One plausible view of Socrates is that he did not really have a philosophy, in the sense of a body of doctrine, so much as a method of philosophical enquiry. This volume contains four Platonic dialogues. Three of them are usually taken to be canonical ‘dialogues of search’—that is, dialogues in which Plato portrays Socrates using his method of enquiry—while the fourth, Meno, shows Socrates above all working through issues thrown up by his method of enquiry. And it is usually thought that the three dialogues of search (Charmides, Laches, and Lysis)1 belong to Plato’s earliest period of writing, when he was concerned mainly to portray his mentor at work, while Meno belongs on the borderline of his middle period, when he was beginning to reflect upon aspects of his Socratic inheritance. Thus, while the first part of Meno (up to 79e) looks very like a dialogue of search, tackling the question ‘What is excellence?’,2 the bulk of the dialogue raises fruitful questions which are designed to overcome difficulties raised by the search.
Socrates’ method of enquiry is also known as the elenchus. The word is a transliteration of a Greek word at whose heart is the idea of ‘challenge’ or of ‘testing’. In Plato’s Apology of Socrates—his largely fictitious version of the defence speech Socrates delivered before the Athenians at his trial in 399 BCE—Socrates explains that, in response to the famous Delphic oracle which declared that there was no one wiser than Socrates, he began to question people, to see if they really knew what they thought they knew. He challenged them, then, and invariably found that they had no more than superficial knowledge, or beliefs inherited from somewhere but not fully thought out, and not part of a coherent system of beliefs; they did not have anything which had the stability and certainty one would expect from knowledge. And so ‘elenchus’ in the sense of ‘challenge’ very often took on the aggressive sense of ‘refutation’, and the dialogues of search tend to end in aporia (‘having no resources’, ‘having no way to progress’, ‘being in an impasse’, ‘being stuck’), as Socrates traps his interlocutors into infuriating self-contradiction. An impasse is more precise than ‘confusion’, as the word is often translated; it refers to the state of mental frustration which results when you have followed a train of thought as far as it can go, and it has failed to take you where you wanted or expected, and you can see nowhere else to go. Socrates clearly believed that aporia was good for the soul (e.g. Charmides 166c; without this belief, an aporetic conversation would just seem futile), and at Gorgias 458a Plato has Socrates even say that it is better to be refuted than to refute others. Our basic and worst sin, he thought, is believing that we know something when we really do not, and aporia, unlike plain ignorance, is a state where we are compelled to be aware of our ignorance and will hopefully be motivated to do something about it (though, oddly, few of the characters in the aporetic dialogues evince much interest in continuing their education).
So we are shown characters in the dialogues who believe they can define a virtue (an aspect of excellence), but are reduced to aporia. As long as they can be reduced to aporia, this shows that they did not really know what they were talking about. This applies even to Nicias in Laches or Critias in Charmides, whose definitions are Socratic and possibly correct, but are only accidentally correct, because they have heard them from someone else, not thought them up, thought them through, and made them known. The possession of a belief is in itself more or less useless: it remains a mere slogan unless one can defend it.3 Equally, the imparting of information by a teacher is more or less useless without understanding. Hence Socrates did not teach, but explored the contexts of his interlocutors’ answers, even if that meant creating aporia in them. Later in his life, Plato looked back on Socratic enquiry and picked out as its chief features that, through cross-examination, it points up inconsistencies in a person’s beliefs, and thereby makes people angry with themselves and more tolerant of others, purges the soul of the conceit of knowledge, and leaves the soul believing that it knows only what it does actually know (Sophist 230a–d).
The dialogues of search need some introduction, because there are aspects of them that can seem puzzling. They all end inconclusively: what has been the point? Charmides and Lysis are particularly odd, in that sometimes Plato seems to fall into little more than silly word games. The arguments Plato put into Socrates’ mouth are sometimes bad, in the sense that they are formally fallacious, but these were still the early days of rational argumentation and the mistakes are often quite subtle. In any case, Plato was more concerned to show Socrates puzzling over critical issues, and to leave readers some room to work things out for themselves. The point of the arguments is to change people’s lives, to make us better people, in the sense that, especially where moral issues are concerned, thinking something through and identifying underlying assumptions are always preferable to the unthinking and uncritical acceptance of society’s injunctions. From this point of view, Plato was more interested in the process and in the conclusions than he was in providing formally valid arguments. Formal validity helped him only because it was persuasive, and he was concerned above all to persuade individuals—the kinds of people we see interacting with Socrates in the dialogues—which often gives the arguments an ad hominem feel.
The moral nature of these three dialogues is clear in the first place from their subject matter. Charmides investigates the nature of self-control, Laches does the same for courage, and Lysis explores the concept of friendship. Some readers might meet a moment’s puzzlement here: why was friendship considered worthy of philosophical investigation? The goal of all Plato’s dialogues, and his teaching in the school he set up (the Academy), was to get his audience to improve the quality of their lives—to live the good life, to fulfil themselves as human beings, to attain happiness, to live as godlike a life as is humanly possible. For Plato, these were just different ways of saying the same thing. Without friends, one’s life would not just be emptier, but would hardly be a human life at all. Interaction with other people is part of what it is to be a human being, living in the real world, and friendly interaction improves the quality of one’s life. So friendship has been investigated not just by Plato, but by a number of philosophers from all eras.4 Moreover, the ancient Greeks invariably took friendship to be based on reciprocity: you are my friend if you scratch my back, and you expect me to do the same for you. This made friendship far more volatile and uncertain than we normally take it to be nowadays—and so, given the assumption that friendship improves the quality of life, it was more urgent for Plato to explore it.
The aspects of excellence examined in the other two dialogues were standard members of any ancient Greek list of cardinal virtues. Courage has been regarded as a virtue at all periods of human history. Of course, it is particularly relevant in a society such as that of ancient Athens, which was more or less constantly at war until membership of various empires sidelined it, but (as Plato was quick to point out) courage is important even outside of any military context: it takes courage to stand up to injustice and aggression in all walks of life, to face threats, and to preserve one’s integrity in situations which would damage it.
Self-control, by contrast, is one of the quiet virtues, to do with co-operation rather than competition. The Greek word sōphrosynē means, originally, ‘being of sound mind’ (and so Plato implicitly defines it as mental health at Charmides 157a) and referred primarily to self-knowledge in the sense of knowing one’s proper place in society. It covered a range from self-reliance and moderate self-restraint in the face of one’s emotions and bodily appetites, through discretion, prudence, politeness, and good manners, to the kind of humility and self-effacement that patriarchal Greeks required of their womenfolk and the younger generation. In a political context, it had overtones of aristocratic conservatism; in a philosophical context (such as that of Heraclitus, Fragment 112), it could even mean the ability to see things as they are, without imposing one’s own views.5 All these shades of meaning play a part in Charmides, as Plato struggles to find some common core which underlies them, points out how difficult it is to understand the concept on traditional lines, and perhaps paves the way for a new understanding.
It is peculiarly appropriate to the moral nature of these three dialogues, and to the ad hominem nature of the argumentation, that Plato devotes so much care to characterization and scene-setting.6 In Laches, half the dialogue has passed before Socrates gets down to investigating courage; in Charmides and Lysis the characterization not just of Socrates, but of the teenagers who are Socrates’ interlocutors, is brilliantly and deftly handled, often with great charm and subtlety. The upper-class men and boys we find Socrates talking to in these dialogues are typical of his interlocutors; the other main category of interlocutor, not represented in this volume, consisted of professionals such as Sophists. These three dialogues of search are remarkable not just for the similarity of their structure and their portrait of Socrates at work, but for the artistry of their composition. Scene-setting and vivid interludes, brilliantly executed, occupy several pages in each dialogue, and it is clear that Plato cared about such things—though perhaps not so much as a way of communicating or supporting some philosophical point, but as a way of flexing his artistic muscles.7
Another common feature to these three dialogues is the focus on young boys (the homoerotic aspect has been briefly covered in the Explanatory Notes). Although this is by no means universal in the dialogues—Socrates converses with people of all ages—in this respect Plato was surely giving us a faithful portrait of the historical Socrates. For almost all of Socrates’ life, his native city was involved in a cold war, which often reached boiling point, with its great rival, Peloponnesian Sparta. The aristocratic youths who were Socrates’ interlocutors in these dialogues would form the next generation of power-possessing politicians. It must have seemed crucial to Socrates that they should not unthinkingly accept the moral prescriptions of earlier generations, but should have reflected upon underlying principles, which could then be translated into prescriptions appropriate for the new world into which post-war Athens would emerge.
