Plato’s Philosopher

It is a fine question how the aim and method of the philosophical enterprise is to be related to the beliefs we bring to that enterprise. It is bootless to pretend we can start by somehow setting aside the equipment with which we approach philosophy, for then there would be nothing with which to work. We can, however, ask whether the main point of philosophising is to examine, clarify, reconcile, criticise, regroup, or even unearth, the convictions or assumptions with which we began, or whether something more is possible: a search which might lead to knowledge or values that were not in sight at the start, and not necessarily implicit in what we then knew.

Each of these enterprises has its obvious difficulties. No one can object to the attack on confusion, conflict, obscurity, and self-deceit in our everyday beliefs; these defects in our views of ourselves and the world exist in profusion, and if some philosophers can with skill or luck do something about reducing them, those philosophers deserve our respect and support.

But it would be disappointing to suppose this is all philosophy can do, for then philosophy would seem to be relegated to the job of removing inconsistencies while entering no claim to achieve truth. Consistency is, of course, necessary if all our beliefs are to be true. But there is not much comfort in mere consistency. Given that it is almost certainly the case that some of our beliefs are false (though we do not know which), making our beliefs consistent with one another may as easily reduce as increase our store of knowledge.

On the other hand, it is not easy to see how to conduct the search for truths independent of our beliefs. The problem is to recognise such truths as we encounter them, since the only standards we can use are our own. Where the first approach makes no attempt at fixing objective standards, the second can seem to succeed only by illegitimately re-labelling some portion of the subjective as objective.

There is an obvious connection between the two pictures of the method and aim of philosophy and two traditional concepts of the nature of truth: one method goes naturally with coherence theories of truth, the other allies itself with correspondence theories. A coherence theory in its boldest and clearest form declares that all beliefs in a consistent set of beliefs are true; coherence is the only possible test of truth, and so coherence must constitute truth. So stated, a coherence theory of truth can be taken as a defence of a philosophical method which claims only to remove inconsistency: for once inconsistencies have been excised, the coherence theory assures us that what remains will be an unadulterated body of truths.

Correspondence theories, on the other hand, maintain that truth can be explained as a relation between a belief and a reality whose existence and character is for the most part independent of our knowledge and beliefs. Truth of this sort is just what the second approach to philosophy seeks. But unfortunately correspondence theories provide no intelligible answer to the question how we can in general recognise that our beliefs correspond in the required way to reality.

No theme in Plato is more persistent than the emphasis on philosophical method, the search for a systematic way of arriving at important truths, and of ensuring that they are truths. Yet I think it is safe to say that Plato not only did not find a wholly satisfactory method, but did not find a method that satisfied him for long. In the early dialogues, in which Socrates takes charge, the elenctic method dominates, and there is nothing in those dialogues to promote the suspicion that Plato, or Socrates, sees the need to add anything to it. Yet it seems clear that it is a method that at best leads to consistency; if it is supposed to yield truth, the ground of this supposition is not supplied. In the middle and some of the late dialogues Plato suggests a number of ways in which the elenchus might be supplemented or replaced by techniques with loftier aims. But what is striking is that Plato does not settle on any one of these methods as a method guaranteed to achieve objective truth: one by one the new methods are discarded, or downgraded to the status of mere useful devices. Plato often makes it clear that he recognises the inadequacy of his methods for achieving his aims; and the inadequacy can be painfully apparent to the modern reader.

Plato and Aristotle are often held to be paradigms of the contrasting methods. Aristotle insisted, at least in moral philosophy, that views that are widely shared and strongly held within our own community must be taken seriously and treated as generally true. But Plato, we are told, ‘throughout the middle dialogues ... repeatedly argues against the philosophical adequacy of any method that consists in setting down and adjusting our opinions and sayings. It is Plato who most explicitly opposes phainomena, and the cognitive states concerned with them, to truth and genuine understanding. It is also Plato who argues that the paradeigmata that we require for understanding of the most important philosophical and scientific subjects are not to be found in the world of human belief and perception at all.’1

This is, indeed, the standard view, and when, as in this passage, it is restricted to the middle dialogues, it is roughly correct. Even with this restriction, though, it needs to be taken with a grain of salt. Nussbaum gives, as a striking example of the opposition of methods in Plato and Aristotle, their views on akrasia, or weakness of the will. Socrates, as we know, paradoxically maintained that akrasia was impossible; he argued that if an agent knows what is good, he cannot fail to act in accord with that knowledge. Aristotle, on the other hand, held that the common view must be right: despite Socrates’s arguments, there are cases of akrasia.

