2. Austin's critique

A detailed scrutiny of Prichard's argument is found in J.L. Austin's posthumously published paper.Footnote ⁴ A uniform consensus among the later commentators is that Austin had decisively refuted Prichard's view.Footnote ⁵ Austin makes a number of claims, some of them involving a painstaking textual analysis both of NE and Prichard's article. But the central objection relates to the meaning of ‘eudaimonia’. In Austin's view, Prichard mistakenly identifies the lexical meaning of the Greek term ‘eudaimonia’ with ‘feeling pleased’, or ‘feeling happy’. This is, according to Austin, the main implicit premiss of Prichard's entire argument. Given this premiss, Prichard is able to maintain that Aristotle's discussion is supposed to yield the factors producing a certain psychological state (viz. eudaimonia). Aristotle could not be interested in discovering the ingredients of eudaimonia itself, since its nature should have been clear at the outset.

This rings true, as we find the following passage:

And similarly another fragment:

Austin remarks, first, that this latter fragment is strange in that Aristotle is made to interpret eudaimonia in the English metalanguage.Footnote ⁶ More important of course is the fact that ‘happiness’ is apparently taken by Prichard to refer to pleasure, or feeling happy, or feeling pleased. Austin cites three additional passages in Prichard in which the locutions of ‘pleasure’ or ‘state of happiness’ – in short, of psychological feelings – are apparently substituted for ‘happiness’ tout court.Footnote ⁷

Austin goes on to claim that the lexical meaning of ‘eudaimonia’ is given by to eu zēn kai to eu prattein (‘living and acting well’), that Aristotle declines to define agathon, but that agathon is certainly not lexically equivalent to telos. These claims, interesting and important as they are, do not carry the burden of Austin's argument. They are auxiliary propositions intended to buttress the main contention which lies in rejecting the eudaimonia/happiness equivalence.

Now, suppose for the moment that Austin's charge is sound. Then, I think, there are some uncomfortable questions one will have to face. In the first place, how plausible would it be to believe that Prichard had such a mistaken understanding of the lexical meaning of eudaimonia? Surely his presumed linguistic incompetence cannot be attested by the mention of the English word ‘happiness’, since this is the translation adopted, e.g. by Ross and Rackham, neither of whom could possibly be suspected of such a gross mistake.

Secondly, if Prichard indeed thought eudaimonia to mean something like ‘feeling pleased’, then his whole argument is no more than a tedious superfluity. To prove Aristotle's psychological hedonism, he could fasten on the idea from Book I (putting together 1095a19 and 1094a22-23) that all actions are done for the sake of eudaimonia, replace ‘eudaimonia’ with ‘feeling pleased’, and draw the desired conclusion. To anyone who has this lexical meaning of eudaimonia, the proof of psychological hedonism would be entirely straightforward.

Thirdly, it would be a major challenge for such a commentator simply to make sense of numerous passages such as ‘the question is asked, whether happiness is to be acquired by learning or by habituation’ (1099b10), ‘must no one at all, then, be called happy while he lives’ (1100a10-a11), ‘happiness, then, is the best, noblest, and most pleasant thing in the world’ (1099a24), ‘any chance person – even a slave – can enjoy the bodily pleasures no less than the best man; but no one assigns to a slave a share in happiness’ (1177a7-9). In all of them Aristotle addresses himself to the audience which is supposed to share the same linguistic competence, but which may, or may not yet, be converted to his own theory of eudaimonia. The bulk of Prichard's argument, had Austin's objection been sound, had to be occupied with explaining away these passages, rather than with the proof of psychological hedonism. Granted, Prichard believes Aristotle to have misrepresented the point of his philosophical enquiry, so that this confusion may provide an explanation of a passage like 1095a21 where Aristotle asks ‘what happiness is’. But Prichard never makes an undeniably preposterous accusation that Aristotle misrepresented his, or his audience's, linguistic competence.

It has to be noted that Austin recognises (e.g. in his quotation from Wordsworth) the monumental absurdity of the view he attributed to Prichard. But oddly he never pauses for a moment to reflect that the very fact of such an absurdity should put his own critique in doubt.

