II. CONTENT-INDEPENDENCE

A (practical) reason is a fact that counts in favor of an agent acting in some way.Footnote ⁶ So, for example, the fact that exercising makes me healthier counts in favor of me exercising, so that fact is a reason for me to exercise.Footnote ⁷ This example demonstrates the way reasons normally work. When considering some act, such as exercising, I examine the merits of exercising, such as the fact that it will make me healthier, as well as the demerits of exercising, such as the fact that it will be time-consuming. These features of exercising are reasons for me to exercise or to refrain from exercising. I then balance my various reasons to determine whether, all things considered, I should exercise.

Content-independent reasons work differently. The most discussed example of a content-independent reason is a command issued by an authority. Imagine a parent exercising parental authority over her child and commanding the child to go to bed. If the child asks why she should go to bed, that is, if the child asks for reasons, a paradigmatic parental response is “Because I told you to.” With this response the parent does not appeal to the merits of going to bed, such as the benefits of a good night's sleep, or the demerits of not going to bed, such as being tired in the morning. Instead, the parent appeals to something that has nothing to do with the merits or demerits of going to bed at all: the fact that the parent commanded the child. The fact of the command is a reason for the child to act, but it is not one of the merits or demerits of what was commanded.

The most distinctive and important characteristic of content-independent reasons is the way they sever the normal connection between the merits and demerits of acting in some way and whether one has a reason to act in that way.Footnote ⁸ If I am commanded to exercise, I now have a reason to exercise that is not one of those valuable features of exercise that I normally consider. While we understand why the health benefits of exercise count in favor of exercising, it is unclear under why the “extraneous” fact of the command should count in favor in the same way.Footnote ⁹ Indeed, this is one reason that theorists have embraced philosophical anarchism, for they do not see how an autonomous moral agent could be justified in ignoring the character of the action she was taking and relying on extraneous facts, especially extraneous facts regarding the fallible will of others.Footnote ¹⁰

For all its oddness, content-independence appears to be a theoretically useful concept. Most influentially, H.L.A. Hart uses content-independent reasons to analyze the commands of an authority, following insights from Hobbes and Bentham and followed by many others.Footnote ¹¹ It is also used more widely, for example to analyze promises,Footnote ¹² decisions,Footnote ¹³ and requests.Footnote ¹⁴ Furthermore, the contrasting notion of content-dependence has been used to analyze advice fruitfully.Footnote ¹⁵

Unfortunately, this useful notion is unclear and has often been characterized in a quite hasty and imprecise manner. Here is Joseph Raz's characterization of content-independent reasons: “A reason is content-independent if there is no direct connection between the reason and the action for which it is a reason.”Footnote ¹⁶ Hart says a command “is intended to function as a reason independently of the nature or character of the actions to be done.”Footnote ¹⁷ This is essentially the extent of their discussions of the meaning of the term; early on, Hart expresses his hope that the idea will be “self-explanatory.”Footnote ¹⁸ Yet these characterizations raise more questions than they answer. What is the nature of this “direct connection?”Footnote ¹⁹ In what sense does the reason function “independently?” As noted by Andrei Marmor, far from being self-explanatory, “the notion of content-independence is notoriously difficult to define.”Footnote ²⁰ Thus, despite the evident usefulness of the ideas of content-independence and content-dependence, they have led to considerable confusion, as shown in more detail below.

In order to understand content-independence, we need to get clear on three issues: 1) what the relevant content is; 2) in what sense the reason is independent of said content; and 3) what the reason is dependent on, given that it is not dependent on content.

I begin with the third issue, which will shed light on the first two. The parental example is instructive. When saying that the reason for the child to go to bed was “Because I told you to,” the parent is highlighting a feature of the command that is not about the content of the command. In this case it is about who issued the command, or what I call the command's source.Footnote ²¹ This hearkens back to Hobbes, who writes that “Command is where a man saith, ‘Doe this,’ or ‘Doe not this,’ without expecting other reason than the will of him that says it.”Footnote ²² According to Hobbes, the reason to do as commanded depends on a feature of the source of the command (“him that says it”), not a feature of what was commanded.

To see this more clearly, consider two variations on the parent case. In the first variation, the parent says, “Eat your peas” instead of “Go to bed.” This changes the content of the command but keeps the source constant. In the second variation on the original case, a stranger commands the child to go to bed, changing the source but keeping the content constant. The question is whether the child has an obligation to obey in the variations. The answer is obviously that the child has an obligation in the first variation but not in the second variation. The stranger lacks the parent's standing so cannot issue commands to bind the child; the child does not have an obligation to go to bed just because a stranger tells her to. But the child does have an obligation to eat her peas just because her parent tells her to, just as she has an obligation to go to bed if her parent says to. Changing the content of the command does not change whether the child is obligated, but changing the source does. That is, whether the command is a reason for the child to obey depends on features of the command's source, not on features of the command's content. It is therefore natural to say that a command is a content-independent but source-dependent reason (or what one might call an exclusively source-dependent reason).

The lesson of this example is that content-independence necessarily calls attention to the fact that the reason depends on something other than the content. The fact that you were commanded to act can be a content-independent reason because there is something other than features of the command's content that the reason can depend on, namely features of the command's source. In order to make sense of content-independence, we need this contrast with content. In more general terms, a command can be a content-independent reason only because the command itself is a container. A command is a speech act that has referential content; its content is the act that it refers to. To say that a command is a content-independent reason to obey is to say that its status as a reason to obey depends on features of the container (the speech act), not on features of the content (what the speech act refers to). It is this contrast between container and content that allows us to make sense of commands as content-independent reasons.

