CHAPTER TWO
The Finality Criterion
1. INTRODUCTION
I begin by raising a problem about Aristotle’s practical teleology. In NE I.7 Aristotle lays out some criteria that any account of the human good must meet. Perhaps the most important is that it be the most final or end-like (teleiotaton) of all the human goods.1 The human good, in virtue of being most final, is choiceworthy for its own sake alone and never for the sake of any other good (1097a28–34). Eudaimonia, or happiness, meets this criterion, which means that whatever happiness turns out to be—whether it be the activity of moral virtue or contemplation—it will be worth pursuing for its own sake and never for the sake of anything else that might be gained through it (1097a34–b5). Aristotle’s ethical theory is meant to be a practical guide in the sense that he believes that the point of examining the nature of happiness is to aid our own pursuit of it (1094a22–24).2 So in describing eudaimonia as the most final end of a flourishing life, Aristotle must suppose that it is possible consciously to live our lives for the sake of whatever good turns out to be the highest. But what would it be to make a good (and in particular, contemplation, the good Aristotle argues is the high-est) our most final end?3 It would mean that all the goods and activities we pursue would be worth pursuing for the sake of eudaimonia, and that we would choose them for that reason. In particular, we would choose the goods we already value for themselves—like friends, moral virtue, and honor—for the sake of the separate good that is eudaimonia. This is the attraction of Aristotle’s emphasizing the “endiness” of the highest good: Our devotion to the highest good gives order to our pursuit of goods that we already value for their own sakes.
But this very attraction of Aristotle’s ethical theory ushers in serious complications. It presupposes that there are goods that are choiceworthy both for their own sakes and for the sake of some independent good, namely eudaimonia. (I take it to be implausible to think that only the eudaimonic good is choiceworthy for its own sake.) Or, to put it another way, it presupposes that there are goods whose ends are in themselves and also beyond themselves, in eudaimonia. I will call these goods middle-level ends.
The problem I want to raise in this chapter is that it is not at all clear that, in Aristotle’s conception of a telos (end) there could be ends of this middle-level sort. The problem will take some time to explain, but let me give a brief outline here. In general, when something happens or exists for the sake of an end, that end sets the standards of success for the process leading to it. It is because that process happened, or that thing came into being for the sake of the end, that the process or thing is the way it is. And we can determine whether the process was successful by discovering the extent to which it achieved its end well. So an end guides the appropriate pursuit of the things leading to it and is their source of value. The problem arises, though, when we consider ends that are choiceworthy both for their own sakes and for the sake of eudaimonia. How could anything be worth choosing both for itself and for the sake of a higher goal?
Note that the problem is not merely that there are things with both an immediate and an ulterior telos . Instruments or tools can be ends like this; immediately, making a tool is for the sake of making the product of the relevant craft. But if the craft product is itself made for the sake of something further, it is right to say that this further goal of the craft is also the ulterior goal of the making of the tool. Notice that in this case of making instruments, the ulterior telos is a fulfillment of the immediate telos. The immediate goal of a computer hardware designer, a computer chip, is created to meet the demands of the ulterior goal of hardware design, computer processing. The ulterior telos is what the immediate telos is for. Since this is so, the interest in the ulterior telos determines what counts as a successful attempt at making the immediate telos. But notice that, according to this model, the way in which the immediate telos depends on the ulterior telos means that the former is not choiceworthy for its own sake. No one would want a computer chip as such unless he cared about using a computer.
By contrast, middle-level ends are genuinely choiceworthy for their own sakes, even though they are also worth choosing for the sake of eudaimonia . Aristotle says we would want them even if they did not lead to anything further (1097b3–4). The problem is to understand how this is possible.
Being choiceworthy for the sake of another end does not always imply that a thing is not also choiceworthy for its own sake. Take, for example, the connection between the disposition to be brave and acts of bravery. The disposition is clearly choiceworthy for the sake of the actions it tends to produce. But Aristotle thinks that it is also choiceworthy in itself (1097b2– 3, re all virtues). It makes sense, after all, to desire a courageous character even if, due to peaceful times, one is not called upon for a prominent display of courage. Bravery and the other virtues of character are goods choiceworthy for their own sakes that find their fulfillment in an end beyond themselves. Upon closer examination, however, it is clear that the relationship between a disposition and its actualization cannot be a model for the connection between the middle-level ends (at least not all of them) and eudaimonia. For, in Aristotle’s theory, the middle-level ends are meant to be independent of the good that is eudaimonia. Although friends, moral virtue, and pleasure are choiceworthy for the sake of eudaimonia, they are all conceivable independently of eudaimonia, and they can be rationally pursued outside the context of a life lived for the sake of the highest good. Dispositions, on the other hand, are not independent of their actualizations in this way. Courage, although it is different from courageous action, is not conceivable in isolation from its actualization. That is because the disposition to be courageous is what it is—it has the form it has—in reference to actual acts of courage. Indeed, insofar as we find the disposition to be courageous worth choosing for its own sake, it is because we value its character as the sort of thing that tends to lead to courageous actions. It is hard to see how all the middle-level ends could depend on eudaimonia for their form. This difficulty arises no matter whether we interpret Aristotle’s eudaimonia as a monistic end, as I think we should, or as an inclusive end.
In the relationship between a disposition and its actualization, the two goods are not sufficiently independent of each other to be a model for all middle-level ends. In the relationship between an instrument and its ulterior telos, the two goods may be sufficiently independent, but the instrument is not choiceworthy for its own sake. What, then, could the relationship between the middle-level ends and eudaimonia be? If the very idea is incoherent, as it begins to seem to be, it is hard to see how it can be of any help in planning our lives to focus on the status of eudaimonia as an end. We already value a variety of goods as choiceworthy for their own sakes and need to find some way to coordinate their pursuit. What good will it do me to be told that, in some mysterious way, these intrinsically valuable goods are also choiceworthy for the sake of eudaimonia?
In the remainder of this chapter I will try to make the nature of this problem clearer and to explain why some apparently promising solutions do not succeed. First, I will examine what it is to be a telos in Aristotle’s sense, as described in his biological and metaphysical works. Next I will examine his concept of a telos, or a “that for the sake of which,” as developed in the Nicomachean Ethics. My purpose here is twofold: I want to show that Aristotle’s conception of a telos in the Nicomachean Ethics is the same one he uses in his biological works and, at the same time, to show the process of reasoning leading him to the introduction of the finality criterion, that is, to claim that the highest good (happiness) must be choiceworthy for its own sake and not for the sake of anything else. It is in this section that I will offer reasons for interpreting eudaimonia as a monistic end. Then, once I have clarified the problem with middle-level ends, I will examine and criticize several possible solutions.
2. WHAT IT IS TO BE AN ARISTOTELIAN TELOS
We are trying to understand the nature of goods that are choiceworthy for their own sakes and for the sake of eudaimonia. So we need to get straight at a more basic level what it is to act or find something choiceworthy for the sake of an end. There is much to be said about Aristotle’s teleology, but for my purposes I want to make clear that the concept of a telos at work in the Nicomachean Ethics is broader than any merely psychological conception. That is to say, a telos is not defined as an object of desire or an appropriate object of desire, where that is understood as one that would allow for the greatest overall amount of desire satisfaction.
In the most concrete and literal sense, a telos is a result and connotes that something has been finished.4 Aristotle appeals to this literal sense of telos when he identifies the end of a change (kinêsis)with its stopping point: “When there is some telos of a change, provided that the change is continuous, this last thing is also that for the sake of which (to hou heneka)” (Phys. II.2 194a29–30). However, it is not right to think of a telos, in the sense of an Aristotelian final cause, as being literally the “last thing” to which a change leads. For, as this passage suggests, if the change were interrupted (i.e., not continuous), according to Aristotle, its literal telos or stopping point would not be its telos as final cause. The distinction between the two senses of telos Aristotle points to in this passage is an intuitive one. We often talk about the interruption of natural processes. In order to do so, we must be able to distinguish ends that are mere stopping points from ends that are natural conclusions or fulfillments. It is the latter sort of telos that Aristotle identifies with the final cause, or “that for the sake of which.”5 Since the Physics is an inquiry into the causes of natural change, we can infer that a mere stopping point is not in itself a telos in the sense that interests Aristotle in the Physics.
A telos need not be a result separable from a process that naturally leads to it, either. Aristotle applies his framework of teleological explanation to goings-on that are never, in principle, over. These are activities that achieve their telos at every moment they are engaged in. Aristotle’s paradigm is the activity of seeing. At each moment that I see, my capacity to see is fully realized and achieves its natural fulfillment (Meta. È.6 1048b22–25).6 (This is an example of an energeia [activity] as opposed to a kinêsis [process]. Unlike an energeia, when a kinêsis reaches its telos it is over and goes out of existence.)
What, then, is a telos or end if it need not be a “last thing” at all? An example from the Physics can help:
And further [there can be a cause] in the sense of the telos and this is that for the sake of which. For example, health is the cause of walking. For why does he walk around? We say, “in order that he may be healthy,” and saying this we think we have pointed out the cause. (Phys. II.3 194b32–35)
Aristotle says that the end of walking is health, but it is clear that health is not a telos in virtue of its being the point at which walking comes to an end—or the result upon the achievement of which the walking ceases. Health does not bear such a direct chronological relationship to walking. When Phaedrus’s doctor told him to walk in the countryside for the sake of his health he did not mean that Phaedrus should keep walking until he had achieved health. Nor did he intend for Phaedrus to quit walking simply when he became healthy. Rather, he meant that Phaedrus ought to walk for the reason that it would make him or keep him healthy.7 In other words, health is a telos because it is the benefit to be achieved by walking. Furthermore, because health is the benefit aimed at in walking, it will affect the manner in which Phaedrus walks—how quickly, for how long, and over what terrain he walks. This is a feature that walking for the sake of health shares with all natural changes and activities. The nature of the goal determines the features of the process leading to it or the activity expressing it. If a certain manner of walking is especially productive of health, then it is, to that extent, a good way of walking. Thus, in Aristotle’s Physics, a telos is a normative standard for a process or activity. It is the point at which a change achieves its good.8
Aristotle makes this point explicit at Physics II.2 194a30–33, immediately following the passage with which we began. It is ridiculous, Aristotle says, for poets to write that a person who dies (teleutên) has “achieved that end for whose sake he was born.” “For ‘telos’ does not mean any last point, but the best.” The fact that a change stops at a certain point does not mean that the change is complete. It is only when the good of the process—that is, the result that sets the standard for the success of each stage and of the whole—is achieved that the process is complete.9 A little later in the Physics, when he is enumerating the different kinds of causes, Aristotle says, “and then there is the end and the good of the other things; for the that for the sake of which will be the best and the end of the other things” (Phys. II.3 195a23– 25). And, to take one last example, Aristotle says in the Metaphysics that wisdom is “the science of the end and [i.e.,] the good” (Meta. B.2 996b12).
It might be thought that, although Aristotle intends his conception of a telos to be normative for all cases of natural process or activity, he is only entitled to that result if he assumes that the ends of these changes are desired or intended.10 Aristotle’s example of walking for the sake of health might seem to provide a case in point. It is only because Phaedrus wants to be healthy that health is the end of his walking and so determines what counts as success in that activity. Thus, even in the Physics, where Aristotle tries to explain the natural changes of rocks and plants as well as of animals, a telos ought to be an object of desire. Questions then arise about what sense it makes to attribute desired ends to rocks and plants.
This objection misses the point, however. True, we noticed before that the character or manner of an instance of walking depends on what its end is. If the end is health, the walking will be quite different than if its end is transportation. But this does not show that the end of one and the same type of activity, in this case walking, varies with the agent’s desire. From an Aristotelian perspective, the fact that walking for the sake of health and walking to the bank have different ends and thus different standards for success just shows that they are two different kinds of activity. This conclusion does no outrage to our intuitions. Walking for exercise (particularly speed walking) does in fact seem like quite a different affair from walking to the bank. The fact that both activities involve putting one foot in front of the other is a superficial similarity.11 (Their dissimilarity is reflected in our language: There is a difference between going for a walk and walking to the bank.) Desire, of course, plays a role in establishing health as a telos for a particular person at a particular time.12 But this is only to say that desire for the telos is the means by which the telos guides the agent’s action; it has no bearing on the fact that health is the correct normative standard for exercise walking, and that all instances of that activity are to be deemed successful to the extent that they achieve this telos. (I will return to the connection between desire and ends in human action in section 4.)
