Divine command theory and the (supposed) incoherence of self-commanding

Abstract

Theological voluntarism is a family of metaethical views that share the claim that deontological statuses of actions are dependent on or identical with some divine feature. Adams's version of this theistic metaethical view is a divine command theory (DCT). According to Adams's DCT, the property being-morally-obligated is identical to the property being-commanded-by-God. Thus, a natural consequence of Adams's DCT is that an agent is morally obligated to do something just in case God commands that agent to do such a thing. From Adams's DCT, it follows that God is morally obligated to act just in case God commands himself to act. Quinn argued that commanding oneself is incoherent and, therefore, that God cannot be morally obligated to act. The claim that commanding oneself is incoherent has seldom been discussed in the divine command theory literature. This article is an attempt to change that. Here, I argue (contra Quinn) that that no constitutive rule (or condition of satisfaction) of self-commands is incoherent, from which it follows that self-commanding is not an incoherent speech act. I conclude that divine command theorists can, without the charge of incoherence, affirm that God can be morally obligated because God can command himself.

Introduction

Since the latter half of the past century, there has been among theists a revival (Quinn 1990b) of an old theistic metaethical view, namely, theological voluntarism. According to this view, deontological statuses of actions (obligation, permissibility, prohibition) are dependent on or identical with some divine feature. One well-known version of theological voluntarism is Adams's (1979, 1987a, 1987b, 1999) divine command theory (DCT for short), which is the view that deontological statuses of actions are identical to divine commands. Thus, according to Adams's view, the property being-morally-obligated is identical to the property being-commanded-by-God. Adams's theory implies, then, that for any moral agent, that agent is morally obligated to act just in case God commands her to act.

Given this theory of moral obligation, it is natural to ask: does God have moral obligations given the truth of divine command theory? Philip Quinn, a DCT-ist, says we must answer negatively. For given DCT, God would be morally obligated to act iff God commanded himself to act. But, Quinn says, commanding oneself is incoherent; it does not make any sense to command oneself. Therefore, not even God can command himself since being omnipotent does not empower a being to do what is logically impossible. Hence, given the truth of DCT, we must conclude that God cannot be morally obligated to act.

It has been seldom, if at all, discussed in contemporary literature on DCT whether Quinn's claim that commanding oneself is incoherent is plausible. My impression is that it is very prima facie plausible to believe that commanding oneself is an odd and weird concept. However, to my knowledge, no DCT-ist has offered an explicit argument for the conclusion that commanding oneself is incoherent or that God cannot command himself. Thus, my goal here is to defend the claim that commanding oneself (or self-commanding) is not an incoherent notion. I will, therefore, also argue that a DCT-ist can, without the charge of incoherence, accept both DCT and the proposition that God can be morally obligated to act precisely because God can command himself.

Our discussion proceeds as follows. First, I present Quinn's argument that God cannot be virtuous nor be morally obligated. Second, I argue that commanding oneself is not an incoherent notion, giving special attention to whether we can have authority over ourselves. Third, I generalize the conclusion arrived at previously and argue that the concept of authority over oneself is necessary for moral theorizing. In the penultimate section I address some objections and conclude in the final section.

Quinn on the incoherence of commands to oneself

Many theists claim that the property being morally obligated is either identical to or dependent upon the property being commanded by God. This view is called divine command theory (DCT). Adams (1999, 252–258) claims that these two properties are identical.¹ Thus, on the face of it, DCT entails that:

(P) for any subject S and action φ, S is morally obligated to φ iff God commands that S φ.

In fact, Quinn (1978) developed a formal system where P is an axiom. Quinn writes:

We can inject theological doctrine into the deontological doctrine formulated so far by equating what is commanded by God with what the moral law makes obligatory. Let ‘Cp’ abbreviate ‘God commands that p’. The appropriate axiom will then be the following: Cp ≡ Lp [where ‘Lp’ abbreviates ‘it is morally required that p’]. (76)

If the quantifier expression ‘for any subject S’ ranges over all subjects that are moral agents, then it follows that God is morally obligated to φ iff God commands that God φ.² In other words, DCT entails that God has moral obligations just in case God commands himself to act.

Some DCT-ists have reasons to reject the view that God has moral obligations. These reasons can be grounded upon considerations that are independent of DCT. For instance, Alston (1990, 304–316) argues that if God has moral obligations, then God is not morally perfect. If Alston is right, then a DCT-ist might claim that P is true but since one proposition connected by the ‘iff’ in P is false, the other proposition (viz., that God commands himself) is also false.

Philip Quinn, however, takes the proposition ‘God commands himself’ to be false and therefore affirms that God does not have moral obligations since he thinks P is true. He even uses the proposition that God cannot command himself as a premise in his argument that God cannot be virtuous. The argument runs as follows. One may think that a DCT cannot accommodate the fact that humans have virtues. This is false, according to Quinn. If we take virtues to be dispositions to act in certain ways, then virtues – given the truth of DCT – would be dispositions to obey divine commands. For instance, S would be just if S had a disposition to obey those divine commands that require S to act justly. Generalizing this point, Quinn notes that, provided DCT is true, the fundamental virtue is obedience to God. If S obeyed all divine commands, S would live a complete virtuous life since being virtuous is to have the disposition to obey God's commands (Quinn 1978, 130).

