Chapter 2 - The Place of Concepts in Socratic Inquiry

Summary

In ‘The Place of Concepts in Socratic Inquiry’, Terence Irwin examines Socrates’ question ‘What is F?’, which is often taken to be a request for some sort of definition or account of what F is. When Socrates asks, ‘What is courage?’, ‘What is piety?’, ‘What is temperance?’, does his discovery that everyone, including himself, cannot answer such questions in a satisfactory manner imply that these answerers do not know what the words mean? If one cannot answer the ‘What is F?’ question, does it follow that one lacks the concept of F? Irwin argues that conceptual argument has an indispensable role in the arguments that lead to Socratic definitions, but it will not take us all the way to them. To understand Socratic definitions, Irwin compares them with Aristotelian real definitions, and with Epictetus’ views on the articulation of preconceptions.

1 Two Questions About Concepts

Plato’s Laches gives us one example of a familiar pattern of argument in the Socratic dialogues.¹ Socrates and his interlocutors discuss questions that are relevant to bravery. Socrates asks each interlocutor whether he can say what bravery is. The interlocutors try various answers that offer different attempted definitions. Their answers turn out, under Socrates’ questioning, to be unsatisfactory. After a series of answers are offered and found to be unsatisfactory, the participants in the discussion agree that none of them, including Socrates, knows what bravery is. They have been looking for definitions, but they have not been able to find them.²

I intend to ask two questions: (1) Does the argument of these dialogues suggest that Socrates is interested in concepts and that he argues about them? (2) When Socrates tries to find definitions, is he trying to define or to analyse concepts? These two questions are evidently connected. If we answer Yes to the second question, we have good reason to answer Yes to the first question. But if we answer No to the second question, we may still reasonably answer Yes to the first. For even if the definitions that Socrates seeks are not definitions of concepts, conceptual argument may still be useful, even necessary, for finding them.³

These are the answers that I will defend. Once we see what kind of definitions Socrates is looking for, we will find that conceptual argument has an indispensable role in the arguments that lead to Socratic definitions, even though it will not take us all the way to them. To find a Socratic definition, we need to apply both conceptual argument and a type of systematic argument that I will describe later. We can understand the nature of Socratic arguments better if we attend to the interpretations that are offered by Aristotle and by Epictetus. According to both, Socrates is looking for real definitions, as opposed to analyses of concepts, and in order to find them he has to engage in systematic inquiry into virtue, practical reason, and happiness. The fact that some of Socrates’ ancient readers interpret him in this way does not show that they are right. But we will find that their interpretation explains some features of Socratic argument that are difficult to explain on the assumption that he seeks to analyse concepts.

2 Is a Socratic Definition a Means to Grasping a Concept?

First, we need some idea of what concepts are and of what a conceptual argument is like. Questions about the nature of concepts, and indeed about whether there is any such thing as the nature of concepts, are matters of philosophical dispute.⁴ If we had to settle these questions before we could inquire into Socrates’ interest in concepts, we would never get to our main question. But we can perhaps begin with one fairly intelligible use of ‘concept’, and see whether this allows us to ask useful questions about Socrates.

According to one use of ‘concept’, if I grasp the concept of F, I thereby grasp the meaning of the term ‘F’, I understand ‘F’, and I grasp something about what it takes to be F, or about the conditions for being F. I manifest this grasp by my competent use of ‘F’ in appropriate conversational contexts, and by my capacity to recognise Fs in an appropriate range of circumstances. To find out whether I grasp a concept, you might ask me (e.g.) ‘What is a vixen?’. If I grasp the concept, I reply that a vixen is a female fox. I need this grasp of a concept if I am to find out whether a vixen was in the back garden last night.

This approach to concepts seems relevant to Socrates and Plato. Socrates asks, ‘What is bravery?’ (etc.), and invites his interlocutors to answer (e.g., Lach. 190b3–c7). The interlocutors assume that Socrates’ ‘What is it?’ question should be easy, since they are familiar with brave actions and brave people. They are surprised, therefore, to find that their answers are unsuccessful because they do not manage to say what bravery is. Since they think Socrates’ question is easy, we might reasonably suppose it is a question about the meaning of the word ‘bravery’, or about the concept of bravery, as we have described it. Since they seem to talk about bravery all the time, and seem to communicate about it, they suppose that they must grasp the relevant concept. If Socrates’ question asks about the meaning of ‘bravery’, his negative conclusion asserts that they cannot state the meaning of the word. When Socrates says they do not know what bravery is, he means that they do not know what the word ‘bravery’ means, because they cannot give an account of the concept of bravery.

When we say that someone uses a word without knowing what it means, we imply that they use it without any determinate meaning, even though it may have a determinate meaning. If I am not sure what some contemporary (or once contemporary) slang expression means, I may use it to show I keep up to date with current usage, but I may give myself away by using it wrongly, for instance, if I use ‘funky’ to describe something that no one else would call funky. Or if I have some very vague idea of what it means, I may sometimes accidentally use it in the right situation and mislead people into thinking I mean what is usually meant. We might suppose that Socrates believes that this is the predicament of his interlocutors who do not know what bravery is, and therefore do not know what ‘bravery’ and ‘brave’ mean. If neither of us knows what ‘funky’ means, I may say that a three-piece pin-striped suit is funky, and you may agree with me, but neither of us has a clue what we are saying about it. If Socrates believes that this is what Laches and Nicias are doing, we can see why he thinks it urgent to find out what bravery is.

