1.1 Theological compatibility problems

In philosophy and theology, moral responsibility motivates the debate around free will. The core issue is the compatibility of responsibility and freedom with other beliefs about nature, God, and the universe. It seems obvious that humans make choices between options and can make a difference with respect to their moral character. At least some of our actions are ‘up to us’. We can preliminarily define ‘free will’ as this ‘up-to-usness’, that is, the agent’s capacity to control her actions via decisions and choices that are based on reason. Moreover, free will is often associated with self-control: we exercise some degree of self-control over our emotions, desires, and intentions.

In this context, agential terminology is used: act/acting, choice/choosing, will/willing, and decision/deciding. The Greek terminology of Plato and Aristotle was quite varied: to hekousion (voluntariness), boulesis (wish), and prohairesis (choice). It is far from clear that these terms cover the same territory as the later Latin notions of voluntas (the will) and liberum arbitrium (free will or free choice, especially as it relates to the issue of salvation). The ‘up-to-usness’ condition seems central to our practices of moral judgment and moral responsibility. Only actions that the agent controls – those that are ‘up to the agent’ – are suitable for moral assessment and can thereby ground moral judgments about the praiseworthiness or blameworthiness of the agent. Without some degree of control, it is difficult to see how any agent could be blamed or praised for an action.

There are two kinds of challenges to this kind of control. First, there are challenges that emerge from some universal feature of our cosmos, mainly because of the control that God, gods, nature, or fate exercise over human lives. Here, the features that cast doubt on action control come from the outside, as it were. Second, there are mental or psychological challenges that have to do with the internal structure of the agent herself. Here, the threats to the ability of the agent to control herself and her actions come from the inside and take the form of irrational desires, emotions, or some other incontrollable psychological failure (like original sin), for instance.

Universal or external compatibility problems emerge when some general beliefs about nature, gods, or God seem to clash with the human ability to exercise control over their actions or self-control over their conflicting desires, intentions, and emotions. This clash casts doubt on whether humans in general can be blamed and praised for their actions and whether they are in control of their lives. For instance, if one believes that God, gods, or fate ultimately controls the destiny of the individual agent, the agent’s control over her actions and herself is called into question. When accused of an evil deed, someone might say, ‘it is not my fault, it wasn’t up to me: God/fate/destiny made me do it’. Similar problems arise on some accounts of nature as well. If we believe that nature ultimately consists of non-rational physical particles that collide randomly in the void and we are a part of that nature, there seems to be no sense in which we can exercise control over our actions. Such ‘metaphysical issues’ are very well represented in Christian theology. They often arise from God’s purported foreknowledge, omniscience, election, and providential control over creation.

Another set of issues has more to do with human psychology and philosophy of mind. If free will means that I control my actions, and myself, it immediately raises further questions: what am I and which part of me should be in control – my reason, my will, or my emotions? This is why free will debates are closely related to philosophy of mind. As we will soon see, the traditional ideal was a psyche guided by reason: the agent controls her actions when the actions have their roots in the agent’s reason. In Christian theology, such an account faces various compatibility problems. The problem of free will, sin, and grace is a good example. Many accounts of original sin claim that there are certain psychological defects that render humans incapable of controlling some of their morally and spiritually significant actions. This raises immediate questions about blameworthiness: how can humans be blamed for their sinful actions if they do not exercise control over those actions?

Both external and internal compatibility problems have their roots in the scriptures. Neither the Hebrew nor the Christian scriptures offer a systematic treatment of free will and moral responsibility. It can be argued, however, that a basic notion of human moral responsibility and the accompanying human freedom is clearly (but not uncontroversially) assumed. This assumption is reflected in how God chooses to relate to humans. This is especially apparent in the Hebrew scriptures, where the core narrative is that of God’s election of the people of Israel and Israel’s response to this election in various circumstances. God’s covenant with Israel calls out for a response: Israel expresses willingness to remain in the covenant by following God’s law. In this relationship, some measure of autonomy on the part of Israel is clearly assumed (e.g. Deut 30:15–20). While God sometimes unilaterally elects certain individuals and groups for special treatment, God nevertheless treats humans as responsible agents. He promises to reward those who do his will and punish those who do not. He issues commandments, expresses his will to people, and expects them to comply. God also calls people, persuades them, and makes promises to them. If human beings were assumed to be mere automata, such actions on God’s part would make very little sense.

Christian scriptures of the New Testament give more reasons to assume that humans are free, responsible agents. This is especially pertinent when considering the increased stakes in one’s relationship to God. Rather than just secular success and failure, one’s earthly life determines how one spends one’s eternal afterlife. In other words, one’s actions in this life – especially accepting God’s gracious offer of mercy or rejecting it – have eternal implications (e.g. Matt 23:37). Given this, it is very difficult for Christian theologians to completely reject human moral responsibility. Humans are agents, who are accountable for their actions in the eyes of God and their fellow humans. Not only are humans apt subjects of ordinary praise and blame for their moral actions, but they are also accountable in the ultimate sense to God. This sense of responsibility in the eyes of God forms the background for most Christian discussions on free will.

However, human moral responsibility is called into question because of other tendencies apparent in Christian and Jewish scriptures. Some facts about God and his actions seem to undermine human free will. In the Hebrew scriptures, we can observe the theme of God’s unilateral and sovereign action towards his creation and his people. One central theme is the election of Israel in relation to God’s providence. First, God chooses a special people for his purposes, and this choosing is not based on any kind of merit on the part of Israel. Second, the fate of the chosen people depends on God’s active providence and goals. Lives of individuals and groups have roles to play in God’s plans. Third, there are cases where God seems to act in such a way as to unilaterally elect some individuals for suffering and evil. Such actions seem somewhat surprising, since God is supposed to be perfect love, but both Christian and Jewish scriptures contain examples of this. The most famous example is the hardening of the heart of the Pharaoh in (Exod 10:12), where God seems to intentionally cause the Pharaoh to have an evil will. The theme also emerges in Paul as an explanation for why Jews refused to believe in Christ (Rom 9).

In the New Testament, the themes of providence and election are similarly important. A central assumption is that God’s providential plan underlies the events in history and individual lives. Again, there are events in salvation history that are unilaterally God’s work: Jesus calls Israel to repentance and renewal of their relationship with God in the coming kingdom; the coming of the kingdom depends on God only, not on humans. Moreover, Paul emphasizes God’s giving of grace and liberation from sin through the work of Christ. Paul highlights God’s agency in providing salvation. Faith in Christ is a gift from God for which the faithful receive no merit or credit (Rom 9; Eph 1:11).

1.2 Sourcehood, intellectualism, freedom of choice, voluntarism

Before we can examine how various compatibility problems could be solved, we need to look more carefully at the notions of the will and free will. Contemporary philosophers have usefully drawn a distinction between two types of control (Timpe 2017). ‘Leeway’ accounts locate the ‘up-to-usness’ condition in the human ability to choose between alternatives. An agent controls an action when the agent can act or refrain from acting. We might call this kind of control the freedom of choice, or the freedom of indifference. Another type of control is that of ‘origination’ or ‘sourcehood’. Here, control is defined in terms of the free action having its roots in the agent’s desires, intentions, and volitions. An agent controls her actions when those actions have their origins in the agent’s will and reason, rather than originating outside the agent. In the contemporary context, this kind of control is often called sourcehood, but traditionally it was associated with the freedom of spontaneity: namely, the ability to do what one wants without external constraint.

Both sourcehood and the freedom of choice emerge in Greek philosophy. Later on, Christian theologians inherited these notions and made extensive use of them. Actions are goal-directed movements. They can be understood in the light of some goal, or telos. Many animals have the capacity for basic agency in this sense: to act in goal-directed ways. They are driven by various desires and yearn to fulfil those desires. Both animals and humans have the basic capacity to will: to have a will is to act in pursuit of some goal, which is governed by some ‘internal principle’ rather than some ‘external principle’, like coercion or other outside force.

Aristotle and others acknowledged that many non-human biological organisms are capable of acting from internal rather than external impetus in pursuit of some goal. However, this is not free will. The extra ingredient for responsibility-allowing action control is the supposedly uniquely human capacity of reason and intellect (Nielsen 2011). Aristotle sees human actions as products of practical reasoning and various other, non-rational desires and emotions (Nichomachean Ethics, Book III). Reason has some measure of control over the extent to which non-rational desires and emotions trigger actions, but there seems to be no one, single psychological system responsible for bringing actions about. Reason allows for agent to identify and evaluate various possible goals, which, for Aristotle, is the process of practical deliberation and reasoning. Whereas a non-rational agent simply acts on its desires, a rational agent can deliberate between different desires and ways of reaching goals. So, here we have a core notion of rational action control: the agent is in control of those actions that are guided by the agent’s rational evaluations and practical reasoning.

Plato, Aristotle, and other ancient philosophers had a strong interest in the psychological features of action control and self-control, because they were central in their account of the ‘good life’. Both Aristotle and Plato took the life of virtue and wisdom as the best possible life for a human being. Achieving such a life requires a significant degree of self-control. For Plato (The Republic, Book IV), freedom is an achievement that comes through a life of virtue and reason. Freedom is a state where the agent has little or no inner conflicts between various desires, which are outputs of the intellect and emotions. The ideal life is a life where one has developed virtues and learned to control one’s impulses and desires according to the outputs of reason. So, the higher part of the soul (reason) subdues or directs the lower parts (desires).

