Anscombe, Anarchism, and Authority

1.1 A Priori Anarchism

Robert Paul Wolff, perhaps the most ardent and prominent proponent of philosophical anarchism, attempts to show that there can be no legitimate political authorities in principle. An entire body of literature responds specifically to Wolff as the paradigmatic anarchist, and most regard his formulation of the Anarchist’s Dilemma as authoritative.⁴ In the contemporary literature, A. John Simmons and M.B.E. Smith represent positions closest to Wolff, though as we will see they adopt slightly different forms of anarchism.⁵

Philosophical anarchists articulate a wariness about someone having the right to be obeyed because obedience would consist in performing actions in compliance with an authority’s directive, rather than as the result of self-directed deliberation. Traditional philosophical anarchists put this conflict as follows:

Anarchist’s Dilemma:⁶

• (1)  

  A practical authority’s directive generates a duty to obey.

• (2)  

  Subjects have a duty to act autonomously.

• (3)  

  In obeying an authority, one thereby fails to act autonomously.⁷

The dilemma assumes that either an agent acts because the authority says so, in which case she acts obediently but heteronomously, thus not autonomously; or she does what the authority says for the reasons she thinks support doing so, in which case she acts autonomously but not obediently. So long as we assume that authority consists in a right to be obeyed and correlative duty of subjects to obey, the above entails that a right to be obeyed amounts to standing to demand that subjects violate their fundamental duty to act autonomously. Anarchists like Wolff argue that there can be no such right, thus, no legitimate political authorities.

It is worth spelling out the problem in more detail and looking at what traditional anarchists like Wolff and Raz have to say in defense of each premise. Because many contemporary anarchists, like A. John Simmons, see the problem differently. They argue that if a state has a right to be obeyed, that would potentially threaten a right to exercise individual autonomy. For a legitimate state might then issue a command that happens to conflict with a subject’s right to choose if the commanded act diverges from what the subject would have freely chosen. In turn, this undermines a subject’s ability in the particular case of conflict to “choose for themselves the life they wish to live without coercive interference” (Adams 2018: 90).⁸ But on this framing of the problem, the issue arises only after the fact of conflict of the command and what the autonomous subject would have chosen; thus this version of the view is called a posteriori anarchism (Wendt 2020: 39, 528).⁹

But Wolff, Raz, and traditional a priori anarchists’s concern with deference to authority is something that persists even if the state does not issue commands that conflict with what subjects judge to be choiceworthy.¹⁰ Rather, the very existence of an authority to whom one owes obedience conflicts with the duty to exercise autonomous choice:

In obeying the authority they abdicated their autonomy, says the philosophical anarchist, they abdicated their responsibility to decide on the balance of reasons themselves. (Raz 1990: 6)¹¹

The deep problem a priori anarchism identifies, then, is that obedience involves abdicating responsibility for one’s actions, and there is a non-overridable and fundamental duty we have as agents to take responsibility for our actions by choosing them for ourselves.

Pace the contemporary anarchists, then, the problem with practical deference to political authority has to do with a moral duty to decide for oneself what to do, not just a right that might be violated if the authority commands something incompatible with one’s autonomous choice. Wolff insists:

The fundamental assumption of moral philosophy is that men are responsible for their actions … Being able to choose how he acts makes a man responsible, but merely choosing is not in itself enough to constitute taking responsibility for one’s actions. Taking responsibility involves attempting to determine what one ought to do. (1990: 25)

Wolff (1990: 26) puts this in terms of the “duty to engage in the reflective process” of trying to “ascertain what is right.” A person who acts because the authority says to seems to circumvent the reflective process. This pits obedience against the norms governing agency in a profound way.¹² And Raz (1988: 39) maintains that genuine obedience to authority consists in “a surrender of judgment.”¹³ This surrender supposedly undermines our responsibility for the actions we perform on the basis of another’s command.

As I mentioned at the outset, we might think that a state can fail to be legitimate and so have authority over subjects for a myriad of other reasons, even if it does not require this surrender of judgment. I want to be clear that the puzzle we get from the traditional anarchists is not the only serious objection to political authority. But in my view, it is perhaps one of the most daunting threats because it suggests there is something conceptually confused about the notion of a duty to obey the state, given the fundamental duty to act on our own judgment and reason.

1.2 Moral Testimony Pessimism

The concern that lies at the heart of a priori anarchism also emerges in recent arguments for pessimism about moral testimony.¹⁴ Moral testimony pessimists generally argue for the asymmetry thesis, namely, that “while testimony is a perfectly fine source for non-moral belief, there is something problematic about basing one‘s moral beliefs on it” (McShane 2021: 134). In nonmoral cases where I do not know what to believe and do not have sufficient evidence to figure out what to believe, I may have most reason to seek out and defer to the testimony of an expert. In moral cases, though, I have reasons to not defer but figure things out for myself.¹⁵

The moral testimony we will focus on here explains the asymmetry by claiming that practical deference exhibits something defective in a person’s agency. For instance, one virtue-ethical argument for pessimism starts with the idea that agents ideally are related to their own beliefs in a special way, such that when they act on those beliefs, their actions express their character and agency. Robert Howell defends the asymmetry based on this idea, arguing that we bear special responsibility for our beliefs in the moral domain: what we believe and do in this domain is supposed to be a function of our character, whereas what we belief and subsequently do in other domains need not be so closely tied to our character. When a person defers to another’s moral testimony, she severs that connection between her own beliefs and action, on the one hand, and her character, on the other. He asserts, “We can push off the work of knowing and discovering non-normative facts to the relevant experts, but we are each of us responsible for the knowledge and discovery of moral facts” (2014: 401).