At least one ancient critic takes this to be the point of Socratic requests for definitions, though we have only a very brief statement of his interpretation and objection. According to the orator Theopompus, as reported by Epictetus, Plato looks for a definition because he assumes that anyone who cannot provide it does not mean anything by the relevant term, and therefore cannot communicate significantly by using the term. Failure to provide a Socratic definition of (e.g.) bravery would betray our inability to say or think anything by using the term ‘bravery’. On the assumption that this is Plato’s reason for seeking definitions, Theopompus replies that it is a bad reason. If Plato were right about the point of finding definitions, we would have to say that when we use terms we cannot define, we use them without any meaning. But it is obvious that we can mean something by our use of terms without any definition.⁵ Since we are not in the predicament that Socrates thinks we would be in if we had no definition, Socrates’ search for a definition tries to solve a non-existent problem.

We might take Theopompus’ argument a step further. Suppose that Socrates could find a definition. He could show that it is the correct one only if he could show that it matches our correct judgments about the extension of the relevant concept and about the criteria that determine the extension. But if we can already make these judgments, we already grasp the concept before we have a Socratic definition that is supposed to make it possible to grasp the concept. If we can see that a Socratic definition is correct, we can also see that it is useless.

But is Theopompus right about why Socrates looks for definitions? Socrates does not say so, and the dialogues do not imply that he holds this view. On the contrary, they give us reason to deny that he holds it. For Socrates insists that he, no less than his interlocutors, cannot answer his own questions, and therefore cannot define the objects of his inquiries. If anyone who fails to define F lacks the concept of F and cannot think or communicate about F, Socrates cannot do these things either. But he clearly assumes that he can do these things, because he thinks he can argue, as he does in the Crito, about whether (e.g.) it is just to break the law, and that he can communicate with Crito on this question. Theopompus is wrong, then, about why Socrates looks for definitions. Socrates is not looking for definitions because he thinks competent users of the relevant concepts can produce definitions. His search for definitions does not reflect this sort of interest in concepts.

This is not a decisive refutation of Theopompus’ objection. Theopompus is wrong only if Socrates’ position is consistent. But perhaps it is not consistent. Perhaps the demand that Socrates imposes on his interlocutors rests on an assumption that is inconsistent with his other beliefs. We might even infer that Plato (the author) intends us to see this inconsistency in Socrates (the character), and thereby to see that it is misguided of Socrates (the character) to demand definition.⁶

3 Are Socratic Definitions Conceptual Analyses?

Before we accept any of these conclusions, we ought to consider other possible reasons for being interested in concepts and definitions. Even if we are familiar with a concept and can use it competently, to say appropriate words, think appropriate thoughts, and make appropriate inferences, on the correct occasions, we may not be able to analyse the concept. We may be able to make the appropriate moves, but we may not be able to state the general rules that make these moves appropriate. These rules may be complex and difficult to state.

Neglect of this familiar fact about the analysis of concepts underlies G. E. Moore’s argument to show that ‘good’ is indefinable. In Moore’s view, a proposed definition ‘G’ of a concept ‘F’ is acceptable only if we cannot significantly ask whether Gs are F. This test works all right in the case of ‘vixen’ and ‘female fox’. It seems obvious that if ‘female fox’ is the meaning of ‘vixen’ we cannot significantly ask ‘Are female foxes vixens?’. The question is insignificant because we know it would be self-contradictory to suppose that a female fox is not a vixen, and therefore the proposed question cannot arise. According to Moore, no correct definition leaves an ‘open question’ of the form ‘Are Gs F?’. Since all proposed definitions of ‘good’ leave an open question (according to Moore), they must all be rejected, and ‘good’ is indefinable.

As critics of Moore have pointed out, this argument against definitions that leave open questions ignores the possibility that some concepts may be less easy to articulate than ‘vixen’.⁷ If we do not know the right account of ‘F’, we do not know whether the question ‘Are Gs F’? is significant or not. If an open question is one in which the question appears significant, open questions are not a good test for definitions, since some questions that appear significant are not really significant. If, alternatively, open questions are really significant, we cannot tell whether a question is open until we have a correct definition of the concept in question, and therefore we cannot appeal to open questions to tell us whether a proposed definition is correct.

Moore’s mistake results from inattention to the possible interest and difficulty of finding accounts of some concepts. The gap between the use and the analysis of concepts explains the title of philosophical works such as The Concept of Mind, The Concept of Law, and The Concept of a Person.⁸ In the first of these works Gilbert Ryle argues against the belief in internal mental events and states, but not by arguing that there are no such things. He argues about the concept of mind because he argues that when we speak of thinking or believing or enjoying, we do not mean that different sorts of internal mental events are happening. However elaborate and controversial Ryle’s account of the concept of mind may be, it is intended to tell us about what we mean when we use mental vocabulary, not about the facts in the world.

