Aquinas on Free Will and Intellectual Determinism

Aquinas on Free Will and Intellectual Determinism

Tobias Hoffmann & Cyrille Michon

From the early reception of Thomas Aquinas up to the present, many have interpreted his theory of liberum arbitrium (which for Aquinas is free will specifically as the power to choose among alternatives) to imply intellectual determinism: we do not control our choices, because we do not control the practical judgments that cause our choices. In this paper we argue instead that he rejects determinism in general and intellectual determinism in particular, which would effectively destroy liberum arbitrium as he conceives of it. We clarify that for Aquinas moral responsibility presupposes liberum arbitrium and thus the ability to do otherwise, although the ability to do otherwise applies differently to praise and blame. His argument against intellectual determinism is not straightforward, but we construct it by analogy to his arguments against other deterministic threats (e.g., the one posed by divine foreknowledge). The non-determinism of the intellect’s causality with respect to the will results from his claims that practical reasoning is defeasible and that the reasons for actions are not contrastive reasons.

Are our choices unavoidable, given our reasons for making them? If our choices follow our reasons strictly, one might worry about whether our reasons proceed from sources that do not depend on us, such as our innate dispositions, our upbringing, and particular events that happened to us shortly or long before we made a choice, determining the choice unavoidably. In contrast, if our choices do not follow our reasons strictly, then they are irrational. In either case, free will is threatened, and apparently moral responsibility with it. These concerns are one way of approaching the traditional problem of free will and determinism, which we call the problem of intellectual determinism.

In this paper we will investigate Thomas Aquinas’s answer to the problem of intellectual determinism in his account of liberum arbitrium (Aquinas’s and his contemporaries’ term for free will) and moral responsibility.¹ He clearly maintains that liberum arbitrium exists and that it is presupposed for moral responsibility. What is less clear is whether he adopts intellectual determinism, in which case he would be a compatibilist. Prima facie, it seems that he is not a compatibilist, because he repeatedly says that our choices are not necessitated. But Aquinas’s early critics accused him of intellectual determinism, and some contemporary scholars give his account of liberum arbitrium a compatibilist interpretation. The question, then, is whether Aquinas’s account of liberum arbitrium implies intellectual determinism, and if so, whether it does so contrary to his own intention or whether he consciously subscribes to intellectual determinism.

We argue, however, that Aquinas is most plausibly seen as (1) an indeterminist in general and (2) an intellectual indeterminist in particular — in other words, that he has an indeterminist conception of liberum arbitrium. We offer separate and original evidence for both theses, while at the same time arguing that his general indeterminism enhances the plausibility of his intellectual indeterminism. In order to show the significance of indeterminism for his action theory, we will also clarify the ways in which he speaks of freedom and analyze the role he attributes to liberum arbitrium in moral responsibility.

In section 1, we will explain Aquinas’s concept of the will and two senses in which it is free. For Aquinas, free will in the broad sense does not require alternative possibilities, while free will in the narrow sense, which is liberum arbitrium, is precisely the power to make choices between alternative possibilities. In section 2, we will investigate whether for Aquinas moral responsibility presupposes liberum arbitrium (and hence alternative possibilities). We will first discuss an interpretation that altogether denies liberum arbitrium as a necessary condition for moral responsibility; then we will consider whether Aquinas posits an asymmetrical relation between liberum arbitrium and moral responsibility, according to which blame presupposes liberum arbitrium, while praise does not. In section 3, we will investigate how Aquinas responds to several threats to liberum arbitrium, not only by intellectual determination. We will discuss, above all, his arguments that divine foreknowledge and God’s ‘efficacious will’ do not necessitate human choices. His strategy is never to adopt a compatibilist solution, but always to show that these threats only appear to imply that our choices are necessary. Aquinas hardly discusses the threat of intellectual determinism, but the way he handles other threats of necessitation makes it highly plausible that he would not accept a compatibilist solution to the threat of intellectual determinism either. In section 4, we will review contemporary interpretations of Aquinas that see him either as a compatibilist or as an incompatibilist and — given his belief in liberum arbitrium — as a libertarian; and we will argue against a paradigmatic compatibilist interpretation. Finally, in section 5, we will address intellectual determinism directly: with the help of Elizabeth Anscombe’s analysis of practical reasoning, we will argue that Aquinas’s account of practical rationality is exempt from the threat of intellectual determinism, since practical reasoning, as Aquinas conceives of it, does not lead to necessary conclusions and thus to necessary choices, because practical reasoning is defeasible. One remains always free to revise one’s practical inference by changing one of its premises or by adding a further premise.

1. Freedom of the Will and Liberum Arbitrium

Aquinas calls certain persons free, even though they cannot do otherwise. He sometimes even calls them freer than those who can do otherwise. And yet he repeatedly insists on the possibility to do otherwise as a condition for moral responsibility and as a criterion for a choice freely made. These apparently inconsistent affirmations have caused some confusion among his readers. In order to gain a clear understanding of Aquinas’s theory of liberum arbitrium and of its relation to moral responsibility, we must first clarify different senses in which he speaks of freedom in general and of freedom of the will and liberum arbitrium in particular. We will show that, for Aquinas, freedom of the will conceived broadly (libertas voluntatis, libera voluntas) requires sourcehood (that is, the agent’s will must be the source of her action) but not alternative possibilities, while liberum arbitrium, which is freedom of the will conceived narrowly as freedom of choice, requires alternative possibilities in addition to sourcehood. Since sourcehood and alternative possibilities can be understood differently, we will clarify how we take Aquinas to understand them.

1.1. Freedom of the Will and Sourcehood

Aquinas does not offer a general definition of ‘freedom’ which would explain various senses in which he calls a person or a person’s power (such as the will) free.² Some of the senses of freedom he employs are rather remote from the contemporary notion of free will. For instance, he occasionally uses ‘freedom’ in the Augustinian and ultimately biblical sense of freedom from sin (cf. John 8:34–6). This freedom can be had to different degrees; Augustine famously called it a greater freedom to be unable to sin than to be able not to sin.³ Aquinas acknowledges that this is a very important sense of freedom, but he rarely discusses it.⁴

The senses of freedom to which Aquinas devotes most attention are opposed not to sin, but to necessity; they fall under the traditional label of ‘freedom from necessity’, which he implicitly identifies with freedom of the will, and which according to him cannot be increased or diminished (QDV 22.5 ad 14, QDV 24.8 ad 5, QDV 24.10 ad 7). In accordance with the earlier medieval tradition, Aquinas frames discussions about freedom more frequently in terms of the necessity or non-necessity of acts than in terms of alternative possibilities or lack thereof; and when he discusses threats to freedom, he is concerned with different types of necessity.

The meaning of ‘necessity’ determines the meaning of ‘freedom from necessity’; but how precisely Aquinas understands necessity in his discussions about freedom is just as controversial as his understanding of alternative possibilities, for he defines necessity in terms of non-possibility: “‘Necessary’ is that which cannot not be [quod non potest non esse]” (ST 1a.82.1). He also defines necessity as “determination to only one outcome”, that is, the exclusion of any alternative possibility: “Something is called necessary, precisely because it is immutably determined to one outcome [immutabiliter determinatum ad unum]” (QDV 22.6 c. lines 69–70). A fuller account of Aquinas’s conception of necessity will emerge later in this paper. For now, suffice it to mention that, in the context of his discussions of freedom, he analyzes the notion of necessity with the help of Aristotle and Augustine. Building upon Aristotle’s treatment of necessity in Metaphysics 5.5.1015a20–b15, Aquinas distinguishes between (1) intrinsic necessity (which he also calls natural or absolute necessity) and (2) extrinsic necessity.

Intrinsic necessity belongs to a thing because of its nature (e.g., it is necessary for a triangle to have three angles that are equal to two right angles). Extrinsic necessity is twofold: either (2.1) something is necessary because of an end (e.g., eating is necessary for life; a ship is necessary to cross the sea), or (2.2) something is necessary because of an efficient cause (e.g., someone who is coerced to do something with-out being able to do otherwise).⁵ From Augustine’s De civitate Dei 5.10, where Augustine explicitly addresses the compatibility of freedom and necessity, Aquinas takes the distinction between natural necessity (which corresponds to Aristotle’s intrinsic necessity) and necessity of coercion (which corresponds to Aristotle’s necessity from an efficient cause).⁶

For Aquinas, which of these necessities threatens the freedom of the will depends on whether the will’s freedom is taken in the narrow sense as liberum arbitrium (that is, as freedom of choice), or in a broad sense (that is, as a freedom that can be enjoyed even apart from choice). If freedom of the will is broadly conceived, then it is compatible with all the forms of necessity except for necessity of coercion. According to Aquinas, necessity of coercion is incompatible with any sense of freedom of the will because although a person can be coerced to do something, it is impossible that she be coerced to will something. To will something means to will it of one’s own accord; just as one cannot make a stone move upward by its own inclination, so one cannot coerce someone to will something of her own accord (ST 1a2ae.6.4). In contrast, as we will see in detail below, if freedom is understood in the narrow sense as liberum arbitrium, then it is only compatible with the necessity of an end, because this necessity is conditional, inasmuch as the agent is free to abandon the end. For example, if I want to cross the sea, it is necessary that I take a ship; but I remain free not to take a ship, because I can abandon my intention of crossing the sea.

Aquinas’s understanding of freedom of the will broadly conceived (i.e., as compatible with all forms of necessity except for necessity of coercion) and of liberum arbitrium is based on the definition of the will as rational appetite. For Aquinas, the fact that the will is a rational appetite means that one can have a rational desire or make a choice only in accordance with what one understands to be good. Analogously, one can have sensory desire only if one’s senses perceive something as good. So there may well be a conflict between sensory and rational desire: a person with gluten intolerance may crave beer but, by rational desire, want to avoid it. Of course, one can also mistake something good as bad, and thus be turned away from it; vice versa, one can consider something bad to be good and thus pursue it. An alcoholic may consider abstinence to be bad and overdrinking to be good. Yet for Aquinas, even when someone pursues something bad, she can do so only under the guise of the good (sub ratione boni), for willing evil under the guise of evil falls outside of the scope of the rational appetite (ST 1a2ae.8.1).

According to Aquinas, precisely because the will is rational appetite, it wills certain things of necessity and other things contingently. Acts of liberum arbitrium are contingent acts. Before discussing these, however, it will prove helpful, in view of the next section, to consider in further depth the will’s necessary acts. Since the will as rational appetite is ordered to the good understood by reason, when something inevitably appears to us as good from every perspective, we cannot but desire it. Such a thing is happiness (ST 1a2ae.10.2 c., QDM 6 c. lines 429–35). Similarly, the blessed human beings and angels — who, according to Christian teaching, see God as he is — necessarily understand that God is the essence of goodness, and for this reason they cannot but love him (ST 1a.62.8, ST 1a.82.2). Furthermore, God, too, cannot but love himself, for the divine will has the divine goodness as its proper object (SCG 1.80 n. 677, QDV 23.4 c., QDP 1.5 c., ST 1a.19.3 c.). In all of these three cases, Aquinas calls the will free and reminds us that only necessity by coercion, but not natural necessity, is incompatible with the will’s freedom.7 To express the kind of freedom at playing these cases of necessary willing, Aquinas usually uses the term ‘free will’ (libera voluntas).8

Clearly, then, for Aquinas freedom of the will does not require alternative possibilities. Yet it does imply that the agent’s will is the source of the action; indeed, as we have seen, since the will cannot be coerced, the will is essentially the source of its own acts — or, more precisely, by means of the will the human person is the source of her acts. Aquinas’s expression that comes closest to our contemporary understanding of sourcehood is ‘voluntariness in the perfect sense’ (voluntarium secundum rationem perfectam). Perfect voluntariness must meet three criteria: first, the source (principium) of the act is in the agent rather than imposed from the outside (this applies also to natural objects — e.g., a stone in free fall is the source of its motion, but a stone thrown upward is not); second, there must be cognition of some end (this applies also to non-rational animals — e.g., a lion seeing deer recognizes it as prey); third, there must be an understanding of an end qua end and an understanding of the relation between the end and the things that promote the end (this applies only to humans) (ST 1a2ae.6.1–2; cf. QDV 24.1–2). Perfect voluntariness opens up alternative possibilities when rational deliberation shows that there are different available means to an end (ibid.); but if it is understood that there is only one available means to the desired end, and if it is impossible not to desire the end, then there can be perfect voluntariness without alternative possibilities. This is the case, for example, of the blessed, who are unable not to desire happiness and who, because of the divine vision, fully understand that happiness can be had only through love of God (ST 1a.82.2 c.; cf. ELP 1.14 lines 501–5).

We can identify perfect voluntariness in Aquinas with sourcehood as long as we qualify the term ‘sourcehood’. For Aquinas, the human will is, metaphysically speaking, not the absolutely first origin of its act. Like all created causes, it can cause something only insofar as it is in turn moved by God, who is the creative cause not only of the human being but also of her volitions (QDP 3.7 c., p. 58; ST 1a2ae.6.1 ad 1 and ad 3). Aquinas distinguishes between the primary cause, which is God, and secondary causes, which are created causes; and he holds that secondary causes can act only inasmuch as they are acted on by the primary cause (e.g., QDP 3.7, ST 1a.105.5). Contemporary philosophers prescind from primary causality and distinguish only between sourcehood (in general) and ultimate sourcehood: a willing agent is the source of her action only if the act is produced by her will; a willing agent is the ultimate source of her action only if she is the source of her action and her will is not caused by something else. Aquinas’s account is more complex, because he introduces the distinction between primary and secondary causality. We may say that, in his view, God is the absolutely ultimate source of a human being’s actions. Yet, for Aquinas, God does not coerce the human being’s acts, for if he did, God would not move her will, but he would move the person against her will (ST 1a.105.4 ad 1, ST 1a2ae.6.4 ad 1). So Aquinas leaves room for the agent being the ultimate source of her own action in the order of secondary causality, although she is not the absolute ultimate source of her action (the terminology is ours). Aquinas’s explanation of how God and the human will are both sources of the person’s act is not very developed. He merely argues that God moves voluntary causes in such a way that he does not take away the voluntary character of their actions, but rather causes them to be voluntary (ST 1a.83.1 ad 3). We will return to this below in section 3.2.

