Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE ( © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

date: 14 October 2019

Agency in Plato’s Republic

Abstract and Keywords

This article discusses some of the most important recent controversies in the psychology of Plato’s Republic. These include its views on akratic action, the capacities of the parts of the soul, and the distinction between the rational part of the soul and the nonrational parts. It argues that the Republic accepts the possibility of synchronic akratic action, that is, action contrary to the agent’s belief about what is overall best at the time of action. It then considers some recent arguments that the lower parts of the soul, especially the Appetitive part, are cognitively primitive. Against these views, this article argues that the Appetitive part is capable of means-end reasoning and of forming a conception of its own good. Finally, this article argues that Plato’s distinction between the rational and the nonrational parts of the soul is to be understood in terms of the intelligible versus sensible distinction.

Keywords: akrasia, akratic action, ancient psychology, intelligible, means-end reasoning, parts of the soul, Plato’s Republic, rationality, sensible

Starting around the beginning of the twenty-first century, there has been a resurgence of interest in the Republic’s psychological theory. In this article, I focus on those controversies that are most important for understanding it. These include Plato’s views of akratic action, the capacities of the parts of the soul, and the nature of the distinction between the rational part of the soul and the nonrational parts.

Akratic Action in the Republic

It has long been widely thought that Plato rejects the possibility of strict akratic action in the Protagoras but accepts it in the Republic. Recently, however, some scholars have argued that the Republic does not, after all, accept strict akratic action.1 To fix ideas, suppose that S knows or believes that she can do either X or Y (but not both) and that X is better for her overall than any other option. If S, deliberately, without changing her mind about the options’ goodness, and without compulsion, does (or tries to do) Y, this would be a case of strict and synchronic akratic action.2 This is distinct from, for example, diachronic akratic action (one variety of weak akratic action).3 In diachronic akratic action, the agent at t1 believes that X is better for her overall than any other option, but at some time between t1 and t2 irrationally changes (say under the causal influence of a desire) her belief about what is best for her overall so that she comes to believe that Y is best for her overall. At t2, while retaining her belief that Y is best for her overall, she does Y. Typically at some later time, t3, S reverts to her original judgment that X was best for her overall and regrets doing Y. I call this a case of weak akratic action, because S does not act contrary to the judgment of what is overall best for her that she possesses at the time that she acts. Other varieties of weak akratic action include, for example, impetuous akrasia: S, under the influence of a desire or emotion, does Y before forming a belief about what is overall best for her, although if she had rationally formed such a judgment, she would have judged that X was best for her overall.

So does the Republic recognize the possibility of strict akratic action? Gabriela Carone, followed by Joshua Wilburn with qualifications, answers negatively. On her view, what scholars have taken to be strict akratic action turns out merely to be weak akratic action. Paradigmatically what happens is that powerful desires or emotions cause irrational changes in a person’s judgment of what is overall best for her so that at the time of action she acts in accordance with a new judgment of what is overall best for her.

So what evidence is there? Before turning to the Republic itself, we should note that Aristotle tells us that Socrates denied the possibility of akratic action contrary to one’s knowledge of, or belief about, the good.4 Given Aristotle’s interest in akrasia and in the differences between Socrates’ and Plato’s views about nonrational motivations, it is reasonable to expect that if Plato also held a view that is so “at odds with what patently appears to be the case” (Nicomachean Ethics 1145b27-8), Aristotle would have mentioned it.5 That he does not is some evidence that Plato did not share Socrates’ view.

The Republic’s strongest specific evidence for strict akratic action comes in the Leontius passage. Leontius, passing the public executioner, found himself with a desire to look at the corpses there.

He had a desire to look at them but at the same time he was disgusted and turned away. For a time he fought with himself and covered his face, but, finally, overpowered by the desire [kratoumenos … tês epithumias], he pushed his eyes wide open and rushed towards the corpses saying, “Look for yourselves, you unhappy ones [kakodaimones], take your fill of the fine sight!”