These few examples suggest why Socrates might regard his search for definitions as an inquiry into concepts. Such an inquiry may reveal features of our concepts that are not obvious to competent users when they are asked to say what they mean.⁹ If we are asked how knowing differs from believing, our first answer may be that to know something is to be sure of it. Further questions may be needed to convince us that we cannot know what is false, and that if we believe something for no reason whatever, we do not know it. The questions that might make it clear to us what we mean when we say that we know something are quite similar to the sorts of questions that we encounter in Socrates’ discussions with his interlocutors. This is some reason to believe that Socrates’ inquiries are about concepts. If this is what he is doing, he is not open to Theopompus’ objections. The fact that we can use a concept competently is no reason to believe that a Socratic definition will be a waste of time if it provides a conceptual analysis.

This is not the only sort of inquiry, however, that might aim to say (e.g.) what a given virtue is. We can find examples in twentieth-century philosophy that illustrate a different sort of inquiry. After the mid-1960s, we find fewer books and papers that claim to be about the concept of something. John Rawls’s book is A Theory of Justice, not The Concept of Justice.¹⁰ Though Rawls has something to say about the concept of justice, most of his discussion is about what justice really is – it is about the justice of which the concept of justice is the concept. In contrast to those who discuss concepts, Rawls does not stop short of offering a substantive account of justice. We need not pause to ask whether the theories about meaning, necessity, and the a priori that have been associated with concepts and that have been used to distinguish ‘conceptual’ from ‘substantive’ inquiries are tenable. We can at least form an approximate idea of the difference, real or intended, between inquiry into the concept of (say) justice and inquiry into justice.

Rawls discusses the concept of justice in order to discover some truths about justice itself. Others examine a specific concept to discover that it refers to nothing. John Mackie, for instance, argues that, according to our moral concepts, moral facts are both objective and prescriptive.¹¹ But we know, in his view, that there are no such facts. Hence, we discover that there are no moral facts. We would reach a nihilist conclusion about minds if we argued that our concept of a mental state makes them non-physical causes of physical behaviour, but there are no such causes of physical behaviour. Both constructive arguments, such as Rawls offers, and destructive arguments, such as Mackie offers, make it clear that we may be interested in the concept of F even if our aim is to learn something about F, and not simply about the concept.

If, then, we were looking for new subtitles for some of the Socratic dialogues, should we call them ‘The Concept of Bravery’, ‘The Concept of Virtue’, and so on, in 1950s style? Or should we prefer ‘A Theory of Bravery’ and so on, in 1970s style? To ask these questions is not to assume that Socrates has a clear conception of each type of inquiry or a clear idea of what he is doing. Moreover, the two kinds of subtitles I have suggested may not exhaust the possibilities. But the questions I have asked may help us to grasp at least his implicit aims.

4 Aristotle: Socrates Looks for Real Definitions

The distinction between conceptual and substantive inquiries is not merely a modern distinction. Aristotle recognises something like this distinction and applies it to the interpretation of Socrates. In his view, Socrates is primarily interested in theories rather than in concepts, because he ‘was seeking the universal’ (Arist., Metaph. A.6 987b1–4). Aristotle claims that Socrates wanted to construct a body of scientific knowledge (epistēmē) about ethics, and that this was why he wanted definitions.¹²

This claim implies that Socrates’ inquiries are not purely conceptual. When Aristotle says that definitions provide the principles of scientific knowledge, he has a specific sort of definition in mind. He distinguishes nominal from real definitions. A nominal definition tells us what we take the name ‘F’ to signify at the beginning of inquiry. Let us say that it describes the nominal essence that corresponds to the name.¹³ If our inquiry succeeds, we find a real definition that tells us what the real essence of F turns out to be. If we find the real essence of F, we can explain the features that we initially took ‘F’ to signify (Arist., An. post. 2.10 93b29–94a10). The real essence of F is the non-linguistic universal that the study of F seeks to discover.¹⁴

Aristotle, therefore, implies that the universal that Socrates was trying to define is the real essence, and that the definitions Socrates was looking for are real definitions; for these, in contrast to nominal definitions, are the basis of scientific knowledge. The relation between nominal definitions and conceptual analysis is by no means straightforward, but it is close enough to imply that Aristotle denies that Socrates’ primary aim is conceptual analyses. These would not be appropriate principles for scientific arguments.¹⁵

A real definition describes not the nominal essence that the competent speaker and hearer grasp, but the ‘real essence’ they refer to.¹⁶ If we associate ‘gold’ with the nominal essence of being a shiny yellow metal, we may nonetheless discover that some things satisfying this description are really bits of ‘fool’s gold’ (iron pyrites). In this case the natural kind with the inner constitution of gold does not include all the examples that satisfied the initial description.