This implies an answer to our first question about the nature of content. Content-independence applies only to reasons that have containers as constituent parts. Recall that (on my view) practical reasons are facts that stand in a certain relationship to agents and actions. As examples of reasons, we have seen the fact that exercise promotes health and the fact that my parent commanded me to go to bed. The former is a reason for me to exercise and the second is a reason for the child to go to bed. These facts (states of affairs, as I understand them) have constituent parts. The first fact has exercise and health as some of its constituent parts, the second has the command. The fact that my parent commanded me to go to bed can be a content-independent reason because it partially consists of a speech act that has referential content, so we can usefully distinguish the content from the container. The fact that exercise makes me healthier cannot be a content-independent reason because none of the constituent parts of the fact are containers. There is no content in the relevant sense. I have a reason to exercise due to the fact that exercise makes me healthier, but this reason is neither content-independent nor content-dependent on my understanding.Footnote ²³ The distinction simply does not apply to reasons that do not have containers as constituent parts.

Interestingly, we can apply content-independence to other sorts of containers, such as cups: the fact that this cup was a gift from a loved one can be a content-independent reason to drink from the cup.Footnote ²⁴ This application makes sense because we can contrast features of the container (who it reminds you of) with features of the container's content (the liquid, or whatever, in the cup). Normally, the reasons to drink from a cup are a function of features of the content of the cup, as when you drink from a cup filled with water in order to quench your thirst. But drinking from a cup that was a gift can be an expression of gratitude or connection quite apart from whether the cup contains water or tea or whatever else, so you can have a content-independent reason to drink from the cup. The paired notions of content-independence and content-dependence apply only when the distinction between container and content can be usefully drawn.

Here I do not provide a definition of a container or necessary and sufficient conditions on being a container.Footnote ²⁵ I intend to use the notion of a container in a commonsensical way. There may be marginal cases where such a definition would be helpful, but my main goal here is simply to show that content-independence and content-dependence depend on the idea of a container and that not all reasons have containers as constituent parts. This allows us to see when the notions are useful and what they are useful for.

We next need to know what it means for the reason to be independent of content. The main point of clarification concerns what features of the reason are independent of content. I take the relevant sense of dependence to be a kind of supervenience relation: if some feature of the reason is independent of content, then changing the content does not change that feature of the reason.

We can distinguish three features of a reason that might be independent of content: the reason's existence, its weight, and the action it counts in favor of.Footnote ²⁶ On my understanding, what is important about content-independence is that the reason's existence and weight do not depend on content.Footnote ²⁷ That is, features of the container, not features of the container's content, determine whether the agent has a reason of a certain weight.Footnote ²⁸ The content can determine what action a content-independent reason counts in favor of.Footnote ²⁹ What this entails is that changing the content of a content-independent reason does not change whether the agent has a reason or the weight of the reason, but it does change what the agent has a reason to do.

An example will help. Above I give the case of a parent commanding a child to go to bed and a variation where the parent commands the child to eat her peas. If my explanation of content-independence is correct, then the change in content should not change whether the child has a reason or the weight of that reason, but it should change which action the command counts in favor of performing. And this is exactly what we see. The parent's commands are a reason for the child to act in both cases, and in both cases the reason is of equal weight but the favored actions are distinct because one is a reason to go to bed and the other is a reason to eat peas, as determined by the content of the respective speech act.

If we change the source of the speech act, as in the stranger case, the existence and the weight change but the favored action does not. The source of the command is wrong, so the child has no reason (of no weight) to go to bed. Note that the favored action is still determined by the referential content of the command: it is precisely a reason to obey and go to bed that the child does not have due to the fact of the command. The stranger's command can be a reason to act in other ways, for example a reason for the child to tell a parent about a nearby, interfering stranger.

In sum: a reason is content-independent iff it has a container as a constituent part and changing the content of that container does not determine whether the agent has a reason and does not determine the weight of that reason.Footnote ³⁰ The existence and weight of a content-independent reason instead depend on features of the container. The paradigmatic case of a content-independent reason is a reason constituted by a speech act (including commands, promises, and requests). In the case of speech acts, we are most often concerned with the existence and weight of the content-independent reasons that are determined by features of the source of the speech act.Footnote ³¹

III. THE PUZZLE OF UNJUST LAWS

This explanation of content-independence allows me to respond clearly to one of the idea's most vexing puzzles. If a law imposes a content-independent moral obligation, then according to the understanding just articulated, changing the content of the law will not change the existence or strength of subjects’ reason to do as the law commands.Footnote ³² It is a feature of the container that directly determines the fact that subjects have a moral obligation, not any feature of the content of the particular law. So if citizens have a general content-independent political obligation, then they are morally obligated to obey every law, regardless of the content of the law.