1.2. Liberum arbitrium: Sourcehood and Alternative Possibilities

As we have seen, the will is not always able to choose; but, for Aquinas, liberum arbitrium is by definition the ability to choose. Aquinas understands liberum arbitrium not as a power distinct from the will, but rather as the will insofar as it chooses the means to an end (ST 1a.83.4). Sourcehood, which characterizes all acts of the will, thus applies also to liberum arbitrium. In addition, liberum arbitrium, as the power of choosing the means to an end, also implies having alternative possibilities. Furthermore, Aquinas makes liberum arbitrium a necessary condition for moral responsibility. He takes the fact that human beings have moral responsibility for their actions as a given and argues from this premise to the fact that human beings have liberum arbitrium: if they lacked liberum arbitrium, “advice, exhortations, precepts, prohibitions, rewards and punishment would be pointless” (ST 1a.83.1 c.).9

We can formulate this account of liberum arbitrium and its relation to moral responsibility in three propositions:

(LA1) Acts proceeding from liberum arbitrium originate in the agent.10 [Sourcehood condition]

(LA2) Acts proceeding from liberum arbitrium are avoidable by the agent. [Leeway condition]

(LA3) Liberum arbitrium is a necessary condition for moral responsibility.

The first two are defining and necessary conditions of liberum arbitrium; the third is very similar to the Principle of Alternative Possibilities (PAP).11

Apart from the argument for liberum arbitrium from the assumption that we have moral responsibility, Aquinas also repeatedly offers arguments from moral psychology — that is, on the basis of the nature of intellect and will, their respective objects, and their interrelation. While a fuller discussion of his account has to wait until sections 4–5 below, it is helpful to outline his argument as he presents it in his later writings. Aquinas distinguishes between exercise and specification: the will moves itself and it moves the intellect as to the exercise of the act (wanting after not wanting, considering something after not doing so — e.g., starting to deliberate about how to become healthy), whereas the intellect moves the will as to the specification of its act (wanting this or that — e.g., wanting to take this medicine now). In the order of specification, only happiness and what is understood to be indispensable for achieving happiness move the will of necessity, and so (as mentioned in section 1.1) we do not have liberum arbitrium with regard to the desire for happiness. But most things can be seen as good from one perspective and as deficient from another perspective, and hence they can be either desired (or chosen) or not; hence, with regard to these things, we have liberum arbitrium (ST 1a2ae.10.2, QDM 6 c.).

The fact that Aquinas links liberum arbitrium to precepts and prohibitions, and to rewards and punishments, might suggest that liberum arbitrium essentially concerns choices between good and evil; but this is not in fact the case. While for Aquinas liberum arbitrium must involve alternative possibilities, it is incidental that the choice be between good and evil. In fact, in his view liberum arbitrium is essentially ordered to the good (since it is not a distinct power from the will), and, as we have seen, we can choose something evil only if we falsely take it to be good. He distinguishes between three ways in which liberum arbitrium may regard alternative possibilities: (1) the choice between different means to an end; (2) the choice between good and evil; (3) the change of preference — that is, desiring now one thing and now another. Only the first of these is essential to li-berum arbitrium. It is in this way that God has liberum arbitrium (QDM 16.5 c. lines 215–300; see also In Sent. 2.25.1.1 ad 2). In fact, according to Aquinas, God is free to create this world or not, and to create a better or a worse one (e.g., one containing more species of animals — or fewer).12 Likewise, the blessed in heaven — who, like God, can no longer sin — have liberum arbitrium by which they can choose between different means to an end (QDM 16.5 c. lines 249–52, 266–70). Examples might be the blessed angels who, in their communication with other angels, can choose between different addressees and different topics (cf. ST 1a.107.1 c.), and who presumably can decide on different ways to assist those on earth who are entrusted to their care (cf. ST 1a.113.5 c. and ST 2a2ae.52.3 ad 1).

1.3. Compatibilist and Incompatibilist Interpretations of Sourcehood and Alternative Possibilities

As is evident from what we have seen so far, unlike contemporary discussions of free will, Aquinas’s discussions are not framed in terms of causal determinism — that is, the idea that all events or states of affairs, including actions and choices, are necessitated by the conjunction of some past state of the world and the laws of nature. In Aquinas’s conception of nature, efficient causes are things endowed with causal powers rather than events subsumed under some kind of laws, and in his view, causation and necessitation do not imply each other: some causes are not necessitating, and some forms of necessitation (e.g., by logical necessity, as in the presumed necessitation by divine foreknowledge) do not involve causation. Causation by another secondary cause — even if not necessitating — does, however, undermine ultimate sourcehood, and necessitation — even if not causal — does undermine leeway. Accordingly, we take the notions of ‘compatibilism’ and ‘incompatibilism’ in a broader sense than in the contemporary free will debate, as applying to theories that consider liberum arbitrium to be compatible or incompatible with necessitation, what-ever may ground it.13

With this clarification in mind, we can say that there may be either compatibilist or incompatibilist interpretations of Aquinas’s LA1 and LA2 and, as a result, of LA3. A compatibilist interpretation of LA1 emphasizes that liberum arbitrium does not exclude the causation of choice by factors independent of the agent.14 For compatibilism, commonly understood, all the conditions for free agency are compatible with causal determinism, which implies that any choice is ultimately caused by factors that do not depend on the agent. Compatibilism requires that, in order to qualify as a free choice, an act has to be elicited willingly and according to the agent’s beliefs — that is, the causal route to choice must pass through the agent’s mind in a non-deviant way. (In addition, no pathology must affect the agent, the choice must conform to the agent’s own history, it must be sensitive to reasons, etc.15) But just as the agent’s desires and beliefs may be caused by factors that are not in the agent’s control, so also with the act of choice itself. In that case, the agent would still be the source of her action. But the libertarian account of LA1 requires more: it requires that the agent be not only the proximate source, but the ultimate source — at least in the order of secondary causality, for, as we have seen in section 1.1, although Aquinas holds that God and not the human will is the absolute ultimate source of human actions, he considers the human will to be the ultimate source among created causes.16 We will return to the issue of how the exclusion of absolute ultimate sourcehood from the human will accords with libertarianism in section 3.2.

Similarly, according to the compatibilist interpretation of LA2, leeway, avoidability, or, more generally speaking, the ability to do or choose otherwise, could be interpreted according to the ‘conditional analysis’ and hence be considered compatible with necessitation. Thus, saying that the agent could have chosen otherwise is merely to say this: Had the agent had other desires, beliefs, or reasons, she would have chosen otherwise. Again, the libertarian account requires more: LA2 requires absolute leeway, which is incompatible with necessitation. Hence, for the libertarian, either the agent has leeway regarding the causes of her choice, or those causes do not determine her choice, or her choice has no cause.

One can be fully compatibilist or fully incompatibilist concerning liberum arbitrium, thinking that conditions LA1 and LA2 are jointly either compatible or incompatible with determinism, or one can hold that one condition is compatible and the other is not. But some incompatibilists consider LA2 to be completely irrelevant to the problem of free will. In fact, some libertarians have adopted the position of source-incompatibilism, requiring only LA1 as a necessary condition for free will.17 Furthermore, one can admit that LA2 is a necessary condition for liberum arbitrium, but not for moral responsibility.18

While our full argument that Aquinas has an incompatibilist understanding of LA1 and especially of LA2 must wait until later in this paper, we think there is a good prima facie case to be made for his incompatibilism. As concerns LA2 in particular, its incompatibilist interpretation is the most natural reading of Aquinas’s repeated affirmations that the will does not choose of necessity any particular goods (e.g., QDV 22.6, ST 1a.82.2, ST 1a2ae.10.2, QDM 6) and that sins are avoidable (see section 2.1 below). It is also the most natural reading of Aquinas’s affirmation that the future is open to free agents in such a way that it depends on them which alternative possibilities are real-ized (e.g., ST 2a2ae.49.6 c.). Finally, there appears to be no prima facie reason for ascribing to Aquinas any commitment to determinism or to compatibilism.

Having focused, in this section, on sourcehood as the necessary condition for freedom of the will (LA1) and on alternative possibilities as an additional necessary condition for liberum arbitrium (LA2), we have mentioned only briefly the relationship between these two conditions and moral responsibility (LA3). To this we now turn.

2. Liberum Arbitrium and Moral Responsibility

In order to grasp the full significance of liberum arbitrium and hence of alternative possibilities in Aquinas’s moral theory, we will now clarify the relation of liberum arbitrium to moral responsibility. We will first discuss an interpretation that, like ours, considers him to be a libertarian, but, contrary to ours, denies that he sees liberum arbitrium as a necessary condition for moral responsibility. Then we will consider the hypothesis that Aquinas posits an asymmetrical relation between liberum arbitrium and moral responsibility: blame, but not praise, would presuppose the agent having alternative possibilities. We will argue on the contrary that for Aquinas moral responsibility is always anchored in liberum arbitrium: a present action deserves praise or blame only if the action might not have been done, or at least if a previous action which led to the present action might not have been done. Yet we also intend to show that Aquinas does admit a certain asymmetry, albeit one which accords with the consistent requirement of alternative possibilities for moral responsibility.

2.1. Moral Responsibility without Alternative Possibilities?

As mentioned in section 1.2, for Aquinas, one line of argumentation goes from the existence of moral responsibility to that of liberum arbitrium (LA3) and from there to the availability of alternative possibilities (LA2). But Eleonore Stump has argued that while Aquinas is fully incompatiblist regarding liberum arbitrium, he is only source-in-compatibilist concerning moral responsibility.19 In other words, while Stump agrees that for Aquinas alternative possibilities are required for liberum arbitrium (LA2), she denies that they are also a necessary condition for moral responsibility, and thus she holds that, in Aquinas’s view, PAP and hence LA3 are false. She argues that Aquinas would give the same verdict as Frankfurt in scenarios meant to show that an agent who cannot do otherwise is still morally responsible for her actions.20 Stump’s interpretation relies on texts in which Aquinas admits that some sins are unavoidable at the very moment when they are performed.21 If they are sins, moral responsibility is implied (QDM 2.2 c. lines 128–42). If they are unavoidable, alternative possibilities are excluded. Aquinas says in fact that in some circumstances one may sin under the influence of sudden passion, without being able to submit the sudden passion to the control of reason.

A sin or its avoidance can exceed the power of liberum arbitrium insofar as some sin occurs suddenly and unexpectedly, and thus it escapes the choice of liberum arbitrium, although liberum arbitrium could commit the sin or avoid it if it directed its attention and effort toward it. (QDV 24.12 lines 301–8)

Stump takes Aquinas to hold that, in these cases, sin still implies moral responsibility, but not the ability to do otherwise, and so she concludes that there can be moral responsibility without liberum arbitrium. Thus liberum arbitrium would not be a necessary condition for moral responsibility (Stump 2003, 297–300).

This interpretation does not square well with Aquinas’s repeated insistence that the denial of liberum arbitrium would ruin moral responsibility.22 It also does not square with his affirmation that

The notion of voluntary sin requires only that a human being be able to avoid each individual sin. (ST 1a2ae.74.3 ad 2)

We therefore favor another explanation. The texts Stump refers to belong to a larger consideration of the question: Is sin necessary? Aquinas’s constant position is that it is inevitable over a certain stretch of time to commit some sin (because one cannot constantly pay enough attention to avoid sinning), but that each particular sin is avoidable (because, with respect to any given case at hand, one can make the effort to pay sufficient attention).23 And this does not go against LA3, since every sin-token remains avoidable. Just after the text of QDV quoted above, Aquinas goes on to say:

In the state of corrupt nature [i.e., after the Fall], it is not in the power of liberum arbitrium to avoid all such sins, because they escape its act. Nevertheless, if it makes an effort, it can avoid any one of these movements. But it is not possible that a human being continuously make an effort to avoid such movements, because of the various occupations of the human mind and because of the rest this requires.24 (QDV 24.12)

Accordingly, what Aquinas expresses is a general, as it were, statistical limit to the power of liberum arbitrium, rather than the idea that liberum arbitrium is exceeded in any individual case. The reason he gives (namely, that inattention is unavoidable in the long run) should be understood as a feature of the human condition inevitably leading to sin eventually. Another way to put it is to say that the sin committed under sudden passion was unavoidable at the moment the passion occurred, for at that moment deliberation had become impossible. But it can be traced back to a time when the agent could have avoided it by preparing against the situation where she was subject to such a sudden passion. In that case one can say that some sins are not directly acts of liberum arbitrium.25 Aquinas, in fact, repeatedly argues that people can be held responsible for sins they can no longer avoid, because of their own fault. The most momentous case he discusses concerns the fallen angels, who, according to Christian teaching, are perpetually in a state in which their will cannot but be turned away from God, after having freely chosen to disobey God when it was possible for them to remain obedient (ST 1a.64.2, QDM 16.5 c. lines 301–40). Although Aquinas holds that they can no longer change their disobedient wills, he thinks that they are voluntarily in this state of disobedience (QDM 16.5 ad 8) and are responsible for it, just as Aristotle argues that drunken people who lose control are responsible for their actions, because it was up to them to get drunk or not.26 So all these cases of unavoidable sin proceed from liberum arbitrium, in the sense that the unavoidable sinsare due to previous acts of liberum arbitrium.27 And so liberum arbitrium remains a necessary condition for moral responsibility, and Aquinas’s adherence to LA3 is vindicated.

2.2. An Asymmetrical Account of Moral Responsibility?

As we have seen in section 1.2, Aquinas affirms that both praise and blame (or reward and punishment, merit and demerit) presuppose that the agent has liberum arbitrium (and hence alternative possibilities). But so far, our discussion of moral responsibility has centered on blameworthy actions, either insofar as they are directly avoidable or as they are traceable to avoidable actions. A good case can be made that avoidability is presupposed in praise and blame in an asymmetrical way and that we must distinguish between two PAPs:28

(PAP-Blame) A person is morally blameworthy for what she has done only if she could have done otherwise.

(PAP-Praise) A person is morally praiseworthy for what she has done only if she could have done otherwise.