This discovery of the difference between the nominal and real essence depends on facts about the nominal essence. The hypothesis I just considered – that the nominal essence belonging to ‘gold’ is simply being a shiny yellow metal – must be wrong, if we are to discover a real essence. We must also assume that ‘gold’ refers to a natural kind of metal. If we had a word that was intended simply to collect particular superficial characteristics (of metal, plastic, etc.) without reference to any underlying constitution, there would be no real essence to be discovered. Not all nominal essences, then, leave room for a real essence. Some nominal essences identify the relevant kinds as purely conventional kinds. There is no real essence of the fashionable that underlies the manifest features of fashionable things. Fashionable things have a nominal essence, but no distinct real essence. Similarly, saucepans and frying pans are conventionally used for different purposes, but there is no fact about the real essence of the saucepan that determines whether a particular pan is a deep frying pan or a shallow saucepan. If it is generally used for boiling and steaming rather than for frying, someone who thinks it is a shallow saucepan is not making any mistake about its real essence.

We believe that ‘gold’ is not such a word, because we regard it as a name of a natural kind. When we see a discrepancy between the real essence and the elements of the nominal essence, we allow the real essence to determine the extension of a term that we take to name a natural kind. If we are looking for the real essence of F, we must believe that F has, or may have, a real essence. If we believe that F may have a real essence, we must rely on some assumptions about the nominal essence of F. If assumptions about the nominal essence of F are assumptions about the concept of F, we must rely on these assumptions.

If Aristotle’s account of Socrates’ search for definitions is correct, the definitions that Socrates looks for are not nominal definitions, and he is not trying to define concepts. But, for the reason I have just given, he ought to be interested in concepts. If he is looking for real definitions, he relies on assumptions about the concepts that introduce the properties he inquires into.¹⁷

5 Epictetus: Socrates Seeks to Articulate Preconceptions

Aristotle’s interpretation of Socratic inquiry is worth comparing with Epictetus’ interpretation. Epictetus deserves attention because he often appeals to Socrates both for a guide for life and for a pattern of philosophical method and argument.¹⁸ In his view, Socrates’ search for definitions is not open to Theopompus’ objection, because Theopompus does not distinguish the use of preconceptions from the articulation of them, and therefore he misunderstands Socrates’ questions.¹⁹ We can speak significantly and communicate about the good and the just because we share thoughts (ennoiai) and preconceptions (prolēpseis). Socrates agrees that we can do this, but he still thinks his inquiries are necessary. We need Socratic definitions because our preconceptions are not yet articulated and complete. Similarly, we have a preconception about the healthy, but that does not make the work of Hippocrates unnecessary. Once we have an articulated and complete conception, we can reach reasonable judgments about the different treatments that people prescribe. These judgments result from ‘applying’ (or ‘fitting’, epharmozein) our preconceptions to the particulars.²⁰

To achieve this result, we must recognise that one preconception does not conflict with another. If we believe that both p and q are preconceptions and that they conflict, then either (1) at least one of them is not really a preconception, or (2) we have not articulated them completely, and we are wrong to suppose that they conflict. We can avoid these mistakes if we persist with the Socratic search for definitions.²¹ But if we try to apply our preconceptions without articulation, we falsely suppose that they conflict.

At first sight, Epictetus’ description of Socratic inquiry seems quite different from Aristotle’s description. (1) He says nothing about Socrates’ desire to engage in deductive reasoning from basic principles. (2) He implies that Socratic definitions will help us to apply our preconceptions to particular cases, but Aristotle implies that Socrates is interested in a science (epistēmē), and therefore in universals rather than particulars. (3) Epictetus explains Socrates’ inquiries by reference to preconceptions, which Aristotle does not mention.

To see how significant these differences between Aristotle and Epictetus might be, we can look more closely at Epictetus’ view of preconceptions. He does not use ‘preconception’ for just anything we might believe or assume about the object of our inquiry. Some preconceptions are basic principles of crafts or professions.²² We violate the preconception of a craft if our actions are so contrary to the aim and principles of the craft that we do not really practise it at all. Ethical preconceptions are so basic in our ethical thinking that they constitute the principles in the light of which we assess our other ethical beliefs. We argue from preconceptions in so far as we argue from the principles that define what ethics is trying to achieve.

This use of ‘preconception’ clarifies the Stoics’ attitude to ethical preconceptions. They are not just appealing to widely shared beliefs. They maintain that ethics is a systematic discipline with an aim in the light of which we can correct various claims about what ethics requires. We can tell whether a carpenter is right to say a hammer should be made of iron rather than wood if we consider what hammers are properly used to do in carpentry. What, then, would be the analogue in ethics?

It would be uncontroversial to say that ethics is about the human good, and about the relations of one human being to others. The Stoics also claim that it is about the good for human beings as rational beings. The question is about the relevant aspects of rationality. Epictetus’ answer to that question appeals to the special constitution of the human mind. We ought to think about what is especially appropriate to a rational agent who is capable of seeing the connections in things.²³

In the Stoics’ view, then, the apparent paradoxes of their position result from adherence to preconceptions. Sometimes Stoic views seem to violate common sense, and the contrary view may seem persuasive. But the Stoics answer that their views are really closer to common sense, properly understood, than rival theories are. If some other beliefs conflict with preconceptions, we ought, in the Stoic view, to give up the other beliefs rather than the preconceptions. The preconceptions serve as rules or ‘standards’ (kanones) for settling disagreements.²⁴

If this is the role of preconceptions, the difference between Epictetus and Aristotle on Socratic definition is less sharp than it initially seemed to be. Each of them describes Socratic definition in his own way, but both believe that it is intended to support a systematic discipline. An Aristotelian science is systematic because it presents the principles and derived propositions that apply to a specific area, in their proper logical and epistemological order. A science of this sort provides the rules and standards that Epictetus finds in preconceptions.