Aristotle emphasized the role of human choices in developing virtues (Nichomachean Ethics, Book III). Non-rational agents cannot make moral choices that in time create habits and shape the agent’s moral character. Rational beings such as humans, however, are capable of living a moral life and striving towards happiness, which is the telos of all rational beings. Through self-control the agent can resist conflicting desires and develop virtues, like justice, courage, and wisdom. A virtue is a consistent set of dispositions to act and feel in a way that is conducive to human happiness. However, if an agent does not use her powers of deliberation and choice consistently, she will develop various vices that hinder her happiness and lead to undesired consequences. In this way, reason-guided action will also shape the individual moral agent herself. A wise and happy person is psychologically well integrated – an enkratic, strong-willed person as opposed to an akratic, weak-willed person who often succumbs to desires that are against her best judgment. An enkratic person does not suffer from the weakness of the will, but is the master of her own soul and follows her own best, reasonable judgment.

Ancient and later Christian accounts of free will were nested in a larger network of ideas about human nature and the human good. Although the idea is common to many ancient philosophers, it was Aristotle who had the lasting influence. The idea is that human beings share a nature that gives all humans a set of goals, sometimes identified with the good. Human reason is a rational desire: it is naturally directed towards the good even when humans fail to achieve the good or try to reach it in misguided ways. For human beings, this good is happiness (in Greek, eudaimonia). Freedom consists in being able to rationally identify what leads to happiness and act in such a way as to achieve it. In other words, proper action control is the condition in which the agent (or a group) can pursue the in-built goal of happiness unimpeded by outside or inside forces.

Aristotelian ideas are fully integrated into Christian theology by Thomas Aquinas. For Aquinas, human willing is a rational desire and directed towards what is perceived as the good. In this sense, the will cannot act at random, but it can only act when reason has presented something to it as good and attractive (Summa Theologiae I q. 80 a. 2). What humans take as good can vary from situation to situation, of course. Nevertheless, the ultimate happiness, which is the telos of human nature, is God. God has created human beings such that they find ultimate fulfilment only in God. However, while the human will cannot choose its ultimate goal, it can choose between different means of reaching the goal. No one way of reaching happiness appears to humans so appealing that it would automatically command assent. So, it is up to the judgment of human reason to discern between them (Pope 2002).

Such accounts of the generation of action are rather intellectualist in nature. Free action consists of the agent acting rationally and the rule of reason in the agent’s soul. Rational actions also have a goal that is not ultimately determined by the person. The will is automatically directed towards the good or God. Moreover, intellectualism is evident in the fact that ‘the will’ as a faculty of the soul has no independent role in action production. The properly-working will always follows the outputs of reason. The intellectualist thinks that reason provides evaluations of the extent to which some intention or desire is in accordance with the good. The will subsequently moves the person to action according to the final output of reason. While desires and emotion contribute to action, this happens in conjunction with reason. Some have even suggested that the notion of the will as an independent faculty is an invention of Christian theology and have dubbed Augustine (354–430) as its father (Dihle 1982). Against this, others locate the emergence of the will as a central feature of Stoic moral philosophy and psychology (Frede 2011).

In the second and the third century, the will as an independent faculty took on a more important role in both philosophy and Christian theology. During this era we begin to see the emergence of leeway accounts of control and voluntarist accounts of the will. The voluntarist maintains that the will can also initiate spontaneous actions that are somewhat independent from the outcomes of reason’s judgments. Free actions are not necessitated by prior evaluations and reasons, but the will retains the ability to make spontaneous, non-determined choices between good and evil alternatives. Such choices are the grounds for judgments about moral responsibility.

It is sometimes suggested that Christian thinkers introduced this idea of leeway freedom, the freedom of choice (Dihle 1982). It is true that some Christians assigned a larger role for the underdetermined choice of the will than some ancient philosophical schools. Origen (184–253), for instance, maintained that since the Church teaches the coming of final judgment, the existence of human moral responsibility is implied in this teaching (On First Principles). Because humans can only be praised and blamed for that which is under their control, this, in turn, implies that humans have free will – the choice between morally relevant actions is under their control.

However, Greek and Roman philosophers had extensively discussed leeway freedom long before Origen and other Church Fathers. They were particularly cognizant of one very basic compatibility problem: the problem of free will and God’s providence. The uncompromising determinism of the Stoic philosophers drove this debate (Bobzien 1998). According to Stoicism, the fate of all humans is governed by the same rational, natural laws as everything else in nature. As a consequence, humans have no control over the circumstances of their lives. They can only control their psychological attitude towards it. Despite having no control over their fate, human individuals can still be morally responsible for their actions because they can be sources, originators of their own actions. According to Chrysippus (279–206 BCE), for instance, humans can be responsible for their actions, because at least some human actions are not determined by external circumstances but rather by the rational process internal to the agent. Here the notion of assent played a central role. An action is triggered by an impression of something desirable detected by the senses. However, the agent’s intellect can assent to, or refrain from assenting to, the impression. In the Stoic view, this assent is the locus of control and grounds responsibility. Later on, these accounts found their way into Christian theology, especially through Augustine.

However, many resisted the compatibilism of the Stoics already in Hellenistic times. Alexander of Aphrodisias (c. 150–215) was a commentator of Aristotle who, in his close readings of Aristotle’s texts, developed the philosopher’s notions further. Aristotle already mentions that a person can be held responsible for only those actions that the person could have refrained from performing. However, he does not offer a clear analysis of this. Alexander, however, is probably the first to explicitly insist that such a power to ‘act otherwise’ – as it is now called – indeed requires that the agent have two metaphysically possible alternative courses of action available to her at the moment of action. This is what contemporary philosophers call leeway freedom, or the principle of alternative possibilities (van Inwagen 1983). Given this notion, a free action is incompatible with prior necessity, because it would rule out the possibility of leeway (an alternative choice) from the agent. In Alexander’s thought, we see the emergence of a libertarian conception of free will. Libertarians argue that humans have free will, and that free will is incompatible with there being sufficient prior causes to the agent’s actions. As such, they hold that determinism must be false and indeterminism true.

1.3 Divine determinism and theological compatibilism

The term ‘determinism’ emerges in the seventeenth century. Prior to that time, philosophers talked about various kinds of necessity, including necessities introduced by logic, fate, laws of nature and God or gods. Above, our discussion identified a number of philosophical and theological compatibility problems that cast doubt on the possibility of human action control. These challenges can come in the form of universal, metaphysical necessities or psychological necessities – both of which seem to constrain human actions. One appealing way to solve both external and internal compatibility problems is to argue that human free will and moral responsibility are compatible with determinism.

Determinism can come in many different kinds. First, there is the idea of fatalism, according to which a person can contribute nothing to her future. The future will remain the same regardless of what the person does. For the purposes of this article, theological fatalism is the most central issue. As we will see later, theological fatalism is what produces the problem of divine foreknowledge and free will. According to theological fatalism, there are some facts about God’s knowledge that logically fix future events. This issue is independent from causal determination. While fatalism is often confused with determinism, it is important to keep these two notions apart.

Second, physical or causal determinism is a general metaphysical thesis that the totality of the past up to the given moment, combined with the complete set of laws of nature, only allows for one unique future. Notice that determinism does not entail fatalism: it is perfectly possible to hold that determinism is true and that what the person does now has an effect of how her future will turn out. Third, psychological determinism is the thesis that a particular act or a choice of an agent is determined by the prior mental fact that pertains to the agent. So, if the agent’s reasons, desires, beliefs, and other relevant mental features are fixed, they always produce the same action.

Finally, Christian theologians have been particularly worried about a specific kind of determinism known as divine determinism. The exact mechanism of divine determinism can be understood in different ways (causal or logical, for instance), but the core idea is that all created events and facts are ultimately made necessary by God’s will, power or knowledge, including the free acts of creatures (Furlong 2019). This necessity can be a product of God’s complete and infallible foreknowledge or God’s comprehensive providential control over creation. Divine determinism has been rather attractive in the history of Christian theology. Its proponents include Augustine, Thomas Aquinas (1225–1274), Martin Luther (1483–1546), John Calvin (1509–1564), and Jonathan Edwards (1703–1752), just to mention a few.

Divine determinism might be an attractive position for Christian theologians for many reasons. Many divine determinists invoke the scriptures as evidence for its truth. Some passages clearly imply that God is in full providential control over his creation (Ps 103:19; Dan 4:35; Matt 10:29). Second, the scriptures also seem to imply that God has perfect knowledge of the future, including human free actions (Isa 42:9; Acts 3:18). Because God is necessarily infallible, it follows that human future (and present and past) actions are, in fact, not contingent but necessary. Aquinas and his followers argue that the truth of divine determinism is entailed by the very nature of God as the omnipotent and sovereign creator of the universe. According to the classical theistic view, God is the source of every contingent being and those contingent beings depend on God for their existence. It is God who is the only being that exists ‘by itself’ (Latin: a se). This aseity of God is often thought to lead to a radically unilateral causal relationship between the created and the creator. Like Aquinas maintains, God is the primary cause of everything in the created realm. Not only does God constantly give being to the created world, but God also participates in all created causal relations. God is said to concur with creaturely causes (Latin: concursus Dei). In contemporary debates, this divine property is often called God’s omnicausality. Combining the view of God’s aseity and omnicausality with his immutable will and his atemporality, divine determinism becomes an almost unavoidable consequence.

