1.3: Morality Here and There

1. Kant Among the Sherpas

A Sherpa man was once accused of murder by the younger brother of his alleged victim:1

Having been accused and brought to court under suspicion of murder, the accused was given the obligatory flogging — one that apparently left him crippled for life. . . . [E]ven under duress, he maintained his innocence; just as steadfastly his accuser pressed his demand for confession and the payment of an indemnity. . . . The court now had both the accuser and the accused flogged simultaneously, but both men stuck to their stories. The court, unable to reach a definitive solution, sent the case back to the prime minister. He recommended that the issue be resolved by throwing dice. . . . The accused won two out of three throws and was declared innocent. The accuser was required to pay a token gift to the accused and probably considered himself lucky not to have been tried for the murder himself.

It sounds like the Sherpa have a perverse, even perverted sense of justice. A man is flogged before trial; next, his accuser is flogged; his innocence is finally established by three throws of the dice; and then his accuser gives him a token gift. Oddly, everyone goes away satisfied, if not exactly happy.

Maybe the Sherpa are simply misinformed about psychology and metaphysics. They think that flogging is a means of extracting the truth from litigants, failing which, the truth can be extracted from the spirit world by divination. In fact, however, the Sherpa are not primarily interested in the truth:

We [Westerners] solve a crime by finding out exactly what happened. But among the Sherpas, this is the last thing anyone wants to happen. Precisely the opposite: the object of their conflict-resolving efforts is to make sure that the brute facts do not emerge, because this would make much more difficult the task at hand: arriving at a new socially defined reality which will allow social life to go on with minimal disruption.2

[O]ngoing social harmony is protected by a tacit agreement not to speak too plainly about certain aspects of reality. Since antagonists in most disputes have to live out the rest of their lives in proximity to each other, it is more useful to construct new versions of what happened that save face, reduce shame, and redescribe conflict as misunderstanding or in some other new way than to insist on the whole truth and let the chips fall where they may.3

Thus, the flogging in this case is part of “an effort to persuade the disputants, through intimidation and duress, into getting on with the process of agreeing on a story and settling the dispute”.4

The “tacit agreement” among the Sherpa “not to speak too plainly about certain aspects of reality” covers not only testimony in criminal trials but ordinary conversation about people’s motives and intentions:5

As a general rule, Sherpas are not very informative in response to questions of motive. If one asks why somebody did something, one often gets a one-word answer (‘merit’, ‘prestige’, ‘money’, etc.), or even a hostile response: ‘How should I know, we can’t see into other people’s heads?’

In this respect, the Sherpa observe what anthropologists now call “opacity of mind”,6 a rigorous avoidance of discourse about people’s mental states.

Another group who avoid discourse about mental states are the Ilongot of the Philippines:7

Among Ilongots, personality descriptions are extremely rare, as are strategic reckonings of motivation. Accounts of why particular persons acted as they did refer almost exclusively to public and political concerns — surprising actions giving rise to the despairing claim that “one can never know the hidden reaches of another’s heart.” In general, Ilongots do not discern intentions, trace responsibility, or reckon blame by asking if offenders “knew” that they wronged others through their actions.

Here is a third example, the Baining of Papua New Guinea:

The Baining exhibit a pervasive avoidance of modes of discourse about psychology. . . . They are reluctant to speculate about the personal motivations, actions, and feelings either of themselves or others. They do not offer interpretations of the meanings of the behavior and events around them in these terms.8

Informants do not readily speak or proclaim personal opinions about either their own actions or those of others. Evaluations of events do not invoke an internal, emotional explanation. The most common response to questions such as “Why did he do that?” is “I don’t know about him.”9

In these cultures, psychological explanation in terms of inner feelings and attitudes is avoided in favor of explanation in terms of roles and scenarios. The ethnographer of the Baining writes that “[they] are not prone to describe themselves or explain others in terms of personal experience or subjective states; their descriptions are much more dependent on aspects of social roles, interpersonal interaction, and the nature of social behavior and action.”10

This folk psychology — or lack of folk psychology — yields a distinctive conception of crime and punishment: punishment does not aim at reforming the inner person: “What matters to disputants in the end is not the kind of moral change we seek within the criminal or guilty human heart, but . . . the establishment of bonds of kinship wherein all violent, selfish, and disruptive acts are seen as ‘shameful’ and at odds with an assumption of cooperation, ‘sameness,’ and autonomy.”11

An emphasis on shame is common in cultures that avoid mind-reading in the interest of social harmony; for unlike guilt, shame is directed at outward appearance and behavior irrespective of motive and intention. Of necessity, however, shame in these cultures is not the private emotion that it is among Euro-Americans:12

Shame is . . . a situation as much as a private emotion. . . . It represents a situation, or a state, of powerlessness and rejection. The legitimacy of one’s basic posture of assertion or of appeal has been removed.”

To “feel” shame is thus like “being in disgrace”. The emotion is primarily an aspect of a social situation rather than an unseen feeling in the individual heart.

Returning now to the Sherpa, we find that they also regulate behavior primarily by shame rather than guilt, and for them too, “shame is primarily a social status, and only secondarily an inner state.”13 The disadvantage of guilt is that it tends to leave a permanent mark:14

No priority is given in Sherpa justice to designating a particular person as the guilty one and punishing him. In our society, there are jails to hold such stigmatized persons. Among the Sherpas, everyone must live with everyone else, and setting off a permanent category of guilty criminals is a social impossibility. Instead of a permanent guilt which sets one apart as a criminal, the operative sanction is shame; the resolution of the conflict, thus, involves the elimination of shame.

Isolating anti-social members would interfere with the overriding goal of peaceful coexistence; “shame has the advantage over guilt that it can be wished away by social maneuvers”:15

A solution consists of finding terms upon which the disputants may agree — the payment of a fee, the return of a stolen object, or the mediation of an outsider in a land dispute. But this must almost necessarily be done by means of constructing a fictionalized version of the past and a tolerably well papered-over construction for the future.

Thus, the use of shame as a readily dissoluble sanction goes hand-in-hand with a relative indifference to the truth.

These passages of ethnography portray a coherent suite of ideologies and practices very different from our own. Where social harmony is at a premium, attributions of motive and intention would be dangerously disruptive; blame and guilt therefore have no purchase, and they pose the additional danger of being difficult to dispel; behavior is regulated instead by shame, which requires no psychological speculation and can be dispelled without remainder by face-saving fictions; and conflicts can then be resolved — if necessary, under coercion — so that harmony can be restored.

Is it permissible to flog an accused and his accuser in order to make them agree on a fictional version of their conflict? We would certainly say no if the dispute were taking place in Europe or the English-speaking world. But we wonder whether to say the same of the Sherpas, whose self-conception makes these practices intelligible, and for whom our practices of blame and guilt would make no sense.16 Such reluctance is the root of moral relativism.

Moral philosophers urge us nip relativism in the bud.17 They argue, to begin with, that relativism is self-undermining, because it supposedly applies to itself and implies of itself that it cannot be universally true. I will respond to this objection, but my response will follow trivially from my response to a second objection, namely, that relativism is unstable. It purports to be distinct from absolutism, on the one hand, and nihilism, on the other; the objection is that there is no third hand between those two.

If we say that the Sherpa permit the flogging of recalcitrant litigants and Anglo-Europeans don’t, we will be reporting their mores without making any normative claims about what is permissible or impermissible. These ethnographic reports will be compatible with the nihilist thesis that calling something permissible is an ontological error, like calling someone a witch. The relativist wants to rule out nihilism by saying that there is not just a cultural difference between groups with respect to what they permit but also a normative difference with respect to what is permissible for them — which entails that permissibility is real.