Epictetus agrees with Aristotle’s belief that Socrates is looking for real definitions, and that he tries to discover them by beginning with concepts. According to Epictetus, inarticulate preconceptions are sufficient for saying something significant and communicating it to others. But the conditions for significant speech and communication do not ensure the correct application of our preconceptions to specific circumstances.

Epictetus acknowledges that preconceptions may appear to conflict, but he believes the appearance is misleading; it results if we fail to distinguish the real content of a preconception from the hasty conventional assumptions that we rely on in applying preconceptions to particular situations. If, however, we restrain hasty conventional assumptions, and we apply preconceptions systematically, we find, in Epictetus’ view, that their application to specific types of situations undermines our unreflective conventional judgments about these situations.

We need to apply preconceptions systematically because consideration of one preconception at a time may not remove conflicts between different apparent preconceptions, and therefore will not tell us what the real preconceptions are. It follows that we cannot, for instance, complete an inquiry into one virtue at a time. If we think bravery is a virtue, we need to consider not only our beliefs about bravery, but also our beliefs about virtue. These beliefs may conflict, but we will not know how to resolve the conflict unless we consider our other ethical beliefs. We need to consider them all if we are to articulate our apparent preconceptions, and thereby to uncover the genuine preconceptions.

6 Socrates Relies on Concepts

I have discussed Aristotle and Epictetus on Socrates because they present an interpretation of Socratic inquiry and definition that is worth discussing. If they are right, Socrates should be interested in concepts and in nominal essences as a means to discover real essences, but his primary interest is in real essences.

Conceptual arguments rely on judgments about contradiction, incompatibility, and logical impossibility. If we suppose – or claim to suppose – that bravery is the virtue that shows itself in writing neatly and legibly and expressing oneself in few words, we have not discovered an interesting fact about bravery. Nor have we made a false claim that needs to be refuted by examination of different brave people and their handwriting. If we recognise that this is the wrong sort of refutation, we recognise the kind of argument that is needed to show that something follows from, or is excluded by, a specific concept.

In this sort of argument, the source of the contradiction is not one’s other beliefs as a whole, but rather specific beliefs about the relevant concept. To affirm that a vixen is not a fox is to contradict oneself implicitly, because substitution of the meaning of ‘vixen’ (i.e., ‘female fox’) for ‘vixen’ creates an explicit contradiction (‘a female fox is not a fox’). Similarly, to say that A is a faster runner than B, but A is not capable of running a given distance in less time than B takes, is self-contradictory. Similarly, we might argue that it would be self-contradictory to assert that A is brave, but A has no virtue, or to assert that A is brave, but A cannot face the smallest danger without being terrified. Someone who asserts these things does not seem to grasp the concept of bravery.

This feature of conceptual arguments gives us an idea, however imprecise, of what to look for in Socrates’ arguments. If he asks what follows from the agreed features of a concept, or what they exclude, he relies on conceptual argument. On this basis we can find some indications of the features that Socrates attributes to his interlocutor’s grasp of a given concept. The inquiry in the Laches begins from the assumption that virtue is such that the soul of anyone who acquires it becomes better (Plat., Lach. 190b3–c2). If we said that someone has acquired a virtue, and has thereby become worse in every respect, we would not be inquiring into a virtue. Socrates proposes a restriction of the discussion to one part of virtue. He takes it to be obvious that virtue has parts (Plat., Lach. 190c8–d1). He does not say what follows if virtue has parts, but he assumes that there are several virtues, and that it will be easier if they discuss just one of them. He takes the plurality of the virtues to follow from the belief in parts.

Socrates continues to list assumptions that he takes to be undisputed, by suggesting that they discuss the part of virtue that bears on fighting in armour. The many, he says, think that the relevant part is bravery (Plat., Lach. 190d3–6). It is not immediately obvious why this is so. They have been talking about professionals who offer specialised training in fighting in armour. For these purposes, we might think, the relevant qualities are (say) strength and agility. But Socrates relies on our recognising that we care about fighting in armour not for its own sake (as we might be interested in fencing without supposing it is any use in twenty-first-century combat), but because it is a way of fighting battles in dangerous situations. The virtue that we need to face the relevant dangers is bravery. That is why the discussion soon moves on to facing dangers and to standing firm in frightening situations. Socrates expects that we will recognise that bravery is the virtue that faces danger in the right way. We could make a similar list of assumptions about the other virtues that Socrates treats as uncontroversial. An interlocutor who did not assent to such assumptions would not be a suitable inquirer into a virtue.