Self-control forms the background to this dialogue, as well as its overt topic. Faced with a dialogue in which Socrates converses with Critias and Charmides, Plato’s readers would immediately have been reminded of the subsequent careers of these two aristocratic members of Plato’s own family (see the Index of Names). As prominent members of the brutal and bloody oligarchic regime which briefly ruled Athens in 404–403 BCE, they could hardly be said to have practised self-restraint,8 as normally understood, though paradoxically, as opponents of Athenian democracy, the politically conservative overtones of sōphrosynē would have appealed to them. Socrates is distanced from them, not just by the fact that he disagrees with what they say, but by his display of self-control at, especially, 155d–e: he does not allow his desire for Charmides’ body to distract him from the task at hand, improving Charmides’ soul.
Once they get down to discussing self-control, Charmides first proposes that it is a kind of unhurriedness (159b). Socrates’ refutation of this idea is typical of many arguments in the dialogues. He points to several counter-examples, enough to show that Charmides’ idea does not satisfy the criterion that a good definition should be universal: it should not be too narrow (it should not exclude things rightly held to fall under the concept in question) nor too broad (it should not include things rightly held not to fall under the concept in question). Socrates’ approach is to gain Charmides’ admission that self-control, whatever it may be, must be a good thing; the counter-examples, then, are simply cases (chosen from those familiar to Charmides’ own experience) where unhurriedness is not or not always a good thing.
Charmides next moves from external behaviour to the internal state that might prompt such behaviour, and proposes that self-control is modesty (160e), but Socrates disposes of this by means of a single counter-example, gleaned from the authority of Homer: modesty is not always good, and therefore, given the assumption that self-control is always good, the two cannot be the same. These two definitions, and the next, were well entrenched in the everyday meanings of sōphrosynē, but Socrates is not necessarily claiming that the common understanding of sōphrosynē was entirely wrong; the elenchus is pointedly personal, and so we can take Plato to be claiming only that Charmides has failed to defend the common understanding, or that these ideas do not encompass all there is to say about self-control.
At this point Critias enters the conversation. In all three of the dialogues of search translated here, the entry of a new interlocutor indicates a rise in the level of sophistication of the search. And so Charmides’ third definition (voiced by him, but attributed slyly to Critias at 161b–c and overtly at 162c–d) is that self-control is ‘doing what pertains to oneself’ (161b). There can be no doubt that Plato was attracted towards this definition: it became the definition of social justice in the later dialogue Republic (433a). Here too it is placed in a social context, and Socrates refutes it by pointing out that in any such context people are involved to some extent in other people’s business: teachers teach others, artisans make things for others, and so on. A society in which people did only what pertained to themselves, without any interaction with others, would hardly be a good society; but self-control must be something good, and so the definition is taken to be refuted.
Critias now enters the discussion in person (162c), prompted by the teasing of Charmides and Socrates. He proposes to reveal the idea underlying the definition of self-control as doing what pertains to oneself. This is important: it explains why Plato does not have Socrates subject his ideas to the same kind of blunt critique that Charmides’ first two definitions received. Plato was attracted to the idea that self-control is doing what pertains to oneself, but he wanted to see it refined until it was unassailable, and so could stand as the definition of self-control. So the conversation with Critias can be seen as consisting of successive attempts to refine the definition.
Socrates first tries to substitute ‘making’ for ‘doing’ in the definition, but Critias rightly resists this move (162e–163c). He does so, however, in a peculiarly snobbish fashion, by sneering at artisans for ‘making’ things and commending the ‘doing’ of things as admirable. This allows Socrates to force a refinement of the definition: it is not ‘doing what pertains to oneself’, but ‘the doing of good things’ that is self-control (163e). This is a pretty desperate first attempt at refinement, but rather than demolishing it or considering what the vague phrase ‘good things’ might mean, Socrates simply uses it as an excuse to introduce the topic of knowledge, which will dominate the rest of the dialogue. The introduction of knowledge is significant because, as we will see more fully in Laches, Socrates appears to think that every aspect of excellence is or involves knowledge.
Someone may do good, or act beneficially, without knowing that he is doing so; therefore, Socrates claims, Critias’ definition of self-control as doing good implies that self-controlled people do not know that they are self-controlled (164a–c). Socrates could have pulled this rabbit out of the hat whatever Critias had said, but Critias alters his definition of self–control to ‘knowledge of oneself’. Socrates tries to argue (as elsewhere in the early dialogues) that all branches of knowledge have a product, but Critias rightly resists this move (165e–166c). By the time this piece of Platonic self–criticism is over,9 ‘knowledge of oneself’ has become ‘knowledge of itself’—that is, knowledge of knowledge (166e), which is the definition with which the remainder of the dialogue is occupied.
Socrates tries to generate an impasse by pointing out that, if there is such a thing as knowledge of knowledge, it is an oddity: there is no such thing as sight of sight, or desire of desire. Critias could simply respond that it may be a unique case, but Plato only allows him to end up just as puzzled as Socrates (169c). All this is Plato’s way of getting us to think about knowledge of knowledge, to remember that it is a familiar human experience that we can not just know things, but be aware that we know things. On such occasions the human psyche is split, so to speak, into higher and lower levels, with the higher level overseeing the lower; in fact the Greek word sōphronistēs, cognate with self–control, meant ‘supervisor’. It is surely essential to self-control that a person has the ability to stand back from whatever emotion or desire is moving her, or is about to move her, in order to resist it. This must be what Plato is driving at.
Over the subsequent pages, ‘knowledge of knowledge’ is interpreted in several different ways. First, it is taken to be ‘knowing what one does and does not know’. Intuitively, this is a reasonable notion of sōphrosynē, which involved knowing one’s limitations. But Socrates gradually whittles away at the idea that knowledge of knowledge has any true content: he first reduces it to ‘knowing that one does and does not know’, and then to a kind of vague awareness that one knows something, or at best a kind of general supervisory knowledge, which ensures the smooth operation of all other branches of knowledge. The chief assumption governing the discussion is that all types of knowledge are the same, each having just one domain and one product. On this assumption, knowledge of knowledge is just that, knowledge of knowledge, and cannot involve knowledge of anything else: it is not knowledge of health or knowledge of knowledge-of-health.
Moreover, on this assumption, knowledge of knowledge (if it can do anything) can only guarantee the smooth operation of the various other branches of knowledge; it cannot guarantee that such branches of knowledge will lead to happiness, because that will be a different branch of knowledge, namely knowledge of good and bad (174b–175a). Knowledge of knowledge, therefore, may make one efficient, but it makes no real contribution towards human life or happiness. It cannot be the same as self-control, then, because self-control must enhance one’s life. Put another way, Plato seems to be pointing to the difficulty of understanding how knowledge of knowledge, understood as a purely cognitive state, can be equated with an aspect of moral excellence.
The final assumption controlling the argument, then, is that the possession of any excellence is bound to be beneficial,10 and more specifically to contribute towards our happiness. The precise relation between excellence and happiness is a thorny topic in Socratic studies, but at any rate Socrates saw a very close link between the two, such that either excellence is both necessary and sufficient for happiness, or is at least necessary.11 And the link is generated by the close connection between excellence and knowledge: in a famous argument (found in two minimally different versions, at Euthydemus 278e–282a and Meno 87d–89a), Plato has Socrates argue that it is always knowledge which guarantees success and therefore happiness. It is the fact that each aspect of excellence is some kind of knowledge that guarantees that it contributes towards human happiness. Charmides makes this more precise: it is knowledge of good and bad that is a necessary and sufficient condition for happiness. There is even the suggestion that people with this kind of knowledge should, in an ideal world, have political power—and so the irony of having Charmides and Critias as interlocutors comes full circle: Charmides was generally held to be self-controlled, and Critias used the virtue as a political slogan; they both came to hold political power in Athens, but they were not truly self-controlled, because they lacked knowledge of good and bad.