Yet this cannot be the case. To modern eyes it is obvious that clearly and egregiously unjust laws cannot bind citizens.Footnote ³³ If the United States repealed the Thirteenth Amendment and thereby made slavery legal again, laws defining the legal standing of slaves and laws supporting slave ownership would not morally obligate United States citizens to consider others as slaves or support slavery. This would be absurd. Yet we are clearly making this judgment on the basis of the content of the particular law: laws legalizing slavery do not morally bind precisely because slavery is egregiously morally wrong. So content-independence looks like it cannot account for the fact that laws like this do not bind.Footnote ³⁴ George Klosko goes so far as to claim that the notion of a content-independent political obligation is at odds with the entire idea of limited government.Footnote ³⁵

To be as clear as possible about the puzzle: according to the understanding of content-independence just given, the reason is independent of content in the sense that changing the content does not affect the existence or weight of that particular reason. The objection claims that this cannot be the case because egregiously unjust commands do not constitute binding reasons: changing the content of a command to an egregious injustice affects the existence or weight of the reason. So the explanation of content-independence given cannot account for a considered judgment about commands.Footnote ³⁶ Something has gone wrong: perhaps content-independence is conceptually incoherent or perhaps it is coherent but fails to explain commands. In either case, the search for content-independent reasons in the political sphere is bound to fail, and the whole debate needs to be reframed.Footnote ³⁷

Two notes about this puzzle. First, it is important that it is limited to egregious injustices. The claim is not that every unjust command fails to bind.Footnote ³⁸ As Raz emphasizes, this would contradict the very nature and purpose of authority. Necessarily, some mistaken commands bind.Footnote ³⁹ Otherwise people should just do what the balance of reasons supports regardless of authority, and authorities cannot play the mediating and coordinating role that they do. Understanding authority in this way—as making no normative difference—invites confusion and chaos. This limitation also seems consonant with our experience with the law. We tolerate some perceived injustices, but at a certain point we reject the law's claim to authority.

Second, it will not do to respond to the puzzle by saying that commands are “partly” content-independent.Footnote ⁴⁰ As Green notes, content-independence is not a matter of degree.Footnote ⁴¹ A partly content-independent reason is just a content-dependent reason.Footnote ⁴² The point of identifying content-independent reasons is to separate them from disagreement over the quality of content and so to establish their unusual but possibly helpful role. This role is undermined if content comes back into play to any degree. This is a conceptual point. Content-independent and content-dependent reasons will often have to be weighed against one another in practical matters, and just because a reason is content-independent does not mean it is overriding.Footnote ⁴³ But we still need to keep those reasons that are features of the container distinct from reasons that are features of the content, and categorize them accordingly.

In the remainder of this section I solve this puzzle. In short, the solution is that the content affects the container. Egregiously unjust commands can affect the existence and weight of a reason because egregiously unjust commands affect our assessment of the qualities of the source of the command. This does not make such commands into content-dependent reasons. Features of the container, not features of the content, still directly determine the existence and weight of the reason. Unjust commands can fail to bind yet still be content-independent reasons. In these cases the content is relevant, but only indirectly so.

Importantly, however, this solution arises from a particularly instrumentalist, Razian perspective on the grounds of content-independent reasons and authority. I leave it to defenders of other theories (for example, hypothetical consent or democracy) to determine whether content-independence can be consistent with limited government on their accounts.Footnote ⁴⁴ On Raz's service conception of authority, A has genuine authority over B when B would more likely conform to her reasons by trying to follow A's directives than by following her own private judgment.Footnote ⁴⁵

How could we assess whether B would be more likely to conform to her reasons by following A's directives? We cannot consider the content of a particular command and compare it to B's preexisting reasons. First, this would replace the notion of likelihood with a guarantee that B would better conform and would make the reason content-dependent. Second, this is impossible for limited beings like ourselves, who often lack knowledge and appreciation of the relevant reasons. Third, it would render authority useless by making it impossible to coordinate actions in conditions of uncertainty and disagreement about content.Footnote ⁴⁶ The assessment of whether B is more likely to conform to her preexisting reasons must instead be made indirectly.

Not coincidentally, this indirect assessment is similar in many ways to the assessments that novices must make about experts.Footnote ⁴⁷ If novices were in a position to judge whether someone is an expert by examining the quality of their work, they would no longer be novices, they would be experts themselves. So instead novices use indirect indicators of expertise. For example, in the case of medical doctors, we use indicators like degrees, certifications, reviews by other trusted experts, and so on. One of these indirect methods is looking at a track record, that is, at the pattern of success and failure in particular instances.Footnote ⁴⁸ A flawed track record is indicative of some failure that is relevant for the assessment of present ability to act well.Footnote ⁴⁹ As a part of a track record, then, the content of a command can be relevant as one indirect way that assessments of authority get made. This opens the door for a solution to the puzzle.

In the case of egregious injustice, the problem is not just that this particular command is mistaken. Egregious injustices infect our assessment of someone's judgment on the whole.Footnote ⁵⁰ If the U.S. government were to repeal the Thirteenth Amendment, it would not be only that particular command that failed to bind. We citizens would then have excellent grounds for doubting the judgment of the government in future cases as well, not just in this case. If the U.S. government thought that allowing slavery is permissible, especially after our horrible history, then its judgment would be fundamentally flawed. Failure to appreciate the clear and decisive reasons against slavery would show that the government cannot appreciate the most basic reasons of the kind it needs to rule, let alone the nuanced and difficult reasons that are relevant to political questions.Footnote ⁵¹

When A issues a command to perform an egregious injustice, it is true that this command is not a reason for B to perform an egregious injustice. But that is not because this one command fails. It is because A loses her authority in the very act of commanding an egregious injustice. Future commands from A would also fail to constitute reasons for B.Footnote ⁵² The content of the egregiously unjust command is relevant to the overall assessment of A's judgment and so is relevant to whether A has authority over B. The egregiously unjust command makes it the case that A loses her authority, so that particular command and future commands all fail to bind.

