Normative Reasons, Epistemic Autonomy, and Accountability to God

Abstract

According to many, human autonomy is necessary for moral action and yet incompatible with being morally accountable to God’s divine commands. By issuing commands that ground normative facts, God demands our accountability without understanding our normative reasons for moral action, which crushes human autonomy. Call this the Autonomy Objection to Theism (AOT). There is an unexplored connection between models of normative reason and AOT. I argue that any plausible AOT must be stated in terms of an adequate model of normative reason. There are two broad metaethical categories for models of normative reason: anti-realist or realist views. I defend the thesis that both anti-realism and realism about normative reasons fail to support AOT by means of a dilemma. If the AOT defender adopts anti-realism about normative reasons (subjectivism and constructivism), AOT loses its force. However, if the AOT defender adopts moral realism, they face the same problem as the theist, as normative fact constrains autonomy. Consequently, AOT is a problem for all moral realists, including non-theists, such as Russ Shafer-Landau, David Enoch, and Erik Wielenberg, among others.

1. Introduction

We can enjoy morally significant autonomy or theism, but not both. So says a longstanding Kant-inspired object, that any version of theism according to which we are accountable to God by divine command.1 This objection is based on two claims: (1) human autonomy is necessary for moral action, and (2) human autonomy is incompatible with the standard theistic claim that humans are morally accountable to God apart from understanding their reasons for obeying God. Call this line of reasoning the Autonomy Objection to Theism (AOT).2

Among the many replies to AOT, none explore the connection between the nature of normative reason (reasons for moral action) and AOT. Exploring this connection sets up a new way of answering AOT. I show that any plausible AOT must be stated in terms of an adequate model of normative reason. There are two broad metaethical categories for models of normative reason: anti-realist or realist views. I defend the thesis that both anti-realism and realism about normative reasons fail to support AOT by means of a dilemma:

The AOT Dilemma: If the AOT defender adopts anti-realism about normative reasons (subjectivism and constructivism), AOT loses its force. But, if the AOT defender adopts moral realism, they face the same problem as the theist, as normative fact constrains autonomy.

If correct, a surprising thesis is revealed: AOT is not a problem unique to theism but for all moral realists. This would be a serious problem as several non-theists embrace realism. For example, Michael Huemer writes, “I would say the objectivity of morality has nothing to do with God. As to “how” it is objective, it is objective by not constitutively depending on the attitudes of observers” (Huemer 2021, pp. 156; see also Huemer 2013). In explaining the thesis of his book, Wielenberg writes:

I accept moral realism yet I believe that God does not exist. I also find it unsatisfying, perhaps even “lame” as Mackie would have it, to posit mysterious, quasi-mystical cognitive faculties that are somehow able to make contact with causally inert moral features of the world and provide us with knowledge of them. The central goal of this book is to defend the plausibility of a robust brand of moral realism without appealing to God or any weird cognitive faculties.

(Wielenberg 2014, p. ix)

The AOT Dilemma, if correct, poses a serious problem, especially for those AOT defenders who adopt moral realism. So why think the AOT Dilemma is defensible?

2. Autonomy Objections to Theism

2.1. Divine Command Theory

AOT targets divine command theory, according to which at least some normative facts depend on or are grounded in facts about God’s commands.3 Following Robert Adams and C. Stephen Evans, I will limit the scope to moral obligations as follows (Adams 1987; Evans 2013).

Divine Command Theory (DCT): At least some normative facts—facts about moral obligation—are grounded in facts about God and God’s commands.4

The fact that any action, φ, is morally wrong is grounded in facts about φ being contrary to facts about God and God’s commands. The obligation and wrongdoing statements of DCT are compatible with, if not consequent of, each other. If one has an obligation to φ, then bringing about φ is morally right or good, while failing to φ is morally wrong.

I understand divine commands as communications from God regarding our collective and individual accountability to God for action and cultivating virtues to live a particular kind of life (perhaps one like Jesus’s) and become a particular kind of person (perhaps like Jesus). One might think that in following God’s commands that one cultivates the virtue of accountability to God. This virtue is a trait of those who are rightfully held accountable, who welcome and embrace being accountable and thereby show a sensitivity to what the relation requires (Evans and Rickabaugh 2022; Evans 2023). A sincere theist (notably Jews, Muslims, and Christians) might seek to cultivate the virtue of accountability to God, as they are accountable to God for their lives as a whole. Of course, the fact that one has a moral obligation to cultivate the virtue of accountability to God is grounded in facts about God and God’s commands.