So a relativist has to say that flogging litigants is not just permitted by the Sherpas but permissible for them; that it is not just forbidden by Anglo-Europeans but impermissible for them. And when he says that such behavior is permissible for the Sherpas and not for Anglo-Europeans, he cannot simply mean that it is permissible according to the one group but not the other, which would be just another ethnographic report.

Nor can the relativist mean simply that the act of flogging litigants is permissible when carried out by Sherpas but not when carried out by Anglo-Europeans. The circumstances of the two groups may differ in ways that yield different verdicts even when subjected to a single, universal norm. Even an absolutist can say that what’s permissible for one group to do may not be permissible for another to do provided that they live in circumstances that differ in morally relevant respects.

Thus, the relativist faces a dilemma. If in calling behavior permissible for the Sherpas, he means “permitted by the Sherpas” or “permissible according to the Sherpas”, then he fails to rule out the possibility that permissibility is a myth. If he means that the behavior is permissible for the Sherpa to engage in, he fails to rule out the possibility of a universal norm that yields that verdict when applied to their circumstances though not to ours.

What a relativist should say is that statements of permissibility are perspectival: behavior can be permissible for one group but not another in the sense that its permissibility depends on their disparate perspectives. Relativism would then imply that being permissible is like being to the north. From a given perspective, some things are to the north — really and truly to the north — but nothing is to the north absolutely. Nor is anything permissible absolutely, according to relativism.

Of course, it can be true absolutely, independently of any perspective, that point B is to the north of point A; and so it can be true absolutely that, from the perspective of A, point B is to the north; but it cannot be true that point B is to the north, full stop, independently of any perspective. For when we speak of something as to the north without qualification, we are actually leaving an indexical qualification unspoken, since we mean that it is to the north of here, northwards from where we are. In this usage, ‘to the north’ contains an implicit indexical whose reference is relative to our perspective.

What is perspective-dependent, notice, is not the fact we assert in saying that B is simply to the north. If we are at A, the unspoken ‘here’ in our assertion refers to point A, and so we are asserting the perspective-independent fact that B is to the north of A. What’s perspective-dependent is only the way we express that fact. But that mode-of-expression makes a significant difference. For if I tell you that B is to the north of A, you won’t know which way to go in order to get there — not unless you know that you are at A, so that you can infer that the way to reach B is to go north, that is, north from here. The latter, indexical statement is, as John Perry puts it, essentially indexical: what it conveys cannot be conveyed by a perspective-independent statement of the same fact. Specifically, it conveys practical guidance that a perspective-independent statement cannot convey, since the latter cannot tell you how to get to point B.18 Perry, being Perry, puts the point more vividly, as follows. The fact that David Velleman’s trousers are on fire gives me no practical guidance unless I know that I am David Velleman, so that I can infer that my trousers are on fire — “my trousers” being the indexical, hence action-guiding, way for me to pick out the burning britches.

The same point applies to objective norms. “Lying is wrong” is action-guiding because it implies that one may not lie, but “One may not lie” is action guiding only because it can be transposed, within any given perspective, into “I may not lie.” In order to be guided by “Lying is wrong,” I must interpret it as implying that I, who am considering whether to lie, am forbidden to do so. So even an absolutist, who believes in universal moral norms, must acknowledge that they can provide practical guidance only by way of their first-personal instances.

If the absolutist must acknowledge that practical guidance is perspectival, what differentiates him from a relativist? The difference lies, to begin with, in the order of determination between the moral status of an action-type and practical guidance about it. An absolutist believes that I am forbidden to lie and you are forbidden to lie because lying is wrong. A relativist believes the converse: lying is wrong because it is forbidden from my perspective and yours. Thus, a relativist is a perspectivalist not just because he thinks that practical guidance is perpectival — the absolutist thinks so too — but because he thinks that right and wrong are ultimately grounded in practical guidance.

There is more to relativism than perspectivalism, however. One can think that practical guidance comes first in the order of constitution while also thinking that practical perspectives differ only with respect to the circumstances in which they are embedded. One may think, in other words, that the perspectives of a Nepalese pastoralist and an American academic differ only with respect to the local facts from which they must draw practical guidance. A relativist must believe that these agents draw guidance from their circumstances differently, so that they would derive different guidance even if their circumstances were exactly the same.

I suggest that we formulate the relativist’s thesis in terms of the basic units of practical guidance, namely, reasons for acting. Formulated in terms of reasons, relativism says that lying is wrong insofar as there is overriding reason against it, but that one and the same set of circumstances can present such a reason from one perspective but not from another.19

This difference would have to be based in a difference between perspectives with respect to the reasons-action relation — the relation between facts and a type of action in virtue of which those facts count as reasons for performing an action of that type. The relativist must claim that an agent’s perspective is somehow implicated in the reasons-action relation, so that one and the same totality of facts can present a Nepalese pastoralist and an American academic with reasons for different actions.

There are some necessary conditions that any account of the reasons-action relation must satisfy. Arguments for these conditions were put forward by Bernard Williams and Donald Davidson, but I will simply assert the conditions without argument.

First, a reason for someone to act must be a consideration that could become a reason for which he acts, that is, a something on which he could potentially base the action. And a reason for which the agent acts must be a consideration that causes him to do so, and causes this precisely in virtue of being a reason for so doing. Finally, behavior’s being caused by a reason, in virtue of its being a reason, is what raises that behavior from the status of mere behavior to that of an action.

Action transcends mere behavior in that it is our doing, whereas mere behavior happens in us or to us. We act when we take that behavior into our own hands, if you will, undertake it in such a way as to make it our own doing. The difference, to take a homely example, is that between the coughs that we cough intentionally, to clear our throats, and the coughs that escape from our throats almost despite us. It is manifested in the difference between cases in which we choose between alternatives and cases in which we have no alternatives — a difference, say, between choosing whether to eat a meal and being so hungry that we fall upon the first food in sight. It is also manifested in the difference between steps that we base on an antecedent decision and steps that are reflexive or impulsive — the difference, for instance, between walking along a planned route and dodging an oncoming car.

What makes the difference, in each instance, should be whether the behavior is caused by something in virtue of its being a reason for the resulting action. When a cough is a reflex triggered by a tickle in the throat, it is mere behavior; when it is caused instead by the fact that coughing will relieve the tickle, and caused by that fact in virtue of its being a reason for coughing, then the cough is an action. The reasons-action relation must therefore satisfy the following condition: that behavior caused by a something in virtue of their being so related is thereby raised from the status of mere behavior to that of an action.

In this section, I will consider one account of this relation, the one found in Kant’s Groundwork for the Metaphysics of Morals. In the next section, I will consider an account implicit in Aristotle’s Nicomachean Ethics. My thesis in either case will be that, protestations of absolutism notwithstanding, these authors hold conceptions of the reasons-action relation that contain a variable whose value must be supplied by a cultural perspective.

According to Kant, mere behavior is motivated solely by a desire whose satisfaction it is believed to promote. Behavior rises to the status of action when it is contingent on our proposing that this desire be a good enough reason for the resulting action. And we can rationally make this proposal only if it is consistent with the fact that a good enough reason must be recognizable as such to any rational agent. So our behavior amounts to an action when it is contingent on a self-consistent proposal that its motivating desire be universally recognizable as a good enough reason for the action — or, as Kant puts it, when it is based on a proposal of universal law.

