The typical (but not exclusively) Christian way of conceiving the creative act has been to understand it analogically as an act of divine free willing or divine free choice: God freely chooses to create this world, but God could have chosen to create another or none at all. (For the purposes of argument, ‘world’, ‘world order’, ‘universe’, and ‘creatures’ are used synonymously.) ‘Free willing’ or ‘free choosing’ (also synonymous here), as an act or capacity, is taken to be a reality in God as it is in us, something properly predicated of God albeit only analogically. (‘Proper’, ‘literal’, and ‘analogical’ divine predication are all synonymous here as well; see below.) Thomas Aquinas stands as a foremost representative of this tradition of thinking about divine creating. But this essay argues against this majority report position on creating-as-divine-free-choosing by critiquing Aquinas’ attempt to predicate free choosing (liberum arbitrium, electio, and equivalents) properly of God. Our argument in short is that, if ‘free willing’ is understood in the specific way Thomas defines it (and not in more minimal ways we might try to conceive), and if ‘God’ is understood according to Thomas’ classical theist view, then free willing cannot be properly in God. If God is the purely actual, simple, perfect reality Thomas understands God to be (Summa theologiae [henceforth ST] 1.3.1, 1.4.1; Dolezal Reference Dolezal2011), then all the essential features of free willing as Thomas defines it will have to be denied of God. For Thomas, as we will show, it is of the essence of free choosing, one of its intrinsic rationes, that an agent elects some means that conduce to the end the agent has willed. And this is precisely what God cannot do vis-à-vis creatures, not only because no finite reality can be a means conducing to God’s end (the divine goodness), but also because God simply has no means to God’s end. God ‘achieves’ God’s end simply by identically being that end, being the divine goodness. Because Thomas clearly and invariably defines free choosing as being the selection of means that conduce to ends, the divine end is something about which there simply cannot be anything like free choosing, not even analogically. But in that case, nothing intrinsic to ‘free willing’ survives to be properly attributed to God after the concept’s several rationes have been purified by the fires of remotion.
But the fact that ‘free choosing’ fails as a proper predication (and so as a genuine analogy for the divine creative act) does not mean that it cannot succeed as a metaphorical divine predication, which, in Aquinas’ understanding of divine predication, nonetheless reveals something true about how God’s willing of creatures is derivative of the act by which God wills the divine essence. For a metaphorical predication need not preserve the identical rationes between the analogates: the similarity between the two can lie instead in what is associated with the primary analogate, usually a likeness in effect. And so in these pages we critique ‘free willing’ as an analogical predication, even as we suggest that it is an important and religiously useful metaphorical way of speaking of how God causes creatures to be.
We select Aquinas as our test case for a few reasons. First, Thomas’ defence of the free-willing divine predication is among the most sophisticated and conceptually comprehensive in the Christian tradition, particularly because of his rich doctrine of human free willing that serves as the foundation for his analogical understanding of divine free willing. Thomas’ view is also representative of much of Christian (and Jewish and Muslim) thinking on the question, both before and after him (Burrell Reference Burrell1993).Footnote 1 Thus, the problems we identify in his proper predication of ‘free willing’ to God will be found, mutatis mutandis, in many other articulations of classical theist positions that God freely wills creatures. Our critique therefore suggests that the divine-creating-as-free-choosing analogy should be more widely re-evaluated by classical theists, not just by Thomists. But we have also chosen Thomas’ defence as our illustration because we broadly share Thomas’ metaphysics, and especially his philosophical doctrine of God (divine simplicity, pure actuality, divine infinity, etc.), and because he provides, with exceptional rigor, the principles that lead us to judge that ‘free willing’ cannot coherently be properly predicated of God while also maintaining a classical theist doctrine of God. Certainly the question of the internal consistency between various aspects of Aquinas’ doctrine of God (especially his doctrine of simplicity) and his position that God freely wills has been studied before (e.g., Kretzmann Reference Kretzmann and MacDonald1991a, 208–28; Stump Reference Stump2003, 100–101, 108–115; Lovejoy Reference Lovejoy1960, 73–82), and our work here builds on these analyses and debates. But our novel contribution is to focus the discussion on whether ‘divine free willing’ can pass Aquinas’ own stringent standards of proper divine attribution.
Our conclusion, then, is that classical theists should deny that ‘free willing’/‘free choosing’ so understood is in God. This does not mean, however, that the analogical predication of ‘will’ or voluntas to God is illegitimate, or that the divine will is unfree, or that ‘freedom’ ought not be properly predicated of God. It can and must be. Thomas distinguishes free choosing, as only one kind of volitional act, from voluntas as a power; so to deny free choosing of God is not to imply that God is non-volitional, or not an agent, or creates in a mindless way. All of these concerns are addressed below. The object of our critique here is simply and only the proper attribution to God of this specific kind of willing: ‘free choosing’ as understood by Aquinas.
A final word. This article does not engage directly with contemporary philosophical debates on the doctrine of divine simplicity and modal collapse (e.g., Andani Reference Andani2022; Craig Reference Craig2025; Grant Reference Grant2019; Huneeus Reference Huneeus2025; Mullins Reference Mullins2016; Nemes Reference Nemes, De Silveria and Tat2020; Schmid Reference Schmid2022; Tomaszewski Reference Tomaszewski2019) or theological discussions on the contingency or necessity of God’s creative act (Hart Reference Hart2022, 104–105; Wood Reference Wood2022, 76–77; Rooney Reference Rooney2024). But it bears implications for these debates, for if Aquinas’ doctrine of God is one that contemporary classical theists find representative, and if we are correct in our critique of his proper predication of free willing to God, then this poses a challenge for theists who claim the mantle of Aquinas’ doctrine of divine free willing as an analogical account of divine creating.
We proceed as follows. First, we briefly analyse Thomas’ remarks in the Summa Theologiae on proper (analogical) divine predication and metaphorical divine predication. We next examine Aquinas’ understanding of divine willing in its two aspects: necessary and natural willing, and ‘free’ willing, the willing that is through liberum arbitrium. The following section analyses the four central rationes of the concept of free willing in finite agents as Thomas understands it, and we then explain Thomas’ own (in our view inadequate) remotion of the concept when applying liberum arbitrium to God. Next, as the heart of our argument, we identify four ways in which Thomas’ doctrine of God must, if applied in a proper remotion of free willing, evacuate divine free willing of all the essential features shared with finite agents’ free willing. This leaves the attribution of free willing to God ultimately equivocal, and therefore not proper. As a consequence, we suggest that Thomas’ attribution of free willing to God is better treated as what Thomas calls a metaphorical predication, a form of meaningful equivocation (see the next section). Speaking of divine creating metaphorically as free willing still communicates an important difference between how God wills the divine essence and how God wills creatures, but it does not establish that God in fact performs acts of ‘free choosing’. We then conclude with some reflections on how the denial of ‘free choosing’ to God impacts a classical theist conception of divine freedom.
Proper substantial divine predication
In ST 1.13.2.c., Thomas states that the names that speak of God’s relation to creatures in no way refer to God substantially; rather, they speak of the creature’s relationship to God. Names such as ‘creator’ fit the bill here. To say that God is creator, therefore, is not to predicate anything substantially of the divine essence but instead to predicate of creatures that they originate from God. But when it comes to positive names that contain no reference to creatures in their very concept, Thomas argues that these terms indeed refer to the divine essence, to God as God is in Godself. Thomas excludes two interpretations of how the substantial attribution of positive names to God works. First, he rejects Maimonides’ view that these positive names, contrary to what the speaker seems to intend, in fact say only something negative. If we predicate that God is alive, argues Maimonides (in Thomas’ telling), then we mean only that God is not like inanimate things. This position maintains that positive predications are nothing but remotion, removing a predicate from God. Thomas denies this position. The second interpretation is that positive names of God, such as ‘living’, tell us merely that God is the cause of life in creatures. In other words, this position holds that divine predications speak primarily of creatures and only secondarily of God. This too Thomas rejects: When we speak of God analogically with proper names, we are doing more than just talking about creatures in their relationship to God (ST 1.13.6.c).
The foundation for our capacity to speak of God analogically, and so to properly predicate anything of God, is in the relation creatures have to God as their cause. As cause of creatures, God must possess all the perfections of creatures, yet in a supereminent mode (ST 1.4.2.c; Rocca Reference Rocca2004, 49–76, on the Dionysian background; Te Velde Reference Te Velde2006, 85–90). But in Thomas’ account of proper or literal divine predication, two further cognitive operations must be performed in moving from creatures to God: remotion and eminence (see Barnes Reference Barnes2014, 436–442 for an overview). In remotion, one removes from the predicated concept whatever is bound up with materiality, finitude, and potency per se. If the creaturely perfection is a pure perfection that exists in God, it does not exist in God in that ‘imperfect mode proper to creatures’ (ST 1.14.1.ad1). In predicating according to eminence, we understand these perfections to exist analogically in a higher mode in the divine essence, and as the divine essence, yet in a manner we cannot intellectually comprehend or express adequately in language (Wippel Reference Wippel2000, 519–20). Therefore, all literal or proper divine predication will be analogical predication.