2.2. Two Versions of the Autonomy Objection to Theism

It is worth distinguishing AOT from a related objection. Jean-Paul Sartre argued that if God created us, we would be mere objects, like a “paper knife”, unable to form a meaningful life by creating ourselves (Sartre [1946] 2007, pp. 20–23). Similarly, according to Kurt Baier, God is “a kind of superman … a sort of playwright-cum-legislator-cum-judge-cum-executioner” (Baier [1957] 2008, p. 83). These objections from self-creative autonomy are far weaker than AOT. For one thing, full self-creative autonomy does not guarantee a meaningful life. Plausibly, only the kind of autonomy God allows is necessary for a meaningful life (see e.g., Penner (2015)). Secondly, God could overcome any degradation of human life by bringing about a much more significant meaning to life overall (see e.g., Kraay and Dragos (2013)).

AOT requires a substantive notion of autonomy, a theory of normative reasons, and the relation between the two. To clarify these concepts, I will consider the autonomy objections of Nowell-Smith and James Rachels, as they are the most developed and discussed.

2.2.1. The Infantilizing Argument

The first version of AOT is what I will call the infantilizing argument (Nowell-Smith [1961] 1999). Nowell-Smith’s infantilizing argument employs Jean Piaget’s theory of moral development, especially the second developmental stage, where children (ages 5–9) obey their parent’s normative commands as unbreakable authoritative directives. Understanding what these moral rules are for or what makes them worth following does not guide the child’s obedience. In this respect, they are infantile. Children become autonomous in the third developmental stage by evaluating the reasons for obeying normative commands. The child’s non-autonomous infantilizing obedience dissolves as they gain an equal footing with adults (Piaget 1965).

According to Nowell-Smith, the stage two parent/child relationship is analogous to the God/human relationship. Just as the normative authority of the parent infantilizes the child, the normative authority of God infantilizes the human person.5 Similarly, Walter Sinnott-Armstrong points to cognitive science studies showing that raising children to obey God’s commands simply because God commanded them undermines their moral development (Sinnott-Armstrong 2009, p. 110). “Religious morality”, says Nowell-Smith, “is infantile” (Nowell-Smith [1961] 1999, p. 403). If DCT is true, so the objection goes, all are infantilized by God, subjugating us to the psychological development of children in Piaget’s second stage.

2.2.2. The Worship Argument

The argument defended by Rachels, what I will call the worship argument, attempts to show that the moral duty to worship God destroys human dignity. Kant’s words are often conjured for this point: “Kneeling down or prostrating oneself on the ground, even to show your veneration for heavenly objects, is contrary to the dignity of humanity…” (Kant [1797] 1996, p. 437).6 According to Rachels, to acknowledge God is to recognize God’s unlimited authority and our ultimate accountability to worship God. “That God is not to be judged, challenged, defied, or disobeyed”, writes Rachels, “is at bottom a truth of logic; to do any of these things is incompatible with taking him as One to be worshiped” (Rachels 1971, p. 333). Why think this is true? According to Rachels, this conception of God is self-evident and self-explanatory.7 The fact that God exists, if it is a fact, entails that we are accountable to God in such a way that we cannot come to know apart from God why we are so accountable. We are not autonomous in gaining moral knowledge about our accountability to God.

2.2.3. The Nature of Autonomy in AOT

Defining ‘autonomy’ is notoriously difficult. After surveying numerous notions of autonomy, often used interchangeably, Gerald Dworkin concludes that conceptions of autonomy are so diverse that they share only two traits: “autonomy is a feature of persons and that it is a desirable quality to have” (Dworkin 1988, p. 3). Some thirty-three years later, this problem remains.8

For my purposes, I need only consider the aspects of autonomy employed by defenders of the autonomy objection. Consider the claim of autonomy objector Robin Le Poidevin:

The truly moral agent is one who wishes to be his own master, not the instrument of some other power, and not to trust the deliverances of some supposed authority, but to work out for themselves the rightness of certain kinds of behavior.

(Le Poidevin 1996, p. 84)

Poidevin echoes Kant’s definition of autonomy as “the property of the will by which it is a law to itself (independently of any property of the objects of volition)” (Kant [1785] 1997, p. 440). Autonomy objectors also endorse Kant’s thesis that self-legislation and self-government are necessary for autonomy and human dignity (Ibid., pp. 435–36). Although, AOT does not reduce the ability to self-legislate or self-govern regarding the freedom to choose to φ or not to φ. One can be free to φ or not to φ and make one’s choice irrationally, yet not self-governing in any relevant way. Likewise, not being externally controlled or coerced to φ does not entail that one acts autonomously. Freedom is necessary but not sufficient for autonomy.