Question: What is the reasons-action relation, according to this theory of action? Answer: A reason for an action is a desire that the action will tend to satisfy and that can be proposed as a reason consistently with the requirement that, if it is a reason, it be universally recognizable as such. Next question: How does a desire cause an action in virtue of bearing this relation to it? Answer: A desire causes an action in virtue of this relation because we have an additional motive that would prevent it from causing the action if they weren’t so related. Last question: What is that motive? Answer: That motive is respect for our own potential agency (or, as Kant would put it, our autonomy). We make behavior our doing by conditioning it on a proposal that its motivating desire be a good enough reason for it; and we condition it on that proposal out of respect for the ideal of thereby making it our doing.20

Yet the ethnographies of the Sherpa and similar groups suggest that they do not share Kant’s psychologized conception of agency, based on desires and proposals of their recognizability as reasons. I don’t claim that members of these groups have no desires, or that they are unaware of the desires they have. But they are reluctant to cite their mental states in explaining or justifying behavior; they tend instead to cite their social roles and the social scenarios in which those roles are embedded. It would be odd if Sherpas couldn’t make behavior their own without committing themselves to a self-conception that they are enculturated to suppress.

Kant’s own strategy of analyzing action in terms of a commitment to agency therefore militates against analyzing Sherpa agency in terms of a commitment to the Kantian conception of it, to which they, who are clearly agents, are not committed. A Sherpa must be guided by considerations relevant to his own conception of agency, which are considerations of the scenarios in which he participates and the roles he plays in them.

I therefore suggest that agency itself doesn’t require the Kantian conception of agency nor any other particular conception; it requires only a commitment to agency plus some conception thereof. Every agent does things with a commitment to their being his doing under some conception or other, but conceptions differ, and the commitment to agency lends the force of reasons to different considerations accordingly. For someone whose group observes opacity of mind, reasons will consist primarily in the scenario in which he is participating and his role within that scenario. The scenario may call for him to pursue an end or express an emotion, but the end or emotion will not be given by his inner experience; it will belong to a role that he is playing.

Here is an example from, the Kaluli of Papua-New Guinea, whom I cited earlier:21

First, for the Kaluli, anger almost always bears the implication that the angry person has suffered a loss of some kind, even if only in the form of a frustrated desire or disappointed hope. Second, because loss in a scheme of reciprocity implies one is entitled to return, the person who is angry is in some sense owed something: he has a legitimate (if often hopeless) expectation that he is due redress. . . .When a man has suffered wrong or loss (and where the culprit has made himself scarce), he may stamp furiously up and down the outside yard or inside hall of the longhouse yelling the particulars of his injury for everyone to hear in order to arouse their sympathetic attention and inspire their backing for redress. Thus anger attains a particular rhetorical force, a certain kind of measure and legitimacy, and a set of implications, from the way it is situated in the scenario of reciprocity.

Felt anger is not the man’s reason for stamping up and down the yard; his reason is that he has suffered a loss, and stamping up and down is the way to “be angry”, that is, to enact the role of the injured party. And there can of course be scenarios that include sentiments and behaviors far more alien to Anglo-Europeans than stamping around in anger. The Baining feel a sentiment that combines loneliness and hunger.22 As for behavior, it suffices to note that the Ilongot hunt heads.

Now, the project of using the philosophy of action to identify the relation between reasons and action may seem to be misguided, because the philosophy of action is a theoretical inquiry, whereas the relation between reasons and action is normative. Kant’s conception of agency is a conception of what agency is, whereas his Categorical Imperative dictates how it ought or ought not to be exercised. His strategy of identifying normativity by analyzing agency would therefore seem like an attempt to derive an ‘ought’ from an ‘is’.

Kant does not in fact try to derive an ‘ought’ from an ‘is’; what he does is rather to identify an ‘ought’ with an ‘is’. An ‘ought’ is an action-guiding hence necessarily perspectival expression of the same fact expressible with an ‘is’. According to Kant, action is behavior conditioned by a proposal that its motivating desire be universally recognizable as a reason — in other words, behavior conditioned by a proposal of universal law. That’s an action-theoretic fact, which has no practical import. It’s like the directional fact that B is to the north of A. The action-theoretic fact acquires practical import only when presented from the perspective of someone capable of being the agent of behavior, like so: The way for me to be the agent of such behavior is to condition it on a proposal of universal law. That’s like the directions from A to B: The way to get there from here is to go north. And if someone is committed to undertaking something as his doing, and he holds the Kantian conception of agency, he is thereby committed to conditioning it on a proposal of law, just as someone at A committed to reaching B is committed to going north.

Kant’s factual characterization of agency is framed from a standpoint outside the agent’s perspective, a standpoint from which it does not give the agent directions for being an agent. But when that same characterization is formulated by the agent in the first person, it turns into directions that he can follow from his perspective; and then his commitment to being an agent commits him to follow those directions. So the ‘is’ has become, if you will, a ‘how’ and then an ‘ought’, without any derivation or deduction.

Thus, Perry’s account of the essential indexical explains how the facts about agency, on the one hand, and directions for being an agent, on the other, can be two sides of the same coin. The philosophy of action then explains how those directions can have the force of an ‘ought’ for any would-be agent, who is already committed to following them.23

With this explanation in hand, we can see that the relativist can take a third way between making ethnographic reports that are compatible with nihilism and making normative claims that are compatible with absolutism. On the one hand, the very nature of action falsifies the nihilist brief against normativity: in order to act at all, we must be guided by a commitment to agency as we conceive it, which when combined with a conception of agency can give considerations the normative force of reasons. And the nihilist can take no comfort from the lack of guiding force in ethnographic reports about the conceptions of agency held by particular groups. Such conceptions can be action-guiding from the perspectives of those who are committed to enacting them — namely, agents, who are after all the only ones susceptible to the normative force of reasons. Normativity is missing from these reports only because they are framed from outside the agential perspective.

On the other hand, differences of what’s permissible or impermissible for different groups are not the result of substituting their disparate circumstances into universal norms; rather, the differences arise from substituting circumstances into disparate conceptions of agency. The results are local norms, which falsify absolutism by being local, and falsify nihilism by being genuinely normative.

Finally, a word about the objection that relativism is self-undermining. Relativism is said to be self-undermining because it has normative implications and hence applies to itself, thereby implying, of itself, that it is relative to a perspective. The answer to this objection is that relativism has no normative implications. It is a statement of fact, which is not relative to a perspective according to relativism itself.

Now, you may have suspected a circularity in my analysis of action. For if action is behavior guided by a commitment to agency, then agency must be conceived in terms of a commitment to itself. And the concept of a commitment to embodying that selfsame concept, and merely that concept, would be vacuous.

But a conception of agency can be self-referring without being vacuous, since it must include substantive specifications for agency in addition to the concept of a commitment to them. It may contain, for example, the concept of enacting particular roles, so long as the roles are conceived as including a commitment to enacting them. It can specify how an injured party behaves, that is, while also specifying that the injured party undertakes the behavior partly out of a commitment to that role.

Indeed, only a self-referential role can save its occupant from bad faith. Sartre’s famous waiter is in bad faith because he enacts the role of being a waiter anyway rather than out of a commitment to being one. In other words, the role to which he is committed is the role of being simply a waiter, not the role of hereby being a waiter — being one, that is, out of a commitment to this very role.

If a role isn’t self-referential in this way, you cannot genuinely fulfill it. If you aren’t already simply a waiter, you cannot become one by playing the role of someone who is. But by playing the role of someone hereby playing a waiter’s role, you can become just that. And that’s what an authentic waiter is: someone who styles himself a self-styled waiter.