The triplex via of causation, remotion, and eminence cash out differently with distinct names applied to God in scriptural revelation, depending on whether the substantial attribution should be taken as metaphorical-equivocal or proper-literal-analogical (ST 1.13.3). Proper-literal and metaphorical attributions are mutually exclusive (Wippel Reference Wippel2000, 507; on analogy and metaphor in Thomas, see McInerny Reference McInerny1996, 116–137; Rocca Reference Rocca2004, 318–322). Although we rightly say that ‘the Lord is my rock’ (Psalm 18:2), God cannot properly (proprie) be called a ‘rock’, because materiality is intrinsic to the substance of a rock. As a creature, a rock receives the perfection of ‘substance’ in this deficient mode of corporeality, and the very concept of ‘rock’ expresses this creaturely mode of receiving the perfection (ST 1.13.3.ad1). Corporeality is part of a rock’s core intelligibility (ST 1.19.11.c). But because remotion requires removing everything intrinsically bound up with corporeality, if ‘rock’ were predicated to God properly (again, not metaphorically), everything of ‘rock’s’ intrinsic intelligibility or conceptuality as material substance would have to be remoted. Accordingly, there could be nothing left to attribute to God supereminently in a proper analogical predication. In such an instance of total remotion, the predication becomes equivocation, not analogy, for analogical predication requires that the rationes of the distinct analogates not be wholly diverse (ST 1.13.5.c; Wippel Reference Wippel2000, 569–570). There must be a ‘floor’ below which the remotion of the concept does not go – negatively so that the distinction between analogical and equivocal predication does not disappear, and positively so that the predication according to eminence can launch.
But none of this precludes us from predicating ‘rock’ of God metaphorically (metaphorice). Such metaphor is an equivocation, Aquinas writes (see Aquinas, In V Metaphysic. lect. 14, §974; In IX Metaphysic. lect. 1, §1773 [McInerny Reference McInerny1996, 123–124, 132], where Aquinas refers to the ‘potential’ of the number 3 to become 9 as ‘metaphorical’ and ‘equivocal’, respectively), but not completely so, to the point where the predication would become pure paronym. Such metaphorical divine predications, despite not being ‘proper’ and ‘analogical’, are still religiously useful and revelatory (ST 1.13.3.ad1, ad3; see Scriptum Super Sententiis II d. 13.1.2.c for another instance of Aquinas identifying metaphorical and equivocal speech). This metaphorical divine attribution can be justified by some likeness between the metaphor’s referent and God, usually a likeness in effect (McInerny Reference McInerny1996, 134; see ST 1.13.6.c, 1.19.11.c; Aquinas, Summa contra gentiles [hereafter SCG] 1.92.11; Aquinas, De potentia [henceforth DP] 7.5.ad8.). As rocks produce stability, so God produces stability; therefore the Psalmist speaks rightly of the Lord as a ‘rock’ (ST 1.19.11.c), even though there is nothing of the intrinsic conceptuality of ‘rock’ as material substance that can be applied to God in Godself. This equivocation-as-metaphor nonetheless speaks substantially (although not properly) of God and is fruitful, as long as the hearer keeps in mind its form of usage. Still, to predicate of God ‘rock’ metaphorically on the basis of a likeness in effect will not alter the fact that a creature remains the primary instance of the term (ST 1.13.6.c). ‘Good’, on the other hand, is said of God properly, for the concept of ‘good’ implies no necessary corporeality, potentiality, or finitude, and God is the primary instance of the term (ST 1.13.2.ad2; SCG 1.30.2; Aquinas, De veritate [DV from here on] 2.11.c). By removing finitude or potentiality from ‘good’, one is not stripping the term of its core intelligibility.
Aquinas applies these same three operations (causality, negation, and eminence) to predicate properly of God names referring to rational powers and their acts. Since neither the concept of ‘will’ (voluntas) nor that of ‘understanding’ (intellectus) entails corporeality, potentiality, or finitude, these may be predicated properly of God. To do so, one removes whatever belongs to imperfection to head towards an analogous concept of these perfections in God. Central to this remotive operation is Thomas’ doctrine of divine simplicity, derived as a necessary consequence of his definition of God drawn from his natural theology (e.g., ST 1.3). Because there is no composition whatsoever in the divine essence, if God is said to have intellectus, then to predicate intelligere of God one must also affirm that God’s operation of understanding is identical with God’s power of understanding as well as with God’s essence. Further, one must say that what God primarily understands is identically the essence doing the understanding, and that this understanding is fully actual, with no potential. If in a finite intellectual substance an act of understanding is a passage from potency to act and thus not identical with that substance’s esse, this motion is denied of God. We thereby purge intelligere in God from any suggestion of finitude. Then by predicating of God an intellectus that is identical with God’s esse, we praise God by supereminently attributing the most excellent form of understanding to God – and not only to God, but also as God, which is to say that God will neither be composed with nor participate in the perfection of understanding but will simply be that perfection (Wippel Reference Wippel2000, 519–520). Again, in analogical divine predication, God serves as the primary instance of the perfection in which creatures merely participate in some restricted fashion (ST 1.13.3.c). Ultimately, it is our knowledge of what cannot be true of the divine nature per se – knowledge drawn from the results of Aquinas’ arguments for the existence and attributes of God – that allows us to judge whether a given predication can be literal-proper-analogical or equivocal-metaphorical (Rocca Reference Rocca2004, 319).
The rationes of liberum arbitrium for Aquinas
Here, we present the core concept of ‘free willing’ in Aquinas. To do so, we begin by discussing voluntas as a rational power in Aquinas’ account, before turning to the particular act of voluntas that Aquinas identifies as liberum arbitrium. Voluntas is a rational power that follows upon possessing an intellect, for in understanding being with the qualification of good, one inexorably desires it or is inclined towards it (ST 1.19.1.ad1 and SCG 1.72; Knasas Reference Knasas2002, 423–424; Shields and Pasnau Reference Shields and Pasnau2016, 130–137). The formal object of the will is the good, and agents will the good necessarily because willing is by definition ordered towards the good (DV 22.6.ad5; ST 1-2.8.1.c). Properly speaking, voluntas or ‘willing’ in terms of its act concerns ends, not means (SCG 1.88.4; this is why choosing, or electio/eligere, is not an act of the will when the latter is considered absolutely [DV 22.15.c] – see below). But with respect to voluntas considered qua power, its object can also include the means selected for the end (ST 1-2.8.2.c.). Voluntas in any intellectual agent is a ‘natural’ and necessary inclination towards its proper end (DV 23.4.c); it moves towards an object suited to its nature and rests in delight upon attaining it.
For an intellectual nature, the necessity of this willing-as-natural-inclination is a formal and natural necessity. This natural necessity is not a violent necessity imposed by an external agent (necessitas coactionis). Formal, natural necessity of this kind is not violent because nothing natural to a thing, nothing in principle intrinsic to its very intelligibility, can be violent or unnatural to it (DV 22.5.c; ST 1.82.1.c; DP 10.2.ad5). Just as a triangle must have three sides by the intrinsic necessity of its nature, so too must natural voluntas will an end suitable to the agent’s nature (ST 1.82.1.c; cf. DV 23.4.c.). The natural end we necessarily will is happiness. Importantly for our argument, our inability to do other than will our happiness is not repugnant to voluntas for Thomas (ST 1.82.1.c), and neither is such natural necessity contradictory to our will’s being free: ‘the natural necessity under which the will is said to will a thing of necessity – happiness, for instance – is not incompatible with the will’s freedom’ (DP 10.2.ad5, emphasis ours: Naturalis necessitas secundum quam voluntas aliquid ex necessitate velle dicitur, ut felicitatem, libertati voluntatis non repugnat; translation revised). Analogically, in God’s natural willing, God necessarily wills the divine goodness because the divine goodness is the divine will’s sole proper object, the only end proportionate to the divine nature. God cannot not will the divine goodness: God does not choose freely to will the end of the divine goodness but instead does so with natural necessity. This, again, is no mark against the freedom of the divine will, nor is it evidence of an extrinsic coercion of God, for necessarily willing the divine goodness as end is just what it means for God to have a rational will. This point is important for properly interpreting the conclusion of our argument. If we are correct that Aquinas’ doctrine of God entails that free willing is logically impossible for God – just as it is impossible for God to grow or to learn – then denying that free willing is properly in God does not imply that the divine will is not essentially free.
In predicating natural, necessary voluntas to God analogically, Aquinas applies the requisite techniques of remotion and eminence. In finite agents, what is willed naturally is happiness, and yet no finite agent by nature possesses or is identical with its happiness. So we move towards happiness – but God does not (ST 1.82.1.c). Thus, if we attribute natural willing to God, then God by nature and necessity wills one equal and adequate object, an object with which God’s willing must be ontologically identical due to divine simplicity. Granted that the intelligible structure of finite rational willing is movement towards the good and then delight in the good once possessed, then only this delightful rest, and by no means movement, can be predicated of the simple God (ST 1.19.1.ad2). Simply put, God’s willing of the divine goodness involves no goal-directedness; God has no goal to attain or realize. What remains once the notion of ‘natural willing’ has been analogically purified is therefore God’s will as ordered to and delighting in the infinite good that God is. Necessarily willing this one, equal object must be identical with God’s already possessing and resting in that adequate object. For both simply are the divine essence, and the divine willing and divine essence are distinguishable only in thought.