AOT also requires that insofar as one is autonomous, one must be able to step back from one’s desires and assess moral actions rationally. In conjunction with the freedom requirement, the rationality requirement allows us to act intentionally. This tracks with the standard (procedural) view of autonomy in bioethics, according to which an agent is autonomous concerning an action if it is performed intentionally, with understanding, and without determining influences.

Taken together, we get the following thesis:

Personal Autonomy: An agent S has personal autonomy, iff, S’s actions are performed intentionally, with understanding, and without conditions that determine S’s actions.

The autonomy objector might agree that the features of personal autonomy are necessary conditions for human autonomy in general, yet insufficient for the infantilizing and worship arguments. A child in Piaget’s second developmental stage can act intentionally, undetermined, and with understanding, for example, that her parent knows more than she does. So, if one is infantilized, it is not for a mere lack of personal autonomy.

AOT is concerned with a more specific thesis than Personal Autonomy. It requires epistemic autonomy or what Robert Adams calls total inner directedness: the power to rely exclusively on one’s reasoning or feelings in adopting moral principles, values, and priorities (Adams 1999, p. 271). More precisely:

Epistemic Autonomy: An agent S has epistemic autonomy, iff, S is able to form S’s beliefs by S’s cognitive resources, and S’s actions are performed intentionally, with understanding, and without conditions that determine S’s actions.

We can distinguish this from epistemic interdependence, the ability to think with and consult other agents as external cognitive resources. An important aspect of intellectual autonomy is that it provides an ideal context for forming true beliefs and gaining understanding and knowledge.

The autonomy objector does not require that one have epistemic autonomy tout court. As Elizabeth Fricker has shown, an ideal autonomous knower, someone who only believes what she can find out through her cognitive resources, without the testimony of another, does not exist (Fricker 2006). One cannot gain knowledge apart from the cognitive resources of experts. Fortunately, for the autonomy objector, AOT only requires epistemic autonomy regarding accountability to God.

2.3. Epistemic Autonomy and Normative Reasons

AOT requires that one’s reasons for moral action, one’s normative reason, must be adopted as a consequence of one’s Epistemic Autonomy. I understand the nature of normative reason as follows:9

Normative Reason: An agent S has a normative reason to φ at time t, iff, some fact about φ (concerning S’s situation at t) plays a particular role in explaining what S ought to do at t.

Accordingly, the explanation of what one ought to do at t supervenes on one’s reasons for and against the available courses of action at t. To seek out an explanation for why one should obey God is to seek out one’s normative reasons for obeying God. When free to do so, one has the following kind of autonomy:

Normative Reasons Autonomy: An agent S has normative reason autonomy with respect to φ at time t, iff, S decides to φ at t because S has a normative reason to φ at t, and S comes to know S’s normative reason with Epistemic Autonomy.

Accordingly, one is autonomous with respect to obeying God just in case one decides to obey God in light of one’s normative reasons for doing so. One comes to know those reasons through one’s cognitive resources intentionally, with understanding and without determining conditions.

Normative reasons autonomy is the point of contention in the infantilizing and worship arguments. Consider Kai Neilson’s rhetorical question:

Is it really hubris or arrogance or sin on our part to wish for a life where we make our own decisions, where we follow the rules we do because we see the point of them and where we need not crucify our intellects by believing in some transcendent purpose whose very intelligibility is seriously in question? Perhaps by saying this I am only exhibiting my own hubris, my own corruption of soul, but I cannot believe that to ask this question is to exhibit such arrogance.

(Nielson 1991, p. 22)

The autonomy objector claims that if DCT is true, then seeking to understand one’s normative reasons for obeying God entails that one disobeys God. The thought is that to seek one’s normative reasons to obey God requires refraining from obeying God until those normative reasons are found. Withholding obedience to God, even for the sake of understanding one’s normative reasons for doing so, is to judge, challenge, and defy God. Thus, DCT entails the impossibility of possessing the autonomy necessary for coming to know (with one’s cognitive resources alone) one’s normative reasons for obeying God and not disobeying God. That is, exercising one’s Normative Reasons Autonomy by seeking to know one’s normative reasons for obeying God is itself an act of disobedience to God. Without Normative Reasons Autonomy, we are infantilized and degraded by the inability to discover our normative reason for obeying God. AOT, therefore, depends on the nature of normative reason, especially the close connection between normative reason and autonomy.