This discussion leads us to another feature of Aristotelian ends important for my purposes. In general, Aristotle thinks that a thing’s natural end is intimately associated with its essence or form. Indeed, he says that if you want to study a thing’s telos, you will study its form (Phys. II.2 194a27– b15).13 This is particularly clear in the case of growing things. Aristotle says that the nature or principle of change in something that is born and develops is the form of the mature being toward which it is developing (Phys. II.1 193b12–18). So, the nature of a baby is a human being; the nature of an acorn is an oak tree. And this form that is the telos of growth is also its (not fully actualized) form now. A growing thing develops in a particular direction because that is what it is. The connection between form and telos also holds for things that do not undergo processes of growth. It is part of the nature of fire, for example, that it moves up (DC I.2–3). We can tell that upward movement is the end of fire by its natural tendency, when unimpeded, to move in this direction. Because the natural motion of a thing aims to realize its nature or essence, Aristotle virtually equates form and telos on several occasions (Phys. II.2 194a27–b15, II.7 198a25 ff., 198b1–4, II.8 199a30–32, II.9 200a14–15). I mention the connection between telos and form now because it will become quite important for our discussion of the hierarchy of ends in the Nicomachean Ethics. A chain of teleological dependence ought, from what we see in Aristotle’s physical works, to imply some sort of chain of formal or essential dependence.
According to Aristotle’s technical understanding of a telos as presented in the Physics, an end is a normative standard for the activity undertaken for its sake. The end determines what counts as success in the activity. For this reason, it is closely associated with the nature of that thing whose end it is. As Aristotle says, “What a thing is, and what it is for, are one and the same” (Phys. II.7198a25–26). Furthermore, the end lends its value to the process leading to it; insofar as the end is good, the things leading to it are good. These are the features of ends as they figure in Aristotle’s biology and metaphysics, and, as the close association between an end and a good suggests, it is this very same technical concept of an end that Aristotle uses in the Nicomachean Ethics.
3. TELEOLOGY IN THE NICOMACHEAN ETHICS
3a. NE I.1: Ultimate Ends and Hierarchies
When we turn to the Nicomachean Ethics we see that human ends, like ends in the rest of nature, are goods determining the form and conditions for success of the things leading to them. Furthermore, like ends in the rest of nature, human ends are sources of value for the things leading to them.14 This close connection between goods and ends is familiar to readers of the Nicomachean Ethics, for Aristotle appeals to it in the very first line: “Every craft and every inquiry, and likewise every action and every choice, seem to aim at some good, for which reason people have rightly concluded that the good is that at which all things aim” (NE I.1 1094a1–3, my emphasis). Carpentry aims at houses, and houses are what make carpentry a good skill to employ. My decision to run three miles a day aims at making me healthy, and health is also the benefit that comes from running. This is what Aristotle has in mind by saying that an action’s end is also its good. An end is the source of value for the process leading to it.15 From this it follows that since there is such a thing as the good, it too will be an end of a special variety. It will be the end at which all actions and crafts and so forth aim.16 From this point of view, such a good would most of all be an end.
Immediately afterward Aristotle draws our attention to a distinction among ends with which we are familiar from our examination of Aristotle’s scientific works. Some ends, he says, are in the activities—goings-on that attain their end at every moment—while other ends are results (erga) beyond the actions that produce them (1094a4–6). In the latter case, the ends are better than the activity. This will become an important claim, and it is one with which we might disagree. For instance, we tend to recommend productive activity for its therapeutic value. From this perspective, it doesn’t matter if your clay pots are a bit wobbly or your mosaic made of broken dishes is rather hideous. The value of undertaking handicrafts is in doing something—anything—with your hands. Similarly, we think it is an open question whether the excellent exercise of carpentry is less valuable than the house it produces. After all, such activities can constitute a valuable life. There is no doubt that for Aristotle, however, crafts’ production of useful results—results that are for the sake of something else—detracts from their intrinsic value. For the moment, let us notice only that according to Aristotle, in every case the good is in the end, whether that end is the activity or something beyond the activity. So to this extent, at least, Aristotle’s talk of ends in the Nicomachean Ethics is consistent with his treatment in the scientific works.
Now what particularly impresses Aristotle about the ends of human activity is that they tend to fall into hierarchies. Bridles are the end and good of bridle making, but saying that tells only an incomplete story about why bridle making is good. For bridles—along with stirrups and saddles—are in turn good for the sake of cavalry riding. And the end of cavalry riding, even though it is one of those ends immanent in the action, is itself for the sake of the general’s craft, whose end is victory.17 Just as an ulterior telos is better than the activity that produces it, so too Aristotle notices that whenever ends fall into hierarchies, the higher ends are better (more choiceworthy) than the subordinate ones. In particular, the highest end is the most choiceworthy of the ends in the hierarchy, for it is for the sake of the highest end—the end of the architectonic craft—that the lower ends are pursued in the first place. The value of the highest end makes the subordinate ends worth pursuing (1094a9–16).
What is it for an end to be for the sake of another end? Unless Aristotle has relaxed his definition of a telos, he must mean at the least that the higher ends provide the criteria of success for the subordinate ones.18 This is confirmed by Physics II.2 194a36–94b7. Aristotle distinguishes between the craft of making a tool and the craft of using a tool. The craft that uses a tool determines the tool’s form for the craft that makes it.19 The job of the craft that makes the tool is to know about the materials required—what material can best realize that form. In other words, the telos of the higher craft, what the craft knows how to make and do, determines the form of the telos of the subordinate craft (i.e., the tool it produces). The purpose of riding, for example, determines the form of the end of bridle making; because riding aims at a certain kind of activity, bridles must be constructed in a certain, appropriate way. This means that, in general, the higher end is not just any chance (or even regular) result of a subordinate end, or of our using a subordinate end; it must be that for which the subordinate end was, so to speak, naturally directed all along and by reference to which it was, all along, to be evaluated.20 Of course, this does not mean necessarily that the lower end (the tool) was made with the conscious intention of achieving the higher end (though in the human case this is often true). All that’s necessary is that the lower end’s standards for success are determined by the contribution it makes to the higher end. To take one of Aristotle’s examples (Phys. II.2 194b5–7), a rudder that does not allow the pilot to steer accurately (perhaps because it is warped) is a bad rudder. Thus the process of making a warped rudder is an example of bad rudder making.21
Now higher ends do not set limits for subordinate ones (such as tools) only by determining their form or nature. Higher ends also determine the extent to which subordinate ends are worth pursuing. So, for example, if I want to make a dinner, that end not only determines that I should have a dessert; it also determines how much dessert to have. Thus, we might say that in chains of ends, higher ends limit lower ones in terms of (1) the form of the lower end and (2) the appropriate extent of pursuit. (Again, this is not limited to the human case. Leaves are for the sake of photosynthesis, which is for the sake of the existence of the tree. This higher end determines not only that the tree should have leaves of a certain sort but how much of it ought to consist of leaves [since all leaves and no roots make a hungry tree].)
It seems clear that Aristotle intends the higher ends he discusses in NE I.1 to limit their subordinates in both these ways. The general’s craft determines not only that there should be a cavalry (and hence bridles) and what counts as a good cavalry (and so a good bridle); it also determines how many riders, and hence how many bridles, there need to be. And though Aristotle must be aware that bridle making can have more than one result, he identifies as the highest end that result of bridle making that plays both normative functions.
A lower end in a hierarchy is a genuine end, then, because it provides the normative standard or criteria for success for the process that leads to it. But its value, in turn, is given to it by the higher end toward which it aims. In the first chapter of the Nicomachean Ethics Aristotle suggests that the human good is the end of all choiceworthy human activity in just this way. Because the human good is our ultimate object of pursuit, all the things we make and do are as they are. And to the extent that these lower pursuits succeed in achieving the human good, they are good themselves.
3b. NE I.2: Two Kinds of End
Actually, Aristotle has not quite said everything in NE I.1 about what the good qua ultimate end must be like. For there may be some good upon which all our chains of ends converge but which is itself for the sake of something further. A good like this would certainly be the source of value for our subordinate ends, but it would play this role in the way that cavalry riding is a source of value for bridle and saddle making: It would be incomplete. That is to say, the choice worthiness of such a convergent end could not be fully explained without reference to the further end for whose sake it was worth choosing. In NE I.2 Aristotle argues that given that ends derive their value from that for whose sake they are worth choosing, hierarchies of ends cannot go on ad infinitum. Or, as Aristotle puts it, “we do not choose everything for the sake of something else” (1094a19–20). Thus, there is a difference between ends choiceworthy for the sake of something further and terminal ends at the apex of the hierarchy. The highest good will be an end of the terminal sort.22
We should pause a moment to examine why Aristotle thinks there must be terminal ends. The reason he gives in the Nicomachean Ethics is that without a final resting point, “our desire (orexis) would be empty and vain” (1094a20–21). Aristotle uses the same sort of argument in Metaphysics á.2. There he says that an infinite series of ends would undermine final causation altogether, and that this is equivalent to saying that an infinite hierarchy would destroy the very notion of the good. For “no one would try to do anything if he were not going to come to a limit (peras)” (Meta. a.2 994b13– 14). I take it what Aristotle means is that purposeful endeavor aims at success that is unconditional. But when a chain of ends continues indefinitely, each success is always conditional on the success of the next end down the line. Under these circumstances, all the work seems to lead one nowhere. We need an end that is sufficient of itself to give purpose to the subordinate ends, whose value is not itself dependent on a higher end.23 In a similar vein Aristotle continues that without an ultimate end, “there would be no reason in the world; the reasonable man, at least, always acts for the sake of something; and this is a limit, for the end is a limit” (994b14–16).24 The very possibility of practical reason depends on there being limiting ends, conceivable outside the context of a further purpose to be achieved by them, which can serve to justify the lower aims.
I must confess that as a point about practical reason as related to desire, I do not think Aristotle’s arguments succeed. All we need for desire is an end that is desirable for its own sake, an end we need not desire for some further purpose in order to lend value to the means for and ends subordinate to it. As far as desire is concerned, I see no reason why it matters whether this end is in turn desirable for the sake of further ends. What is important is that it be genuinely desirable for its own sake. If it is, my desire and action will not be in vain.25 In other words, what would be objectionable about an infinite chain of middle-level ends? Aristotle does have a point, however, when we think of ends outside the context of the psychology of human desire. If lower goods (even goods in themselves) derive their value from higher goods, then there must be some unconditional good that is an ultimate source of value. Otherwise it would be indeterminate whether the subordinate goods really are good.26 If X is a means to Y, then we can only know if X is good of its kind by seeing how well it contributes to Y. And if Y, in turn, is by nature a means, then its value will be indeterminate until we see how well it contributes to its further end. Until we do this, we will not know whether Y is good of its kind. And so long as that is in question, we will be unable to judge the value of X. (This is one of the many places where I think Aristotle’s tendency to discuss ends in terms of desire can be misleading.)
Nicomachean Ethics I.2 shows us, then, that there are two kinds of ends: those choiceworthy for the sake of something further and those choiceworthy for their own sakes alone. The good must be of the latter sort. If it exists, it will be the final resting point for all chains of ends; the end that everything aims at but does not look beyond itself for a source of value.
3c. NE I.4: Happiness as a Convergent End
I have been talking about the human good as if it were obvious that Aristotle intends it to be the goal of all chains of ends. This is an assumption that appears to be supported by the very first line of the Nicomachean Ethics: “[I]t has been rightly concluded that the good is that at which all things aim” (1094a2–3, my emphasis). Admittedly, however, Aristotle is here reporting a respected opinion, which he may modify before endorsing. Furthermore, the argument of NE I.2 requires only that for each chain of ends there be an ultimate end. Nothing Aristotle has said in the first two chapters requires that all chains of ends terminate at the same point.27 Of course, Aristotle does compare knowledge of the good to having a target at which to aim; but the metaphor does not force us to think of the good as a bull’s-eye toward which all our actions shoot.