In this respect, conceptions of agency are like roles. If you aren’t already a creature that simply conditions its behavior on a proposal of law, you cannot become one out of a commitment to being one. But a creature that so conditions its behavior out of a commitment to being such a creature — a creature that’s autonomous out of a commitment to autonomy — is a sort of creature that you can become out of a commitment to being one. (And what sort of autonomy would it be if you were stuck with it anyway, without being committed to it?)

So Kantians can really be Kantian agents, and Sherpas can really be Sherpa agents. Both are right about what agency is in their case. Even so, the possibility remains that in practice, being one sort of agent works better than being another — works better in the sense that one thereby takes fuller ownership of one’s behavior, makes it more fully one’s doing. Embodying that specific conception of agency may better enable one to embody the general concept. In that case, conceptions of agency would not be on a par, and neither would the associated normative perspectives. Whether this result would undermine relativism is a question I must postpone until the end of the next section.

2. Aristotle in Bali

A social requirement like “When in Rome, do as the Romans do” may sound relativistic but isn’t. It’s a particular instance of a more fundamental requirement to follow the conventions prevailing wherever one is, and that requirement is universal.

A relativistic social requirement along similar lines would be a fundamental requirement to do as the Romans do, applicable only in Rome. It would be a requirement that didn’t apply elsewhere, and not just because one wasn’t required to do like Romans elsewhere but because, when elsewhere, one wasn’t even required to do like Romans when in Rome. No requirement to do like Romans, not even one conditional on being in Rome, would be in force anywhere else; otherwise, a universal requirement would lurk in the background.

There may indeed be a requirement local to Rome of acting like a Roman, but it is a merely social requirement that gets its force from social sanctions. In New York one is under no pressure from vacationing Romans to follow Roman customs when one goes on vacation in Rome. But social sanctions lack the rationally binding force of morality. The challenge for relativism is to explain how there can be requirements that are rationally but only locally binding — a possibility that seems at odds with the universality of practical reason.

I hope to explain this possibility, beginning with an analogy.

Consider the two statements “In Quintopino they speak Spanish” and “We speak Spanish here,” uttered in Quintopino. The latter statement is action-guiding in a sense that the former is not. The second statement implies that the hearer ought to continue the conversation in Spanish. (Presumably, the speaker knows no English other than “We speak Spanish here,” which he uses to squelch conversational overtures in English.) By contrast, the former statement — “They speak Spanish in Quintopino” — cannot guide the listener’s linguistic activity unless it is supplemented by something like “This is Quintopino,” a statement that introduces the indexical ‘this’, meaning “this place”, or in other words, “here”.

When uttered in Quintopino, “We speak Spanish here” expresses the proposition that Spanish is spoken in Quintopino, which is true not only in Quintopino but even in Podunk. Of course, “Spanish is spoken here” is false when uttered in Podunk, because it then expresses the proposition that Spanish is spoken in Podunk, which is false even in Quintopino. The truth-values of the propositions don’t vary between the two locations; what varies is the proposition expressed by the indexical sentence. So much is obvious. Less obvious is that no practical guidance is provided by statements that make explicit which proposition they express; practical guidance is provided only by the indexical statement, which leaves the proposition expressed to be determined by the context of utterance. Even less obvious is that it then makes no difference for practical purposes which proposition is expressed. If Spanish is spoken here, then one had better speak Spanish, irrespective of where here is.

The action-guiding force of “We speak Spanish here” is in some sense conditional. The statement gives guidance only to someone who wants to be understood by inhabitants of the location indicated as here. For someone who doesn’t want to be understood by the locals, the statement is of no practical import. Yet the aim of being understood is internal to language: it is what language is for, in general if not always. You may say something in Spanish to your spouse in order not to be understood by your monolingual English-speaking children; you may speak Spanish to an English-speaking mugger in order to make him think that you cannot understand his threats. But these uses of language are the exceptions, and even they depend on the general rule that language is used as a means of communication.

“We speak Spanish here” is therefore internally action-guiding for the activity that it is about — the activity of linguistic communication. It’s like saying, “This lock takes a skeleton key,” a statement that tells the hearer how to use something for a purpose internal to its nature. The internal purpose of a lock is to control access; the way to control access with this lock is to use a skeleton key. The internal aim of language is to communicate; the way to communicate with language hereabouts is to use Spanish. Either statement is action-guiding for its stated topic as such.

The action-guidingness of “This lock takes a skeleton key” also depends on an indexical. “The lock on John Silver’s treasure chest takes a skeleton key” would not be action-guiding, not even when said to someone who had stumbled upon Silver’s chest, unless he knew this chest, so indicated, to be Silver’s. “They speak Spanish in Quintopino” would not be action-guiding, not even to someone in Quintopino, unless he knew this location, so indicated, to be Quintopino. And in either case, the indexical would be sufficient for guiding action and the name would be superfluous, because it wouldn’t matter for practical purposes whose chest this was or which place was here.

In Quintopino, then, the way to speak is in Spanish; in Podunk, the way to speak is in English. Generalizing on this example, we can state linguistic relativism: The language to speak in one social context may be different from the language to speak in another. By analogy with linguistic relativism, we can state moral relativism: What is morally the thing to do in one cultural context may be different from what is morally the thing to do in another.

But this statement is ambiguous and, as we have seen, one of its readings is not relativistic. On the non-relativistic reading, it says that morality can require a person to do different things in different cultural contexts because cultural context is a circumstance relevant to what one is morally required to do. Relativism is something else. It says that occupants of different contexts are subject to different fundamental requirements, not the same requirement conditioned by different circumstances. The distinction between these readings can be formalized in terms of the scope of moral requirements. On the first reading, the statement says that one may be morally required to do the following: A in context X and B in context Y. On the second reading, it asserts that there are two independent requirements: in context X one is morally required to do A, whereas in context Y one is morally required to do B.

The linguistic analogy can once again be helpful. As you board the plane in Podunk, a friend may say, “When you get to Quintopino, speak Spanish.” When you get to Quintopino, the locals may say “Speak Spanish here.” The former implicitly applies a universal imperative to your situation: Always speak the local language; the language of Quintopino is Spanish; so speak Spanish in Quintopino. The latter is an imperative addressed only to those in Quintopino.

If indeed there is no universal requirement to speak the local language, then conversation in Podunk can include no normative statements about the language to speak in Quintopino. Conversation in Podunk can of course include the true statement that the language of conversation in Quintopino is Spanish, but that remark would be a statement of mere linguistic fact: it would carry no recommending or guiding force. If someone in Podunk were to say, “One ought to speak Spanish in Quintopino,” he would be speaking out of turn (or, more precisely, out of place), since ex hypothesi there is no requirement applicable in Podunk as to which language to speak elsewhere. Only in the context of conversation in Quintopino can one say that Spanish is required.

This requirement, though local, is not merely social. One is required in Quintopino to speak Spanish not just by social pressure but by the internal aim of language. When one speaks, one is committed to the aim of being understood, and the force of one’s own commitment requires one to speak in the language of one’s interlocutors. So we might describe the requirement in Quintopino to speak Spanish as a local linguistic requirement.

Why are linguistic requirements local? The explanation is that the internal aim of speech is to communicate with the people one meets. This explanation depends on two simplifying assumptions: first, that one’s meets only those in one’s geographic vicinity and, second, that both parties are monolingual. (It was in order to suggest these assumptions that I chose the names ‘Podunk’ and ‘Quintopino’, which in their respective languages connote geographical isolation, as one might say “the sticks” or “the boondocks” or “the back of beyond”.) Together, these assumptions imply that people within a geographic region must coordinate on a single language in order to communicate.