Due to the fact that divine determinists have held onto human moral responsibility, they have been forced to provide accounts of responsibility in which determinism and human control over actions are compatible. Incompatibility problems can arise in two ways that correspond to the two kinds of control outlined above. On sourcehood or origination accounts, determinism could be argued to rule out the possibility of an agent being the source of her own actions. The truth of determinism rules out the possibility of ultimate self-control: if determinism is true, no human agent can be the source of his or her own psychology and moral character. Within leeway accounts, determinism can be argued to rule out the possibility of choosing between alternative courses of action. Freedom of choice entails that a free action is such that an agent can either act or refrain from acting, while all the background conditions stay the same (the same psychological history, etc.). If determinism is true, it seems obvious that no human agent can make such choice: determinism entails that the past up to the given moment entails only one possible future. If one wants to defend the compatibility of free will and determinism, one must provide an account of control that avoids both compatibility problems.

Let us briefly examine two stereotypically compatibilist accounts emerging from Christian theology. The first comes from the influential church father Augustine. Due to the great significance of Augustine’s views of free will in theology up until today, there have been many interpretations of the details of Augustine’s account. There is disagreement over whether Augustine was a compatibilist about free will and responsibility. Eleonore Stump interprets him as a sourcehood theorist and a libertarian incompatibilist (Stump 2001), whereas many others maintain that at least Augustine’s mature views were consistent with compatibilism (Bonner 2007). The disagreements might be partly explained by the fact that Augustine’s views on free will seem to change over his lifetime and he tended to emphasize different aspects in changing contexts, especially in debates with different adversaries. Despite various interpretations, it is safe to interpret Augustine’s view leaning towards intellectualism regarding the nature of the will and compatibilism regarding free will and divine determinism (Couenhoven 2017a). It also seems that Augustine was committed to some form of compatibilism as far as divine foreknowledge was concerned. However, Jesse Couenhoven (Couenhoven 2017a: 252) notes that Augustine explicitly denied complete divine determinism (e.g. in Predestination of the Saints), since, according to him, God cannot be the ultimate cause of sin and evil. It was very important for Augustine that God does not cause or create evil. Thus, God cannot determine evil. Of course, this leaves Augustine in a precarious position: the ultimate source of evil becomes a complete mystery. Some evils in the created world are due to human free will, not God. However, there still remains the question of how evil emerged in the first place. Nevertheless, for Augustine, divine necessity actually guarantees (rather than threatens) human freedom.

Augustine’s account of the will follows compatibilist and intellectualist lines. First, Augustine seems to accept something like rational self-control as the core of free will and to reject the notion of ultimate responsibility. This picture is complicated by Augustine’s view of ‘the fall’. Before the fall, Adam and Eve had the ability to choose between good and evil. However, after the fall, all subsequent humans lack this capacity and are destined to act in evil ways. While humans can shape their characters to some extent, their moral nature is beyond their control. Both saints and sinners do what they love, but have little control over the objects of their love. Second, true freedom consists in having the right objects of love, not in having access to alternative possibilities by themselves. So, when God heals the human soul from sin, God might actually narrow down the range of possible human actions, but this is nevertheless agency enhancing: it becomes psychologically impossible for the saint to love and do what is evil.

Finally, Augustine’s view on moral responsibility has some internal tensions. On the one hand, it seems that being under original sin is clearly out of the control of the individual person. In their sinful state, humans have no control over their wills. Given Augustine’s insistence that one can only be blamed about facts that are in one’s power to choose or not to choose, it seems that sinful humans are not culpable for their sin. On the other hand, Augustine accepts that sinners are in fact morally accountable for their actions, because the scriptures say so. Sinners have no libertarian free will and no control over their moral nature. Nevertheless, sinners are responsible, because those actions emerge from their wills, not out of external pressure or compulsion. Humans sin ‘according to their will’, not ‘against their will’. This is enough for responsibility.

Augustine’s views had an immense impact on later theology, in part because the church later followed his views on grace and original sin while rejecting his competitors, ‘Pelagians’, as heretical (Council of Carthage 418). Not only were Augustine’s views followed and developed by medieval theologians, like Thomas Aquinas, but they were crucial for many reformers of the sixteenth century.

Some scholars (Couenhoven 2013: 130–134) have suggested that what emerges here in Augustine’s thought is what contemporary philosophers call an attributionist theory of moral responsibility. Here the locus of responsibility is not in controlling the relevant action, but a form of ownership over action, where the responsible action can be properly attributed to the agent. Even while the sinful agent lacks control over sources of her actions (will, character), she nevertheless acts out of deeply held attitudes and judgments. Her actions reflect her beliefs, desires, moral judgments, and character, thereby allowing an action to be attributed to the agent. This basic ownership, rather than control over the action, is what grounds responsibility.

Another example of an explicitly compatibilist account of free will comes from the American Puritan theologian Jonathan Edwards (1703–1758). As Aristotelian natural philosophy transformed into modern physical science from the seventeenth century onwards, theologians had to deal with the mechanistic and physicalistic views of nature that resulted from the shift from Aristotelian and Ptolemaic worldview to the mechanistic and Copernican one. One response to this development was to incorporate the determinist and mechanist view of nature into a thoroughgoing divine determinism and compatibilist account of free will. Edwards formulated a modern version of divine determinism and free will in his treatise Freedom of the Will. His view is rather radical and some scholars suggest that it is occasionalist (Crisp 2012). According to occasionalism, there really are no secondary causes in the created world at all. When water kettle boils, for instance, we vulgarly attribute this as the effect of physical laws and the essence of water. However, such explanations do not track the actual causes, or cause, which is God. The fact that water boils when it is heated is a result of God acting in predictable ways, not due to some natural powers or inclinations of the water. Such a view also seems to lead Edwards towards idealism: he seems to be rejecting the existence of matter altogether.

Edwards is known for his adamant critique of libertarian and voluntarist accounts of free will. For Edwards, such views are ultimately incoherent, because they assume that the will can be the cause of its own action. Free actions, according to the voluntarist, are spontaneous and independent from prior causes. This is incoherent, Edwards maintains, since nothing can be independent from prior causes: everything has a cause for its existence. There is no self-moving power of the will, as libertarians conceive it. If the will is truly supposed to be free in this sense, it should be preceded by a spontaneous choice that has generated it. That choice should also come from a will shaped by spontaneous choice and so on, leading to infinite regress. Furthermore, Edwards asks, how can free actions be, on the one hand, caused by the will, but on the other hand spontaneous? Again, the libertarian account seems incoherent. So, in order for any kind of human freedom to be possible, determinism must be true rather than false.

Edwards foreshadowed a debate that emerged in philosophy. In particular, the Scottish philosopher David Hume (1711–1776) argued that the libertarian account of free will is incoherent, or at least implausible, because it entails that free actions are products of chance. Indeterminism only undermines control, rather than enhancing it. The problem can be stated as follows: it seems that choices between alternative possibilities seem far from being free – rather, they are ultimately random. We make sense of the action of an agent by invoking the agent’s reasons, desires, and deliberative process. Free actions are controlled or guided by rational considerations. Only actions that are produced in such a way express our will. However, all this requires that there is a close causal connection between an agent’s will and her actions. The more spontaneous the outputs of the will are, the less understandable and more inexplicable our actions become. If there is indeterminism in the process that leads to action, that indeterminism can only add more randomness and thereby loosen the agent’s control over the action and undermine moral responsibility.

Edwards’ view of free will is clearly intellectualist and compatibilist in nature. For Edwards, free actions are caused by the prior outputs of the agent’s reasons and other faculties. The agent always does what she perceives as the best course of action. An agent can be held morally responsible for her actions when those actions are caused by her choices. However, those choices have their roots in the agent’s will and reason, which are not under the control of the agent.

The combination of divine determinism and theological compatibilism has to face significant challenges. These challenges will be examined in detail later in this article, but they can be briefly summarized here. First, Augustine already struggled with the problem of evil: it is difficult to blame humans for evil and excuse God from blame, if human free will is compatible with divine determinism. This is because the compatibility makes it possible for God to determine what all humans want without overriding their freedom or undermining their responsibility. God could create a world in which every human only willed what is good. Second, one might raise the question whether compatibilism can save human moral responsibility after all. Perhaps the notions of ownership and attribution are too slim to ground responsibility in the human case. Moreover, responsibility is not a zero-sum game: two agents can be simultaneously responsible for an action. So, it is not enough for a theological account of moral responsibility to defend human moral responsibility. It must also provide a way to exculpate God from any evil and wrongdoing. This latter task is a challenge to theological compatibilists, because the view raises questions about divine manipulation. First, God determines that humans act in evil ways or refuse to believe in him, and then he, nevertheless, judges them for their evil and unbelief. So, the ultimate control over a person’s faith and moral character rests with God. This seems to lead to a fully predestinarian account of grace and faith: as Augustine explicitly argues, the eternal destiny of an individual is purely up to God. However, it seems to follow from this claim that God, not the individual, is to blame for the individual’s unbelief. How can God blame the sinful human for his unbelief and sinfulness, if they are ultimately in God’s own control?