I do believe that Aristotle thinks of the good as a convergent end from the beginning of the Nicomachean Ethics, and certainly before he formulates the finality criterion in NE I.7. For instance, he claims it is the business of the political craft to study the nature of and the means for producing the human good because it is the most architectonic craft. All of the most honored crafts—strategy, rhetoric, household management—are subordinate to it, meaning that their products are choiceworthy for the sake of the good that politics produces (1094b2–3). Furthermore, according to Aristotle, it is the job of political craft to determine the shape of childhood education (1094a28–b2) and indeed of our entire lives, telling citizens what to do and not do (1094b5–6). We may wonder whether it is literally true that the end of the political craft encompasses all other ends (1094b6), but Aristotle certainly intends to say in chapter 2 that it orders the ends of all other crafts. This is sufficient to establish the political telos as the human good (1094b6– 7). As described at the beginning of the Nicomachean Ethics, then, the good, as the end of politikê (political craft), is the unique goal of at least a significant portion of human striving. As far as I can see, this is an opinion Aristotle never withdraws. For example, later in book I he says that “we all do everything else for the sake of happiness” (1102a2–3), and in the last book, when he argues that contemplation is the human good, he urges us to do everything for it (NE X.7 1177b33–34).
According to NE I.2, then, there is an architectonic craft, and when someone acts as the architectonic craftsman, he arranges all human rational activity toward the production of a single end. Aristotle’s point is deeper than this, however. In NE I.4 he suggests that people typically organize their own lives with respect to a single goal. Everyone can agree to call this goal happiness, though naturally people cannot agree among themselves about what happiness is (1095a14–22). This may seem implausible. But Aristotle does not mean that each person pursues a unique goal throughout his entire life. He says that a single person may change his mind about the nature of happiness depending on his circumstances. When he is poor, he will think that happiness is wealth; when he is sick, he will think it is health (1095a23– 25). How can we tell that a sick person thinks happiness is good health? Presumably it is because sick people make regaining health the focus of their lives. They think and talk about it; they endure painful treatments for it; they decide how much and what to eat with an eye to it. In other words, we can determine a person’s conception of happiness by discovering the single end where all, or at least the most important, of his activities converge. So Aristotle is not making the obviously implausible claim that everyone’s life aims at the same goal throughout. Rather, he is making the more likely (though still controversial) suggestion that at each point, everyone’s life aims at some one good thing or other. At least it does if the person has a substantive conception of happiness. (If he has no idea what happiness is, his choices may lead him in a number of different directions and nowhere in particular.) In NE I.4, then, happiness is being treated as the name for whatever end is the terminus of all a person’s choices (or all the important ones). And it’s for that reason, Aristotle suggests, that everyone can agree that it is the good. The good just is the convergent end of a life worth living.
Aristotle never explicitly justifies this assumption, but it should not surprise us if we consider it against the background of Aristotelian natural teleology. All living things have a single telos, according to Aristotle. This is the thought at work in the famous function argument (NE I.7), where Aristotle draws implications from his claim that the ultimate end of a living thing (or its parts) is its ergon, or characteristic work. The ergon of a flautist, for example, is flute playing (1097b25), while the ergon of an eye is seeing (1097b30). Everything that a flautist (qua flautist) or an eye does is for the sake of realizing this end. Now when something has an ergon, there is another way of describing it. We can say that a flautist is a person who plays the flute, and that an eye is the part of the body which sees. In other words, naming a thing’s characteristic work is a way of naming its essence. This is a point we saw earlier in the discussion of ends in Aristotle’s biology. A thing’s telos determines its essence and form. If this is so, then we can see why Aristotle might have assumed that there is only one ultimate end toward which a flourishing human life aims. For if a thing has only one essence, it is not possible for it to have more than one ergon and ultimate end.28 (What we may think of Aristotle’s claim that human beings as such are the kind of thing to have an ergon is another question. But it is not outlandish to think that the unity of a person’s identity hinges on his valuing a certain activity as ultimately good and essential to his self, and that happiness is the expression of one’s identity by successfully acting for the sake of that good.)
I want to leave open for the moment whether or not Aristotle’s teleology requires that the human function be a monistic end. For all I have said so far, the human function might be an appropriately organized set of activities through which human nature is expressed.29 My point for now is that whatever the human good is like, Aristotle’s teleology gives him reason to think it must be a convergent end. Any activity not leading to that end would be, in an important sense, outside the life of the human being. Thus, despite Aristotle’s failure to argue that the good is a convergent end, there is every reason to expect that this is what he believes.30
3d. NE I.5: Using the Concept of an End to Specify Happiness
We have seen that Aristotle’s approach to ethics emphasizes the endiness, or finality, of the human good, or happiness. Given what Aristotle thinks an end is, this approach is promising. An end provides the standards for success for the processes and goods subordinate to it. If happiness is the end of a flourishing human life, then understanding its form should provide insight into what subordinate goods we ought to pursue and how we ought to pursue them. There is reason to think there is such an ultimate end, according to Aristotle, both because all chains of ends must terminate somewhere and because the unity of our essence depends on there being a characteristic activity through which it is expressed. The only task left for Aristotle seems to be figuring out which good activity plays this role.
In NE I.5 Aristotle considers three conceptions of happiness closely tied to three prominent visions of the good life: pleasure (the end of the voluptuary’s life), honor or virtue (possible ends of the statesman’s life), and contemplation (the end of the theoretical life).31 Pleasure is dismissed more by slander than by argument: Aristotle says that the life of pleasure is slavish and fit for cattle (1095b19–20). And he defers discussion of the end of the theoretical life (1096a4–5). But in his discussion of the political life, Aristotle uses the concept he has been developing of the human good as an end to argue that neither honor nor virtue can be the human good. Neither of these goods has the degree of finality that the good must have.
Refined men of action (some of whom would have been in Aristotle’s audience), those who believe the political life is best, are inclined to equate honor with happiness, so perhaps it is the telos of the happy life (1095b22– 23). However, “they seem to pursue honor for the sake of believing themselves to be good” (1095b26–28). Since even honor-lovers choose honor for the sake of something else, it cannot be an adequate account of happiness. For happiness, the human good, must be an ultimate end, choiceworthy only for its own sake. Thus, the admirers of the political life show by their actions that they, at least, consider virtue to be better than honor (1095b29–30).32 But not even virtue is sufficiently final to be the ultimate end of the political life. Aristotle says that virtue seems atelestera, not fully completed or finished (1095b32). Presumably this is because the value of virtue points to the value of virtuous activity. The refined men of action do not want to be good simply in the sense of having a capacity, as if one could be happy even if one slept through life; they want to be good in action. The statesman pursues honor for the sake of virtue, but there is reason to think he also pursues virtue for the sake of performing virtuous deeds.33 Thus, another potential candidate for happiness is shown not to be final enough to be the ultimate end of a life worth living.
We see Aristotle employ the same strategy in his criticism of the moneymaking life. Aristotle says it would make more sense to suppose that any of the ends he has already discussed—pleasure, honor, virtue—is the human good than it would to propose wealth. At least those former goods are loved for their own sakes (1096a7–9). Wealth, on the other hand, is for the sake of something else (1096a6–7). As such, it does not meet the requirement that the human good be an ultimate end, not worth choosing for anything beyond itself.34
By the end of NE I.5 Aristotle has shown that his approach to the human good as an end bears fruit. By keeping in mind that the good is an ultimate end, we can reject many very common accounts of happiness on the grounds that the goods prized in them are worth choosing for the sake of some further end. But the discussion has also revealed an incompleteness in Aristotle’s account. Notice that the problem with pleasure, honor, and virtue is not that they are not genuinely final ends. He says (and seems throughout the Nicomachean Ethics to agree) they are all loved for themselves. The problem is that they are not final enough. Choiceworthy as they are in themselves, they are also worth choosing for the sake of a further end. In other words, NE I.5 points the way to Aristotle’s articulation of middle-level ends in NE I.7. (In fact, pleasure, honor, and virtue are his very examples of what I call middle-level ends [1097b2–5].)
3e. Inclusivism and the Approach to the Finality Criterion
Aristotle’s discussion of the finality criterion is the most sophisticated treatment in the Nicomachean Ethics of the concept of an end and its connection to the good. It also depends on a significant innovation in Aristotle’s understanding of how ends fall into hierarchies. Whereas previously (NE I.2) Aristotle divided ends into those choiceworthy for the sake of something further and those choiceworthy for themselves, he now explicitly distinguishes a third, middle group: ends choiceworthy both for their own sakes and for the sake of something further. These are the perplexing middle-level ends. We have seen why Aristotle should mark out this third category of ends. These are the goods that due to their relatively high degree of finality, are likely to be mistaken for the good. The finality criterion is intended to help us make the distinction. But as I suggested earlier, it is not clear how these goods that have their ends both in and beyond themselves can fit into Aristotle’s theoretical framework. How can a good both determine its own standards for success and have them determined from the outside by a higher good?
In recent decades many scholars have thought that the relationship between middle-level ends and the highest, most final good—at least in NE I.7—ought to be interpreted as the relationship between part and whole.35 At the end of this chapter I argue that this is an ingenious, though ultimately flawed, solution to the problem. Mistaken as I believe them to be, however, inclusivist interpretations of eudaimonia (which say that happiness includes middle-level ends as parts) have become so commonplace among contemporary readers of the Nicomachean Ethics that even those of us who do not accept them tend to give them more credence as interpretive possibilities than I think the text warrants. Thus, before I discuss the finality criterion itself, I want to argue that if Aristotle does intend the finality criterion to specify happiness as an inclusive end, this ought to surprise Aristotle’s readers, given everything he has already said.
Consider the passage immediately preceding the discussion of the finality criterion:
But let us return again to the good we are seeking and see what it might be. For it seems to be different in different actions and crafts since it is one thing in medicine, another in military science, and so on for the rest. What then is the good in each case? Surely it is that for the sake of which the rest [of the actions comprising the craft] are done? This in medicine is health, in military science, victory, in house-building, a house, and in other cases something else. And in every action and choice the end is the good, since it is for the sake of the end that everyone does everything else. So if there is some end of all the things done, this would be the practicable good,36 and if there are several ends, these would be. Coming around from a different angle, the argument has arrived at the same point. And yet we must try to clarify this further. (I.71097 a15–25)
Aristotle’s discussion here flows directly from considerations he raised in NE I.1–2.37 Indeed, the passage seems to be a summary of the argument of these first two chapters. Although he says nothing explicit about whether the good is a monistic or an inclusive end in I.1–2, the ends of the crafts upon which Aristotle’s conception of the good is modeled are all monistic goods—health, ships, victory, wealth, cavalry riding, bridles. Furthermore, nowhere between NE I.2 and the passage quoted above does Aristotle suggest that the good might be an inclusive end. On the contrary, while we learn in NE I.4– 5 that everyone disagrees about what happiness is, all the prominent accounts of the good specify monistic ends—health, wealth, pleasure, honor, virtue. Inclusive ends simply are not on the table in the chapters preceding the finality criterion.38 In other words, the implication of the entire discussion prior to the introduction of the finality criterion is that eudaimonia is a monistic end.
Of course, Aristotle may think that an inclusive good would avoid problems faced by the monistic goods proposed in NE I.4–5. Since he himself argues in the Eudemian Ethics that the highest good includes all the moral and intellectual virtues, the possibility of an inclusive end was at least available to him.39 And in fact, many scholars think he refers to this possibility in the penultimate line of the passage from NE I.7quoted above: “If there is some end of all the things done, this would be the practicable good, and if there are several ends, these would be” (1097a22–24). The implication seems to be that several ends might together compose the practicable good.
We should hesitate to accept this reading, however. First, Aristotle makes it clear that he is still summarizing the results of his previous investigation during the sentence in question. Immediately afterward, he says, “Coming from a different angle, we have arrived at the same point.” But as we have seen, there is no hint of an inclusivist conception of happiness in the previous discussion. We should, if possible, avoid interpreting Aristotle as including a novel idea in what is intended to be a summary of well-established facts. Second, the Greek leaves us free to read the sentence as saying that if there are several ends of everything we do, these would be the practicable goods. In fact, the only other time Aristotle has used the term “practicable good” it was in the plural (NE I.4 1095a16–17).40 It is not obvious there which goods are the practicable goods, but it is likely they are final ends that have a serious claim to being happiness. Thus, more likely than an inclusivist interpretation of 1097a22–24 is one that reads it as saying that if only one good is an end—that is, is worth choosing for itself—then it will be the practicable good, but if several goods are ends, these will be the practicable goods. The discussion from NE I.4–5 has shown that the first option is not a realistic possibility, however. We already know there are, in fact, many goods worth choosing for themselves.