7 Socrates Seeks Real, not Nominal, Definitions

Some of Socrates’ remarks about definitions, however, suggest that he is looking for real, not nominal, definitions.²⁵ He asks what F is, and what all Fs have in common, not about what the word ‘F’ means. He wants to be told ‘that very form by which all the piouses are pious’ (Plat., Euthphr. 6d10), or something that ‘is the same as itself in every <pious> action’ (Plat., Euthphr. 5d1). Similarly, he asks for the single power that is common to all cases of bravery (Plat., Lach. 191e9–192b8). The explanatory function that he attributes to the one F ‘by which’ and ‘because of which’ all Fs are F is fulfilled by a real essence, but not by a nominal essence.

Socrates does not say that he is looking for the principles of a science from which one can argue deductively. He claims neither that he argues demonstratively nor that his eventual aim is to do this. Still, some remarks tend to support Aristotle. These remarks raise large questions about Socrates’ views on knowledge and virtue, which we can mention only briefly. It will be enough to show that they make Aristotle’s claims intelligible.

Sometimes Socrates appears to identify one or another virtue with knowledge. Aristotle takes these passages seriously, since he criticises Socrates for supposing that theoretical knowledge is all that we need for virtue of character (Eth. Eud. 1216b2–16). The end of the Laches seems to identify bravery with knowledge (Plat., Lach. 199d3–e4). The Charmides seems to say the same about temperance (Plat., Chrm. 174b11–d7). The Protagoras seems to identify all the virtues with knowledge of good and evil (Plat., Prt. 361b7–c5). The Euthydemus implies that all we need for happiness is the royal craft, which is the knowledge of good and evil (Plat., Euthyd. 282a1–7, 291b1–c2). The Protagoras claims that we need the measuring craft if we are to free ourselves from the power of appearance (Plat., Prt. 357a5–b5). All these conclusions encounter some objection that prevents Socrates from endorsing them unreservedly at the end of these dialogues, but the fact that they often emerge from the argument makes it reasonable to believe that Socrates seeks to achieve knowledge.

It is reasonable, therefore, for Aristotle to suppose that Socrates is looking for some sort of science. On this basis it is reasonable to assume that the definitions that Socrates is looking for might belong to the science he is looking for. According to Aristotle, the definitions that provide principles for a science are those that state the relevant essence. The essential properties are those that explain the other properties of the thing or event in question. If, then, Aristotle is right to believe that Socrates is looking for essential properties, Socrates should be looking for explanatory properties.

Though Socrates’ starting points are undisputed assumptions, his conclusions seem to everyone, including himself, to be controversial. When he describes some of the conclusions he has come to through his cross-examinations, he mentions his view that living finely, living justly, and living well are the same (Plat., Cri. 48b3–10). Only a few people, he says, agree with his conclusions (Plat., Cri. 49c10–d5). In the Protagoras he implies that the denial of incontinence will seem incredible to the many (Plat., Prt. 352a8–c7). In the Gorgias Polus comments that even a child could refute Socrates’ views about virtue and happiness (Plat., Grg. 470c1–8).

If we compare Socrates’ undisputed starting points with his paradoxical conclusions, we might try to use this comparison to identify the extent of his reliance on concepts. We might suggest that his undisputed starting point relies on the concept of bravery (say) and that his paradoxical conclusion (e.g., that bravery is knowledge of good and evil, and is therefore virtue as a whole) relies on some non-conceptual argument. If that is so, not all of his argument is about the concept of bravery.

This argument, however, does not show that Socrates’ conclusions are not conceptual. The mere fact that the conclusion is paradoxical and would not be accepted by most of those who grasp the relevant concepts, does not show that it is not reached by conceptual argument. As we saw earlier in our discussion of open questions, giving an account of a concept may be difficult, and the account we reach may be in some respects counterintuitive.

Perhaps, however, we can find a better argument to show that Socrates’ paradoxical conclusions are not conceptual, if we recall Epictetus’ claim that preconceptions do not conflict. This claim requires us to examine all the virtues (e.g.), because we cannot reasonably be confident that we have found a preconception that provides an appropriate standard for bravery until we have compared our apparent preconception with apparent preconceptions about the other virtues. If the comparison of apparent preconceptions shows us that one of the initial beliefs that fixed our concept of bravery is false, our inquiry is no longer conceptual. If we can find this sort of systematic argument in Socrates, we have some reason to believe that his inquiries are not merely conceptual but aim to provide standards for virtue and virtuous action.

We have some reason to suppose that Socrates engages in systematic inquiry if we consider the beginning and end of the Laches. We noticed that Socrates identifies bravery as the object of inquiry by saying that they should not examine virtue as a whole but confine themselves to one part of virtue. They agree that the part that is relevant to their present concerns is bravery, because it is the specific virtue that deals with facing dangers in battle. The later stages of the dialogue, however, argue that bravery is inseparable from knowledge of goods and evils as a whole, so that it has to be identified with the whole of virtue. In that case the initial assumption that it is a proper part of virtue is false.