3. The logical structure of Prichard's argument

Unless, therefore, we are content with denying Prichard a very basic philosophical ability, Austin's refutation is itself anything but immediately plausible. It is, I think, worth taking another look at Prichard's argument. Even if Prichard's conclusion were absurd, it is necessary to see just what reasons led him to maintain these absurdities. In fact I am going to argue that there is much more to Prichard's argument than the happiness = pleasure equation, and that, while his argument is fallacious, the error he committed is not at all self-evident.

According to our reconstruction in §1, Prichard's argument should (roughly) run as follows:

1. 1. Some values, Aristotle believes, are pursued for their own sake (di' auto) (104–105).

2. 2. However, Aristotle's other claim is that they are also pursued ‘as good’ (hōs agathon) (105, 107).

3. 3. Hence, according to Aristotle, the ultimate goal must be the ‘good’ (agathon), rather than the values such as honour or practical wisdom.

4. 4. The nature of agathon is not clear (passim).

5. 5. Agathon in NE amounts to an egocentric ‘goodness-for-me’ (109).

6. 6. But ‘goodness-for-me’ should be identified with ‘my gratification’ (109).

7. 7. There is evidence that Aristotle endorsed the claim #6:

   1. (a) Eudaimonia is supposed to be our ultimate goal (111).

   2. (b) But eudaimonia amounts to the agent's gratification (112).

8. 8. Therefore, Aristotle is a psychological hedonist. (113).

Contrary to Austin's objection, the identification of eudaimonia with gratification does not come as a premiss of Prichard's main argument. It rather is supposed to offer direct textual support for the claim that Aristotle in fact endorsed psychological hedonism in NE. As our reconstruction makes clear, the main argument has a different form – namely, that in the light of some other views he indisputably held Aristotle is left virtually no choice but to be a psychological hedonist.

Prichard's argument naturally falls into three parts. In ##1–4 we are told that the real ultimate goal in Aristotle is the good, that perplexing agathon. The point of ##5–6 is to present a perfectly general dilemma – either agathon is ‘goodness’, or else it is ‘personal gratification’ – and to maintain that Aristotle cannot embrace its first horn. And the point of ##7a–7b is to provide textual evidence for the idea that Aristotle indeed embraced the second horn of the dilemma.

If this is how Prichard's reasoning proceeds, his alleged linguistic misconception of ‘eudaimonia’ plays a rather minor role in it. The most it enables us conclude would be that Aristotle actually endorsed a version of hedonism. There are though two other claims more ambitious in scope and more important for the evaluation of Prichard's argument as a whole: that there is a dilemma of irreducible goodness and personal gratification, and that Aristotle ought to have endorsed a version of hedonism, perhaps malgré lui.

4. Good-for-me, nobility, sacrifice

Like Austin, I will have little to say about the claims ##1–4. It seems fairly plausible, judging from Book I of NE, that Aristotle's final goals have to be ‘good’, and that there must be a tight connection between purposeful activities and the ‘good’. At all events I think the most interesting and important statements, as far as the theory of hedonism and Aristotle's interpretation are concerned, are the statements ##5–7.

To begin with #5, a couple of preliminary comments. It is clear from Prichard's text that the dilemma he examines concerns only goodness-for-me and goodness-for-someone. A Moorean option of intrinsic and wholly impersonal goodness simpliciter is not on the cards. Also, Prichard does not claim that ‘good’, according to Aristotle, is to be interpreted in absolutely every context as ‘good-for-me’. His concern is solely with the goodness of our goals. So in general the formula:

(1)

is in no way concerned with just any X. It features in the argument so far as X is my action, goal, or any other element of my ethical practice. My desire has a certain object, and I desire this object as good (per ##1–4). That is, I desire it as good-for-me. This restriction to ethical practice should also explain why goodness simpliciter is ignored. The fact of ethical practice, of acts directed at different individuals, is taken for granted. Even if some element of my practice were good/bad simpliciter (and this goodness were to be recognised as a distinct category), there would necessarily have to be someone for whom it is good/bad. This leaves open the question whether, say, a painting could be good simpliciter without being good for anyone.

The choice, then, is between the claim (1) and a weaker claim:

Now ascribing to Aristotle the claim (1) has to sound illogical. A virtuous man certainly desires his virtuous activities. And it is plausible that some of these activities are desired by him precisely because they are good for him. A typical example would be temperance. But there are other virtues, such as generosity and courage, which in many instances are desired by the virtuous man for the sake of someone else – that is, because they are good-for-someone-else.