1.4 Divine openness and theological incompatibilism

Another way to solve the theological compatibility problems is to reject divine determinism and accept a libertarian account of human freedom. On this account, human free will and moral responsibility are considered incompatible with divine determinism. Such a view would provide a robust answer to the problem of evil: at least some evil in the world is due to humans and God cannot control it. Similarly, it would explain why there is unbelief and sin in the world: humans choose to abuse their free will in such a way, and it is not up to God. The obvious challenge here is to make sense of God’s providence, foreknowledge, grace and sin non-deterministically. These issues will be considered in the following section. First, let us briefly consider some examples of what a libertarian and indeterminist theology look like.

Incompatibilist accounts of the will emerged very early in Christian theology. Origen and his followers, for example, have already been mentioned above. Early medieval theology saw many versions of explicitly voluntarist and libertarian accounts of free will. Not all medieval theologians followed Augustine in his intellectualist account of human action as closely as Aquinas. Anselm of Canterbury (1033–1109), for instance, offers an early libertarian account of free will. According to Anselm, God gives free will as a gift to humans, because without free will humans can be neither just nor happy. Anselm is clearly a voluntarist and an incompatibilist: the will has the power to determine itself and to act spontaneously. Not even God can make the will of a human agent to will something necessarily. This leads Anselm to reject the notion that salvation is completely up to God. He maintains that although God’s grace is necessary for a life pleasing to God, maintaining and preserving just moral and spiritual life requires the application of free choice (Rogers 2008).

There is considerable debate whether Thomas Aquinas was a compatibilist or an incompatibilist. According to some scholars (e.g. Zoller 2004), Aquinas was a compatibilist and his account of the psychology of will and reason was deterministic: given the totality of prior mental states, an agent is determined to will whatever she wills. The will makes no spontaneous choices, because practical reasoning is not under its control. However, a proper liberum arbitrium, Aquinas maintains, necessarily requires some independence from external secondary causes and influences. In other words, a free will is not causally determined by factors outside its own operations. Freedom of the will is not about having alternative choices available, but rather being free from prior sufficient causes that would compel the will. Against this, some scholars (e.g. Hoffman and Michon 2017) have insisted that Aquinas indeed argues for an incompatibilism and the alternative possibilities condition.

So, even if Aquinas offers a sourcehood-based account, it is unclear whether it is compatible with thoroughgoing causal determinism. It is clear that some sourcehood accounts of free will are indeed compatible with determinism (Frankfurt 1971) – but Aquinas’ account might not be among them. Some contemporary philosophical theologians have been inspired by Aquinas’ approach (Stump 2003) and argue that it represents what they call ‘source incompatibilism’. They argue that the alternative possibilities condition (leeway) is not necessary for free will. In this sense, free will is compatible with determinism. However, determinism might rule out free will in another way, by undermining the ability of the will to operate independently of prior causes external to it (sourcehood). So, a free will can be determined to act by prior mental conditions internal to the agent, stemming from the character, virtues, and reasons of the agent. However, the agent cannot be said to have freely acquired such features without some indeterminacy in the world.

Despite the fact that Aquinas is working within the legacy of Augustinianism, his views of human sin might differ from Augustine. While Aquinas affirms the sinfulness of fallen humans and the absolute requirement for God’s supernatural grace (Summa Theologiae I-II q. 85 a. 3), he also affirms that sinfulness does not necessitate the human will to making only evil choices (Summa Theologiae I-II q. 109 a. 2). Furthermore, if the libertarian commentators are correct about Aquinas’ source incompatibilism, Aquinas will ultimately reject all deterministic accounts of grace. According to some interpreters (Stump 2003), Aquinas maintains that the process of salvation requires something from the human side that God cannot control: namely, that the human agent ceases resisting grace. God is the only cause of grace and salvation; but it is up to the human agent to bring her will into a quiescent state, where it is neutral with respect to God. Only then can God’s grace work on the agent. So, the human agent is never the cause of her salvation, but she, nevertheless, exerts some control over it.

From the fourteenth century onwards, Franciscan theologians such as John Duns Scotus (1266–1308) and later William Ockham (1387–1447) develop powerful voluntarist and libertarian accounts of free will. For Duns Scotus (Questions on the Metaphysics, IX), the will is a faculty that is in principle independent from the outputs of reason. Indeed, genuine freedom of the will consists of the power of the will to determine itself and even act against reason. In this sense, Scotus argues, human choices between alternative possibilities cannot be determined by prior causes; nor can they be made necessary by any intra-mental states, like the outputs of reason. So, Scotus rejects both causal determinism and psychological determinism. Instead, the will has an inexplicable power to choose: choices are almost like brute facts that have no ultimate explanation apart from the will (Cross 1999: 83–89).

Libertarian accounts also emerged in the aftermath of the Reformation in the seventeenth century. One example of this is the Calvinist/Arminian debate over grace and free will (Mueller 2017). The French-born Swiss reformed John Calvin (1505–1564) adopted a radical Augustinian view: uncompromising divine determinism accompanied by human moral responsibility. Calvin also developed an explicit account of double predestination and limited atonement, where the atonement only pertains to those individuals that God has predetermined to save (Institutes of the Christian Religion). Following Jacob Arminius (1560-1609), a Dutch reformer, Arminians reacted against Calvin and defended a libertarian – and thus an incompatibilist – notion of free will. While God’s providence is affirmed, many Arminians maintain that God cannot make human choices necessary and free simultaneously. Like other theological libertarians, Arminians also hold that, although salvation is impossible without God’s grace, there is always the possibility that a human person rejects the offer of faith. This possibility of rejection is such that God cannot decree a person staying in faith, as it is ultimately a choice made by the human will. As a consequence, God can only free the human will from the bondage of sin; he cannot remove the possibility that humans might, nevertheless, resist God’s offer of grace. Through the teachings of John Wesley (1703–1791) and others, these ideas had a significant influence on many Protestant movements, such as Methodists and Baptists, especially in the United States.

2.1 Foreknowledge, providence, and free will

Ancient philosophers already knew the problem of divine foreknowledge and human freedom. Christian theologians took up this problem in their own context. If God has perfect knowledge of the future, does this knowledge not make the future inevitable and human future actions unfree (Zagzebski 1991)? Even if God does not know the future in the sense of providentially causing it, there is still the worry of fatalism. If God knows all future truths, the things that make those truths true (human free actions, for instance) are logically necessary. A prominent attempt to solve this problem has been made by those who insist that logical necessity, unlike causal necessity, does not hinder free will. Augustine already argued that God’s foreknowledge does not work in a causal way. Most famously, Boethius (477–524) argued that God’s foreknowledge is – like God – atemporal. Aquinas (Summa Theologiae I q. 14 a. 13) develops this idea: he argues that, because God is atemporal (has no temporal attributes), God knows all contingent truths – including truths about human free actions – by having access to all past, present and future contingent truths. God is outside time, so all created contingent events are constantly present in God’s atemporal moment. God is like someone who stands on a cliff and watches over a large plain. He can see all the paths people in the plain can take. This atemporal access, in and of itself, does not amount to causal necessity, or so the argument goes (Goris 2005). Finally, accepting divine determinism and theological compatibilism also provides a way out from the problem of divine foreknowledge and free will. If human moral responsibility and free will are compatible with logical necessity, then foreknowledge does nothing to undermine them. While divine determinism and theological compatibilism has its contemporary defenders (e.g. Helm 1994), it would be fair to say that contemporary philosophers and theologians have, on the whole, preferred other models of providence.

Because of the intense focus of the Reformers on divine determinism and the emphasis of the emerging Renaissance on individual agency, many Roman Catholic thinkers developed libertarian and voluntarist accounts of free will as a response. Jesuit Luis de Molina (1535–1600) and his followers developed an ingenious way of combining comprehensive divine providence and foreknowledge with human libertarian free will. They attempted to offer an account of providence and foreknowledge that would meet the requirements of a divine determinist, but without implicating God in human evils and maintaining libertarian free will.

According to Molina (On Divine Foreknowledge), such an account is possible if we posit the existence of a very special form of knowledge he calls ‘middle knowledge’. Middle knowledge differs from God’s natural knowledge, which consists of God’s knowledge of necessary truths (like those of mathematics), and free knowledge, which is based on God’s creating the world and all the contingent facts therein. Middle knowledge pertains to truths about various possibilities: that is, what would happen in different circumstances. For discussions of free will, the most important truths of this kind pertain to what libertarian free agents would freely choose in different conditions. Knowledge of this kind is about the truth of counterfactuals of freedom. Such counterfactuals are of the form: ‘if a person P is in conditions C, that person will freely choose to do A’. By having access to the truths of the counterfactuals of freedom before creation, God can choose to create that possible world where the created libertarian agents in that world act according to God’s will. God can guarantee a complete and comprehensive providence and foreknowledge without actually causing human actions or making them necessary as the divine determinist is forced to do. If coherent, Molinism looks like an effective and fruitful alternative to divine determinism as an account of freedom, providence, and foreknowledge. The theological usefulness of Molinism is highlighted by the fact that during the Jansenist disputes in the eighteenth century, Molinism was also applied to the problem of divine grace and free will.

Molinism has been extensively developed and debated in the contemporary context. Some see it as a means to solve a host of vexing theological problems, including the problems with hell, providence, and salvation (Flint 1998). A vast literature has been generated defending and critiquing Molinism (Perzyk 2012). In brief, questions have been raised about the notion of middle knowledge itself (what makes it true, for instance), as well as whether Molinism actually provides any benefits in solving, say, the problem of evil, compared to divine determinism.