Now, given DCT, can God be virtuous in the sense explained above? Quinn answers negatively. He offers the following argument.³

1. 1. S is obedient to God in doing φ = _df S and φ are such that God commands that S φ and S's belief that God commands that S φ is S's reason for φ-ing. [definition]

2. 2. S exercises the moral virtue of character trait M in doing φ = _df φ manifests M and S is obedient to God in doing φ. [definition]

3. 3. For all S and for all M, if S has the moral virtue of M, then S and M are such that it is logically possible that, for some action φ, S exercises M in φ-ing. [premise]

4. 4. It is not logically possible that, for some action φ, God commands that God φ. [premise]

Therefore,

1. 5. It is not logically possible that God obeys himself. [from 1, 4]

Therefore,

1. 6. It is not logically possible that some action of God manifests the virtue of M, for all M. [from 3, 5]

Therefore,

1. 7. Necessarily, God is not virtuous. [from 6]

Premises 1 and 2 are uncontroversial since they are stipulated definitions given by Quinn. Premise 3 is plausible since it can be regarded as a version of an ought-implies-can principle. So, this argument rests on the truth of 4.

It is unfortunate, however, that Quinn does not give an explicit argument for 4. Rather, he dismisses the possibility of self-commanding because of its oddness and unintelligibility. He writes:

But, within [the framework of DCT], the divine command theorist does not seem to be able to speak coherently of God having moral virtues. For it is very odd, and perhaps unintelligible, to suppose that God, or anyone else for that matter, commands himself to do certain things and then obeys the commands he has addressed to himself. If a certain man is captain of a ship, then the crew is under his command, and he may command them and they must obey him. But he does not command himself. (Quinn 1978, 130, my italics)

Quinn seems to be arguing that self-commanding is incoherent because it is odd and unintelligible. Because of this, the DCT-ist cannot speak coherently of God having moral virtues since he believes that, under DCT, to have a virtue is to (have a disposition to) be obedient to the commands of God. But since one is obedient just in case one follows the commands of a commander, and God cannot command himself, it follows that God cannot be obedient to himself, and hence that he cannot be virtuous. Quinn continues:

If God goes ahead and does what he tells himself to do, then he is not obeying a command which he has addressed to himself; instead he is simply doing what he has decided to do. And, if God does something other than what he tells himself to do, he is not disobeying a command he has addressed to himself; rather he has changed his mind and is only, after all, doing what he has finally chosen to do. (Quinn 1978, 131)

Quinn, then, thinks that no-one – including God – can command oneself. With this claim and other auxiliary premises, he concludes that God cannot be virtuous. Moreover, it follows that God cannot have moral obligations, because S has a moral obligation iff God commands S to act. The moral of the story is that if DCT-ists are going to accept Quinn's arguments, they should be ready to commit to the conclusions above.

Some theists believe that God can (and in fact does) have moral obligations.⁴ In fact, Abrahamic religious texts give the impression that God has obligations since he has made covenants and promises with certain people. There is, however, an obvious tension for those who, like me, want to entertain DCT and the possibility of God having moral obligations. For it seems that the only way to hold both views is by rejecting 4, that is, by claiming that God can command himself. Though this seems a radical move, I think it is viable. To my knowledge, no DCT-ist or philosopher discussing DCT has addressed the issue of self-commanding, perhaps because of Quinn's huge influence in the DCT literature and the prima facie plausibility of the claim that self-commanding is incoherent.⁵ In what follows I try to show that Quinn's claim that self-commands are incoherent is false and, therefore, show that DCT-ists can hold both DCT and the claim that God can have moral obligations because God can command himself.

Speech acts and commands to oneself

Self-commands and incoherence

Note that Quinn grounds his appeal to incoherence in the ‘oddness’ and ‘unintelligibility’ of self-commanding. He also offers examples that (presumably) show that commanding oneself is incoherent. But giving an example and just stating that in such an example no self-commanding has taken place is hardly an argument. The same goes for the appeal to its oddness and unintelligibility. Many philosophical theories strike one as odd but that is not a reliable indicator that the theory is incoherent. Moreover, unintelligibility may well say something about us and not the theory, namely, that we are not in a position to understand it, just as theists say that we are not in a position to understand God's nature, or as physicists say that we are not in a position to understand the fundamental structure of the universe. I submit, then, that Quinn's arguments are, at best, inconclusive. Despite this, the consensus among theists who endorse DCT seems to be that Quinn is correct. Against this consensus, I will defend the claim that self-commanding is not incoherent.

What do we mean here by ‘incoherent’? At first it is hard to say. We can say that a proposition is incoherent just in case it is a self-contradiction, or it entails a self-contradiction. This is not completely right. The proposition ‘some triangles have four sides’ is incoherent, but it is not a self-contradiction, nor does it entail one. The sentence ‘some triangles have four sides’ is incoherent because it is conceptually impossible. Let us say that a proposition is conceptually impossible just in case substitution of synonymous or analytically equivalent terms results in a contradiction. In the sentence ‘some triangles have four sides’ the word/concept ‘triangle’ is synonymous with ‘closed three-sided two-dimensional shape’. If we substitute the latter with the former, we get a contradiction. So, let us say that a proposition is incoherent iff it is self-contradictory, it entails a self-contradiction, or it is conceptually impossible.

We still have one problem. What is it for the concept commands to oneself to be incoherent? Self-commands are a species of the genus ‘commands’. In turn, commands are speech acts and, certainly, speech acts are not mere propositions. So, what do we mean when we say that a speech act is incoherent? Green defines speech acts as ‘those acts that can (though need not) be performed by saying that one is doing so’ (2021, 4). For example, I can command Ana to study by saying ‘Ana, go study!’ but I can also command her without uttering any words, say, by pointing to her philosophy books while looking at her. In both cases, we may safely assume I have commanded Ana to study, the difference being that in the former case I said that I commanded her while in the latter I commanded her without saying anything.