My interpretation is supported from another direction. Remember I said that the only previous use of the phrase “practicable good” comes in NE I.4. It occurs at the very beginning of NE I.4 in a passage where Aristotle says he is taking up the argument he left off in NE I.2. But as we have seen, our passage from NE I.7also takes up the argument left off in NE I.2. In other words, both references to practicable good, in chapters 4 and 7, occur at the same logical moment. But notice how Aristotle characterizes that stage of the argument in NE I.4:
Since all knowledge and choice seeks some good, what is the good at which we say the political art aims, that is (kai) what is the highest (akrotaton) of the practicable goods? (1095a14–17)
According to Aristotle in NE I.4, the project after NE I.1–2 is to discover which one of the practicable goods is the good. It is likely, then, that this is the same project he points to in NE I.7when he says, “Coming from a different angle, we have arrived at the same point.” In other words, the point at which we have arrived is the stage of determining which of the practicable goods is the highest. But if this is correct, we should interpret Aristotle as saying in the previous sentence that if there are many ends, these would be the practicable goods, not that all together they would constitute the practicable good. Figuring out which of a plurality of goods is the good is a task that still faces us. (As I said, after NE I.4–5, the possibility that there is only one practicable good is no longer viable.)
Finally, an inclusivist reading of 1097a22–24 must make sense of Aristotle’s apparent dissatisfaction with the state of the argument in the last sentence of our passage: “And yet we must try to clarify this further.” He goes on to argue that of genuinely final ends—ends worth choosing for their own sakes—some are more final than others. The highest good will be the one that is most final (teleiotaton). In other words, Aristotle gives a more precise account than he has so far of which of the practicable goods is the best. In the inclusivist reading of 1097a22–24, according to which Aristotle says that if there are multiple ends they will together comprise the practicable good, it is less certain why Aristotle thinks he hasn’t been sufficiently clear about the nature of the highest good already. For in this reading he has already told us that the highest good is all the practicable goods, however many of them there happen to be.
Nothing I have said disproves that only an inclusive end satisfies the finality criterion. All I want to show is that nothing in the discussion leading up to the finality criterion suggests an inclusivist conception of eudaimonia. 41 Thus, when we learn in NE I.7about the middle-level ends, it is not at all obvious that we should understand these goods to be constituents of happiness.
3f. NE I.7: The Finality Criterion
So we should expect Aristotle to try to find a way to narrow down the multitude of practicable goods to the single good we are seeking. And this is precisely what the finality criterion does.42 If there are many final (teleia) ends, the highest good will be the most final (teleiotaton) among them (1097a30).43 As we should expect, the relationship between more and less final ends is one of subordination.
And we say that [an end] which is worth pursuing (diôkton) for its own sake is more final (teleioteron) than one which is worth pursuing for the sake of something else; and [an end] that is never choiceworthy for the sake of another is more final than those which are choiceworthy for their own sakes and for the sake of that other; and [dê] [an end] which is always choiceworthy for its own sake and never for the sake of another is final without qualification (haploŝ teleion). (NE I.7 1097a30–34)
Some ends are worth choosing merely for the sake of some further end, which they serve. These are the subordinate craft products, like bridles and ships, that Aristotle discussed in the first chapter. A feature of their being instruments is that one cannot give an adequate account of them without reference to their higher end. Other ends, as we know from NE I.2, are chosen always and only for their own sakes, being that for the sake of which the instrumental goods are worth choosing. These ends, Aristotle says here, are unqualifiedly and most final. In achieving them, our goal-directedness is entirely successful. Clearly, a most final end is more final than a mere instrument. But there are other ends more final than instruments but less final than eudaimonia. These are the middle-level goods, goods choiceworthy both for their own sakes and for the sake of the most final good. The examples Aristotle gives make his meaning clearer:
And eudaimonia most of all seems to be this sort of thing [i.e. a final end without qualification]. For we always choose this because of itself and never because of another, but honor and pleasure and intelligence (noûs) and all the virtues we choose both because of themselves (for even if nothing resulted from them, we would still choose each of them), and we choose them for the sake of happiness, supposing that through them we will live happily. But no one chooses happiness for the sake of these things, nor in general because of anything else. (NE I.7 1097a34–b6)44
(Notice that Aristotle uses the finality criterion to rule out goods like honor and moral virtue, which, as we already saw in NE I.5, seem to be but are not the highest goal of the happy life.) The middle-level ends are genuinely final or terminal. That is to say, they are ends that can bring a chain of activity to a close in the sense that (unlike most craft products) on their own they provide a sufficient normative context for justifying the pursuit-worthiness of the goods subordinated to them. This is the point of Aristotle’s counter factual: If nothing further resulted from these goods, we would still choose them.45 And yet middle-level goods do not top the hierarchy of ends. Though they are intrinsically valuable, they are also choiceworthy for the sake of the most final end: happiness.
4. TELEOLOGY, DESIRE, AND MIDDLE-LEVEL ENDS
4a. What Kind of Goods are Middle-Level Ends?
If goods like honor, friends, and, moral virtue are worth choosing for their own sakes, how can they also be worth choosing for the sake of something else—that is, something quite separate from these, in the way that basking in honor, enjoying our friends, and acting virtuously could not be independent of them?
It might seem that the answer to this question is obvious. Goods choice-worthy for their own sakes and for the sake of something else are valuable for two independent reasons: They are intrinsically valuable, and they are also (broadly speaking) instrumentally valuable. We might say that their value for us is overdetermined.46 So, for example, I desire moral virtue and pleasure each for its own sake. Even if happiness is impossible for me to achieve, I would still want each of them (1097b2–4). But, as a matter of fact, moral virtue and pleasure also tend to lead to happiness, either because they are parts of happiness (an inclusive conception of eudaimonia) or because they contribute to or are instruments to some further good that is identified with happiness (a monistic-end conception of eudaimonia). Under these circumstances, it is an additional reason why I value and pursue moral virtue and pleasure that they have this result.
In effect this interpretation draws a parallel between this passage of the Nicomachean Ethics and the threefold division of goods in book II of the Republic, and it might be thought that the precedent provides some reason to interpret the Nicomachean Ethics along the same lines. There, Glaucon asks Socrates to distinguish (1) the kind of good, such as the harmless pleasures, “that we welcome not because we desire what comes from it, but because we welcome it for its own sake”;47 (2) the kind of good, like knowledge and health, that “we like for its own sake and also for the sake of what comes from it”; and (3) the kind of good that is valuable only instrumentally, while being in itself unsavory. Glaucon’s examples are medicine and moneymaking (Rep. 357b–c). In the interpretation we are now considering, Aristotle’s middle-level goods, choiceworthy for their own sakes and for the sake of eudaimonia, correspond to Glaucon’s second class of good desired for their inherent benefits and for their results.48
But if this interpretation is correct, why would Socrates and Aristotle disagree about which group of goods is best? If happiness belongs in Glaucon’s first category, as it would if the division were equivalent to the one in the Nicomachean Ethics , then Socrates would be preferring the (constitutive?) means to happiness to happiness itself. This is most unlikely.49 In fact the threefold division of goods in Republic II is not equivalent to the threefold division of goods in NE I.7.50 Notice that Socrates says that the second group is the finest because it provides its possessor with happiness by two routes: The good thing will itself make him happy, and it will lead him to other sources of happiness (Rep. 358a). This suggests that Glaucon’s groups are distinguished by the different ways they supply us with happiness.51 Since Glaucon’s middle group is more useful in causing happiness, it is the best according to the terms of his division. Aristotle’s groups, on the other hand, are distinguished by different levels of teleological subordination. From Aristotle’s point of view, insofar as the goods in all three of Glaucon’s groups are valuable because they provide us with happiness, they are choiceworthy for the sake of happiness. That is, none of them is valuable as something haplôs teleion.52 Thus, despite superficial similarities, Plato’s threefold division of goods as objects of desire gives us no reason to think that middle-level ends in the Nicomachean Ethics are essentially ends desired for themselves and for their results.
Nevertheless, as I have said, it is attractive to interpret Aristotle’s middle-level ends as ends desired for two reasons. It is clear from NE I.1 that Aristotle thinks that human practical reasoning is shaped by the perception of hierarchies of ends. Our perception of the desirability of victory, for example, sets in motion a process of discovering the means to that end which, due to their instrumental value, we also find choiceworthy. And it is a feature of our practical reasoning that we can desire an end for itself and for some separate result it produces. Nevertheless this very feature of this interpretation—that it ties the hierarchy of ends very closely to our habits of practical reasoning and desire—makes it an inadequate reading of the Nicomachean Ethics. It is a presupposition of this interpretation that what it is to be an end is to be an actual object of desire, or at least an appropriate object of desire, where appropriateness is judged in terms of general or overall desire satisfaction.53 Thus it presupposes that to understand Aristotle’s talk of hierarchies of ends all we need do is to study the ways in which our actual desires may be subordinated to one another. But the concept of an end in Aristotle is much broader than any such psychological conception. As we saw, this is particularly evident in Aristotle’s natural philosophy, where he is quite willing to talk about the ends of plants, material elements, and a host of other things to which he clearly does not attribute desire or any kind of conscious intentions (e.g., Phys. II.8 199a20 ff.). Even in the animal and human cases, where desire psychology is integral to the workings of natural teleology, an end is not by definition an object of desire or even an appropriate one. It is rather what some creature is naturally for and what therefore it ought naturally to desire.54 Aristotle must believe, therefore, that he can explicate the finality relation in terms that do not make reference to desire or conscious intention. Our interpretation of the various levels of ends must be sensitive to this fact.
We can press the point further. Although it may seem that middle-level goods are those goods desired for themselves and for their results, our examination of Aristotelian ends showed that this apparent solution is not in the spirit of Aristotelian teleology. For we can desire to perform an action to get any of its results. But as we now know, an end is not any chance result. Rather, an end is the good of an activity that determines its form and what it is to be a successful instance of that activity. So, for example, a doctor may treat patients not because he wants to make them healthy but simply because he wants to make a buck. Nevertheless, we judge the quality of his care not by whether it causes the patient to give him money—a horrible quack may be quite adept at satisfying his desire for money—but by whether the patient is cured. To use Aristotelian terminology, desire aims at the apparent good, but the apparent good can come apart from what truly is good (NE III.4; DA III.10). Thus, the set of all objects of desire is wider than the set of genuine ends. The genuine ends are those results in terms of which the action in question, as such, truly is subject to evaluation.
So Aristotle’s scientific conception of an end is not consistent with thinking of middle-level ends as goods desired for two independent reasons. It is conceivable, I suppose, that Aristotle relaxes his conception of an end in his ethical works. It might be that, so far as the Nicomachean Ethics is concerned, a telos is simply an object of desire, or an appropriate or correct such object (as that might be judged relative to satisfaction of desire in general). (Of course, as we saw in section 3a, Aristotle does seem to employ that concept at the beginning of the Nicomachean Ethics in the same way as he does in the Physics.) But this is not a move those of us who are sympathetic to Aristotelian ethics should be happy to see him make. After all, as Aristotle is himself aware, different people can desire different things for all kinds of reasons. Indeed, people can even desire for their own sakes things that are quite clearly instruments. (Misers and those who emulate the money-making life do this, for example.)55 The whole point of examining teleology in an ethical work, however, is that discovering what our ends really are will reveal to us what we ought to desire. If Aristotle relaxes his definition of telos now to include any object of desire or objects of desire authorized by the overall satisfaction of desire, we have no reason to think that the structure of our ends (i.e., objects of desire) can tell us anything about how we ought to guide our lives. In particular, Aristotle will not be able to argue from the (quite implausible) fact that people do aim at a unified conception of the good throughout their lives to the conclusion that it would be a great folly not to do so (EE I.2 1214b10–11).56 Any realist framework that uses teleology as a means of discovering the good should resist collapsing the distinction between an end and an object of desire.
4b. Human Desires and Human Ends
It is important to realize that Aristotle’s vocabulary in the Nicomachean Ethics is often normative. He often talks about ends as being diôkta and haireta, and often he does intend to point to our actual patterns of desire to support his claim. But although diôkta and haireta can mean ‘pursued’ and ‘chosen’, respectively, they can also mean ‘worth pursuing’ and ‘choiceworthy’. It is the normative claim that is primarily important to Aristotle. He draws our attention to the fact that, by nature, we tend to desire in certain ways as a piece of his argument that we ought to desire in these ways, that is, that these really are our ends. But the ends exist independently of desire, whether or not we happen to want our true end on a given occasion.