3. Anti-Realist Models of Normative Reason and Varieties of AOT

AOT is a thesis about normative reasons, namely, that God should provide the autonomy necessary for us to seek and discover what normative reasons we have to be accountable to God, to live according to his divine commands. In order to know the kind of autonomy necessary for seeking and discovering these normative reasons, at least two things must be clarified: (a) the nature of normative reasons and (b) the relationship between normative reason and autonomy. Thus, the AOT advocate must posit a model of normative reason that is capable of supporting AOT. Moreover, the posited model of normative reasons must be closely linked to autonomy in such a way that God must provide this autonomy for the sake of our seeking and discovering the normative reasons for obeying God. Advocates of AOT have not clarified (a), and therefore have not clarified (b). In what follows, I will explore the most popular anti-realist models of Normative Reason in order to clarify (a) and (b). In doing so, I will argue that both models fail to support AOT.

3.1. Subjectivism about Normative Reasons Autonomy

Subjectivism about normative reason is a desire-fulfillment theory, grounding facts about normative reasons in facts about specific mental states: desires, attitudes, or aims. More precisely:

Normative Reasons Subjectivism: The fact that an agent S has a normative reason to φ at time t is grounded in facts about S’s desires.

What explains the fact that one has a normative reason to φ is the fact that φ-ing will satisfy one’s attitudes or desires. So, Hume was correct: “Reason alone can never be a motive to any action of the will”, and reason alone “can never oppose passion in the direction of the will” (Hume [1739] 1967, 2.3, p. 413). Put more famously, “Reason is, and ought only to be the slave of the passions and can never pretend to any other office than to serve and obey them” (Ibid. p. 415).

Prima facie, normative reason subjectivism seems to champion autonomy. Normative reason is only an instrument. The direction of explanation is from desire to reasons for action (Sobel 2021, p. 307). Facts about one’s desire, not any authority, divine or otherwise, grounds facts about one’s normative reasons. More specifically:

Subjectivist Normative Reasons Autonomy (SNR-Autonomy): An agent S is autonomous with respect to φ at time t, iff, the fact that S has a normative reason to φ at time t is grounded in facts about S’s desires.

Consequently, one is autonomously accountable to God just in case one decides to obey God because one is in the mental <desiring to obey God>. On SNR-Autonomy, what one ought to do at t supervenes on one’s desire(s) at t. Therefore, one’s normative autonomy can go only where one’s desires allow it.

SNR-Autonomy is certainly at odds with divine command theory. According to DCT, facts about one’s accountability to God and the facts that ground one’s normative reasons to obey God have nothing to do with one’s desires. Normative facts are iconoclastic with respect to one’s desires as they are grounded in facts about God and God’s commands.

A Problem for Subjectivist Normative Reasons Autonomy

As previously explained, AOT requires a close connection between autonomy and normative reasons. On SNR-Autonomy, however, it is not clear why normative reason should be closely related to autonomy. It may make more sense to identify autonomy with the ability to realize one’s preferences. In most subjectivist views (especially Humeanism), the weight or justificatory force of a normative reason covaries with the strength of one’s desire for x and the likelihood that S’s φ-ing will realize S’s desire for x.10 Thus, facts about normative reasons are always grounded in facts about one’s desires, although facts about one’s normative reasons autonomy are grounded in facts about desires in conjunction with facts about the likelihood of an action fulfilling one’s desire. However, now, paradoxically, a person’s autonomy is a function of external circumstances not totally within the person’s control, violating the non-determined condition of personal autonomy. Moreover, it is not clear how much, if any, autonomy one has when reason is a slave of one’s passion. If reason only obeys desires, normative reasons are disconnected from personal or normative reason autonomy.

Perhaps the subjectivist can modify their thesis to avoid these aforementioned problems. Some subjectivists (e.g., David Hume and Henry Sidgwick) limit the scope of relevant desires to informed desires that reveal one’s true concern (see e.g., Hume ([1739] 1967, p. 460), and Sidgwick ([1907] 1981, pp. 111–12)). Some characterize an informed desire as a second-order Frankfurt desire, that is, what we desire to desire within the context of our life projects (Frankfurt 1971). Others hold that facts about an ideal agent’s desires, not an agent’s actual desires, ground facts about normative reasons. If any of these restrictions hold, one might argue that Normative Reason Autonomy requires only that normative reasons are determined by informed desires, which are not determined by anything outside the agent. Perhaps one’s desire for x is a function of external factors, although their second-order desire, their desire-to-desire x, is not a function of external factors.

This move will not work for at least two reasons. First, plausibly, this reply assumes compatibilism. The second-order desires are autonomous (within one’s control) by being internal mental states, while the first-order desires are not within one’s control but a function of external circumstances. However, compatibilism is disputed, and therefore not difficult to avoid for incompatibilists and determinists. Second, Mark Murphy has persuasively argued that a desire-based theory is rightly understood as a simple unrestricted subjectivism according to which, if one desires x, then one has a normative reason to act to satisfy one’s desire for x, as there is no adequate rationale for informed desire limitations (Murphy 2001).