How clear is the contrast here between Plato (really Socrates) and Aristotle? In the early dialogues we meet with the most emphatic cases of conclusions that plainly contradict common conviction: yet nothing is said to show that the elenctic method is capable of more than revealing inconsistencies. In the middle dialogues there are the strong representations just mentioned that philosophy can arrive at truths not dreamed of by ordinary men, and not to be tested by experience: yet in these same dialogues much less is made of the paradoxical character of the doctrines that emerge. In particular, the Socratic denial of the possibility of akrasia is explicitly dropped. To make our own small paradox: you might say in the early dialogues dogmatic claims are based on a method that cannot support them; in the middle dialogues rather tamer results flow from methods which are advertised as leading to absolute and objective truth.

I think that in the end Plato lost faith in the ability of these methods to produce certified eternal truths that owed nothing to the serious goals and convictions of most people, but came to have a renewed confidence in the elenchus, supplemented and refined in various ways, to arrive at truth by way of consistency: in other words, he returned to something like the Socratic method and its approach to the philosophic enterprise.

The line of thought I am pursuing was inspired by a brilliant and provocative paper by Gregory Vlastos called ‘the Socratic Elenchus’.2 Viewed logically, the elenchus is simply a method for demonstrating that a set of propositions is inconsistent. In practice, the elenctic method is employed by Socrates, or some other interrogator, to show that an interlocutor has said things which cannot all be true (since they are inconsistent). If this were the whole story, the function of the elenchus would be no more than to reveal inconsistencies; such a revelation should, of course, be interesting to anyone tempted to believe all the propositions in the inconsistent set.

There is no obvious reason why a philosopher – or anyone else – should be concerned with inconsistent sets of propositions only when they happen to be believed: after all, one can prove a proposition true, and hence worthy of belief, by showing its negation inconsistent. This is no help in establishing substantive, or moral, truths as opposed to logical truths. Nevertheless, it is often helpful, when trying to decide where the truth lies, to appreciate the inconsistency of a set of propositions to which one is not yet committed

It is therefore surprising, and instructive, that in the Socratic dialogues, Socrates usually insists that the interlocutor be seriously committed to the propositions being tested. It is one of the merits of Vlastos’s article that he notices this striking feature of Socrates’s method, and appreciates how important it is. Vlastos quotes from the dialogues: ‘By the god of friendship, Callicles! Don’t think that you can play games with me and answer whatever comes into your head, contrary to your real opinion’ (Gorgias, 500b). ‘My good man, don’t answer contrary to your real opinion, so we may get somewhere’ (Republic I, 346a). And when Protagoras says in answer to a question of Socrates, ‘But what does it matter? Let it be so for us, if you wish,’ Socrates angrily replies: ‘I won’t have this. For it isn’t this “if you wish” and “if you think so” that I want to be refuted, but you and me. I say “you and me” for I think that the thesis is best refuted if you take the “if” out of it’ (Protagoras, 331c).

This last quotation brings out another feature of the elenchus, and helps answer the question why Socrates is so concerned that the people he questions should express their real opinions. Socrates is interested in refutation. The typical elenchus begins with Socrates asking a question, to which the interlocutor gives an answer. Socrates then elicits some further views from the interlocutor (not infrequently by putting them forward himself, and getting the interlocutor to agree), and proceeds to demonstrate that these further views entail the falsity of the original answer. Unless the person being questioned accepts the propositions which refute his original answer, he will have no reason to give up his opening proposal; no particular thesis will have been refuted.