Now to liberum arbitrium (free willing or free choosing), which Thomas synonymously calls electio (from eligere; compare the easy exchange between eligere and liberum arbitrium in DV 24.1). Thomas discusses free willing, both in God and in finite rational agents, in many places, so our presentation here must be selective, focusing only on those aspects of free willing that are critical to the analogical predication of free willing to God. This presentation is furthermore synthetic, and necessarily so, for it attempts to unearth the deep structure of Thomas’ thought on what constitutes free choosing. On Thomas’ account, free willing is the willing through which an agent elects (eligere, electio) some means over others ‘to the end’ (ad finem) that is willed (SCG 1.88.4). ‘Free willing’ thus picks out a particular kind of act that intellectual agents perform; it is to this specific act that we refer below when we use the phrase ‘free willing’ or ‘free choosing’, and not to the fact that the will is, formally speaking, free. Even in natural and necessary willing, the will is free, according to Thomas. When speaking more loosely, we may say this act of free choosing proceeds ‘from’ liberum arbitrium, although there is no faculty of liberum arbitrium really distinct from the faculty of voluntas or will (DV 24.6.c).
Let us break down in some further detail the central components of free willing in Aquinas’ account as we have stated it above (viz. the willing that elects some means over others ‘to the end’ that is willed). First, ‘the reason for choosing something [as means] is because it conduces to the end’ that has been willed (ST 1-2.13.5.c: Ratio eligendi aliquid est ex hoc quod ducit ad finem). This is the rationale for choosing anything as means. Because finite agents are not identical in being with their end (as God is), the end that motivates and provides the rationale for their choosing, they move towards attaining that end by choosing means. To clarify, when Thomas says that an agent acts to attain or achieve an end, it does not mean that this end must be in every instance their own benefit or improvement; he simply means that free action is for an end not already possessed. And so finite agents exercise free willing whenever they choose some means over others to attain a penultimate end distinct from the final and natural end of happiness, although this choosing is always under the formality of seeking that natural end (thereby rendering every proximate end a means to our final end; see ST 1-2.1.6.c; 1-2.13.3.c). So central is the means-end schema to Aquinas’ account of free choosing that when he argues for the reality of liberum arbitrium in humans but not in animals, he appeals to our capacity to understand the meaning of an ‘end’ and a ‘means’ to that end, which enables us to pass judgement on our own free judgements about whether to act or not to act (DV 24.1.c).
Furthermore, because free willing involves ordering means to ends, the exercise of this electio always presupposes a prior intellectual act. This act is reason’s act of consilium, which orders means to ends by understanding that a particular means conduces to the set end. That the means do in fact conduce is what is meant when Thomas speaks of the ‘proportion’ of a means to an end. Most basically, consilium is that intellectual act that both understands means as means (in relation to an end; ST 1-2.12.1.ad3; 1-2.12.5.c; 1-2.14.1.c) and deliberates about the best means to the end desired. In the volitional act of approving of and desiring the various means proposed by reason for achieving the end in consideration, the agent consents (consentire) to the means (ST 1-2.15.3.c). If there is more than one viable means, then consenting to any means as in fact desirable to attain the proposed end is, in finite agents, really distinct from electing one of those means over another (ST 1-2.15.3ad3). But when there are multiple good and possible means understood in consilium, then logically prior to the choosing (electio), there occurs a liberum iudicium, in which we judge that the means which reason has proposed to the will as the best (in whatever aspect of the object is currently under consideration) is in fact so in comparison to others: ‘election implies a previous comparison’ (ST 1-2.13.1.ad1: Electio importat collationem quandam praecedentem; cf. DV 22.13.ad16: electio est actus voluntatis in ordine ad rationem comparantem ea quae sunt in finem ad invicem). The agent then freely chooses that (perceived) best means as ordered to the willed end (ST 1-2.12.4.ad3 and 1-2.13.1.ad2). In every instance of free choosing, we desire and choose the means because we understand it to produce for us the desired end: this is simply what it means for something to be understood by the intellect as volitional ‘means’ – a point to which we will return several times below.
In cases where there is only one means to an end, we choose with the ‘necessity of the end’ (necessitas finis). In such instances of choosing, the means are the only ones proportioned or adequate to achieving the end (DV 23.4.c; Wippel Reference Wippel and Wippel2007, 232). If I will to cross the sea, I must (in the thirteenth century) will to board a ship (ST 1.82.1.c). Yet in any instance where multiple distinct means are adequately proportioned to achieving the end, we are said to choose those means by free willing or electio. For example, we may freely choose the means of a horse for a journey we can also make on foot (ST 1.19.3.c).
Thomas’ basic and repeated analogy for predicating liberum arbitrium to God is this: God wills the divine goodness as God’s end (and does so with necessary-but-free voluntas), and God freely chooses creatures as means adequate ‘to the end’ (ad finem) of the divine goodness (variants in ST 1.19.3; SCG 1.81.3 and 1.88.5; DV 23.4.c and 24.3.c; cf. SCG 1.82.c.9). But because no particular created means are necessary for God to possess God’s end, God’s willing of creatures must be an act of free willing. God could choose this universe, an infinite number of other created universes, or no universe at all.
Now, insofar as free choosing presupposes a nature and a cognitive power, the particularities of God’s nature and cognitive power will determine Aquinas’ analogical predication of free willing to God. First, because God’s nature is the divine goodness in its perfection, divine free willing cannot tend towards evil as creaturely free willing can. Second, because the divine cognitive power lacks discursivity (because it has no potential to actualize), God’s free choosing will be exempt from deliberation about means. That means it will be exempt from consilium in the sense of working to discover means or wavering between alternatives once discovered (DV 24.3.c, ad4; see below), even though it will still presuppose a divine understanding of select means as adequate to God’s end. Divine liberum iudicium, Aquinas concludes, is thus a judgement to freely choose ‘this or that’ means – for free willing requires indifference towards options – simply in accord with the divine goodness that could be manifested otherwise and that would be complete in itself without any created manifestation at all (DV 24.3.c). As such, God exercises free will by creating this world, creating another, or not creating at all (SCG 1.81.3). And so we can say that God, angels, and humans all share in liberum arbitrium (DV 24.3.c; compare Aquinas, De Malo 16.5.c) in their analogically distinct modes.
Why divine creating cannot properly be free choosing in God
So the components of free choosing in finite rational agents are as follows: (i) Grasping the proportion of a means to the end willed (consilium); (ii) selecting (electio) a perceived best option; and (iii) a motivating final cause or rationale for the free choice, namely, an end, (iv) towards which the means are ordered (ad finem) in consilium and electio. These are the rationes of free willing in creatures. If Aquinas is going to properly and not just metaphorically attribute free willing to God, some one of these elements needs to survive the remotion.
The problem for this proper predication, however, is that Aquinas’ doctrine of God effectively remotes all these integral components from divine free choosing. If even one element were retained, a case could be made that Aquinas has merely purified ‘free willing’ in properly predicating it of God. But Aquinas’ rigorous doctrine of divine perfection leaves nothing here untouched. For in his account there is (i) no possibility of God’s understanding finite realities as proportionate volitional means to attain God’s end; (ii) no best option for God’s free willing of creatures; (iii) no motivating rationale for God to will creatures; and (iv) a fallacious equivocation between volitional and ontological final causality when explaining how creatures are ‘ordered’ to God. But to empty God’s free willing of all the essential features of finite free willing renders the attempted analogical predication equivocal. This equivocation still functions as a valuable metaphor that indicates something true about divine creating, as we detail in our conclusion, but it is Thomas’ equivocation on the conceptuality of ‘free choosing’ that we exposit in these next four subsections.
No proportionate means to the divine end: The logical impossibility of divine consilium’s grasp
In free choosing, an agent is moved by an end to understand possible means as proportioned to that end. This is the activity of consiliari. Now, Aquinas explicitly denies that consilium or consiliari is essential to electio as such, insofar as consilium involves enquiry (inquisitio) or reason’s deliberation (deliberatio) and this kind of movement from ignorance to knowledge is proper to rational natures like humans, but not to purely intellectual natures, of which God is the supereminent instance (DV 24.3.ad4; on consilium generally, ST 1-2.14). We grant with Aquinas that discursive enquiry or deliberation of reason is accidental to free choosing, but in this section we refer to the first and most basic result of consilium, namely the intellectual content grasped by the deliberating agent: the means understood as proportioned to one’s desired end, which is just to say volitional means qua volitional means. This aspect of consilium is presupposed in every act of electio, because electio chooses among means already understood by the rational agent as means to (as ordered to; ST 1–2.12.1.ad3) the desired end. And so without the result of consilium, there can be no electio. If God is to freely choose this universe, another, or none as means, then God still must understand each of those options as means appropriately proportioned to God’s end, even if God understands this nondiscursively.
Now, Aquinas holds that voluntas cannot be moved to the means qua means unless moved towards the end (ST 1-2.8.3.c; cf. DV 22.5.c and ST 1.83.4.c). The agent is first moved to those means she discovers in reason’s act of consilium that precedes (if only logically) and is presupposed in her exercise of electio. That is, means are only recognized as means if the intellect understands them (rightly or wrongly) as proportioned to the end that is willed (DV 22.13.c; cf. DV 22.14.c). This involves either the proportion of necessitas finis, when there is only one means to a desired end, or the proportion of means to end in free choosing, when the agent comprehends that other possible means could conduce to the same end (DV 23.4.c; ST 1.19.3.c).