There are some views according to which God’s foreknowledge and providence can be conceived without a commitment to divine determinism. These views, which have historically been minority positions, were further developed in the twentieth century. They focus on the ‘openness of God’s plans’: God’s providence might consist of directing the overall course of history rather than determining all the details of it. To put it in terms of a metaphor, God would guide the story to the end he wanted but not determine each and every twist along the way.

Such a view would have several benefits over divine determinism. First, it would make the problem of evil easier to deal with. The divine determinist faces the rather difficult task of explaining why every single instance of evil is necessary, and how it is possible that God decrees evil but does not cause it or become culpable for it. If divine determinism were abandoned, it could be argued that at least some amount of evil (usually moral evil) could be attributed to human free will and chance. Second, a world where divine determinism were false would be much more hospitable to free will as libertarians construe it. In such a world, created agents could choose between alternative options and develop their moral characters as they wished.

Such a view emerged in the late twentieth century, sometimes dubbed ‘open theism’. According to open theism, humans have libertarian free will, and it follows from this that God’s foreknowledge and providence cannot be as complete as Molinists and divine determinists have thought (Pinnock et al. 1994; Hasker 2004). In open theism, God provides a providential plan, or perhaps a great set of contingency plans, to prepare himself for the randomness, chance, and free choices of persons that come into existence in the created world. The future and the actual path of the created world is not completely known to God before creation and the created world could end up developing in ways that God has neither foreseen nor intended.

As critics of open theism often contend, it seems that if open theism is true, God takes risks. Because of libertarian free will and the contingency of the created world, things could turn out in ways that God does not want. Defenders of open theism do not see this as an insurmountable problem. First, they contend that even if God lacks complete foreknowledge, he has access to various probabilities. Because of God’s knowledge of the past, he can form extremely accurate predictions and update his probabilities. Second, open theists often maintain that certain contingency and risk-taking characterize God as he is depicted in the scriptures. God’s interaction with the world is not strictly unilateral, but instead dynamic and reactive. God is not the only ultimate cause of all in the created world, but instead he has given the created world significant causal independence. Here the open theist is motivated by God’s love rather than power and sovereignty: God respects the autonomy and freedom of created persons and the created world.

One criticism that Molinists and divine determinists might level against open theism is that it denies God’s omniscience and foreknowledge. Open theists have a reply ready to this and it relates to the tricky issue of what makes future propositions true in the first place. Consider this proposition: ‘Aku will jet ski tomorrow’. According to divine determinists, God knows this to be true simply by virtue of his will: he intends to cause it tomorrow (or causes it eternally in his atemporal existence). On the Molinist account, God knows this to be true, because he has access to counterfactuals of freedom regarding what Aku would or would not freely do in certain circumstances. Open theists deal with this problem in different ways. One possibility is to argue that propositions about future free actions have no truth-value at all before the free agent actually makes her choice. In other words, ‘Aku will jet ski tomorrow’ is neither true nor false right now. Its truth or falsity is determined by Aku’s choice tomorrow. If there is no truth to be known regarding such a proposition, it takes nothing away from God’s omniscience if he does not know it. Indeed, it would be impossible for God to know it, because it cannot be known. Open theists have also provided other options: perhaps there are matters of fact about future contingencies, but they cannot be known; or perhaps all future contingent statements are false right now. Be that as it may, the open theist has multiple options at her disposal.

Another critical point that Molinists and divine determinists make against open theism is that it leaves God’s plans vulnerable to contingency emerging from chance in nature and freedom of created persons. It might turn out that God’s plans will not be realised after all. This worry concerns the central motivation behind divine determinism: in order for the faithful to trust God, they must be able to believe that God is in ultimate control over everything and nothing can stand in the way of God’s benevolent will. Open theists are forced to reject this view, at least to some extent. Because God’s plans involve beings with libertarian free will, there is no way God can make the outcome certain. In this sense, God indeed takes risks and can fail to achieve to his purposes. However, this possibility is only a metaphysical one, not necessarily actual. In the actual world, God has a plan to counter any contingency introduced by chance or human free will. God is like a chess master playing against a novice player. The master cannot know how the novice will move the pieces, but the master has a response to all possible moves. This enables the master to win the game regardless of how the novice decides to move the pieces.

It is not difficult to see why open theism might be attractive to those who want to adopt a free will theodicy. Since humans have significant moral autonomy and libertarian free will, it is much easier to lay the responsibility for moral evil in human hands. Open theism makes it possible to invoke human moral autonomy as well as randomness and chance in nature in order to explain the existence of evil. Some evils might have no reason whatsoever for their existence (van Inwagen 2008). Open theism also makes sense of the problem of evil created by eternal punishment, as discussed further below.

There has been some debate about whether some form of open theism might be compatible with divine determinism after all. This would be a workable option, if we could distinguish divine determinism, which preserves freedom and randomness (or so it is argued), from causal determinism, which apparently does not. Kathryn Tanner, for instance, has argued that divine determinism functions in a non-causal way, so it is not a threat to human leeway freedom at all (Tanner 2005). The roots of this discussion can be traced back to Aquinas. Combining Aristotle’s philosophy to produce an account of God and creation, Aquinas maintained that God is the primary cause (Latin: causa prima) of every contingent event. God is – in a very strong sense – the cause of everything else apart from himself. Aquinas also puts forward an account of concurrence, where God participates in all created causal relations. However, God’s causal activity in the created world is of a different nature than normal, ‘intra-world’, secondary causation (Latin: causa secunda). All contingent events, such as human free actions, have secondary causes that are based on the created natures of substances that produce those events. The primary/secondary causation distinction is made to highlight the transcendence of God and the difference between immanent causation and God’s transcendent causation. God’s activity is not simply a link in a causal chain of other causes. It is transcendental, beyond the created world and its system of causes (In I Perihermeneias 1. 14 nr. 22). Moreover, because of this omnicausality, the causal traffic between the creator God and the created world is radically unilateral. There is nothing that created substances can cause ‘on their own’; nor can they have an effect on God, since it is God who is the cause of everything else. Aquinas argues that God’s primary causality does not interfere with human free will, because God’s ‘causation’ is transcendent, not a part of the network of secondary causality (Summa contra Gentiles II c. 21).

If Aquinas is correct, the truth of divine determinism does not entail the truth of physical causal determinism. The created world could (arguably) have various contingencies and even randomness while God’s primary causality would be the source of everything (Stump 2003: ch. 9). For such reasons, divine determinism could be freedom-preserving whereas physical determinism might not be. Following this idea, it could be argued that God’s concurrence and omnicausality are compatible with robust libertarian accounts of free will (Grant 2016). If God’s causal activity allows for secondary contingency, randomness, and indeterminacy in nature, it could allow for indeterminate choices between alternative possibilities in the human case.

Finally, some theological views of providence and the God/world relationship have been inspired by various developments in twentieth-century natural science and process philosophy. Philosopher and mathematician Alfred North Whitehead (1861–1947) attempted to create a new metaphysic that would take into account the dynamic aspects of nature that modern physics and biology had revealed. According to Whitehead, we should not view the world as a collection of objects, substances, as most Western metaphysics had done so far, but as processes. Whitehead’s ideas had a deep impact in theology, especially through the work of Charles Hartshorne (1897–2000) and John Cobb. On the Roman Catholic side, process thinking inspired Pierre Teilhard de Chardin’s (1881–1955) views. According to process theology, God lacks the traditional attributes of omniscience and omnipotence. God is neither in full providential control over creation nor has complete knowledge of the world’s future.

Process theologians also tend to reject the radical difference between God and the created world but without identifying God with the totality of created processes (panentheism). In a way, God is a part of the process of the created world, ‘luring’ or ‘enticing’ it into following his will. Such an account of God would naturally make room for human free will and contingency in the created world. It would also solve the problem of evil, at least to some extent: God does not have the power to remove all evil from the world. Finally, many have argued that process theology as an account of human autonomy and God’s resistible love is ethically preferable to that of traditional divine determinism. Recently, process theology has given rise to many modified versions of theism and panentheism (Oord 2015).

2.2 Sin, grace, and free will

The role of human agency in the process of salvation has emerged as one of the most central contexts of the theological free will debate. While St Paul and other New Testament writers did not use the language of free will, they nevertheless put forward claims about the extent to which humans can act in ways that please God. For Paul, these questions are intertwined with his discussions on the relationship of the Christian gospel and the life according to Mosaic Law. The sinfulness of humans is such that no one can act in ways that would merit salvation by God. Moreover, sin is such that humans have no ultimate control over it. Humans are subject to evil forces that are beyond their abilities to overcome (Rom 1). These forces also exert influence on humans’ internal life: not only do they occasionally commit sins, but they also desire to commit sins and end up committing them inevitably.

The doctrines of grace and justification were central to the Pelagian controversies in the time of Augustine and later the sixteenth-century Reformation. The main question was the extent to which sinful human individuals can contribute to their coming to saving faith and fulfil God’s righteous demands towards them. Augustine argued strongly against Pelagius and others, who (apparently) held that human beings could, at least in principle, live a life pleasing to God without God’s supernatural gift of grace. For Augustine, medieval theologians, and later theologians of the Protestant Reformation, this was ruled out as impossible. Human beings in their sinful state cannot act in such a way as to prepare for or act in ways that would merit salvation. Augustine himself maintained that humans, nevertheless, have free will. While humans had a libertarian free will before the fall (how could humans be culpable for sin otherwise?), sin subsequently corrupted human desires and rational capacities such that humans are unable to resist acting sinfully (compatibilist free will). However, through God’s unmerited grace, human mental capacities are purified so that they can once again function properly.