More needs to be said, however, about the connection between human ends and desires. For in some sense the apparent good can play the role of a telos, as Aristotle is well aware (MA 6 700b28–29).57 After all, the fact that a person wants X will shape his behavior in pursuit of X. And to some extent the nature of what he wants will determine what counts as a successful attempt to get it. But only to some extent, for even though a particular representation of the apparent good may aim at what is not in our interests, the apparent good as such does aim at our true interests. Our capacity for desire (which as such aims at our apparent good) is the natural means by which we fulfill our nature, that is, our telos. This capacity would not fulfill its function if it pointed us to the genuine good only by chance. (Indeed, in Aristotle’s understanding, this could not be desire’s function if the connection were only a chance one.) Thus, there is a perfectly good teleological sense in which our actions may not be successful even though they achieve the particular end the agent intended. If a person has an incorrect image of his telos and thus of what actions are appropriate under the current circumstances—think, for example, of Don Quixote—what he does can be successful so far as his conscious intentions go, but strictly speaking it is not successful at all. But even when people are wrong about what particular thing they ought to pursue, it is still true in a general sense that they are pursuing their human telos; they are exercising a capacity that, by nature, is for that.
This is all quite abstract. The crafts provide a concrete manifestation of the way in which normative goals can come apart from individual desires. The different elements of, for example, medicine—cutting and burning, alterations in diet—were developed to achieve particular results, which the practitioners took to be good. And because the results were in fact good, a pattern of activity emerged as the craft of medicine. The telos of medicine and the standards for right medical practice are now cut loose from the desires of the individuals who practice it. A particular doctor may effect a cure for any number of reasons; indeed, he may not care about the health of the patient at all. Nevertheless, so long as his actions are guided by the medical craft, they are for the sake of health. If the doctor makes a mess of cauterizing a patient, it will be no good for him to say, “Look, I don’t really care about whether this person lives or dies. I only wanted money and my “cure” certainly got me that.” It is still appropriate for us to judge his practice of medicine by whether it produces health, because medicine is for the sake of health.58 An activity that is not correctly judged by its conduciveness to health, whether or not a doctor happens to perform it, is simply not the exercise of medicine.
The independence of a craft’s ends from the desires of the individual craftsman is even more starkly displayed in hierarchies that span over several crafts. To return once again to an example from NE I.1, the craftsman whose actions are directed toward the production of bridles may have no knowledge of the cavalry. Indeed, if the bridle maker is a slave, the freedom that good horsemanship will achieve under the direction of a good general may not even be open to the craftsman to enjoy; and we do not deliberate about how to produce impossibilities. Thus it simply is untrue that every craftsman makes bridles because he wants the peace that follows in the wake of victory and which victory itself is for. Nevertheless, Aristotle says that the bridle maker acts for the sake of the general’s end (1094a15–16). That is because the norms by which he knowingly guides his bridle making depend ultimately on the victory that is the general’s end. What these considerations reveal is that the norms established by the ends that guide our actions may not be determined by the results we hope to achieve by them.
Nevertheless the results we hope to achieve by our actions, and with reference to which we consciously make our decisions, are in some sense their ends. And the standards for the agent’s actions will appear to the agent to be set by their contribution to this apparent good. (The craftsman may judge his performance as a craftsman by the number of bridles he sells rather than by whether expert riders approve and buy them.) The point is just that his actions would in truth be better judged by another standard.
Aristotle seems to assume we can see true ends behind the actual objects of desire throughout the Nicomachean Ethics. For instance, in NE I.5 he says that the refined men of action take honor to be their telos. As a matter of fact, however, the end of their actions is moral virtue, since the honor they knowingly want depends on that. Aristotle’s idea is not that these gentlemen unconsciously desire virtue as their most final end.59 (Tomy knowledge, Aristotle never discusses unconscious desire.) Rather, they desire honor as their most final end and lead their lives in its pursuit. But since they desire honor only when given as a reward for virtue, Aristotle believes that virtue is their true telos; virtue is what sets the standard for their actions. (And, as I mentioned before, even this may turn out not to be their true end.) Again, at NE X.6, Aristotle says that many people act for the sake of relaxation as their final end, but relaxation is for the sake of work (1176b32–1177a1). So these people have misjudged their end and thus the norms for their behavior. (We can easily see here what the bad consequences would be. If I think that life is for the sake of relaxation, I may be proud to have reduced myself to a state of such lassitude that I can hardly lift a pinky. But since relaxation is for the sake of work, in reality I will be an utter failure.)
We should not be misled, then, by the connection between desire and natural goal-directedness. While our desires can provide important evidence about the nature of the human good, they do not in themselves determine what the human telos and good are. An end, or “that for the sake of which,” is not, strictly speaking, an object of desire. Thus, the fact that a good can be desired for two reasons does not explain how it can have an end in itself and an end in some other, independent good.
5. THE PUZZLE IN NE I.7 AND TWO POSSIBLE SOLUTIONS
The upshot of the examination of Aristotle’s notion of an end, or “that for the sake of which,” is this: It’s not at all clear how there could be goods that are worth choosing both for their own sakes and for the sake of eudaimonia, if eudaimonia is something other than just another name for particular middle- level ends. For goods like these would be goods that, in being of value for some further end, have a normative limit outside themselves on whose value their own value depends, and at the same time, in being of value just in themselves, need no such external source of value but set for themselves the norms of their own goodness. They are goods that somehow or other contain within themselves the standards for what it is to be a good of that kind and, at the same time, have normative standards imposed from the outside. But how could this be? It seems as if any intrinsic value a good might allegedly have would be denied by its being for the sake of something further;60 and conversely, any instrumental value a good choiceworthy for its own sake might have would seem to be a mere happy coincidence and irrelevant to the natural, intrinsic value of the good as determined by its telos.
We have seen that we cannot analyze middle-level goods as goods that are desired for two independent reasons—for themselves and for some separate thing that they serve as instruments for achieving. There are two other possible solutions that do not make the mistake of confusing a telos and an object of desire.
First, perhaps the two ends coincide by chance. That is to say, it might be that middle-level goods have their ends in themselves but, in addition, regularly promote some further goal. As it so happens, what it takes to be a good instrument to that further goal is exactly coincident with what it takes to be a good thing of that kind for its own sake. In this case, we might call the chance result a further end of the intrinsically valuable activity. This account of middle-level goods is unpromising as an interpretation of the Nicomachean Ethics because it posits a regular but chance connection between their two ends. In the Physics Aristotle denies regular chance connections (II.8 198b35). But at a deeper level, this interpretation is questionable because it supposes that a good may have two separate, teleologically unconnected ends. Although I have not yet found evidence to this effect, I doubt very much that Aristotle would admit that the proper pursuit of or natural coming to be of a thing could have two entirely independent ends. For if we say that a thing has two ends, we are saying that it has two separate natures or forms. (That’s because, as we saw above (pp. 22–23), a natural telos is closely tied to the form or essence of the thing.)61
Another and more interesting interpretation of middle-level goods is to say that, for these goods, the two normative functions of the telos come apart into the function of determining the form and success conditions for activities in pursuit of it and the function of determining how many such pursuits there should be or how frequent they are (p. 18 above). Intrinsic goods might determine their own natures but have the limits of their appropriate pursuit set by a further end. In other words, the most final good might function as a side constraint on the pursuit of other intrinsically valuable goods—the middle-level goods.62 So, for example, a good dinner party is choiceworthy for its own sake; what it is to be a good dinner party makes no reference to some external result it might have. Nevertheless, how often we ought to have dinner parties is not a question we can decide without looking to some further goal. In particular, different conceptions of happiness will endorse the joys of fine dining to different degrees. In this way, a middle-level end could be choiceworthy for its own sake, in the sense of not depending on some further end for its form, while yet being choiceworthy for a separate further end that imposes the limits for when and how much it is to be pursued.
For two reasons, however, this interpretation of middle-level ends is unsatisfactory. First, in this reading, eudaimonia no longer seems to be a convergent end. The highest good is a special kind of intrinsic good, but it is no longer the ultimate object of all the happy person’s activity. But as I have argued, in the Nicomachean Ethics a good is to be conceived of as an end or “that for the sake of which,” and the highest good is to be conceived of as the super end on which all chains of ends converge.63 In the example above, the person does not aim at the highest good when having a dinner party. Rather, the highest good trumps all others in a conflict—including the good of having a dinner party. And that raises the second problem with this interpretation. In what acceptable sense do I pursue X for the sake of or for the end of Y just because the way I pursue X on this occasion does not obstruct my pursuit of Y? Treating a more final good as a trumping good stretches the meaning of “acting for the sake of an end” beyond recognition. It is not, at least, what Aristotle usually means by “acting for the sake of an end” in his technical sense. Thus, middle-level ends cannot be related to eudaimonia as to a side constraint, or at least this does not capture the sense in which middle-level ends are choiceworthy for the sake of eudaimonia. In this interpretation, eudaimonia is not a convergent end for all the happy person’s choices; indeed, it hardly seems to be a telos at all for his choice of middle-level goods.
There may be some confusion on this point. The supreme finality of eudaimonia is not to be understood as its carrying more weight than other, subordinate goods in our deliberations. A craft product, insofar as it is an end, does not carry more weight in our deliberations than the means leading to it do. There is no conflict between means and end that is resolved in favor of the end because, as an end, it is more important than the means (in the way that a conflict between health and tasting something sweet might be resolved by an appeal to the superior value of the former). Rather, its being an end specifies what those means should be and is the source of their value. A means does not simply have less value than the end (1094a5–6); it has no value outside the context of aiming at its end. If someone were to pursue a means beyond the point that was conducive to achieving the end, we would urge him to stop on the grounds that what he was pursuing was a means to something else. But the force of this advice would not be that the end is more important than the means. Rather, we would tell him that in pursuing the means excessively, he would be defeating the good of having those means in the first place. So we should not imagine that the supreme finality of eudaimonia can be grasped in terms of its weightiness. The highest good is more important and more valuable than other goods, but that is not the significance of its being most final. To call eudaimonia most final is to specify the way in which it is more important than the other goods: It is what our pursuit
6. ACKRILL’S INCLUSIVIST SOLUTION
It seems, then, that we must find some other way of interpreting the relationship between middle-level ends and the highest good. It was just our problem of figuring out how ends that are choiceworthy for their own sakes could also be for the sake of eudaimonia that John Ackrill addressed in his famous article, “Aristotle on Eudaimonia.” Ackrill thinks it is mysterious how, in NE I.1, “one action or activity [can be] for the sake of another, in cases where the first does not terminate in a product or outcome which the second can then use or exploit” (1974, 18).64 And in general Ackrill wonders how “some things may be done for their own sake and may yet be done for the sake of something else.”
Ackrill suggests that if eudaimonia is an inclusive end—in this case including all intrinsic goods—then a middle-level end will be choiceworthy for eudaimonia’s sake, as a constituent is for the sake of the relevant whole. Now, although I disagree with Ackrill’s particular solution to this problem, I believe there is something importantly right in the motivation that underlies it. For, in essence, Ackrill says that Aristotle’s position about the relation between final and most final goods is coherent so long as the relation holds precisely because middle-level goods are intrinsically valuable.65
. . . [O]ne can answer such a question as “Why do you seek pleasure?” by saying that you see it and seek it as an element in the most desirable sort of life. . . . The answer to the question about pleasure does not imply that pleasure is not intrinsically worth while but only a means to an end. It implies rather that pleasure is intrinsically worth while, being an element of eudaimonia. Eudaimonia is the most desirable sort of life, the life that contains all intrinsically worthwhile activities. (1974, 21)
Ackrill has discovered a very natural way of explaining the idea that we can seek intrinsic goods for the sake of the highest good without downgrading their status as intrinsically valuable. Their instrumental value depends on their being intrinsically valuable.66 From this point of view, some form of inclusivism is quite attractive.