The dialogue ends in the usual aporia—the interlocutors fail to arrive at a definition of self-control which survives Socrates’ challenges —but several features of the discussion stand out as those which come closest to satisfying our intuitions about self-control. First, Plato makes it absolutely clear that he finds the notion of self-control as ‘doing what pertains to oneself’ highly promising: the entire latter half of the dialogue is devoted to trying to explicate this idea; it is just that neither Charmides nor Critias found a way to defend it. Second, the idea that self-control involves layers or levels of awareness and executive control within the human psyche must be right. Third, a little reflection on human happiness might have brought Socrates and Critias closer to the realization that self-control (or any of the virtues) involves not just an inner state (say, knowledge of knowledge), but external activity: if Critias had said that self-control was knowing what you do and do not know (i.e. knowing one’s limitations) and acting accordingly, Socrates might have found it hard to challenge him. Since in other dialogues Plato is aware that excellence involves both an inner state and external action, he may even be expecting his readers to pick up on the strange omission and supplement the text accordingly.
The interaction between the five protagonists that occupies the first half of the dialogue leads naturally towards an enquiry into courage. Lysimachus and Melesias want to hear the opinion of Laches and Nicias about the value for their sons of an education in combat. In the late 420s (the time when the dialogue is set: see the second note to 182a), Nicias and Laches were at the height of their power in Athens, but they were to lose their lives in campaigns that helped to seal Athens’ defeat in the Peloponnesian War, and Nicias, at any rate, gained a reputation for prevarication, if not outright cowardice. For Plato’s fourth-century readers, then, there was no little irony in the choice of protagonists for this dialogue. Anyway, Socrates argues that what Laches and Nicias would have to say would be useful only if they were experts in education. The function of education is always to improve the pupil’s soul, and therefore Laches and Nicias should demonstrate that they are experts in excellence. That is too grandiose a topic, however, and so they propose to focus on the ‘part’ of excellence which is presumably relevant to combat, and that is courage.
Laches first suggests that courage is remaining at one’s post in battle (190e), but this restricts courage to the battlefield alone, whereas there are plenty of other situations in which courage can be displayed. Socrates wants to know what is common to courage in all situations (191e). Laches suggests, more plausibly, that it is ‘mental persistence’ (192b), but Socrates argues, first, that unintelligent persistence may be bad, and, second, that sometimes unintelligent persistence may be more courageous than intelligent persistence. Since everyone assumes that courage is a good thing and stupidity is a bad thing, Laches’ definition fails. The first definition failed because it was too narrow (it failed to accommodate many cases of courage); the second failed because it was too broad (it failed to separate off unintelligent persistence). Laches may have failed, but he has allowed Socrates to introduce the idea that some kind of knowledge or intelligence is essential to courage, and it is the purpose of the rest of the dialogue to explore what kind of knowledge that might be.
So Nicias now enters the fray, with the explicitly Socratic idea that courage is a kind of knowledge (194c–d)—specifically, knowledge of what is and is not threatening in every situation requiring courage (194e–195a). As in Charmides, we move from traditional, Homeric ideas about the virtue in question to something more sophisticated, more theoretical, and more tinged with the learning current in Athens in the last quarter of the fifth century. After some preliminary sparring between Laches and Nicias, who have become rivals for Socrates’ approval, Socrates initiates a more objective examination of this idea (196d). A threat lies in the future, so Nicias’ definition implies that courage involves knowledge of the future. There is no branch or kind of knowledge, however, where knowing the future is different from knowing the past or present. Courage must therefore be knowledge of what is good and bad for oneself at any time, but that looks more like a definition of excellence as a whole, not just a part of it (199d–e; compare Charmides 174b–c). And so they have failed to define courage.
Laches is a more straightforward work than Charmides. We do not have to dig deep to uncover some positive lessons from the apparently negative course of the dialogue. Minor points of some importance include the distinction of morally neutral kinds of knowledge from those with moral relevance (195c–d) and the distinction between courage and fearlessness (196e–197d), both of which remain unchallenged. Most important, however, is the Socratic notion that goodness and knowledge always go together. It is not just that this idea informs the most serious part of the enquiry, but also that the conclusion it entails is endorsed by Plato’s Socrates in other dialogues. In the dialogue Protagoras, Plato has Socrates argue precisely that every aspect of excellence is the same as every other, because they are all knowledge of good and bad, or at least that all the aspects of excellence are mutually entailing (see Protagoras 332a–333b, 349e–350c).12 In other words, Nicias’ definition of courage as knowledge of what is good and bad for oneself can stand; the reader has only to understand that Nicias’ argument fails only if courage is taken to be merely an aspect of excellence. The main passage which has been taken to tell against this view of the purpose of the dialogue is 192b–193d, where Socrates argues that sometimes unintelligent or at least uninformed action is more courageous than intelligent action. But this does not really tell against the intellectualist definition of courage: it simply awaits the distinction between courage and fearlessness (196e–197d). What appears to be unintelligent courage is actually fearlessness or recklessness, not courage. And then we should note that at Meno 88b boldness plus knowledge seems to be a candidate for courage. Perhaps Plato means us to read the entire latter part of the dialogue as an attempt to find what kind of knowledge needs to be added to a disposition such as Laches’ ‘persistence’, so that we arrive at a definition of courage as ‘persistence based on knowledge of what is good and bad’.
Socratic ‘intellectualism’ or ‘rationalism’—the idea that excellence is knowledge—also manifests in certain paradoxical views attributed by Plato to him. Above all, Plato’s Socrates held (1) that no one deliberately does wrong, and in fact that it is better to have wrong done to one than to do it oneself; (2) that no one wants anything bad, only good, and that any pursuit of anything bad is therefore involuntary.13 The most paradoxical consequences of the paradoxes are that they appear to deny two things: (a) that a criminal or anyone does wrong deliberately; (b) that anyone can suffer from akrasia—weakness of will—such as knowing that I should not have that sixth glass of wine, but having it anyway. But deliberate criminality and weakness of the will are common occurrences. Nevertheless, Socrates did not think he was being paradoxical; he thought he was stating plain facts.
The denial of deliberate criminality is not a denial of the existence of crime as a social phenomenon; rather, it is an assertion that if the criminal knew what he was doing, he would not do wrong at all. All people aim at happiness or pleasure. The criminal’s mistake is that he supposes that his happiness lies in committing crimes. In fact, though, happiness is a function of the soul, not of the possession of material goods. And doing wrong harms the soul (even if it may benefit the body), so that the soul is not really being made happy by crime. Hence in Gorgias Plato goes so far as to argue that it is better to have wrong done to you than to do wrong yourself. Thus the criminal thinks he is acting in his own interest, but is not; he is acting from false belief, not from knowledge. If he acted from knowledge, he would not be a criminal.
The paradox of the denial of weakness of the will can be resolved in much the same way. If I really had knowledge of what was good and bad for me, I would not fail to act on it;14 therefore the fact that I do have that sixth glass of wine shows that I do not really have knowledge. Wanting something that is (in actual fact) bad for me is a clear sign of lack of knowledge: we want only things that are good for us, but sometimes we foolishly mistake bad things for the good things we want. The difference between a man of knowledge and a fool is not that a fool wants bad things—that is impossible, according to Plato’s Socrates—but that he mistakes what things are bad for him. In this context, in Protagoras, Plato has Socrates talk of a calculus of happiness. The person of knowledge weighs up pleasures and pains. He knows that the present pleasure of that extra glass of wine is going to be vastly outweighed by pain the next morning (or even sooner). And so he avoids the sixth glass (and probably stops at two, anyway).
