Aquinas thought that the Second Way established the existence of a single uncaused cause of all effects, it is worth keeping in mind that Aquinas did not think that polytheism had ever had much of an intellectually respectable following. He considered that many pagans had in fact believed in one supreme God as the cause of all other existents, while giving the name ‘Gods’ to all immortal substances – a view that reflects the Neoplatonic sources of much of his knowledge of pagan thought. He asserted that the Manichees were the principal supporters of polytheism (S.C.G. I ch. 42) and that the presocratic Greek philosophers were the initiators of polytheism among Gentiles (S.C.G. I, chs. 20, 42; In Met. book 1 lectio 4, lectio 5). None of the causes he gives to explain the existence of polytheism provide any rational basis for believing in it (Commentary on the Apostles’ Creed, art. 1; 2a2ae q. 94 a. 4 co., ad 2). He accepted the euhemerist explanations of polytheism given by the biblical Book of Wisdom (Wisdom ch. 14), which explained it by such non-rational factors as the desire to flatter rulers. Aquinas would not have considered the possibility of there being more than one uncaused cause to be a live or pressing issue. This helps to explain why he does not address the question of the number of uncaused causes in the Second Way.

The absence in the Second Way of any argument for there being only one uncaused cause has led some interpreters to claim that the Second Way omits any discussion of the number of uncaused causes because this subject falls outside its scope.Footnote ¹⁴ On this view, the argument only concludes that every effect is ultimately produced by some uncaused cause; Aquinas seeks to establish that only one uncaused cause can exist in the extensive discussion of divine simplicity and unity that follows the Five Ways. In favour of this interpretation, it can be pointed out that Aquinas gives arguments for there not being more than one God after giving arguments for the existence of God. He does this in Summa contra gentiles book I ch. 42, Summa theologiae, 1a, q. 11, and Compendium theologiae part 1 ch. 15; the arguments for the existence of God in these works occur earlier, in Summa contra gentiles book I ch. 13, Summa theologiae, 1a, q. 2 a. 3, and Compendium theologiae part 1 ch. 3.

However, Aquinas makes it clear that he considers the Second Way to establish the existence of a single uncaused cause. His statements of the argument for an uncaused cause clearly describe the cause in question as a single cause;

In addition to flatly asserting that the uncaused cause whose existence is established by the argument of the Second Way is a single cause, Aquinas applies the term ‘God’ to this single cause. He makes it clear that this term ‘God’ refers to a single being. All of question 2 of the Prima pars, which is the context in which the Five Ways occur, clearly asserts that the questions being asked and the arguments being provided concern the existence of a single divine being. The same is true of chapters 9–12 of book 1 of the Summa contra Gentiles, which provide the context for the argument for an uncaused cause in that work. The assertion that the uncaused cause is what all men refer to as God is significant here. It is not claiming simply that the uncaused cause is in fact the same being as God: it is saying that the uncaused cause in the conclusion of the Second Way is understood as being God, that is, as being the sole creator and principle of all being.

Aquinas’s arguments against there being more than one God are thus not intended as a completion of a demonstration of the existence of a single God that is begun, but not brought to a conclusion, in the Five Ways. These arguments occur in question 11 of the Prima Pars, after seven questions in which God is spoken of in the singular and discussed as a single entity. If Aquinas had thought it necessary to complete the Five Ways by a subsequent demonstration that there is only one God, the logical place to do this would have been in question 3 on the divine simplicity, since the conclusions arrived at in that question mean that it is senseless to suppose that there is more than one God. In both the Summa theologiae and the Summa contra gentiles, Aquinas’s examination of the question of the number of Gods is given in the course of consideration of the divine unity. He provides arguments for there being only one God in question 11 because he is discussing the divine unity, and the question of the number of Gods is pertinent to the divine unity; not because he seeks to prove for the first time in his exposition in the Summa that there is only one God. One might ask why there is a need to discuss the divine unity if it is already taken as established that there is only one God. The reason for this is that Aquinas considers the divine unity as involving more than the fact that there is only one God. This unity is understood not only as a numerical fact about the number of Gods that there are, but also as a characteristic of the divine nature which excludes all internal division in the being of God. This lack of internal division is metaphysically more fundamental than the fact that there is only one God, and explains the latter fact.

To understand why Aquinas thought that the Second Way establishes that there is only uncaused cause for all the effects that exist, we should not begin with the question ‘Why hold that there is only one uncaused cause?’ Instead, we should begin with the prior question ‘Why should we consider that a number of different effects are brought about by different entities?’ We do not always start from the assumption that different effects might have different causes, and seek proof for the assertion that these effects are produced by a single cause. If we see a line of footprints in the sand, in normal circumstances we assume that they were made by the feet of one individual; we do not try to rule out the possibility of their being made by a man with only his left leg and a man with only his right leg hopping together in tandem. We ask if effects were caused by more than one thing, or assume that they are caused by more than one thing, if they are relevantly different in their intrinsic properties, in their location in time and space, or in some other extrinsic property. Absent any relevant difference, we do not ask or need to ask if they have different causes. The character of the cause in question also affects which questions we ask. Different effects may have different proximate causes but the same remote causes. Much scientific investigation is concerned with attempting to discover a single remote cause, or at least fewer remote causes, for a mass of effects that all have different proximate causes.

Obviously, there are a multitude of different proximate causes of effects. But the question when it comes to the Second Way is not whether we should consider that a number of different effects are brought about by different proximate causes. It is whether or not we should ask if all effects have more than one ultimate cause. Aquinas did not see any reason for thinking that there was more than one such ultimate cause and thus did not see any rational motivation for asking this question in the first place. This was in accordance with his principle of ontological economy, which stated that when something can be explained by a single thing, it is better to posit only that one thing rather than many things as an explanation – ‘quod sufficienter fit uno posito, melius est per unum fieri quam per multa’, S.C.G. book 1 ch. 42 n.4. That is why Aquinas takes it for granted that the Second Way establishes the existence of a single uncaused cause of all effects.

7. Conclusion

We can now see that Aquinas’s belief that the Second Way is a demonstration of the existence of God is philosophically intelligible and can be shown to follow from his metaphysical positions. Aquinas makes no pretence to originality in his Five Ways, and most of their contents have been traced to his philosophical sources, which include Aristotle, Neoplatonist Aristotelian commentators, Muslim and Jewish philosophers, and patristic and scholastic Christians. This examination of Aquinas’s thought on the uncaused cause of all effects may suggest avenues of investigation into these sources.