We might suspect that Socrates does not accept his conclusion, since he points out that it conflicts with the undisputed initial assumption that bravery is only a part of virtue. This initial assumption is part of the concept of bravery; Socrates relies on it to fix the object of inquiry. If (the argument continues) Socrates accepted the conclusion that bravery is the whole of virtue, he would be rejecting one of the assumptions that belong to the concept of bravery. He would therefore be saying that bravery, as we conceive it, does not exist. He would therefore be presenting an eliminative argument about bravery and about every other virtue that we take to be a proper part of virtue as a whole. But he never says that there is no bravery, no temperance, and no justice. Hence, he cannot really endorse the apparent conclusion that bravery is the whole of virtue.

This argument is worth considering because it suggests a route that Socrates does not take. We noticed earlier that some philosophers who take conceptual analysis to be an important part of philosophical argument use it to draw nihilist conclusions. If, for instance, we find that common sense accepts some assumptions that imply incompatibilism about determinism and freedom, and we think determinism is true, we might conclude that there is no freedom, according to our common concept of freedom.²⁶ If other people defend a compatibilist account of freedom, we may protest that they are introducing a different concept of freedom, and hence are not really disagreeing with the nihilist conclusion that was drawn from our common concept of freedom. Similarly, if Socrates denies that bravery is a proper part of virtue as a whole, he denies that there is any such thing as bravery, as it is commonly conceived.

Socrates recognises the possibility of eliminative arguments. Indeed, the Protagoras offers such an argument against the reality of incontinence. Socrates argues from the concept of incontinence to the conclusion that there is no incontinence (Plat., Prt. 353c–356c). When he concludes that lack of self-mastery is simply ignorance (Plat., Prt. 357c6–d3), he is not affirming the reality of incontinence and saying what it is; he is denying that there is any such thing. About incontinence, then, his attitude is eliminative. But he does not seem to take an eliminative attitude to the virtues that are said to be proper parts of the whole of virtue. Should we, then, infer that he does not take bravery to be the whole of virtue?

We will not draw this conclusion if we pay attention to Epictetus. The assumption that bravery is only a part of virtue is part of our inarticulate preconception, which may conflict with other inarticulate preconceptions. But the articulated preconception implies that bravery is the whole of virtue. We cannot articulate our preconceptions without examining the other virtues as well; for if we follow the rule about the non-conflict of preconceptions, we have to examine our preconceptions about virtue as a whole in order to remove conflicts between our apparent preconceptions. Once we consider all the relevant preconceptions, we find that bravery appears to be the proper part of virtue that is about some specific dangers, but we do not endorse this appearance.

Socrates’ attitude to apparent preconceptions is illustrated by the discussion with Thrasymachus in Republic 1²⁷. In response to Socrates’ questions Thrasymachus denies that justice is a virtue (Plat., Resp. 348e1–4), even though this is one of the features that might reasonably be taken to belong to concept of justice. No argument is needed to show that justice belongs on the normal list of virtues (see, e.g., Plat., Euthyd. 279b4–c1). In Epictetus’ terms, it is an apparent preconception that justice is a virtue. But Thrasymachus uses the Socratic principle that a genuine virtue must be beneficial for the agent, and therefore he denies that justice is a virtue. An alternative response argues that justice is beneficial for the agent, and that therefore it cannot require actions that benefit others. This is the response that Callicles endorses in the Gorgias (Plat., Grg. 483b4–484b1). A third response argues that the actions that Thrasymachus takes to be harmful to the just agent are actually beneficial. This is Socrates’ response in the rest of the Republic. None of these responses is eliminative, since they all assert that there is such a thing as justice. A decision between them requires a decision about the content of the genuine preconceptions about justice.

According to Epictetus, we would be wrong to insist, in advance of inquiry, that any one set of presumed truths about an ostensible presumed virtue or justice, belongs to the concept. If Socrates and his interlocutors share a concept of the ostensible virtue, they agree, over a range of examples, about the extension of (e.g.) ‘bravery’, and about the sorts of traits that a brave person should display. But these initial agreements do not fix the contours of the concept in such a way as to require Socrates to maintain an eliminative view of bravery, as commonly conceived. The appropriate comparison and articulation of apparent preconceptions, and especially of apparently conflicting preconceptions, is needed for us to decide whether Socrates and his interlocutors are really talking about the same virtue.

The Laches, then, suggests the possibility that Socrates follows Epictetus’ rules for systematic inquiry. But should we take this possibility seriously? A reason for rejecting it might be the character of the Socratic dialogues. Some of them deal with one virtue at a time and reach an aporetic conclusion about it. If Socrates had really intended a systematic inquiry, would he not have discussed several virtues at a time, and would he not have marked his positive conclusions more clearly? He does not endorse the conclusion that bravery is knowledge of goods and evils as a whole. He simply notes that it is inconsistent with the initial assumption that bravery is just one part of virtue. We might even say that he reaches the aporetic conclusion because some of the arguments have led us away from the concept of bravery that we began with.