If technology puts people into contact between geographically remote locations, of course, then geographic proximity and conversational proximity will not coincide; and if people learn more than one language, then they may not need to coordinate on a single one. But for most of human history, most people have known only one language and conversed only with others nearby. That’s why individual languages have tended to be points of convergence within geographic regions but not between them.

Technologies for communication and travel have scrambled the linguistic phenomena, but they have left the underlying exigencies of communication unchanged. My assumptions are therefore harmless, provided we keep in mind that they unscramble the phenomena only for the sake of simplicity.

If we want to explain the phenomena of moral relativism, albeit unscrambled for the same purpose, we will have to find similar exigencies, albeit under similar assumptions. That is, we will have to explain why there appear to be different moralities in different places, modulo the complications of modern multi-culturalism. And then, as I’ve said, we’ll have to explain how each of these moralities can really have normative force that is both rational and moral.

I believe that this challenge can be met, because action, like language, has an internal aim that favors coordination with others in one’s vicinity. Just as the aim inherent in language is linguistically normative in that it determines which language to speak, so an aim inherent in action as such is practically normative, determining what to do. And just as the language to speak follows from the local linguistic conventions, so the thing to do can follow from the local conventions of behavior, because the aim inherent in action, like the one inherent in language, turns out to favor social coordination. The goal of moral relativism is therefore to explain how mores can determine the moral thing to do, required by any agent’s commitment to an aim that is inherent in action as such.

How can an aim be inherent in action as such? An aim can be inherent in action because it makes action what it is: when added to mere behavior, it transforms that behavior into action. Plenty of goal-directed behavior falls short of action, because it isn’t authored and controlled by its subject but is rather reflexive, impulsive, or unwitting. When you blurt out an obscenity, dodge a blow, or simply cough, your behavior may be governed by impulses, reflexes, and skills that operate without engaging your powers of self-governance. But similar behavior with the same first-order aims may amount to autonomous actions, as when you eloquently curse someone out, deftly side-step a blow, or politely clear your throat. What I am now suggesting is that the difference is introduced by an additional aim that transforms your behavior into action.

Each of the resulting actions has its own first-order aim — delivering an insult, avoiding an injury, loosening phlegm. Nevertheless, these actions can share a second-order aim with respect to the manner in which they promote their first-order aims. An example of a second-order aim of manner would be speed. (I’m not saying that speed is the constitutive aim of action; I’m only citing it as an example of a second-order aim.) You cannot aim at being speedy by itself; you can only aim to be speedy in pursuing other aims, speed being a manner of pursuing them. What I suggest is that mere behavior is transformed into action by the presence of some such second-order aim with respect to the manner of behaving. I will now consider a theory of action according to which it is constituted in this way.

In the first sentence of his Nicomachean Ethics, Aristotle declares that every praxis aims at some good. Then, swapping the quantifiers, he suggests that there may be some good at which every praxis aims. That universal end, he later says, is eudaimonia, which he defines as “living well and doing well”.

Here the phrase for “doing well”, eu prattein, sounds like it has the secondary sense of that phrase in English, the sense of “faring well”. And indeed eu prattein can be translated “faring well”. But Aristotle now turns to a discussion of “the well” for human beings, and he begins that discussion by asking what it is characteristic of humans to do, since doing that thing well is what “the well” will be in the human case. So he conceives of eu prattein as a matter of doing well not just in the sense of faring well but also, at the same time, in the sense of doing something well.

What we ultimately aim at doing well, according to Aristotle, is the activity of the rational soul, but in the practical sphere that activity aims at doing well at more specific things. And in all of these more specific doings, the end is doing well, as Aristotle explains in Book VI when he compares making, poesis, with doing, praxis:

Whoever makes something does so for some purpose, and what is made is not an end in itself but of and for something. But what is done [is an end], for eupraxia [is] the end, and desire is for that. (VI.ii.5, 1139b)

Making [poiēsis] has some further end, but not praxis; for eupraxia itself is the end. (VI.iv.4-6, 1140b)

Eupraxia is just the nominal form of eu prattein. Thus, the aim of doing as such — doing as opposed to making — is doing well.

In the past, I have presented this material as a lecture; I tried to do that well. The way to present a lecture well is, first, to compose a text (I tried to do that well) and then to read it out (I tried to do that well). The way to read out a lecture well is to speak clearly in an animated voice; I tried to do that well. And so on. The hierarchy of activities extends in the other direction, too. Presenting a lecture is a way of practicing the profession of philosopher well; practicing the profession of philosopher is a way of pursuing a career well; and so on, all the way up to the activity of living well, which constitutes eudaimonia. At each stage, the rational soul identifies what to do in order to do well whatever one is already trying to do.

Returning from the sublime to the mundane, consider how the second-order aim of doing well might generate my earlier examples of action when added to the corresponding instances mere behavior. Cursing someone out goes beyond spewing curses at him by being aimed at insulting him well; side-stepping a blow goes beyond reflexively dodging it by being aimed at avoiding injury well; clearing one’s throat aims not just at loosening phlegm but at doing so well. In sum, action can be analyzed as behavior that, in addition to pursuing some primary aim, has the second-order aim of pursuing it well — an aim imparted to it, according to Aristotle, by the deliberative activity of the rational soul.

You might think that you can do something without aiming to do it well, for example, when you play tennis with a child and try to play poorly in order to let the child win.24 But in that case, you aren’t really playing tennis after all; you are entertaining a child on the tennis court by pretending to play — which you are trying to do well, so as not to let on that the child isn’t really beating you. So what you are doing is determined by what you are trying to do well. If you try to do something poorly, you end up doing something else, which you are trying to do well.

The way to act, then, is to aim at doing things well under rational guidance, because that’s the constitutive aim of action, just as communication, being the internal aim of language, determines the way to speak. Internal to every praxis is the aim of eupraxia, and eupraxia is therefore the fundamental practical norm. Eudaimonia is just the most comprehensive form of doing well.

Now, this result may seem trivial — worse than trivial, circular — because it purports to derive a practical norm from a theory of action that helps itself to a normative concept, namely, the concept of doing something well. Isn’t it obvious that “the well” is normative for action, and trivially so?

Well, I think that to eu, for Aristotle, is not a merely formal aim of the sort that would be expressed by a “thin” normative term meaning “in the right way (whatever that is)”. Remember that Aristotle limits deliberation to matters that call for judgment because there is no well-developed method for dealing with them — matters that are more of an art than a science. When we moderns think of the right way to do things, we think of the efficacious way, the way that will get things done, which is often a well-developed method. But matters that call for virtue, in Aristotle’s eyes, are matters such as how much fear or anger to manifest, how large a gift to give, how much pain to bear, how retiring or outgoing to be, how boastful or self-deprecating, even how far to go in joking. Most of these are matters of proper form rather than efficacy.

Let me cast this aspect of Aristotle’s ethics into relief by means of comparison with a very different culture, in which doing things well is a different matter entirely. Here I will digress into a bit of ethnography, drawn from Clifford Geertz’s eloquent study titled Person, Conduct, and Time in Bali.25 Doing well looks very different in Bali of the 1960s than in Athens of the fourth century BC, and yet there are illuminating similarities.