Again the intuition about agency and responsibility is at work. If we surrender our moral judgment to another, we may act on the reasons that make an action right but still fail to be connected to the action in the right way. We give up that connection when we avoid taking responsibility for the judgment about what to do, and thus what is done. Or so says Howell. The act of deferring will “bypass moral character in that the beliefs and actions do not stem from the agent’s present virtues”; as a result of the deference, the agent may have “a virtuous belief, and [perform] a virtuous response, without the agent possessing the relevant virtue” (2014: 403). Moreover, the belief acquired on testimony threatens to stunt further moral growth if it “fails to integrate with the rest of the agent’s present beliefs” and cannot serve as a ground for further belief formation or change.

Other pessimists put the concern more broadly, arguing that deference exhibits some failure of our practical agency. When we accept a moral claim on testimony, they say, we fail to act based on our own perception of and responsiveness to the reasons and requirements our moral belief is about. But this is to shirk responsibility for what we are doing.

Alison Hills (2013) illustrates the idea with the case of Eleanor, someone contemplating becoming vegetarian. Eleanor talks to a friend who says that eating meat is wrong and directly believes her. Hills argues that Eleanor may have the right belief, but for the wrong reasons; she lacks understanding of why eating meat is wrong and subsequently will not be able to apply her knowledge correctly in similar but novel circumstances.¹⁶ Her testimony-based belief may be correct, but it does not inform or exhibit her agency in the right way.

Roger Crisp also suggests the person who forms moral beliefs based on testimony displays defect in their agency. He argues that “for an adult to form a view on this important question on the basis of testimony strikes me as, arguably, epistemically and morally defective. Even doing this on a single issue is detrimental to autonomy, unless we set the standards for rational self-government quite low” (2014: 138, my emphasis). The talk of the “standards for rational self-government” conjure up the notion of a norm or duty, rather than a permission or a right.

We hear echoes of a priori anarchism in the refrain unifying these moral testimony pessimists’ complaints. The agent relying on moral testimony fails to “do one’s job” when it comes to acquiring moral beliefs, integrating them into one’s own character, and acting from that character.¹⁷ A person who defers to someone else about what ought to be done or avoided relinquishes something of her agency in letting her belief and subsequent action be determined by the testifier’s say-so. An action she performs on the basis of a belief acquired by testimony will lack a proper connection to her will, or agency, or character. She thus evades the kind of responsibility for her action that she would bear ordinarily.

1.3 The Responsibility Problem

A common concern underwrites both one prominent form of moral testimony pessimism and a priori anarchism.¹⁸ I will call this the Responsibility Problem: Believing or acting out of practical deference contravenes a duty we have to make up our mind for ourselves and act from judgments we formed, in such a way as to be responsible for what is done. Any demand that conflicts with that duty, whether a demand to be believed or to be obeyed, then, will be problematic.

If I am right about this being the central complaint about practical deference of the a priori anarchist and certain moral testimony pessimists, then what follows? A satisfactory response to the Responsibility Problem should have two features. First, it should acknowledge in some way the importance of our responsibility for what we do and believe as agents. Second, it should convince us that a demand to be believed or obeyed, and the correlative duty to defer, can be reconciled with whatever fundamental duty flows from this fact about the importance of taking responsibility for our actions. Whether or not any given state is legitimate or positioned in other ways to demand obedience from us, or whether there are individuals positioned to demand belief from us, may depend on a host of other facts having to do with further conditions on authority. But fulfilling these further conditions will matter not at all if it turns out that a duty to obey cannot be made compatible with a fundamental duty to act autonomously.

2.1 The Teaching Authority Puzzle

The little-discussed essay “Authority in Morals” opens with a reflection on something perplexing about a mundane phenomenon—namely, that some people, like teachers and parents, seem to be in a position to demand that others believe them about “morals.” Here she uses the term “morals” in the broad and Thomistic sense—she glosses it later as “right and wrong, e.g. good and evil in conduct, of what is virtuous and vicious to do” (Anscombe 1981b: 46).”¹⁹ “Teaching authority,” she says, is “the right to declare to someone else what is true … and to demand that he accept what one says and act accordingly” (46).

On the one hand, nothing could be more commonplace than this kind of authority. We all grow up deferring to someone or other about what to believe about math and science and morals, even though eventually most of us move beyond that dependency (See, for instance, Wiland 2017). On the other hand, the people we believe about math and morals can get things wrong, and believing them when they are wrong might seem rather problematic. She suggests that most of us think this especially troubling in the case of teaching about morals:

Teaching Authority Puzzle:

• (1)  

  Teaching authority in morals is necessary.

• (2)  

  This gives rise to a right to be believed about moral matters, since “a necessity can hardly fail to be a right” (Anscombe 1981b: 43)

• (3)  

  What makes it correct to believe p is simply that p is true.

• (4)  

  Parents and other teachers are fallible about morals.

• (5)  

  Thus, “it looks both necessary and impossible that there should be teaching authority on the part of fallible people” when it comes to morals. (45)

In sum, we must but cannot have a right to be believed about moral matters, where this is construed as anything in the range of ordinary truths regarding what is to be done or avoided in human life.

Admittedly, premise 3 may strike some contemporary readers as foreign or implausible. But here Anscombe simply follows Aquinas (and Aristotle, NE VI.2) in saying that truth is the good of the intellect, and so a proposition’s being true makes it a right thing to believe (Aquinas, SCG I.59). This understanding of what we ought to believe, together with the fallibility of ordinary teachers of morals, generates her puzzle.