However, Prichard reasons, when we look closer into Aristotle's discussion of both courage and generosity, we will discover a thoroughly egoistic picture. The key textual support comes from the discussion of nobility (to kalon) in IX.8. Aristotle there seems to claim that a generous person in his acts of generosity assigns a greater good to himself, and that herein should lay the explanation of such acts. But a one-line comment does not do justice to this very complex passage. It deserves first to be quoted at length:

Aristotle here is on record stressing the priority of egocentric good. But no one should fail to observe how strange his view is. A man sacrifices his life to assign a greater good – not to his friends or his country, as one might believe, but to himself. To make sense of this, let us examine the passage in more detail. (i) That reason chooses what is good for itself is of course no proof of any egoistic nature of virtuous motivation. Apart from the fact that in the text Aristotle does not make reason choose what is good only for itself, it is also possible that reason chooses what is good for NN, but what is truly good for NN may simply coincide with some non-egocentric good. Further, (ii) Aristotle explicitly asserts that a virtuous man can work for the sake of other people alone, and even (iii) sacrifice his life for them. But the explanation offered in (iv) is cryptic. As its first part involving pleasure makes clear, Aristotle rejects a summative conception of value, on which no matter how great (finite) value an action X has, a sufficient number of actions – each having lesser values than X – will, in total, exceed the value of X. The way this sentence is constructed suggests that the claim about nobility is supposed to obtain by analogy from the claim about pleasure. However, both that latter claim and the analogy itself lack any obvious justification.

One tempting thought is that Aristotle is committed to the Millian idea of incommensurable values. A day of philosophical study is more enjoyable than a year of relaxing on the beach not because it carries more quantity of pleasure, but because the first activity has the higher intrinsic quality. Some textual support can be found in 1175b39-1176a3 where pleasures are differentiated on account of their source. Pleasures of sight will be superior in ‘purity’ to pleasures of touch, since the activity of sight is ‘purer’ than the activity of touch (both types of pleasure are genuine pleasures, unlike morally evil pleasures declared earlier fake pleasures altogether). However, be that as it may with pleasures, there is as yet no ground to extend the solution to the noble.

Taking a different route (though with little textual support), one could say that the passage reveals Aristotle's indifference to the idea of maximizing pleasure. This in turn would provide a reason for distinguishing his view from any recognizable form of hedonism. The problem with this move is, again, that, whatever the appeal it carries for the theory of pleasure, it makes not a lot of sense for the theory of nobility. Why should we abstain from the greater amount of nobility delivered in successive small steps and prefer the amount of nobility delivered in one go? Perhaps dying for your country is individually the most noble action. But why should we not prefer repeated acts of generosity – each less noble than sacrificing your life – made possible by saving our life in that situation of sacrifice?

A possible way out could be found were we to consider, if only for the sake of argument at this stage, a hedonist reading of the passage.Footnote ⁸ A hedonist presumably must value nobility to the extent it generates enjoyment. The analogy mentioned earlier would rest on the effective reduction of the noble to the enjoyable. If we had such a reduction, we could e.g. employ the idea of incommensurability. We could say that, since the act of sacrifice carries with it such a great amount of enjoyment, it is preferable to many minute noble acts. However, this move is equally baffling: how could anyone actually enjoy the sacrifice of his own life?

Our best hope, I think, is the discussion of courage in III.8. Aristotle there describes the courage of a citizen-soldier that most closely resembles the ‘true courage’ described in III.6–7:

It is somewhat obscure just why the courage of citizen-soldiers is supposed to be not exactly equivalent to ‘true courage’. But whatever the difference, small as it could be, Aristotle is fairly explicit in insisting on the virtuousness of the citizen-soldier's courage. It appears, then, that the explanation why citizen-soldiers choose to sacrifice their lives is in their desire to avoid various punishments for cowardice.

Given this result, (v) is interpreted easily. The prize the good man chooses for himself is the avoidance of dishonour. That is why he chooses the prize for himself. For the same reason Aristotle finds it natural to run heroism and charity together. Doubtless charity can carry with it tangible experienceable honours, whilst a hero may be denied them in death. But in (vi) they are put together, since a hedonist resolution has already been accepted in (iv). According to this reading, then, the justification cited in (vii) should not be taken at face value. The choice of to kalon is not ultimate. It is itself explained by the desire to avoid dishonour and shame.