As Vlastos points out, all that Socrates has shown is that the interlocutor’s beliefs are inconsistent, so at least one of those beliefs must be false. But there is nothing about the elenctic method to indicate which belief or beliefs should be abandoned. In the event, it always turns out to be the original proposal. For pedagogic and dramatic reasons, this is clearly the right strategy. But it is a strategy that will lead to true conclusions only if one is careful to start with a false belief, and then to draw upon additional, but true, beliefs to disprove the starting claim. How can Socrates know in advance of using his method what is true and what false? Is the method, after all, just a device for persuasion?

What is clear is that Socrates trusts that the elenctic method does lead to moral truths: the negations, in general, of beliefs held by interlocutors at the start of a discussion. Where p is the original claim, Socrates repeatedly sums up the result of an elenctic argument by saying that the negation of p ‘has become evident to us’, or the interlocutor now ‘sees’ or ‘knows’ that not-p.3 In the Gorgias, Socrates says that his thesis (the negation of the interlocutor’s p) has been ‘proved true’. (Here not-p is: to suffer injustice is better than to commit it.) Presently he puts it even more strongly: ‘These things having become evident in the foregoing arguments, I would say, crude though it may seem to say it, that they have been clamped down and bound by arguments of iron and adamant ...’

According to Vlastos, whose argument I have been following closely up to this point (as well as using his translations), the last two quotations differ substantially in strength from the earlier claims; proving something is more than simply making it evident. Vlastos thinks the claim that the elenchus can ‘prove’ truths is not Socratic; he believes Socrates did assume that the elenchus leads to truth, but it was an assumption on which he did not consciously reflect, and which he therefore felt no need to defend. Throughout the dialogues which precede the Gorgias Plato depicts Socrates arguing for his views in much the same way as other philosophers have done before or since when trying to bring others around to their own view: he picks premisses which he considers so eminently reasonable in themselves and so well-entrenched in his interlocutor’s system of belief, that when he faces them with the fact that these premisses entail the negation of their thesis he feels no serious risk that they will renege on the premisses to save their thesis ...This being the case, the “problem of the elenchus” never bothers Socrates in those earlier dialogues.’ Plato, however (to continue Vlastos’s theory), did come to wonder what ensured the choice of true premisses: and well he might have, given how often what seemed common knowledge to almost everyone else was found to be false according to Socrates. Although the words are put in Socrates’s mouth, it is Plato who, according to Vlastos, realises what must be assumed if the elenchus is to produce truths: the assumption is that, in moral matters, everyone has true beliefs which entail the negations of his false beliefs. It follows from this assumption that all the beliefs in a consistent set of beliefs are true, so a method like the elenchus which weeds out inconsistencies will in the end leave nothing standing but truths. Therefore Socrates did not need to know in advance which beliefs were true, nor did he have to worry that upon discovering an inconsistency, the inconsistency might be removed by inadvertently throwing out the true. For the retained falsity would itself be found inconsistent with further beliefs.

I shall not consider the textual evidence in the Gorgias for this doctrine, since it is clear, on the one hand, that something very like this assumption is necessary if the elenchus is to be defended as a way of reaching truths, whether or not Plato or Socrates realised it; and, on the other hand, it is equally obvious that there is absolutely no argument in the Gorgias or any of the earlier dialogues to support the assumption. Vlastos believes that Plato realised this, and that it is for this reason that in the next three dialogues, written just after the Gorgias, the Euthydemus, Lysis and Hippias Major, the elenchus makes no appearance. In these dialogues there is philosophical argument, but Socrates carries on essentially by himself, acting both as proposer and as critic.