2.2 The Asymmetry Thesis?

Anscombe does two things in her subsequent discussion of this puzzle. First, she uses the puzzle to unearth and examine the impulse to think that beliefs about “morals” differ from those about math or science or other subjects, making teaching authority about morals especially problematic. Second, she considers at length the kind of view we have been discussing, namely that we have a fundamental duty to form our own beliefs insofar as they involve what ought and ought not to be done, so a right on the part of a teacher to be believed will generate a contravening duty.

In just a few paragraphs, Anscombe countenances and attempts to dismantle three reasons why someone might think that teaching authority is exceptionally problematic when it comes to morals. In doing so, she anticipates explanations for the asymmetry thesis that have surfaced in the contemporary moral testimony literature, though she finds them all wanting.

First, we might think that there is a sort of special danger that awaits a person who receives moral teaching from a bad teacher. “We sometimes imagine someone with a terribly bad upbringing, who is taught all sorts of misbehaviour as right, and taught to despise much that is good, and we think: what about such a person?” By contrast, one rather bad teacher in math does not a hopeless mathematician make, since a student will likely quickly be corrected either by some other teacher or by experience. Anscombe thinks this explanation of asymmetry exaggerates the effects of bad moral upbringing. Instead, even people with a bad upbringing “know quite well how to cry out against injustice and lying and treachery,” and thus even the students of the worst moral teachers “know quite a lot” (1981b: 45). So this cannot be the ground of a genuine asymmetry.

Another tempting suggestion is that ethical matters are more necessary to human life and wellbeing than other subjects. We may think that children need to be taught morals, in a way that they do not need to be taught organic chemistry or particle physics, to flourish. But Anscombe points out that plenty of other subject matters—say, basic math, or health—need to be learned, too, for humans to live well. This explanation of the asymmetry will not do either.

Finally, we might think there are fewer moral experts, or that those in position to teach morals are more likely to be incorrect than their nonmoral counterparts. After all, teachers of other subjects typically operate in systems that hold them accountable to teaching what is correct: “A professional teacher” of a subject like mathematics, Anscombe says, “naturally cannot justifiably claim his teaching commission in support of his teaching when what he teaches is untrue” (1981b: 44). But those in a de facto position to demand that their subjects believe them about moral matters —parents, religious educators, caregivers—are more likely to get it wrong and maintain their positions. “Morals, I suppose, are what is most universally taught by fallible teachers informally at least” (44). While Anscombe will end up agreeing that moral teachers are fallible, she does not think this is unique to the subject matter of morals. So this explanation of an asymmetry cannot generate a distinctive puzzle about teaching authority in morals.

3.1 The Mistaken “Kantian” Ideal of Belief Formation

Anscombe begins the second half of the essay by acknowledging that a commitment regarding what morality fundamentally consists in might support the belief that morals are not teachable the way other subjects are. Specifically, she diagnosis a recalcitrant belief in the asymmetry thesis as stemming from the view, which she attributes to Kant, that “one’s morality must be something one has formulated for oneself,” and that “to take one’s morality from someone else, that turns it into a bastard sort of morality, marked by heteronomy” (1981b: 45) On the so-called Kantian view, “one’s morality must be something one has formulated for oneself, seeing the rightness and wrongness of each of the things one judges to be right or wrong.”

Consider an illustration of the “Kantian” ideal:

Elias and AI: Elias, an undergraduate student, hears his friends talking about using generative AI to write essays. He is in danger of failing his literature class and so considering using generative AI to write his next paper. He calls his mom and she tells him it would be wrong to use AI to write even part of his essay.

Maybe Elias simply believes her but without appreciating why it would be wrong—for instance, that it would count as lying to submit the essay as his own, or that it would constitute cheating based on the university’s newly adopted policy on AI. If Elias does not see why using AI would be wrong, the belief that it is wrong fails to be a proper moral belief on the “Kantian” view. For Elias to have a properly moral belief, he would need to consider whether he could conceive of or will a world in which he and everyone else uses the generative AI in this way and conclude he could not.

Notice how such a thought gives rise to the idea that fulfilling one’s basic duty to act autonomously cannot be compatible with fulfilling a duty to obey or believe a practical authority. An action done for a maxim one gets from someone else is heteronomous, and thereby not a moral act, on the so-called Kantian view. So we could never act morally by either forming a belief on the basis of which we act by testimony, or by acting because it has been commanded. This is the problem with surrendering one’s judgment about what to do, as Raz put it.

3.2 Practical Deference as a Human Action

In response, Anscombe argues that the distinction between moral and nonmoral (or immoral) does not track the distinction between acting because of thinking a thing “on the strength of what someone else says” and acting because of thinking it “left to oneself” (1981b: 46). Thus, deference to teaching authority in morals does not end up violating some fundamental and non-overridable norm of human agency. In fact, Anscombe will go so far as to say that it can be dangerously misguided and psychologically unrealistic to attempt to act always and only on the basis of only one’s own “unaided” judgment, even when one worries about the fallibility of others who might offer moral testimony.²⁰

After laying out the so-called Kantian worry, Anscombe asks us to consider how we form beliefs not just about morals but in other areas of life as adults: by some combination of relying on others and our own judgment. She takes the case of memory to be instructive:

I can make no judgment about the past without my own memory. But only a fool thinks that his own memory is the last word … Of course [a man] would not have any basis for such judgments if he did not already rely on his own memory to some extent, but it would not be reasonable to argue from this that his own memory must after all be for him the last word about what has happened in the past. (1981b: 46)

For example:

Faculty Attendance: My colleague asks whether I attended all of last year’s faculty meetings. Initially, I try to recall each of the monthly meetings; then I worry I’ve forgotten whether I attended in December. I remember there was a fastidious note-taker at the meetings and decide to check the minutes.