Recall that content-independent reasons depend on features of the container. In this case, the content is relevant to the features of the container, and so indirectly changes whether there is a reason.Footnote ⁵³ But the existence of the reason is still determined by a feature of the container, not a feature of the content. When the content does not affect the container despite the demerits of the content, as in the case of slightly unjust commands, the subject still has a reason because the content is not the determinant.

So the content is relevant to the existence of the reason, but only indirectly by being relevant to whether A has authority at all. This does not undermine the claim that the commands of genuine authorities constitute content-independent reasons.Footnote ⁵⁴ This is, I believe, the only satisfying explanation of how content-independence can be consistent with the fact that egregiously unjust commands fail to bind.

IV. CONCEPTUAL CHALLENGES

My understanding of content-independence also provides the resources to respond to two challenges to content-independence in the literature. P. Markwick claims that the idea of content-independence is either conceptually incoherent or useless.Footnote ⁵⁵ Stefan Sciaraffa claims that it can be made coherent and useful, but only by jettisoning any reference to the idea of content.Footnote ⁵⁶ I show below that my account is not subject to Markwick's criticisms and that Sciaraffa's account mischaracterizes some reasons.

Markwick's main argument is that content-independence is a feature of all reasons so cannot be used fruitfully to analyze concepts like commands because it does not identify a distinguishing feature of those concepts.Footnote ⁵⁷ Compare the following reasons for Alice to ϕ: because she was commanded to ϕ and because ϕ-ing maximizes utility. It appears that the command is a content-independent reason for Alice to ϕ because we could specify ϕ-ing in any way and Alice would still have reason to ϕ. Whether a reason to ϕ exists depends solely on whether the property “being commanded” obtains, so is independent of the particular specification of ϕ.

But Markwick notes that the same is true of Alice's other reason as well.Footnote ⁵⁸ For any ϕ, if ϕ maximizes utility then we have a reason to ϕ. The existence of the reason depends on whether the property obtains and not on how ϕ is specified, so is independent of its content. Maximizing utility suddenly appears to be a content-independent reason as well. Furthermore all reasons, not just maximizing utility, will pass this test, seemingly due to the simple fact that they are reasons, so apply across cases. If this is true, then content-independence is a feature of all reasons and does not distinctively identify reasons given by commands or anything else, so should not be used to characterize or analyze some subset of reasons.

My understanding of content-independence is not subject to this argument precisely because Markwick and I disagree about what counts as content in the relevant sense.Footnote ⁵⁹ Markwick does not restrict the notion to reasons that have containers as their constituent parts, instead assuming that the relevant content is the action that the reason counts in favor of rather than the content of some constituent part of the reason. Markwick argues that the fact that ϕ maximizes utility is a content-independent reason to ϕ because whether ϕ maximizes utility is independent of ϕ. On my view, this reason cannot be content-independent because it does not have any containers as constituent parts.

Why not think the reason itself is a container? The standard treatments of content-independence seem to imply this; discussion of the content of the reason is rampant. But, as I show below, these discussions occur in the context of reasons that are themselves speech acts and so have content on my understanding. Reasons in general are facts, that is, states of affairs, like the fact that I was commanded to ϕ. The facts themselves have no content on their own (even if our statements of the facts do). The fact that I was commanded to ϕ does not “contain” the command, it is simply partially constituted by the command. Loose talk of the content of reasons in contexts where the relevant reasons are containers should not lead us to the view that reasons per se have content.

There seem to be two main reasons that Markwick applies content-independence to all reasons: first, in defining content-independence no theorist has heretofore explicitly restricted its application to containers. Recall Raz's definition: “A reason is content-independent if there is no direct connection between the reason and the action for which it is a reason.”Footnote ⁶⁰ Since there is no restriction, this appears to apply to any reason whatsoever.

Second, Markwick claims there are instances where some theorists have ascribed content-independence (or -dependence) to noncontainers. It may be the case that we could find some instance of this. But the two cases Markwick relies on are infelicitous. Markwick claims that Raz refers to penalties as content-dependent reasonsFootnote ⁶¹ and that Leslie Green ascribes content-dependence to considerations of utility.Footnote ⁶² In context, however, both are considering containers.