Even if subjectivism can be limited to informed desires, the same problem arises. Suppose informed desires are second-order desires. In this case, one is autonomously accountable to God just in case one obeys God because one has a second-order desire: to have the desire to obey God. Nevertheless, there still exists a gulf between autonomy and normative reason. If one is led by second-order desires, then what role does reason have to play?

Alternatively, suppose that the fact that S has a normative reason to φ is grounded in facts about an idealized version of S, S₁. Consequently, one is autonomously accountable to God just in case they decide to obey God because an idealized version of them would desire to obey God. The same problem as before arises: any desire of S₁ is a function of factors external to S. One decides to obey God, not because of their desire, but because an idealized version of them would desire to obey God. However, one is never in control of the desire that an idealized version of them would have. This fact undermines personal autonomy and, thus, normative reason autonomy. Therefore, the autonomy objector must reject SNR-Autonomy for an alternative model of normative reason.

Lastly, consider the assumption that one should pursue Epistemic Autonomy, without which AOT cannot get off the ground. On subjectivism (SNR-Autonomy), one has a normative reason for pursuing Epistemic Autonomy just in case one is fulfilling their mental state ‘desiring to pursue Epistemic Autonomy.’ How does this motivate a conflict between DCT and Epistemic Autonomy? What follows from the fact that God’s commands conflict with one’s desire to have Epistemic Autonomy? That one’s desire is frustrated by normative facts grounded in facts about the normativity of God and his commands does not entail that one has a right to fulfilling one’s desires, including one’s desire for Epistemic Autonomy. The use of desire in Normative Reasons Subjectivism deflates AOT to a mere desire-based objection. However, that one desires Epistemic Autonomy to seek one’s normative reasons for obeying God does not entail or make plausible that one ought to fulfill that desire or that one has a right to disobey God’s commands. The connection between Normative Reason and Epistemic Autonomy required by AOT is not secured by Normative Reasons Constructivism. It provides no reason for which God must provide Epistemic Autonomy for the sake of our seeking and discovering the normative reasons for obeying God. Therefore, Normative Reasons Subjectivism is incapable of supporting AOT.

3.2. Constructivism about Normative Reasons Autonomy

Like subjectivists, constructivists hold that normative facts originate in certain mental states (attitudes or desires) (see e.g., Markovits 2014; and Street 2009). Unlike subjectivism, the fact that S has a particular mental state attitude does not by itself ground the fact that S has a normative reason to φ. Instead, for a normative reason to be correct is, in the words of Sharon Street, “for it to stand up to the specified sort of reflective scrutiny; the normative judgment’s correctness is constituted by the fact that it withstands this scrutiny” (Street 2008, p. 209). More precisely:

Normative Reasons Constructivism: The fact that an agent S has a normative reason to φ at time t is grounded in facts about the rational evaluation of the attitudes and activities of S or S’s community toward φ at time t.

The constructivist model grounds facts about normative reasons in facts about the rational responses of one or more human beings. In this way, the standards of normative reasons are procedurally self-authenticating.

The relationship between constructivism about normative reason and autonomy seems straightforward.

Constructivist Normative Reasons Autonomy: (CNR-Autonomy): An agent S is autonomous with respect to φ at time t, iff, the fact that S has a normative reason to φ at time t is grounded in facts about the rational evaluation of the attitudes and activities of S or S’s community toward φ at time t.

On CNR-Autonomy, what one ought to do at t supervenes on the rational evaluation of one’s attitudes or those of one’s community at t. This includes whether or not one ought to pursue Epistemic Autonomy.

At this point, one might reply that constructivism is not committed to a specific theory of autonomy, much less one that ties normative reasons to autonomy. However, I am concerned with the conjunction of constructivism about normative reason and the autonomy objection, which requires a close connection between normative reason and autonomy.

Problems for Constructivist Normative Reasons Autonomy

Constructivism comes in two forms: social and individual. Social constructivists see normative facts as grounded in facts about the attitudes and activities, the values of society, and facts about society’s circumstances, both of which are determined by the rational decision procedure of a society. Consequently, one is autonomously accountable to God just in case one decides to obey God because one’s society has determined that it values obedience to God by their chosen rational decision procedure.

It is not clear how social constructivism is compatible with Normative Reasons Autonomy. Normative reasons grounded in procedural produced agreements are binding only on those participating in the agreement. However, suppose facts about normative reason are grounded in facts about one’s mental state and attitude toward agreeing to socially constructed normative facts. In that case, we are back to a subjectivist view.