It would border on absurdity for my colleague to protest, “But I want to know what you think!” After all, I am relying on my memory of the note-taker to get the information. I am still at least partly the source of the belief when I rely on someone else’s notes.

So too with beliefs about “practical matters,” or morals, in Anscombe’s terms. An individual “must put some reliance on his own conscience: that is to say, on those judgments of right and wrong which he makes for himself” (1981b: 46). But it would be a fool’s errand to try to insulate that conscientious judgment from all other sources of input.

The main claim, then, is that when we engage in practical deference—that is, when we believe someone with a view to our own action—we do something from our own character and agency. It is something we do in the distinctively human way—a human action. And as such, we can do it well or badly, we are responsible for it, and our act can be evaluated as right or wrong, good or evil, virtuous or vicious. After all, she explains, while it is possible to believe or obey someone “without reflection, consideration or interpretation” (1981b: 48), when we act, we do something for which we are responsible. For “doing what one is told is an interpretation and so with doing, however obedient one is, one can hardly escape being one’s own pilot” (48, emphasis mine).

Further, she says, “There is a sense in which only the individual can make his decisions as to what to do, even if his decision is to abide by someone else’s orders or advice” (1981b: 48). I suggest we read her as insisting that, while what we do may differ depending on whether we rely on testimony, or act in obedience, or act from a belief we form “unaided,” in each and every case we are doing something intentional. It is not as though the act of forming a belief unaided is any more an exercise of our agency than the act of relying on someone’s testimony about morals. In each case, we perform some human act or other for which we can be held to account. And so it is rather absurd to think that practical deference results in the believer “getting off the hook.”

3.3 Deferring Well or Badly

At this point we have learned that on Anscombe’s picture, believing someone else with a view to acting, that is, in the practical mode, is a human action. She aims to convince her audience that not only is believing moral testimony something for which we bear responsibility and can be evaluated, but also that it can be done well or badly. That is, good or moral action does not always consist in action done on the basis of judgments or beliefs formed “unaided,” and nonmoral or bad action does not always result from judgments or beliefs formed by testimony. The norms governing human action, then, do not bottom out in some norm about acting and judging for oneself.

Imagine the following:

Evan the Ethical Vegetarian: Evan reads an article on ethical vegetarianism in his philosophy class. He can follow the argument in defense of becoming vegetarian. In class, some fellow students raise objections that seem equally compelling. Evan thinks he is not well positioned to know which view is correct, so asks his professor what she thinks.

On Anscombe’s account, it would be natural and unobjectionable for Evan to judge that he should be a vegetarian on the strength of her say-so, even though he could not on his own assess specific arguments for and against it. Evan’s case parallels the Faculty Attendance case. Just as it would be a misuse of my intellectual agency to refuse to consult the notes about my attendance, it would be a misuse of Evan’s practical agency to refuse to consult the opinion of the person he considers an expert about ethical eating.

Some moral testimony pessimists think that offloading a question about what to do in such a case constitutes a failure of agency, or will at the very least show that Evan lacks a key component of virtue. But Anscombe’s memory analogy helps give the lie to this assessment. Not only is Evan plenty involved in his own belief formation; but it might be foolish, or vicious, of Evan to refuse to defer to his professor. Consulting another person, or someone’s notes, is a choice, after all. Moreover, it is a choice that one can only make by deploying one’s own agency, just as figuring out who to ask about the forgotten information itself requires deploying one’s own memory (e.g. remembering who the good note-taker at the meeting was).

So now we can ask how good or bad, defective or permissible Evan’s practical deference is. And the answer, Anscombe wants us to see, will not be settled by whether Evan was the ultimate and sole source of the judgment on which he acts. Instead, our evaluation of Evan’s act of deference comes down to how Evan engages in deference, the circumstances surrounding his deferences, and what kind of character he acted from in deferring.

Suppose Evan’s classmate Devan does not read the assigned pieces on ethical vegetarianism and comes to class ready to defer not out of a concern for the truth and conviction that the professor knows better, but out of laziness. She might even defer without it occurring to her that she is believing a moral expert or deferring; instead thinking she is just going along with what she hears. But if these are recognizably coextensive descriptions to Devan, then we can hold her accountable for forming the belief and acting on it in a way that is defective (perhaps because it is an act of sloth).

To see the plausibility of this, we can look to what Anscombe says about the class of human actions we perform from “knowledge without realisation.” Consider the example of St. Peter’s denial of Christ from Intention. In the gospel story, Christ says to Peter that he will deny him before the rooster crows. Peter insists that he would never do such a thing. Anscombe writes:

And St. Peter might perhaps have calculated ‘Since he says it, it is true’; and yet said ‘I will not do it’. The possibility in this case arises from ignorance as to the way in which the prophecy would be fulfilled; thus St. Peter could do what he intended not to do, without changing his mind, and yet do it intentionally. (Anscombe 2000: 94)

St. Peter would not describe himself as denying Christ, but he is not prevented from acting voluntarily or coerced when he tells three people that night that he never knew him. Nor is his denial a function of some feature of his circumstance he couldn’t have known about. As John Schwenkler (2019: 208–209) helpfully explains, “While Peter does not bring his actions explicitly under the concept ‘denial,’ at the time he speaks he has practical knowledge of what he is doing under a description that he understands to be equivalent to it: for he knows that he is telling someone that he is not a follower of Jesus.”²¹

We learn from St. Peter’s denial that what a person ends up being responsible for goes beyond what she explicitly entertains in thought prior to performing some action. It includes descriptions the agent “understands to be equivalent” to what is done intentionally. Thus, if Devan can see that declining to think through the arguments for vegetarianism herself is an equivalent description of her going along with what the professor says, she is on the hook for both the former and the latter.