On the whole, it seems that the younger Augustine of On Free Choice of the Will (written in periods between 387–395) was more optimistic about the power of human natural agency to adhere to God’s will and resist sinful mental movements. Augustine maintains that the person’s will can always either assent or resist the outputs of reason and perception. So, it seems that humans would have the power to choose between morally significant alternative courses of actions and they cannot be held accountable without it. Such a view certainly looks like a libertarian one. However, in the later parts of the book, he also argues that the assent has its source in the desires and outputs of the mind. One’s will (Latin: voluntas) is so essential to one’s nature that one has no power over its moral nature (Couenhoven 2017a).

Later on, Augustine had to revisit his views when he vigorously engaged Pelagius (c. 354–418) and others, whom he considered heretics (De Peccatorum Meritis et Remissione libri III; Contra Julianum). While Augustine himself denied that he had changed his views (Retractiones), many scholars doubt this claim. The later writings of Augustine adopt a more pessimistic position: the human will in its sinful state is incapable of making morally relevant choices between good and evil actions. Without God’s intervention through grace, the human person is only capable of choosing evil actions. At this point, Augustine also develops his influential notion of original sin as a corruption of the orders of human love. The root of sin is the inability of the sinful human person to put the love of God and other human beings above all else. The sinful will, however, loves itself too much, often succumbing to pride and other vices.

The views of mature Augustine (which had the most influence in medieval and Reformation thought) were formed against his very robust view of human sinfulness. Apparently, Pelagius and his followers argued that humans had the ability – if they decide to exercise their free will – to live a pure and sinless life (Letter to Demetrias). Augustine responded to this by invoking St Paul, who especially in the letter of Romans seems to be rather sceptical as to whether human moral agency could fulfil the law. Augustine argues that human moral agency is severely impaired: while humans have not lost the capacity to make free choices, original sin prevents them from exercising this capacity properly or rightly (Couenhoven 2005). Sin introduces conflicting desires in the human mind, which lead to fragmentation. Sin is like a disease of human moral capacities that God’s action cures. God’s unilateral action, the gift of supernatural grace, is required in order to free human agency from the grip of sin so that the power of free choice can yet again be exercised in the right way (Nisula 2012).

An extension of the Pelagian controversy was the debate between Martin Luther and Erasmus of Rotterdam in the sixteenth century. The Reformation brought traditional Augustinian themes of sin, grace, predestination, and free will to the forefront again. Luther had reacted strongly against late medieval voluntarism of Gabriel Biel (1425–1495) and others in his Assertio. Luther argued that the voluntarist account of the freedom of the will had resulted in multiple spiritual and institutional wrongs in the Roman Catholic Church. Salvation, Luther maintained, was no longer a work of God but had become an affair of the human will. In 1524, Erasmus of Rotterdam (1466–1536) – a Dutch humanist sympathetic to the cause of the Reformation – wrote his De Libero Arbitrio Diatribe Sive Collatio as a response to the Assertio, criticizing its radical conclusions about the lack of free will. If humans had no free will, there would be neither political peace, nor morality, nor the just punishment of sinners and rewarding of saints by God. Luther then answered this in 1525 with his hefty De Servo Arbitrio.

The contested issue was the role of human agency in salvation, but this is not the only disagreement (Visala and Vainio 2020). Erasmus and Luther disagree on the role of biblical interpretation in theological methods, as well as on the authority of the Church and its theological tradition. Treading in familiar territory already mapped out by Augustine, both Luther and Erasmus hold a strict Anti-Pelagian stance: salvation is the work of God and grace is an unmerited gift that God gives to the sinner. Sinful humans cannot save themselves by means of their own meritorious actions, nor is there anything that the sinner can do without God’s help to move herself from the sinful state to that of godliness. Moreover, Luther and Erasmus both agree that humans are morally responsible agents in God’s eyes. It is God’s right to punish sinners and reward the saints.

However, there is significant disagreement between Luther and Erasmus as to the role of human will in the overall process of salvation and the grounding of human moral responsibility. For Erasmus, some measure of undetermined active human contribution is required in the process. After God’s initial action towards the sinner, God works in cooperation with the individual’s will in a way that cannot be preordained, predetermined, or otherwise made necessary by God. If God is the sole cause of salvation, there can be no liberum arbitrium (free will or free choice). Against this, Luther flatly denies any kind of undetermined human contribution to the process of salvation. The will that cannot choose the good in its sinful condition does not deserve the name of liberum arbitrium.

Erasmus presents a number of arguments for the conclusion that Christians should accept a libertarian account of free will. First, if humans had no choice as to whether they acted morally or not, or whether they believe in God or not, it would undermine human moral responsibility. Without the assumption that humans make undetermined choices between morally and spiritually significant alternatives, it is difficult to justify God’s judgment and the blameworthiness of humans for their sins. But because God righteously rewards the saints and punishes the sinners, humans must have free will. Otherwise God’s judgments would be unjust. Furthermore, if humans had no libertarian free will, it seems that God would not only be unrighteous in his judgment, but also the cause of evil and sin in the world. Without human free will, the problem of evil would become intolerable.

Erasmus also invokes the language of the scriptures in defence of libertarian free will: God treats humans as moral agents by issuing commandments and laws, and occasionally calling on them and persuading them. Such actions seem meaningless and irrational, Erasmus maintains, if there is nothing humans can do to respond to God. Interestingly, Erasmus also invokes the role of freedom of the will in moral and political affairs. If humans cease to believe in free will, they will also cease to strive to live better in moral and spiritual affairs. Belief in free will serves an important social and moral function, and Erasmus is worried that Luther’s vehement criticisms of free will might lead to unwanted moral and political consequences.

Luther’s reaction against Erasmus was strong in style and content (Kolb 2017). Luther explicitly denies any kind of human free will with respect to spiritual matters. For liberum arbitrium, a person needs to be able to choose between morally significant alternatives. With respect to faith in God, no human being has such ability, since coming to faith depends on God only. In its sinful state, the human being cannot even will anything pleasing to God, not to mention generating salvific faith in God. In addition, Luther invokes a strong doctrine of original sin, according to which humans in their sinful state have no control over their will. They cannot make spiritually and morally significant choices; moreover, they cannot shape their character or nature from which their actions emerge. Rather, the ultimate control over a person’s character resides either with the Devil or God. In this sense, the human will is always in bondage, a servant or a slave to good or evil.

Luther’s main reason for such a radical account of human sinfulness is that if humans had even some power to act in such a way as to be pleasing to God, it would follow that the gospel, God’s unmerited gift of grace, would be in vain. If humans had free will, they could – at least in principle – incur some merit from God for their good deeds. However, the gospel tells us that no human action whatsoever can earn merit in God’s eyes. So, it is the truth of the gospel that compels Luther to reject any kind of human free will.

Finally, Luther also responds to Erasmus’ claim about human moral responsibility and the need for theodicy. Luther argues that God is within his rights to hold humans responsible for their actions even while humans lack free will. From a contemporary perspective, Luther could perhaps be understood as semicompatibilist). According to semicompatibilism (Fischer and Ravizza 1998), free will might not be compatible with determinism but human moral responsibility is. Following Augustine, Luther maintains that humans need no ultimate control over their actions in order to be blameworthy. Instead, they can be blamed or praised by God because their actions have their sources in their own intentions, desires, and wills. A sinful person does not act against her will, but simply according to her will (her deep-seated values and judgments). Sinful natures are not God’s fault (they are Adam’s fault), so God can judge the evil action stemming from them, even while he concurrently causes their evil actions. Luther maintains that God is not responsible for evil in any sense while he participates in causing evil actions, so it is in God’s rights to eternally damn the sinners and punish the saints.

The best way to characterize the differences between various accounts is to distinguish between deterministic and non-deterministic doctrines of grace (Timpe 2014: 51–52). According to deterministic doctrines of grace, like the one Luther defended, the necessary and sufficient condition for an agent to come to faith and remain in faith is God’s act of giving and maintaining faith. In this sense, whether the human agent has faith or not is completely up to God. No human choice, decision, or any other kind of action (understood in terms of incompatibilism) is necessary. Deterministic accounts of grace often emerge in the Augustinian tradition, especially from Lutheran and Reformed theology. The doctrine of irresistible grace was accepted by the synod of Dort (1609) and, subsequently, many Reformed theologians (historically Edwards and others; also modern scholars such as R. C. Sproul and J. R. Packer). Opposed to this is the non-deterministic doctrine of grace defended by Erasmus and Arminius, as well as most Catholic and Eastern Orthodox theologians. According to this view, God’s giving of grace is a necessary but not a sufficient condition for salvation. In addition to God’s giving of grace, an additional assent or act of human will is required. Here assent or act of will is understood in terms of incompatibilism: hence it would be impossible for God to determine such assent or act of will.