Alternatively, if the aforementioned is not the case, that is, if facts about normative reasons are binding because they are constructed by society, then personal autonomy and normative reasons autonomy are lost. What one ought to do, and the normative reasons one has for doing so are determined not by the individual but by one’s society. For example, in defending social constructivism, David Copp recognizes that “in some cases, cultural or social acceptance is needed for a person to have a sense of self-respect, and in different cultures, different things contribute to social acceptance.”11 These “background facts” are imposed on the individual by their society. Thus, like subjectivism, there is no close connection between normative reasons and autonomy.

However, will not the constructivist say one can rationally reflect on the societal agreement and judge that one should keep it? Therefore, the agreement comes from others, but normative reason tells one that they ought to keep it. This will not evade the problem altogether. Suppose morality holds because one ought to agree to live by the socially constructed agreement. In that case, at least one normative fact precedes the agreement, one that is not constructed; namely, that one ought to keep the agreement, which contradicts social constructivism (Copp 1995, p. 167).

Alternatively, individual constructivists see normativity as created by each person, perhaps to avoid the previous problem. Some follow Kant’s view that normative reason is a formal principle—a categorical imperative—allowing us to govern our inclinations, deciding which must be followed, which must not be followed, and which are permissible but not required, which apply without reference to adopting any ends or goal for ourselves. One evaluates proposed principles of action by discerning which can rationally be willed as universal. In this way, normative facts arise from a free rational will standing under the categorical imperative. However, whether such a formal principle can provide concrete guidance is unclear. Can one will a universal law that she should not step on sidewalk cracks? Probably not, but only because we recognize no value in such a principle. Formal reason itself does not tell us this.

Alternatively, Christine Korsgaard proposes that individual constructivists supplement Kant’s categorical imperative by adding to it a principle of personal identity. We must choose what she calls the moral law—what Kant calls a kingdom of ends—a maxim we must give ourselves regarded as a law according to which we act only on maxims that “all rational beings could agree to act on together in a workable cooperative system”(Korsgaard 1996b, p. 99; see also Kant [1785] 1997, 4: 433). Facts about the principles that one can rationally will as a universal law are partially grounded in facts about one’s sense of who they are. Korsgaard explains:

But part of the normative force of those reasons springs from the value we place on ourselves as human beings who need such identities. In this way, all value depends on the value of humanity; other forms of practical identity matter in part because humanity requires them. Moral identity and the obligations it carries with it are therefore inescapable and pervasive.

(Korsgaard 1996b, pp. 21–22)

Therefore, normative reason is constructed. Facts about normative reason are grounded in facts about the value of humanity, and facts about the value of humanity are grounded in facts about our assignment of value regarding our individual sense of identity. However, this makes morality subjective since people have wildly different senses of identity. Think, for example, of the differences between the white nationalist and the individual whose life is rooted in opposition to racism.

Korsgaard tries to limit the subjectivity. Constructivists typically characterize their view as entailing qualified anti-realism or what Christine Korsgaard calls procedural realism: “values are constructed by a procedure, the procedure of making laws for ourselves” (Korsgaard 1996b, p. 112). As Julia Markovits and Kenneth Walden explain, normative facts exist, but they do not exist independently of the parts of which they are constructed (Markovits and Walden 2021, p. 318). Normative facts are, therefore, not sui generis. In agreement with realists, constructivism affirms that there are normative facts, properties, and relations and that normative facts can be known. This is compatible with subjectivism. Moral facts are determined by mental states—desires—and one can know moral facts by knowing one’s desires. However, like the subjectivist, the constructivist denies that normative facts are mind-independent. As Sharon Street explains:

Normative truth, according to the constructivist, does not outrun what follows from within the evaluative standpoint, but rather consists in whatever is entailed from within it.

(Street 2010, p. 371)

Still, Korsgaard tries to limit the subjectivity of constructivism by arguing that every reasonable person’s identity should include recognizing the value of every human person. Thus, according to Korsgaard, if one is to value one’s humanity, one must value the humanity of all human persons. However, this implies that humans have an objective, mind-independent value independent of normative constructions, which contradicts constructivism.

At the same time, Korsgaard tries to avoid positing an objective value for human persons. In her view, one necessarily values one’s self. So to be rationally consistent, one should value others like one’s self. But why be rationally consistent? In Korsgaard’s view, the will must have a law, but it is its own law, so “nothing determines what that law must be. All that it has to be is a law” (Korsgaard 1996b, p. 98). Thus, no one should be rationally consistent in an objective sense. The collapse into subjectivism is unavoidable.