I see no problem with supposing that X can be for the sake of Y in virtue of being one of Y’s constituents.67 This seems to me a perfectly legitimate application of the “for the sake of” relation. Nevertheless, as I have argued, a careful reading of NE I.1–7makes it most unlikely that Aristotle intends all intrinsic goods to be constituents of eudaimonia, as Ackrill suggests.68 It is a further strike against an inclusivist interpretation of NE I that it is hard, if not impossible, to reconcile with NE X, in which Aristotle says that contemplation is true eudaimonia.69 But there is also a philosophical reason for denying that a set of goods can perform the function of a higher end of giving the form to its subordinate goods. That is because the set in question is conceptually dependent on its members. Suppose happiness is the set of all intrinsic goods. There cannot be a set (with members) of goods choiceworthy for their own sakes unless there are already various determinate (or determinable) goods to fill it. Thus, the idea of the set in no way shapes the essence of its members. But in that case, an inclusive end cannot function as an end, in Aristotle’s sense. This is a rather crude version of inclusivism. But even more-sophisticated versions run into trouble. Terence Irwin (1991, 389), for example, has argued that eudaimonia is the set of all intrinsic goods structured by the demands of moral virtue. Moral virtue limits the amount of the other intrinsic goods the happy person should pursue. Even here, though, what goods we ought to pursue are not determined by the nature of eudaimonia. We can know what things are intrinsically valuable only independently of eudaimonia (since eudaimonia just is the appropriately structured set of intrinsic goods). Thus happiness, even in a more sophisticated version of inclusivism, does not play the value- and form-conferring role I have argued is essential to its being the most final end.70 In fact, insofar as morally virtuous action determines how much of the other goods the happy person ought to pursue, it (and not happiness as a whole) is playing something like the role of a most final end. As we saw, a telos determines (1) the form of subordinate goods and (2) the appropriate extent of their pursuit (p. 18). Thus a structured inclusivist end, such as Irwin proposes, does not fulfill (1) at all and fulfills (2) only in virtue of one of the members of the set: morally virtuous activity.
Still, although Ackrill’s interpretation of NE I.7is, I think, incorrect, we can learn something from his argument. What is needed is an interpretation of the relationship between the middle-level goods and the highest good that makes the intrinsic value of the lower goods a necessary part of their being for the sake of the highest good. That way, when the happy person pursues eudaimonia through honor and pleasure and the like, he must do so by recognizing their final value.
In this chapter I have argued that, although the human good or happiness is understood by Aristotle as a most final end for the sake of which ultimately all other goods are worth choosing, it is unclear exactly how middle-level ends—those goods choiceworthy for their own sakes and for the sake of something else—are related to happiness. But this problem is particularly acute if it turns out that morally virtuous activity is a middle-level end, choiceworthy for the sake of something beyond itself. For not only is morally virtuous action worth choosing for its own sake, it is part of the definition of such action that it be chosen for its own sake (NE II.4 1105a31–32). An action just is not fully virtuous unless the agent desires it for its intrinsic value. Thus, not only is it unclear how morally virtuous action could be subordinated to an independent end, such as contemplation, it seems as if no life could be morally virtuous unless it made the exercise of moral virtue and not some independent good its ultimate goal.
Let me be clear. Nothing that Aristotle says in the opening book of the Nicomachean Ethics suggests that morally virtuous action is a middle-level end. (He says that moral virtue is a middle-level end, and that is presumably in part because it is worth choosing for the sake of its actualization.) And yet it is important for us to see that there are suggestions in NE I that some virtuous activity or other may turn out to be teleologically subordinated to the highest good.
First, there is the function argument of NE I.7. Aristotle argues that whenever something has a function or characteristic work, performing that function well is its good (1097b25–27). This is not really a new approach to the subject of the human good. The function of a thing just is the end to which it is naturally suited.71 Playing the flute just is the activity for the sake of which the flautist qua flautist does all that he does. Seeing is the end for the sake of which the eye is directed. Aristotle’s point is not that a thing’s function is the only activity it has a capacity to perform. A carpenter is capable qua carpenter of doing many things in addition to building a house—selecting screws and wood, building sawhorses, reading architectural plans. Rather the point is that where there is a function, all the other typical capacities and activities are worth choosing for its sake. In fact, it is more correct to say that excellent functioning is the end, since it is the excellent activity that serves as a normative standard for that kind of activity in general. This is why, although Aristotle does not say so explicitly, if a thing has a function, excellent or virtuous functioning will be its good. Virtuous functioning is a thing’s ultimate end.
Now, Aristotle claims that activity in accordance with reason is the human function (1097b34–1098a5); thus, activity in accordance with rational virtue or excellence is the highest human good (1098a16–17). But after this he adds a phrase that has been the object of much scholarly debate. Punctuating in accordance with Bywater’s text, it reads: “If it is thus, the human good is activity of the soul in accordance with virtue, and if there are many virtues, in accordance with the best and most final (teleiotatên)” (1098a16–18). What sort of activity is activity in accordance with the most final virtue? In thinking about this question, it is important to bear in mind the connection between functioning and ultimate ends. Since Aristotle has so recently been defining most final as “choiceworthy for its own sake alone and never for the sake of another,” there is a strong presumption that we should understand “most final” here in the same, teleological sense.72 Of course, in the function argument, Aristotle is talking about the most final virtue, whereas in discussing the finality criterion he was talking about the most final end. But it is likely that the most final virtue is the virtue whose activity is a most final end. At least it is unlikely that there is any virtue that is worth choosing only for its own sake and never for the sake of the activity it makes possible. Thus it is unlikely that any virtue (as opposed to virtuous activity) could be most final, strictly speaking, in the sense specified by 1097a30–34. And, as I said before, excellent functioning is a thing’s ultimate end. It is probable then that the most final virtue is the virtue whose actualization is the most final activity and end. If I am right, we can rephrase the conclusion of the function argument this way: The human good is virtuous rational activity or, if there are several virtues, the virtuous activity that is a most final end.73
We should notice that this qualification at the end of the function argument is reminiscent of the finality criterion: “So that if there is only one [end] that is final, this would be the [good] we are seeking, but if there are several, the most final (teleiotaton) of them [would be]” (1097a28–30). As we have seen, Aristotle goes on to explain that one end will be more final than other final ends if it is choiceworthy for itself alone while the other ends are choice-worthy for its sake. Thus, if one form of virtuous human functioning is most final, that ought to mean that the other forms of virtuous activity are teleo-logically subordinated to it. Aristotle does not emphasize the subordination of the other forms of virtuous activity, but it is entirely appropriate to infer that this is what he has in mind.74 Excellent functioning is an ultimate or final end; thus if one such kind of activity turns out to be more final than another, that means that the latter is worth choosing for the sake of the former. That is just what it means for one end to be more final than another. Thus we see Aristotle leave room in NE I.7for the possibility that some virtuous activities are middle-level ends.
The possibility seems to be suggested again at NE I.8 1099a29–31: “And these [best activities], or one of them—the best—we say is happiness.” If only one of the excellent activities is happiness and the best, then only one of them will be the most final end. The others ought to be subordinated to it. Of course we do not know yet which one exercise of virtue is best. I do not mean to suggest that in NE I.7–8 Aristotle is already saying that there is certainly a teleological hierarchy of rational virtuous activities, much less that the exercise of some other virtue is more final than the exercise of those moral virtues with which Aristotle’s well-bred audience is familiar. That argument does not come until NE X.75 But in retrospect we can see that he is laying the groundwork for his argument that this one highest virtuous activity is contemplative. And that means that morally virtuous action is choice-worthy for its own sake and for the sake of something higher.
In the last book of the Nicomachean Ethics, Aristotle denies that morally virtuous action is the ultimate aim of the happiest life. Instead, the best life aims at theoretical contemplation as perfect, most final (teleia) happiness (1177a17, 1177b24). Since Aristotle clearly believes that the philosophical life will be morally virtuous (1178b5–6), and since he also thinks that all choices ought to be made for the sake of contemplative activity (1177b33– 34), it is likely he thinks that in the best life, morally virtuous activity is chosen for the sake of contemplation.
So, if we cannot come to terms with middle-level ends, we cannot understand—despite eight books of the Nicomachean Ethics devoted to the subject—how, in Aristotle’s account, moral virtue figures into the happiest human life. I suspect there will be a variety of ways in which middle-level ends are choiceworthy for the sake of happiness. The account of how the happy person seeks honor for the sake of happiness may not be the same as the account of how he chooses his friends for the sake of happiness. But since there is a special problem about the status of moral virtue, I will devote the rest of this discussion to understanding how it is that the activity of moral virtue can be choiceworthy for the sake of an independent good, such as theoretical contemplation, without undermining its own intrinsic desirability.
1 The second major criterion is self-sufficiency. I discuss this criterion in chapter 3.
2 The human good, happiness, is an end for the statesman as well, but it is not his end alone. Indeed, when at 1094a24 Aristotle conceives of the good as a target (skopos), he has not even mentioned the standpoint of the politikos yet. It is only after Aristotle claims the good is the target of human life that he declares that the political art aims to produce it for all citizens. Thus, while I agree with Bode´u¨ s (1993) that the NE is addressed to future statesmen whose job will be to create happiness for others, I believe Aristotle also intends to address his audience as individuals seeking their own happiness. See Pangle (2003, 8–16) for a good discussion of this issue.
3 Although the problem I will raise is particularly clear when we interpret eudaimonia as a monistic end, it arises in an inclusivist interpretation as well.
4 One verb for dying is teleô, literally “come to one’s end.”
5 Throughout I will use telos and “that for the sake of which” interchangeably. The two terms are not exactly coextensive. All ends in Aristotle are thats for the sake of which, but it would be odd to call certain thats for the sake of which ends. For instance, a beneficiary is a that for the sake of which but not a telos. (Aristotle himself notices this difference in thats for the sake of which; see chapter 4, below.) I believe, however, that everything I say about thats for the sake of which in this chapter applies to those that are ends.
6 Energeiai (activities) have their ends in themselves and therefore, unlike kinêseis (processes) whose ends are outside themselves, they continue going on even though their end is already achieved (Water low 1982, 186). See Charles (1986, 132–139) and Reeve (1995, 101–106) for further discussion of the different ways in which activities and processes are related to their ends.
7 At least this would be right on our own understanding of why one ought to walk. There is evidence in the Physics, though, that walking was believed to produce health by loosening the bowels (Phys. II.6 197b22–28). In that case, health might be the telos of walking qua stopping point.
8 We may wonder how an activity can contain its own normative standard, i.e., its own end. It may help to think of the end that determines an activity as the instantiation of a certain type of activity. So, for example, it is because I want to dance a waltz that I move my feet in a certain way, but this process of dancing is itself the waltz. The end is in the activity and sets its standards for success (Reeve 1995, 104–105).
9 See Meta. Ä.16 1021b23–25: “We call complete (teleion) . . . (3) The things which have attained a good end are called complete (teleion); for things are complete in virtue of having attained their end” (Ross translation from Barnes 1984).
10 See Broadie (1990) and Cooper (1982), who argue against this interpretation. I leave aside for the moment whether there is a useful distinction to be made between good qua telos and good simpliciter.
11 Of course, there is a reason speed walking is called a type of walking: It was derived from normal walking. Its inventors took the ordinary activity of walking and transformed it, almost beyond recognition, into an exercise activity.
12 It may be that we need to posit desire in order for a telos ever to be a genuine cause of human action on a particular occasion. Desire is not necessary for all human teleological behavior, however. For example, digestion is for the sake of the perpetuation of the human form (DA II.4), but we do not achieve this end by desiring it. There is a further question how final causes can be genuine causes and not merely explanations. See Furley (1996) for discussion and bibliography.
13 Charlton 1970, 97 not. ad 194a21-b9. See also his introduction, p. xvii.
14 I take this way of phrasing Aristotle’s point from Korsgaard 1996b. Richardson denies that ends, at least in the NE (he acknowledges that the Prime Mover may be a special case), are sources of value. Instead, he argues, they play an action-guiding role. What I have argued is that Aristotelian ends are sources of value because they play a regulative role. Richardson himself admits that his separation of these two ways of conceiving an end depends on thinking of ends of action as essentially objects of desire (1992, 329 n. 4). This is the view of Aristotelian human ends against which I will argue in the remainder of this chapter.
15 “And [happiness] seems to be this way [i.e., one of the honorable and final (teleiôn) things] also because it is a first principle (archê); for it’s for the sake of this that we all do everything else, and we suppose that the principle and cause (aition) of goods is something honorable and divine” (NE I.12 1102a2–4). Here eudaimonia is both the goal of all other activity and the source of the goodness of all human goods.
16 Aristotle is not innovating here; Plato and Eudoxus also thought of the good as an end. The emphasis on the fact that the human good, or happiness, is an end is peculiarly Aristotelian, however, and was enormously influential. Think, for instance, how natural it is for Cicero in De Finibus to expound the different Greek moral philosophies in terms of a dispute about the nature of the human end.
Note also that I do not think that the first line of the NE offers an argument for the existence of the highest good. This is something Aristotle takes for granted. What he argues here is that we should conceive of the good as a special kind of end.