In the Meno Plato finds a new way of defending the elenchus: the doctrine of recollection. According to this theory, everyone is born knowing everything, but the vicissitudes of life have caused them to forget what they know, and to come to believe falsehoods. Once again, it is clear that a method that claims no more than that it can remove what is inconsistent with what is known is adequate to achieve truth. Vlastos describes the theory of recollection as a ‘lavish present’ by Plato to Socrates. ‘By the time this has happened,’ Vlastos concludes in ‘Afterthoughts’, ‘the moralist of the earlier dialogues has become the metaphysician of the middle ones. The metamorphosis of Plato’s teacher into Plato’s mouthpiece is complete.’

Vlastos sees a vast difference between Plato’s two ways of saving the elenchus: the way of the Gorgias, which merely assumes the existence of enough ineradicable truths in everyone, and the way of the Meno, with the transmigration of the soul and the theory of recollection. Years passed between the writing of these dialogues, years during which Plato lost all confidence in the elenchus. ‘Then, one day’, writes Vlastos, the theory of recollection came to Plato.

This is a fascinating story, and Vlastos makes it plausible with a wealth of references to the texts, and a shrewd consideration of the human and logical probabilities. I have no intention of arguing against it, except on one point which is not explicit in his article, but is strongly implied. That implication is that after the Gorgias, Plato permanently lost faith in the idea that moral truths can be elicited from anyone by something like the elenchus. In any case, I want to put forward the hypothesis that at a certain point late in his career Plato returns to (if he ever departed from) the Socratic concern with the good life, the right way to live, and that he depends on the assumption that there is enough truth in everyone to give us hope that we can learn in what the good life consists.

First, we ought to notice that though it is certainly true that the doctrine of the transmigration of the soul and of recollection is new in the Meno, that doctrine is closely related to the methodological assumption which Plato realises, in the Gorgias, is needed to defend the elenchus. Indeed, the doctrine of recollection doesn’t supplant the assumption of the Gorgias: it entails it. Viewed solely as a supplement to the elenchus, the theory of recollection has no need of the doctrine of the transmigration of the soul. Like the methodological assumption of the Gorgias, the theory of recollection postulates that there is enough ineradicable truth in each of us to ensure that the elimination of inconsistencies ultimately results in the elimination of error: when all inconsistency is removed, what remains will be true. From a strictly methodological point of view, the chief difference between the two doctrines is that while the assumption of the Gorgias suggests that the only sure route to knowledge is the elenchus, the theory of recollection places no premium on the elimination of inconsistencies and so invites us to consider methods other than the elenchus in the search for truth.

And of course other methods do come to the fore in the middle and late dialogues: the various methods of ‘ascent’ in the Republic, Symposium and Phaedrus; the method of ‘hypothesis’, of ‘collection and division’; and the method or methods of the Philebus. There is an obvious transfer of interest from moral problems to epistemological and ontological problems, a new concern (in the Theaetetus) about the possibility of perceptual knowledge, and a persistent worry about philosophical method.

Plato did not abandon the elenctic method: what philosopher would? Our concern is with the question what can be expected of that method, and what Plato thought it could deliver. The essential problem is the one with which we began: can philosophy hope to transcend what is inherent in the beliefs and values with which it begins? If not, consistency but not truth is all we can trust it to deliver, and this is what the unaided elenchus promises. But if the theory of recollection is true, the elenchus can do more. The trouble is that the theory of recollection, treated as an essential assumption needed to support knowledge claims, drops out of sight in the dialogues almost as soon as it appears. It is crucial in the Meno, and plays an important role in the Phaedo. But by the end of the Phaedo the doctrine of recollection has been superseded by the method of hypothesis: and this is not a method better suited to prove truths, as we shall see in a moment. The important thing is that Plato no longer seems willing to trust the theory of recollection. Recollection is introduced in the Republic, but it is obvious that quite different methods are the ones on which Plato relies in that work; there are further mentions of the theory of recollection in the Phaedrus, Philebus and Laws, but in none of these dialogues does it serve an important epistemological role.

It seems clear that Plato was not willing for long seriously to embrace the theory of recollection as a source of substantive moral truths. What, then, did he think to put in its place? If, as Vlastos convincingly insists, the ‘problem of the elenchus’ obsessed the methodologically-minded Plato, how can he have relinquished the substance of his ‘lavish present’ to his teacher without finding a suitable substitute?