Moreover, in Korsgaard’s view, autonomy is the source of all normativity (Ibid., p. 91; Korsgaard 1996a). However, if autonomy is the source of all normative value, then the autonomy objector cannot explain why human autonomy is valuable in the first place.12 The autonomy objection cannot be defended or even motivated by constructivism. Consider the AOT assumption that one ought to pursue Epistemic Autonomy. On Normative Reasons Constructivism, one has a normative reason for pursuing Epistemic Autonomy just in case the rational evaluation of the attitudes and activities of one’s community decides so. Normative Reasons Constructivism, therefore, deflates AOT to an objection from the normative constructions of a community. However, that one’s community constructs normative reasons that presumably conflict with God’s normative commands does not entail that one’s community and its citizens have the right to secure Epistemic Autonomy. That is, Normative Reasons Constructivism provides no reason for which God must provide Epistemic Autonomy for the sake of our seeking and discovering the normative reasons for obeying God. Therefore, Normative Reasons Constructivism is incapable of supporting AOT.

4. The Real Issue: Realism about Normative Reasons

Now I will argue that the real issue driving AOT is moral realism, specifically realism about normative reasons. This shows that moral realism about normative reason motivates AOT. Therefore, what was thought to be a problem for divine command theory is a problem for all moral realist theories or, at least, all realist theories about normative reasons.

4.1. Moral Realism

According to Moral Realism: Facts about normativity are (a) grounded in objective, mind-independent facts and (b) apply universally (to rational beings). Moral Anti-Realism rejects (a) and (b). Ralph Wedgwood points out that the fundamental questions about normative reason concern the dispute between the realist or what he calls the “recognitional” view and the constructivist. On realism, normative reasoning “fundamentally consists in attempting to figure out what is a good thing to do and what is not and then choosing accordingly” (Wedgwood 2002, p. 139).

Applied to normative reasons, we get the following:

Normative Reason Realism: Facts about S’s normative reason to φ at t are (a) grounded in objective, mind-independent facts and (b) apply universally (to rational beings).

Normative Reason Anti-Realism: Facts about S’s normative reason to φ at t are not (a) grounded in objective, mind-independent facts and (b) do not apply universally (to rational beings).

Normative reasons, on a realist view, are grounded in normative facts, while on an anti-realist view, they are grounded in procedural facts (constructivism) or desire facts (subjectivism).

It must be understood that moral realism does not stand at odds with autonomy but is only a theory of autonomy, according to which facts about the value of autonomy are grounded in mind-dependent facts. The self-referential aspect of autonomy and authenticity, in the words of Charles Taylor, “doesn’t mean that on another level the content must be self-referential: that my goals must express or fulfill my desires or aspirations, as against something that stands beyond these” (Taylor 1991, p. 82). Failure to understand this, in the words of Taylor, “lends legitimacy to the worst forms of subjectivism” (Ibid.)

4.2. Why Moral Realism Is the Actual Problem

I have argued that the most popular anti-realist models of normative reason cannot plausibly support AOT. Formally, these anti-realist views appear to support AOT insofar as they view normative rationality and autonomy as merely instrumental; no facts about authority, divine or otherwise, ground facts about one’s normative reasons. However, as I argued in Section 2, in their specified material form (for example, subjectivist and constitutive models), the anti-realist views fail. Consequently, realist models of normative reason are all that remain to formulate AOT. This is enough to raise the AOT Dilemma. However, first, I want to raise two further points that show why Moral Realism is the actual problem.

4.2.1. Virtual Reality, the Simulation Hypothesis, and Accountability to God

Here is another way to clarify my point. I want to argue that David Chalmers’s argument about the worship-worthiness of a virtual god shows that AOT is likewise a debate that assumes moral realism. Chalmers offers the following as one reason for his atheism:

Even if the Abrahamic God exists, with all those godlike qualities of perfection, I will respect, admire, and even be in awe of him, but I won’t feel bound to worship him… I don’t think any qualities can make a being worthy of worship. As a result, we never have good reason to worship any being. No possible being is worthy of worship.

(Chalmers 2022, p. 144)

Setting aside the claim that no qualities can make a being worthy of worship, I will focus on the metaethical theories working behind the scenes in Chalmers’s reasoning.

Now, Chalmers’s argument is couched within virtual realism and simulation realism, that is, his view that virtual reality and the objects of a computer simulation are just as real as non-virtual or physical reality and non-virtual objects.13 He provides five criteria for x being real: x exists, has causal powers, is mind-independent, is non-illusory, and is a genuine x. However, if the virtual is not real, then the simulation is not worthy of worship because things that are not real are not worthy of worship. Without that assumption, the question about worshiping a simulator god is not relevantly different from the question about worshiping God.