Geertz explains the nature of conduct in Bali in terms of the Balinese conceptions of personhood and time, both of which he characterizes in terms of their nomenclature — how people and times are named. These systems of naming are complex and, to the Westerner, profoundly foreign, and they are indicative of what is, to us, a profoundly foreign form of agency.

Names for persons in Bali tend to efface their individuality and emphasize their changing position in the life of the community. Children are given personal names, but those names are almost never used. Personal names are nonsense syllables, not drawn from any pool of common names: no one is named after anyone else. But children are not usually called by these unique personal names; they are rather called by birth-order names, of which there are only four, so that the fifth child has the same birth-order name as the first. Similarly, kinship terms are assigned by generation, not by lines of descent. Thus, one uses the same term to denote all of one’s siblings, cousins, spouses’s siblings, and so on; another term to denote all of one’s parents and their siblings and their siblings’ spouses, and so on. What’s more, great-grandparents and their great-grandchildren denote one another by the same kinship term — they call each other kumpi — as if closing the circle of the four generations that can be alive at the same time.

Once children have children of their own, they are no longer called by their birth-order names and are called by teknonyms instead — that is, names such as “mother-of-John”, where John is her eldest child and the name “John” is his personal name, by which he himself is almost never called. Once the mother has a first grandchild, whether born to John or to one of his younger siblings, her name changes, say, to “grandmother-of-Jane”, although Jane herself is called by a birth-order name until such time as she is called by a teknonym, as mother of her first child, who may not be the one after whom its now great-grandmother is named, if her first great-grandchild was born to one of Jane’s younger siblings. Denominating adults by descent rather than ancestry — that is, as “mother …” or “father of …” rather than “daughter …” or “son of …” — locates them in relation to the living rather than the dead, and unlike the latter, requires their names to change throughout the life-course as they ascend the scaffold of generations.

Because of the way these names change to reflect one’s shifting positions among the living rather than one’s fixed relations to the dead, they de-emphasize one’s individuality. Persons are identified not by individual names but, at first, by their position in the generative life of their parents, and then by their position in the generational scaffolding of the family and village. More important than the persisting individual is his or her changing location in the family.

Even more complex than this nomenclature for persons is the Balinese nomenclature for times. Whereas Westerners have one cycle of day-names, marking the days of a seven-day week, the Balinese have ten different cycles of day-names, each of a different length, running concurrently. Hence a single day has a name in the nine-day week, a different name in the eight-day week, and so on. Most significant are the weeks of five, six, and seven days, which yield a cycle of 210 trinomially designated days. There are also binomially designated days that recur every 39, 35, and 42 days. Geertz writes:26

[T]he nature of time reckoning this sort of calendar facilitates is clearly not durational but punctual. That is, it is not used . . . to measure the rate at which time passes, the amount which has passed since the occurrence of some event, or the amount which remains with which to complete some project: it is adapted to and used for distinguishing and classifying discrete, self-subsistent particles of time — “days.” The cycles and supercycles are endless, unanchored, unaccountable and, as their internal order has no significance, without climax. They do not accumulate, they do not build, and they are not consumed. They don’t tell you what time it is; they tell you what kind of time it is.

“What kind of time it is”, given by the bi- and trinomial name of the day, determines whether a temple’s festival is celebrated (there are some 20,000 temples in Bali), whether the time is propitious for building, traveling, marrying, burying, and so on. And the name of the day one was born functions like a sign of the zodiac, determining one’s character, suitability for a particular marriage, prognosis in illness, and so forth.

These two systems of nomenclature, for persons and times, account for what Geertz calls “the ceremonialization of Balinese social interaction”.27 The effacement of individuality requires interaction to be highly formalized: “To maintain the (relative) anonymization of individuals with whom one is in daily contact, to dampen the intimacy implicit in face-to-face relationships … it is necessary to formalize relations with them to a fairly high degree,” emphasizing surface appearances. The ceremonialization of life is further intensified by the atomization of time, which causes what Geertz calls “absence of climax”:28

[S]ocial activities do not build, or are not permitted to build, toward definitive consummations. Quarrels appear and disappear, on occasion they even persist, but they hardly ever come to a head. Issues are not sharpened for decision, they are blunted and softened in the hope that the mere evolution of circumstances will resolve them, or better yet, that they will simply evaporate. Daily life consists of self-contained, monadic encounters in which something either happens or does not — an intention is realized or it is not, a task accomplished or not. When the thing doesn’t happen — the intention is frustrated, the task unaccomplished — the effort may be made again from the beginning at some other time; or it may simply be abandoned.

“In short,” Geertz concludes, “events happen like holidays. They appear, vanish, and reappear — each discrete, sufficient unto itself, a particular manifestation of the fixed order of things.”29

Ceremonialization of daily life has a further consequence: “This ceremonialization takes the form of an earnest, even sedulous, kind of ‘playing’ with public forms; … religion, art, and etiquette are then but differently directed manifestations of an overall cultural fascination with the worked up semblance of things; and … morality here is consequently aesthetic at base.”30 Once ceremonialized, in other words, daily life becomes aestheticized: “Social acts, all social acts, are first and foremost designed to please — to please the gods, to please the audience, to please the other, to please the self; but to please as beauty pleases not as virtue pleases.”31 Social behavior is therefore regulated, not by guilt or shame, but what Geertz translates as “stage fright”:

[S]tage fright consists in a fear that, for want of skill or self-control, or perhaps mere accident, an aesthetic illusion will not be maintained, that the actor will show through his part and part thus dissolve into the actor. … What is feared — mildly in most cases, intensely in a few — is that the public performance that is etiquette will be botched, that the social distance etiquette maintains will consequently collapse, and that the personality of the individual will then break through to dissolve his standardized public identity.

Geertz’s depiction of Balinese daily life makes clear that the three categories of his title — person, time, and conduct — are fundamentally different from the corresponding categories in the West. And it suggests that the Balinese have a specific and quite distinctive conception of doing things well. For the Balinese, apparently, doing something well consists in doing it in a formalized and aesthetically pleasing way; achieving the action’s first-order aim may even be secondary.

Aristotle’s conception of doing well is both like and unlike the Balinese’. After all, his term for the mean between extremes is to kalon, the primary meaning of which is “the beautiful”. This use of the adjective kalos does not mean “beautiful” in a purely aesthetic sense, and yet a translation shorn of aesthetic connotations, such as “noble”, strikes me as too far afield. I prefer “spendid”, which has its root in the Latin verb “to shine” and therefore puts me in mind of Kant’s statement that the good will “shines forth like a jewel”. In fact, Aristotle himself says that in adverse circumstances, to kalon “shines through” (dialampei).32 And we sometimes use the word ‘beautiful’ for things done well: when someone hits the bullseye, the basket, the ace we exclaim “Beautiful!”. That’s what Aristotle says when someone hits the mean between extremes.

So in matters that call for virtue — matters of fear and daring, aggression and concession, giving and taking money, self-promotion and -deprecation, joviality and gravity — in such matters, to kalon is indeed a kind of beauty, though not the Balinese kind. Both are socially constructed ideals of well-ordered behavior, an aim that is the hallmark of action.

In the previous section I argued that an analysis of action in terms of a second-order aim can also be found in Kant’s Groundwork for the Metaphysics of Morals. The Categorical Imperative encodes a conception of action in terms of autonomy: action is behavior conditioned by a proposal that its motivating desire be such as anyone would recognize as a sufficient reason; in other words, behavior conditioned by a proposal of universal law; in other words, behavior that is autonomous. We are moved to follow this recipe because we respect our own potential autonomy and therefore aspire to realize it. Autonomy is thus the second-order aim that is constitutive of action, each action also having a first-order aim given by its motivating desire.