And so, there are two kinds of proportion that can be grasped in the agent’s act of consilium. But on Thomas’ own account, neither of these proportions can apply to God’s free willing. Due to his account of divine transcendence and perfection, there can be no proportion whatsoever, as volitional means conducive to the divine end, between any world order and the divine goodness, precisely because of the sui generis, transcendent nature of that divine end. ‘[T]the divine goodness is not such an end as could be produced from what is for the sake of the end’ (Non enim divina bonitas est talis finis qui efficiatur ex his quae sunt ad finem; DV 23.4.c; translation revised). If this end (which is God’s end in all divine willing) cannot be produced or attained by means, then it follows that there can be no divine consilium about creatures. For no world order can be understood by the divine intellect as possible volitional means to the end of divine goodness in God’s free willing, since means are so grasped only as bearing some proportion to the end willed. It is not just, as Thomas says in DV 23.4.c., that creatures are volitional means to God’s end that are ‘proportioned to the end’ (fini proportionatum) of the divine goodness, just not ‘perfectly’ (perfecte) proportioned. In this volitional context of choosing, ‘perfectly proportioned’ means for Thomas that what is chosen would be the only means by which the agent may attain the agent’s end (something chosen with necessitas finis).Footnote 2 The issue is rather that, because of the transcendent perfection of the divine goodness as end, Aquinas ought to deny to created effects any possible volitional proportionality to the divine essence, for by definition, finite realities cannot be understood as a means to achieve or attain infinite divine goodness. That is, to be truly created goods, these goods must be infinitely disproportionate in value to the unrestricted value of the divine goodness, goods that do not contribute to, comprise, or enhance that goodness. ‘[T]the divine goodness is perfect in itself, even if no creature existed, for he does not need our goods’ (DV 23.4.c).
Thomas communicates that there is no such volitional proportion between God and creatures anytime he states that God’s willing the divine end (which, again, means not desiring but resting in delight in the divine essence) does not entail God’s willing any particular world or any world at all. If it did, this would indicate that some proportion could be grasped between the transcendent good as end and finite good as means to it, just as in willing to eat lunch I grasp the proportion between that end and the various food items in my fridge. In fact, this appeal to the ontological disproportion between any world order and the divine goodness is a crucial feature of Thomas’ argument in ST 1.19.3.c that God freely, and not necessarily, wills creatures. Here Aquinas both invokes the infinite disproportion of created goods to God’s goodness – in order to show that since God need not will any creatures at all with necessitas finis, God only wills them freely – and yet also speaks of God’s freely choosing creatures (or not) on analogy with someone who, wishing to travel to a nearby place, might choose either walking there or riding a horse, two means that are proportioned to the end of making the journey or arriving at the destination, and that do genuinely conduce to it:
Now in willing an end we do not necessarily will things that conduce to it, unless they are such that the end cannot be attained without them; as, we will to take food to preserve life, or to take a ship in order to cross the sea [i.e., instances of choosing with necessitas finis]. But we do not necessarily will things without which the end is attainable, such as a horse for a journey which we can take on foot, for we can make the journey without one. The same applies to other means. Hence, since the goodness of God is perfect, and can exist without other things inasmuch as no perfection can accrue to Him from them, it follows that His willing things apart from Himself is not absolutely necessary (ST 1.19.3.c).
We are not arguing against Thomas’ conclusion that divine creating cannot be an instance of choosing with necessitas finis, but notice the contradiction here. Either there exists some proportion between the end and the possible means to it (as horse and walking to completing a journey), or there does not. To establish his analogy that God freely wills creatures, Thomas invokes the utter disproportion of all created means to the perfect divine end, but takes it merely to establish that no creatures bear necessitas finis to God’s end, and thus that God may still ‘freely choose’ vis-á-vis them (go by horse, or go by foot; choose this world, that one, or none). But what Thomas’ understanding of divine perfection really entails is that creatures cannot be a means to God’s end at all, both because of their infinite disproportion to God’s perfection (their not contributing to, comprising, or enhancing it), and because God already perfectly possesses God’s end, and so can have no ‘means’ to it at all. By Thomas’ own principles, this divine ‘free choosing’ of creatures is essentially and radically nothing like the way a traveller freely chooses between going by horse or by foot, two means that genuinely are proportioned to the end, and genuinely do conduce to it.
Again, the disproportion of creatures to the divine goodness does in fact make it logically impossible that God could choose creatures as means with necessitas finis, where God must choose creatures to achieve or reach the divine end; here, we agree with Aquinas. But the fact that God cannot choose creatures with necessitas finis (as necessary means to the divine end) does not thereby automatically generate the conclusion that God must freely will creatures as means to the divine end. Instead, divine perfection (God as agent does not need to attain God’s end) entails that God cannot freely choose creatures either, because to freely choose them would still require them to be means to achieving the divine end – just one among many such means. Aquinas’ understanding of divine goodness as the end of divine willing renders it logically impossible that the divine intellect might understand some proportion between means and end in consilium (again, without any need for divine inquiring to discover or divine deliberating about these means). But without consilium about means there is simply no electio of means: ‘Insofar as the movement of the will is to the means, as ordained to [because understood as proportionate to] the end, that movement is electio’ (ST 1-2.12.4.ad3, translation revised: Inquantum motus voluntatis fertur in id quod est ad finem, prout ordinatur ad finem, est electio). This is just what electio means – choosing something understood as means for attaining an end. To say that God is ‘choosing’ creatures, but not as a means conducing to God’s end, is just to say that God is not choosing them at all, on Thomas’ definition of choosing.
If this line of reasoning is correct, then ‘free choosing’ when applied to God has been purged of a constitutive element by virtue of Aquinas’ doctrine of divine perfection. Arguably, the remotion of this element on its own renders any analogical attribution of free willing to God untenable, insofar as all electio presupposes a grasp of means as means to some desired end. But let us see whether Aquinas’ understanding of divine free choosing can preserve some of the other rationes of free willing.
No best option: The problem of divine equipoise
In Aquinas’ account, God freely wills one world order from among an infinite number of such possible orders. This is not to imply that these possible orders exist in some (actual) ‘realm of the possibles’, either in the divine mind or beyond it (Zedler Reference Zedler1948, 139–144). It is merely to say that we recognize that God could have willed orders other than this one. For if it is correct to say that God can do other than God does in relation to the divine power considered absolutely (and not as bound conditionally by divine foreordination and knowledge [ST 1.25.5.c, and especially ad1]); and if God necessarily knows all possible finite realities by knowing the infinity of the divine power (ST 1.14.9.c.); then these two judgements together entail that God could have willed another world. Although God can always make a world better than the last with respect to its intrinsic excellence (‘God could make other things, or add something to the present creation; and then there would be another and a better universe’: ST 1.25.6.ad3), what all possible orders share is their finitude in comparison to the infinite divine essence that transcends them all equally. Here lies the problem: if these orders are all equally disproportionate to the transcendent divine essence, then none possesses any quality making it more choiceworthy to God as a finite representation of that essence (DP 1.2.c; ST 1.25.6; DV 23.4.c). This presents a problem for analogically attributing electio to God, insofar as electio for Thomas is in its essence comparative.
For Thomas, free choices are always intelligible in a very particular way. Specifically, they are intelligible insofar as rational agents choose what they choose because they understand it as good, that is, as a fitting means to one’s desired end (ST 1-2.8.1.c). Choosing is thus always under the aspect of the good as understood by the intellect (ST 1-2.1.6.c). But as Aquinas writes – crucially for our argument – when one option is chosen from among multiple good options, it will be elected over others because it is understood as better in some fashion: ‘Electio includes something that consent has not, namely, a certain relation to something to which something else is preferred’ (ST 1-2.15.3.ad3, translation revised: Electio addit supra consensum quandam relationem respectu eius cui aliquid praeeligitur; cf. DV 22.15.c). Thomas dispenses with the possibility of true ‘equipoise’ for the will (a problem first reported in Aristotle [Leggatt Reference Leggatt2015, 159]). That is, he denies that two options truly could be wholly identically attractive to an agent: ‘If two things be proposed as equal under one aspect, nothing hinders us from considering in one of them some particular point of superiority, so that the will has a bent towards that one rather than towards the other’ (ST 1-2.13.6.ad3, emphasis ours: Si aliqua duo aequalia proponantur secundum unam considerationem, quin circa alterum consideretur aliqua conditio per quam emineat, et magis flectatur voluntas in ipsum quam in aliud). For Thomas, then, the equipoise thought experiment overlooks the crucial fact that in free choosing reason has already compared and proposed some option as better in some respect.