According to the asymmetrical account of moral responsibility, PAP-Blame is true but PAP-Praise is false. Some philosophers maintain asymmetrical responsibility above all on the basis of the following two considerations:29 First, it seems that we can praise rightly an agent who performs a good action that the agent could not avoid. Luther’s statement “Here I stand; I can do no other” is often given as a paradigm of freedom, resolution, and moral endorsement of a decision.30 Susan Wolf has made famous the scenario of a woman who jumps into the water immediately upon seeing a child drowning. Though it might be true that she could not do otherwise, her action is still morally ascribable to her, and praiseworthy.31 Wolf argues that determination “is compatible with an agent’s responsibility for a good action, but incompatible with an agent’s responsibility for a bad action”.32 She expresses her view succinctly as follows:

When we ask whether an agent’s action is deserving of praise, it seems we do not require that he could have done otherwise. If an agent does the right thing for just the right reasons, it seems absurd to ask whether he could have done the wrong. “I cannot tell a lie,” “He couldn’t hurt a fly” are not exemptions from praiseworthiness but testimonies to it. (“Asymmetrical Freedom”, 156)

Second, Dana Nelkin points out that PAP-Blame can be derived from the Kantian Ought-Implies-Can principle:

When people perform blameworthy actions, they do what they ought not to do and, instead, they ought to have done otherwise. Then, according to the Kantian thesis, it must be that they can do otherwise. Therefore, blameworthy actions require that their agents have been able to do otherwise. (Nelkin 2011, 99)

She argues that an analogous reasoning cannot be made for praiseworthy actions; there is no evidence that the fulfillment of an obligation presupposes the ability not to fulfill it (pp. 99–103). Moreover, in case of supererogatory acts, there is praise without obligation, so it is not the case that an action is praiseworthy only if it meets an obligation, and hence an obligation the person could have failed to meet (p. 103). Accordingly, Nelkin argues that moral responsibility for a good action does not require that a person could have acted badly (p. 15).

To these two arguments against PAP-Praise, one might add a theological consideration which has its roots in Augustine, was made famous by Anselm of Canterbury, and was well known to Aquinas. God and the blessed angels ought to be credited with the highest degree of freedom, and yet they cannot sin nor do evil; hence freedom cannot be defined as the ability to sin or not.33 Anselm’s solution is to define freedom as the “ability to maintain the rectitude of the will for the sake of rectitude” (De libertate arbitrii 3 and 13). Keeping the rectitude of the will for the sake of rectitude is certainly praiseworthy, and so God and the blessed angels are free and deserve praise although they cannot sin. In contrast, on Anselm’s account, sinners are able to maintain the rectitude of the will at least prior to their first sin, and so they are free and can be blamed for sinning precisely because they could have done otherwise (De libertate arbitrii 2). So Anselm, like Wolf and Nelkin, seems to deny PAP-Praise while affirming PAP-Blame. What about Aquinas? Does he likewise deny PAP-Praise, thinking a person can be morally praiseworthy although she did not have access to alternative possibilities when doing a praiseworthy action or the actions that be-long to its causal history?

2.3. Aquinas on Moral Praise and Alternative Possibilities

Answers can be found within several contexts. Aquinas’s most direct statement in favor of PAP-Praise is found in the argument for the existence of liberum arbitrium as a presupposition not only for blame, demerit, and punishment, but also for praise, merit, and reward (see section 1.2 above).

Another endorsement of PAP-Praise is found in a discussion of whether one merits in what one desires by necessity. The first opening argument³⁴ denies this, on the principle that we do not merit by what we desire naturally (QDV 22.7 arg. 1). The argument assumes that what we desire naturally, we desire by necessity, that is, not by liberum arbitrium. The context for this discussion is the desire for happiness, which according to Aquinas is a natural desire. In the response, Aquinas implicitly endorses the axiom on which the opening argument is based — that is, that we do not merit in what we do naturally and hence necessarily. He argues that the desire for happiness in general is not meritorious, while the specific desire for happiness in the divine vision, rather than in, say, bodily pleasures, is meritorious. The reason is that the generic desire for happiness is given by natural necessity, while any specific desire for happiness in this rather than in that depends on the person herself (QDV 22.7 c. lines 61–82).

There is a third context that shows that Aquinas accepts PAP-Praise, at least implicitly. Following Aristotle’s Nicomachean Ethics (1.12.1101b10–25), Aquinas distinguishes between praise (in the narrow sense) and honor (which is also praise in the broad sense). Aquinas maintains that praise (in the narrow sense) is owed to people because of the goodness they have in relation to something else, while honor is owed to people because of the goodness they have by themselves (secundum se). Thus human beings are to be praised because of their virtuous choices, while God is to be honored because of his intrinsic goodness, not because of his choices. God, too, makes choices, for example when he decides the manner in which he leads human beings to their final end; but with regard to these choices God deserves praise, not honor. Aquinas also says that virtuous acts are praiseworthy, since they are acts that promote the end, while the act of enjoying happiness is honorable (In Sent. 3.9.1.2 qc. 1 c.; QDV 23.1 ad 4). According to this account, honor does not require alternative possibilities, but praise does. For apart from situations in which there is a single possible means to an end (e.g., to cross the sea, one can only employ a ship, ST 1a.82.1 c.), the fact that praise concerns the means to an end implies that it concerns situations in which the agent has alternative possibilities.

A context that manifests a more nuanced attitude of Aquinas to-ward PAP-Praise is his discussion of promises, notably the vow given by a minor to enter religious life upon attainment of the canonical minimum age — a problem historical circumstances led Aquinas to discuss with particular fervor.35 Aquinas not only defends the legiti-macy of minors’ vows, but also thinks that it is altogether praiseworthy to make such a vow. Most interesting for our purposes is his discussion of an objection, which Aquinas formulates as follows:

For they say that something is more praiseworthy and more meritorious the more it is voluntary; but the more something is necessary, the less it seems to be voluntary. Hence it seems more praiseworthy and more meritorious that one does the works of virtue of one’s own accord, apart from the necessity of a vow or of obedience, than that one be compelled by a vow or by obedience to do such works.36 (DPSV 13 lines 53–60)

For Aquinas, however, the praiseworthiness of an action is not inversely proportional to its necessity, as the objection states, but rather directly proportional to the goodness of the will from which the action issues.37 The more firmly the will adheres to the good (stabilis in bono), the more an action is praiseworthy, just as the more obstinate the will is in evil, the more the action is detestable. By means of a vow, a person confirms her intention to fulfill the vow. Hence an action that is done from such a firm intention is more praiseworthy than the same action done without it.38 Aquinas’s direct response to the objection is this:

The point that what is necessary is less meritorious is to be understood about the necessity that is imposed against one’s own will. But when people impose on them-selves the necessity to act well, then this makes [the ac-tion] more praiseworthy, for they thus make themselves as it were slaves of justice, as the Apostle admonishes in the Letter to the Romans 6. Hence Augustine says in the Epistula ad Paulinam et Armentarium: “Fortunate necessity, which compels to what is better!”39 (CDR 13 lines 12–21)

It is important to note three things. First, the kind of necessity at hand is not absolute necessity, but necessity “from the end”, which, as mentioned above, is only conditional necessity, and the necessitating condition is in the agent’s own power: if one wants to remain faithful to the promise, then one must act so as to keep it (cf. SCG 3.138 n. 3121, ST 2a2ae.189.2 ad 2). Second, in this quotation Aquinas explicitly admits that a certain kind of necessity does make the action less praiseworthy: a necessity that is imposed on people against their will. Third, the necessity under which a person who made a vow finds herself in the present is the consequence of a freely made decision in the past — that is, it can be traced to an act of liberum arbitrium which was not subject to necessity. So this case is not a denial of PAP-Praise. Aquinas merely argues that a diminishment or lack of alternatives in the present does not imply a diminishment or lack of praise.

Acts of virtue are an analogous case. In language which recalls Susan Wolf’s lifesaving example, Aquinas speaks of the virtuous person’s necessity to act well, which is rooted in a virtuous disposition that makes the will pursue the good of virtue with greater force and less risk of failure. In fact, according to Aquinas, virtues make a person act firmly and promptly.40 This type of necessity he calls “necessity from an inner inclination” (SCG 3.138 n. 3120). Yet such necessity does not destroy liberum arbitrium; indeed, acts of virtue are precisely acts of liberum arbitrium: “An act of virtue is nothing else than a good use of liberum arbitrium” (ST 1a2ae.55.1 ad 2). This is reflected in Aquinas’s point mentioned earlier, that acts of virtues are those acts for which we are praised, but not honored, since they concern choices about the means and not the attainment of the end. The “necessity from an inner inclination” is in fact not absolute necessity, for in this life, no one is entirely exempt from sinning (QDV 24.9 c.) and so a virtuous person remains capable of acting contrary to her virtuous disposition.41 Fur-thermore, just as in the case of promises or vows, necessity from inner inclination can be traced to a past in which one was not yet under such necessity.

Aquinas denies PAP-Praise in none of the previous cases. But neither does he ever affirm that praise requires the ability to act either well or badly. In fact, he denies this directly in two further contexts. The first concerns the so-called confirmation of one’s liberum arbitrium in the good by divine grace. As just mentioned, according to Aquinas, one cannot be fully confirmed in the good in this life in such a way that one becomes absolutely unable to sin. This is a privilege of the blessed (QDV 24.9 c.). But divine grace can confirm a person in the good such that sin is made difficult and practically impossible. Aquinas raises an objection: it seems that one has more merit and is more praiseworthy in avoiding sin while being able to sin. He quotes Holy Scripture (Sir. 31:10) speaking of the just man who could disobey the law but did not (QDV 24.9 arg. 5).42 Aquinas replies that the notion of merit does not require the ability to sin; the ability to sin only makes merit manifest, because it shows that the good (i.e., meritorious) action is voluntary (QDV 24.9 ad 5). This is not a denial of PAP-Praise, however, for it leaves open the possibility that merit presupposes other kinds of alternative possibilities — not between sinning or not, but between doing or not doing a praiseworthy action that is not obligatory. An example for such an action is almsgiving (ST 1a2ae.108.4 c.).

Aquinas explicitly argues that praise presupposes alternative possibilities other than those between good and evil when he discusses the merit of Christ. According to Christian teaching, Christ was unable to sin, and yet he acted meritoriously. In an early text, Aquinas raises the objection that Christ could not merit, since to merit presupposes that one’s liberum arbitrium be not determined to a single outcome; this was, however, the case for Christ, since his liberum arbitrium was determined to the good (In Sent. 3.18.1.2 arg. 5). Aquinas’s reply contains two answers. In the first, he maintains PAP-Praise and merely denies that merit presupposes the ability to either sin or not:

It must be said that Christ’s liberum arbitrium was not de-termined to numerically one outcome, but to generically one outcome, namely to the good, because it could not tend to evil. Nonetheless, it can do this [particular act] or not do it. And this does not preclude liberum arbitrium, since the ability to sin is not the freedom of choice nor a part of freedom, as Anselm says. And this determination occurs from the perfection of liberum arbitrium, inasmuch as it is fixed by a disposition of grace and glory in that to which it is naturally ordered, namely the good, for al-though in us liberum arbitrium relates to both good and evil, liberum arbitrium does not exist for the sake of evil but for the sake of the good. (In Sent. 3.18.1.2 ad 5)

The first reply is coherent with PAP-Praise, because Christ had alternative possibilities in many of his individual actions, such as to choose Peter as his apostle but not Luke, or to turn water into wine or not, or to heal a person or not.

But in the same text, Aquinas proposes a second viable answer in which he denies PAP-Praise:

Alternatively, it must be said that even if [Christ’s liberum arbitrium] were determined to numerically one outcome, such as loving God (which it cannot not do), it neverthe-less does not forsake freedom, or the character of praise or merit, for it does not tend toward it by coercion, but of its own accord; and thus it is master of its act. (Ibid.)

In this second reply, moral responsibility receives only a minimal condition: absence of coercion. While this reply coheres with Aquinas’s general statement that the ability to sin is not a presupposition for merit (QDV 24.9 ad 5), it does not cohere with his denial in QDV 22.7 that one can merit in actions one does by natural necessity. In fact, later, in his QDV (the only instance in which he takes up this question again), Aquinas abandons the view expressed in the second reply; there he discusses the exact same objection as In Sent. 3.18.1.2 arg. 5, but without allowing for Christ’s merit apart from alternative possibilities:

It must be said that although Christ’s soul was determined to one outcome within the genus of morality, namely to the good, it was however not determined to one outcome simply speaking. In fact, it could do this or that, or not do it. Hence freedom remained in it, which is required for meriting.43 (QDV 29.6 ad 1)

The one exception from Aquinas’s early writing aside, then, Aquinas affirms PAP-Praise. Yet an asymmetry between blame and praise does appear: an individual is blameworthy for an evil action only if it was possible for her to do the good action instead; but it is not the case that, conversely, a person is praiseworthy only if it was possible for her to do the evil action instead. Praiseworthiness merely requires the ability to do this or that good action, or, when a good action is not obligatory, to do it or not. And it is not necessary for praiseworthiness that such alternative possibilities apply to the present action itself; the present action can be praiseworthy even if alternative possibilities apply only to the causal history of this action. Thus Aquinas agrees with Wolf and Nelkin that praiseworthiness does not require that the agent could have acted immorally instead of morally. But, contrary to Wolf and Nelkin, Aquinas holds that praiseworthiness does require that the agent had some kinds of alternative possibilities, and hence liberum arbitrium.

So the question of whether acts proceeding from liberum arbitrium are compatible or incompatible with necessity remains — not only for blameworthy but also for praiseworthy actions. In the next section, we will argue for an incompatibilist interpretation of both LA1 and LA2, on the basis of Aquinas’s consistent concern to refute determinist objections in various contexts in which liberum arbitrium is at stake.

3. Threats to Liberum Arbitrium

The first case we make in favor of an incompatibilist reading of Aqui-nas is an argument from silence. When discussing different threats to liberum arbitrium, which are threats of determination of human acts, Aquinas never adopts the compatibilist move of accepting determination while showing its compatibility with liberum arbitrium. By analogy, we argue that he does not adopt a compatibilist position with regard to the threat of intellectual determinism either. The threats Aquinas discusses take the form of objections to liberum arbitrium from the necessitation of choices by antecedent conditions. Aquinas explicity discusses at least five such threats. Two of these are transcendent threats — that is, concerning the relation of liberum arbitrium to God: the threats from God’s foreknowledge and from his efficacious will. The other three are immanent threats — that is, concerning the relation of liberum arbitrium to created causes: the threats from celestial bodies, from human passions, and from intellectual determination.44 Aquinas’s formulations of the transcendent threats have structural similarities to the “Consequence Argument” for incompatibilism.45 His formulations of the other threats do not follow this structure explicitly, but can be reformulated according to it. Aquinas can quite easily dismiss the threats from the celestial bodies and human passions, while the transcendent threats and the threat of intellectual determinism pose a more serious challenge to his incompatibilist stance; therefore they deserve more detailed scrutiny.