With this, we can say that the goodness or badness of belief formation regarding morals is orthogonal to whether it is done by deference to testimony or unaided. In fact, she points out, if a theory were to offer a blanket vindication for any action according to one’s unaided conscience, then a person could too “easily get away with having thought good what was bad and bad what was good, and acted accordingly” (Anscombe 1981b: 47). A person does not necessarily display more seriousness about doing what she ought to do by being more meticulous about following her own sense of duty over and against what others say is to be done or not done: “an endlessly conscientious style in one’s behavior … endless bellyaching,” she says, “isn’t seriousness” (47).

The arguments above undermine one assumption that generates the Teaching Authority Puzzle, namely, that teaching authority in morals is impossible. The best defense of that premise, in her view, is the Kantian ideal according to which one simply cannot do anything moral unless one forms the judgment on which one acts for oneself and by oneself. She offers compelling reasons for rejecting that ideal. If, then, the best argument for the impossibility of teaching authority in morals fails, then we do not have warrant to think such teaching authority is “both necessary and impossible.”

4.1 No Per Se Revealed Moral Truths

If certain moral truths were per se revealed—that is, “are revealed … could not be known otherwise,” then there would be teaching authorities in morals who possessed a right to be believed in each and every case of testimony (Anscombe 1981b: 49). In other words, if a person were in possession of moral truths that others could not infer, intuit, or grasp independently, then such a person would be in a position to demand belief from the moral learner about each discrete claim she communicated, and the learner would be under a perpetual duty to believe her.

Anscombe (1981b: 49) argues that no human being possesses this form of teaching authority because there could be no per se revealed moral truths, in principle:

There is no such thing as a revelation that such-and-such is good or bad not for any reason, not because of any facts, not because of any hopes or prospects but simply: such-and-such is good to do, this is to be believed, and could not be known or inferred from anything else.

This is quite a strong claim, and a bit surprising coming from Anscombe, who is Roman Catholic. Wouldn’t Anscombe assume that some religious dogmas are exactly examples of such things?

Furthermore, putting to the side religious dogmas, we still might think there are other cases where certain moral truths have to be revealed by testimony, and that the testifier about such truths would be in a position to demand belief about these claims. For example:

Bystander Brent: Imagine my coworker, a heterosexual man, routinely makes remarks to me when we are alone that make me feel uncomfortable. He never does this when others are around, and so his inappropriate behavior is not something others could think out for themselves. Further, our other coworker, Brent, is also a man, and has never navigated a workplace in a female body so cannot anticipate that certain remarks would cause discomfort.

The fact that the remarks constitute discrimination is something Brent cannot work out for himself, it seems. In this case, it cannot be the case that the best moral course of action is to rely on our own judgments about right and wrong, independent of input from any others. It seems like I might have the right to demand not just that Brent believe me in general about such moral matters, but that he believe me in each and every case of my testimony to him.

Anscombe argues against such a possibility for two reasons. First, she says, moral beliefs acquired on this sort of authority would never become part of the kind “characteristic of the understanding of a virtuous person” (1981b: 50).²² If grasping such truths were truly a matter of morals—part of virtue or being a good human being—then it would be both impossible and necessary for us to grasp them. (For recall on the Thomistic Aristotelian notion of “moral,” the moral is demarcated by Aristotelian necessities—claims about what is needed to flourish and act well, and so anything that gets folded into what is needed becomes part of the moral.)

Second, and relatedly, if there is no in principle reason from which a person like Brent could infer the truth in question, then Brent would never be able to apply the concepts of discrimination or harassment flexibly. If the fact that some remark is discriminatory is per se revealed, then at best, I or another woman will have to point out to Brent every case of discriminatory speech or behavior because the men cannot learn to recognize a pattern from the outside.²³ But if there is a pattern, then in principle, someone like Brent could learn to discern it. And then it would not be per se revealed.

Anscombe concludes that we must reinterpret cases that look like per se revealed moral truths as per accidens revealed. That is, either the truth is not a moral truth—not something to do essentially with living a human life well—or the person who does not already have the belief about the moral matter in question “could have thought it out for himself” (1981b: 49). In many cases, getting to the belief by testimony is just an ordinary human act, not morally significant but not defective. But if the belief is really a moral belief in her sense, then it must be the sort of thing one could be evaluated as defective or admirable for acquiring, and so there must be a way for human beings in general to acquire it.

This allows for cases where “some of the facts, of what is the case, will help to determine moral truth” and “some such truths about what is the case are revealed,” but they are simply not moral truths (Anscombe 1981b: 49). For example, the permissibility of religious ascetism might seem to be “morbid” on a reasonable view of human life, but religious revelation might make intelligible the goodness of moderate ascetism by providing nonmoral, metaphysical facts about the human soul or eternal life from which we can work out for ourselves the consequences of a revealed nonmoral truth. When we draw a moral conclusion—the implications of the nonmoral truth for our own action—it is properly our own. No moral truth is in itself the sort of thing that has to be sourced wholesale from another. If this is correct, then there are no teaching authorities who possess a distributed right to be believed. And any individual learner or agent has the latitude to make up her own mind about whether to believe a teaching authority in morals about a particular claim. This leaves us permission always to think for ourselves, even when being regularly disposed to believe the person in the position of authority.