In the text Markwick refers to, Raz does not refer to penalties at all. That aside, Raz does consider threats to be a sort of reason. And in the case of threats, we can see how penalties do constitute a content-dependent reason on my view. The threat is a speech act, and the threatened penalty is part of its content. The reason to comply with the threat changes if we change the threatened penalty. If the robber says, “If you do not give me your wallet, I will kill you,” you have very strong reason to hand over your wallet. If the robber says, “If you do not give me your wallet, I will weakly poke you in the arm,” you have hardly any reason at all to hand over your wallet. So penalties can be content-dependent reasons when they are the content of a threat. Insofar as Raz is concerned with penalties at all, it is when they are part of a threat, so do not stand as a use of content-independence or -dependence outside the context of a container.Footnote ⁶³

Markwick's reading of Green encounters the same problem. Green does indeed refer to considerations of utility as content-dependent. But he does so in the context of considering the reasons subjects have to obey a command or follow a social norm. Consider a social norm to ϕ; the norm is the container, and the referential content of the norm is ϕ. What reasons do you have to conform to the norm? You have the content-independent reason to conform to the norm because of the fact that a social norm requires ϕ-ing, but you also often have content-dependent reasons to conform to the norm because of the nature of ϕ–ing, like the fact that ϕ–ing generates a net gain in utility. The former reason is content-independent because changing ϕ, that is, changing the container's content, does not change the fact that a social norm requires it, so does not change whether you have a reason or that reason's weight. The latter is content-dependent because whether you have a reason and the weight of your reason depend on how ϕ is specified, because that will change whether and to what extent there is a gain in utility.

The contrast between these sorts of reasons is common, which Green highlights with his discussion of indirect consequentialism. The whole advantage of indirect forms of consequentialism is that I can take the content-independent reasons I have to obey a rule or follow a norm (like the fact that it is part of a utility-maximizing set of rules) as decisive, thus avoiding the costs of calculating the utility-based reasons to ϕ that are constantly changing according to how ϕ is specified and its features in a given context. These utility-based reasons to ϕ are content-dependent when considered in relation to a rule or norm to ϕ, not simply content-dependent in general. It is only in the context of considering the content of containers that Green concerns himself with considerations of utility. This is precisely where content-independence comes into play and the controversy over political obligation has its bite.

That said, Markwick may be correct that some people have used content-independence in the way he describes and that usage may well be incoherent in the way he argues. But we need not follow the way the idea has been used by others, so I conclude that Markwick's arguments do not show that my understanding of content-independence is incoherent or useless.Footnote ⁶⁴ His arguments do not apply to an understanding of content-independence that is restricted to container reasons, and we have no good reason to extend the notion of content-independence beyond containers.

Interestingly, although content-independence and content-dependence are applied almost exclusively to speech acts, there are two instances that I know of where they are not. Raz says that rules are content-independent reasons but explicitly denies that they are speech acts,Footnote ⁶⁵ and Marmor refers to content-dependent reasons to follow a map.Footnote ⁶⁶ It is clear that rules are referential containers just like speech acts, and we can make sense of a map as a referential container as well, as its features are meant to refer to features of the area. These applications of content-independence are consistent with my understanding that restricts itself to containers but not solely to speech acts.

My understanding of content-independence also shows why Sciaraffa's proposed account is off track. Sciaraffa finds inspiration for his analysis in Hart's discussion of commands. For Hart a command both expresses the authorities’ intention that subjects act in particular way and expresses the authorities’ intention that the subjects act in that way because the authority so intended. So it is not enough to do what authorities say; you must do what they say because they say it. Call this feature of commands their “reflexive intention.” Sciaraffa argues that reflexive intention is what defines content-independent reasons.Footnote ⁶⁷ (I think reflexive intention is importantly characteristic of commands, following Hart; my disagreement with Sciaraffa is whether reflexive intention captures content-independence.) According to Sciaraffa, you have a content-independent reason to ϕ if and only if someone's reflexive intention is a reason for you to ϕ.

This move appears to lead Sciaraffa to restrict his view to speech acts, so content-independence applies only to containers on his view. There are two problems. First, Sciaraffa's view applies only to speech acts, but content-independence is a sensible notion to apply to other sorts of containers as well, as I show above in the case of the gifted cup. Second, and more important, the focus on intention is misleading because a speech act can be a content-independent reason without meeting Sciaraffa's test.

Consider the following case. A man works for an investment fund that happens to be run by his mother, and he advises her to purchase a certain stock, stock A. The mother knows the advice to be bad but purchases stock A anyway in an attempt to soothe her contentious relationship with her son. The son did not intend for his mother to purchase the stock because he said so, he intended her to purchase the stock because purchasing the stock is a good idea on its merits, as revealed by his good advice. He was trying to show that he was a quality employee, not attempting to encourage nepotism. As advice, his speech act was supposed to constitute a reason for his mother to act only if its content had merit. On Sciaraffa's account, then, the son's speech act cannot be a content-independent reason because the son did not intend the mere fact of his speech act to be his mother's reason for acting, that is, the son lacked the relevant reflexive intention.

However, it is clear that the speech act was a content-independent reason for the mother. The mere fact that it was her son's speech act gave her reason to buy that particular stock even though it was a bad stock. If her son had said to buy stock B, she would have had the same reason to buy stock B as she had to buy stock A because she can achieve her goal of appearing to respect her son's judgment regardless of the stock. And if some other employee had said to buy stock A, she would not have had reason to buy it because the source of the speech act was wrong, given her aims.

This aligns perfectly with my understanding of content-independence. It is a feature of the container, the speech act's source, that determines the existence and strength of the reason she had to buy the stock. Changing the content does not change those features of the reason, but changing the source does. So the son's advice is a content-independent reason for the mother even though he did not intend it to be.

Of course it may well be the case that all things considered, she should not have acted on her content-independent reason. The facts that she is deceiving her son, that she is risking him finding out and worsening their relationship, that she is using her professional position for personal reasons, and that she is risking losing money on a bad investment all count against her acting on her content-independent reason to buy stock A. But those facts do not show that she lacks a content-independent reason, just that her reason is outweighed.