Let us clarify that for Aquinas free choosing is not efficiently caused or necessitated by reason’s proposing some option as best. It is not ‘automatic’ in any fashion, precisely because the intellect can consider any object besides beatitude as deficient in some respect and so as not fully choiceworthy, obviating any automaticity of choice. This is true even with choices characterized by necessitas finis: if I will life, I must will food – but I can instead decide it would be better to die. So too can a person simply choose not to consider an object that reason proposes as choiceworthy in some fashion (ST 1-2.10.2.c; ST 1-2.13.6.c; De Malo 6.c). But any free choice of any object as means is nonetheless informed by reason’s proposing that means as better in some respect, and absent this comparative act of reason, there is no electio. This is simply part of the rational structure of free choosing for Aquinas:
[C]hoice is not an act of the will considered absolutely but rather in relation to reason, because there appears in choice what is proper to reason, namely, to compare or to prefer one thing to another, which is indeed found in the will’s act from the influence of reason inasmuch as reason itself proposes something to the will, not as useful simply, but as more useful to an end (DV 22.15.c, emphasis ours, translation revised: Est tamen electio actus voluntatis non absolute, sed in ordine ad rationem, eo quod in electione apparet id quod est proprium rationis, scilicet conferre unum alteri, vel praeferre: quod quidem in actu voluntatis invenitur ex impressione rationis, in quantum scilicet ipsa ratio proponit voluntati aliquid non ut utile simpliciter, sed ut utilius ad finem…eligere est actus voluntatis, secundum quod ratio proponit ei bonum ut utilius ad finem).
In this light, even choosing not to choose between two options is itself a choice that will be undertaken because reason has proposed that, in this instance, not choosing is better than choosing. And there is no danger of infinite regress here, for consilium has a finite starting and end point (ST 1-2.14.6.c).
Although in Aquinas’ view the problem of equipoise cannot obtain for finite agents, it seems that it must obtain in the case of God’s willing of creatures. This is because a ‘better’ choice cannot be presented for divine choosing, for God can always will a better world from the infinity of possible worlds God understands. And so God’s creation of this world over these other worlds appears unintelligible when analysed as a free choice. (N.b. we are not arguing that divine creating is unintelligible per se, only that it is unintelligible when we consider it under the formality of ‘free choosing’.) Why does God choose this world, and not another, or another better, to be, even though that too is within the remit of the divine omnipotence? For the reasons just surveyed, the answer cannot be that God has God’s good reasons, even if they are unknown to us. Neither can the question be dodged even if Thomas held that God freely wills all possible world orders (a position he explicitly denies: SCG 1.81.4). For as seen above, God equally exercises liberum arbitrium in not creating, and so even if God willed all worlds to exist, the question then transforms into why willing them all is a better choice than not so willing (see ST 1.19.10.ad2). Michael Liccione (Reference Liccione1995, 231–233) acknowledges that on Aquinas’ view there can be no greater reason for God to will to create than to not, but he thinks it does no injury to the coherence of Aquinas’ position on divine free willing, because in Liccione’s view, not all choice of means need include a contrastive judgement that one option is better than another (Huneeus Reference Huneeus2025, 97–97 argues similarly). Whatever the truth on that question, it is clear that on Aquinas’ account, contrastive motivation is central to free choosing. So to give up on contrastive explanation for choice is just to deny another constitutive element of free choosing for Aquinas, and so to render the divine choice of this world arbitrary according to Aquinas’ understanding of free choosing. On that understanding, to ‘choose’ something without the intellect presenting it as in some fashion better is just not to choose it at all, for on Aquinas’ view, an ‘arbitrary choice’ is a logical contradiction.
The transcendent perfection of the divine essence over any finite likeness of it thus renders incoherent, in Thomas’ view, the notion of a ‘best possible world’ (Kretzmann Reference Kretzmann and MacDonald1991b), as well as the notion of ‘divine free choosing’ to which it is connected (compare Daeley Reference Daeley2022 for a contemporary defence of a ‘best possible world’ for divine choosing). But this fact makes a divine ‘free choice’ of this world profoundly unlike a free choice made by finite agents. Aquinas’ doctrine of God thus effectively remotes from the notion of liberum arbitrium another of its central components: that free choosing always occurs as informed by reason’s proposal of what appears best.
No motivating end as rationale: Defusing the Dionysian principle
Grant that God’s choice to create (this world) lacks the typical intelligibility of electio, namely reason’s consideration of one option as better than another. Allow, too, the force of the argument above that there can be no divine understanding of a world as a proportionate means to attaining the transcendent divine end (= no consilium), and so no movement of the divine will to the means as a function of a divine intending of the end, i.e., desiring the end as acquired by means (ST 1-2.14.4.c). Still, this need not imply that there are no grounds for properly predicating free willing of God. There remains another essential feature of free choosing, namely that choice is for the sake of an end, that it has a motivating or final cause. If the motivating rationale or final cause of creating can be found to survive the remotion required by Aquinas’ classical theism, then the divine proper predication of ‘free willing’ may be justified. And it seems there is such a motivating rationale for divine creating: to communicate divine goodness.
After all, volitional act per se is not always ordered to attaining an end; that is true only of imperfect (i.e., metaphysically complex) agents. But God is a perfect agent, and so God acts only to communicate the divine goodness, not to achieve it: ‘But it does not belong to the First Agent…to act for the acquisition of some end; He intends only to communicate his perfection, which is his goodness’ (ST 1.44.4.c; compare Aquinas, On the Divine Names 4, lect. 5, §352). In the light of that rationale for divine willing of creatures, is not this world adequate, or proportioned, to that goal of communicating God’s goodness, insofar as creatures are finite representations of divine goodness? In that case, there appears to be no problem in speaking of a world as a fitting ‘means’ freely chosen by God for the goal of communicating or manifesting God’s glory and goodness, even if we cannot speak of God choosing created means in order to attain the end of being God. It is the nature of divine goodness, in fact, to communicate itself: call this (with Kretzmann Reference Kretzmann1996, 223) the ‘Dionysian principle’, after Ps. Dionysius’ development of the Platonic theme of the self-diffusion of the good.Footnote 3
Thomas effectively blocks this proposal, both explicitly in his account of the divine end and implicitly through other aspects of his classical theism. The reason why communicating divine goodness to creatures cannot be the final cause of God’s choice of creatures is twofold. First, the final cause of any possible divine free choice must be identical with the divine end of God’s necessary willing: God does not have two ends, unless God is metaphysically composite. But Aquinas states categorically that communicating the divine goodness is not God’s final end: ‘The ultimate end [for God] is not the communication of the divine goodness, but the divine goodness itself’ (DP 3.15.ad14, emphasis ours: Communicatio bonitatis non est ultimus finis sed ipsa divina bonitas; cited in Blankenhorn Reference Blankenhorn2002, 834). This immediately disqualifies ‘communicating divine goodness’ as a motivating rationale for a divine choice to create.
Second, Thomas’ classical theism, particularly his account of divine perfection, requires him to deny that communicating divine goodness could be a motivation for God freely choosing creatures. The divine goodness is not enhanced or realized by communicating itself to or in other things, for that would imply that the divine goodness were not already maximally and supereminently communicated internally in God (Aquinas, Scriptum Super Sententiis 1.2.1.4, cited in Blankenhorn Reference Blankenhorn2002, 819). If in order to diffuse divine goodness God must create a world (or an infinite number of them), then to be fully expressed – which means, to fully realize its self-diffusive nature as the good – God’s goodness requires creatures to supplement it. In that case, this ‘infinite’ goodness would be restricted and finite – an absurdity. And so: ‘It is not necessary, if God wills his goodness to be [viz. possesses and delights in the divine end], that He should will things other than himself to be produced’ (SCG 2.28.10, translation revised; see DV 23.4.c. and Kretzmann Reference Kretzmann and MacDonald1991a, 223). As Bernhard Blankenhorn (Reference Blankenhorn2002, 826, 835) writes, only for creatures who are striving for their final end – meaning, who are not ontologically identical with that end – can the diffusion of their goodness, in their performance of self-perfecting acts (ST 1.14.4.c), represent an intermediate end on the path to that last end. God, however, is no such imperfect agent. Hence Aquinas’ firm denial that communicating the divine goodness might be God’s last end, and the corollary fact that on account of divine perfection, God does not have any intermediate ends either (for intermediate ends are always ‘means’ towards attaining a final end; but this cannot apply in God’s case, for God already possesses God’s end). But then communicating the divine goodness cannot be a volitional motivation, or final cause, for God’s choice to create (SCG 1.81.3; DV 23.4.c.; Keane Reference Keane1975, 106–107; Liccione Reference Liccione1995, 227–228), either as an intermediate or final end for God.
Someone may say, ‘Granted, God does not create to realize a goal, because God already possesses and is the divine goodness. So in creating, God acts just in order to communicate the goodness God already has and is, and not to achieve that goodness. So creating can still be analogically conceived of as a choice, even though the agent’s final cause in choosing is not something the agent is aiming to attain.’ But a ‘choice’ with no goal or end to attain, a choice with no motivating rationale, is a square circle. If the objector’s ‘in order to communicate’ is to carry genuine volitional meaning here, then it can only mean one of two things: either communicating the divine goodness is God’s final motivating end, or it is a means conducing to God’s end, an intermediate end that will contribute to realizing God’s final end. But neither can be true for God. That means that the objector’s premise (that ‘God does not act to attain a goal’ in creating) already eliminates any sense in which creating could be a choice. For it could only be conceivable as a choice if it had a genuine motivation – if God were acting ‘for the sake of’ realizing something not already realized by God simply being the divine essence. It does not matter that the posited ‘goal’ here is ‘expressing liberality’, something God does not for God’s own benefit, but for creatures. God could only be genuinely motivated by this as a goal if ‘expressing liberality’ were not already perfectly realized in God’s simply being the divine goodness. But it is, for otherwise, God would not be infinite perfection. This is not to deny that God’s creating is an instance of mysterious liberality. It is only to say that liberality with no possible motivating rationale lacks the intelligibility of a ‘choice’.