(Rep. 439E8-440A4)6

This case is intended to show that the Appetitive and the Spirited parts of the soul are different, so it is reasonable to take Leontius’ disgust and turning away as manifestations of his Spirited part. On the traditional interpretation, the Reasoning part judges that it is best overall not to look at the corpses both before and at the time of action. But because Plato is not explicit about the Reasoning part’s role here, is it possible that Leontius never forms or does not have when acting an overall best judgment? First, the grammatical subject of “overpowered by desire” is the person, but it is common for Plato to allow the Reasoning part to count as the person, for example, in the deviant persons’ choice of lives in Books 8 and 9. We should thus expect that the Reasoning part (along with the Spirited part) is being overpowered and thus that prior to this the Reasoning part must have judged that it was overall best not to look. It is still open to Carone to say that such overpowering consists not in getting the person to act contrary to the person’s overall judgment, but in getting the Reasoning part to change its judgment. Second, even leaving the first point aside, there has been no previous discussion of cases in which the person acts without a judgment by the Reasoning part nor are they explicitly discussed in the rest of the Republic. Moreover, the paradigm of such cases of acting without an overall judgment is impetuous akrasia and Plato here stresses the length of Leontius’ struggle. So it is unlikely that Leontius did not form an overall judgment.

Is it possible, then, that the Reasoning part judges, throughout the episode, that it is best overall to look? In addition to the first point about being overpowered by desire, there are two other reasons to reject this possibility. First, the word used to condemn the eyes, kakodaimôn, is rare in Plato and is the literal opposite of being happy, eudaimôn.7 Whether the spoken judgment is the direct expression of the Reasoning part or is the expression of the Spirited part that has taken it over from the Reasoning part, we should see this as a judgment of happiness by the Reasoning part. The eyes, or more exactly, the Appetitive part is unhappy precisely in virtue of having these desires.8 The Spirited part would not make such a judgment by itself; it is instead part of the defining task of the Reasoning part to judge what is best for each part of the soul (442C5-7). So if this is right, the Reasoning part’s overall judgment at least at the time of action must be against looking at the corpses.9

Second, from 440A9 to 441A4, Plato repeatedly stresses the natural affinity that the Spirited part has for the Reasoning part. Indeed, the task that Socrates faces at 441A5-6 is showing that the Spirited part is distinct from the Reasoning part. But if the Reasoning part had judged that it was better overall to look, this would in itself show the distinctness of the Reasoning and the Spirited parts, and there would be no need to find another case of their opposition at 441A7-B1. This second point also rules out the possibility that Leontius’ Reasoning part changes its judgment about what is overall best, that is, from having no prior judgment or a prior judgment that it was better not to look to judging at the time of action—as Carone suggests—that it was best to look. Even a temporary difference in judgment between the Reasoning and the Spirited parts would suffice to prove their distinctness. Because we cannot allow the judgments of the Spirited and the Reasoning parts to diverge without undermining 441A7-B1 and not even Carone suggests that Leontius’ Spirited part ever fails to reject looking, Leontius’ Reasoning part, too, must have rejected looking throughout the entire episode. Thus, the traditional interpretation that Leontius is a case of strict akratic action is sustained.

This reading of the Leontius case is confirmed by the immediately following passage that generalizes it:

And do we not on many other occasions notice that when desires force [biazôntai] someone contrary to calculation, he reproaches himself and gets angry with that in him that is doing the forcing [biazomenôi], so that of the two factions that are fighting a civil war, so to speak, spirit allies itself with reason? (440A9-B4)

This would be a highly misleading way of making the point that although the Appetitive part uses force on the Reasoning part, it never actually gets its way unless the Reasoning part gives in and changes its overall best judgment and endorses the Appetitive part’s preferred option.

More generally, we can ask what determines the outcome when there is a conflict of desires. Plato does not answer explicitly, but his description of the thirsty man who refuses to drink suggests that the strongest desire always wins out (439A9-B8). In this case, it is hard to see how Plato could, in a principled way, avoid admitting cases of strict akrasia. The point here is simply that the causal strength of some desire of the Spirited or the Appetitive part might be greater than that of the Reasoning part’s desire for the overall good. (Note that this is consistent with the idea that within each part what is judged best is always most wanted.) Both the thirst example and the Leontius case suggest that Plato does not hold that merely being a desire for what is overall best makes a desire extraordinarily stronger than all other desires. Simply stipulating that a nonrational motivation could never be stronger than a desire for the overall best seems implausibly ad hoc, so some commentators have suggested that Plato’s “Hydraulic Model” of desire guarantees (or comes close to guaranteeing) that an agent will not be subject to akrasia.10