To perform these speech acts, however, there are certain conditions that need to be the case for one to felicitously perform such an act. For instance, to absolve a person from their sins I need to be a Catholic priest. J. L. Austin (1962) called these conditions felicity conditions (also conditions of success or satisfaction) the violation of which prevents the speech act from being performed. These conditions can be described in declarative sentences that express a proposition. We can then say that a speech act is incoherent iff at least one of its felicity conditions is incoherent in the sense explained above. My claim is that no felicity condition of self-commanding is incoherent. Formally, we can state the argument thus:

1. 8. A speech act, SA, is incoherent iff some SA's conditions of (performative) success are incoherent. [definition]

2. 9. Commanding oneself (CO) is a speech act. [premise]

Therefore,

1. 10. CO is incoherent iff some CO's conditions of (performative) success are incoherent. [from 8, 9]

2. 11. No condition of (performative) success of CO is incoherent. [premise]

Therefore,

1. 12. CO is not incoherent. [from 10, 11]

In what follows I explain premise 11 using Searle's speech act theory and defend it.⁶

Searle (1979, 13) classifies commands as directives which is a specific kind of speech act. When we say, then, that self-commands are incoherent, we are saying that a speech act of a certain kind is incoherent. Furthermore, following Searle (1969, 33–34), we can say that a speech act is the sum of its constitutive rules/conditions which are the following:

• Essential Conditions. A speech act is done for some end or purpose. For example, directives attempt to bring about a desired behaviour of the person directed. Searle (1979, 2–3) also calls these conditions the illocutionary point. If S does not fulfil this condition, the speech act is not performed.

• Propositional Content Conditions. Some speech acts require the utterance of a sentence with a certain form. For example, promises are necessarily about future acts. So, when S promises that p, p needs to be about the future.

• Sincerity Conditions. For speech acts that have propositional content, generally, S must have an attitude towards the proposition contained in the speech act. For example, if S promises that p, then S intends that p. If one doesn't intend what was promised, one has promised insincerely and, therefore, infelicitously.

• Preparatory Conditions. These are non-linguistic features such that, when they hold, they make bringing about a speech act possible. For example, I cannot christen a boat without the relevant authority to do so; I cannot absolve from sins if I am not a priest, etc.

Each kind of speech act is constituted, then, by its constitutive rules/conditions described above (see Searle 1969, 62–71).

Since commands are speech acts, we can say that commands are the sum of its constitutive rules/conditions. Let ‘S’ and ‘H’ refer to a speaker and a hearer, respectively. The constitutive rules of commands are the following.

Essential Conditions → S's commanding H counts as an attempt to get H to do A in virtue of the authority S has over H.

Content Conditions → The commanded action must be a future act A of H.

Sincerity Conditions → S wants H to do A.

Preparatory Conditions → H is able to do A. S believes H is able to do A. S is in a position of authority over H. (Searle 1969, 66)⁷

Note that the constitutive rules are propositions or, at least, expressible as propositions. Thus, let us say that (the concept of) a particular speech act, SA, is incoherent just in case some of its constitutive rules or conditions are self-contradictory, entail a self-contradiction, or are conceptually impossible in the sense defined above.⁸ Also note that self-commanding is a speech act identical to its constitutive rules which are the ones above when we replace each ‘H’ with ‘S’. Thus, Quinn's objection amounts to this claim: some of the constitutive rules of self-commands are incoherent in the sense explained above. I will argue that this claim is false.

The sincerity condition of self-commanding – that S wants S to do A – is perfectly coherent. Many, if not all, people want themselves to act in certain ways. In fact, one might just say that sentences of the form ‘I want to do A’ express the proposition that the utterer wants the utterer to do A. The sentence ‘I want to eat’ entails ‘someone wants to eat’, and that someone is me. So, if I say ‘I want to eat’ that just means that I want that someone, namely myself, eats. Thus, the sincerity condition is not incoherent.

I can't see that the content condition leads to a contradiction. The restriction that the content condition introduces to commands is that the commanded action must be a future action. That is, the propositional content expressed by the command must be of the form S will do A. This seems to be true even if, ex hypothesi, I self-command. Certainly, I cannot command myself to do something that I already did or that I am doing in the present. But the fact that someone else or myself commands me to do A does not lead to any incoherent result. Therefore, the content condition does not show that self-commands are incoherent.

The two prospects for deriving an incoherent result from self-commanding are the preparatory condition and the essential condition. I think we can dismiss the essential condition thusly. Replacing ‘H’ for ‘S’ in the essential condition we get ‘Counts as an attempt to get S to do A in virtue of the authority S has over S’. Let us ignore for the moment the part of having authority over oneself – since that is what the preparatory condition says – and let us focus on attempting to get S to do A, namely, attempts to get oneself to do something. Is this incoherent? No. I recall myself going to an amusement park and wanting to ride a gigantic rollercoaster. But I was so afraid that I was not able to ride it. In this case, it is extremely plausible to say that I was attempting to get myself to ride the rollercoaster. This is ubiquitous in the lives of many! People try to get themselves out of bed, break their lifting record, write more, not fall into temptation, etc. All these cases are instances of trying to get oneself to do something. Therefore, attempting to get oneself to do something is far from being incoherent.