The notion that communicating the divine goodness might be a motivating final cause for God’s free choice to create runs into other problems. First, ‘creating’ and ‘communicating the divine goodness by creating’ are co-extensive, identical in re (if not in concept). In other words, to will creatures to exist just is to will to communicate divine goodness to them.Footnote 4 Consider why. If God could will creatures to be without thereby willing to communicate divine goodness to them, then creatures could exist without participating in divine goodness. But God diffuses divine goodness to creatures precisely by being their ontological final cause, and by so being their final cause, God moves them to act. To say that God is creatures’ final cause is therefore convertible with the claim that creatures exist as ordered to God or to divine goodness (Liccione Reference Liccione1995, 226, 231). And so to imagine a divine intention to create creatures that is not also an intention to communicate goodness to them is to imagine creatures that do not have God as their ontological final cause, to hypothesize a creature that does not share in, and so represent, divine goodness. But such a ‘creature’ is a contradictio in adjecto, a triangle with only two sides. Thus, ‘God’s representing his goodness by creating is necessarily equivalent to his communicating his goodness by creating’ (Liccione Reference Liccione1995, 226). So to say that God creates in order to communicate the divine goodness is simply to say ‘God creates in order to create’. Such a description can serve as a good definition of creating; but as an explanation of the motivation or final cause for a divine choice to create, it is a tautology, because it does not provide a volitional motivation, or final cause, for that act.
An objector may grant this immediately prior point but consider it the very solution to the problem we are posing: ‘Of course “communicating divine goodness” refers only to the fact that creatures should exist, and not to some other goal over and above their participation in God. And so to say that God freely chooses creatures to communicate divine goodness is just to say that God acts for their benefit and not to attain anything for God’s benefit.’ But notice the implication. If ‘communicating divine goodness’ is a true final cause for a divine choice to create, and yet it refers only to the existence of creatures themselves, then creatures themselves, as that communication and representation of divine goodness, would constitute the final cause for God’s choice to create them. This would be equivalent to saying ‘God wills creatures (= wills to communicate the divine goodness) because of creatures (= communications of the divine goodness)’. But this cannot be the case on Aquinas’ account of divine perfection, for God is not caused in any fashion by any other entity (Blankenhorn Reference Blankenhorn2002, 834; Wippel Reference Wippel and Wippel2007, 229). Thomas is explicit that, for this reason, creatures do not move (viz. explain, cause) the divine will (ST 1.19.2.ad2). Creatures, then, cannot be the ‘because’ or ‘in order that’ of God’s choosing. But then neither can ‘in order to communicate the divine goodness’ be such a motivation or final cause for divine free choosing, for this is simply equivalent to saying that the divine will is motivated by (moved by) creatures. In short, any agent that is motivated to choose (and such motivation is intrinsic to liberum arbitrium) is extrinsically caused. But God is not caused or ‘motivated’ at all, for God always already possesses God’s end by being simply identical in essence to it.
If Thomas is right about this, then a venerable way of making sense of divine creating (viz. God creates to share God’s glory or goodness) becomes incoherent. Hence the tension in some of Thomas’ formulations that simultaneously invoke ‘communication of divine goodness’ as an apparent final cause for a divine free choice of creatures and yet also attempt to analogically purify the idea that God is acting for the end of communicating divine goodness:
The ultimate end is not the communication of the divine goodness, but the divine goodness itself, for love of which God wills to communicate it. For he works for his goodness’ sake not as desiring to have what he has not but as wishing to communicate what he has, for he acts not from a desire for the end but from love of the end (DP 3.15.ad14: communicatio bonitatis non est ultimus finis, sed ipsa divina bonitas, ex cuius amore est quod Deus eam communicare vult; non enim agit propter suam bonitatem quasi appetens quod non habet, sed quasi volens communicare quod habet: quia agit non ex appetitu finis, sed ex amore finis).
How should we interpret Thomas’ volitional analogy for creating here? First, Thomas importantly insists that ‘communicating the divine goodness’ cannot be taken as a goal for God. He affirms that the divine will is indeed willing to communicate the divine goodness – but only for the sake of, or for the love of, the divine goodness, and not for the communication as an end in itself (compare ST 1.19.2.ad2). In that case, what does ‘for the sake of’ or ‘for the love of’ the divine goodness as end mean here? Does it mean that God wills to communicate the divine goodness ‘as a means which conduces to attaining that end’? Thomas denies this here, for God is not ‘desiring to have what he has not’: communicating the divine goodness is not a means to attaining the divine goodness. Creatures do not contribute to or comprise the divine end in any respect, and that end is not one attained by any means anyway. But in that case, as we have argued above, this divine ‘choice’ of creatures is already severely unlike a free choice, because to freely choose for Thomas is to select means which conduce to an end.
What, then, can Thomas’ language here mean? The same passage provides the clue: Thomas glosses the notion of God’s ‘choosing’ creatures as ‘means’ to God’s end as ‘God will[ing] things apart from Himself insofar as they are ordered to His own goodness as end’ (ST 1.19.3.c: inquantum ordinantur ad suam bonitatem ut in finem; translation revised). Likewise, in ST 1.19.2.c, Thomas first evokes a version of the Dionysian principle (that it belongs to agents, inasmuch as they are perfect and in act, to communicate their goodness as widely as possible), but then says that, in God’s case, what this means is that God
therefore wills both Himself to be [viz. the divine will eternally delights in the divine essence], and other things to be [viz. wills them to bear the act of existence; these two “wills to be” are not univocal]; but Himself as the end, and other things as [ordered] to that end, inasmuch as it befits the divine goodness that other things should be partakers therein (Sic igitur vult et se esse, et alia. Sed se ut finem, alia vero ut ad finem, inquantum condecet divinam bonitatem etiam alia ipsam participare; translation revised).
It is clear why Thomas would want to purify the meaning of the ‘communication’ intention in this way: in order to clarify that in communicating the divine goodness (= creating), God is not pursuing any goal, and that God is not motivated by that communication, either as an intermediate or final end. God is not acting to attain God’s end, or acting with creatures as God’s end, or willing creatures as a means conducing to God’s end. Instead, in creating, God is merely establishing them in being as ordered to God as their end. And this is not something God is motivated to do ‘because of’ (propter) the divine goodness in the volitional sense of acting to attain or enhance the divine goodness as end (for Thomas denies all this; see ST 1.19.5.c below), but rather in the sense that the existence of creatures ordered to God as their end is merely congruous with (condecet) the divine goodness. But in that case, although Thomas is using volitional choosing language here (speaking of velle, finis, and things which are ordered ad finem), the upshot of this formulation is not volitional at all – for its whole point is to maintain that in creating, God has no motivating rationale, since ‘it would be utterly false to Aquinas’s notion of the divine perfection to suggest that God could be after anything by creating’ (Liccione Reference Liccione1995, 227). All that creating is about is instead God’s causing there to be finite realities ordered to God as their end.
Other readers of Thomas, too, have noted that Thomas carefully controls the meaning of the statement that God ‘intends’ to communicate the divine goodness, or that God creates ‘for the sake of’ the divine goodness, or that the divine goodness is the ‘reason’ for God creating. As Kevin Keane has persuasively argued (Keane Reference Keane1975, 107–108), to say that the divine goodness is the ‘reason’ for God willing creatures (the ratio volendi: ST 1.19.4.ad3; SCG 1.86.2) is not to identify a motivating rationale for why God would create. It is only to claim that the divine goodness is the ratio or ‘formality’ of God’s willing of creatures, insofar as the resultant creation will be intrinsically ordered and rational, reflecting divine goodness as the final cause to which that creation is ordered. Again, it does not identify any motivating rationale for God to create; it is just a description of what creating and creatures are: if God creates, by definition there will exist finite reflections of the divine goodness ordered towards it as their end. This is why, as Keane points out, Aquinas avoids calling the divine goodness the causa for the divine will, meaning, in this volitional context, the motivating rationale that would explain (move) the divine will. ‘We might fairly sum [Aquinas’] position up as follows: it is true that God’s creation of the universe [viz. God’s creative act] has no rationality [viz. creative rationale], but the universe he creates is rational because internally ordered’ (Keane Reference Keane1975, 106n19). Liccione (Reference Liccione1995, 229–230) agrees, noting that when Thomas says ‘the divine goodness’ is that for the sake of which God acts in creating, or is the ratio or final cause of God’s creative act, this must not be taken to ‘explain that act in any clear and non-trivial way’, i.e., to identify a motivating rationale, for God can have none in creating. It only means that ‘[w]hat God creates is…intelligible’ and ‘well-ordered’. If we tried to construe Thomas’ claim that God creates ‘for the sake of the divine goodness’ as if it were an answer to the question why God creates rather than not, or creates this world rather than another, ‘the answer would come to: “Because the world is well set up”’ – in other words, a non-answer that only ‘supplies a ratio of creation, but no rationale’ for God creating (230).