A final comment: despite the appearance to the contrary (‘a twelvemonth of noble life preferred to many years of humdrum existence’), Aristotle can hardly intend to embellish self-sacrifice per se. A life which required sacrificing your life or health in a battle must be less good than an otherwise virtuous life which did not require any such action. People are not expected to chase an opportunity of giving their lives for the country, nor is the very fact of such opportunity a necessary condition of a genuinely happy life. On the other hand, people must use every opportunity to develop their virtuous dispositions, including courage. But the actual circumstance of sacrificing your life should be seen as a result of a misfortune caused by factors outside of your control. Given that circumstance, however, a virtuous agent will try to make the best of it (see also I.10). And the best outcome for the virtuous agent in that circumstance is the avoidance of shame.

The nobility discussion provides support for #5, as well as for Aristotle's identifying ‘good-for-me’ with ‘my gratification’ (claim #6). So far, however, we have seen no reason why anyone would accept any such identification in the first place. Prichard says:

A little later he talks of an ‘almost irresistible’ appeal of this turn of thought. The wording suggests that the good-for-me interpretation of agathon is perceived by Prichard as a honey trap. While he does not say why this interpretation is in fact false, he explains the ‘irresistible’ appeal of the honey trap. Suppose we have a list of goals which we believe we desire for their own sake (call them ‘u-goals’). The list may include theoretical knowledge, moral action, or some kind of professional accomplishment. We then enquire into the nature of our desire of those u-goals. And we notice, by way of our enquiry, that not only do we consistently enjoy each and every one of our u-goals, but also that it would be impossible to desire them unless they provided enjoyment. We are then apt to conclude (perhaps erroneously) that our desires had as their ‘real’ ultimate object not the items on that list of u-goals, but rather the pleasure we obtain by achieving them. This idea of the causal link between our u-goals and pleasure gets perverted when we succumb to another temptation and proclaim pleasure a property of u-goals. The analogy Prichard draws is with the secondary qualities such as taste or smell. Apples and garlic produce certain sensations of taste and smell in us. But many people tend to attribute, correctly or not, the properties of taste and smell to apples and garlic themselves.

A similar perversion – or at least what Prichard regards as perversion – occurred to Aristotle's account of u-goals. Aristotle catalogued certain u-goals, subsequently noticing that we derive pleasure from all of them. But rather than to insist on a causal relation between those goals and pleasure, he directly attributed a property of pleasantness to them. And his way of expressing this idea comes by way of the claim #2, i.e. by saying that all these goals are ‘good’, and that we pursue them ‘as good’. The locutions of ‘good’ are equivalent to ‘good-for-me’ (claim #5). Thus to infer Aristotle's psychological hedonism we have to establish his endorsement of the connection between ‘good-for-me’ and ‘my gratification’, i.e. his endorsement of the claim #6.

5. ‘Eudaimonia’ and eudaimonia

As I have already indicated, the best textual evidence one can get for the good-for-me = my gratification formula is by looking into Aristotle's treatment of those hard cases where egocentric goodness and egocentric pleasure apparently have to come apart. The possibility of self-sacrifice is precisely such a case. In examining this kind of situations we can show, I think, that for Aristotle the equation always holds. But Prichard elected to go in a totally different direction: he hoped to extract the same conclusion from Book I.

We now, therefore, have to attend to the final part of the argument dealing with eudaimonia. In the passage quoted back in §2 Prichard says that by ‘eudaimonia’ Aristotle meant happiness. One way to understand this statement is to say, with Austin, that ‘happiness’ in English should mean something like ‘feeling pleased’, and then to argue that the lexical meaning of ‘eudaimonia’ is in Prichard's view equivalent to ‘feeling pleased’. For the reasons mentioned in §2 this interpretation leads to obvious puzzles. It is strange to think that Prichard could be unaware of them. Even if his view cannot be sustained in the end, its evident fatuity should be accounted for.