Notice that regardless of the son's intention, the advice constitutes a variety of content-independent reasons.Footnote ⁶⁸ I focus above on the mother's content-independent reason to buy the stock. But the fact that my son advised me to buy stock A also constitutes other content-independent reasons. For example, even if the mother treats the advice merely as advice, the fact that her son gave her the advice means that she has a content-independent reason to treat her son as responsible for giving the advice, she has a content-independent reason to inform her son, not anyone else, whether she took the advice, and so on. These are content-independent reasons because their existence and weight are determined solely by features of the container, including features of the source such as his identity and institutional role, not by the content of the speech act. Whether the advice is good or bad, she has reason to inform her son of her response because he was the source of the advice. Had some other employee given her the same advice, she would have a reason to inform them instead, because the reason to inform depends on features of the container, not the content; had some random person that she stands in no institutional relationship to given her professional advice, she might not have a reason to inform them of her response at all. All speech acts will constitute a variety of both content-independent and content-dependent reasons of this rather banal kind.

It is important to see the application of content-independence in these sorts of cases, apart from the context of commands and the accompanying complicating factors of illocution and preemption. Standardly understood, commands demand obedience, not merely in the sense of doing as commanded but doing as commanded because one was commanded to do so, that is, by recognizing and appropriately responding to the speaker's reflexive intention. Reasons to obey, then, can obtain only when one has been commanded; if you do as someone says but there was no command, you have not obeyed in the strict sense. So the reflexive intention that is partly constitutive of commands is necessary for a speech act to be a content-independent reason to obey because of the nature of obedience, but content-independent reasons for other sorts of actions are commonplace, especially with respect to speech acts, and do not depend on the speaker having a reflexive intention.

My overall point that Sciaraffa's account is mistaken is established simply by the fact that the content-independent reason exists without the son having a reflexive intention. Sciaraffa's account also distances itself from the original understanding of content-independence by ignoring the notion of content entirely, claiming that the term is a “misnomer” and “misleading.”Footnote ⁶⁹ My account avoids the unfortunate implication that all the theorists who defined content-independence in terms of content and thought that content was important were badly mistaken.

V. POLITICAL OBLIGATION

The final issue I address is Klosko's attempt to reframe the debate over political obligation. He “criticize[s] content independence directly” and ultimately concludes that we should reject the search for content-independent political obligations, instead focusing on content-dependent political obligations.Footnote ⁷⁰ His discussion is wide-ranging, and a full treatment would take us far beyond the scope of this paper. I do not address many of the considerations he raises because my goal is not a complete discussion of or response to Klosko. (Indeed, I agree with some of Klosko's substantive conclusions but take exception to his framing of the problem in terms of content-independence.) Instead, my point is that the understanding of content-independence we have come to in this paper brings some clarity to his assertions and ultimately suggests that the debate over political obligation cannot be fruitfully recast in the way he claims.

A helpful way to enter the debate over political obligation is with a very practical question: When faced with a law that requires me to act in some way, what reasons do I have to act as the law requires? The most obvious reasons are the content-dependent reasons, having to do with the merits and demerits of acting in the particular way specified by the law. To take the philosophical anarchists’ favorite example, when faced with a stop sign, there are often reasons for me to stop because stopping prevents me from rushing into an intersection at high speeds, risking the safety and welfare of myself and others. But—and this is part of the anarchists’ point—these content-dependent merits of obeying the law requiring me to stop change from context to context. If I am in the middle of a flat desert, in the light of day, without any sight obstructions, and I can see that there is no risk of an accident, it appears that there are no merits to obeying the law and stopping. It is just a waste of time and gas. Do I have any reason to obey when I lack content-dependent reasons?

If my reasons to do as the law says are exhausted by content-dependent reasons, a variety of problems arise. First, because content-dependent reasons change according to context, citizens must constantly reevaluate whether they have reason to do as the law says, resulting in significant inefficiencies. Second, and more worrying, the law loses its ability to coordinate actions in situations of fundamental pluralism and disagreement. We disagree about what the content-dependent reasons are, so we disagree about when we have reason to do as the law says we must. Yet this undermines the coordinative goals of political institutions, which require most citizens to obey most laws most of the time.Footnote ⁷¹ The government achieves coordination by upholding the law even when citizens deny that they have an obligation to obey on content-dependent grounds (and even sometimes when the citizens are correct about this). Third, the content-dependent reasons will often be obscure to ordinary citizens, who do not have the time or expertise to evaluate the grounds of the huge volume of complex laws in modern society.

These problems for relying solely on content-dependent reasons are connected to the nature of authority. Authorities demand that subjects defer, or that they surrender their will, or that they replace their private judgment with the authority's judgment.Footnote ⁷² As Raz emphasizes, the whole point and value of authority are captured by the mediating role that deference enables.Footnote ⁷³ Authority abstracts away from the merits of particular commands and generates reasons based on who issued those commands. It thus bypasses debate and disagreement about the nature of underlying, fundamental reasons. There is “enormous” value in this mediating role, not only for purposes of coordination and efficiency, but for the possibility of a “pluralistic culture.”Footnote ⁷⁴ A genuine authority achieves these advantages by issuing commands that constitute content-independent (and preemptive) reasons.Footnote ⁷⁵