3.1. The Threat from Divine Foreknowledge

To show that Aquinas adopts an incompatibilist stance with regard to the threat divine foreknowledge poses to human liberum arbitrium, we will focus on a particularly lucid argument he formulates as one of the opening arguments of his major discussions of divine foreknowledge.46

It is structured as follows. First, the objection explicitly invokes a principle of transfer of necessity (PTN):

(PTN) If A is necessary, and if it is necessary that B follows from A, then B is necessary.47

That A is necessary implies that it is not up to X to make A not be the case, whatever kind of necessity be under consideration. Next, the objection maintains that this consequence is necessary: “If God believed p, then p”, because it belongs to the divine nature that he is infallible.48 Divine infallibility constitutes a law-like element (L) in the argument. Finally, the objection claims that the antecedent, ‘God believed p’, is not up to the agent because of the necessity or fixity of the past or because of the necessity of atemporal eternity.49 The antecedent concerns the relevant circumstances (C). In conclusion, the consequent, the thing or event known by God, is also necessary.50 Since God’s foreknowledge is comprehensive due to his omniscience, the conclusion implies that all future choices are necessitated. Manifestly, this objection has the structure of the Consequence Argument:51

(F1) It is not up to X at t1 to prevent that, if God believed at t0 / believes eternally that X would do A at t2, then X will do A at t2. [Divine infallibility]

(F2) It is not up to X at t1 to prevent that God believed at t0 / be-lieves eternally that X would do A at t2.52 [Factual premise and fixity of the past]

(F3) It is not up to X at t1 to avoid doing A at t2. [(F1), (F2), (PTN)]

Aquinas does not deny the validity of this argument, since it relies as a rule of inference on the PTN, which he accepts.53 And since this is a form of the Consequence Argument for incompatibilism, we can say that he accepts incompatibilism. But if foreknowledge implied the necessity of the foreknown object, as the argument suggests, this would exclude that the foreknown actions are done by liberum arbitrium. (Note that while this threat bears on leeway, sourcehood remains unaffected.)

Confronted with the threat of necessitation by divine foreknowledge, Aquinas’s way out is neither to give a compatibilist account of LA2, nor to reject LA2 and to stick only to LA1. He rather rejects the conclusion F3; in other words, he rejects the idea that the choice is necessitated, and he argues that the argument is unsound because of some fault in the premises. In fact, the known event, which is future and contingent in itself, is present and necessary as known by God’s eternal knowledge.54 Aquinas is using here an idea he borrows from Boethius, according to which a thing existing in its own mode (i.e., in the case at hand, temporal and contingent) is known according to the knower’s mode of knowledge (i.e., eternal and necessary).55 To this idea he adds that, in a conditional proposition — such as F1 — when the antecedent means something pertaining to an act of knowledge, then the consequent must be interpreted according to the knower’s mode of knowledge, not according to the mode of the known thing in itself. His example is this: “When I say ‘when the mind understands something, it is immaterial’, then this must be taken to mean that the thing is immaterial insofar as it is in the intellect, not insofar as it is in itself.”56 Accordingly, Aquinas adopts only qualified versions of F1 and F2: they should use the formulation ‘God believes eternally’ (and not at t0), and they should say “that X does A at t2 insofar as ‘X does A’ is present to God’s knowledge”.

From these modified premises, the fatalist conclusion F3 no longer follows, but only a different conclusion:

(F3′) It is not up to X at t1 to avoid doing A at t2 only insofar as ‘X does A’ is present to God’s knowledge.

Liberum arbitrium is no longer threatened, for Aquinas holds that we can avoid an event that is future in itself and for us, even though the same event is unavoidable insofar as it is present to God. The two frames of reference (temporality for us, eternity for God) not only make the same event future to us while present to God, but they also introduce two different modalities: contingent events are contingent in themselves, but necessary with reference to God (because of the necessity of eternity, which for Aquinas parallels the necessity of the present).57

However objectionable this solution may seem, the fact that Aquinas does not adopt compatibilism in his response is all that matters for our overall argument.58

3.2. The Threat from God’s Efficacious Will

Aquinas sees a parallel between divine (fore)knowledge and the divine will: both are “supremely certain and infallible” (certissima et infallibilissima), and yet he claims that they do not destroy the contingency of what is known and willed, including the acts of liberum arbitrium (ST 1a.23.6 c.). Even so, the so-called divine ‘efficacious will’, which entails that everything that happens is willed by God and that it is necessary that everything God wills actually happens, constitutes a threat which is probably the greatest challenge to Aquinas’s defense of liberum arbitrium. Aquinas discusses the compatibility of liberum arbitrium with the divine efficacious will principally in his treatments of the divine will, of divine providence (by which God orders all creatures to their final end), and of predestination (that special part of providence by which God orders the rational creatures — i.e., humans and angels — to the supernatural end of beatitude).59

The efficacious will of God is a threat not only to leeway, but also to sourcehood, for the divine will is the absolute ultimate source of human choices. As was mentioned in section 1.3, for Aquinas, every secondary (created) cause, such as the human will, depends for its causality on the primary (divine) cause and thus on God’s will, which is causal with respect to creatures. Thus they are subject to divine providence “like the particular cause is under the universal cause” (ST 1a.22.2 ad 4). But God’s causality with regard to human choices is more than just generic; since “divine providence produces its effects through the activity of secondary causes … that which happens by liberum arbitrium happens from predestination” (ST 1a.23.5 c.), i.e., from the particular way in which God exercises his providence with regard to rational creatures.

Aquinas does not explicitly address how sourcehood of the human will is safeguarded despite predestination in particular and the divine efficacious will in general; but we think it plausible to interpret Aquinas as an incompatibilist concerning sourcehood. For him, the fact that the divine will is, at the transcendent level, the absolute ultimate source of acts of human will does not eliminate the sourcehood of the human will, and thus the human will remains the ultimate source of its act in every sense relevant for free will. In fact, Aquinas does not conceive of the relation between the primary cause and secondary causes in the same way as a relation between two created causes.60 If a created cause other than myself caused my choice, then my choice would be totally caused by some circumstances that are not up to me, and I would not be its proper source. But God’s causality does not substitute itself for the causality of secondary causes, as occasionalist theories of divine causality maintain. Rather, just as when wood is burning, it is the fire (the secondary cause), not God (the primary cause), that burns the wood (QDP 3.7 c.), so also when I make an act of choice, it is I who make that choice and not God in my place (QDP 3.7 ad 12 and ad 13, QDM 6 ad 4). So God’s primary causality does not remove the will’s own causality. If it did, moral responsibility (which, as we have seen, Aquinas takes as given) would be annulled:

If the will were moved by another in such a way that it were not at all moved by itself, the acts of the will would not be imputed for merit or demerit. But since by the fact that it is moved by another, it is not excluded that it is moved by itself, as was said, therefore the ground for merit or demerit is not taken away.61 (ST 1a.105.4 ad 3)

As regards the threat for leeway, Aquinas has more explicit discussions. He considers in different places (notably in QDV 23.5 and ST 1a.19.8) the question of whether God’s will imposes necessity on all the things willed by God, and he always answers that God’s will imposes necessity only on some, but not all, things. In fact, he argues that if the divine will imposed necessity on all things, then liberum arbitrium would be destroyed (QDV 23.5 s.c. and ST 1a.19.8 s.c.). So Aquinas’s explicit statement about the threat of the divine efficacious will is an incompatibilist and libertarian reading of LA2.

           

The opening arguments to his most extensive discussion of God’s efficacious will, in QDV 23.5, all maintain that because the divine will is efficacious, it imposes necessity on all things, including human choices. Thus it is impossible for God to will X to happen without X happening (see esp. arg. 3 and arg. 4). In contrast, Aquinas himself argues that it is precisely because the divine will is universally efficacious that it does not impose any necessity on all of created nature. The efficacy of the divine will implies not only that certain effects happen, but also that they happen in the modality of either necessity or contingency; to produce necessary effects, God arranges necessary causes, and to produce contingent effects, he arranges contingent causes (QDV 23.5 lines 100–18).62 In the response to the third argument, Aquinas concedes that it is impossible that God wills X to happen and that X does not obtain, e.g., that God wants this person to be saved and that the person be damned. But he argues that it can be the case that God wills X to happen and wills for there to be a deficient power to produce the effect; in other words, Aquinas seems to say, God can want there to be a power — such as the human will — that can be deficient in producing its effect. So God can will Peter to be saved and yet it is possible in principle that Peter be damned.63 Aquinas’s expression is parsimonious in the extreme; more concretely, he might intend to say that it is possible in principle that Peter be damned because of Peter’s own will, which can be deficient, but that it is, as a matter of fact, impossible that he be damned, because, in point of fact, God wants him to be saved.

Later, in the Summa theologiae, Aquinas repeats the same solution in less detail, but using a new opening argument which formulates the threat of the divine will to contingency according to the structure of the Consequence Argument. Aquinas again begins by invoking a version of the PTN: “That which has necessity from something prior to it is absolutely necessary; for example it is necessary that an animal die because it is composed of contrary parts” (ST 1a.19.8 arg. 3). The divine will’s infallibility is a law-like premise (L): ‘If God wills p, then p.’64 The relevant circumstance (C), God’s actual will, is past or eternal, and hence necessary.65 In conclusion, “it follows that all that God wills is absolutely necessary” (ST 1a.19.8 arg. 3). Aquinas denies this conclusion. He accepts the PTN invoked in the opening argument, but he clarifies that that which derives necessity from something prior has necessity according to the mode of what is prior, which, in the case at hand, may be absolute or conditional necessity. Hence it is not true that everything is or happens by absolute necessity; some things are necessary only conditionally (ST 1a.19.8 ad 3). Conditional necessity is compatible with contingency, for among things that are condition-ally necessary, only those with absolutely necessary antecedents are absolutely necessary.66 Aquinas’s reply thus coheres with his answer in QDV 23.5, which he also repeats in ST 1a.19.8 c.: Because God wants certain effects to happen contingently, he arranged for them contingent causes (such as liberum arbitrium).

Aquinas has recourse to the same solution in his account of how divine providence does not threaten the contingency of the effects of contingent causes; to the contrary, God ordered things in such a way that contingent effects may have contingent causes in the first place (ST 1a.22.4 c.). And for the same reason, God’s predestination does not destroy liberum arbitrium (ST 1a.23.5 c.).

What Aquinas does not explain, however, is how the human will can will X contingently even though God’s will that X obtain cannot be forsaken. In other words, even though God wills acts of liberum arbitrium to be contingent, no human being can frustrate the divine will nor do anything to change it — not even by prayer (ST 1a.23.8) — and so it seems that the human will cannot will otherwise than God wills it to will, after all.67 It seems, then, that with regard to an act of human liberum arbitrium, Aquinas’s clarification regarding the PTN is pointless: though not absolutely necessary, what God wills in particular is beyond a human being’s control; hence the divine will’s necessity transfers on account of the PTN to the particular determination of a human being’s will here and now. In fact, we can construct an argument in analogy to the one in ST 1a.19.8 arg. 3, now specifically concerning an act of liberum arbitrium:

(W1) It is not up to X at t1 to prevent that, if God willed at t0 / wills eternally that X does A at t2, then X does A at t2. [God’s infallible will]

(W2) It is not up to X at t1 to prevent that God willed at t0 / wills eternally that X does A at t2. [Factual premise and fixity of the past]

(W3) ∴ It is not up to X at t1 to avoid doing A at t2. [(W1), (W2), (PTN)]

At this point, the interpreter might argue that Aquinas has no resources to address this argument and that his assumption that we control our acts of liberum arbitrium is incoherent with his theory of the divine efficacious will.68 Alternatively one might attempt to formulate an answer on the basis of other discussions by Aquinas. For example, he seems to say at least in one place that although the divine will is infallible, it does not directly determine each act of a human being. To make sure that a certain person attain her ultimate end of beatitude, he maintains, God provides a series of aids so that the person may either not sin or repent after having sinned (QDV 6.3 lines 225–36). Thus God can ensure that his will for the person’s salvation be fulfilled without determining at each step the person’s decision leading to it.

Whatever interpretation one might prefer, it remains that Aquinas’s explicit claim is that the compatibility between the divine efficacious will and human freedom is not to be understood as compatibility between necessity and human freedom. Once again, our concern is not the critical assessment of Aquinas’s solution. What is important for our argument is only that, in his view, God’s efficacious will does not destroy human liberum arbitrium as defined by LA1 and LA2, and that he defends his view without adopting a compatibilist interpretation of LA1 or LA2.

3.3. Threats from Created Causes

Aquinas also considers two threats to liberum arbitrium from created causes. One is a threat of external necessitation, which he discusses as causal determination of human choices by the “celestial bodies”.

Behind this idea is the cosmological view that all the events in the sublunary sphere, that is, on earth — including human choices — are determined by the constellations of the higher celestial spheres. Aqui-nas dismisses this threat by arguing that while the celestial bodies may have a necessitating effect on the senses (and hence on non-human animals), they do not necessitate the acts of intellect and will, since intellect and will are immaterial powers that are not directly affected by material objects.69

Aquinas also considers internal threats, based on the functioning of the psychological powers, dispositions, and states. If certain con-crete psychological circumstances beyond the individual’s control were causing the choice, then the choice would once again be neither free nor imputable, and sourcehood and leeway would be threatened. The real source would be in the external factors and out of the agent’s reach, and the choice would be necessitated by them. In considering the threat of necessitation by the passions in particular, Aquinas ac-knowledges that the sensory appetite, which is aroused by an external object independent from the agent, might move the will by presenting a certain object under a good and hence choiceworthy aspect that, apart from the passion of the sensory appetite, would not be perceived as good (e.g., ST 1a2ae.9.2). But Aquinas holds that, apart from mad-ness, the passions cannot move a person in such a way as to influence the agent beyond her control. In general he does not say that passions move the will, but rather that they may bind or absorb reason. Since the will, being a rational appetite, relies on the functioning of reason, the will’s movement ceases if the binding or the absorption of reason is total, as in the case of insane anger or lust (ST 1a2ae.10.3 and ST 1a2ae.77.7; see also QDM 6 lines 472–81). This is a major difference between human beings and other animals, which do in fact act accord-ing to their passions and are determined by them (QDV 24.12 and ST 1a2ae.10.3).