The answer which appeals to most Plato scholars is that Plato did find other methods in which he placed confidence. This may be true: certainly many methods are mentioned, some of them a number of times, and some of these methods come highly recommended by a character named ‘Socrates’ or ‘the Eleatic Stranger’, etc. Nevertheless, there are three good reasons for not accepting this answer as the last word on Plato’s last word on the method of philosophy. The first is that we can find arguments in the dialogues that show why these methods are in one way or another inadequate or incomplete. The second is that when, in the Philebus, Plato returns once more to the question of the nature of the good life, these methods are not seen to provide the answer; despite several long and difficult discussions of method, it is the elenchus that provides the basic argument. The third reason, which the historically-minded may consider irrelevant, is that we can see for ourselves that none of the alternative methods can provide a firm basis for moral truths, while there is, after all, support for the assumption which, in the Gorgias, is recognised as sufficient to defend the elenchus – the assumption that (in ethics at least, and perhaps metaphysics more generally) there are enough truths in each of us to make it plausible that once our beliefs in these matters are consistent they will be true. It would not be foolish to suppose that Plato figured this out for himself: but the direct evidence is no more than suggestive.

The Philebus is a significant test case. It is a very late dialogue – probably only the Laws is later. It is directly concerned with a Socratic question (one might say the Socratic question), the nature of the good life. Even the phrasing of the problem reminds us of earlier works. The Philebus asks: ‘What is the best possession a man could have?’ It enquires after ‘the proper goal for all living things’. In the Republic, Socrates says: ‘Our argument is over no chance matter but over what is the way we ought to live.’

The Philebus begins with a double elenchus. Socrates sums up the two starting positions: ‘Philebus holds that what is good ... is enjoyment, pleasure, delight, and all that sort of thing. I hold, by contrast, that intelligence, thought, memory ... are, for anything capable of them, preferable and superior to pleasure; indeed to all those capable of a share of them, whether now or in the future, they are of the greatest possible benefit.’ As in the early dialogues, it is assumed without question that ‘everything capable of knowing pursues’ the good: this is taken to be an infallible test of what is acceptable as the good life. With the aid of this assumption, Socrates is able to prove that both Philebus’s position and his own position are false.

PROTARCHUS: Neither of these lives seems to me worth choosing ... and I think anyone would agree with me.

SOCRATES: What about a joint life, Protarchus, made up of a mixture of both elements?

PROTARCHUS: One of pleasure, thought and intelligence, you mean?

SOCRATES: Yes, and things of that sort.

PROTARCHUS: Anyone would choose that in preference to either of the other two, without exception.

SOCRATES: We are clear what follows for our present argument?

PROTARCHUS: Certainly. There are three possible lives before us, and of two of them neither is adequate or desirable for man or beast.

SOCRATES: Then it’s surely clear that neither of these at least can be the good ... For if any of us chose anything else he would take it in defiance of the nature of what is truly desirable ...4

The argument has exactly the pattern Vlastos found problematic in the early dialogues. An interlocutor answers a question; Socrates gets him to agree to further premisses; the original answer is shown to be inconsistent with the further premisses; it is then agreed by all hands that the negation of the original answer (or answers in the Philebus) has been proven true. The correctness of the conclusion depends on the truth of the unexamined further premisses: it is just this that made Vlastos decide that Plato could not accept the elenchus as leading to truth. Yet here in the Philebus there seems no room for doubt that Plato uses the elenchus, and accepts its results.

Two other features of the early elenchus are also conspicuous in the Philebus. There is, first, the feature on which Vlastos laid great stress: the insistence that the interlocutor sincerely express his own beliefs. One may guess that it is just in order to emphasise this point that Plato substitutes Protarchus for Philebus as the chief interlocutor: Philebus is too stubborn and crude to be counted on to answer honestly. But Socrates repeatedly makes certain that Protarchus is expressing his own views. At one point, Socrates turns to Philebus to ask if he agrees. We get this exchange:

PHILEBUS: My view is, and always will be, that pleasure is the undoubted winner-but it is for you, Protarchus, to decide.