This nest of ideas connects with the doctrine of the unity of the virtues. All virtue or excellence, as Laches suggests, is or involves knowledge of what is good and bad for me. Virtues are necessarily beneficial to the virtuous person, and it is the element of knowledge in them that makes them beneficial. So a person of knowledge, such as Socrates, stops after two glasses of wine, and is praised for his self-control; he refrains from criminal action, and is praised for his justice; he can be courageous, and so on. In all these situations, he is acting only from knowledge of what is good and bad for him, and from this point of view the common distinction between the virtues or aspects of excellence is meaningless. The knowledge involved in excellence, then, is self-knowledge, and this is the point of Socrates’ frequent recommendation to ‘look after one’s soul’ (e.g. Plato, Charmides 156e–157a, Laches 185d–e, Apology 29d–30a).15
Lysis is a discussion of friendship—or rather, it is a discussion of philia. ‘Friendship’ is often an adequate translation of philia, but the Greek term was used to describe not just a human relationship, but being fond of certain pursuits or things: ‘philosophy’, for instance, is literally ‘love of knowledge’. Where human relationships are concerned, philia is not just friendship, but drifts into ‘love’ (passionate or otherwise), ‘affection’, and even ‘loyalty’, since the Greeks were always pragmatic about friendship: it was defined as much by ties of mutual obligation as it was by any feeling or emotion, and often carried distinctly political connotations, in that your ‘friends’ were those who helped you in your political career and who expected to be repaid once you had gained a position of influence.16 Hence in Lysis Plato has Socrates link friendship with need or lack (215a–b, 221 d–e), and the background to several of the arguments is a broadly instrumentalist view of friendship. The semantic situation is enormously complicated by the fact that philos, the adjective cognate with philia, could bear an active sense, a passive sense, or both at once. That is, the same word could mean ‘friendly towards’ or ‘liking’, ‘a friend of’ or ‘dear to’ or ‘liked’, or ‘friend’ in the sense that two people are each other’s friends. Plato has fun and games with these ambiguities (see the Explanatory Notes for examples).
As with Charmides and Laches, the scene-setting of Lysis is peculiarly relevant to the subsequent discussion. There is a lot of banter at the start about how Hippothales is in love with Lysis. The term translated ‘love’ here is not philia, but erōs, which was the word for ‘passionate love’ or even ‘lust’, and especially for what an older man felt for an attractive teenage boy. We have at least two models of affection in the characters of the dialogue: Hippothales’ one-sided love for Lysis, and the friendship between Lysis and Menexenus. We also hear in the early stages of the dialogue of the philia felt by parents for their children (207d ff.), which is oddly said to depend on the children’s usefulness to their parents. This would be the case only if the relationship between parents and children consisted solely in the parents’ allowing or disallowing their offspring to do certain things.
Some interesting points emerge from this discussion about parental philia. A friend, it is implied, wants the best for his friend. He wants him to be happy, or to be free to do what he likes (short of harming himself in any way). But this freedom is granted only to those with knowledge, because it is knowledge that makes someone useful to others, and it is knowledge that guarantees that someone will not be harmed but benefited. It follows —though this is not explicitly brought out—that a true friend will educate his friend if he can, to give him the knowledge that brings happiness. There is no doubt that Plato means us to see Socrates’ attitude towards his young associates as a model of this educational friendship, and the implication that a man of knowledge will be everyone’s friend (210c–d) is a nice tribute to Socrates. Socrates embodies wisdom or knowledge; we have philia for what we lack; if we are aware of our lack of knowledge, we will want knowledge, which is to say that we will be philosophers. Socrates made his young associates philosophers.
There is one respect in which Lysis differs from Charmides and Laches (and other dialogues of search): Socrates himself comes up with all the ideas about friendship which are then examined. The dialogue, then, is not ‘maieutic’—Socrates does not act as a ‘midwife’ for the birth of others’ ideas.17 We do not need to exaggerate this difference, however: there is no reason to think that Socrates did not use the elenchus to examine his own views (see Plato, Apology 21b; both Critias in Charmides and Nicias in Laches are mouthpieces for Socratic or quasi-Socratic views; and Plato came later to define thinking as an internal dialogue18). In any case, the arguments progress in a standard dialectical fashion: even though it is Socrates who proposes definitions or ideas for discussion, he does not examine them until he has gained a measure of agreement from his interlocutor. It is true that the dialogue does not start with a question of the form ‘What is F?’, as do both Charmides (159a) and Laches (190d),19 but this is a mere formality: the investigation of friendship is in all important respects parallel to those of other dialogues which search for a definition. The topic is ‘what it is to be a friend’ (216c, 223b), even though the first question was not ‘What is a friend?’, but ‘Who is whose friend?’ (212b). Plato is looking for the cause of friendship, with his usual definitional aims: so that he can understand the concept and identify genuine cases.
By 216c, we want to say, ‘So far, so good.’ Plato has set out to examine friendship as a human relationship, explored the ambiguities inherent in the term (211d–213c, a flawed but very systematic passage), and come across difficulties in two contradictory but initially plausible notions: that people who are similar are friends (214a–215c), and that people who are opposites are friends (215c–16b). The argument has been guided above all by the instrumentalist and egoist principle that friendship is a kind of lack or need; friendship is based on reciprocity and mutual utility and perceived value. But the dialogue now takes a curious dog-leg. Instead of talking about friendship as a relationship between two humans, much of the subsequent discussion is concerned with the nature of the befriended object, whether that is a person or something inanimate. Philia changes from being a mutual relationship to attraction towards something. A lot of the puzzlement some readers feel is due to the fact that we think of friendship as a mutual relationship, whereas Plato spends much of the dialogue exploring one-sided attraction. This is because he sees human friendship as a species of desire or attraction, which is the more general concept.
Given that the previous discussion has already excluded the possibilities that good is attracted towards good, or bad towards bad, or good towards bad, or like towards like, we are left with the possibility ‘that what is neither good nor bad may be the friend of [i.e. attracted to] what is good’ (216e–217a). These are the terms of the ensuing discussion. Plato first argues (217a–218c) that it is the presence of something bad that attracts what is neither good nor bad towards the good. That is, for instance, illness (bad) causes a body (in itself, neither good nor bad) to be attracted to health (good). The thing that is neither good nor bad cannot be totally or essentially corrupted by the presence of the bad thing, because then it would be bad, not neither good nor bad, and bad things cannot be friends of good things; but something bad must be non-essentially or contingently present to it. This is the first occurrence in philosophical literature of the critical distinction between essential and non-essential properties.
Again, the argument rests on the egoist assumption that a person is attracted towards something because of the good it can do him. This attitude towards friendship has been harshly criticized —should we not value friends for their own sake (whatever that may mean), not for what they can do for us?—but Plato is exploring the foundation of friendship. It is not clear that he is wrong that the foundation of friendship is some kind of need, and he still leaves room for affection and a less self-centred type of relationship to develop on the utilitarian foundation: one of his examples of ‘friendship’, for instance, is the love of parents for a baby (212e–213a). Moreover, the criticism anachronistically applies a post-Romantic conception of friendship to the ancient Greeks, for whom the value of friends was uppermost.
The interlocutors rest on their laurels only momentarily, however, before Socrates introduces a complication. After summarizing the previous argument in somewhat different terms (218d–219b), he draws out from it, as an implication, the notion that anything attractive is found attractive only as a means to some further attractive end (219b–c). This process will necessarily either go on ad infinitum, or end with a ‘primary lovable object, the final end which makes everything else that is lovable lovable’ (219d). And if there is such a primary lovable object, the model of ‘friendship’ they arrived at before—that it is the presence of something bad that attracts what is neither good nor bad towards the good—needs qualification, because only the primary lovable object is a true friend (object of desire), while everything else is a subordinate or second-rate friend, a friend not in itself but only as a means to a further end (219d–220b). Plato thinks that we should now be in a position to see what is essential to friendship or attraction.