Now, we surely know that, when someone’s desires incline strongly for one thing, they are thereby weakened for others, just like a stream that has been partly diverted into another channel. (485D6-8)

Some scholars hold that this passage shows that whichever part has the highest “flow” of desires exercises rule in the soul. Let us grant this. This does not, however, show that in all people the distribution of desires is so lopsided that a desire for something other than what is best could never win out. Suppose that one’s stream of desires runs slightly more to the Spirited part than to the Appetitive part and not at all (or nearly so) to the Reasoning part. Let us grant that the Spirited part rules in this person. Taking over the Spirited part’s goals, the Reasoning part deems that it is best to forego large amounts of money, food, sex, and drink for the sake of some minor honor. The Hydraulic Model may entail that the net strength of all the desires in the Reasoning part and the Spirited part is greater than the strength of the desires in the Appetitive part. But for any given action, all that matters is the strength of the desires in direct competition, and the Hydraulic Model is too general to entail that the desire for the overall good plus whatever supporting spirited emotions there may be would be stronger than any appetitive desire.11

Before leaving this topic, let me note one important issue that I cannot explore fully here. If Plato is committed to the principle that the strongest desire always wins out, we might have two worries: (1) the person’s actions might seem to be psychologically compelled, and (2) even if (1) is not the case, the person’s agency seems much reduced; she seems to make no choice or decision that is an effective determinant of what she does. Although I do not have the space to explore these issues, both of these worries deserve an answer and (2) is especially troubling. I have argued elsewhere that Plato returns to this topic in the Laws, where he seems to give the person an ability to intervene in the competition of forces.12

In sum, although Plato is not entirely explicit, it seems fairly clear that the Republic recognizes the possibility of strict akratic action. Accepting this is consistent with allowing that the lower parts primarily influence the Reasoning part not by causing strict akratic action, but by causing irrational belief change.13 We can also appeal to dialogues that are close in time to the Republic.14 In particular, the Phaedrus recognizes strict akratic action without ambiguity. There the black horse, representing the Appetitive part, forces the Reasoning and the Spirited parts, while they are unwilling, to approach the beloved boy for sex (Phdr. 254D2-7).15 It seems to me more plausible that Plato would announce a change of mind about akratic action during Republic 4’s account of the division of the soul and the first explicit recognition of its nonrational parts rather than en passant during the Phaedrus’ account of love and Recollection. But in any case, the Phaedrus passage does show that Plato unequivocally recognizes strict akratic action near the time of the Republic.

Let me close this section by putting akratic action in a broader perspective by considering its relation to Plato’s partitioning of the soul. The parts of the soul are essential to Plato’s account of akratic action. But they have a much broader significance and play more roles than that for Plato.

  1. (1) Partitioning enters Plato’s thought along with a radical distinction between philosophers and nonphilosophers in terms of their goals and capacities. Partitioning explains these distinctions. In particular, the different capacities, characteristics, and goals of philosophers and the types of nonphilosophers are explained by which part rules in the individual.16

  2. (2) The virtues are characterized in terms of parts of the soul and their features. Thus, the nature of the virtues depends on the nature of the soul parts, and the goodness of the virtues is dependent on the ways in which the soul parts can be good. On the other hand, once Plato allows akratic conflict and akratic action, it is a primary task of the nonwisdom virtues to prevent them.17

  3. (3) Because all people have all three parts at least while incarnated, the partitioning theory provides Plato with a general theory of human psychology.18 His account of the Spirited and the Appetitive parts provides him with a general account of nonrational motivation in both the virtuous and the unvirtuous.

Solving the puzzle of akratic action was not likely to be the driving motivation for partitioning the soul, nor was it the most important result of it from Plato’s point of view. Nevertheless, akratic action is sufficiently puzzling and Plato’s proposed solution is sufficiently insightful that it is, I think, reasonable to see this theory of akratic action as a significant accomplishment.19

Means-End Reasoning and the Appetitive Part

In this section and the next, I consider questions about the capacities of the parts of the soul. Here I discuss whether the Appetitive part can engage in means-end reasoning and in the next section whether it can have desires for its own good. I first consider means-end reasoning in connection with the Appetitive part’s love of money and then turn to some broader issues.