That was my interpretation of Kant. I then argued that this analysis of action cannot apply to agents who do not understand themselves in the psychological terms in which the Categorical Imperative is framed, and who therefore cannot be assumed to have any commitment to Kantian autonomy. I described several groups, such as the Sherpa of Nepal, whose members avoid attributing mental states to people, including themselves, thereby observing what anthropologists call “opacity of mind”. They think of themselves, not as manifesting inner mental states, but rather in terms of stereotyped roles and routines. For them, being angry consists in acting the part of an injured party, and being ashamed consists in a social status akin to being in disgrace. For members of these societies, respect for Kantian autonomy is unlikely to play a role in the constitution of action.

I therefore divided the Kantian analysis of action into two parts, one constant, the other variable. What remains constant is an analysis of action as behavior with the second-order aim of being its agent — or, as I put it, of making it one’s own doing. What varies is the operative conception of agency, Kantian autonomy being only one such conception.

I believe that the analysis of action that I have now attributed to Aristotle can fit into the same framework. The notion of doing something well under rational guidance is a conception of agency. Aiming at doing things well under rational guidance is thus a version of aiming at agency, the Aristotelian version. So what makes eu prattein normative in Aristotle’s theory is what makes acting on universalizable maxims normative in Kant’s: each is that philosopher’s conception of action’s constitutive aim.

I said at the outset that a relativist must explain how the mores of a group can place its members under requirements that differ from those placed on the members of groups with different mores, and how such requirements can be rationally binding rather than just socially enforced. I turn now to those explanations.

Conceptions of agency are local for much the same reason as languages. People who interact need to interpret one another, and be interpreted by one another, in their behavior as well as their speech. Such mutual interpretation is essential to collective action, which includes not only active cooperation but also the passive cooperation of minding one’s own business so long as the business of others seems harmless enough, and then making one’s business seem harmless so that others will go on minding theirs. And even if one needs to seem dangerous rather than harmless — and even if seeming dangerous requires acting unintelligibly — well, then, one still needs to know how others will try to interpret one’s behavior, so as to ensure that they can’t. Thus, people need a shared schema for making sense of and to one another — or not, as the case may be. Such a schema is supplied by their conception of agency, which structures not only how they package their own behavior but also how they unpack the behavior of others. That’s why there is pressure within a social group to coordinate on a conception of agency, whereas there is no pressure for coordination among mutually isolated groups.

Kant thought that the requirements of morality follow directly from his conception of agency as autonomy. By contrast, all that follows from Aristotle’s conception of agency is the form of what’s required, namely, the form of to kalon as a mean between extremes. How to find the mean, he left for the agent to learn in the course of a good upbringing.

Aristotle seems to have thought that this education is authoritative because it cultivates a knack for seeing where the mean really lies, depending of course on the particulars of each case. We might think — I do think — that there is no fact as to where the mean lies, only a socially constructed sensibility as to what would be excessive or deficient in anger, acquisitiveness, candor, or joviality, and what by contrast would hit the bullseye and be the splendid thing to do. But I also think that this sensibility carries an authority over and above that of the teachers who inculcate it. An Aristotelian agent can make some behavior his doing only by aiming to do it well, and he must have some way of distinguishing between “the well” and “the badly”. Since he and his fellow Athenians interpret behavior by figuring out what the subject is trying to do well, the Aristotelian agent must be able to draw the distinction between “well” and “badly” along the same lines as them; otherwise, he won’t be able to make sense of them or to them. (Someone who sees you pretending to lose at tennis won’t know what you’re doing if he has no sense of what counts for us as entertaining a child well.) The point is not that the Aristotelian agent tries to impress his fellow Athenians by doing what they will regard as well done; it is rather that he can make something his doing only by trying to do it well, and he cannot draw his own private distinction between doing something well and doing it badly. And having been taught the distinction, the Aristotelian agent finds it authoritative not because he learned it from authority figures but because it is an indispensable accessory to his conception of agency and therefore shares in the authority of his commitment to being the agent of his behavior.

Where did we humans get the idea of agency, much less any particular conception of it? Where did we get the idea of taking our behavior into our own hands and making it our doing? We must have experienced that accomplishment in order to get the idea of it — who could have imagined it beforehand? — and yet we couldn’t have experienced it without first having a commitment to it, which we couldn’t have had without the idea.

I believe that what I have all along been calling conceptions of agency are in the first instance schemas for interpreting behavior and making it interpretable, or as I put it, for packaging our own behavior and unpacking the behavior of others. Even if other creatures are capable of interpreting one another’s behavior, humans are the only creatures capable of considering themselves from the standpoint of a prospective interpreter. That capacity enabled us to develop forms of sociality that entail making ourselves interpretable to one another. One such form of sociality is language, of course, but there are others; for example, shared intentions, which are the basis of human cooperation.

In order to elicit an interpretation of our behavior, we must mold that behavior to an interpretation schema that we share with our intended interpreters, and such a schema amounts to a shared conception of agency. I call it an interpretation schema by analogy with the term ‘explanation schema’ as used in reference to Hempel’s theory of explanation. Just as Hempel’s D-N schema yields explanations when filled in with particular descriptions and laws, so Aristotle’s conception of agency yields interpretations only filled in with ways of doing particular things well, and Kant’s conception of agency yields interpretations when filled in with particular maxims. I believe that we developed conceptions of agency, in the first instance, not because we needed to understand agency itself — it didn’t yet exist — but because we needed to interpret human behavior. Having developing schemas for interpreting it, and aiming to satisfy them, we became agents, and they became conceptions of agency. So the same revolution in self-awareness that was crucial to our becoming language-users was equally crucial to our becoming agents.

This genealogy suggests that conceptions of agency are not necessarily on a par. There can be better and worse schemas for mutual interpretation — better and worse ways of packing and unpacking behavior so as make ourselves understood to others and understand them in return. The standard of better and worse is not moral: it’s not that some conceptions of agency yield mores closer to or further from the true morality. The standard is rather set by the original purpose of such conceptions, which was mutual interpretation and it associated forms of sociality.

Even if some conceptions of agency are better than others, however, agents cannot take guidance from conceptions other than their own. Consider an extension of the geographic analogy that I introduced in the previous lecture. Occupants of point C cannot get to point B by following the directions for getting there from point A. But point C may still be superior to A as a location from which to go places: it may be a travel hub with many spokes whereas A is situated on a single dirt track. For travelers, then, occupying the perspective of C would in general better serve a commitment to travel, the commitment shared by anyone who wants to go places. A frequent flyer who lives in Podunk might even consider moving to the big city. Until he does, however, he had better not follow city-centric directions. Action-guidance is always guidance from where one is.

Where one is, normatively speaking, is one’s conception of agency. It may be a one-track conception; I think Aristotle’s was. Or it may be a conception from which one can get to any of the others. But as I have said, our conception of agency harbors relativism within itself by subsuming the other conceptions. In that respect, at least, I think that we have made progress.

---

1 Robert A. Paul, “Act and Intention in Sherpa Culture and Society”, in Other Intentions: Cultural Contexts and the Attribution of Inner States, ed. Lawrence Rosen (Santa Fe, NM: School of American Research Press, 1995), 15-45, pp. 28–29. Citation omitted. The case was reported by C.W. Cassinelli and Robert B. Ekvall, A Tibetan Principality (Ithaca: Cornell University Press, 1969), 171–177; the quotation is from pp. 174–175.