Prichard's basic idea is plain enough: Aristotle did not intend to give us a novel lexical meaning of ‘eudaimonia’. Of course he must have assumed that ‘eudaimonia’ was a word understood by his audience. And to his audience it meant the same as the word ‘happiness’ means to an English speaker. A native speaker, Greek and English, must have some competence with the words ‘eudaimonia’ and ‘happiness’ respectively. Unlike a question ‘Who is glücklich?’, the questions ‘What is happiness?’ or ‘Who is a happy man?’ should be understood by that speaker by virtue of his linguistic competence.

The term ‘eudaimonia’, then, is native to the language of Aristotle's audience in the same way the term ‘happiness’ is native to the language of the English speakers. But if so, how could there be such serious disagreements about the extension of the term? We disagree about how to use the terms ‘happy life’ or ‘happy person’ in our language. Some say that happiness is physical pleasure, and that a person is happy when and only when he has lots of such pleasures. Others might analogously link happiness with political success or virtue. These are the kinds of disagreement registered by Aristotle in I.V.

Given the fact of a disagreement so radical, one could regard these speakers as users of different dialects. In one such dialect the word ‘happiness’ would mean pleasure, in another honour, and so on. But if we accept the idea of different dialects, it would be difficult to understand the purpose of Aristotle's enquiry. It would then most naturally be taken to aim at introducing a novel use for the term ‘eudaimonia’. Some people, on this view, mean by ‘eudaimonia’ pleasure, while others honour. And Aristotle simply asks us to mean by it a virtuous activity of the soul. That is, I think, the conclusion against which Prichard protests. Rightly so: if this is how we understand the debate, the matter at stake would be narrowly philological and lack any major philosophical import.

There must be sufficient common ground shared by the debate participants in order for that debate not to be terminological. Thus the question becomes: what are the assumptions about eudaimonia made by Aristotle in NE? Aristotle's audience is supposed to consist of the men of ‘good upbringing’. The facts ‘evident to us’ – that is, evident to Aristotle and the audience – are the ‘starting-point’ of the whole discussion (1095b4-b6). There is a set of ethical beliefs that is supposed to be shared by its members. And the fact of good upbringing is commonly thought to imply that, for instance, sceptical challenges posed by Callicles or Thrasymachus are not to be addressed: everyone understands that injustice is bad for you, or that a tyrant is not the happiest man of all. But while there are ethical beliefs agreed on at the outset, there is no theoretical account capable to justify those beliefs. The audience may know that an action or a life is good, but not know why. Good upbringing may equip you with patterns of good behaviour in a range of circumstances, but it is not expected to hand you a reason why this particular act is to be judged good. Aristotle's task is to provide the missing reason.Footnote ⁹

If this sort of picture is taken on board, then one might say that people in Aristotle's audience already lead broadly happy lives.Footnote ¹⁰ Their knowledge of happiness is purely practical, taking the form of particularised precepts tailored to specific circumstances. But as far as the nature of happiness is concerned, then, on the contrary, the job is to be done by Aristotle. The audience does not consist of philosophers, or in any event, of accomplished philosophers. As they are not credited with the theoretical knowledge of happiness, they are not thought to know the nature of happiness. Indeed, if we were to think otherwise, it is obscure what the point of the relevant parts of Books I and X is supposed to be. On any sensible reading they are supposed to yield exactly the concept of happiness.

No less sensible, however, is to assume, as Prichard did, that the audience had linguistic knowledge of the term ‘eudaimonia’. It was a word, as we said, in their native tongue, rather than a specialised newly invented technical term. Nor did Aristotle intend to invent a new meaning and attach it to the familiar string of shapes and noises. And if the question is, ‘What did the audience mean by the term “eudaimonia”?’, then, Prichard reasons, the answer is obvious: ‘By “eudaimonia” they meant whatever the term “happiness” means in English – that is, happiness.’ Since we have already admitted that the speakers in Aristotle's audience did not know the nature of happiness, we seem to have painted ourselves into a corner here. On one hand, the speakers do not know what happiness is, while on the other they know the meaning of the term ‘happiness’. I am willing to speculate that Prichard, in slipping eventually towards the happiness = pleasure formula, saw no way out of this dilemma and chose to embrace its second horn at the expense of the first. Thus he says that ‘we’ – i.e. Aristotle and his audience – know the nature of happiness, because ‘we all know the nature of that for which the word “happiness” stands’ (112–13).