Self-commands and authority

I have shown that the content, preparatory, and essential conditions of self-commanding are not incoherent. What about the preparatory condition? If any condition looks suspiciously incoherent, it is this one. I argue that not even the preparatory condition of self-commanding is incoherent. Let us turn to this issue next.

The preparatory condition of self-commanding is ‘S is able to do A. S believes S is able to do A. S is in a position of authority over S’. There are three parts of which the first two are obviously not incoherent. The fact that I am able to do A does not entail a contradiction or a conceptual impossibility, nor does the fact that I believe myself able to do A. It is well-established that I can have a doxastic attitude about myself, for instance, I believe that I exist. Nevertheless, the third part of the preparatory condition is not innocent. For it is not obvious at all that we can have authority over ourselves. In fact, the opposite seems to be the case since authority seems to require a distinction between the authoritative figure and the subordinate figure. Put differently, authority seems to require that x be authoritative over a subordinate y; but since one and the same individual cannot be authoritative and subordinate at the same time, in the same respect, then x and y must be distinct.

To show that authority is or is not incoherent, we first need an account of that authority is. After we have understood what authority is, we can assess if, under that understanding, having authority over oneself is incoherent. In what follows I consider three accounts of authority and argue that self-commanding is coherent given their truth.

Authority as constitutive actualization

According to Murphy's (2002, 8–19) account, ‘[H] has authority over [S] iff [S's] commands actualize a reason for [H] to φ; the [propositional] content of the command and what the command actualizes a reason to do must be identical; the commands must be themselves parts of the reason for action actualized’ (Murphy 2002, 15). Put more simply, one has authority when one's dictates and commands actualize reasons for action. This does not mean that my command is both necessary and sufficient to actualize reasons for action. For Murphy, reasons must be complete, that is, ‘a fact must include within it all that makes an action choiceworthy in some way if that fact is to count as a reason to perform that action’ (Murphy 2002, 9). A reason, R, may be, then, written out as a long conjunction r ₁ & r ₂ & . . . & r _n the conjuncts of which are those things that make an action choiceworthy, one of them being that S commands H to do the action for which R is a reason.

Let us consider the reflexive case where S =  H. By replacement in Murphy's schema above we get:

• S has authority over S iff (i) S's commands actualize a reason for S to φ; (ii) the [propositional] content of the command and what the command actualizes a reason to do must be identical; (iii) the commands must be themselves parts of the reason for action actualized.

The only part that references the commands of the agent is condition (i). Is (i) incoherent? I do not think so. First, it is intuitively plausible that commanding myself to behave in a certain way actualizes a reason for me to behave in that way. It is less intuitive that this is so from one agent to another. This is because we have intuitions that moral agents are autonomous. Second, the first-person perspective seems to give us ‘special access’ to ourselves so that it gives us primacy over what we do. Third, we do things with our bodies. And our bodies are our own. These three points make plausible the claim that if authority is constitutive actualization, then self-commanding is one, if not the one, practice that has support from this theory.

Here's an illustration that puts these points in context.⁹ Suppose Jones suffers from a disease that exercise will help eliminate. Jones goes to his doctor and the doctor confirms that some exercise will help him. Jones has the following reasons to exercise: exercise being good for Jones's health, the doctor prescribing an exercise routine, the doctor's being an expert on health and medicine, Jones's desiring to be healthy, and Jones's dictating himself to do things to promote his health. Note that if Jones commands himself to do things that promote his health, it is plausible to assume that, under Murphy's account, Jones has an obligation to do such things. But if Jones did not command himself to do so, it is also plausible to say that he was doing something desirable, rational, etc. In the former case, if I fail to do the things to promote my health Jones has failed to give what he owed himself; in the latter case, if Jones does not do these things, he does not fail to give what he owes himself. In other words, Murphy's account gives us the tools to accommodate self-commanding as a way in which we actualize reasons that generate obligations. There is nothing in this account that excludes this possibility. Hence, under Murphy's account, self-commanding is not incoherent.

The service account of authority

Raz (1985, 1990, 2005) account of authority attempts to capture what happens in cases where one has authority over the will of another. Although Raz is not explicit about what he takes authority to be, we can infer his view when he presents the problem of justifying authority over others. For instance, presenting the problem of justification Raz writes that ‘[t]he moral question is how can it ever be that one has a duty to subject one's will and judgment to those of another?’ (2005, 1012, my italics). For Raz, then, authority is subjecting one's own will and judgement to the will and judgement of another person. He then tries to capture when it is justified to submit one's own will to another's will with the following criteria: practical authority [submitting one's own will to the will of another] is normally justified if (a) the subject is likely to conform better with reasons which already, independently, apply to her if she intends to accept the authority's directives and treat them as valid; and (b) the relevant situation is one where it is better that the agent conforms with the reasons that externally apply to her, rather than decide for herself what to do without the aid of an authority (Raz 2005, 1014). One example given to purport this point is the following.