I begin with some clarifications. Suppose that while thirsty I see a Coke machine. I believe that if I put $1 into the machine, it will give me a Coke. I thus form the desire to deposit $1 and so act. I have moved from having a certain noninstrumental desire and an instrumental belief to having an appropriate instrumental desire. If, however, we think of reasoning as a process that takes me from beliefs to beliefs, I have not engaged in a process of reasoning: I have moved from a belief and a desire to another desire.20 Because the literature on the Republic tends to include such transitions from beliefs and desires to other desires as instances of means-end reasoning, I shall follow this usage, although not all philosophers see such transitions as cases of means-end reasoning for the reason just noted. Those who accept that the Appetitive part is capable of means-end reasoning hold that it is capable of having the relevant instrumental beliefs and making the appropriate transitions. Those who deny means-end reasoning to the Appetitive part typically deny that it can have instrumental beliefs at all.21

Let us begin with the textual evidence. At 442A6-7, the Appetitive part is said to be “by nature [phusei] most insatiable [aplêstotaton] for money.” In Book 9 Socrates says of the soul’s third part:

we called it the [A]ppetitive part because of the intensity of its desires for food, drink, sex, and all the things associated with them, but we also called it the money-loving [philochrêmaton] part, because such desires are most easily satisfied by means of money [hoti dia chrêmatôn malista apotelountai hai toiautai epithumiai] … Then, if we said that its pleasure and love are for profit, would that not best determine its central feature … and would we not be right to call it money-loving and profit-loving [philokerdes]?22 (580E2-581A1)

I do think that these passages combined with the implausible consequences of denying means-end reasoning to the Appetitive part give us good reason to attribute it to the Appetitive part. But before adopting that conclusion, let us consider the opposing arguments.

Hendrik Lorenz denies means-end reasoning to the Appetitive part on two grounds: Socrates (a) characterizes the Appetitive part as “unreasoning” (alogistos) at 439D7, and (b) never explicitly says that the Appetitive part itself recognizes that money is a means to the satisfaction of its other desires.23

Because the aforementioned passages are hardly evidence against attributing means-end reasoning to the Appetitive part, (a) must carry the weight of Lorenz’s argument. But I do not think that it can bear such a burden, because Plato does not say at 439D7 what he means by “irrational” and Lorenz simply relies on his intuition—which may not be shared by Plato—that an ability to engage in means-end reasoning is sufficient for Platonic rationality. (I return to rationality in the fourth section.)

Even if the Appetitive part does not love money as a means, Lorenz accepts that it loves money in itself and thinks that he can explain why this is so without attributing means-end beliefs to it. Lorenz suggests that the stories that children heard praising money led them to become attached to it as a direct source of pleasure.24 But such an explanation is unsatisfying. Children also heard stories praising justice, and many fewer of them came to take direct appetitive pleasure in it. Lorenz’s view also leaves it mysterious how one comes to acquire appetitive desires for things such as cheating on one’s taxes that are not culturally praised but are means of making money. What we need is an account that connects the Appetitive part’s love of money and its inherent desires for food, sex, and drink. James Wilberding offers such a proposal. An associative link forms, he suggests, between having money and the pleasure of satisfying one’s desires. After significant experience of having money followed by the pleasure of the satisfaction of my appetitive desires, my Appetitive part comes to take pleasure in having money.25

Although this proposal is an improvement, it is still not satisfactory. First, this associationist link is not hinted at in the text and the Republic does not otherwise show signs of associationism. (Evidence for some kind of associationism comes from the much later dialogue, the Philebus.) Second, there is the standard problem in associationist explanations of determining which properties are associated with which other properties. As Anthony Price points out, it is unclear, for example, why “appetite comes rather to love having money (a state that precedes spending and makes it possible) than to love having had money.”26 Spending money, after all, is more closely associated with the resulting pleasure than is merely having money. Third, associationist learning is slow because it requires repeated experiences of the stimulus. But cannot the Appetitive part learn quickly about some relevant means? For example, cannot the Appetitive part develop a desire to cheat on the person’s taxes after one success? Indeed, is experience necessary at all? Is it not possible for the oligarch merely to hear about the benefits of tax cheating, form an overall aversion to it, and still have an appetitive desire to cheat? Here we cannot appeal to associationist memory links to explain such a desire, but it seems that we must attribute to the Appetitive part some grasp of means-end connections as well as the resultant instrumental desires.