If we consider the shorter ‘dialogues of definition’ – those in which Socrates asks his interlocutors ‘What is it?’ about a specific virtue – it is difficult to answer our questions, because it is not clear why he expects the interlocutor to give the answer that Socrates wants. We may reasonably wonder, for instance, how Socrates manages to drop any reference to any non-cognitive component of bravery and temperance, and how he can dismiss the objection that since bravery is only about certain kinds of danger it cannot be identical to virtue as a whole.

A reason for raising questions at these points in Socrates’ arguments is the fact that the answers he elicits from his interlocutors fall short of the claims that he relies on. Interlocutors agree that a particular virtue is fine (kalon), and that if it is fine, it is beneficial (Lach. 192c4–d8). But this point of agreement does not settle who benefits from it. Socrates assumes that the virtue must benefit the agent. If we are studying an individual dialogue, we might reasonably accuse Socrates of making an unjustified leap from what the interlocutor has accepted to what Socrates needs for his argument.

Socrates, however, does not simply introduce unsupported assumptions into his argument. If we are puzzled by the movement of the argument in one dialogue, we can resolve our puzzle by reference to another dialogue. In the cases I have mentioned, arguments in the Protagoras, Gorgias, and Euthydemus explain why Socrates believes that virtues have no non-cognitive elements, and that virtues are beneficial to virtuous agents. His arguments are implicitly systematic, in so far as he sets out in different places the different elements of the theory that makes the argument of each dialogue more intelligible than it would otherwise be.

8 Socrates’ Conclusions do not Rest on Conceptual Arguments

Does this systematic aspect of Socrates’ argument help us to decide whether, and at what points, he relies on conceptual arguments? Let us consider how he argues that a virtue must be beneficial to the agent. According to the Euthydemus (278e–282e), we all want happiness, and whatever else we want, we want for the sake of happiness. It follows that if we want the virtues, we want them for the sake of our happiness. Moreover, happiness is not simply the end of what we want; it is also the end of what is worth wanting, so that if anything is worth wanting, it is worth wanting for the sake of happiness. Since we agree that each virtue is worth wanting, we ought to agree that it is worth wanting for the sake of our happiness and is therefore beneficial to us.

This argument is not about the analysis of concepts. If we are trying to describe the abilities and assumptions that underlie the competent use of ‘brave’, we have no reason to include the assumption that it is always good for brave people to be brave. We might agree on the sorts of actions that are to be expected of brave agents. We might even agree that they are often good for the agent; we are often better off if we resist some danger or hazard for the sake of some greater gain (e.g., we may be in a less dangerous position if we stick together and do not run away from the first attack). But we may still reasonably disagree on whether they are always good for the brave agents; we may doubt, for instance, whether brave actions that cause your death are none the less good for you. We generally agree about the extension of ‘brave’, and about some generalisations that determine the extension; people show their bravery in resisting dangers because bravery is about resisting dangers and not about running away from them out of fright. This general agreement about the extension and the generalisations that determine it indicates our sharing the same concept of bravery.

We might argue that Socrates insists that all the elements in his account of bravery are also elements in the concept of bravery, if we could find him saying that anyone who rejects his account is not really talking about bravery at all, or that their beliefs are about some other condition than bravery. But he does not say anything like this. If we consider the Socratic dialogues together, we have reason to conclude that Socrates engages in systematic inquiry, and that he follows Epictetus’ rule that preconceptions do not conflict. Since Socratic definitions result from systematic inquiry that follows Epictetus’ rule, they are not meant to be nominal definitions that express our grasp of the relevant concepts.

9 The Republic Looks for Real Definitions, Through Systematic Inquiry

To reach this conclusion, we need to examine different Socratic dialogues and draw some cumulative conclusions from them. These conclusions are partly confirmed by the Republic. In Books 1–2 Socrates and his interlocutors agree about the concept of justice, since they largely agree about the extension and about the principles that determine the extension. They disagree (even if only for the sake of argument) about whether justice promotes happiness. To resolve this dispute Plato considers not only justice, but also the other cardinal virtues, the questions in moral psychology that are discussed in the Protagoras, and the features of the human good that support the view that the just person is better off than the unjust. The eventual account of justice comes at the end of this systematic inquiry that follows Epictetus’ rule.²⁸

If the Republic combines in one systematic argument the elements that are dispersed in different Socratic dialogues, we have some reason to believe that Socrates also argues systematically. The different threads that we have identified in the Socratic dialogues are combined into a single argument in a single dialogue. Republic 4 reaches a definition of justice that would provide an implausible nominal definition, but it is a much better candidate for being a real definition. This also shows us the sort of definition that Socrates aims at.

It would be unwise to use the Republic as the sole basis for the claim that Socrates argues systematically, following Epictetus’ rules, in the Socratic dialogues. But it is reasonable to turn to the Republic to confirm that Socrates argues in the way we have described. Both in the Republic and in the Socratic dialogues, Plato’s attempts to define the virtues would be open to severe criticisms if they were offered as accounts of the relevant concepts, but they are at least not open the same criticisms if they are understood as attempts at real definitions. Plato in the Republic disagrees with Socrates (and with Epictetus) about how the apparent preconceptions should be revised to reveal our real preconceptions about the virtues, but he is still trying to answer Socratic questions.