17 As Kraut has pointed out, Ackrill is wrong to argue that those ends found in the activity that “produces” them cannot be for the sake of something else (Kraut 1989, 213–217; Ackrill [1974] 1980, 18–19). For such activities can have results that can be used by superior crafts. Or as I would put it, such activities in action can be used by a superior craft (see note 20, below). (Aquinas [I.i.18] suggests gymnastics as an activity subordinated to another end, viz., health.) I think, however, that Kraut is wrong to agree with Ackrill that these activities are always choiceworthy for their own sakes, if by that he means that they are ceteris paribus worth pursuing regardless of any relation they may have to a higher end. It is true that in Meta. È.6, the locus classicus for the distinction between activity and process, all of Aristotle’s examples of activities are choiceworthy for themselves. But the activities described by Aristotle in NE I.1 suggest a different picture. Though the end that governs the performance of the activities mentioned in NE I.1 is the activity itself, these activities (at least in I.1, where ends choiceworthy for their own sakes are not mentioned or explicitly envisaged) are only means for the production of some further end. That is to say, they cannot be properly understood or sensibly aimed for outside the context of the higher crafts whose ends they serve. Ends choiceworthy for their own sakes, on the other hand, are precisely those ends that need not be for the sake of something further. For, as Aristotle says in I.7, they would be choiceworthy even if they led to nothing else. Perhaps there is additional support for my claim that activities are not all choiceworthy for themselves in Phys. II.3 (and Meta. Ä.2): “The same is true of all the means that intervene before the end, when something else has put the process in motion (as e.g. thinning or purgingor drugs or instruments intervene before health is reached); for all these are for the sake of the end, though they differ from one another in that some are things the doctor does (erga) while others are instruments (organa)” (194b35–195a3). If some of the things a doctor does are activities, then Aristotle describes activities here as for the sake of an ulterior end. Since a doctor’s activities are unpleasant, they would hardly be choiceworthy for their own sakes.
It is a trick of language that all activities, because they have their ends in themselves, seem to be choiceworthy for their own sakes in the strict sense. Activities have their ends in themselves in the sense that the correct engagement in that activity is guided by the end of producing that activity itself. In the same way, a craft product guides its proper production. Therefore, in the first instance at least, just as the craft product makes its production valuable, the activity qua end makes the engagement in that activity worthwhile. But notice that this does not commit us to the thought that the activity is necessarily valuable regardless of what it leads to. That would be as unfounded as saying that the craft product, because it functions as an end for the craftsman, must be intrinsically valuable.
18 Kraut (1989, 200–201) also partially analyzes the “for the sake of” relation as having “a normative component: when A is pursued for the sake of B, then B provides a norm that guides A.” In addition Kraut believes that a causal connection between A and B is essential if A is to be chosen for the sake of B. Although, of course, I believe that there is a causal connection when the “for the sake of” relation is instrumental, not all “for the sake of” relations are (efficiently) causal in Aristotle’s teleology. (See chapter 4, below.) Kraut (1989, 13) believes that all “for the sake of” relations in Aristotle are instrumental.
19 Aristotle is following Plato’s Cratylus 388a–390d.
20 Thus we cannot explain the fact that activities (whose ends are in themselves) are subordinated to other ends as Kraut (1989, 213–217) does in his response to Ackrill (see note 17,above). Kraut says that, although activities are choiceworthy for their own sakes, they have results that can be used by other crafts, so they are also choiceworthy for the sake of those ulterior craft products. But now we see that an activity,
cannot be subordinated to Y simply in virtue of Y’s being a result of 
21 The craft example might cause us to worry that subordinate ends are not really bad or good in themselves depending on their usefulness; it might be that they are only bad or good relative to a purpose. After all, I can use a stone as a hammer, and it might be good or bad for that purpose. But that is hardly relevant to whether or not it is good as a rock. It is to avoid this worry that we should remember the role crafts play in giving a new form to matter. The usefulness of found objects is not the product of craft since their forms are not changed simply by our using them. Craft can fashion a stone (and wood) so that it has a new form. When it has a new form—of a hammer or of a stone wall, for example—it also has a new telos, dictated ultimately by our well-being.
22 Notice, I think Aristotle is assuming that there is a good, and that it is a convergent end. B See section 3c for further discussion.
23 See Frankfurt (1992), who uses this premise from Aristotle in an interesting argument that final ends have instrumental value. For the record, Aristotle might agree that treating certain (correct) ends as final produces valuable results and so is instrumentally valuable in that sense. What he could not agree with is the implied further claim that the final ends themselves are, in a technical sense, choiceworthy for the sake of their instrumental value. That is to say, these goods do not take their form and standards of success from the nature of the good they do us when treated as final.
24 Translation of these lines modified from Ross (in Barnes 1984).
25 Thus, while I agree with Broadie (1991, 13) that “[m]y desire must settle at some point because until it does I do not actually desire anything for the sake of which I then desire the things through which I can accomplish it,” this does not settle the question of why there needs to be some absolutely ultimate end.
26 This suggests another solution, proposed to me by John Cooper. The sort of orexis (desire) that Aristotle has in mind in NE I.2, where he says that it cannot go on forever without being empty, is boulêsis, or rational desire (1094a19). (Note the connection between reason and the existence of terminal ends in Meta . á.2, cited above.) Whereas other kinds of desire might be readily satisfied by achieving any good choiceworthy for its own sake, rational desire might not be satisfied by goods of that kind if they are also choiceworthy for the sake of something further. For so long as a good derives its value from some other, higher, good, it will be rational not to be satisfied until we can, in some way or other, possess that higher good. Our wanting the lower final end will give us reason to aspire to the higher end, which is its source of value. This seems to me a promising suggestion.
27 Richardson 1992, 344–349. Richardson is particularly concerned to show that convergence is not assumed before the NE I.7discussion of degrees of finality. In other words, “convergence is [not] built into Aristotle’s notion of the highest good.” He admits that eudaimonia, as a matter of contingent fact, may be a convergent end (349 n. 51).
28 Aquinas (I.ix.106) makes a similar point. (I thank Chris Shields for bringing this passage to my attention.) See pp. 43–45 for further discussion of the function argument.
29 However, I do think that what I have said shows how unlikely it is that Aristotle would think of the human good or function as an inclusive end. What would it be to have a set of activities as one’s function? Since the realization of one’s function expresses a single form or self, the members of an inclusive end would have to be connected closely enough for a unified conception of the self to be expressed through them. This would be possible if, for instance, the activities making up the set were species of the same genus. (Notice that the function argument is sometimes read as saying that the human function is all the various activities of reason.) But in that case the pursuit of the various species would be choiceworthy for the sake of the genus; there would in fact be a single good, not several, for the sake of which everything is to be done.
30 See note 42, below, on convergence in NE I.7.
31 See Joly (1956) for a fascinating account of how the lives were a trope in philosophical and poetic discussion of happiness.
32 If the refined men of action choose honor for the sake of thinking themselves to be good, why does Aristotle say they think being good is better than honor? The fact that they seek approval from those who are practically wise and who know them (1095b28–29) shows that they care about the truth of the matter and not simply about having high self-esteem. We need not argue that justified self-esteem is instrumental to being virtuous in order to claim that virtue is functioning as a telos here. The goal of actually being virtuous is functioning as a normative standard when it determines the conditions under which the refined person thinks it appropriate to approve of himself.
33 Aristotle does not go so far as to say that virtuous activity is the end of the political life until he returns to the topic of the prominent happy lives in NE X.6–8. It is interesting to speculate about why Aristotle is so coy in NE I.5 about the true end of the political life. Aristotle is rehearsing the common opinions in this chapter, so perhaps he thinks of the emphasis on the activity of virtue as his own innovation. See Kraut (1989, 17–18) for a discussion of Aristotle’s inconclusive attempts to locate the proper end of the political life in NE I.5.
34 Notice that Aristotle is here dismissing an account of happiness suggested at NE I.4 1095a22–23.
35 See, e.g., Ackrill 1974; Irwin 1991.
36 The practicable good versus the good simpliciter: The contrast Aristotle has in mind is with the good of the whole ordered world (e.g., Plato’s Form of the Good, criticized in NE I.6), which presumably no human being can rationally pursue.
37 The list of crafts and their ends in the NE I.7passage is virtually identical to the one in NE I.1, except that whereas I.1 says that wealth is the end of household management (oikonomikê : 1094a9), I.7says that houses are the end of housebuilding (oikodomikê: 1097a20).
38 Kraut (1989, 225–227) also argues that NE I.5 fails to support an inclusivist interpretation of eudaimonia. However, he thinks that this passage, and in particular 1097a22–24, shows that Aristotle still keeps the inclusivist possibility open.
39 EE II.1 passim, and in particular 1220a3–4.We should notice that in this chapter Aristotle compares the parts of the human good to the parts of physical health. This suggests that the parts of human goodness form an organic whole. If so, we can ask what is the principle of their unity. The last chapter of the EE suggests that this principle is the contemplation of god. Thus, even if Aristotle’s conception of the good in the EE is inclusive of several goods, I doubt it provides a precedent for the sort of inclusivism commentators claim to find in the NE, according to which eudaimonia is a set of independent goods, structured (perhaps) by moral virtue.
40 At NE I.6 1097a1 Aristotle speaks of the “attainable and practicable goods (ta ktêta kai praktatôn agathôn).” See also NE VI.71141b9–14, where Aristotle says all deliberation aims at a practicable good (prakton agathon) and that the practically wise person aims at the best of these. This best of the practicable goods is, of course, happiness.
41 See DC II.12 292b3–10, where Aristotle seems to take it for granted that the human good is a monistic end: “For man is capable of attaining many goods and so is capable of doing many things, and for the sake of still other things. . . . But the other animals have less [variety of action than man]; . . . For either there is but one attainable good, as indeed man has, or, if several, each contributes directly to the best.”
42 As I argued in section 3c, Aristotle always assumes that the good is a convergent end. As several people have rightly pointed out, all that is necessary for a good to be most final is that it be haplôs teleion (final without qualification). And as Aristotle describes it, a haplôs teleion end is merely one that is not subordinated to any other end; he does not say it must be at the top of a hierarchy of ends, much less that it must be at the top of all chains of ends. Thus, there might be several most final ends of an individual life, each one most final relative to its own hierarchy of ends. (Kraut 1989, 204–205; although ultimately he argues that Aristotle’s most final end is meant to be convergent.) Now, it is certainly true that Aristotle does not provide a definition of the most final end that guarantees there will be only one, convergent end. Be that as it may, it is clear from the way Aristotle argues that eudaimonia is the most final end that this is what he has in mind. First of all, he says that eudaimonia is especially a most final end not simply because it meets the haplôs teleion requirement but also because pleasure, intellectual wisdom, and the moral virtues are choiceworthy for its sake. That is, its claim to be the good is determined not only by the fact that it is not subordinated to some other good but also by the fact that other important final goods are subordinated to it. Apparently, then, Aristotle assumes that the most final end tops a chain of ends. Second, the goods that Aristotle claims to be subordinated to happiness even while loved for their own sakes—pleasure, honor, virtue, and intellectual wisdom—are the very goods mentioned in NE I.5. He does not argue there or anywhere that these are the only goods loved for their own sakes. But they are at least the most important ones. And in NE I.7he claims that they are all choiceworthy for the sake of eudai-monia. This suggests that for Aristotle, happiness is meant to be an end upon which all chains of ends—or at least the most important ones—converge. Finally, there seems to be only one most final end (1097a29–30). Since Aristotle has ruled out the possibility that there are chains of ends that continue ad infinitum, that means all chains of ends must come to a stop in the one, most final good.
Although Aristotle does not justify his assumption that the most final end is unique, this is not as grave an omission as it might at first seem. It would be a problem if Aristotle thought that, under the aegis of the finality criterion, he could prove a good was uniquely eudaimonia simply by showing that it is not choiceworthy for the sake of something else—that it is merely haplôs teleion. For, as I have said, there might conceivably be several goods fitting this description, given all Aristotle has shown us. Happily, however Aristotle does not use the finality criterion in this way. In NE X.7he tries to argue not only that contemplation is not choiceworthy for the sake of anything further but also that it is the only candidate good in this position, since pleasure and moral virtue are choiceworthy for its sake.
43 The translation of this word is highly disputed. Those who espouse an inclusivist interpretation of NE I often translate it as ‘complete’, while those who favor a monistic-end reading often prefer ‘perfect’ or ‘final’. These are all legitimate translations of the Greek. I prefer to translate teleia here as ‘final’. This is meant to leave open which of the two interpretations of eudaimonia is appropriate. It also calls our attention to the fact that Aristotle is here talking about the endiness of the good (finis = end). It is impossible, however, to translate teleion as ‘final’ in all contexts in the NE. In particular, we must translate it as ‘complete’ in the phrase en biôi teleiôi (1098a18). Still, this should not encourage us to translate teleion as ‘complete’ elsewhere. I believe that the emphasis in this line (1098a18) is on having happiness through to the conclusion (or telos) of a natural length of time, not, primarily, on having all the different parts of a whole. (So we might translate en biôi teleôi as ‘in a finished life’.)