PROTARCHUS: As you have handed the argument over to me it is no longer for you to say whether or not to agree to Socrates’s proposal.

A second striking point is Socrates’s insistence at every point that what he takes to be true, every untested assumption on which further results depend, be agreed upon. As we have just seen, he does not do so well with Philebus. But the agreement on which he depends is the agreement of those engaged in the dialectic. In the passage quoted above, Socrates secures Protarchus’s agreement no less than three times to ‘what anyone would choose’; when the question is asked in the right way, what everyone would choose is the good; and when the question is asked in the right way, everyone will agree that the mixed life of pleasure and intellect is better than the life of pure pleasure or the life of pure intellect. Early in the dialogue, Socrates makes clear that his own position is as much at risk as that of Protarchus. He says: ‘I take it that we are not now just vying to prove my candidate or yours the winner, but shall join forces in favour of whatever is nearest the truth.’ And at another point he remarks: ‘Should we then register our agreement with earlier generations, and instead of just citing other people’s opinions without risk to ourselves, stick our necks out too ...’ Towards the end, Socrates asks once more whether everyone would agree with the essential assumption on which the argument rests, and with the conclusion that the mixed life of pleasure and intelligence is best. He even suggests that ‘if anyone thinks we have overlooked anything ... I hope he will now go back and state the matter more accurately.’ At the very end, Protagoras proclaims: ‘We are now all agreed, Socrates, on the truth of your position.’ This is not, of course, the position with which Socrates began: it is the position reached by following the elenctic argument.

The central argument of one of Plato’s last dialogues concerns, then, a major Socratic problem, employs the Socratic elenchus, and unambiguously endorses the outcome of the elenchus. Why, after entertaining such profound doubts about it, does Plato unquestioningly return to Socrates’s method?

Those who are familiar with the Philebus will have noticed that I have been discussing a very small proportion of the material in that dialogue: the defeat of the pure lives of pleasure and intellect and the victory of the mixed life is assured in the first fifth of the dialogue, and the actual argument uses up barely more than three Stephanus pages. That is the heart of the dialogue: but there is, after all, a great deal more. And much of the rest is remote from anything we find in the early dialogues, or, for that matter, anywhere else in Plato. There is much confusing discussion of methods, and ontological doctrines are put forward that sound vaguely Pythagorean and perhaps something like the doctrines Aristotle attributes to Plato in the Metaphysics. Little of this material can be attributed to Socrates; at one time, scholars were reluctant to attribute it to Plato. We need not be concerned here with how these strange views are to be reconciled with the rest of what we think we know of Plato’s late philosophy, nor with how they can be reconciled with one another. For I think it is reasonably clear nothing in the rest of the Philebus solves Vlastos’s ‘problem of the elenchus’ – that is, shows why it yields truths – and if this is so, the central argument and result of the dialogue depends on nothing but the unadorned Socratic elenchus.

One important argument is not really elenctic (I have already mentioned it). In proving by the elenchus that neither the life of pleasure nor the pure life of the intellect is the good life, Socrates uses, as we have seen, two unexamined assumptions, which he and everyone concerned agrees are true. These are 1. the good is what everyone (‘capable of knowing’) prefers or would choose if he could, and 2. no one would, on reflection (and aided by Socrates’s arguments), choose either the pure life of pleasure or the pure life of the intellect. This completes the elenchus. Socrates then wins agreement to a further claim: 3. everyone would prefer the mixed life of pleasure and intelligence to either of the other two lives. This implies the negative conclusion of the elenchus, and so could have been used in place of 2. It is introduced, however, as an independent conclusion. But nothing here departs from the elenctic method: 3., like 1. and 2., is accepted once agreed to.