So on the supposition that creatures exist, then divine goodness will be communicated to them and through them – i.e., divine goodness will be their ontological final cause. God’s ‘intention’ to communicate the divine goodness, then, is only a description of what would be true of any world God might create: that it will appropriately express the divine goodness in its own way. This is the reason why, when Thomas deploys his free choosing analogy for creating, he frequently glosses it by saying God ‘chooses’ creatures just inasmuch as creatures’ existing merely ‘is becoming to’ (competit) the divine goodness (SCG 1.37; Kretzmann Reference Kretzmann and MacDonald1991a, 218–219), or that it ‘befits’ it (ST 1.19.2.c.: condecet, pertinet; DV 23.4.c.: decet; compare SCG 1.86). He does so because his doctrine of divine perfection requires him to purify away all other possible senses in which God might ‘will’ creatures: not as an end in themselves (whether final or intermediate), nor a means to attain God’s end – but merely inasmuch as they fittingly express the divine goodness. The creative act might thus be seen as importantly volitional, because it brings forth a finite expression of the divine will’s delighting in the divine goodness (as Aquinas suggested above by saying that in creating God acts not to attain God’s end but ‘out of love for that end’). But that does not make it a choice, because it bypasses the ratio eligendi defined above, namely that means are chosen because they conduce to an end. It only affirms that, if creatures exist, the divine goodness is their exemplary cause (ST 1.6.4.c: ‘everything is called good by reason of the similitude of the divine goodness belonging to it’) and their final cause. It does not establish that creating has the sort of final causality, or volitional rationale, that characterizes choosing.
But if communicating the divine goodness to creatures cannot serve as a final cause, a motivating rationale, for God’s free choice of this world, then one further ratio of ‘free choosing’ has been remoted by Aquinas’ doctrine of divine perfection. None of this is to deny that God as perfect agent communicates divine goodness in whatever God does, or that God creates out of love of that divine goodness and not for its attainment. But it is to say that this communication simply cannot be analogically conceived of as a free choice, because ‘free choices’ with no motivating end – no ‘reason why’ or rationale – are not choices at all. To speak of this sui generis mode of divine acting as ‘free choosing’ appears ever more equivocal than analogical.
What ‘means’ means: Equivocating between volitional and ontological final causality
In this transposition of the Dionysian principle, Thomas redefines the ‘self-diffusiveness of the good’ to prevent the entailment that communicating divine goodness is a final cause or a motivation of God’s choice to create. But he also consistently says that any created effects will also be ordered to divine goodness as their end (e.g., DV 21.1.ad4; Kretzmann Reference Kretzmann and MacDonald1991a, 220), and speaking of ‘order to an end’ certainly sounds like free-choosing language, where the means are chosen because ‘ordered’ to the end one wills. Is this fact that creatures are ordered to divine goodness as their end not sufficient to ground the analogical predication of liberum arbitrium to God? For then we would be saying, analogically, that God ‘chooses’ creatures in the sense that in creating, God ‘orders them to’ the divine goodness (to God’s end) as their end. The ‘ordering to the end’ here is not God attaining God’s end through them (as with volitional means), but just God causing there to be finite realities ordered to God as their final cause.
As he writes in ST 1.19.5.c, ‘in Him, to will an end is not the cause of His willing the means, yet he wills the ordering of the means to the end. Therefore, He wills this to be as means to that; but does not will this on account of that’ (ita velle finem non est ei causa volendi ea quae sunt ad finem, sed tamen vult ea quae sunt ad finem, ordinari in finem. Vult ergo hoc esse propter hoc, sed non propter hoc vult hoc). Thomas expresses here his restriction of the Dionysian principle: God’s willing the divine goodness does not cause or motivate God to will any creatures as ‘means’ ordered to that end. That is, God does not will creatures ‘for the sake of’ (propter) the divine goodness in the sense of ‘in order to attain’, or of fulfilling any goal. Rather, he wills them ‘for the sake of’ the divine goodness only in the sense that, in creating, God is causing there to be finite realities ordered to the divine goodness as their end, as their raison d’être.
So Thomas denies here any ‘ordering’ of creatures to God of the kind that features in free choosing, where means are chosen in order to attain the end – what we can call volitional final causality. As we saw in the last section, the kind of ‘ordering’ of creatures to God which Thomas then affirms is instead just ontological final causality, simply the fact that all creatures are ordered to God as their end. (More precisely, it is that individual creatures are ordered to the finite common good of the universe, and the universe itself is ordered to the transcendent good that is God.) Thomas makes this shift – from volitional to ontological final causality – often in these discussions. We have already seen him make it in ST 1.19.2.c and 1.19.3.c (and see SCG 1.81.5), where he says that God wills God’s own goodness as end but ‘freely chooses’ other things just in the sense that they are ordered to the divine goodness as their end (inquantum ordinantur ad suam bonitatem ut in finem). The slippage is Thomas’ oft-repeated way of arguing that, since the basic scheme of something-being-ordered-to-an-end obtains both in free choosing and divine creating, divine creating, analogically, is God freely choosing creatures.
But this slippage from volitional final causality to ontological final causality is an equivocation. In a finite agent’s (free or necessary) choosing, means stand (conceptually) between the agent and her end: Agent → Means → End. In God’s willing of creatures, the ‘means’, viz. creatures, stand beyond or outside of God’s end, simply as ontologically ordered towards that end: [God as Agent → God as End] ← Means. As such, God’s ‘electing’ creatures in this way is not ‘choosing’ creatures at all, because creatures are related to God and God’s end in a way nothing like the way volitional means are related to agents and their ends. To say that God ‘chooses’ creatures not to conduce to God’s end but merely ‘inasmuch as they are ordered to his goodness as end’ (ST 1.19.3.c) is just to say that God does not ‘choose’ them at all. That is simply not what volitional ‘means’ are on Thomas’ own highly developed account of free choosing. Again, ‘the reason for choosing something [as means] is that it conduces to the end’ (ST 1-2.13.5.c). It is equivocal for Thomas to consistently use the means-end schema to explain how God freely chooses creatures but also to empty the schema of any volitional meaning (see SCG 1.88.4 and DV 23.4.c for similar instances).
Thomas’ conceptual equivocation here is concealed by a lexical equivocation. The English translations we have been citing speak of ‘means to an end’ in volitional contexts. But Thomas’ Latin speaks in volitional contexts simply of ends (finis) and of ‘those things which are towards/for an end’ (ea quae sunt ad finem), as well as what is ‘because of/for the sake of’ (propter) an end (for propter, see also DV 23.4.c). For this reason Thomas can, at crucial points, slip with lexical ease from a volitional register (speaking of means as ‘those things which are for an end [ea quae sunt ad finem]’, and which an agent wills in order to attain the end) into the conceptually distinct but lexically identical register of creatures’ ontological final causality, the register of ‘those things which are for an end’ (ea quae sunt ad finem) in the sense of being ontologically ‘ordered towards an end’ (ordinantur ad finem). And he uses propter ambiguously across both registers, such that in DV 23.4.c and ST 1.19.5.c, he denies that creatures represent any kind of necessary volitional means for God (namely, necessitas finis), yet also says, in the language of volitional final causality, that they are created propter suam bonitatem. Yet as we have seen (in the prior section and in ST 1.19.5.c just above), Thomas constrains this to mean not that God wills creatures to attain God’s end, but simply that the divine goodness is creatures’ exemplar and final cause (that creatures reflect the divine goodness and are ordered to it as their end). Though Thomas repeatedly deploys volitional choosing language (presenting creating as God willing Godself as end and creatures as means), on his doctrine of divine perfection, that language’s upshot must be restricted so that it comes to have a non-volitional meaning (just that creating is there existing creatures ordered to God as their end).
It may seem that this transposition somehow predicates to God the volitional meaning of free choosing in a mysterious and higher analogical mode. For the notions of ‘ends’ and of ‘those things which are for the end/for the sake of the end’ have been preserved. But in reality it simply erases the volitional meaning of free choosing and so makes the attribution of ‘free choosing’ to God equivocal. This equivocation functions in the attempted analogical attribution by the logical fallacy of the undistributed middle: (i) In free choosing, an agent chooses means (volitionally) ordered to an end; (ii) creatures are (ontologically) ordered to the end that is God; (iii) therefore creatures are (volitional) means God freely chooses. (All x’s are z. All y’s are z. Therefore, all x’s are y.) Non sequitur. This conflates the notion of something being elected as means to an end in liberum arbitrium with the ontological notion of creatures having God as their ontological final cause. ‘Ordering’ is operative both in ontological and volitional final causality, but the orderings are not conceptually equivalent.Footnote 5 This equivocation on the ‘means’ of divine free choosing, however, is to be expected given Aquinas’ position (analysed above) that the divine goodness is not the sort of end that is produced or attained through any means (DV 23.4.c). But then the divine goodness is just not an ‘end’ of the sort that features in free choosing. Therefore, the notions of both ‘means’ and ‘end’ have been remoted to the point of equivocation under pressure from Aquinas’ rigorous doctrine of divine perfection.