2 Robert A. Paul, “The Place of the Truth in Sherpa Law and Religion”, Journal of Anthropological Research 33:2 (1977) 167–184, p. 174.

3 “Act and Intention in Sherpa Culture and Society”, p. 21.

4 Ibid.

5 Sherry Ortner, High Religion: A Cultural and Political History of Sherpa Buddhism (Princeton: Princeton University Press, 1989), p. 216 n. 17; quoted by Paul, “Act and Intention in Sherpa Culture and Society”, p. 19.

6 See the special issue of Anthropological Quarterly 81:2 (2008): “Social Thought and Commentary Section: Anthropology and the Opacity of Other Minds. Edited by Alan Rumsey and Joel Robbins”.

7 Michelle Z. Rosaldo, “Toward an Anthropology of Self and Feelings”, in Culture Theory: Essays on Mind, Self, and Emotion, ed. R.A. Shweder and R.A. Levine (Cambridge: Cambridge University Press, 1984), 137–157, p. 146. The passage continues (p. 147): “Deviance, illness, madness, and failure to perform are typically attributed to things outside the self: Spiritual forces may cause crops to fail or make a person wild or weak by taking the heart out of one’s body. But no one sees in deviant acts the telling symptoms of a person’s character or worth. Nor do Ilongots in their self-reflections speak of personal histories or distinctive psychic drives to account for the peculiarities of deeds or dreams. . . . In short, it seems misleading to identify individuality with the Ilongot sense of self, first, because Ilongots do not assume a gap between the private self and public person and, second, because the very terms they use in the accounts of how and why they act place emphasis not on the individual who remains outside a social whole but rather on the ways in which all adults are simultaneously autonomous and equal members of a group.”

8 Jane Fajans, “The Person in Social Context: The Social Character of Baining ‘Psychology’”, in Geoffrey M. White and John Kirkpatrick (eds.), Person, Self, and Experience: Exploring Pactici Ethnopsychologies (Berkeley: University of California Press, 1985), 367–397, p. 367.

9 Ibid., p. 383.

10 Fajans, op. cit., p. 371.

11 Michelle Z. Rosaldo, “The Shame of Headhunters and the Autonomy of Self”, Ethos 11:3 (1983) 135–151, p. 141.

12 Edward L. Schieffelin, “Anger and Shame in the Tropical Forest: On Affect as a Cultural System in Papua New Guinea”, Ethos 11 (1983) 181–191, p. 189. Schieffelin describes opacity of mind among the Kaluli as follows (p. 184): “Kaluli . . . avoid making statements that attribute feelings, motivations, or intentions to people (even if they have a pretty good idea what they are) unless these feeling have been in some way already publicly expressed. This reluctance to talk about the feelings and motivations of others is part of a more general reluctance to paraphrase or present interpretively others’ statements of claim or purpose when reporting them to others.” On versions of shame, see also Rosaldo, “Toward and Anthropology of Self and Feeling”, p. 149: “[T]he error of the classic ‘guilt and shame’ account is that it tends to universalize our culture’s view of a desiring inner self without realizing that such selves — and so the things they feel — are, in important ways, social creations. ‘Shames’ differ as much cross-culturally as our notions of ‘shame’ and guilt.’”

13 Paul, “The Place of the Truth in Sherpa Law and Religion”, p. 171.

14 Ibid., p. 174.

15 Ibid.

16 See Paul, “Act and Intention in Sherpa Culture and Society”, p. 36: “One quickly sees that there would be no point in asking why the Sherpas, or many other people like them, do not talk ‘truthfully’ and ‘objectively’ about intention and inner states. Why should they? What on earth would they gain by it? In a social situation where an ongoing basis for at least tolerable cooperation and absence of conflict is desirable, what point could there be in translating their already adequate abilities and mechanisms for coping into a blunter and more destructive language, the ‘psychologizing’ language of intent, fault, blame, and inner state? So long as people act reasonably well within the bounds of expectation and acceptability, no one gains from insisting on knowing whether anybody really meant what they did or said way down deep in their heart.”

17 See Paul Boghossian, “Three Kinds of Relativism”, in A Companion to Relativism, ed. Steven D. Hales (Oxford: Wiley Blackwell, 2011), 53–69; “What is Relativism”, in Truth and Realism, ed. Patrick Greenough and Michael P. Lynch (Oxford: Oxford University Press, 2006), 13–37; “Relativism About the Normative” (MS).

18 Of course, “B is to the north” won’t help you get to point B unless you know which way is north, right or left, forward or backward. Thus, the implicit reference to A as ‘here’ is not sufficient; the designation ‘north’ must also be translated into indexical terms — in effect, as “there”. But this necessity is just another illustration of the same point, that practical guidance must be perspectival.

19 This thesis must be stated carefully. When the relativist says that the same circumstances can present a reason from one perspective but not from another, his point is not that there may be additional facts available in one but not the other perspective. The relativist’s thesis is that different practical guidance can issue from different perspectives when all of the facts are taken into account from both.

20 This version of Kant is similar, perhaps identical, to Christine Korsgaard’s “self-constitution” account of action, which is of course inspired by Kant. At the end of the next section, I will suggest that it should be subsumed under a more general account, but as I have said, I am reaching for an off-the-shelf philosophy of action for present purposes.

21 Schieffelin 1983, pp. 186–187.

22 Fajans 1985.

23 This explanation also clarifies the constitutivist’s reply to the “shmagency” objection raised by David Enoch (“Agency, Shmagency: Why Normativity Won’t Come from What Is Constitutive of Action”, Philosophical Review 115 (2006): 169–198). The question “Why should I be an agent?” is nonsensical, because ‘should’ has force only from the perspective of agency. There may be a perspective of shmagency, and maybe agents can ask themselves, from their perspective, whether to migrate into that one. But there is no perspective-independent question about which perspective to occupy.

24 The point made in this paragraph is similar to that made by Christine Korsgaard in Self-Constitution: Agency, Identity, and Integrity (Oxford: Oxford University Press, 2009), 31–32.

25 Person, Time, and Conduct in Bali: An Essay in Cultural Analysis (New Haven: Yale University Press, 1966).

26 Geertz (1966), p. 47.

27 Ibid., p. 54.

28 Ibid., p. 60.

29 Geertz (1966), p. 61.

30 Ibid., p. 56.

31 Ibid. Ward Keeler disputes Geertz’s account of “stage fright”, though in a way that only underscores the Balinese denial of individuality: “The Javanese and Balinese recognize that much of what they do and say in encounter covers feelings that they keep hidden, but they do not think that secrets make up a special or “true” self. They do not feel that any person stands apart from interaction, that someone has a private self which threatens to “break through to dissolve his standardized public identity”. Neither wishes nor quirks constituted a concealing, informing self. Rather, they are either so universal as to be presumed or so idiosyncratic as to be discounted. A person’s failure to act suitably in encounter, therefore, does not reveal a human face behind a social mask. Instead it calls in into question the integrity of the speaker — who shows himself without regard for his own and others’ status — and his interlocutor’s power — which has been called into question by the speaker’s failure to register its effects.” (Ward Keeler, “Shame and Stage Fright in Java”, Ethos 11 (1983): 152–165, pp. 161–162.)

32 On this issue see Gabriel Richardson Lear, “Aristotle on Moral Virtue and the Fine”, in Richard Kraut (ed.), The Blackwell Guide to Aristotle’s Nichomachean Ethics (Oxford: Blackwell, 2006), 116–136.