In short, Aquinas’s strategy in discussing threats from created

69. See In Sent. 2.15.1.3, QDV 5.10, ELP 1.14 lines 270–320, and more briefly QDM 6 c. lines 392–406. See also ST 1a2ae.9.5.

Aquinas on Free Will

causes is to insist on the non-material nature of the will, which entails that nothing material (celestial bodies or the senses) can necessitate the will. Once again, rather than adopting a compatibilist stance, his solution is to dismiss the conclusion that the will is necessitated.

3.4. The Threat of Intellectual Determinism

There is a further internal threat to liberum arbitrium, this time coming from the intellect: the threat of necessitation by reasons, that is, by the agent’s judgment about which option is best. Aquinas agrees that a sort of necessity could come from the object presented to the intellect and judged to be good — that is, desirable — if the judgment did not ultimately depend on us:

Likewise some wanted to destroy the other root of contingency which the Philosopher posits here — namely from the fact that we deliberate — because they wanted to show that the will is moved of necessity by the desirable object. For since the good is the object of the will, it cannot, it seems, turn away from it so that it does not desire that which seems good to it, just as reason cannot turn away from assenting to that which seems true to it. And so it seems that choice, which follows deliberation, occurs always of necessity, and thus everything of which we are the source by means of deliberation and choice would occur of necessity. (ELP 1.14 lines 462–74)

Aquinas’s direct response is that only happiness and what is under-stood to be indispensable for happiness is desired of necessity, while necessity to choose these. And for this reason the Philosopher explicitly designated the root of contingency in the things done by us on the side of deliberation, which concerns the things that promote the end and which are not [already] determined. (ELP 1.14 lines 506–17)

Yet this response does not effectively show that the deliberation leading to this choice rather than that is not intellectually determined after all. Though in principle deliberation may be open-ended, one may still worry that the whole psychological process by which the agent deliberates is in fact intellectually determined.

For the sake of clarity, we will now recast the argument for intellectual determinism along the structure of the Consequence Argument. The law or necessary conditional (L) now at stake expresses the link between an agent’s previous judgment concerning her best option and the actual choice. It has this form: It is not up to X to prevent that, if X judges definitively that A is to be done here and now, then X will choose to do A. (In other words, the choice of A is implied by the judgment that A is to be done: JA → CA.) What interpreters generally refer to as definitive (or last) practical judgment is the conclusive, all-things-considered judgment of practical deliberation (e.g., “All things considered, I should accept this job offer”), as opposed to a judgment that is made, but reevaluated as one continues to deliberate (e.g., “This job has a lot of advantages, but…”).70 The antecedent of this consequence, the judgment made about what to do, is supposed to occur causally and even temporally prior to the choice. Hence once the de-finitive judgment of what must be done here and now has occurred, it is subject to the fixity of the past. Thus the argument from rational necessitation can be formulated thus:

(R1) It is not up to X at t1 to prevent that, if X definitively judges at t0 that A is to be done, X will choose at t2 to do A. [Intel-lectualist premise]

(R2) It is not up to X at t1 to prevent that X definitively judges at t0 that A is to be done. [Fixity of the past]

(R3) It is not up to X at t1 to avoid choosing at t2 to do A. [(R1), (R2), PTN]

In short: I cannot choose otherwise because I cannot undo the practical judgment that entails my choice. So we can say that the judgment necessitates the choice, or that the intellect necessitates the will.

In order to consistently reject the threats of necessitation of liberum arbitrium, Aquinas has to address this argument from intellectual determinism. We think that it is plausible that he rejects a compatibilist solution with respect to intellectual determinism, as he has done with respect to the other threats; but whether he truly intends to do so, and whether he is successful, is controversial. So we will now briefly review current rival interpretations and defend our interpretation of Aquinas as an incompatibilist.

4. Four Rival Interpretations

In order to argue for the plausibility of a libertarian interpretation of Aquinas against those who read him as an intellectual determinist, it is crucial to study his account of intellect and will and their interaction, for, according to some, this account implies intellectual determinism. It is notoriously difficult to interpret, at two principal levels. First, Aquinas’s moral psychology of free choice has been interpreted differently, either as intellectualist or as voluntarist. We define intellectualism and voluntarism as follows:

(Intellectualism) A theory according to which every act of the will is fully accounted for by some act(s) of the intellect.71

(Voluntarism) A theory according to which there are acts of the will that are not fully accounted for by some act(s) of the intellect.72

A second factor complicating the interpretation of Aquinas’s account of intellect and will is the fact that Aquinas’s modal theory of free choice has received different interpretations. Those who see him as an intellectualist tend to view him as a compatibilist, while those who view him as a libertarian tend to consider him at bottom a voluntarist. Yet an intellectualist interpretation does not have to be compatibilist, and conversely even stressing the role of the will does not inevitably result in a libertarian interpretation. Accordingly, Aquinas can be and has been interpreted as an intellectualist compatibilist (IC), an intellectualist libertarian (IL), a voluntarist libertarian (VL), or a voluntarist compatibilist (VC). We will now discuss these four positions in order to show that the question of whether Aquinas is a compatibilist or a libertarian is not necessarily linked to the question of whether he is an intellectualist or a voluntarist. This helps bring into clear view what is essential to compatibilist interpretations of Aquinas, and thus it will allow us to argue more clearly that a compatibilist interpretation is at odds with some of his explicit pronouncements.

4.1. Intellectualist Interpretations

What is the relation between the intellect’s definitive, all-things-considered judgment that A is to be chosen (JA) and the will’s choice of A (CA)? Intellectualists hold that the will chooses A if and only if the intellect definitively judges that A is to be chosen (JA ↔ CA). Intellectualists furthermore hold that the definitive practical judgment JA is fully accounted for by the intellect, without any contribution originating in the will. Put differently, for intellectualists, the fact that one judgment rather than another becomes a ‘definitive’ judgment is to be traced to the intellect, rather than to the will. The implication CA → JA expresses what has been labeled by Scott MacDonald the ‘Doctrine of Essential Motivation’: everything that is willed is willed for a reason. In fact, as we mentioned in section 1.1, the will is a rational appetite; and hence whatever is willed by the will has to be apprehended by the intellect as good, that is, as worthy of pursuit. This is a principle that Aquinas affirms throughout his writings.73 The gist of the intellectualist reading resides in the converse implication JA → CA: the choice follows necessarily from the practical judgment, which is not only a necessary but also a sufficient condition for the choice.

Intellectualist interpretations split into intellectual determinist (and hence compatibilist) and indeterminist (libertarian) accounts: interpreters attribute to Aquinas either the view that a person’s judgments are not in the individual’s control because they are necessitated by past states of the universe, or the view that they are in her control and that they are not so necessitated.

The first, IC, interpretation was attributed to Aquinas by his 13th-century critics such as Henry of Ghent.74 It has also been proposed, at least as a well-grounded interpretation, by Jeffrey Hause.75 Its main argument is that the intellect does not control its own act, for it is not up to the intellect whether it understands something, whether it assents to a proposition, whether it is convinced or doubtful, etc.76 Even if it is granted that the will may control whether the intellect thinks of something or not,77 that the will can turn the intellect’s attention from one object to another,78 or that the will can make the intellect assent to a non-evident proposition,79 this is of no avail. For if the intellect does not by itself control its own act, then neither does the will, since, according to the IC interpretation, all activity of the will is fully accounted for by the intellect in the first place.

The second, IL, interpretation is defended by MacDonald on the basis of QDV 24.1–2.80 There Aquinas argues that human beings have liberum arbitrium because they control their definitive practical judgment (JA) thanks to their ability to judge their own judgments. Therefore, because the meta-judgment (JJA) can be otherwise, the practical judgment (JA) can be otherwise, and hence the choice (CA) can be otherwise. What this account does not explain, however, is why meta-judgments are not themselves necessitated. MacDonald mentions this problem but offers no solution (pp. 327–8).

4.2. Voluntarist Interpretations

We have seen some difficulties a strictly intellectualist account of liberum arbitrium faces in making room for alternative possibilities. For those who nonetheless read Aquinas as a libertarian, one strategic move is to introduce voluntarist elements into his account of the relation between practical judgment and choice.

Voluntarists would not deny CA → JA — in other words, they too claim that choice presupposes a rational consideration, but they would deny that CA is fully accounted for by the intellect. Either they deny the reverse implication JA → CA altogether — that is, they admit that I could definitively judge that I should choose A now and nonetheless not choose A (we label these “type-1 voluntarists”) — or they hold that JA has been produced with some involvement of the will that cannot in turn be fully traced to the intellect (we label those “type-2 volunta-rists”). Type-1 voluntarists hold that, given the same initial knowledge conditions, one can definitively judge in only one determinate way, but one can choose differently since the choice need not follow the de-finitive judgment; type-2 voluntarists hold that, given the same initial knowledge conditions, one can definitively judge differently and only for this reason can one choose differently.81 In any event, for voluntarists of both types, the will controls whether or not to make the choice. In one text, Aquinas seems to affirm type-1 voluntarism:

No matter how much reason gives preference to one option over another, one of them is not yet accepted over the other as to be done until the will inclines more to one than to the other. For the will does not follow reason of necessity. (QDV 22.15 lines 51–6)

But shortly thereafter, Aquinas appears to deny type-1 voluntarism, for he writes that there can never be discord between the intellect’s definitive judgment and the response of the will:

But a judgment concerning this particular action to be done now can never be contrary to our [rational] desire [appetitui]. (QDV 24.2 lines 79–81)

Elsewhere, Aquinas likewise binds the will’s choice strictly to the intellect’s definitive judgment.82 Accordingly, voluntarist interpreters of Aquinas tend to see him as a type-2 voluntarist. The two quotations can be easily harmonized: the one from QDV 22.15 indicates that a judgment does not become definitive without the will’s adherence to it — or, put differently, that judgment is the definitive judgment to which the will actually adheres. Like MacDonald, some voluntarist interpreters argue that the agent controls JA, but unlike MacDonald, they explain this by means of the will’s involvement in giving content to JA. Yet, they argue, for being involved in giving content to JA, the will does not in every respect require full determination by some previous judgment. This interpretation affirms voluntarism, for it holds that there are acts of the will that are not fully accounted for by acts of the intellect; and the interpretation is libertarian, because it denies determinism. Such a VL interpretation has been argued for by David Gallagher.83

Interestingly, there can be a voluntarist compatibilist reading of Aquinas as well; such a reading has been proposed by Robert Pasnau.84 He opposes an intellectual determinist reading of the relation between intellect and will: the will can always reject a practical judge-ment if it conflicts with the will’s higher-level volitions. Yet what ul-timately shapes the higher-level volitions — long-term goals, dispositions, and desires which are independent from reason’s judgments — is in turn subject to necessity. And these higher-level volitions ne-cessitate our choices.

4.3. Compatibilism and Conditional Freedom

Compatibilist readings of Aquinas have to clarify how his affirmations that our choices are not necessitated (LA2) cohere with the assump-tion that Aquinas would subscribe to determinism, for determinism implies that our choices are subject to necessity after all.

One forceful compatibilist interpretation of Aquinas’s account of liberum arbitrium is the one just mentioned, by Pasnau. The problems Pasnau addresses concern Aquinas’s modal theory of choice, not his moral psychology; they are therefore common to IC and VC readings. As Pasnau points out, the fact that our choices are not necessitated can be given a strong, libertarian sense, or a weak, compatibilist sense. The strong sense is that our choices are not necessitated given the previous entire state of the universe. The weak sense, which Pasnau defends, is that human beings do not always and unchangeably do the same thing (p. 231). At the core of Pasnau’s interpretation is the distinction between absolute and conditional necessity. Certain acts of the will occur by absolute necessity, especially our desire for happiness. Other acts of the will, namely our choices of this or that particular good, are subject to conditional necessity (without the necessitating condition being in the agent’s power): “Given the entire state of the universe, including an individual’s higher-level beliefs and desires, a certain choice will inevitably follow” (p. 232). As Pasnau reads Aquinas, the actions of humans and the behavior of non-human animals85 differ only from the psychological viewpoint, not from the modal perspective, for he takes Aquinas to hold that humans and animals are equally subject to conditional necessity. They differ psychologically, because humans, but not animals, are able to subject their immediate judgments and appetites to higher-level beliefs and desires. Pasnau notes that this makes all the difference for the way in which humans and animals act (pp. 232–3).

Though Pasnau makes his case well, his reading does not seem to us the natural way of reading Aquinas’s claim that our choices are not necessary and that those of animals are necessary. Aquinas argues that an animal’s “judgment” (that a certain good is worth pursuing or a certain evil is worth avoiding) is “determined to one outcome” (determinatum ad unum), which means that, in a given situation, an animal cannot but judge in one particular way. Since an animal can judge in only one particular way, it cannot but act in one particular way. For this reason, Aquinas ascribes to animals only ‘conditional freedom’; that is, those alternative possibilities they have are conditional upon a given set of circumstances. In general, sheep can run or stand still; dogs can bark or be quiet. But within a particular set of circumstances, animals have no longer any alternative possibilities: when spotting a wolf, a sheep cannot but flee, and when excited, a dog cannot but bark (QDV 24.2 lines 104–37).

This notion of conditional freedom expounded by Aquinas comes close to the traditional ‘conditional analysis’ of “can” (cf. section 1.3 above), which had for a long time been a major tool for compatibilism. According to the conditional analysis, the word ‘can’ in the phrase ‘X can / could do otherwise’ should be analyzed as ‘X would do otherwise, if …,’ that is, if a certain condition is met: if X so decided, if X so willed, etc. Of course the condition itself may well be necessitated by antecedent factors, and so would be the corresponding action. With-in the conditional analysis of ‘can’, in one sense there is conditional possibility, for the agent could have done otherwise, and in different circumstances she would in fact have done otherwise; but in another sense, there is conditional necessity, for the agent could not do other-wise all things considered, because she could not change the overall circumstances surrounding her choice. Conditional freedom does not require anything more, and therefore is compatible with determinism.