The bulk of the dialogue is devoted to a detailed examination of the various kinds of pleasure, with arguments designed to show why most pleasures are unsuited for inclusion in the good life. As a first step, Socrates undertakes to show that although all pleasures are alike in being pleasures, it is possible that they differ in that some are good and some are bad. The demonstration of this elementary point begins with a fairly lengthy, but confusing, description of a method. Protarchus asks Socrates’s help in finding a ‘better way to conduct the argument’. Socrates obliges: ‘There could be no finer way than the one of which I have always been a devotee, though often it has slipped through my fingers and left me empty-handed and bewildered ... It is not difficult to expound, but it is very difficult to apply. It has been responsible for bringing to light everything that has been discovered in the domain of any skill ... it was a gift from the gods to men ...’ The description of the method that follows is open to many interpretations, especially when one tries to square it with the subsequent fourfold division of all things into limits, unlimiteds, mixtures of the two, and the cause or causes of mixtures. But the uses to which Socrates puts this method, or these methods, is easier to understand. The first use is this: if we start with a collection of entities, we must first ‘posit a single form’, then subdivide it into two or three more, until no more organised divisions are called for. This takes skill; it is easy to go wrong. So far the method sounds like the method of collection and division defended and practised in the Sophist and Politicus. And that is exactly the use to which it is put. Socrates divides up the species of pleasure with the ultimate aim of distinguishing the good from the bad pleasures. But the method of collection and division cannot make these distinctions: at best it shows that there is no contradiction involved in saying pleasures are one in being pleasures, but may differ in other respects.

As the method is elaborated, the importance of limits, measures, proportion and symmetry is increasingly stressed. Good things – good lives, for example – are the result of the imposition of a limit or measure on one or more indefinite continua: such are the mixtures, among which the good life is to be found. Most pleasures, though not all, are argued to be unsuited for inclusion in a properly balanced and stable life; most forms of wisdom and even practical skills are worthy of inclusion. Mind plays a dual role: its functions and objects are admirable and desirable in themselves, and mind, being akin to the cosmic cause which accounts for all that is good in the universe, is itself the cause of the measured life.

I call attention to only two aspects of these somewhat tortured passages in the Philebus. The first is that no interpretation of the ‘god-given method’, at least none with which I am familiar, can reasonably be said to provide either a substitute for or a supplement to the elenchus of the kind provided by the assumption of the Gorgias or the theory of recollection. The method of collection and division does not itself provide a guide as to which the true ‘units’ are, nor how to tell when a division has been made ‘at the joints’. But even if it did, no substantive moral (or other) truths could emerge; nor does Socrates suggest that they could. The fourfold classification of ontological features of the world does far more work, for it is clear that both the categorisation of many pleasures as basically ‘unlimited’, and the principles that guide the construction of the good life draw heavily on the necessity of the presence, in all that is desirable, beautiful or stable, of a limit. What the method entirely fails to do is to provide criteria for telling when a mixture is a good one. Sometimes Socrates talks as if every true mixture is good, and every limit a principle that produces a mixture. But this is no help, for we then want to know how to tell a limit from some other arrangement of parts: how to tell a mixture from a mere grab-bag of ingredients.

Plato seems aware of the fact that the ‘god-given’ method gives no clear guidance in these matters. Although he stresses the superiority of the pure sciences, like mathematics, to the applied arts such as flute-playing and building, he compares the choice of ingredients for the good life, and their blending, to the work of a craftsman. When it comes to describing the good life, Socrates says: ‘Well, then, it would be a fair enough image to compare us to builders in this matter of the mixture of intelligence and pleasure, and say we had before us the material from which or with which to build.’ Protagoras: ‘That’s a good comparison.’ Socrates: ‘Then our next business must be to try to mix them?’ One is reminded of the detailed examples drawn from music and phonetics earlier in the Philebus, which illustrate that producing a pleasing or acceptable product depends on more than the analytic methods which discriminate the ingredients: it depends as well on the skill and knowledge of the craftsman. The theme is familiar from the early Socratic dialogues.