In fact, this final transposition of ‘means and ends’ we have just analysed catches Aquinas in the trap he identifies as one wrong account of divine predication, namely, thinking that predicating something of God tells us only that God is the cause of that predicate’s existence in creatures (for example, the claim that ‘God is living’ means only ‘God is the cause of life in creatures’; ST 1.13.2.c). If creatures are not ‘means’ in a volitional sense but instead are ‘means’ only insofar as they, qua creatures, are ontologically ordered to that divine end, then to say ‘God freely chooses creatures’ is not a substantial predication about God, for it speaks only of creatures’ relation to God and does not speak of God as God is in se. As this subsection and the previous one have shown, Thomas’ gloss on creatures being chosen ‘because of’ the divine goodness, or of creatures being a ‘means to’ the divine goodness, reduces to talk about creatures, not about God: namely, that creatures’ exemplar and final cause is God. All that Aquinas’ language about God ‘freely willing’ creatures can mean is an affirmation that any created realities that happen to obtain will be ordered to the divine goodness, and that their existence (or non-existence) is compatible with, and reflects, the divine goodness. ‘God wills creatures by free choosing’ thus does not succeed as a divine analogy, for as Thomas maintains, a successful proper (viz. analogical) divine predication says something primarily about God, not primarily about creatures.
Thomas’ account therefore eliminates this final ratio of free willing: an agent’s ordering of means to a willed end.
Divine free willing: Metaphor, not analogy
And with that, there is nothing left of the intrinsic conceptuality of liberum arbitrium to attribute properly and analogically to God. Aquinas’ account of divine perfection has required us to remote all four elements noted above: (i) Grasping the proportion of a means to the end willed (consilium); (ii) the election (electio) of a perceived best option; and (iii) a motivating final cause or rational for the free choice, namely, the final end, (iv) towards which the means are ordered (ad finem) in consilium and electio. If even one element had survived the remotion, there would be a case that ‘free willing’ can be properly predicated as literally, albeit analogically, in God. But what remains after Aquinas’ purgation of ‘divine free willing’ is an equivocation.
If our argument is correct, then Aquinas is incorrect that ‘free willing’ is a pure perfection whose primary instance is God, with creatures only secondarily possessing it. It is instead a perfecting operation for finite agents like us, an act bound up with the finitude proper to rational creatures. Put most simply, the argument we have prosecuted rests its case on the fact that because God already possesses God’s end, and is in fact identical with it (on account of simplicity), the conditions cannot exist under which God could elect any means to reach God’s end, which is of the essence of free willing under Thomas’ specific account of the latter. Free choosing thus appears to be impossible for God in the same way that it is impossible for God to develop or be distended in space. Yet this fact – that the conditions for free willing do not and logically cannot obtain for God – reflects no deficiency in God, for it bespeaks not a limitation but rather plenipotential fullness of being for the divine will, which rests in and is identical with its infinite end.
In this light, we can see why Aquinas’ arguments for free willing in God falter. It does not follow, for example, that if we deny that ‘free willing’ is properly in God, then this implies that God lacks intellect or creates through natural instinct (SCG 1.88.3). This argument takes a necessary condition for the exercise of free willing (having an intellect) and confuses it with a sufficient one. It also ignores the fact that, in Thomas’ own view, intellectual beings naturally and necessarily will the end proper to their nature (happiness in our case, the divine goodness in God’s), and that this natural, necessary willing is both free and intellectual (ST 1.82.1.co; Knasas Reference Knasas2002, 424). Recall that on Aquinas’ account, the primary instance of God’s voluntary activity is the divine will’s necessary ordering to and delight in Godself, which is not a discrete intentional volitional act that could have been otherwise (Blankenhorn Reference Blankenhorn2002, 831). But this does not demote this primary divine volitional and agential operation to the level of the mindless or non-intellectual. God does not need to freely will, in the sense of free choosing, in order to be supremely intellectual and supremely free. It does not follow, then, that God possesses free will merely because God is an intellectual nature. And neither is it the case that God is not master of God’s own acts if God does not exercise free willing (as Aquinas argues in SCG 1.88.5), for if this were true, then God would not be master of God’s own act of willing the divine essence.
Nor, then, will the classical theist who accepts this picture of divine willing have reason to deny freedom to God as a result of rejecting the logical possibility of divine free choosing. Only if divine freedom (the freedom of metaphysically composite agents is a separate matter) is taken to be identical with or to imply the capacity to do otherwise can this conclusion follow (as in Mullins Reference Mullins2013, 194). But we would affirm with David Bentley Hart (Reference Hart2020, 105) a different understanding of divine freedom, one signifying instead the lack of any extrinsic coercion or restraint on God’s perfection and actuality:
[i]nasmuch as God is not a finite being, in whom possibility exceeds actuality, but is infinitely actual, and is in fact the eternal act of all things actual, in him there is no meaningful modal distinction between freedom and necessity. He is the supereminent fullness of all that is, and his freedom is the unhindered actuality of being who he is.
But there are also those within the Thomist tradition itself who have questioned whether the traditional way of counterpoising ‘freedom’ and ‘necessity’ can really be coherent when speaking of the God who is pure act (Crowe Reference Crowe and Vertin2010, 452–53). Most importantly for our argument, however, is Aquinas himself, who when speaking of the Trinitarian processions suggests an account of divine freedom as something other than the ability to do otherwise: ‘God by his will freely loves himself, although he loves himself of necessity’ (DP 10.2.ad5, translation and emphasis ours: Deus sua voluntate libere amat seipsum, licet de necessitate amet seipsum). This coinherence of freedom and necessity in the divine processions suggests that the language of divine freedom can be maintained from a classical theist, and especially Thomist, perspective without positing liberum arbitrium as properly in God.
But what cannot be properly attributed to God, because its primary instance is the creature, may yet be metaphorically predicated of God, on the basis of some likeness associated with the metaphor’s referent, usually a likeness of effect (ST 1.13.6.c: ‘all names applied metaphorically to God, are applied to creatures primarily rather than to God, because when said of God they mean only similitudes to such creatures’; McInerny Reference McInerny1996, 134). What then, is the likeness in effect between finite free willing and God’s willing of creatures, such that the metaphorical predication of ‘free willing’ to God may be justified?
In finite free choosing, the means elected are not what is primarily willed by voluntas; that distinction belongs to the ‘end’. And so there is something derivative, something secondary (cf. DV 23.4.co), about the choosing of the means, which is expressed in saying that these means are ‘ordered to the end’ or ‘for the sake of the end’. Equally so on Thomas’ account, where although God’s willing creatures and God’s willing the divine essence are one self-same, simple act, God’s willing of creatures is nonetheless derivative of God’s willing of Godself (SCG 1.76.4), something like an ‘effect’ of that primary willing (even if Thomas maintains that it is an ‘effect’ that need not have obtained). The same kind of derivative effect flowing from an act ordered primarily to the divine essence is present as well in Aquinas’ account of divine understanding: ‘Now God, by his intellect, understands himself principally, and other things in himself. Therefore, in like manner he wills himself principally, and in willing himself, he wills all else’ (SCG 1.75.7: Sed Deus intellectu suo intelligit se principaliter et in se intelligit alia. Igitur similiter principaliter vult se, et, volendo se, vult omnia alia). The parallel between the derivative nature of means in finite free willing and the derivative willing of creatures in God’s willing of the divine essence suggests the correct way to understand the language that God ‘freely wills’ creatures: it is an equivocation-as-metaphor that says something substantial about God, specifically about the mode of God’s willing of creatures. The metaphor expresses how God derivatively includes creatures in God’s one – necessary yet sublimely free – act of willing Godself, yet without creatures thereby comprising the divine essence, that essence which is the divine end identical in re with the divine will both as power and as the divine act of willing. To speak metaphorically of God’s ‘freely willing’ creatures thus communicates the profound fact that God is not obliged (with necessitas finis or coactionis) to create creatures and that by creating them God gains nothing at all. It furthermore expresses that, although creatures are in some mysterious fashion included in or caused by God’s act of willing the divine essence, creatures are not the proper object of God’s will, and God’s will is not motivated by, or ordered to, or perfected by, them. The free choosing metaphor communicates these denials helpfully by saying that God ‘wills’ creatures in some way, but not in such a way that creatures are the ‘end’ God wills; they are instead only ‘for the sake of’, or ‘ordered to’, that end – even if, as we have shown, this ordering cannot literally be a volitional ordering. We should affirm these denials (and more besides), and use the free willing metaphor to invoke them; but we should not suppose God’s derivative, non-perfective willing of creatures as ordered-to-God is anything essentially like what free choosing is in us. It is far more mysterious than that.
Data availability statement
This article does not have any relevant data set, and thus no statement here is required.
Acknowledgements
We express our appreciation to Sarah Coakley, Karen Kilby, Ross McCullough, Ty Monroe, Taylor Nutter, Jeremy Wilkins, Jordan Daniel Wood, Justin Shaun Coyle, the Mount Angel Abbey Fellowship of Scholars, and the Lonergan Fellows at the Lonergan Institute (Fall 2022) for helpful feedback on this project, as well as to the anonymous reviewers for their incisive criticism of the manuscript.
Funding statement
We received no funding to support the research for, or writing of, this article.
Competing interests
The author(s) declare none.
Ethical standards
The research for, and writing of, this article involved no research with human subjects, and thus no issues where ethical approval or informed consent would be required. No AI tools were used in the generation of the ideas of this article or in its writing.
Open access