Aquinas, however, seems to allow for the conditional analysis of ‘can’ only in the case of animals. He contrasts the conditional freedom he attributes to animals with the kind of freedom he attributes to human beings. The reason why he considers it a mistake to restrict human freedom to conditional freedom (pace Pasnau) seems to be precisely that conditional freedom remains bound by necessity. Aquinas writes that, in contrast to animals, “human beings are not necessarily moved by what happens to them or by the onslaught of passions, because they can accept or refuse them, and for this reason, humans have liberum arbitrium, but not animals” (QDV 24.2 c. lines 133–7). As we argued in section 3 and as we will further substantiate in section 5, for Aquinas, human beings are free not only from necessitation by their passions, but also from other kinds of necessitating circumstanc-es. Thus Aquinas seems to ascribe to humans not merely conditional alternative possibilities, but absolute alternative possibilities. In other words: taking into account that Aquinas makes room for chance, one could admit that there is a real or absolute possibility for an animal to act otherwise than it does. But in the case of animals, this does not come from an “ability” or “power” to do otherwise, since this possibility depends only on chance, not on the animal’s control. In contrast, as we read Aquinas, concerning what is in the agents’ control, given the state of the entire universe, animals are able to do only one thing, but humans are able to judge, choose, and do either one of contrary things.86

It is at this point that Aquinas’s moral psychology of liberum arbitrium affects his modal theory of liberum arbitrium. We agree with Pasnau that, for Aquinas, humans (as opposed to animals) can rise above the immediate impact of the external circumstances and of passions by virtue of higher-level beliefs and desires. But unlike Pasnau, we think that, for Aquinas, it is precisely because of the presence of higher-level beliefs and desires that human choices — at least some human choices — are not subject to necessity. For even granting that those higher-level beliefs and desires might be necessitated, for Aquinas they have no necessitating effect on our choices (as we will argue in section 5). Rather, our capacity for higher-level beliefs and desires can always provide our practical reasoning with alternative premises, and for this reason our practical reasoning and hence our choices have alternative possibilities in the strong sense. By contrast, Pasnau argues (without providing any textual evidence) that for Aquinas our choices do follow necessarily from those higher-level beliefs and desires. But, arguably, if Aquinas had held this position, he would have mentioned this pre-cisely in those texts in which he contrasts human and animal freedom.

So far we have argued for the plausibility of reading Aquinas as an explicit libertarian: he consistently resists the compatibilist temptation when discussing other threats to liberum arbitrium (section 3), and he refuses to restrict alternative possibilities available to humans to mere conditional alternative possibilities (section 4). But since intellectual determinism might be implied in Aquinas’s account of liberum arbitrium against his own intention, we must show that he has the theoretical resources to avoid intellectual determinism. If he failed to avoid it, he would implicitly be a compatibilist. Showing that he is not is the task of the next section.

5. Alternative Possibilities in Practical Reasoning

We have argued that if Aquinas considers determinism in general to be a threat to liberum arbitrium, then he must also consider intellectual determinism a threat. As we have seen, he rebuts an argument for intellectual determinism by pointing out that the will is not necessitated to will particular goods (see section 3.4 above). Thus he is certainly not explicitly advocating a compatibilist account.

And yet his Doctrine of Essential Motivation — that everything that is willed is willed for a reason (see section 4.1 above) — could appear to imply intellectual determinism. The will responds to nothing else but a reason for willing or not willing this or that. So, in the concrete situation, the reason for which the will acts seems to guarantee that it performs just this act. Although a particular good does not move the will to will it necessarily, if the intellect judges here and now that this particular good is preferable, it seems that the will cannot fail to will it. Likewise, although the will does not necessarily exercise its own act or that of another power, if the intellect judges here and now that the will should move itself or another power to its exercise, it seems that the will cannot but do so (cf. section 3.4 above).

What are the options, then? Bite the bullet and accept intellectual determinism? Or make exceptions to the Doctrine of Essential Motivation? In this section, we will argue that Aquinas has a third option. While we do not understand our argument as definitive evidence for an incompatibilist interpretation, we intend to show that such an interpretation is highly plausible, despite his insistence that all acts of liberum arbitrium are rationally motivated. For this, we have to look into his more complete account of choice-making. We will start with some general considerations about practical deliberation and practical inference leading to choice.

5.1. Practical Inference and Deliberation

To perform a deliberate action presupposes most fundamentally that one pursues an end, and then — understanding that it can be attained only by way of some means — that one intends to employ suitable means, after which one chooses among the various candidates (if there are in fact several suitable means). This rational activity has a logical structure, a “practical inference”. It connects the choice with two premises: one that expresses the end pursued (e.g., being no longer thirsty), the other that expresses the belief that a given action promotes the end (e.g., “Drinking this glass of water will quench my thirst”). The choice of means is (or follows upon) the conclusion of the practical inference: the choice is for an activity (I drink this glass of water) that allows me to attain the end.

Deliberation differs from practical inference. A practical inference is a logical form, while deliberation is a psychological process, comprising the whole course of investigation and thus consisting of all mental processes that lead to choosing a means (typically a specific course of action) to reach the proposed end. To be sure, many times deliberation is not made consciously and so the practical inferences are not brought to awareness, but the resulting actions can be explained in retrospect by the agent or by an observer as resulting from the end pursued and the beliefs about suitable means to the end. The difference between deliberation and practical inference is also clear in this: a practical inference has only one conclusion, whereas, according to Aquinas, deliberation is precisely that by which an agent has access to different options:

Since the will moves itself by means of deliberation, and since deliberation is an investigation that is not demonstrative but rather open to opposite routes, therefore the will does not move itself of necessity. (QDM 6 lines 377–81)

Human beings are master of their acts, both of willing and of not willing, because of reason’s deliberation, which can bend toward one side or another.87 (ST 1a2ae.109.2 ad 1)

In the process of deliberation, several candidates for action may be considered but then discarded, and in the end one candidate wins out over the others. In other words, in the course of deliberation, sever-al practical inferences may have been tested, but only one has been deemed successful. The successful one is the reasoning that explains why this particular action is to be done (or why it was done).

But does the agent truly have alternative possibilities? In order to safeguard the contingency of the choice, there are three possible strategies. First, one could argue that the individual could have had different occurrent beliefs — that is, the successful practical inference could have followed from different premises. Second, one could argue that neither the premises nor the conclusion of the practical inference could have been different, but that the choice of the will did not have to conform to the conclusion — in other words, that the will is free to diverge from the definitive practical judgment. Third, one could argue that the practical inference is not necessary: given the premises, the conclusion (that is, the definitive practical judgment that determines the choice) does not necessarily follow.

The first strategy is still compatible with determinism, because the idea that the individual could have had different premises can be given a conditional analysis: for example, had she experienced a different childhood, she would have desired different ends and she would have valued different means to achieve her ends. So this strategy is not robust enough to avoid compatibilism. The second strategy, conceding to the will the ability to act differently from the definitive practical judgment, avoids determinism, but at too high a price. We called this type-1 voluntarism. This strategy faces a theoretical obstacle in that it contradicts the Doctrine of Essential Motivation, for there seems to be no reason to act apart from the definitive judgment, since any reason to do so would simply result in a revision of the definitive judgment rather than in its rejection. As we have seen, it also faces textual obstacles as an interpretation of Aquinas’s thought, for Aquinas holds that the will cannot act at variance with the definitive practical judg-ment.88 Furthermore, Aquinas repeatedly — although not consistently — equates choice with practical judgment.89

In our view, the third strategy is not only philosophically success-ul, but also the strategy at least implicitly adopted by Aquinas: what saves the contingency of a choice is the fact that the conclusion of a practical inference does not necessarily follow upon its premises. We will now develop this idea further.

5.2. The Defeasibility of Practical Inferences

Building upon Elizabeth Anscombe’s analysis of practical reasoning, we will now argue that such reasoning is open-ended because it does

not operate with a definite set of premises.90 Thus the agent controls her definitive practical judgment and thereby her choice.

What makes human choices contingent is that, while the premises of a practical inference might remain unchanged and its soundness undisputed, the practical inference can be defeated. In contrast to sound theoretical inferences, sound practical inferences are non-monotonic (which means that a sound inference can be undone by adding new information). A practical inference can be defeated by the addition of a new premise which blocks the conclusion drawn from earlier premises, so that the considered means (M) toward the intended end (E) is abandoned. Thus the premises that in fact lead to a given conclusion could have led to a different conclusion. Premises that lead to a conclusion do so necessarily only if they are the only premises considered, but not if they are part of a larger set of premises. For example, I need to go quickly to New York. Reasoning from the premises that I need to go quickly to New York and that flying is the fastest means for getting there, I will conclude that I should fly. But I can add a further premise, which expresses a new end: I also want to save money (E′). This end is compatible with the initial premise (E) that I want to go to New York, but not with the conclusion reached: now I conclude that I should drive rather than fly. Or my additional end might be: “I should spend more time with my family” (E ′′). This end is incompatible with the initial premise (E), and I will now conclude that I should renounce my trip to New York and stay home. In both cases, adding a further premise leads to a different conclusion.

The process of deliberation prior to making the definitive practical judgment consists precisely in this: considering certain premises, adding new premises and maybe disregarding previously considered premises, until one arrives at two premises on which one settles, from which one then concludes the definitive practical judgment that makes one choose accordingly. The choice is contingent, because there is no necessity to conclude from this set of premises rather than from that. What is more, the choice is even contingent if one or more premises express a necessary statement (such as the necessity of an end, e.g., “In order to go from Boston to New York in less than two hours, I need to fly”), because it is not necessary to conclude from these premises rather than those. It is for this reason that higher-level beliefs and de-sires, which constitute premises in our practical reasonings, do not ne-cessitate our choices even if those beliefs and desires are themselves necessary (cf. p. 25 above).

It is true that sometimes a simple calculation might suffice to decide what to do in order to achieve a specific end. In this case, one can indeed admit that the reasons necessitate the decision, because they are reasons for doing this (M) rather than that (M'), or reasons for do-ing this (M) rather than not. These are contrastive reasons. In Aristote-lian terms, this is proper to technical reasoning, which operates within a given set of ends (the simplest case would be with one end only), which normally allows for contrastive reasons. A physician makes the (technical) medical decision to abandon the resolution of amputation because a less invasive way of saving the patient is possible. When all the medical considerations of such possibilities have been made, the reasoning is no longer defeasible, and the reasons that remain at the end of the deliberation are contrastive, that is, reasons for doing such-and-such rather than not. Such contrastive reasons indeed neces-sitate the conclusion. Yet this is not so with practical reasoning, that is, reasoning concerning life in general. When, say, an army physician abandons the goal of saving the patient (the technical end of the phy-sician), she does so because the prospects of recovery are dim or be-cause she is more needed in another place. This deliberation implies considerations of a non-technical sort and introduces ends external to the profession, such as political or military ends. This is not a medical decision, nor does it involve technical reasoning, but it is a properly practical decision. The physician is a human being who can abandon her goal as a physician, and thus she might give up the conclusion she reached when reasoning as a physician. Her contrastive reasons to do such-and-such as a physician were not contrastive reasons overall.

One might object that the final conclusion of practical reasoning comes about as the result of a combination of both types of considerations, that is, technical and practical. All means are compared, all ends are considered, so as to have overall contrastive reasons for do-ing this rather than that. But this objection starts from the supposition that a finite collection of available means and ends is at play; and, as we have just seen, this is true only of purely technical reasoning: once practical reasoning comes into play, the reasoning is not constrained by a finite set of ends, and hence by a finite set of means — either be-cause there might be an infinite number of other ends to consider, or simply because the ends the agent considers are not commensurable. There are no overall contrastive reasons if the different ends cannot be rationally ranked by the agent.

5.3. Aquinas on Practical Reasoning

Although Aquinas does not develop the idea of the defeasibility of practical reasoning systematically, the idea is present in several state-ments he makes in various contexts.

The metaphysical foundation for the defeasibility of practical reasoning is the fact that no particular good fully instantiates the ‘univer-sal good’, which is the proper object of the will. “The good is mani-fold [bonum est multiplex] and therefore [the will] is not determined of necessity to one outcome [ad unum]” (ST 1a.82.2 ad 1). Just as the housebuilder can realize the general idea of a house in various ways, so it is possible to will any particular good under the formality of the universal good.91 Therefore, the point of view under which an action is judged as good, and hence to be done, may always be abandoned in favor of another aspect under which it is not good or at least is less good than the alternative.

In certain texts in which the defeasibility of practical reasoning is at work, Aquinas does not explicitly evoke the idea of practical reasoning as an inference from premises, but what he writes could be expressed in this form:

If a good is such that it is not found to be good accord-ing to all particular aspects that can be thought of, then it does not move [the will] of necessity, even as to the speci-fication of the act. In fact, one can will its opposite even while thinking of it, because maybe it is good and suitable according to some other particular aspect that is thought of. For example, what is good for health is not good for pleasure, and likewise concerning other matters.92 (QDM 6 c. lines 441–9)

Expressed in terms of the defeasibility of practical reasoning, this pas-sage states that one may have deliberated about how best to achieve health and may have concluded concerning which is the best means (say, that one should renounce sweets), but then one may reject this conclusion because one introduces a new premise: one ought not to renounce occasions for pleasure. A very similar idea is expressed in this passage:

Not every cause produces its effect of necessity, even if it is a sufficient cause, because the cause can be impeded so that sometimes its effect does not come about. … There-fore it is not necessary that the cause which makes the will will something do so of necessity, because it can be impeded by the will itself, either by removing the con-sideration which makes the individual will something, or by considering the opposite, namely that that which is proposed as good is not good according to a certain aspect. (QDM 6 ad 15)

Removing a consideration means abandoning a premise of a practi-cal inference; considering its opposite means adding another premise to it; either way, the choice is not necessitated.

In other passages, Aquinas evokes directly the idea of practical reasoning from premises. Consider this objection arguing for intellectual determinism, and Aquinas’s response:

Choice follows the judgment of reason about things to be done. But reason judges of necessity about certain things because of the necessity of the premises. Therefore it seems that also the choice follows of necessity. (ST 1a2ae.13.6 arg. 2)

The decree or judgment of reason about what is to be done is about contingent matters that can be done by us. In such things, the conclusions do not follow of necessity from principles that are necessary by absolute necessity, but from those that are necessary only conditionally, such as “if he runs, he is in motion”. (Ibid., ad 2)

Even if Aquinas does not here explicitly explain the contingency of the conclusion by the defeasibility of the practical inference, this is what we take him to have in mind when he says that the conclusion follows from premises that are necessary only by conditional necessity. Con-sider again the example of wanting to go from Boston to New York in less than two hours. There is only one means available to achieve this end: traveling by air. Thus the premise about the means is a necessary premise: in order to achieve this end, it is necessary to travel by air. Nonetheless, the choice does not follow of necessity, because air travel is necessary only if indeed I want to achieve this particular end. But I remain free to renounce this end for the sake of a different end, such as saving money.