4.2 Necessary Teaching Authority

Given the Thomistic dictum that the intellect aims at truth, the fact of our general fallibility about morals implies that no teaching authority could ever be positioned to demand to be believed about every single particular proposition. A parent or teacher cannot have the right to complain, “You did not believe what I said, and that is bad, because it was I, whom you ought to believe, who told you,” she explains. Instead, what teachers and parents have a right to demand is that they not be routinely or regularly disbelieved (1981b: 53).²⁴

Anscombe thus gives us further reason to reject the conclusion of the Teaching Authority Puzzle. For it assumes that teaching authority in morals involves a distributed right to be believed—that is, a right that distributes to each and every instance—and that this is necessary for moral education. But what ends up being crucial to moral education, on her view, is the right to be obeyed.

We learn morals like we learn math, she argues. Just as “one does not learn mathematics … by learning that certain propositions … are true, but by working out their proofs. Similarly, one does not learn morality by learning that certain propositions—ethical ones—are true, but by learning what to do or abstain from in particular situations and getting by practice to do certain things” (1981b: 47).

In Anscombe’s view, teaching authorities in morals do not carry a right to be believed in every instance, nor do they need such a right to accomplish the task of moral education. “The right that a fallible teacher has, in that he has authority, then, is the right that those he has to teach should be generally prepared to believe their teachers” (1981b: 45). Even more important than this is the regular obedience of subjects. For their obedience will get them to practice the very actions that will yield, eventually, an appreciation of what is good and bad, right and wrong. In the meantime, we do nothing defective as agents when we take on beliefs because of a teacher’s say so, with a view to our own action. For as we saw earlier, we maintain responsibility for whether and how we defer.

5.1 The Impossibility Reading

On the first reading—call this the impossibility reading—Anscombe thinks it is impossible to obey an authority without becoming liable for that act of obedience because of the nature of such an action. We can class obeying as a type of action, like “denying” or “promising” or “marrying,” which cannot be understood without reference to the practical knowledge of the person performing it. A person who has been deceived that the ceremony is only a rehearsal is not marrying at all. Anton Ford explains, for Anscombe, we cannot understand intentional actions like this—and what differentiates them from mere events or bodily movements—without reference to the practical knowledge the agent has of what she does. It’s not as though you and your evil twin did the same thing, but she did it “with intent,” and you did not. We cannot even give the true description of what it was you did or what it was the twin did without appreciating whether your knowledge of what you were doing was practical—that is, the very thing that produced the action.

If obeying is this sort of intentional action, essentially, then the person performing an act of obedience does it with practical knowledge. We might acquire a theoretical belief about morals from another person, entertaining the proposition in a theoretical mode. But this sort of thought differs formally from engaging the belief with a view to action in practical thought. When we engage such a belief in the practical mode, it becomes part of our practical knowledge and produces what we do. An agent’s practical knowledge, as we saw in the case of Peter’s Denial, fixes what she is responsible for. So there is no worry that a person’s obedience can absolve her for responsibility for what she does.

In favor of the impossibility reading, consider first this passage from “Authority in Morals”:

There is indeed a sense in which only the individual can make his own decisions as to what to do, even if his decision is to abide by someone else’s orders or advice. For it is he who acts and therefore makes the final application of whatever is said to him. It may be said, concerning his judgments in the field of theoretical morals, that in the same way it is he who thinks what he thinks, and so too only he can make his own decisions what to think. No doubt: but here there such a thing as believing what he is told without reflection; doing what is told is an interpretation and so with doing, however obedient one is, one can hardly escape being one’s own pilot. (1981b: 48, emphases mine)

Anscombe appears to suggest that if an adviser, teacher, or parent tells you that some action is wrong, you cannot help but be the one who applies that description to your own particular action, if you engage the belief in a practical mode. Even if the teacher can induce a sort of passive acceptance of a proposition, the gap between believing the act is wrong and intentionally refraining cannot be filled by anything the authority does. Believing it “with a view to action”—that is, having a moral belief that has taken the form of practical thought, not just theoretical thought—implicates in some measure the person’s own agency.

Further, look at a claim she makes later in the passage: “A man may decline to be his own pilot for certain of his beliefs … this would be possible in moral matters only to the extent that his beliefs were idle, without consequences, i.e. if they concerned matters that he never had to deal with” (1981b: 48). Anscombe thinks that, once someone like Evan walks outside the classroom and starts preparing himself dinner, the theoretical moral belief is no longer “idle.” It figures in what he is doing. As he goes about the kitchen cutting and cooking vegetables and legumes, his roommate might ask, why are you doing that? And when he offers a reason like, “I’m going vegetarian!” or “I’m avoiding eating meat,” this makes sense because the moral belief about vegetarianism is no longer a merely theoretical belief about what is right. The same content now functions as part of Evan’s practical knowledge—his understanding of what he is doing for the sake of some aim or in light of something he desires.

5.2 Idle and Practical Beliefs About Morals

When Anscombe makes the distinction between “idle” and practical beliefs in this essay, she draws on a distinction she employs frequently elsewhere. For instance, in “Thought and Action in Aristotle,” she points out that practical syllogisms differ from theoretical syllogisms, and practical thought from theoretical thought, because they take a different form altogether. The example there discussed, comes from Movement of Animals:

“I need a covering.