44 Contra Broadie (1991, 30–34), Aristotle is not, strictly speaking, proving in this passage that happiness is the good.We can take that for granted because (1) it is the goal of the architectonic craft, politics, and (2) everyone agrees on calling the highest good happiness (NE I.4 1095a14–22). What Aristotle is doing in this passage is demonstrating that the finality criterion can serve to distinguish the good and happiness from other practicable goods.
45 Since Aristotle says we would choose these middle-level ends even if we did not think they resulted in happiness, I do not agree with Weller (2001) that pursuing an end as choiceworthy for its own sake is just to pursue it as a way of realizing eudaimonia.
46 Richardson (1992, 327–351) is perhaps the clearest exponent of this position. See also Kraut 1989, passim; and Reeve 1995, 114–117.
47 Republic translations are from Grube and Reeve 1992.
48 Korsgaard (1996a, 230) connects this attractive reading with Glaucon’s challenge, although she notices that the fit is not good, and takes this fact as demanding a different interpretation of Aristotle’s classification.
49 See Symp. 205a1–3, where Socrates suggests that happiness is the most desirable practicable good. Notice here that happiness seems to be the best as an end (or at least it is the end in a line of justification for action).
50 Both Plato and Aristotle classify goods in more than one way. For instance, both distinguish goods of the soul from goods of the body; and Aristotle further distinguishes goods of the body from external goods (NE 1098b12–16). These classifications are not meant to be the only way to divide the pie.
51 Irwin 1999, 165.
52 There is a further oddity about thinking of Glaucon’s kinds of goods as tracking a hierarchy of ends. His example of the kind choiceworthy for its own sake is “harmless pleasures.” But Plato, unlike Aristotle, thinks that pleasures are processes. That is to say, he believes that pleasures have ends beyond themselves (Gorg. 492e7–493d3; Phil. 54c6-d2, 32a6-b4; Rep. IX 583e9–10; Frede 1993, xlii–xliii). If Plato meant Glaucon to be distinguishing levels in a teleological hierarchy, he ought not have provided as the sole example of a good choiceworthy forits own sake a good that, in Plato’s philosophy, is notorious for not having its end in itself. Of course, Glaucon is not Plato’s mouthpiece, so this point is not decisive in my favor.
53 Richardson (1992, 335), who adopts this interpretation of middle-level ends, is quite clear about this: “something is an end of action, for Aristotle, just in case it is pursued.”
54 Indeed, as Sarah Broadie has pointed out to me, if it is part of an animal’s end to have desires at all, then an animal cannot achieve its end by first desiring it.
55 I take it that the point of claiming that relaxation is for the sake of work in X.6 is to prove that hedonists have been choosing the pleasures of relaxation for the wrong reason.
56 Indeed, the phainomena collected by Aristotle suggest that people do not organize their projects for the sake of a single goal. See NE 1095a20–25, discussed in section 3c, above.
57 Nussbaum 1978, 338 not. ad 700b28.
58 Broadie 1991, 8–12.
59 Although I am not certain, this seems to be the way Kraut (1989, 234) interprets this passage in NE I.5.
60 With the exception, of course, of potentialities, which are conceptually inseparable from their actualization.
61 Cooper (1975, 168–178) and Whiting (1986) have argued that Aristotle’s position in NE X.7–8 is that human beings do have a dual nature. My interpretation of these chapters will show that there is no need to read Aristotle as making this metaphysically suspect assumption. See chapter 8, below.
62 Broadie (1991, 31–32) adopts this sort of strategy. This kind of solution seems to be suggested at EE VIII.3 1249b16–19 (where Aristotle says that contemplation of god is the limit [horos] of the right pursuit of other goods). I would not be certain of this, however, since (1) contemplation of god is not equivalent to happiness in the EE, (2) Aristotle does not stress the sense in which happiness is an end in the EE, and (3) the EE does not contain the discussion of the hierarchy of ends/goods so central to the NE. Kullman (1985, 172–173) thinks there is evidence at Pol. I.8 1256b15–22 for an end that limits only the extent of pursuit of a lower good, but this example of a limiting end is not a telos or “that for the sake of which,” in the sense of an end we have been describing, but is rather a hou heneka tini, i.e., a beneficiary. See chapter 4, section 2, below.
63 We do not have to think of the highest good as a convergent, most final end. My point is just that in the NE Aristotle does, and that this interpretation of middle ends, according to which eudaimonia functions only as a side constraint on their pursuit, does not do justice to this fact. of the other goods ultimately aims at realizing. This sort of weightiness is more profound than being a side constraint.
64 Although Ackrill ([1974] 1980) believes that he finds the problem cropping up already in NE I.1 (Gauthier and Jolif [1970, 6–7] make a similar point about this distinction in NE I.1, although unlike Ackrill they do not mention the problem again at NE I.7), I believe it emerges only in NE I.7. Therefore he explains the problem differently and, I believe, somewhat misleadingly. (See notes 17and 20, above, for an explanation of why I do not think the problem Ackrill finds in NE I.1 is genuine. See also Richardson 1992.) I will make Ackrill’s case only for NE I.7here.
65 Weller (2001, 96–98) makes an interesting point that Ackrill’s examples of an inclusive end—putting as a part of golf and golf as a part of a good vacation—suggest different interpretations of the relationship between part and whole. Putting is intrinsically valuable precisely because it is a part of a whole that is independently valuable. The value of golf makes its constituents valuable. Conversely, a vacation is good because golf and the other parts of the vacation are intrinsically valuable. As Weller rightly says, Ackrill’s considered analysis of the relationship between eudaimonia and its parts is analogous to the relationship between a good vacation and golf. Weller favors the golf/putting model. In other words, he thinks eudaimonia has an independent value that alone makes its constituents choiceworthy for their own sakes. I do not think this can be correct, however, since the model implies that middle-level ends have no intrinsic value except via their connection to happiness. I take Aristotle to deny this at 1097b3–4: We would choose the middle-level ends even if nothing followed from them. (See note 66 on similar issues.)
66 Although Korsgaard (1996a, 231–32) intends her interpretation of middle-level ends to differ from the inclusivist interpretation, it has a similar strategy. Middle-level ends or, in her terminology, conditional ends are “valued for their own sake, given that we are human beings living in human conditions. . . . The unconditional end plays a different role: it is what makes it worth it to be a human being and to live in human conditions. . . . It will be a mark of a conditional end that it is also a means. But this “also” is not merely conjunctive; rather, its being a “means” or constituent of a worthwhile life will be what makes it possible to choose it as an end.” So whereas Ackrill proposes that middle-level ends are choiceworthy for the sake of eudaimonia because they are choiceworthy for their own sakes, Korsgaard proposes that middle-level ends are choiceworthy for their own sakes because they are choiceworthy for the sake of eudaimonia. Unfortunately, Korsgaard does not explain how it is that conditional ends are genuinely final and not mere instruments. According to her interpretation of Aristotle, the constituents of a human life are valuable only on the condition of contemplation. But since Korsgaard analyzes all conditionality as instrumental (if Y is a condition of X’s value, then X is, broadly speaking, an instrument to the production of Y), that means human life is valuable because it is instrumentally valuable. But then where is the intrinsic value of the constituents of a good human life? (No doubt, Korsgaard would think I am illegitimately conflating finality and intrinsic value [1996b]. But if I am, then so is Aristotle. A successful interpretation of the NE cannot act as if he kept these concepts distinct.)
There is a deeper problem with Korsgaard’s interpretation worth mentioning here. She interprets Aristotle as saying that human life, with all its needs, is only conditionally valuable. The philosophical search for happiness is the search for this condition. But it’s not clear to me that Aristotle would agree. The fulfillment of needs—for sex, food, friends—is the expression of our animal and political nature; that is to say, it is the fulfillment of our human nature. As such, these fulfillments are ends, for Aristotle, standing in no need of further justification. They are final even if they are not most final. Of course, Aristotle does think that these soul activities are choiceworthy for the sake of something further. But the concept of these activities as ends does not require that he think so, and that’s part of the mystery of middle-level ends.
67 Cooper [1975] 1986, 19–22. Kraut (1989, 212 n. 10) is quite right that this is not possible on a simple or weak inclusivism, i.e., one that says that Y is an unorganized set of goods. Kraut emphasizes that there is no causal connection between part and whole in this model. Although I do not think the “for the sake of” relation need always be causal, I do agree that it must always be normative. An unorganized set of goods does not play a normative role in the pursuit of its constituents. However, Aristotle does seem to maintain in his biological works that the part-whole relation can be teleological. E.g., the liver is for the sake of the life of the body by being a part that plays a certain role in the functions of the body as a whole. Thus, on a strong form of inclusivism, X could be for the sake of Y by partially constituting Y. Kraut seems to recognize the theoretical possibility of strong inclusivism, but he does not believe (for some reason) that this sort of inclusivism is possible in the ethical case. (See also Kullmann [1985, 169–175] for his discussion of how organs are for the sake of the whole.)
68 See section 3e, above. Of course, Ackrill has a response. Aristotle says that “if there are several virtues, [the human good] will be activity in accordance with the best and most teleion ” [teleiotatên] (1098a17–18). Ackrill says that since, for reasons I have just outlined, the only way to interpret teleiotaton earlier in I.7is as the set of all teleion goods, the teleiotatê aretê must be the set of all the virtues. Ackrill’s reading is bolstered to some extent by the fact that this is precisely the way Aristotle uses teleiotatê aretê in EE 1219a35–39. However, Cooper ([1987] 1999, 222–224) has argued persuasively that there is no reason to read teleiotatê inclusively in the context of 1098a17. (See Purinton [1998, 287–291] for an argument that Cooper’s translation of teleiotatê aretê cannot be supported by appeal to teleiotaton in lines 1097a25– b6. Of course, if Purinton is correct, this does not help Ackrill’s case, as he admits.) Cooper’s argument is strengthened, I think, by looking at Aristotle’s attempts to define the different kinds of soul in DA II. There, Aristotle notices that animals do all kinds of things—perceive, desire, move, etc.—but seems to want to define the function of the animal soul as, strictly speaking, the capacity only to do one of these things (DA II.3).
69 Of course, there have been inclusivist interpretations of NE X as well. I will argue against some of them in chapter 8.
70 Barney (1999) argues that inclusivism is anachronistic insofar as it assumes that, according to Plato and Aristotle, there could be, or we could (philosophically) identify, goods choiceworthy for their own sakes absent the highest good.
71 Cf. Aquinas, I.x.119.
72 Cooper [1987] 1999, 221–224. Ackrill ([1974] 1980, 28) also believes that teleiotaton must mean the same thing in the finality criterion and in the function argument, although of course he interprets it to mean “most inclusive.” My interpretation is closer to Kraut’s (1989, 178–179).
73 I believe the clause at the end of the function argument—“if there are many virtues, in accordance with the best and most final”—is a legitimate conclusion of the preceding argument (contra Roche 1988a, 182–184). If it is part of the implicit understanding of the function of something that excellence in that activity is its ultimate end, then where there are several functions, we should expect that the most final among them will constitute the function and most final end. (See Lawrence [1993, 26–28] for a different view.) I will not insist on this point here, however. What is important, as I argue below, is that the meaning of most final virtue requires that if there is a most final virtue, its activity will be the teleological focus of other, less final virtues. Thus, the activity in accordance with the most final virtue is not merely a particularly important aspect of the human good (Cooper [1987] 1999, Broadie 1991); as the highest end, it is the human good.
74 Purinton (1998, 265) suggests that Aristotle is referring to the subordination of our nutritive and perceptive functions to our rational function. (See also Roche 1988a, 187.) I think this is an entirely plausible interpretation of the text. But if it is plausible, then in my reading the door is open for someone to show that if contemplation is more final than morally virtuous activity, then it is the highest human good.
75 Or rather it is not explicitly tied to conclusions about happiness until NE X. As Lawrence (2001, 447) correctly points out, Aristotle argues in NE VI that theoretical virtue is more final than practical virtue. See also Lawrence (2001, 447–451) for an argument that we ought to read the conclusion of the function argument formally, i.e., as referring to the best and most final virtue, whatever it is.