In explaining incontinence, Aquinas articulates the defeasibility of practical reasoning in syllogistic form.93 Inspired by Aristotle’s account of what makes incontinence (akrasia) possible (Nicomachean Ethics 7.3), Aquinas explains that the syllogism of the continent and the one of the incontinent have more than two premises, because the conti-nent and incontinent are wavering between two rival major premises, one prohibitive, the other permissive. When combined with a corre-sponding minor premise, this leads to either a prohibitive or a permis-sive conclusion:

No fornication is to be committed Everything pleasurable is to

be enjoyed

This act is fornication This act is pleasurable

--------------------------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------

This act is not to be done This act is to be done

Aquinas’s example of the continent and the incontinent considering both major premises is a case where the practical reasoning is defeat-ed by adding another premise. Incontinent people add “Everything pleasurable is to be enjoyed” to the premise “No fornication is to be committed”, whereby they conclude that the act of fornication is to be done. In contrast, the continent dismiss the permissive premise and conclude only from the prohibitive premise that no fornication is to be committed, and so they conclude that the act of fornication is not to be done.

Since the conflict here is between reason and passion, one could of course object that the incontinent act irrationally in that they do not follow their better judgment. But, for Aquinas, the decision of the

incontinent, which can be called irrational, objectively speaking — is rationally made, for it is a conclusion that follows upon a sound practical inference.94 The incontinent could have remained continent by reasoning with the prohibitive major premise, thus arriving at the opposite conclusion. Aquinas’s explanation of the reasoning of the incontinent illustrates nicely how alternative possibilities become avail-able to the agent, defeating her initial practical inference with the introduction of a new (permissive) premise. It also shows that in either case, whether she acted continently or incontinently, the individual acted for some reason.

Compare this with the case of a practical choice like the choice the army physician has to make between immediate care of the patients right here and the better help she would afford to the army by going elsewhere. Just as in the case of the continent and incontinent, we could see the choice as one between two conclusions of two practical syllogisms. Both conclusions and thus both actions would be done for some reason. But now there is perhaps no emotional bias in favor of one reason rather than the other, nor is there a reason that necessitates one over the other by making a definite contrast between them. It is a contingent, but still a rational, choice.

If the prevailing set of reasons does not determine the choice because it weighs more than the alternative set of reasons, then we must attribute to the agent the power of weighting one set over the other. This weighting is what happens when the agent stops deliberating and settles on a specific choice. Aquinas does not make this point explicitly, but without it, nothing would prevent the deliberation from going on indefinitely.

What makes an agent settle on her reasons cannot be fully explained. It would be implausible that it would in turn require a practical inference, for this would imply an infinite regress. Also, it seems clear that for Aquinas it is not chance that bridges the hiatus between non-contrastive reasons and action. It is rather the agent herself, to whom Aquinas refers when he speaks of the ‘will’.96 The language of ‘will’ marks the control Aquinas wants to attribute to the agent. A causal conception of practical reason has the same effect. Through her will, the agent has a special causal power on the world, which is more than the simple absence of determining causality that we associate with chance.

This positions Aquinas as a theorist of agent causation.97 But his account of agent causation only proposes the idea that the agent is a cause of some state of affairs by acting for reasons. This is what sets the agent’s choices and actions apart from other indeterminate events. And this is what Aquinas often expresses as the non-necessitation of the will by reason, which we have interpreted as the non-necessitation of choice by practical reasoning, that is, by the reasons for which the choice is made.

One may object that the absence of contrastive reasons makes choices irrational and that the defeasibility of practical reasoning provides insufficient ground to answer this objection. But libertarians would not find this objection decisive, and they would credit Aquinas’s analysis with being sufficient to account for the rationality of freely made choices. They would in fact argue that demanding contrastive reasons for every choice contradicts the very idea of the agent’s ability to do otherwise. And so they would count Aquinas as a libertarian, in accord with our main argument.

Conclusion

We intended to show mainly two things: that Aquinas does not intend to be an intellectual determinist, and that his account of practical reasoning allows him to avoid intellectual determinism. As a preliminary to the discussion of intellectual determinism, we discussed Aquinas’s distinction between free will broadly conceived as implying only sourcehood, and liberum arbitrium as presupposing in addition alternative possibilities. We also clarified the importance of liberum arbitrium for Aquinas’s moral philosophy and moral theology. We have argued that for Aquinas, an individual is morally responsible for an act only if she had alternative possibilities at the moment of acting or in the causal history leading up to the act. An individual merits / deserves praise or demerits / deserves blame only if she acted from liberum arbi-trium, that is, only if she had alternative possibilities. Yet there is a certain asymmetry. Demerit / blame presupposes alternative possibilities between good and evil — a person demerits / deserves blame only if it was open to her to do what is good instead of what is bad. In contrast, merit / praise does not presuppose alternative possibilities between good and evil, but merely that it was open her to do a different good act or no act at all.

Our first argument against attributing intellectual determinism to Aquinas was that he would most likely reject compatibilism in general and hence also the compatibility of liberum arbitrium with the necessitation of our choices by our intellect, because he consistently avoids adopting a compatibilist solution in his discussions of the other threats of necessitation of liberum arbitrium. For us, the paradigm cases are his refutations of the objections that our choices are necessitated by divine foreknowledge and by God’s efficacious will, because he presents these arguments with particular clarity by adopting a similar form as the Consequence Argument. Throughout his corpus, Aquinas takes an incompatibilist stance: he accepts the validity of the arguments in favor of the necessitation by God’s knowledge and will, but argues that they are unsound, thus avoiding a compatibilist solution. We reconstructed a possible argument for the necessitation of our choices by our practical judgments according to the same structure.

Since Aquinas does not address the threat of intellectual determinism in any detail, and since he expresses himself ambiguously with respect to the extent to which the will depends on the intellect, we discussed paradigmatic interpretations that see Aquinas either as a compatibilist or as a libertarian, and moreover either as an intellectualist or as a voluntarist. We argued that Aquinas rejects an account that sees human freedom as mere conditional freedom, and hence as compatible with determinism.

We finally argued that Aquinas actually avoids intellectual determinism by the way in which he conceives of practical reasoning. We reconstructed his account according to the idea of the defeasibility of practical reasoning. In the interpretation we propose, Aquinas affirms consistently the Doctrine of Essential Motivation: each choice is done for a reason. Yet, at the same time, Aquinas avoids intellectual determinism, because the reason that in fact motivates the choice was not the only reason available to the agent; she could indeed have acted for a different reason.

We do not claim to have demonstrated the impossibility of a compatibilist interpretation of Aquinas. But, given his persistent refusal of compatibilist solutions in responding to various threats to liberum arbitrium (section 3), his rejection of the conditional analysis of free will (section 4), and the fact that his own theory of free will as rooted in his account of practical reasoning is exempt from intellectual determinism (section 5), we find it most plausible to read Aquinas as an incompatibilist concerning intellectual determinism.

1.      Aquinas’s works are cited according to the best available editions — with page or line numbers wherever useful — which for most works is the “Le-onine edition” (Aquinas 1882–). Works cited according to the Leonine edi-tion are Contra doctrinam retrahentium a religione (CDR), De perfectione spiritu-alis vitae (DPSV), Expositio libri Peryermenias (ELP), Quaestiones disputatae de malo (QDM), Quaestiones disputatae de veritate (QDV), Quaestiones de quolibet (Quodl.), and Summa theologiae (ST). Works cited according to other editions are Quaestiones disputatae de potentia (QDP), Scriptum super libros Sententiarum (In Sent.), Summa contra Gentiles (SCG), and Super Romanos (In Rom.). We refer to Augustine and Boethius according to the standard editions contained in Corpus Scriptorum Ecclesiasticorum Latinorum (CSEL) and Corpus Christia-norum Series Latina (CCSL). All translations are our own. The translation of liberum arbitrium poses difficulties, above all because its definition was con-troversial among medieval thinkers, and for this reason it would have to be translated differently for each author. We therefore prefer leaving liberum arbitrium in Latin. Aquinas defines it as a power for making free choices; see for more details section 1.2 below. Shortly after Aquinas, and partly in reaction against him, the preferred term became ‘free will’ (libera voluntas); see Kent 1995, 104–6.

2.      See Spiering 2015.

3.      Augustine, De civitate Dei 22.30, CCSL 48: 863–4; see also De correptione et gratia 12.33 and 12.35, CSEL 92: 259 and 261–2, Enchiridion 28.105, CCSL 46:106.

4.      See In Sent. 2.25.1.5, ST 2a2ae.183.4, In Rom. 8.4 n. 666.

5.      ST 1a.82.1; for the different kinds of necessity, see also ST 3a.46.1.

6.      QDV 22.5, ST 1a.82.1–2, ST 1a2ae.6.4.

7.      As to the free but necessary desire for happiness, see QDV 24.1 ad 20, QDP 10.2 ad 5; cf. ST 1a.81.1 ad 1. As to the free but necessary love of God by the blessed, see SCG 3.138 n. 3120, CDR 13 lines 25–6. As to the necessary but free divine self-love, which results in the free but necessary production of the Holy Spirit, see QDP 10.2 ad 5.

8.      See, e.g., SCG 3.138 n. 3120, QDV 24.1 ad 20, QDP 10.2 ad 5, ST 1a.82.1 ad 1.

9.      Elsewhere, Aquinas uses similar formulations. Without liberum arbitrium, there would be no merit and demerit, no just reward or punishment (QDV 24.1 c. lines 226–30). If freedom of the will (taken here in the sense of liberum arbitrium, as incompatible with necessity) were eliminated, then the praise of human virtue would be void, the person who punishes or rewards would no longer be just, and there would no longer be any thoughtful counsel (SCG 3.73 n. 2491). If human beings were moved by necessity to will something without being able to avoid it, there would not be merit or demerit, and there would be no counsel, exhortation, precept, punishment, praise, or blame (QDM 6 c. lines 248–68).

10.  We understand by “acts proceeding from liberum arbitrium” those acts that are properly acts of liberum arbitrium (choices) as well as the acts derived from them (chosen actions).

11.  As defined by Frankfurt 1969. The PAP is nothing more than the principle according to which LA2 is a necessary condition for moral responsibility. In contrast, LA3 states that not only LA2 but also LA1 is a necessary condition for moral responsibility.

12.  Concerning God’s possession of liberum arbitrium, see In Sent. 2.25.1.1, QDV 23.4, QDV 24.3, SCG 1.81–8, ST 1a.19.10; concerning his ability to create a bet-ter world than the present one, see In Sent. 1.44.2 and ST 1a.25.6. See also Wippel 2007.

13.  If one wants to express this point in terms of laws: it is irrelevant whether the laws in question are laws of nature or some other sort; the point is that, in conjunction with certain circumstances, which are not in the agent’s control, they necessitate the action.

14.  Of course, determining external factors that would themselves depend on the agent’s activity would give the agent an indirect responsibility for the actions and / or choices determined by such factors. We will be concerned hereafter only with cases of direct responsibility.

15.  For a thorough account of the necessary conditions for moral responsibility as compatible with determinism, see Fischer and Ravizza 1998.

16.  If one understands libertarianism as presupposing absolute ultimate sourcehood, then Aquinas cannot be labeled a libertarian. For a detailed discussion of this point, see Shanley 2007, especially 84–9.

17.  For the general statement of source-incompatibilism, see Stump 1996, Hunt 2000, and Widerker 2006.

18.  See note 19 below.

19.  See Stump 2003, 294–306.

20.  For recent discussions, see Fischer 2011.

21.  Stump also refers to texts that explicitly say that some acts of the will are compatible with necessity, but those acts are not proper choices between available alternatives. Recall that for Aquinas not all acts of the will are acts of liberum arbitrium; see above, section 1.1.

22.  See above, p. 6 and note 9.

23.  More precisely: under the condition of original sin (after the Fall and without grace), one cannot avoid mortal sin for very long, and even with the help of grace, one cannot avoid venial sin; see QDV 24.12–3 and ST 1a2ae.109.8. The difference between mortal and venial sin consists essentially in the fact that mortal sin destroys charity, whereas venial sin does not. See QDM 7.1 lines 277–316, ST 1a2ae.72.5.

24.  See also QDV 24.12 ad 8 in opp.: “A human being … can avoid individual sins, although not all.”

25.  This might be the reason why Aquinas speaks about “voluntary sins” (see the quotation of ST 1a2ae.74.3 ad 2, p. 9 above), leaving room for involuntary sins, which are consequent upon voluntary choices.

26.  QDM 16.5 ad 11; see also SCG 3.160 n. 3318, QDM 3.10 lines 79–83. Cf. Aristo-tle, Nicomachean Ethics 3.5.1113b30–3.

27.  For a detailed discussion of Aquinas’s account of tracing exceptions to PAP, see Furlong 2015.

28.  We use the formulation of Nelkin 2011, 98.

29.  We follow here the “Rational Abilities View”, advocated by Nelkin 2011, and the “Reason View”, espoused by Wolf 1990; see also Wolf 1980.

30.  See, e.g., Watson 2004, 100–6, discussing the interpretations of Luther’s statement by Daniel Dennett, Harry Frankfurt, Bernard Williams, and Robert Kane.

31.  Wolf 1990, 58–61.

32.  Wolf 1980, 158.

33.  Augustine, Contra Iulianum opus imperfectum 5.38 and 6.10, CSEL 85/2: 237 and 312; Anselm of Canterbury, De libertate arbitrii 1, ed. F. S. Schmitt, vol. 1: 207–9. Aquinas refers to the famous passage of Anselm repeatedly; see QDV 22.6 co. lines 161–3, QDV 24.3 s.c. 1, QDV 24.10 arg. 10 and arg. 11.

34.  The opening arguments are the arguments by which the scholastics intro-duce a quaestio (or articulus) — that is, the discussion of a particular ques-tion at hand. They are usually objections against the position defended in the quaestio.