Prichard relies, in other words, on the following principle:

(2)

And moreover, Prichard is also apparently committed to a stronger claim:

Because Prichard believes that Aristotle and his audience knew the meaning of ‘eudaimonia’, he is led to conclude that they also knew the nature of eudaimonia, and that since ‘eudaimonia’ means pleasure, the nature of eudaimonia is constituted by pleasure. Neither principle is unassailable. We only need to resist the weaker principle (2). I will argue that there is more than one meaning associated with ‘eudaimonia’. The principle holds for one kind of meaning, but it fails for another. The dilemma that haunted Prichard is spurious: we should be able to embrace its both horns.

6. Two concepts of happiness

We often acquire meanings of terms by being introduced to their extensions. Demonstrative identification is the most ubiquitous medium. ‘What do you mean by “a rabbit”?’, you ask, and instead of giving you a lecture in zoology, of which I am incapable, I point at a nearby rabbit. Understanding acquired is imperfect, but it allows you communicating well on a range of occasions where you are now able to distinguish rabbits from other items. Descriptive identification is another means. If you query me on the meaning of ‘silver’, it is too expensive to produce silver objects to enlighten you. Instead I provide you with a description: ‘That ring you saw on Mr NN's finger is made of silver.’ The purpose is the same: to recognise silver objects in the multitude of other objects. Both methods elucidate meanings by extension. I do not give you a definition of rabbithood or silverhood, much less a reductive definition. Your capacity for formulating what a rabbit is is virtually non-existent. Your narrowly linguistic understanding though is good enough to let you get by on many occasions. It would, I think, be wrong to declare that, after such learning by extension, you remain entirely ignorant of the meaning of the relevant terms.

The learning by extension provides a practical capacity of more or less efficient communication. To test the speaker's extensionalist understanding of the term for X one ought to expose him to the use of that term on different occasions and to see how adequately that speaker behaves on these occasions. We could say that the speaker in such a case is equipped with a recognitional concept of X. There is another method whereby the meaning of the term for X is given by naming some putative essential characteristics of X. Recognitional abilities may, or may not, improve as a result, but the aim at any rate is different. The aim is to link the given concept – of rabbit or silver – to other concepts and eventually to produce a theoretical concept of X. A zoological enquiry will reveal the links between rabbits and other species, while a chemical enquiry will link the properties of silver with those of other metals. This type of conceptual understanding thus corresponds to improved theoretical understanding, similarly characterised by unification and integration.Footnote ¹¹ It involves the possession of a ‘theory’ of X. Rather than smoothening communication, the possession of a theoretical concept ultimately is supposed to improve our understanding of the world. Another, and a less grandiose, consequence is this: to possess a theoretical concept of X is to be able to deal with the question about the ‘nature’ of X. To query the nature of X is to query the defining characteristics of X. The theoretical concept of X is supposed exactly those characteristics. And the way to achieve this is by establishing conceptual links to other phenomena.Footnote ¹²

This kind of talk is to an extent parallel to Carnap's well-known discussion of explication.Footnote ¹³ One could certainly say that a speaker, entirely ignorant of zoology and only acquainted with a few rabbits in his backyard and their pictures in Alice in Wonderland, has an ‘inexact’ classificatory concept of rabbithood. In the process of zoological enquiry the concept gets more ‘exact’. At every stage of his familiarity with the term ‘rabbit’ the speaker is in possession of a concept of rabbithood, less or more exact. With that I am in complete agreement. I would insist, however, that what is essential to the subsequent improvement of exactitude is not, as Carnap suggested, the mathematisation and formalisation of inexact concepts, but instead the network of theoretical links they acquire in relation to other concepts.

Back to happiness. In this instance, one can assume, the speakers in Aristotle's audience had no claim to a theory of happiness, nor did they have any theoretical concept of happiness. Likely introduced to the terms ‘eudaimonia’ or ‘eudaimon’ by the method of extension, they were in possession of a recognitional concept of happiness. Certain individuals or certain lives were picked out and stipulated as ‘happy’. Other individuals or other lives were designated as ‘unhappy’. Solon and Pericles could be recognised as happy, while Priam and Ajax as unhappy. This kind of learning allowed them a semblance of smooth communication in a fairly limited range of circumstances. Many people around them – ‘the mass of mankind’, as Aristotle says in I.V – could judge Sardanapallus or the Persian king happy. People of the Aristotle's audience presumably resisted this use of the predicate ‘is happy’. Such disagreements prompted both sides to search for at least an elementary theory of happiness, thus initiating a proper conceptual enquiry. But for the disagreement to become evident, the original agreement, in the smaller community of Aristotle's audience, should have been possible. And for it to be possible, the samples through which the speakers learned the extension of the term should have displayed similar features. In the case of rabbits and silver such relevant similar features are their perceptual properties. Those properties prompted the recognition of, say, different rabbit-like animals as ‘rabbits’. That could, and often does, happen when the speakers remain entirely ignorant of the biological properties of rabbits.