The second aspect of the methodology of the Philebus to which I want to call attention is the role, or lack of it, of the theory of forms. Critics have argued endlessly over the question whether any of the four elements in the ‘god-given’ method is to be identified with the forms. From the point of view of the present thesis, it doesn’t matter. For as we have already seen, nothing in that method could, or is claimed to, yield substantive moral truths. Nor is it likely that at this stage in his development Plato would have relied on the theory of forms for this purpose. In the late dialogues he found more and more reasons to be dissatisfied with his earlier doctrines about the forms, and no aspect of this dissatisfaction is more evident than the abandonment of any close connection between the forms and value. The unity of the forms which earlier had ensured their purity was given up when it became evident that analysis required that the forms blend with one another (as pleasure does with good and bad – the critical discovery is made in the Sophist). The idea that value depends in some way on being like or resembling a form was recognised by Plato (in the Parmenides) to be incompatible with the epistemological or semantic functions of the forms. Value in the Philebus can no longer be connected with the forms as such; it may be that limits, or mixtures that have limits, are forms and are good: but what makes them good is not that they are forms or limits or mixtures, but that they have symmetry, commensurability and truth – that they are proper limits or appropriate mixtures.

My proposal, then, is that when, in almost his last dialogue, Plato returned to the question of the nature of the good life, he also returned to the Socratic elenchus as the clearest and most reliable method for discovering how we ought to live. So it seems no surprise to me that in this dialogue the leader is once again Socrates. As we know, Socrates practically disappears from all the other late dialogues. In the Parmenides he is portrayed as very young, and it is the great Parmenides who directs the questioning, Socrates who responds. In the Critias and Timaeus Socrates is present, but makes no serious contribution; he is not present in the Laws. In the Sophist and Politicus, the two dialogues most closely related to the Philebus, Socrates turns over the discussion to the Eleatic Stranger. Only in the Philebus is he once again Plato’s spokesman, and, if I am right, again speaks in his own person so far as basic method is concerned. He is Plato’s Philosopher.

In the Sophist we are apparently promised a trio of dialogues – on the sophist, the politician and the philosopher. The first two exist; there is no record of the third. By the most likely dating, the Philebus was written soon after, about when the Philosopher should have been written.5 For a number of reasons we can see why the Philebus could not be called the Philosopher. But I like to think of it as taking the place of that unwritten dialogue. It is about Plato’s philosopher, it comes back to the problem with which that philosopher was most concerned, and it accepts his way of doing philosophy for its most important results.

If, as I have argued, Plato returns in the Philebus to the confident use of the elenchus, it must be because he decided in the end that Socrates was right to trust that method. Clearly Plato had found nothing better – nothing that he did not come to criticise himself. What explains Plato’s renewed confidence in the elenchus? As Vlastos explains, the elenchus would make for truth simply by ensuring coherence in a set of beliefs if one could assume that in each of us there are always true beliefs inconsistent with the false. It is not necessary that these truths be the same for each of us, nor that we be able to identify them except through the extended use of the elenchus. Thus someone who practises the elenchus can, as Socrates repeatedly did, claim that he does not know what is true; it is enough that he has a method that leads to truth. The only question is whether there is reason to accept the assumption.

I think there is good reason to believe the assumption is true – true enough, anyway, to ensure that when our beliefs are consistent they will in most large matters be true. The argument for this is long, and I have spelled it out as well as I can elsewhere. But the argument hinges on a good Socratic intuition: it is only in the context of frank discussion, communication and mutual exchange that trustworthy truths emerge. The dialectic imposes the constant burden of interpretation on questioner and questioned, and the process of mutual interpretation can go forward only because true agreements which survive the elenchus carry a presumption of truth.

In coming to see that Socrates was right to trust the elenchus to do more than ensure consistency, Plato was returning to a point at which he started. In Ulysses James Joyce quotes (or misquotes) Maeterlinck as saying: ‘If Socrates leave his house today he will find the sage seated on his doorstep.’ The same, I have urged, can be said about Plato; or even about philosophy.