Crucial for this discussion is the question of the effects of sin on human moral and spiritual capacities. Most Christian theologians agree that sin has somehow corrupted or impaired human moral agency, as well as the way in which humans relate to one another and God. On the one hand, human culpability for sin needs to be maintained, so that evil actions and sin can be, at least to some extent, attributed to humans themselves. If not, God’s final judgement seems to make no sense. On the other hand, an account of human sin and evil in general needs to be such that God remains blameless and faultless for sin and evil. Striking a balance between these two motives turns out to be rather difficult. If a libertarian account of free will and human original sinlesness are assumed, it is difficult to account for why humans made the first evil choice in the first place. If an agent has a choice and her cognitive system is such that it faultlessly identifies what is right and good, it seems perplexing that such an agent would make an evil choice. However, assuming a compatibilist account also leads to problems. If free human choices were compatible with God’s necessitating activity, it would seem that God could have prevented the fall to sin, or that God actually decreed it. In both cases, it seems that God is, at least to some extent, responsible for human sin and evil.

2.3 Evil, theodicy, and hell

As became clear above, free will has had a central role in debates about theodicy. We have seen three ways in which the problem of evil can be posed. First, there is the question of whether God is just in allowing the degree and amount of evil we see in the world: is it not incompatible with God’s omnipotence and perfect goodness? Second, there is the issue of God’s involvement in human evil and sin. If God is the cause of everything else outside of himself, is God not also the cause of human evil and sin as well? Again, an account must be offered where the evil and sinful aspects of human action are attributed to the human agent, not God. Finally, there is the issue of whether God’s ultimate judgment and the possibility of eternal retributive punishment is righteous.

A handy way to solve these problems would be to postulate the existence of human libertarian free will. Ever since Augustine suggested it, theologians argued that at least some cases of evil are explained by human free will. Libertarian free will could also explain how the causes of evil and sinful human actions could be found in humans alone, not in God. Finally, libertarian free will would also explain God’s everlasting punishment of sinners. Given the benefits of libertarian free will in solving the problem of evil, it is no surprise that so many twentieth-century theologians and philosophers have rejected divine determinism.

The argument that humans are responsible for evil, not God, because humans have free will, can be traced back at least to Augustine (and probably even further). In contemporary context, Alvin Plantinga developed on this basis his well-known free will defence (Plantinga 1974). Plantinga tackles the logical problem of evil (Mackie 1955), according to which the claim ‘God is perfectly good, omnipotent and omniscient’ is logically inconsistent with the claim ‘evil exists’. Plantinga responds to this problem by arguing that given human libertarian free will, there is a possible world in which there are evils perpetrated by humans but, despite his perfect goodness, God allows those evils to take place. It is important to note that Plantinga’s counterargument need not be a full-blown theodicy. It is merely what he calls a defence, that is, a demonstration that the existence of an omnipotent and perfectly loving God is compossible with the existence of evil. Of course, it is possible to develop a more thorough theodicy on this basis. Many take Plantinga’s defence to be so successful that the focus of the debate about theodicy has subsequently moved away from the logical problem of evil towards other versions, like that of the evidential argument from evil. Nevertheless, there is still debate whether Plantinga’s defence is successful.

Most contemporary theodicies, like the soul-making theodicy, assume the existence of libertarian free will (Hick 2010). Plantinga’s free will defence, as well as other free will theodicies, have resulted in widespread acceptance of libertarianism among English-speaking philosophical theologians, even while many come from theological traditions that have largely rejected libertarianism, like Plantinga himself (Timpe and Speak 2016: 4–7). However, there are also philosophers who have argued that theological compatibilism can make use of the free will defence (Turner 2013).

Free will is also central in debates about the problem of evil created by the prospect of God punishing some humans by eternal suffering. Despite a few dissenting voices, the doctrine of eternal punishment has been a core part of Christian orthodoxy. From the eighteenth century onwards, the traditional doctrines of hell have come under heavy attack. Again, the roots of the classical doctrine of hell can be traced beyond Augustine, but it was Augustine that gave it its form. According to the classical doctrine of eternal punishment, there are created persons who will exist eternally, are never saved, and eternally suffer from both the effects of their estrangement from God and additional torments as just punishments for their sins. Moreover, doctrines of predestination have emerged on the basis of deterministic accounts of grace: if salvation is completely up to God, it depends on nothing else than on God’s eternal will, or his decree to save some people and damn others (Couenhoven 2018). This set of doctrines creates a very difficult form of the problem of evil. God could, if he so wanted, give grace to everyone and make sure that they remain in faith. He does not do so. Why? It cannot be because of any benefit for those who are damned, because they will remain infinitely in a state that is objectively the worst possible for any human. Augustinian theologians sometimes suggest that the benefit is the fulfilment of God’s glory and justice. Without the damnation of some humans, God’s righteousness would not be displayed in creation.

At least two factors have driven the criticisms of the traditional doctrine of hell. First, libertarian accounts of free will have become increasingly attractive for understanding moral responsibility. If an agent can do nothing to move herself towards faith and away from sin, it seems natural to assume that the agent has no control whatsoever over her sinful state. Given this, it seems natural to claim that the agent cannot be culpable for sin in the eyes of God and should not be held accountable at all. The other critical point against traditional doctrines of hell arises from theories of punishment. The account of retributive punishment often assumed by defenders of traditional views of hell has also been subjected to strong criticisms. According to the retributive theory, punishment can be deserved simply through wrongdoing, even if the punishment has no benefits for anyone and serves no further purpose. Recently, many other theories of punishment have emerged. These theories invoke various benefits of punishment (it deters future evildoers, it protects the public, and it might function as a corrective to those who are punished) or the signalling effect of it (punishment signals that certain behaviours are unacceptable).

As a consequence of these developments, many (e.g. Walls 1992; Stump 1986) have developed an alternative to the traditional doctrine. This is sometimes called the liberal doctrine of hell or free will hell. This view does not take hell as a retributive punishment that God metes out to sinners, who cannot control their sinfulness, but rather as God giving to humans what they in fact want. The defender of free will hell argues that libertarian free will makes it impossible for God to guarantee the salvation of all human persons, and that it would be incompatible with God’s love of his creatures to override their free will. In both cases, ‘the doors of hell are locked from the inside’, as C. S. Lewis puts it. Indeed, God’s love for his creatures, according to this view, requires that God respect their autonomy and dignity. So, it is God’s love that creates the possibility for some of his creatures to willingly reject him forever. The possibility of hell is an expression of God’s love. Such an account of hell assumes a very robust view of human autonomy and assigns a very high value to it – so high that God is willing to allow humans to reject him infinitely.

Finally, free will has a central role in debates about the possibility of universal salvation. Although a minority view since the times of early Eastern Fathers, like Origen and Gregory of Nyssa, universal salvation has gathered an increasing number of contemporary defenders. Contemporary universalists often assume the existence of some purgatory-like post-mortem punishment and purification, but this is not considered infinite in duration. Given enough time, all humans will freely accept God’s grace and join the beatific vision (Kronen and Reitan 2013). From the point of view of free will hell, it could be argued that universalists are no better guardians of human free will than divine determinists. It seems that in order to guarantee the salvation of all, God would have to make the salvation of all humans necessary by some infallible decree. This, in turn, would entail the rejection of libertarian freedom. While some universalists prefer compatibilist accounts of freedom for such reasons (Hart 2019), others have offered arguments for the compatibility of libertarianism and universalism. One possibility is to invoke Molinism: God can guarantee salvation via his access to counterfactuals of freedom. Another option is to maintain that while humans have libertarian free will, every single human will in fact turn to God eventually, if God gives adequate knowledge about himself and enough time for purifying character development (Talbott 2014).

2.4 Divine will

Discussions and debates about the nature of God’s will and God’s freedom have deep roots in the Christian theological tradition. The reason for this is not difficult to understand: since God is supposed to be perfect, God’s agency should also be conceived of as perfectly as possible. So, whatever theologians write about God’s will and freedom, it represents a form of ideal agency – it is the standard against which limited and finite human agency is contrasted. God is the ultimate free agent; but what does this mean?

Here, the debate between intellectualists and voluntarists can be seen most clearly (Leftow 2017). Intellectualists about the will, like Aquinas and Augustine, have understood God’s free will in the light of his perfect reason. Human free will is limited, because it suffers from various deficiencies: reason often fails to identify what is best and chooses the wrong means; the will is often impeded by internal disintegration (conflicting emotions, desires) or external constraints. God, however, suffers from none of these limitations since his knowledge and power are perfect. God’s reason is perfect, so he always knows the best course of action. God is omnipotent and simple (in the technical sense of lacking psychological and metaphysical parts or divisions) so there can be no internal or external constraints on his actions. Because God is the source of everything else, there is nothing external to God that could constrain his actions or impede the functioning of his reason. So, God’s perfect freedom consists in the power to do what is the best in all situations.

Voluntarists, like Duns Scotus, would see God’s freedom in a different way. For them, God’s ultimate freedom does not reside in the fact that God always acts perfectly rationally. Instead, God is free because God can choose whatever he wants; God has access to all alternatives. When taken to the extreme, this view leads to the conclusion that there is nothing prior to God’s decision that would explain it or make it necessary, not even God’s reason. We see this way of thinking in the case of Luther, whose views are examined above. According to Luther, God is the only being with free will. Free will, properly understood, is a divine attribute; therefore, it is not something that a created being might ever have. God has free will, because his will is not dependent on anything external to it, not even God’s mind. God can do whatever he decides to do and God’s decisions are brute facts that ground all other facts, including moral facts.