The question now is whether we can discover analogous properties of ‘happy men’, such as Solon and Pericles, which allowed their classification as happy men by Aristotle's audience before the enquiry of NE began. These properties cannot describe deep and complex states of the individuals in question: we assume that the learning took place among the people not susceptible to complex reflection. So those classificatory properties would have to be manifest at a superficial level. And one candidate here is the property of enjoyment. The speakers classify ‘happy’ individuals based on how much enjoyment and how little pain these individuals have. In talking about enjoyment we of course should mean manifestations of enjoyment. It is possible for an individual to conceal, on a given occasion, his real psychological state. In any event, enjoyment or observable signs thereof would serve the guide for the application of the predicate ‘is happy’. But in mastering of the use of that predicate the speaker also realises that the property of enjoyment is not equivalent to the property of happiness. The nature of happiness – the question what the property of happiness really is – is, at this pre-reflective stage, left undecided. Perhaps it is equivalent to the property of living well, as the Greek etymology suggests and Aristotle concurs; however, this is no substitute for a solution, since it simply rephrases the problem in terms of ‘wellness’ or ‘the good’. The situation, then, is analogous to the case of rabbithood. Easily observable properties in both cases are, so far as they go, reliable guides in using the term. They are instrumental in fixing the extension of the term. But along with them there are properties not immediately available to perception that will be at work in refining the concept. What those properties are should be the subject of a proper theoretical enquiry.

That the property of enjoyment should be intimately connected with the very meaning of ‘eudaimonia’ is indirectly confirmed by several familiar passages in Book I. A life containing sufferings and misfortunes ‘no one’ would call ‘happy’, unless he were ‘maintaining the thesis at all costs’ (1096a1). Again, no one would call Priam ‘happy’, since his end was tragic and painful (1100a9). In all such instances Aristotle evidently appeals to common intuitions held with regard to eudaimonia. And those intuitions unambiguously tell us, according to Aristotle, that a happy life is largely a pleasant life. It is well to notice that included among those people are not only the theorists of eudaimonia, but equally a man on the street, that is, someone not to be credited with any considered view of eudaimonia. The source of these intuitions, we conclude, should reside in the recognitional concept of eudaimonia shared by average Greek speakers.

In this sense, therefore, Prichard's contention that by ‘eudaimonia’ Aristotle meant something like ‘enjoyment’ is vindicated. But it warrants no smooth transition to interpreting Aristotle's view as hedonism. The reason by now is obvious: generally, having a recognitional concept of X does not entail having any particular theoretical concept of X. For a hedonist, the theoretical concept of happiness (‘eudaimonia’, ‘welfare’, ‘well-being’) has a strong phenomenalist flavour.Footnote ¹⁴ The extension of its predicate should depend entirely on the distribution of psychological states of the subjects. The way a subject enjoys a given state of affairs fixes the amount of happiness he possesses in that state of affairs. Left ignored are the sources of his enjoyment. Their nature matters not at all for the estimation of happiness. Now, someone endowed with the recognitional concept of happiness along the lines described in this section is at liberty to reject such phenomenalist concept of happiness. The perceptual marks of happiness – enjoyment and suffering – can be taken by him to be no more than marks. It is open to him to construe the concept of happiness in a realistic fashion. Happiness will depend not only on the kind of psychological attitudes of enjoyment and suffering the subject has, but also on the states of affairs that produce those attitudes. A parallel move is made in the case of natural classificatory concepts. Rabbits are initially classified by their appearance and superficial behaviour, and this is how, as we have conjectured, speakers learn the recognitional concept of ‘rabbithood’. But a further enquiry classifies rabbits by the kind of zoological properties hidden from a casual observer. Those properties are then posited as the causes of the rabbits' appearance and behaviour.