[A] person driving down a highway might, very plausibly, be thought to have a preemptive reason to drive no faster than a sign-posted speed limit, where this reason also speaks against her attempting to independently determine the fastest speed at which it might be safe for her to drive. (Star and Delmas 2011, 147)

Star and Delmas note that:

Very plausibly, such a person better conforms with independent reasons that already apply to her in the long term by treating the speed limit as authoritative than she would if she considered nonauthority-based reasons, with respect to the issue of determining the fastest speed at which she might drive. (Star and Delmas 2011, 147)

In what follows I show that Raz's account of the justification of practical authority together with his account of the nature of authority presupposes that we have authority over ourselves.¹⁰

Raz's first condition is stated in a way that does not allow a reflexive reading of the conditions for authority. However, one might think about it like this. Suppose there is no distinct authority – i.e., an agent different from me that issues certain commands – such that I would be likely to conform better with reasons which already apply to me if I intended to accept the authority's directives and treat them as valid. In such a case, I will better conform to reasons that apply to me if I accept my own directives and treat them as valid or the situation would be equally good, that is, it would be indifferent whether I take the directives of a distinct authority or of myself. The same reasoning applies to the second condition. Suppose that it is false that the relevant situation is one where it is better that I conform with the reasons that externally apply to me, rather than decide for myself what to do without the aid of a [distinct] authority. This would imply that it is either better or equally good for me to decide for myself what to do without the aid of a distinct authority.

What this shows is that Raz's analysis presupposes authority over oneself. If (a) is not satisfied, then it may be because it would be better or equally good to follow my own directives. If (b) is not satisfied, it would be better or equally good to decide for myself what to do. But this amounts to giving oneself directives to act, that is, self-commanding. In fact, one puzzling aspect of authority is that sometimes it is justified (or even required) to give away the authority we have over ourselves (at least partially) to others, for instance, the state, the teacher, etc. So, far from suggesting that authority over oneself is contradictory, it suggests that the puzzling aspect of authority rests precisely upon the fact that sometimes we give away the authority we have over our own wills to the will of another! In other words, the fact that we are sometimes justified to submit our will to the will of other shows that, when it is not justified, it would be better or equally good to determine our own wills and follow our own judgements, which means following one's own directives and submitting to our own judgements and reasons.

The second-personal account of authority

According to Darwall (2013), for S to have (practical) authority over H is for S to have a standing in a relation with H. This ‘standing’ involves S's valid claims or demands towards H, H's having a (second-personal) reason to comply with S's valid demands, and H's being accountable for not complying with S's demands. In the words of Darwall: ‘[S] has practical authority with respect [H] if, and only if, [H] has a second personal reason to comply with the [S's] valid claims and demands and is responsible to [S] for so doing’ (Darwall 2013, 141).

Darwall's example makes this clearer. Suppose H is stepping on S's feet causing S pain. According to Darwall's account, H has a second-personal reason to comply with S's demand that H stop stepping on him. But this implies, according to Darwall, that A has the authority to demand of H that H stops stepping on him, giving H a reason to refrain from his wrongdoing and becoming responsible for so doing. The authority one has comes by virtue of the relation we have with others as rational, free, moral agents who are part – and representatives – of a moral community. And this relation involves addressing the other as such, that is, second-personally. Therefore, according to Darwall's account, authority is an irreducibly second-personal concept, that is, it necessarily involves addressing another in a relation between persons that take each other to be rational, free agents.

Whether the second-personal account of authority can be not irreflexive depends upon whether a person can address herself second-personally. I think we can establish that an agent can – and in fact does – address herself second-personally. Two pieces of evidence lead to this conclusion. First, since we are dealing with Darwall's account of authority, we would expect that Darwall's theory is not only consistent but suggests that people address themselves second-personally given that people actually address themselves second-personally.

In his book The Second-Personal Standpoint, Darwall takes competence to be one of the features that make us bound to a moral obligation. There is a conceptual connection between being bound by the demands of another and being capable of demanding ourselves to act according to the other's demands. Here's Darwall:

We cannot intelligibly hold someone to a demand as a moral agent without supposing that he could hold himself to that same demand by acting on the relevant second-personal reasons. And if we think that any rational person in that situation would be thus answerable, we are committed to thinking that what makes a rational person subject to moral obligation must itself include a source of motivation to do as he is morally obligated. (2006, 34, my italics)

Furthermore, in explaining what is this competence that is conceptually required by having an obligation to another, that is, being addressed second-personally by the valid demands of another, Darwall says:

What then does second-personal competence consist in? Well, it must consist in something like the capacity to make demands of oneself from a second-person standpoint: in being able to choose to do something only if it is consistent with demands one (or anyone) would make of anyone (hence that one would make of oneself) from a standpoint we can share as mutually accountable persons. (Darwall 2006, 35)

If, as Darwall says, ‘any second-personal claim presupposes a common competence, authority, and, therefore, responsibility as free and rational’ (Darwall 2006, 21, my italics), then the second-personal account of authority is not only consistent but presupposes that we have authority over ourselves (as does Raz's account)! Darwall, then, seems not just to allow but commit himself to the view that we have authority over ourselves. As the main advocate of the view, this gives us strong reasons to believe that, given the second-personal account of authority, we have authority over ourselves.

The second piece of evidence that we find for the claim that people can – and often do – address themselves second-personally is the nature of reactive attitudes. A reactive attitude, r, is an attitude S adopts towards H such that S's adopting r entails S's holding H responsible for doing such-and-such. Paradigmatic examples of reactive attitudes are blame, indignation, and resentment. According to Darwall, personal reactive attitudes are essentially second-personal, that is, they involve making a claim or demand the recognition of which gives one a (second-personal) reason to comply with such demands. In the words of Darwall: ‘[R]eactive attitudes are second-personal in our sense, and that ethical notions that are distinctively relevant to these attitudes – the culpable, moral responsibility, and, I argue, moral obligation – all have an irreducibly second-personal aspect that ties them conceptually to second-personal reasons’ (Darwall 2006, 17). Now, it is known that we can adopt reactive attitudes towards ourselves for doing such-and-such. Take the reactive attitude of blame for example. Parents who give unjust punishments to their children, police officers who act recklessly, people who cause terrible accidents while drunk driving are cases where those people can and do blame themselves for doing such things. Thus, it is very plausible that moral agents can and often do adopt some reactive attitudes towards themselves.