The difference between voluntarists and intellectualists regarding God’s will has significant metaphysical and ethical consequences. One such consequence pertains to the nature of the created world as a whole and our knowledge of God. Christian theologians widely agree that God creates the world freely – meaning that creation is a free decision on God’s part that involves no prior necessity. On the intellectualist picture, God acts out of his perfect reason, so we can assume that the created world is a rational world. The created world is not random or chaotic, but an understandable whole, since it is created by a rational God whose actions are supremely reasonable. It also follows that by examining this world, we can infer the rational principles behind it and learn something about God. The voluntarist would paint a different picture. If the created world is a product of God’s free decision for which there is no prior reason or explanation, he could have created a vastly different world than ours. In a way, the created world is not a product of perfect reason, but a perfectly free choice. God’s choices are brute facts that have no explanation external to God’s omnipotent will, so humans cannot really understand them. It is by no means clear that our reason can discern God’s rational principles in the created world. It is also not clear whether there is anything in such a world that reflects God’s nature or attributes. The way to knowledge of God through the created world is therefore shut.

In recent philosophical theology, some of these convictions have been used to generate arguments against the existence of a Christian God. One argument could be put like this: how can God be perfectly free, when he lacks significant alternative possibilities (Pike 1969)? We humans can choose between good and evil actions, but God cannot. Since God is supposed to be perfectly good, he cannot choose to commit evil actions, so he has fewer choices available to him than we do and therefore lacks power. One possible response is to maintain that God does not lack any significant power by being unable to act in evil ways. We humans have the ability to act in either morally praiseworthy or blameworthy ways, but this is because of our sinful nature. God, however, lacks any kind of sin and his moral nature is perfect. It would be logically contradictory for a morally perfect being to do evil, because there is no reason for such a being to do evil and all the reason to do good. Being able to do what is logically contradictory by no means detract the power of a being, because such states of affairs are impossible. Following this line of argument, God’s omnipotence is not restricted or constrained by his perfect moral nature.

Another challenge emerges when certain accounts of divine will are combined with the doctrine of creation. As mentioned above, creation is supposed to be a free act of God. If this free act is understood in terms of intellectualism about God’s will, it seems that God really did not have the option of not creating, or to choose between different possible worlds he could create. Instead, God’s perfection would compel him to create a world and the only possibility would be to create the one and only best possible world. So, not only does God not create freely, but he must create the best of all possible worlds. But what if there is no such thing as the best possible world? In some sense, the world could always be better. If God cannot create the best possible world (because no such thing exists) but creating such a world is a necessary feature of a perfect being, then there can be no perfect being like God at all (Rowe 2006). Such arguments have been vigorously debated in recent decades. One response is the argument that it is permissible for a morally perfect being to create a world that is ‘good enough’. Moral perfection does not require God to create a world where all value is maximised (Adams 1972).

2.5 Scientific challenges to free will

Finally, let us briefly look at some of the challenges that the results of modern behavioural and natural sciences might present to theologians defending free will. Since the late twentieth century, traditional notions of free will and moral responsibility have been subject to intense criticism from the sciences. We can identify three scientific challenges to free will that correspond to three necessary aspects of it: 1) there must be an agent, whose 2) actions are not determined by prior causes and 3) who is the source of her own will (List 2019).

The first challenge, called the eliminativist challenge, is directed towards the notions of agency itself. According to some philosophers and scientists (Churchland 1988), the combined results of evolutionary biology, cognitive science, and neuroscience suggest that notions like ‘souls’, ‘selves’, ‘intentions’, and ‘beliefs’ are a part of our everyday psychological theory – referred to as folk psychology – which we use to predict each other’s behaviour. However, the progress of these sciences shows that our folk psychological notions are in fact inadequate in explaining human behaviour, because they do not track the real causes of our behaviour. Given the progress of the neurosciences, folk psychological notions will be replaced by scientific ones in the future. The problem is that the concept of free will requires such folk psychological notions: there have to be selves that act, intend, and decide. If there are no such entities and folk psychological descriptions are false, there can be no such thing as free will. Indeed, there are no selves or intentions or decisions either.

The argument from eliminativism raises the question between the connection of self/body dualism and free will. What is the self and how is it related to actions? Traditionally, theological debates about free will have taken place in the context of either Platonic mind/body dualism or Thomistic hylomorphism, according to which the intellectual soul is the form of the body rather than an individual substance independent from it (Goetz and Taliaferro 2011). However, for a number of reasons, such views have fallen under strong criticism and theologians have had to rethink their basic assumptions about anthropology. One reason for this has been the advances of modern psychological sciences and overall criticism of dualism in philosophy, especially by many physicalist philosophers. In addition, from the twentieth century onwards, biblical scholars have also suggested that holistic Jewish anthropology underlying the Hebrew scriptures is quite far from later Platonic dualism adopted by many Church Fathers (Green 2008). As a consequence, some theologians have turned towards physicalist or monist accounts of human nature (Murphy 2006).

Given robust Cartesian mind/body dualism, it is natural to think that what causes the agent’s actions is the agent herself. The agent is an individual, spiritual substance with a specific causal power to initiate actions (Swinburne 2019). Indeed, such a view has been quite typical of free will libertarians for centuries (Thomas Reid, Essays on the Active Powers of the Human Mind, 1788). This view of causation of action assumes the existence of a special type of causation known as agent causation. However, many contemporary libertarians have rejected this view, because of the strong criticisms it has received and its dubious relationship to mind/body dualism. As a consequence, philosophers have formulated libertarian views that invoke normal event causation (Kane 1996). In this view, it is not the agent as a substance that is the cause of her actions, but the mental states closely associated with the agent. It is the agent’s beliefs, desires, and intentions that cause her actions.

The second issue arising from modern sciences is the challenge of determinism. We already alluded to the challenge of early modern Newtonian physics: it seems difficult to reconcile the determinist and mechanist view of nature with that of humans making indeterminate, rational choices. The challenge of global, causal determinism of this kind, however, has been alleviated in the twentieth century due to quantum physics, chaos theory, and various types of emergence. It is not at all clear that global determinism is the final word even in physics (Earman 2004). In any case, the challenge of determinism is not restricted to physics only. Instead, the twenty-first century challenge to free will comes mainly from biology, genetics, and neuroscience. The question in this context is the possible truth of genetic determinism or neurobiological determinism respectively. If genetic determinism were true, the direction of human lives, especially of human characters, would be determined by biological factors outside human control, namely human genetic makeup. Neurobiological determinism is the view according to which all mental states are products of underlying brain states. Given that brain states are physical states sufficiently explained by prior physical causes, there could be no mental causation – that is, the agent’s mental states (intentions, beliefs, desires, decisions) cannot be the causes of the agent’s actions.

One possible theological response to the challenge of various forms of determinism would be to simply accept their truth. If one already accepts divine determinism, perhaps adding new determinisms in the mix would not be that difficult. This, however, would not solve the issue of neurobiological determinism. The problem there is that the denial of mental causation implies that the agent lacks control over her actions. It is the agent’s brain that causes her actions, not her mental states.

This worry is further amplified by the third challenge, the argument from epiphenomenalism. According to this view, there is empirical evidence that mental causation is only apparent causation, an illusion. It turns out – the argument goes – that the conscious decisions and intentions of the agent are rarely, or never, the causes of her behaviour (Wegner 2002). In other words, crucial mental aspects of ourselves (like our beliefs, intentions, and goals) are, in reality, disconnected from our actions. Instead of selves initiating and controlling actions, human beings construct their self-representations on the basis of actions that are initiated with very little input from their self-representations. If this is the case, our consciously accessible mental states make very little difference in terms of our actions. In short, our actions are often driven by unconscious factors outside our control, so we have no free will.

One possible response to such challenges would be to acknowledge that a Christian account of human freedom always includes severe constraints. The social and moral context of human freedom places limits and restrictions on the individual. Sin has fractured this world and its communities, which in turn produces human individuals whose ability to understand moral reasons and act according to them is severely limited. While the results of psychology and neuroscience can function as a reminder to theologians not to overstate their case for human freedom, they go even further: if epiphenomenalism is true, even the most minimal conditions for free will and moral responsibility, like rational self-control, cannot be met by human agents.

Theologically motivated responses to these challenges have taken two distinct paths. The first is to double down on agent causation and substance dualism. Christian theology, it is argued, entails that humans have libertarian free will and robust selfhood, so eliminativism must be false (Hasker 2001). Similarly, genetic determinism and neurobiological determinism must also be false. Indeed, these are often turned around as arguments against naturalism: if naturalism is true, there can be no free will and moral responsibility. Given naturalism, eliminativism about minds and determinism about our nature must also be true. However, if we accept human free will and the self as facts, this will lead us to adopt a supernaturalist position: naturalism is not the final word about us (Goetz and Taliaferro 2008).

The second path adopts a compatibilist and event causal account of free will and allows for the possibility of physical selfhood. It is possible to conceive of human freedom at least in terms of rational self-control, even if the power to choose between alternative possibilities cannot be salvaged (Murphy and Brown 2007). There might be a number of theological reasons to avoid substance dualism anyway. Perhaps a physicalist account of personhood makes better sense of both biblical commitments and scientific results.