The qualification that Plato insists upon, somewhat tortuously (220b–221d), is to eliminate as non-essential the idea that it is the presence of badness which makes something lovable or attractive. In actual fact, he says, it is just desire or lack which makes something attractive. He makes this move, I think, because he wants the primary lovable object to be attractive in itself, not because of the presence of something bad; so there is a hint here of the possibility of altruistic friendship, in which something is found attractive not just for the good it can do the person who finds it attractive. Desire is then analysed, with alarming abruptness (221e), as desire for something close to oneself. Since closeness is a symmetrical relationship (if A is close to B, B is also close to A), then if A loves, likes, desires, or befriends B, B must also do the same for A (222a). We have turned back along the dog-leg to friendship as a personal and reciprocal human relationship, but more importantly this idea conflicts with principles earlier taken to be stable, especially that bad people cannot be friends (222b–d). And so the dialogue ends with the usual aporia—and as usual the aporia is caused by Plato’s own terminological confusion and his insistence on trying to find just one common core to a multi-faceted concept.
It is possible to spot one or two propositions that remain unrefuted in the course of the dialogue, and so to claim that Plato means us to deduce that, for instance, what is good in a person is attracted towards what is good in another person, and that this is the basis of their friendship. Then we can note that, at Phaedrus 255b, for instance, Plato readily agrees with this proposition: ‘It is fated’, he says, ‘that bad men can never be friends and that good men can never fail to be friends’ (see also Laws 837a). In the context of mutual affection between good people, Plato remains convinced that the principle ‘like is friend to like’ is valid (Gorgias 510b, Phaedrus 240c), despite its rejection in Lysis. So if we reinstate the principle, then what is neither entirely good nor entirely bad (i.e. a person) may be the friend of what is neither entirely good nor entirely bad (another person), to which it is similar, in those respects in which the two parties are good. And we can say that the basis of their friendship is that they should be useful to each other (benefit each other, bring out the good in each other), because this instrumentalist assumption is nowhere questioned and guides several of the arguments. All desire is for the good; something perceived as good (whether or not it actually is) is always the object of desire. When a person’s desire for the good is channelled through a relationship with another person, that is human friendship.20 The dialogue is not an incoherent muddle, shifting from the reciprocal to the passive sense of ‘friendship’ and simultaneously from a reciprocal human relationship to one-sided attraction: it is a study of attraction, bracketed by an investigation of the human relationship as a type of attraction.
It also seems clear, on surveying the course of the arguments, that what really interested Plato was what we may call the non-obvious, subliminal, or even metaphysical aspects of attraction. When we find something attractive, what is it about it that we are really attracted to, and why? Does this differ from what we say or think we are attracted to? Plato’s main suggestions here are that we are attracted to something either because of some imperfection in ourselves (as a body is attracted towards a doctor because of its illness) or because of some psychic need which is as natural as physical thirst and hunger. Lysis is an essay in psychology as much as a logical investigation, and it is precisely this interest in the hidden aspects of love that Plato was to take to even more rarefied heights in Symposium and Phaedrus. In these dialogues the ‘primary lovable object’, so barely hinted at in Lysis that we cannot even say what it is, is fleshed out as the domain of those metaphysical entities which are usually called ‘Forms’: the Form of Beauty is the true object of love, and all other lovable objects are pale reflections.
Several questions arise from our survey of these three dialogues, and we may as well start with one of the trickiest. Each dialogue ends in aporia: the interlocutors fail to come up with definitions that satisfy the elenchus. Can the Socratic elenchus do more than point up inconsistencies or other forms of deficiency in others’ views? Is it purely destructive, or can it be constructive?21
The structure and nature of the standard elenchus is pretty straightforward. Essentially, one of the speakers (usually not Socrates) offers a definition of a moral concept. Socrates then shows how this definition D (or its consequences) clashes with some other proposition P (or its consequences).22 Faced with a choice of rejecting D or P, the proposer of the original definition invariably chooses to reject D.23 He prefers P to D because the ideas represented by P are, as the summary of the dialogues above shows, invariably general propositions taken to be self-evident, such as ‘Courage is a good thing’ or ‘Excellence is beneficial’. The interlocutor weighs up the evidence in favour of D and the evidence in favour of P, and rejects D.
This is the bare, logical skeleton of the elenchus. It is not so much a refutation of D as it is a testing of D, by indirect means, and though Socrates may sometimes conclude that D is wrong (e.g. Charmides 161b), he may also say no more than words to the effect that ‘D is not necessarily right’ (e.g. Charmides 160c). Even when he says that D is wrong, he may be saying no more than that D is conditionally wrong: it is wrong if P is accepted. This is important: logically, it is clear that the elenchus can do no more than test for consistency, and it is in the first instance not a refutation of D, but a refutation of the interlocutor’s belief in D and a test of his ability to defend D. At Charmides 162d Plato has Critias protest that Charmides’ failure to defend the third definition of the dialogue shows no more than that Charmides put up a bad defence; it does not show that the proposition is inherently wrong.
The elenchus is not just a logical tool; each elenctic conversation hinges too much on the views and even the character of the interlocutor to attain objectivity. Then again, Socrates often distorts or alters D, or fails to get the interlocutor’s secure assent to all the consequences of either or both of D and P. In short, Plato finds a number of ways to blur the potentially clean logical edges of the elenchus. The personal character of the elenchus is inescapable and is enhanced by a factor called the ‘sincere assent’ constraint.24 If the elenchus is to purge an interlocutor of the conceit of knowledge, which is what Plato takes it to do (as at least one of its primary purposes), then the interlocutor must be present (so at Meno 71d Socrates refuses to engage with the absent Gorgias) and must believe in D, because otherwise the argument will not show that he thought he knew something when he did not, and he must give his assent to P and to the consequences of P which prove to be D’s downfall.
But Plato does not have Socrates consistently insist on sincere assent: he waives the constraint from time to time, or allows interlocutors to get away with qualified assent, or professes himself interested only in the ideas, not the people. In this volume, for instance, we meet the constraint at Meno 83d and Charmides 166d–e, but nowhere else. This suggests that Plato did think that the elenchus could or should be a tool for examining ideas, not just people, and there is plenty of other evidence to support this view, starting with the explicit assertion of this at Charmides 161c: ‘We’re not remotely interested in considering whose idea it is, just in whether or not it’s true’ (see also 166c–e). Then again, in several dialogues (e.g. Euthyphro 11e, Hippias Major 293d) the conversation does not end with the aporia of the interlocutor: Socrates himself makes a suggestion to keep things going. Likewise, several times in Meno (within the passage 74b–76e) and elsewhere (in the dialogues translated in this volume at Laches 192a–b), Socrates provides model definitions to help his interlocutor along. It is true that in all these cases Socrates and his interlocutors end up in aporia anyway, but that is not the point: the point is that they tried; they did not just stop as soon as refutation had occurred. In short, Plato has Socrates take a genuine interest in examining ideas for their own sake, even if they come from him or from some external source such as a poet. The elenchus is above all a method of enquiry, not just a means of puncturing someone’s conceit.
Those who believe that the elenchus can only be destructive point above all to the fact that this is all we are shown in the dialogues: we are consistently shown interlocutors reaching a state of puzzlement, but no more. Plato’s Socrates clearly believes in the positive, cathartic effects of aporia, but those positive effects are not a direct result of the elenchus; the elenchus is no more than a preliminary. ‘In order to make men virtuous, you must make them know what virtue is. And in order to make them know what virtue is, you must remove their false opinion that they already know. And in order to remove this false opinion, you must subject them to elenchus.’25
However, there is overwhelming evidence that Plato did think the elenchus could have constructive results, and could do so directly, not just as a preliminary. In a later dialogue, at Theaetetus 149a, Plato has Socrates say that the interpretation of the elenchus as purely destructive is a sign of ignorance; he can also use it to elicit ideas from people. At Charmides 166d Socrates boldly claims that the elenchus can reveal ‘the nature of each and every existing thing’, and a passage a little later in Charmides also supports the idea that the elenchus can have a constructive purpose. If, as scholars agree, the definition of sōphrosynē as ‘knowledge of knowledge and of lack of knowledge’ (166e) is meant to be, at least in part, a description of the effect of the Socratic elenchus, it is important to note that it is not just the ability to uncover lack of knowledge, but also of actual knowledge: the elenchus does not just prick bubbles of conceit,26 but reveals when someone does genuinely know something. What survives the elenchus may be taken to be true, as Plato says at Gorgias 479e, 505e, 508e–509a, and at Crito 46b, 48d–e, and 49d–e; this is also implied by Sophist 230a–d (paraphrased on p. ix).Charmides 175d is just one of several passages where Plato has Socrates say that he expects the elenchus to come up with the truth.