Given the problems with relying only on content-dependent reasons, the alternative is that I should obey a particular law for content-independent reasons. On the understanding of content-independence that I articulate in Section I, this means that there is some feature of the container that determines whether I have a reason and the weight of that reason.Footnote ⁷⁶ The debate over political obligation comes down to whether there is any feature of a law that can be a reason amounting to a moral obligation, disregarding the features of the content of the particular law.Footnote ⁷⁷

The disagreement between philosophical anarchists and statists can be understood only with the distinction between content-dependent and content-independent obligations to obey the law. For example, the most prominent and influential philosophical anarchist, A. John Simmons, admits that there often are good content-dependent reasons to obey laws, including some obligations to obey.Footnote ⁷⁸ What he denies is that states have the authority to issue laws that correlates to a general content-independent obligation to obey. Statists do not deny that there are good content-dependent reasons to obey many laws, nor do they deny that these reasons should often motivate citizens to obey. They claim that in addition, states have the general authority to issue and enforce laws independent of their content.Footnote ⁷⁹

Why, then, does Klosko think that moving to content-dependent views of political obligation is fruitful? The answer is that he mischaracterizes content-independence and the debate over political obligation.Footnote ⁸⁰ He says that he will understand citizens to be under a content-independent political obligation when they have an obligation to obey the law “because it is the law.” I call this a “purely legal” reason. He distinguishes this from citizens who have an obligation that is “independent of content” but not for purely legal reasons. He calls these “formal reasons.” Examples of formal reasons include considerations of coordination, fair play, gratitude, and democratic decision-making.Footnote ⁸¹ His argument for the move to content-dependent views relies on the claim that formal reasons are not to be understood as part of content-independent views. This is mistaken.

The mistake is not the claim that there is a distinction between purely legal and formal reasons. As Klosko correctly points out, much of the historical debate is framed in terms of purely legal reasons, and there are positions that endorse purely legal political obligations.Footnote ⁸² (These include, for example, positions that ground obligation in delegated divine authority to any extant political leader and positions that ground obligation in the fact that effective political institutions constitute escape from the threats of the state of nature.)Footnote ⁸³ The mistake is to think that formal reasons are not content-independent reasons. Content-independence frames the debate over formal reasons, and to reject this characterization is to misconstrue important elements of this debate.

There is an oddity to Klosko's discussion that brings out his error. He says that formal reasons are “independent of content” and notes that they “satisfy the conditions for content-independence,” but immediately stipulates that formal reasons fall under content-dependent views.Footnote ⁸⁴ Content-independent views, on his definition, are those views that include only purely legal reasons. The question is whether we have reason to follow his way of carving up the conceptual space.

There are good reasons to resist this definition and keep views that appeal to formal reasons in the content-independent camp.Footnote ⁸⁵ First, formal reasons and purely legal reasons share important features, speaking in favor of classifying them together (although not therefore collapsing the distinction between them). Second, Klosko's justification for classifying them separately is questionable.

The main reason to deny that content-dependent views can appeal to formal reasons is that formal reasons and purely legal reasons play the same role in the debate precisely because they are content-independent reasons. This role is distinct from the role of content-dependent reasons and gives rise to distinct challenges and prospects. This point arises again in terms of the practical question I open this section with: Why should I obey this law? When we appeal to content-dependent reasons, we are appealing to exactly the sorts of considerations that people are familiar with and that usually count as reasons in favor of acting in some way. As I state in Section I, the most distinctive and important feature of content-independent reasons is that they sever this normal connection between the merits and demerits of an act (the content-dependent reasons) and the act itself.

Formal reasons and purely legal reasons share this puzzling feature. When someone asks why they should obey some law and act in the way that is required by that law, both purely legal reasons and formal reasons appeal to something other than the merits of the act required by the law. Instead they appeal to some feature of the container. This appeal is puzzling in a way that the normal content-dependent reason is not. The difficult question that both types of reasons must answer is why features of the container are relevant for acting in the way specified by the content.

Solving this problem is notoriously difficult. To see why, consider two of Klosko's examples: the Pauline purely legal reason to obey the law as an act of deference to divinely appointed rulers and the democratic formal reason to obey based on equal respect for others. Due to the fact that both of these reasons abjure the merits of the law in question and instead appeal to a feature of the law (or lawmaking process) in general, the citizen has several open questions: Why should disobeying this law on this occasion be construed as a rejection of divine authority or equal respect? It may be true of some laws, some of the time, but why all the laws, all the time, regardless of content? Is going through the desert stop sign really disrespectful of my fellow citizens? Further, is it disrespectful in a way that could ground a moral obligation? All these questions arise for both the formal reason and the purely legal reason precisely because they are content-independent, appealing to the features of the container rather than merits of the act specified by the content. To see how difficult these questions are, one need only look at democratic theory and its ongoing struggle to answer them.Footnote ⁸⁶

Klosko claims that finding formal reasons that ground an obligation to obey the law “will not be overly difficult.”Footnote ⁸⁷ This mischaracterizes the debate. Simmons's influential anarchist arguments consider the formal reasons that Klosko appeals to, such as coordination, fair play, and democracy. He concludes that none can ground a general obligation to obey the law. He may be wrong, but certainly a fair assessment of the debate cannot claim that finding the connection between formal reasons and obligations to obey the law is not a significant challenge. The power of the anarchists’ arguments against formal reasons in particular is a large part of why the literature is increasingly sympathetic to skepticism about political obligation. Folding formal reasons into content-dependent views thus disguises important features of the ongoing debate.