A cloak is a covering,

I need a cloak;

I must make what I need,

I need a cloak,

I must make a cloak.” (1981a: 73)

If we read the conclusion as “it is necessary that I make a cloak,” the inference is obviously invalid: there may be many coverings besides a cloak. She explains, “You do not get this [necessitation] where various ways of obtaining the end are possible.” As a piece of theoretical reasoning, then, this is no good. But as a piece of practical reasoning, it is unimpeachable. If I go about making a cloak, having recognized all of the foregoing, I act reasonably. This signals that the very same contents can be taken up in two different modes of thought and that they have different formal properties.²⁵ So too a “moral” belief considered in a theoretical mode and one taken up with a view to action, in a practical mode.²⁶

Most importantly for our purposes, when an authority either tells us what to believe about what to do, or tells us what to do, there remains a gap between that say-so and what the obedient agent thinks and does. That gap can be filled only by the subject’s own practical reasoning—her incorporation of the thought or command into her own aims. And this gets the subject on the hook for what is done in a meaningful way.

The authority who tells someone what to do can at best transmit something thought in a theoretical mode to the subject. Anscombe says that two people can entertain all of the same premises, but if one entertains them in the practical mode and the other in the theoretical mode, the conclusion will differ. For instance, consider:

“Owning a Launderette would make me wealthy.

There is scope for opening a Launderette in such and such a place.”

Imagine my financial advisor and I are entertaining these propositions. If “I actively aim at being rich and am working out this one of the many possibilities with a view to action,” then I will conclude the reasoning by doing a thing—opening a Launderette there. My financial advisor, by contrast, “might be doing it idly, or as an academic example,” in which case the conclusion will not be any action (1981a: 74). My financial advisor can supply me with all sorts of premises—about the right course of action, or what would be for my good, what I should do here and now even—but he cannot make it the case that it is my aim to be rich, and thus that I conclude from his advice or command with the act of opening the Launderette. While the content of the reasoning does not contain a desire, the form of practical reasoning presupposes it. And thus it is simply not possible for an authority to get me all the way to action in obedience without some contribution of my agency in the form of my own aims. This way of implicating my agency keeps me on the hook for what I do, and indeed affects the proper description of the intentional description of the act for which I bear responsibility.

The Impossibility Reading aligns naturally with Anscombe’s discussion of aims earlier in “Authority in Morals.” When explaining why, unavoidably, a person contributes something to her action even when what she does is “to abide by someone else’s orders or advice” (1981b: 48), she says, “The human wants things like health and happiness and science and fair repute and virtue and prosperity, he does not simply want e.g. that such-and-such a thing should be in such-and-such a place at such-and-such a time.” We converse and argue about generalities, and about when and where they apply and what it would look like for them to be applied in particular circumstances. But when we act in the characteristically human way, we integrate generalities with our particular circumstances with such aims in mind—whether it is health, happiness, knowledge, or virtue. This will be true even in the case of obedience, and certainly in the case of acting based on moral beliefs gleaned by testimony.

Just as the person who promises has knowledge of what she is doing that is non-observational but produces what she does, so too Evan’s knowledge that he is making a vegetarian meal or doing what the professor said produces the action “making a vegetarian meal” or “obeying the professor.” If tempted to doubt that obedience must involve practical knowledge, consider a case where someone does what an authority says without practical knowledge:

Anwar’s act: Anwar has been told by his father that he must refrain from eating before sundown during Ramadan. Anwar goes on a hiking trip and, halfway up the trail, realizes that he has forgotten to pack his snack and lunch. He cannot get to base until after sundown. In frustration, he says, “I am doing exactly what my parents wanted— fasting!” Here he knows that he is doing what his parents say to do on the basis of observation. This knowledge did not bring it about that he did what they told him to.

Surely Anwar’s act does not count as obedience. Among Anwar’s human acts—his intentional actions—we can enumerate hiking, crying out in distress, but not obeying. Anwar’s foregoing food for the day is not something he does with practical knowledge (or knowledge without realization). It is not among the things for which we would hold him responsible or praiseworthy, for there is no description of what he does with practical knowledge that is recognizable to him as coextensive with obedience.

Generalizing, proper acts of obedience are human acts, acts done from practical knowledge. If someone’s behavior conforms to what an authority says but without practical knowledge—like Anwar’s fasting—it is not obedience. If an act is a proper act of obedience, it is at the least done from practical knowledge—like Devan’s vegetarian eating. But any action done with practical knowledge is action for which the subject is responsible.

Plausibly, then, we can understand Anscombe’s suggestion that one cannot escape being one’s own pilot as a claim that it is impossible to evade responsibility for acts of obedience. This will be so because once a theoretical moral belief comes to be about one’s own actions, it becomes part of one’s practical knowledge. Practical knowledge alone can produce intentional action under a description and give the description of what an agent bears responsibility for. To obey at all, you have to know what you are doing and that knowledge makes the act, properly described, an act of obedience and susceptible to evaluation. It carries with it responsibility for what is done. There will be no conflict between a duty to obey and a duty to act autonomously in this sense.

5.3 The Aristotelian Necessity Reading

The Impossibility Reading suggests that the “cannot” in “one cannot escape being one’s own pilot” be understood as metaphysical impossibility. Some might find this view implausible, though; for it seems that we can fail to think about what we are doing, or can be so self-deceived that they “know not what they do,” while being accurately described as obeying.