35.  Aquinas discussed this especially toward the end of his career, in a polemic with Gerard of Abbeville; see Boureau 1998. We thank Stephen Metzger for suggesting this article.

36.  For a similar formulation, see CDR 11 lines 21–8.

37.  CDR 12 lines 12–5: “Since the praise of a work depends on the root of the will, an external work is rendered more praiseworthy to the extent that it proceeds from a better will.”

38.  CDR 12 lines 15–39, DPSV 13 lines 148–59.

39.  Augustine, Epistula 127.8, CSEL 44: 28: “felix est necessitas, quae in meliora compellit”.

40.  See, e.g., ST 1a2ae.100.9 c., ST 1a2ae.107.4 c., De virtutibus in communi 9 c., ad 13, ad 14.

41.  See Kent 2013.

42.  Similarly, Peter van Inwagen, quoting Mark Twain (“I am different from Washington; I have a higher, grander standard of principle. Washington could not lie. I can lie, but I won’t”), argues that the impossibility to perform a blameworthy action goes counter to its moral responsibility; see van Inwagen 1983, 63–4.

43.  Aquinas uses the word ‘freedom’ (libertas) here in the sense of liberum arbi-trium, as is clear from his insistence on alternative possibilities and from the fact that liberum arbitrium is the term that was used in the opening argument to QDV 29.6 arg. 1, to which this text is the reply.

44.  While Aquinas usually examines these threats in separate contexts, in one place we find him discussing all but the threat from human passions; see ELP 1.14, pp. 75–9.

45.  van Inwagen 1983, introduction and chapter 3, for the canonical version of this argument. He was anticipated by Ginet 1966 and Wiggins 1973. We do not claim that Aquinas’s version is similarly well crafted, but the basic idea is there.

46.  In Sent. 1.38.1.5 arg. 4, QDV 2.12 arg. 7, ST 1a.14.13 arg. 2. For a perceptive discussion of this argument and of Aquinas’s response, see Prior 1962. For a recent study of Aquinas’s doctrine of divine foreknowledge and human free-dom, see Goris 2005. See also Wippel 1984.

47.  See QDV 2.12 arg. 7 lines 58–60: “In every true conditional, if the anteced-ent is absolutely necessary, then also the consequent is absolutely necessary.” See, for an almost identical formulation, ST 1a.14.13 arg. 2. In his earliest dis-cussion of this problem, in In Sent. 1.38.1.5 arg. 4, Aquinas goes into more de-tails: the consideration is made that the truth of the antecedent of a necessary consequence (entailment) does not lead to the necessity of the consequent, but only to its truth, and hence the consequent is only conditionally necessary. But against this it is argued that, if the antecedent of the conditional state-ment is itself absolutely necessary, then the consequent will indeed be abso-lutely necessary.

48.  ST 1a.14.13 arg. 2: “But this is a true conditional: ‘If God knew that this future event will happen, it will happen’, because God’s knowledge concerns only true statements.” See also QDV 2.12 arg. 7 lines 60–2. In In Sent. 1.38.1.5, the conditional ‘If God believed p, then p’ was made in the previous opening argument, arg. 3. We prefer phrasing the argument with the term ‘believes’ rather than ‘knows’, because this states the difficulty with greater precision, see note 52 below.

49.  ST 1a.14.13 arg. 2: “But the antecedent of this conditional is absolutely neces-sary; not only because it is eternal, but also because it is signified as past.” See also QDV 2.12 arg. 7 lines 66–74 and In Sent. 1.38.1.5 arg. 4.

50.  ST 1a.14.13 arg. 2: “Therefore, whatever is known by God is necessary, and thus God has no knowledge of contingent events [contingentium].” See also QDV 2.12 arg. 7 lines 64–5 and In Sent. 1.38.1.5 arg. 4.

51.  In the contemporary version of the Consequence Argument, which considers causal determinism, the circumstances C and the law-like statement L are understood as follows: the circumstances C are any past state of the world, and the laws L are the laws of nature. Using a similar principle of transfer of necessity (PTN), the argument goes like this:

(C1) It is necessary (it is not up to X at t1 to prevent) that L: if some circumstances C are realized, then the choice c occurs at t2. [Premise of determinism]

(C2) It is necessary (it is not up to X at t1 to prevent) that some circumstances C are realized. [Factual premise and necessity of the past]

(C3) ∴ It is necessary (it is not up to X at t1 to avoid) that c occur at t2. [(C1), (C2), and (PTN)]

The truth of C1 is that of causal determinism, and the truth of C2 derives from the fixity of the past.

52.  Instead of ‘to know’, we use the verb ‘to believe’, for two reasons. First, knowledge by itself implies that what is known is true (knowledge is a success verb), whereas belief by itself implies this only on the assumption that the believer is infallible. Hence speaking of belief rather than knowledge separates more clearly the law-like condition F1 (divine infallibility) from the factual premise F2 (what God actually believes). Second, knowledge is an extrinsic characteristic of the knower insofar as it relates the knower to the world, and thus it is dependent on the state of the world it refers to, and when this state of affairs is yet to come, it is dependent on the future; in contrast, a belief is an intrinsic state of the believer, which occurs at a specific time and unless it is true belief is independent of the state of the world it expresses. For this reason, knowledge possessed in the past and bearing on the future is not subject to the necessity of the past, whereas a past belief — even about the future — is subject to the necessity of the past.

53.  That Aquinas invokes the PTN in the opening arguments, which usually do not reflect his own position, in principle leaves open the possibility that he does not endorse it. But he does not reject the PTN in the replies to these arguments, and thus he tacitly accepts it.

54.  In Sent. 1.38.1.5 c. and ad 4, QDV 2.12 ad 7 lines 378–407, ST 1a.14.13 ad 2.

55.  See Boethius, Philosophiae consolatio 5 prose 4.24–5; Aquinas, In Sent. 1.38.1.5 c., p. 911 (where he refers explicitly to Boethius), QDV 2.12 ad 7 lines 381–8, ST 1a.14.13 ad 2. For a detailed and perceptive discussion of Aquinas’s expla-nation and his use of Boethius, see Marenbon 2005, 117–62.

56.  See ST 1a.14.13 ad 2; see also QDV 2.12 ad 7 lines 388–95.

57.  See ST 1a.14.13 ad 2: “When I say ‘If God has come to know something, it will be’, the consequent must be understood insofar as it is subject to God’s knowledge, that is, in its presentness. And thus it is necessary, just like the antecedent, because ‘all that is must necessarily be while it is’, as it is said in book 1 of Perihermeneias [19a23–4].” See also In Sent. 1.38.1.5 ad 4, p. 914, QDV 2.12 ad 7 lines 406–7 and ad 9: “Insofar as it is known by God it is present, and so it is determined to one alternative, even though while it is future it is open to either alternative.”

58.  For a critique of Aquinas’s idea of double modality, see Prior 1962, 128–9, repr. 57–8. For a critique of Aquinas’s reformulation of the fatalist argument from divine foreknowledge (F1–F3) according to the Boethian principle of the modes of knowledge, see Marenbon 2005, 145–7.

59.  For useful studies, see Nicolas 1960, Wippel 1984, 255–63, Paluch 2004, Goris 2005.

60.  Our interpretation of what follows has profited from Shanley 1998.

61.  See also QDV 5.5 ad 1: “That which is up to us, that is, which falls under our choice … is not in such a way determined to one outcome by divine provi-dence as it is in those things that lack the power of free choice.” That for Aquinas a human free choice is not to be traced entirely to God becomes particularly clear in his discussion of the divine causality in human sin. While the act itself, like all secondary causes, has God as its primary cause, its defectiveness has to be traced exclusively to liberum arbitrium of the sinner (QDM 3.2 c., especially lines 85–99, ST 1a2ae.79.2).

62.  For a more detailed discussion, see Frost 2014, 51–7.

63.  QDV 23.5 ad 3: “Although the non-being of the effect of the divine will is incompatible with the divine will, the power to have a deficient effect is com-patible with the divine will. In fact, these statements are not incompatible: ‘God wants this person to be saved’ and ‘This person can be damned’; but these statements are incompatible: ‘God wants this person to be saved’ and ‘The person is damned.’”

64.  ST 1a.19.8 arg. 3: “This conditional is true: ‘If God wills something, it will be’, and every true conditional is necessary.”

65.  Ibid.: “But the things created by God relate to the divine will as to something prior, from which they have their necessity.”

66.  See above, note 47.

67.  On this point, see also Frost 2014, 54, note 18.

68.  So it is argued, for example, by Loughran 1999.

69.  Particular goods, in which human acts consist, are not of this sort … for instance eating this or that food or ab-staining from it; yet they have in them that whereby they may move the appetite according to some good that is considered in them; and hence the will is not induced of

70.  The concepts of ‘definitive’ and ‘all-things-considered judgment’ are key to Aquinas’s account of practical deliberation and choice, but the terms are not his own. He does, however, use the expression “to be chosen … all pertinent conditions considered”; see In Sent. 2.5.1.1 c. We discuss the idea of the definitive judgment and its relation to choice further in sections 4.1 and 4.2 below.

71.  By ‘fully accounted for’ we intend that a given state of the intellect results in one precise state of the will, and conversely a given state of the will results from one precise state of the intellect.

72.  The terms ‘intellectualism’ and ‘voluntarism’ are modern classifications, and the meaning scholars attribute to them ranges widely. Hause 1997, 168 for example, proposes a flexible classification according to which a theory ap-proaches more or less the extremes of intellectualism and voluntarism: “An account of human action is voluntarist to the extent that the will, and not any other power, controls its own activities. Likewise, an account of human action is intellectualist to the extent that the will’s activities are under the intellect’s control”. Our definition is a rigid classification, consisting in a strict disjunc-tion. It allows us to draw more sharply the parallel between intellectualism and determinism, for determinism / indeterminism is a strict disjunction as well.

73.  See, e.g., QDV 22.6 ad 5, ST 1a2ae.8.1, QDM 6 c. lines 420–4.

74.  Henry of Ghent, Quodl. 1.16, ed. Raymond Macken, Opera omnia 5: 98–101, Quodl. 9.5, ed. Macken, Opera omnia 13: 121, lines 36–41.

75.  See Hause 1997. Williams 2012 also considers this a possible interpretation of Aquinas, but does not commit to whether or not it is actually the correct way of reading him see his 2012.

76.  Medieval authors often express themselves imprecisely, saying that, e.g., the “intellect understands”, “controls” etc. We follow their way of speaking for the sake of simplicity. Strictly speaking, it is the person who understands or con-trols something by means of the intellect.

77.  See, e.g., QDM 6 c. lines 343–54.

78.  See, e.g., SCG 3.10 n. 1950 and QDM 6 ad 15 lines 615–9.

79.  See, e.g., ST 1a2ae.17.6, ST 2a2ae.2.1, Sententia libri Ethicorum 3.13 lines 47–54.

80.  MacDonald 1998.

81.  Henry of Ghent, Duns Scotus, and William of Ockham were outspokenly type-1 voluntarists. An example of a straightforward type-2 voluntarist is Pe-ter Auriol. For references and discussion, see Hoffmann 2015, 70–2 and 80–7; and Hoffmann forthcoming, section 1.

82.  ST 1a2ae.77.1 c.: “It is in the nature of the movement of the will … to follow the judgment of reason”; ST 3a.18.4 ad 2: “For what we judge to be done after the inquiry of deliberation, we choose.” See also QDM 16.2 c. lines 261–9. It should be noted that to deny that the will can choose differently from the de-finitive practical judgment is not to deny the possibility of acting contrary to the dictate of conscience, for the dictate of conscience is distinct from the last practical judgment (or the judgment of liberum arbitrium, as Aquinas some-times calls it). In fact, when my conscience forbids me to indulge in inor-dinate pleasure, I can continue to deliberate about aspects that make such pleasure appear worthwhile; see QDV 17.1 ad 4.

83.  Gallagher 1994.

84.  Pasnau 2002, 224–33.

85.  In what follows, we will refer to non-human animals simply as animals.

86.  1 ad 2 and ST 1a2ae.76.1; for choice as simultaneous with the definitive practical judg-ment, see QDM 16.4 c. lines 279–82; for choice as following upon a practical judgment, see ST 1a2ae.13.3.

87.  See also ELP 1.14 lines 513–7, quoted on p. 21 above.

88.  See note 82 above.

89.  For choice as the conclusion of a practical syllogism, see ST 1a.

90.  We draw mainly on Anscombe 1957 and 1989. In these works, Anscombe uses explicitly Aristotelian material, but she also has a Thomistic background which surfaces from time to time. See also Michon 2010, 268–72.

91.  QDM 6 c. lines 288–96. See also ST 1a2ae.13.2 and SCG 2.48 n. 1245. Cf. Gal-lagher 1991, 578–82.

92.  See also ST 1a2ae.13.6: “In all particular goods, reason can consider the aspect of some good or the deficiency of some good, which has the aspect of evil, and accordingly it can apprehend any one of such [particular] goods as to be chosen or to be avoided.”

93.  Aquinas explains the syllogistic structure of the reasoning involved in incon-tinent action in ST 1a2ae.77.2 ad 4 and QDM 3.9 ad 7. For his more general treatment of incontinence, see ST 2a2ae.156. A fine comparison between Ar-istotle’s and Aquinas’s explanations is found in Kent 1989; for a detailed ac-count, see also Bradley 2008.

94.  It is important to note that in Aquinas’s account, contrary to Aristotle’s, the incontinent are making a deliberate choice when acting incontinently. See Kent 1989, 207–10.

95.  Cf. Nozick 1981, 294–300. That the will’s choice is not proportional to the evidence provided by reason is clear in Aquinas’s view of the will’s role in opining, i.e., in assenting to statements that can be true or false: the will can move the intellect to assent to a proposition one is not fully convinced of. See note 79 above.

96.  A similar conclusion, reached by different means, is proposed in Hoffmann 2007.

97.  Roderick Chisholm introduced the terminology of agent causation (as op-posed to event causation), relying occasionally on Aquinas; see Chisholm 2003, 26–37. Aquinas thinks of causes as things endowed with natural powers rather than as events subsumed under general laws, but there might be disputes about the aptness of Chisholm’s use of Aquinas. O’Connor 2000 defends a view of agent causation that would be largely in agreement with the view presented here see his 2000.