But if it is true that people can and do adopt some of these reactive attitudes towards themselves, it seems that this entails that we can and often do have authority over ourselves. For adopting a certain reactive attitude entails having a second-personal reason to comply with the demands made by the person who adopts such an attitude. In fact, Darwall explicitly says that ‘Reactive attitudes invariably concern what someone can be held to, so they invariably presuppose the authority to hold someone responsible and make demands of him’ (Darwall 2006, 17, my italics). It follows directly that – under Darwall's account – when we adopt such an attitude towards ourselves, we presuppose the authority to hold ourselves responsible and make demands of ourselves. Hence, given the second-personal account of authority, having authority over oneself is not incoherent.

Self-authority as necessary in moral theorizing

I have taken three accounts of practical authority and shown that, given each account, having authority over oneself is not incoherent. In fact, the Service and the Second-Personal accounts of authority presuppose that we have authority over oneself. This gives us strong reasons to believe that having authority over oneself is not incoherent and, therefore, commanding oneself is not incoherent. In this section I explain the reason why two of the three accounts presuppose authority over oneself. By explaining this I will be also explaining the reason why authority over oneself is very unlikely to be an incoherent notion in any theory of authority.

Star and Delmas (2011) argue that the puzzling aspects and problems arising from practical authority must be seen from a minimal conception of practical authority, that is, ‘a minimalist conception of practical authority, according to which the problem of authority is the problem of the possible justification of one being subject to directives originating outside of oneself’ (Star and Delmas 2011, 160, their italics). The problem of another will imposing (or creating) pre-emptive reasons over our will becomes a species of a more general problem, namely, justifying the fact that one is subject ‘to directives that originate outside of oneself’ whether or not the source of those directives is another person's will. This is one reason why accounts of authority presuppose authority over ourselves, because subjecting one's will or acting under the directives of another seems surprising given that we are able to direct ourselves to act in the right ways. Why should we become subjects if we started being rulers in the first place? This is something that moral philosophers, especially those sympathetic with Kant, need to explain. But, again, note that here what comes prior to the surprising fact that we (choose to) become subject to directives that are outside ourselves is our wills' not being ruled by others, but being their own rulers. In other words, what makes this fact surprising is the very fact that we can rule ourselves! Thus, there is a relevant sense in which authority over oneself – a will's not being ruled by others by ruling itself – is more fundamental (or prior to) being subjected to directives outside us, especially by those of the will of another. This is why I think authority over oneself is (and perhaps must be) presupposed by any account of authority.

The second reason to believe that authority over oneself is presupposed in by any account of authority comes from Darwall's (2006) second-personal interpretation of Kant's categorical imperative.¹¹ He says:

What then does second-personal competence consist in? Well, it must consist in something like the capacity to make demands of oneself from a second-person standpoint: in being able to choose to do something only if it is consistent with demands one (or anyone) would make of anyone (hence that one would make of oneself) from a standpoint we can share as mutually accountable persons. But that is just a second-personal version of the [categorical imperative]. (Darwall 2006, 35)

In this passage, Darwall is defining what he calls competence, which is required in any second-personal address (Fichte's Point). Competence is what makes us subject to moral obligation and includes ‘a source of the (second-personal) reasons in which moral obligation consists, along with the capacity to act on these reasons’ (Darwall 2006, 34). Darwall continues: ‘In presupposing this, we effectively presuppose autonomy of the will. The capacity of will that make[s] us apt to be held responsible, second-personal competence, is a “law to itself,” since it is the basis of second-personal authority’ (Darwall 2006, 34, my italics). Darwall here is explaining that we presuppose the autonomy of the will because we presuppose competence and competence entails autonomy of the will. Note that competence entails autonomy of the will because it is the capacity of demanding oneself and choosing to act on those demands, which is to have authority over oneself. And it is this aspect of competence that entails autonomy. In a nutshell, from competence we go to authority to oneself, and from authority to oneself, we go to autonomy of the will.

The general point, then, from Darwall is that the autonomy of moral agents is intimately connected to the authority we have over ourselves. Any adequate theory of moral obligation must account for the fact that moral agents are autonomous, and, therefore, would need to appeal to the authority moral agents have over themselves. Thus, authority over oneself must be presupposed by any account of authority because such accounts must be developed against and consistent with the fact that moral agents are autonomous. Whether autonomy is cashed out in terms of competence and authority (as Darwall does) or is just taken to be something important to (but not entailed by) autonomy, in either case one would need to appeal to authority over oneself to have such an account.

These two reasons suggest that having authority over oneself could not turn out to be an incoherent notion on any account of authority (or of obligation generally). The notion of authority over oneself seems extremely relevant to autonomy and needed to account for the surprising and puzzling aspects of the fact that we are subject to directives that originate outside ourselves. Having such a compelling case for authority over oneself, I am very sceptical that one can provide an argument that self-commanding is incoherent just by affirming that the third part of the preparatory condition of self-commanding is incoherent. If it were, autonomy would become irrelevant for morality and facts about subjection would not be surprising at all, consequences that, at the very least, seem suspiciously false from a moral standpoint. I conclude that self-commanding is not an incoherent concept; that it actually seems needed for moral theorizing about relevant concepts or problems; and that those who believe it is, need to provide an argument for such claim.