How, then, does something survive the elenchus? In order to challenge D, Socrates produces another idea, P, which directly or indirectly contradicts D. In accepting P, or at any rate preferring P to D, the interlocutor is accepting that P is right, or is more likely to be right than D.27 In other words, these more general propositions not only regulate the discussion, but also serve as the parameters for possible constructive accounts: in order to survive the elenchus, a proposition must be consistent with P. Even if the dialogues fail to take us past aporia, they do show how a constructive discussion should proceed, and in Gorgias (a later, more reflective work), Plato gives several examples of elenctic arguments throwing up ideas which are taken to be true.
But how can Socrates claim to be searching for truth when all he can reasonably expect to do is test for consistency? As Irwin trenchantly puts it: ‘Whatever Socrates may think, the formal structure of the elenchus allows him to test consistency, not to discover truth. If I survive an elenchus with my original beliefs intact, I have some reason to believe they are consistent; but they may be consistently crazy.’28 The way out of this puzzle is to recognize (as a number of scholars have done) that Plato’s Socrates seems to believe that there are two kinds or degrees of truth.29 On the one hand, there is the knowledge which experts have; on the other hand, there is correctable or fallible knowledge, as exemplified in everyday speech, for instance, when a non-doctor says, ‘I know it’s unhealthy to smoke.’ An expert can be reasonably certain that he has grasped the truth of some matter. Correctable knowledge, however, is what Plato’s Socrates believes the elenchus can produce. It is correctable in the sense that further bouts of the elenchus on the same issue may improve one’s knowledge. To the extent that Socrates himself knows anything (as he occasionally claims to) or steers arguments in particular directions (which presupposes hunches, at the very least), he may claim to have correctable knowledge. At the same time, his constant disavowal of knowledge is sincere: he is a genuine participant in the search for more certain knowledge.30
In short, then, the truth which Socrates searches for by means of the elenchus is the kind of truth which accompanies consistency. If a consistent set of beliefs, which incorporates notions (all those Ps) which are reasonably held to be true, survives repeated elenchi, it has a better chance of being true than an inconsistent set. Consistency is close to being the mark of a set of true beliefs, Plato’s Socrates believes (and he would not be the last coherence theorist to do so);31 and we may go along with him to the extent of agreeing that consistency is both rare and rationally compelling. Nevertheless, Plato is aware that this kind of truth falls short of absolute truth, which he attributes to genuine experts.
The elenchus is not a monolithic enterprise; it is complex, hard to pin down, and constantly surprising. It punctures interlocutors’ conceit of knowledge as a preliminary to replacing false beliefs, challenges the moral views by which they have guided their lives, throws up propositions on which any more defensible views should be founded, and constantly seeks the clarification of ideas and the attainment of beliefs which serve to explain broad moral issues and to allow people to live more moral lives in the future.
Each of the three dialogues paraphrased above is motivated by the search for the definition of a moral concept. Why was definition important for Socrates, and what did he expect of it?32 In Socrates’ time, disputable terms (such as moral terms, above all) were not settled. Are they ever? But there was not even a dictionary or encyclopedia to which one could refer, and people used these terms in very different ways, depending on their social status (a lot of Socrates’ interlocutors, who were always aristocrats, betray their snobbishness), or on how much of the new learning they had imbibed, or on what various poets had said about the concept in question, and so on. Socrates saw his job as settling definitions once and for all, in a rational manner, as a legacy to the future, so that people could then know what these terms meant. Hence he rightly kept emphasizing that one needs first to know what a thing is before trying to determine its properties: one does need a standard to refer to in the case of dispute, and if one also adheres to the principle of univocality, that standard will be single.33
It is important to notice that, in his search for definitions, Socrates is not committed to a view which became known in the scholarly literature as the ‘Socratic fallacy’—that one can not know anything about a concept until one can define the concept and until one knows what it is in itself. On this view, for instance, Lysis 223b was taken to be saying that, despite being friends, Socrates and his companions could not even know if they were friends unless they could define friendship. This position is obviously absurd in itself, and it is not true to the relevant texts. Here, for instance, Plato is saying no more than that friends have the best chance of knowing what friendship is. Other alleged versions of the fallacy occur in our dialogues at Meno 71a–b and 100b, Charmides 159a, 176a–b, Laches 190b–c, and Lysis 212a. But a close look at these passages (and at Euthyphro 6d–e, 15c–e, Hippias Major 286c–d, 304d–e, Republic 354b–c, Gorgias 448e, 463c) shows that, in committing Socrates to the epistemic priority of definition, Plato committed him only to a nest of reasonable points of view: (1) in order to determine whether or not F has such-and-such a disputable property, it is necessary to know first what F is (e.g. it helps to know what excellence is in order to decide whether or not it is teachable); (2) there are certain things only an expert knows about F, and expertise requires knowing the essence of F or being able to define F; the rest of us, while falling short of expertise, may still have true beliefs about F; (3) in certain cases, theoretical knowledge is preferable to experiential knowledge.34
So Socrates asks for a definition of a term, and his interlocutors attempt to answer him to his satisfaction. There are a number of things wrong, Socrates thinks, with the answers he usually gets, and he expresses his dissatisfaction with their first attempts by saying that what he wanted was a definition of F in itself, not a catalogue of F’s properties, nor a list of kinds of F behaviour, nor anything else. But it is extremely difficult to know what Plato was asking for when he asked for a definition of F in itself as opposed to naming any of its attributes. Was he, as some commentators think, trying to introduce a new entity into our conceptual treasury—an entity F which is ontologically distinct from all the things and kinds of things characterized as F? Or was he simply using the term ‘F’ as a convenient way of referring to a universal, the common characteristic of all F things, without implying any or much metaphysical baggage? It seems most likely that Socrates was not a metaphysician (that the theory of Forms was a Platonic development) and that he conceived of the universal as an inherent property of each and every particular F or kind of F: ‘The definiendum of a Socratic definition of F-ness or the F is probably an attribute, which (a) is one and the same in all things that are F, (b) is that by reason of which all F things are F, (c) is that by which all F things do not differ but are the same, and (d) is that which in all F things we call “F-ness” or “the F”.’35
But there are still difficulties. If we define ‘human being’ as ‘animated featherless biped’, that will enable us to identify human beings, but we have named attributes of human beings. One of the perennial problems with definitions is that it is very hard to avoid employing terms in the definition which in turn beg for their own definitions. An infinite regress ensues. Plato shows himself aware of this problem and would allow what we may call working definitions or conceptual analyses (Meno 75b–c; see also the note to Laches 192b). They may not perfectly state the essence of excellence (or whatever), but they allow us to identify cases of excellence, and that is a good start.
Now, there are different kinds of definition. Above all, a ‘nominal’ or ‘dictionary’ definition tells us how people or a privileged group of people use or should use a term, whereas a ‘real’ or ‘essential’ definition tells us about the thing itself, not just about verbal usage (though that may be covered as well, with all the possibility of confusion that may be implied by saying that Plato was looking for two different kinds of definition at once).