More to the point, Chalmers’s argument assumes that facts about one’s normative reasons (specifically concerning worshiping a simulator god) are, at least partially, grounded in facts about reality and the qualities that would make a being worthy of worship. According to Chalmers, one has no normative reason to worship a simulator god because no qualities could make any being worthy of worship. Of course, if a simulation is not real, it could have constructed anti-realist or subjectivist normative facts. However, in that case, Chalmers’s argument will not succeed. I take this to show that debates about normative reasons regarding virtual reality are analogous to debates about normative reasons concerning DCT. Both bottom out in the debate over realism and anti-realism.

4.2.2. Realism and Normative Constraints, and Normative Reasons Autonomy

“The object of systems of morality”, writes Matthew Arnold, “is to take possession of human life” (Arnold [1865] 1902, p. 344). The reality of objective moral truths constrains human autonomy insofar as one cannot determine (subjectively or constructively) the structure of normative reasons or the conditions under which one rightly seeks and obtains knowledge of one’s normative reasons. We are born into this situation. We come into the world without consent to the nature of normative reasons and without knowing our normative reasons. Facts about normative reasons and the nature and attainment of moral knowledge are not grounded in facts about one’s mind or the construction of cooperating minds. As with DCT, moral realism entails that moral knowledge, including knowledge of our normative reasons, requires submission to how normative facts present themselves in reality.14

The ethical demand of reality captured by moral realism entails significant constraints on the normative status of human autonomy. Constructivist Sharon Street understands this well. Street rightly criticizes non-theistic moral realism as resting on “nothing more than an unreasoned faith, with realism about reasons and value, thus becoming a rather odd form of religion” (Street 2016, p. 299). Street understands that it is not theism or DCT alone that places limits on human autonomy but moral realism itself.

5. Conclusions: The Real Dilemma

AOT is not a problem unique to theism but a problem for all moral realists. Moreover, as pointed out in the Introduction, there are plenty of atheists who are moral realists. The following are a few more. Similarly, Russ Shafer-Landau writes:

The laws of logic and rationality are normative. They tell us what we ought to do. But no one invented them. If you have excellent evidence for one claim, and this entails a second claim, then you ought to believe that second claim. If you are faced with contradictory propositions, and know that one of them is false, then you must accept the other… If you are an atheist, you’ll deny that God made up such principles. If any principals are objective, these are. So we have here objective, authorless, normative laws.

(Shafer-Landau 2004, pp. 77–78)15

In explaining the objectivity of morality (moral realism), Walter Sinnott-Armstrong writes:

If what makes an aggressive war morally wrong is that it hurts innocent people, then whether it is wrong does not depend on my desires, such as whether I want to harm those people. It also does not depend on my beliefs, such as whether I believe that the war will hurt those people…Thus, atheists and agnostics can hold not only that there are moral facts but also that these moral facts are objective rather than subjective.

(Sinnott-Armstrong 2009, p. 75)

Consider also how David Enoch explains one’s epistemic access to the correlation between normative truths and normative judgments from a moral realist view:

I argue that the correlation that needs to be explained is not as striking as it seems, and that whatever by way of correlation does need explaining can be explained consistently with Robust Realism, by a godless (and so speculative evolutionary) pre-established harmony kind of explanation.

(Enoch 2011, p. 13)16

Clearly, at least some atheists embrace moral realism.17

Here is a final word from Wielenberg: “The foundation of morality is a set of axiomatic necessary moral truths. No being, natural or supernatural, is responsible for the truth of or has control over these ethical truths” (Wielenberg 2005, p. 66; see also, Wielenberg 2009). The autonomy restrictions on normative reasons are, according to Wielenberg, etched into the necessary moral truths of reality. Thus, the non-theistic moral realists find themselves, just as the theist does, in a universe iconoclastic to our normative reasons autonomy.

The AOT Dilemma remains. If one is to advance AOT, the most plausible models of normative reason are anti-realist, that is, subjectivism and constructivism. Yet, as I argued in Section 2, these anti-realist theories are too weak and inadequate to link autonomy to normative reason, much less in a way that supports AOT. However, as I argued in Section 3, if the autonomy objector adopts moral realism about normative reasons, they face a nearly identical problem as the theist, as normative fact constrains autonomy. Therefore, one can preserve normative facts and lose AOT or keep AOT and lose the objective goodness of autonomy and a place to press AOT. Either way, the Autonomy Objection to Theism (AOT) collapses.