Even if it turns out to be possible to act in a way that deserves the name “obeying” without retaining responsibility—say, because one does it so unthinkingly—we can interpret Anscombe as saying that this cannot be the typical way obedience and authority work. On this reading, the sort of obedience that is needed for ordinary life and moral education to occur, in the sense of Aristotelian necessities, cannot involve acting in this avoidant way. While this reading leaves open the possibility of deviant cases, where a subject obeying so unreflectively that she maintains no responsibility for what is done, obedience in the non-defective, non-deviant case will constitute human action done for an end, for which we can be held responsible.

To get a handle on this interpretation of Anscombe, we first need to characterize the phenomenon of acting or obeying unthinkingly in a way that counts as an escape from being one’s own pilot. Here I want to draw on Barislav Marušić’s description of what he calls epistemic evasion:

As agents, we can, and should, take a different view of our future than mere observers would—even if those observers are as well informed as we are. We might say, echoing Kant, that we act under the idea of freedom. When it is up to us to do something, or perhaps when we (rationally) believe that it is up to us to do something, we should decide what to do, not seek to predict what we will do; we should engage in practical reasoning, not theoretical reasoning. (Marušić 2013: 308–309)

Let me illustrate the stance on our future that Marušić cautions against with a case:

Flora at the florist: Suppose Flora suffers acute depression and forms a belief that if she goes to therapy, she will buy flowers for her partner on the way home. She believes this not because of adopting the end of buying flowers, and thinking of therapy as the means to achieve it, but instead because she notices that the last several times she went to therapy she bought flowers from the street vendor on the way home. Flora goes to therapy, and just as she predicted, she finds herself stopping at the flower stand with a vacant expression, ringing up a bouquet before going home to her partner.

Perhaps in some limited way Flora knows what she does. But she appears to know it by observation. The kind of action performed in this depressive malaise actually alienates Flora from her own agency. Taking an observer’s stance towards what one will do in the future is not paradigmatic of human actions.

As practical agents, Marušić argues, we have a duty to take a certain view of our future actions that we do not take towards future occurrences that are not at all up to us. That is, we should not treat the question of what we will do in a purely evidential or predictive way. When we take up such a predictive stance to our actions, we “deny or distort” our own agency in such actions (2013: esp. §IV). When our knowledge is predictive and not productive of our action, the resultant action is uncharacteristic of human life. It does not fit Aristotelian necessities like, “the human takes the means to ends she adopts” or “the human acts so as to benefit her close associates.”

Some of Anscombe’s remarks in “Authority in Morals” indicate she thinks it is defective, but not metaphysically impossible, to obey an authority completely unthinkingly. Perhaps what she has in mind is a person who acts in a way out of step with typical human action, through something like epistemic evasion. She may avoid responsibility for what she does, because she somehow takes a purely predictive and objective stance towards her action rather than producing the obedience by her practical knowledge of what she does.

Even if it is possible for humans to obey an authority in this way, though, this cannot be how obedience in human life characteristically goes or should go. Nor is it the kind of action or obedience an authority has a right to. For it would generate a duty to act in a way that is inhuman—that is, to engage our agency in a way that undermines what is typical and needed for a flourishing human life. If that is right, then there can be no duty to obey in a way that relinquishing one’s own responsibility for the obedience, even if by epistemic evasion or some other deviant mental cause a person can obey “unthinkingly” and without a view to any end she has adopted or averred as choiceworthy.

On this reading, the “cannot” in “one cannot escape being one’s own pilot” should be read as a stopping modal, indicating practical or normative necessity rather than metaphysical necessity. Stopping modals like “can’t” signify what is out of bounds given some rule or norm or end the person acting has. They do not tell us what is metaphysically impossible. But they do tell us what is impossible if one is to achieve an end, or play a game, or follow a rule. Interpreting “cannot” as a stopping modal would amount to reading Anscombe as claiming that, so long as we want to be performing human actions, and doing the things that are necessary for human beings in the Aristotelian sense, escaping being one’s own pilot (e.g. through epistemic evasion) is off the table.

At the end of the essay, we get another indication that this may be the sense of the bold modal claim about being one’s own pilot. There she claims that asking someone to believe us about a per se revealed moral truth would be like “instruct[ing] an archer to aim at an unseen target” (1981b: 50). Of course, it is metaphysically possible for an archer to hit the mark without seeing the target. But then the archer would be said to have “hit” the mark not in the focal sense, but in a derivative sense. He did not quite stumble onto the mark, since he knew he was shooting an arrow, but he also did not shoot aiming at the mark in the proper sense.²⁷ So, perhaps, the person who manages to obey without contributing their own practical knowledge, and thus being responsible for what they do in obedience, is like the lucky archer who hits the unseen mark. We cannot understand “obedience” and “authority” primarily through such a case. Its intelligibility as an instance of obedience is parasitic on the case of a characteristic human action of obedience—the sort that is produced by practical knowledge that one is obeying, and thus for which one bears full responsibility.

Finally, on this reading, Anscombe’s view also provides for a correctness condition on commands that practical authorities issue. If an authority attempts to command unthinking obedience, then the authority’s order “falls to the ground” in her terms because it rests on faulty opinions about what is possible and what is the case. On the virtue reading, what such an order fails to recognize is that it is not possible for a subject to perform a practically reasonable or virtuous action and defer, or obey, in the way the authority commands. Additionally, the authority has an erroneous grasp of what she is within her rights to demand of others and the sorts of duties her authority generates for subjects. A practical authority cannot make it the right thing to do to deceive oneself, or to fail in one’s practical reasoning, or to believe or act in ways that are incompatible with becoming virtuous. So there can be no genuine conflict between a duty to obey and a duty to maintain responsibility for one’s own actions. Nor can an authority have a right that would generate such a conflict.