It seems clear that Plato was not after just a nominal definition. A true definition, he held, must have enormous explanatory power; it must cover each and every instance of the concept being defined, and it must cover only the concept being defined. It must allow us to identify cases of the concept in question, to understand what makes them such cases, and to deduce further properties of the concept (as in Meno the definition of excellence is supposed to tell us whether excellence is teachable). It must also be such that, every time one uses the term being defined, one could replace it with the definition, but not in the way that a mere synonym would: it must reveal something about the essence of the concept being defined, about what self-control (or whatever) really is, and what makes a self-controlled person self-controlled. It must not simply pick out contingent attributes of the concept: self-control may indeed on occasion be akin to modesty or unhurriedness, but it is not always, and so defining self-control in these terms does not illuminate its essence. It must not implicitly employ the term to be defined in the definition: Laches’ first definition of courage, for instance, is (in brief) resistance. But if courage is resistance, then in answer to the question ‘What is courage?’ he is saying no more than courage is the ability to act courageously. In short, it must articulate the structure of reality, or at least of as much of reality as is encapsulated by the term in question. It seems reasonable to think that Plato was striving towards what would later be called definition per genus et differentiam: in this way F is related to members of the same family, and we also hear what makes it specifically different. So Plato has Socrates qualify Laches’ generic definition of courage as ‘persistence’ by specifying that perhaps ‘intelligent persistence’ stands a better chance as a definition, or he qualifies Meno’s definition of excellence as ‘rulership’ by specifying that it must be ‘just rulership’, or at Meno 76a he offers a model definition of shape as ‘the limit of a solid’.
The rigour of these requirements on a good definition has important moral and methodological consequences. It is precisely because Socrates’ expectations are so stringent and perhaps unrealistic that he can tie his interlocutors up into knots and reduce them to aporia. A core assumption of the elenchus is that knowledge presupposes the ability to give an account: if I know F, I can say what F is. Moroever, if someone possesses an aspect of excellence, one may reasonably expect him to be able to give an account of it, or to know it (Charmides 158d–159a, Laches 193d–e, Lysis 223b). Hence Socrates converses with Charmides about self-control because Charmides is held to possess self-control; he converses with generals about courage, and he converses with friends about friendship. But what happens? They are incapable of producing satisfactory definitions. They thought they knew what F was, but end up being uncertain of this—and even doubting that they possess the quality, because they lack a way of safely identifying whatever it is that they possess as an instance of the quality in question. They are therefore impelled—or so Plato piously hoped (Meno 84a–c)36—to start again, to find some way of possessing the aspect of excellence and of knowing for sure that they do so.
As already remarked, Meno begins as a dialogue of search. Meno asks whether excellence (virtue) is teachable or a natural endowment, 37and Socrates, true to the principle of the epistemic priority of definition (p. xxxiv), converts the question to ‘What is excellence?’ Meno’s attempts at definitions are all unsatisfactory. First (71e–72a), he does not even answer the question. Instead of addressing the question ‘What is excellence?’, he acts as if the question had been ‘What different kinds of excellence are there?’: he lists examples of excellence, and Socrates deploys the assumption of univocality to ask what is common to them all. Meno then suggests that it is the ability to rule (73c), but Socrates argues that this definition is simultaneously too broad and too narrow. First, there are counter-examples (is a slave excellent if he has the ability to rule his master?), and second, if the ability to rule bears any relationship to excellence, it does so only because a good or excellent ruler makes use of justice (in other words, it is not merely a question of what one does, but how one does it). But justice is only one aspect of excellence, so that does not bring us any closer to understanding what excellence is as a whole or in itself. Finally, Meno suggests that excellence is the ability to procure good things for oneself (78c, a modification of 77b), but if the ability to procure good things is to stand a chance as a definition of excellence, it needs a qualification that names at least one aspect of excellence (e.g. ‘procuring things justly, not unjustly’), and so the definition fails for the same reason as the previous one, and is meaninglessly circular.
By 80a, then, the dialogue has reached the usual state of aporia. Unlike Charmides, Laches, and Lysis, however, Meno does not stop there. At 80a–d, wriggling with embarrassment at his failure in the first part of the dialogue, Meno tries to regain the advantage first by blaming Socrates for his own failure (a not unparalleled ploy from a Socratic interlocutor), and then by challenging Socrates with an apparent paradox. What is the point of trying to define anything—of asking ‘What is F?’ about anything—when you necessarily either already know it or you do not? If you already know it, you do not need to undertake the search (this is the ‘paradox of enquiry’); if you do not know it, how will you recognize it when you find it (this is the ‘paradox of recognition’)?
Plato could perhaps have given Socrates easy responses. He could argue against the paradox of enquiry that one knows that one is looking for ‘excellence’ (or whatever it is that is named in the question which initiates the search) without knowing what actual thing corresponds to the word in inverted commas, or in other words that there are stronger and weaker senses of ‘know’: we need to be able only to identify a subject in order to make it a topic for enquiry. And against the paradox of recognition he could argue that he will recognize his quarry when he reaches it because it will enable him to identify what is common to all instances and kinds of excellence. He could even have constructed an answer along the lines of Charmides—that it is possible to know in some sense both what one knows and what one does not know. Instead, however, Plato takes the paradox to strike at the very heart of the Socratic enterprise of definition. In asking ‘What is F?’, Socrates assumes an answer is possible, and that F can be known.38 Plato interprets Meno’s paradox as asking how we can know anything (or at least anything non-empirical, because Meno’s paradox does not rule out empirical knowledge, only the kind of knowledge Socrates was after with his ‘What is F?’ question).
Plato’s answer—an answer of astonishing daring—is that everything we know, everything we seem to ‘learn’, is a truth to which we already have access. The paradox claimed that we either do or do not have knowledge, and that in either case enquiry (or at least the kind of enquiry where the quarry has been set in advance) is superfluous; Plato’s response is to claim that there is middle ground —that we may know something latently without knowing it consciously. The soul is immortal, and at some unspecified time or times in the past it has acquired knowledge of everything.39 As the experiment with the slave (82b–85b) is supposed to show, the correct process of questioning (by oneself or by an external agent) elicits recollection of the relevant truths. At first, this may be insecure, but repeated questioning (85c) will lead a person to understand why things are as they are, and convert insecure knowledge (which is another way of saying ‘true belief’) into certain knowledge (97a–98b).
The experiment with the slave is explicitly a version of a typical Socratic elenchus, and what is more Plato provides us with his own commentary on the proceedings. The conversation is interrupted in order for Socrates to explain what has been going on, and what will happen next. The questioning of the slave is said to produce distinct stages in his progress. First, his false ideas are proved wrong, with the result that he succumbs to aporia (82e, 84a). Second, latent true beliefs are aroused within him; they are based at least partly on clues given by the negative first stage (85b–c). Third (though we are not shown this stage) his true belief could be converted into knowledge (85c). Socrates’ confidence, throughout the dialogues of search, that he will eventually elicit a true belief from his interlocutor is justified, we are told, because the true beliefs are in there, inside the interlocutors: that is the point of the theory of recollection. These innate true beliefs are what allow an interlocutor instantly to recognize the greater evidential value of P over D. The elenchus can then build on these true beliefs to elicit further true beliefs, because of the kinship of all nature (81 d). Socrates, then, would reject Irwin’s logical point (p. xxxii) that it is possible to be consistently crazy or wrong: he believes that we innately have true beliefs which, under dialectical questioning, will lead us to reject false or immoral ideas.40
The paradox of enquiry has been defused: there is a point to the elenchus. There is a point to searching for what one does not know, because the search—the questioning—puts one into contact with innate truths. Plato does not have Socrates directly address the paradox of recognition; presumably such recognition is gained by pitting the knowledge, framed as a definition, against test cases. The definitional formula that encapsulates the knowledge may need refinement as a result of this process, but that too is part of the process of converting true belief into knowledge.
Socrates’ interest in establishing definitions was to come up with something stable in the face of the uncertainty generated by various traditions. These traditions, which count as ‘teaching’, give us the contents of our conscious mind. These contents are our beliefs, which may be more or less organized into a coherent system and which may be true or false; but they are not pieces of knowledge. Knowledge lies latent beneath them, and needs recovering. Then we will have a stable epistemological state and we can come up with definitions or standards to refer to. So the recovery of knowledge is simultaneously the ‘acquiring’ of definitions. Not everyone will do this, presumably. Whereas everyone has some contact with their latent knowledge, because Plato is adamant that all learning is recollection, not everyone will convert their beliefs to knowledge. Perhaps only philosophers do so, and the means of their doing so is by questioning themselves, or being questioned by Socrates. No wonder Plato’s Socrates saw himself as the gods’ gift to Athens (Apology 30a).