However the case may be with natural classificatory concepts, if we choose to adopt a form of realism about happiness, there are some tricky issues to address. On one hand, we say, enjoyment is nothing more than a manifestation of certain realistically construed states of the individual (call them ‘r-states’). But on the other hand, we also know that just about anything may cause a psychological attitude of enjoyment (or suffering). An individual A derives pleasure from doing X (or being X). An individual B abhors X. Suppose that X falls into the class of activities which increase happiness, other things being equal. Then, in engaging in X, do both A and B increase the amount of happiness they have? If, in our happiness judgements, r-states take priority over enjoyment states, as they apparently should, then both of them increase their happiness. But then, in our verdict on the individual B, there is a too big departure from the original recognitional meaning of the term ‘happiness’ linked, in our assumption, to the experience of pleasure. It now turns out that enjoyment influences not in the least the degree of happiness. And then we must have been misrepresented our enquiry. We did not explicate the recognitional concept of happiness, but rather created a purely technical, novel theoretical concept. But if so, in advancing our enquiry we had no right to appeal, as Aristotle did, and almost everyone else still does, to commonsensical intuitions about the use of the term ‘happiness’.

Aristotle's clever way out of this difficulty was to treat pleasure as a necessary attribute of the r-states themselves. What these states are was supposed to be determined by the theory of the virtuous activity of the soul. And any such activity was supposed to yield pleasure. One of course could still wonder why pleasure and r-states, belonging to entirely different categories, cooperate together in advancing happiness. Or one might wonder whether an individual with few r-states and plenty of pleasure – i.e. an evil individual satisfied with his situation – should also have a decent amount of happiness. Aristotle delivers the final blow in X.3: virtuous activities alone could produce pleasure legitimately so understood. Activities that were not virtuous could produce at most ‘bad’ pleasures – that is, attitudes not really pleasant at all. The significance of this idea is in officially making r-states necessary and sufficient conditions of happiness. Pleasure, far from being an independent necessary condition of happiness, is now fully incorporated into the theory of virtue.

Though the theory of bad pleasures is, I think, highly problematic, here is not the place to assess it.Footnote ¹⁵ Suppose we were to accept it. We will even then encounter cases where impeccably virtuous conduct, satisfying all the conditions imposed by the theory of virtue, fails to yield pleasure. One of the more dramatic such cases is self-sacrifice on the battlefield. And this much is clear: if one were wholeheartedly committed to a realistic virtue-based theory of happiness, disposing of any independent contribution enjoyment could make, then the absence of pleasure (and the presence of pain) would have been immaterial for the level of happiness. A wholehearted realist about happiness would gladly bite the bullet. But as the discussion in IX.8 makes plain, Aristotle approaches such cases as a major difficulty. He is at pains to show how even there the subject receives his share of enjoyment. To me this constitutes evidence to the effect that Aristotle does not commit himself to a purely realistic theory of happiness and still treats pleasure as an independent factor.

If Aristotle had any such implicit hedonist commitment, the diagnosis must extend beyond his own particular account of happiness. It should generally be the case that any theory of happiness (or of ‘well-being’) beginning with ordinary language intuitions about the nature of happiness cannot completely sever its link to hedonism. The rudimentary concept of happiness obtained by the method of extension – typically acts of demonstrative or descriptive identification – fixes the link with pleasure. In the course of explicating this concept, driven by our theoretical needs, we can link happiness to r-states. Those will typically lack any conceptual connection with pleasure. But when there are situations in which pleasure and r-states are no longer correlated, the theorist will face a choice of either following his theoretical convictions and dismissing the factor of pleasure altogether, or else preserving the factor of pleasure. The price to be paid in choosing the first alternative is the empirical inadequacy of the theory – that is, the failure to satisfy our ordinary intuitions. While for some theorists this would be a small price to pay, Aristotle is a known staunch defender of ordinary intuitions in ethics. The price to be paid in choosing the second alternative is the return to a version of hedonism, thus largely nullifying the value of the theoretical edifice built upon the doctrine of r-states. The conflict, therefore, originates in the basic epistemic procedure of ethical theory. Aristotle's struggles in NE are a useful illustration of its intensity.