Objections

Inconsistency with theological voluntarism

One immediate problem that the possibility of self-commanding raises for theological voluntarism (specifically, a DCT) is that it seems inconsistent with one central tenet of this view.¹² For one main motivation to endorse voluntarism is to be able to say that God is authoritative over human beings. In fact, this might follow from a strong view of divine sovereignty that is also a reason to endorse theological voluntarism. Nevertheless, the introduction of self-authority seems to undermine the claim that it is God who has authority over humans. Since humans can and (we may suppose) do (at least sometimes) have authority over themselves, then God's being the sole authority of humans is false which entails that theological voluntarism is false.

I think there are many ways in which the DCT-ist can defuse this worry. First, note that having self-authority – as having authority over another person – requires that certain conditions are obtained. For example, a child can have authority over himself. However, the fact he can does not mean he does. By virtue of being the child of his parents – and the parents being their benefactor, guardian, etc. – it is the parents who have authority over the child. Thus, in this case, even if the child can have authority over himself, by being in a certain (normative) relation to his progenitors, it is his parents who in fact have authority over him. And so, even if humans can have self-authority, that does not entail that they do. It is God who has authority over them in virtue of his relation to his creators, not only as a parent, but as creator, benefactor, etc. This account, then, is consistent with the possibility of self-authority.

One might object that it is not sufficient that theological voluntarism is consistent with the possibility of humans having authority over themselves. It must also allow for humans actually having self-authority. I do not think this is a requirement, but I can imagine a way to satisfy this demand. It is known that parents have authority over their offspring. Does this entail that God is not authoritative over the children? No. It could be the case that there is a hierarchy of authoritative figures of which God is the ultimate one. State officials have the authority to take into custody children who are mistreated by their parents. In cases like these one may say that the state officials' authority overrides the parents' authority. So, it doesn't seem to be a requirement of theological voluntarism that God is the sole authority but the ultimate authority. Allowing for hierarchies of authoritative figures, we can allow that God is the ultimate authoritative figure while also allowing that individuals (as also social entities like the state) (actually) have authority. This is consistent with God's being the ultimate authority over everything created.

Divine timelessness

The argument presented here relies heavily on Searle's account of speech acts, specifically on his account of the constitutive rules that define what a command is. According to this account, a command requires that the commanded action A is a future act of H. Thus, the very idea of commanding presupposes that it is logically possible and coherent to talk about the addressee's future. However, many prominent theologians and philosophers hold that God is a timeless being, that is, they claim that God exists in the ‘eternal present’, which is an infinitely extended, pastless, and futureless duration.¹³ If these thinkers are right, then tensed temporal categories and distinctions do not apply to God. Hence, it would follow it is incoherent that God can command himself precisely because it is logically impossible that God commands himself to do a future action since God's mode of existence is timeless. Therefore, even if the notion of, say, humans commanding themselves is coherent, the notion of God commanding himself is not.

Let us, first, mention two indirect ways of getting out of this problem. One way to address this concern is to say that, though the argument presented here uses Searle's framework of speech acts, it is not an essential part of the argument. One could adopt another theory of speech acts that does not presuppose that we can apply tensed temporal properties to God. Another way out is to reject divine timelessness altogether or adopt a weaker version of divine timelessness (see Padgett 1989, 1992, 2011 for example). If either of these two ways seems attractive for some, then the problem dissolves quickly.

However, it would be desirable if what I have argued for here is consistent with such a prominent view as divine timelessness, and I think we can make it consistent with it. An advocate of divine timelessness could say that God's commands are issued simultaneously, that is, everything that God commands himself is commanded at the ‘eternal present’ just as God knows (according to divine timelessness) everything at one instance. Schematically, then, advocates of divine timelessness can amend the propositional content condition of self-commanding thusly:

Content Condition * → If S is temporal, then the commanded action must be a future act A of S; if S is timeless, then the commanded action must be an action in the timeless dimension of S.

Furthermore, this change can be justified along the following lines. When ordinary humans command themselves (and other people), they do it in accordance with what stage of their life they are in, since humans live their lives in a fragmented way, that is, not fulfilled all at once. God, on the other hand, has the most perfect mode of existence. As such, God's life is complete and fulfilled; it is impossible that God's life is fragmented in any way. For this reason, some think that God must be timeless. But for the same reason one could argue that, since commanding oneself is part of one's life, God's commanding activity is complete and fulfilled. It might be that God commands himself to do only one thing, namely, that there is just one command issued from God to Godself. Hence, from the fact that God's life is necessarily complete and fulfilled, it follows that his commanding activity could not be affected by the passage of time just as any other part of God's life cannot be circumscribed by the fleeting passage of time.

The irreflexive character of commands

Another objection is to just deny that we can take the reflexive case of the speech act ‘commanding’. In other words, one might object that it is an implicit assumption on the Searlean illocutionary conditions of commanding that S ≠ H. And this finds justification in that it explains the sense of oddness and unintelligibility of the notion of commanding oneself. Therefore, one can understand Quinn's challenge in the following lines. We have (prima facie) reason to dismiss or at least be sceptical of concepts and notions that are difficult to understand. Self-commanding is a notion of noticeable difficulty. Therefore, we have a reason to be at least sceptical of this notion, and the reason is that one Searlean condition of commands is that the speaker (commander) must be distinct from the hearer (commanded).