Of course, in the case of the slave, Socrates steered the discussion: he was not ignorant of the answer, and he fed the slave leading questions to guide him towards finding the answer for himself (in so far as that is possible). Where both or all interlocutors are ignorant (remember that Socrates usually professes ignorance about any important matter), the way to proceed is to make assumptions (86d–87b). Now, those who believe that the Socratic elenchus was designed purely for negative purposes, for refutation, proclaim the introduction of the ‘hypothetical method’ in Meno as a new departure—a method of enquiry that will allow Plato to aim for positive results. But since (as we have seen earlier in this Introduction) the elenchus is not merely destructive in its effects, and since Plato has explicitly used assumptions earlier (as at Charmides 169d), this must be wrong.
In fact, the ‘hypothetical method’ is another way—in addition to the theory of recollection, that is—in which Plato reflects in Meno on his Socratic legacy. The hypothetical method is little more than a formalization of the practice of the elenchus, in its constructive rather than aporetic mode.41 Just as a geometrician asks what has to be the case for something else to be the case, so Socrates (here in Meno) asks from what proposition it will follow that excellence is teachable, and traces this back to the logically prior proposition that excellence is knowledge, and this proposition in turn is traced back to the supposedly prior proposition that excellence is good (87d). The original assumption is tested not just by seeing what its consequences are, but by seeing whether it stems from propositions to which all the parties agree, and which may even be true. But this is exactly the practice of the elenchus: an assumption (a proposed definition) is tested in the same ways. Laches’ proposed definition of courage as ‘mental persistence’, for instance (192b), is tested by referring back: if mental persistence is to be courage, then courage may be a bad thing, since mental persistence is not always a good thing. Do we agree that courage is a bad thing? No—so let’s start again with a different assumption, one that squares with what we have gained from this, that courage must be good. The same word that is translated ‘assumption’ or ‘hypothesis’ is used at Charmides 163a for the thesis that self-control is doing what pertains to oneself and at 160d for the ‘higher’ hypothesis that self-control is admirable; at Euthyphro 9d the word is used again for one of the trial definitions proposed for piety. Neither Euthyphro nor Hippias Major end with the aporia of the interlocutor: in both cases Socrates himself proposes a new definition or idea for examination, and since Socrates professes ignorance of all such matters, we may fairly take his proposals to be provisional or hypothetical. A hypothesis is simply a proposition that is put forward in order to be tested, and that is what happens throughout the Socratic dialogues.
In Meno, then, Plato suggests that the way out of the aporia that often frustrates moral enquiry is to find a logically prior proposition, which can then be investigated in the same way (see, in brief, the movement of Meno 73a–c). If the interlocutors still fail to agree, the process can be repeated until a point of agreement is reached. That point of agreement may well serve as a definition—albeit necessarily a provisional one—of the concept in question, and Plato tantalizingly dangles the possibility that this process is related to the way to convert true belief into knowledge, by working out the chain of causes that make something the case (98a): the logical process of working back to a point of agreement may well be the same process as that which converts true belief into knowledge, by making true beliefs less provisional and more anchored. Working back to ‘higher’ (more general) hypotheses is meant to cover the ‘different ways’ of looking at a topic (85c) which will convert one’s view of it from true belief to knowledge. This shows that by the time he wrote Meno Plato was aware that Socrates had been operating with two degrees or kinds of knowledge (see p. xxxii): fallible knowledge (here called true belief) and expert knowledge, which is what one increasingly approximates to by working out reasons (98a).
Meno also allows us to clear up another puzzle too. Plato’s Socrates, we know, tries to elicit knowledge from his interlocutors in the form of a definition, but the ability to state even a true definition is not enough on its own to count as or signify knowledge. If it were, Nicias’ definition of courage as knowledge of good and bad would make him a man of knowledge, and so perhaps would Critias’ definition of self-control as doing what pertains to oneself. The ability to state a true definition is necessary for knowledge, but it is not sufficient. When Socrates tests a definition by means of the elenchus, he is concerned to turn the interlocutor’s belief (true belief, if the definition is true) into knowledge (see, perhaps, Laches 194c). Now we see from Meno that it is working out the chain of causes that converts true belief into knowledge. I have already suggested that the ‘method of hypothesis’ is a formalization of the practice of the elenchus and is the same as ‘working out the reason’, in that it anchors the belief with a chain of reasons or higher hypotheses and so converts it into knowledge; and now we have reached the same conclusion from another angle.
When the enquiry into the teachability of excellence resumes, then, they make the following assumptions: (1) if excellence is a kind of knowledge, it is teachable, and otherwise it is not; (2) excellence is good for one. The latter assumption allows Socrates slyly to reintroduce the issue of the nature of excellence (apparently banished at 86d–e, and constantly resisted by Meno throughout the dialogue) and to argue that excellence is knowledge or a kind of knowledge (87d–89a); therefore, ex hypothesi, it is teachable (89b–c). We are not stymied, then, by our inability to define concepts; we can make assumptions or hypotheses and carry on —and one of those assumptions may in any case serve as a definition. We should note, however, that since ‘teachable’ has been redescribed by the theory of recollection, according to which innate knowledge is recovered by questioning, then to say that excellence is teachable is to say that excellence is both innate and teachable. The stark either–or option with which Meno began the dialogue has been mitigated.
In Meno the first set of assumptions led to the conclusion that excellence was teachable, but Socrates now makes a further assumption, that if it is teachable, there must be teachers of it (which is to say that the proposition that excellence is teachable is traced back to the logically prior proposition that excellence has teachers). He then argues at length that there are no teachers of excellence (89d–96c), and that therefore excellence is not teachable. If it is not teachable, people must become good by some other route, and the final suggestion is that excellence is true belief, which is just as good as knowledge for all practical purposes (97a–98c). But true belief, Plato claims, is neither a natural endowment nor a product of teaching, and therefore it is not within our control. The dialogue concludes with the evidently ironic suggestion (99d–100a) that excellence is a miracle, ‘a dispensation awarded by the gods’.42
What is going on? Not only does this contradict the earlier conclusion that excellence is teachable, but it also contradicts a constant aspect of Socratic thought, the intellectualist assumption that excellence is knowledge, which, as we have seen, is operative in Laches and Charmides (and other dialogues). Partly, Plato is concerned to demonstrate the importance of hypotheses, and their danger: different hypotheses lead to radically divergent conclusions. But, apart from methodology, can we also extract any doctrinal conclusions from the dialogue? Is Plato casting doubt on the Socratic idea that excellence is knowledge?
We cannot give absolutely secure answers to these questions, and it seems very likely that Plato himself was uncertain whether or not excellence was teachable, or at least was unwilling to commit himself to its being just teachable (as opposed to a natural endowment or whatever else). It is implied by Socratic thought that it is, but whenever Plato addressed the question directly, he leaves things open. In Protagoras he had Socrates argue first that excellence is not teachable, and then that it is; in Meno this order is reversed. But as long as we are not looking for certainty, we can draw some conclusions.
The elimination of the proposition that excellence is knowledge was achieved by arguing that excellence is not teachable because it is not taught. But Plato holds out the possibility at 99e–100a that one day excellence might be taught by a Teiresias–like figure (Socrates? Plato?), which implies that excellence is knowledge after all. Plato has not abandoned the idea that excellence is knowledge, but he has upgraded it to an ideal. Moreover, since abandoning the attempt to define excellence at 86d, nothing about excellence is taken to be known and everything is hypothetical: if there are no teachers, it is not knowledge—but there may be teachers one day. Even the conclusion may be understood to be provisional, since Plato says that it awaits confirmation (100b); in other words, if excellence is not knowledge and does not come as a natural endowment, it may be true belief and therefore given by the gods. Alternatively, we may understand the conclusion to be referring to ordinary, everyday excellence: even if this is equated with true belief and is due to divine dispensation, ideal excellence is still knowledge.43 Plato may also have intended us to reflect—as the dialogue prompts us—on teaching: conventional excellence may indeed be teachable by a Sophist or a father, but true excellence, true knowledge, is attainable only through recollection.