Further, answering these difficult questions in the case of either purely legal reasons or formal reasons has important implications that are distinct from content-dependent reasons, again precisely because they are both content-independent. These implications are tied to the benefits of authority that I outline above. If we can successfully give an explanation for why obeying the law is a necessary act of respect for the equal standing of others, then we can forego debates about the qualities of particular laws. It allows us the benefits of efficiency, coordination, and pluralism precisely because we do not have to agree about the content-dependent reasons. Instead we agree about features of the law qua container and obey for those reasons. These same benefits are possible if we could explain the purely legal reason that obeying the law is a necessary act of deference to divine authority. It stands at a higher level of generality because it is about features of the container, not features of the content.

The second issue is the questionable grounds that Klosko gives for his classification of the respective views. Recall, I agree that there is some use to distinguishing formal reasons from purely legal reasons and the theories that appeal to them. (I think the distinction is best understood as between purely legal reasons that claim to apply to all legal systems as such and formal reasons that apply only to some legal systems that meet substantive conditions, for example democratic ones.) Klosko, though, does not merely distinguish them; he also claims that formal reasons should be considered a part of content-dependent views. He argues that the moral force of formal reasons “passes through” these reasons to the content of the particular law, unlike purely legal reasons. This “passing through” is what makes formal reasons sufficiently like content-dependent reasons to allow content-dependent views of political obligation to appeal to them while remaining content-dependent views.

For example, Klosko's favored formal reasons are considerations of fair play. On my view, fair play reasons are content-independent because it is a feature of the container (the fact that the particular law is one part of the larger cooperative scheme that citizens benefit from) that determines the existence and weight of the reason, not any feature of the content of the particular law. Are they somehow content-dependent as well? Klosko writes:

There are two problems to address. The first is what moral force “passing through” exactly means. Consider again the law to stop at stop signs. It seems the idea is that if you have a duty of fair play to do your part in a cooperative scheme and the moral force of that passes through to the content of the law, then we can say that it is unfair for you not to stop. The content of the particular law has taken on the moral character of laws in general; that is, it has taken on considerations of fairness.

But this reason is still paradigmatically content-independent. If a citizen asks why he should stop in any particular case, the appeal to fairness is still disconnected from stopping. Why is unfair to not stop? Not because of anything to do with stopping per se. It is only unfair because of some feature of the particular law, the container, namely its place in a cooperative scheme. If stopping was not part of the cooperative scheme in this way, the moral force of unfairness would not have passed through. So the fair play theorist still has to answer the distinctive problem of content-independence, namely how to connect a particular act specified by the content to apparently extraneous features of the container. This will be markedly different from an argument showing why people have an obligation not to murder, for example, which can appeal directly to the features of murder. And so we are right back to the distinct role content-independent reasons play in the debate. Due to the fact that fairness is a feature of the container and so is independent of the merits of acting as specified by the content, anarchists such as Simmons have ample room to argue that fairness considerations cannot ground a general moral obligation and so do not necessarily pass through to every case of, for example, stopping at stop signs.Footnote ⁸⁹

The second problem is whether “passing through” actually distinguishes formal reasons from purely legal reasons. On the Pauline view of purely legal reasons, citizens have an obligation to obey because obedience to the ruler is obedience to God. Surely the moral force of this purely legal reason passes through to the particular law in some sense. If the king says you have to bow before him, his divinely delegated authority makes it the case that not bowing is not only illegal but sinful as a repudiation of divine authority. The moral force of not bowing thus takes on the character of the purely legal reason. If this is right (and we cannot really know without a much clearer understanding of “passing through”), then this is not a helpful way of distinguishing purely legal from formal reasons.

I conclude that we have good reasons to resist Klosko's attempt to include formal reasons in content-dependent views and so we should also reject his larger goal of refocusing the debate over political obligation by abandoning the concern with content-independent reasons.Footnote ⁹⁰ Formal reasons have the distinctive features of content-independent reasons, and the distinction Klosko draws between purely legal reasons and formal reasons does not show that formal reasons are somehow more content-dependent. Further, Klosko's discussion of content-independence mischaracterizes significant features of the modern debate over political obligation and political authority. The interesting question is: What features of laws other than their content could possibly ground a moral obligation to obey regardless of content? Answering this question requires grappling with the distinctive features of content-independent reasons, and solving this question allows one to capture the distinct benefits of abstracting from the controversial and contextual characteristics of content.

To briefly conclude: I argue that a reason is content-independent iff it has a container as a constituent part and changing the content of that container does not determine whether the agent has that reason and does not determine the weight of that reason. This understanding is distinctive in restricting content-independence and content-dependence to reasons that are containers and thereby clarifying how reasons can relate to content. This allows us to solve the vexing puzzle of how content-independent reasons (and especially obligations) can be consistent with the considered judgment that egregiously unjust commands cannot bind.

My account also avoids conceptual challenges to the coherence and usefulness of content-independence and can be used to analyze other content-independent reasons, such as those constituted by requests and promises. Finally, my explanation of content-independence clarifies what is at stake in the debate over political obligations and political authority. This simply clears the way for addressing the more interesting and difficult questions of whether and under what conditions laws actually constitute binding content-independent reasons.