This objection seems to beg the question against the possibility of self-commanding. For if one presupposes that S ≠ H, one automatically rules out self-commanding. A more charitable interpretation of this objection, however, might be that we can come up with independent reasons to affirm that the speech act of commanding is irreflexive. One reason to conclude that self-commanding is incoherent might be that taking the reflexive case of any (or most) speech act(s) results in making that speech act self-defeating in some way. But this cannot be the reason to conclude that self-commanding is incoherent. It is extremely plausible that the reflexive case of a coherent speech act is still a coherent speech act with the obvious exception of speech acts stipulated not to be reflexive. Take questioning as an example. We question ourselves all the time! I take it that, colloquially, ‘to question oneself’ sometimes means to doubt oneself and not to ask a question. Whether it is one or the other, both are speech acts, and we address ourselves acting in such ways, namely, questioning (in the asking sense) and doubting ourselves.

More to the point, promising is a speech act that is closer to commanding (it has normative moral force, it's directed, it binds the promisor, etc.) and that, plausibly, can be done reflexively. Though I do not have the space to argue in detail for the coherence of self-promising here, two points are worth noting. Although it is philosophically puzzling to make sense of self-promising, it is known that people often make promises to themselves. One piece of evidence is that, when a self-promise is broken, we hold ourselves accountable for not fulfilling our promise just as we would hold accountable another for not fulfilling it. Second, there has been a recent development of philosophical accounts that not only argue that self-promising is coherent, but show how morally significant they are, which support the claim that commanding can also be done reflexively.¹⁴ I take it that this is far from being a knockdown argument, but I believe it supports the claim that it is false that the reflexive case of speech acts (even those which are in way similar to commands) are not incoherent.

Are self-commands really commands?

One might be sceptical that self-commands are really commands. For it appears that there are actions such that I have the authority over myself to do (and all the other conditions for self-commanding hold) that no-one (me or anyone else) could command me to do. For instance, it is plausible to think that I have some kind of authority over myself even to do something wrong or foolish. It may be ‘up to me’ whether I indulge in some behaviour that could become a bad habit. It is less clear that I or anyone has the authority to command indulgence in such behaviour. This suggests that even if I have authority over myself to do something, this is different from commanding myself (or another commanding me) to do that action. Self-commanding would, then, be more akin to deciding or persevering in a behaviour. If this is a reasonable thing to say, we might ask, do we want to collapse the distinction between a decision or resolution and a self-command? Or do we want to retain the distinction in some way that is compatible with divine self-commands and therefore with DCT-allowed divine obligations?

One way one could sidestep this objection is found in Adams (1999, ch. 10).¹⁵ He notes that DCT-ists don't claim that any command has the normative force to generate obligations. There are non-deontic considerations that need to be the case for a command to be valid. For example, the content of the command matters; the character of the commander matters; the relationship between the commander and the commanded matters. If any of these go afoul, we do not have a valid command. One can go down Adams's route and say that whether one has authority over oneself and, therefore, whether the commands one issues to oneself are valid depends also on these kinds of factors. For instance, one could say that one has authority over oneself only if one's relation with oneself is a good or virtuous one, since only then one will be adequately responsive to reasons for action. So, in the case of it being up to me whether I smoke fifty cigarettes a day, under this account, it would be true that I can choose to do so, but false that I have authority over myself to command me that I do so precisely because my relationship with myself is anything but virtuous. This way we distinguish acts of deciding from acts of commanding by saying that the former requires conditions that the latter does not (e.g. a virtuous relation with myself).

Interestingly, this works perfectly in the case of God. God has a perfect relationship with himself. In fact, one of the many reasons that Christians give in favour of a trinitarian God is that a perfect being must be self-sufficient but also needs to be loving, and loving is a relational notion. Therefore, God, though one substance, is three persons in perfect loving harmony with each other. This allows us to say that God has – or better, can never fail to have – perfect authority over himself and all of creation partially in virtue of the perfect relation he has to himself.

Conclusion

I have argued that the speech act of commanding oneself is not incoherent. Specifically, I argued that the essential, content, and sincerity conditions of self-commands are not incoherent. This left us with the preparatory condition. I argued that the preparatory conditions are not incoherent. Most attention was given to the preparatory condition of having authority over oneself. I argued that under three views of authority, having authority over oneself is not incoherent. I then argued that authority seems to be presupposed and needed in moral theorizing so that even if the views I have mentioned turn out to be incorrect, we have reason to believe that the correct account does not exclude the possibility of one having authority over oneself. From this, I conclude that no condition of the speech act self-command is incoherent. Since we have defined that a speech act, SA, is incoherent just in case some condition of SA is incoherent – in the sense defined above – it follows that self-command is not incoherent.

Applying our conclusions to DCT, we can make the following conjectures. If one thinks one of the three views of authority is the correct one, then we can apply it to God and conclude that, because no conditions of self-commands (not even the preparatory conditions) are or entail a contradiction, it is not incoherent for God to command himself. If one does not have any inclination to accept one of the three views presented above, one can generalize the point and concede that, whatever the correct account of authority is, it does not exclude the possibility of one having authority over oneself. If so, one may say that it is logically possible that God has authority over himself and, therefore, no incoherence will follow from the fact that (actually and, therefore, possibly) God commands himself. This serves as reason to believe that 4 (the claim that it is not logically possible that, for some action φ, God commands that God φ) is false, hence a reason to believe that God can have (and perhaps does have) moral obligations.