Matters only worsen when we remember that Lorenz not only denies to the Appetitive part the ability to engage in means-end reasoning about money but also the ability to engage in means-end reasoning about anything at all. At the same time, he gives it the role of generating all of our motivations for food, sex, and drink that are not produced by the Reasoning part.27 Such a view is subject to a number of difficulties. I have space to mention only two. First, as Lorenz agrees, the Appetitive part is a language user and has beliefs.28 It is, however, barred from having means-end concepts, and this is a radical limitation. The Appetitive part thus seems excluded from having causal concepts and so desires involving them: so there will be no appetitive desires for French-pressed coffee or kosher food.29 But without causal concepts, the Appetitive part’s cognition will be remarkably primitive. Children two to four years old have a large body of causal beliefs, including some concerning means-end relations, which cannot be accounted for by associationist mechanisms.30 Furthermore, such a denial seems false to our own experience. Ex-smokers are familiar with working out elaborate means of getting cigarettes and, once the plan is finished, of feeling the tug of desire to take the first step all the while having the rational judgment that it is best not to smoke and the rational desire not to smoke. To avoid splitting the Reasoning part, such desires must be attributed to the Appetitive part. Where else should we locate the plan for the means?

The second problem concerns action. The Appetitive part, according to Lorenz, cannot rely on means-end beliefs or desires in the production or guidance of action. How, then, can it produce complex sequences of actions? Lorenz relies here on memories of previous action sequences in the following example adapted from him.31 Suppose I have rationally decided not to smoke and my Reasoning part believes that it is best overall not to smoke and desires not smoking as best overall. My Appetitive part nevertheless desires to smoke. I now find myself with a desire to go to the store and buy some cigarettes. This desire cannot be attributed to the Reasoning part because, as Lorenz agrees, this would result in its subdivision. But neither can it be attributed to the Appetitive part because it is a desire for a means to the end of smoking. Lorenz suggests that

appetite has some kind of cognitive access to reason’s judgments that the way to obtain cigarettes in the circumstances is by going to the shop around the corner and buying them there. This may be by way of a representation that in some way or another presents the whole course of action “going to the shop, buying a pack of cigarettes there, and smoking a cigarette.”32

According to Lorenz, this does not involve any recognition by appetite that going to the shop is a means to smoking and the only desire that the Appetitive part has and acts upon is the desire to smoke; in particular, it has no desire to go to the shop.

Does this succeed in explaining akratic action without attributing any means-end items to the Appetitive part? First, with respect to the informational side of the story, such representations are not found anywhere in the Republic, but only, perhaps, in the Philebus and the Timaeus. (However, it is questionable, even there, how ubiquitous their role is and how specific their action guidance is.33) So it is not clearly appropriate to assume that Plato relies on them in the Republic. Second, the example tacitly depends on there being a match between the memory sequence and the current options open to me. But if it was raining in the memory sequence and is not now, do I still go? If I know that the shop is closed now, why would I refrain from going if my Appetitive part does not recognize that it is going to the store as a means to get cigarettes? Third, the example tacitly relies on there being only one memory sequence that is available. But I also have a memory sequence of flying across country and buying cigarettes in the airport gift shop. Why do I not take that sequence instead of going to the store around the corner? My Appetitive part should not be concerned with considerations of efficiency because on this view it lacks all concepts of means and does not see either action sequence as a means to getting cigarettes.34 Fourth, the example does not take into account the possibility of novel means. I learn, for example, that the store is about to close. I could make it there by driving, but I have never driven there. I realize that by the time I walk there, the store will be closed. Nevertheless, it seems that the only appetitively motivated action I could take on Lorenz’s story is to walk to the store.35

I do not think that the motivational side of Lorenz’s story is any more persuasive. In this example’s latest version, Lorenz does not say why the only desire that the Appetitive part needs to carry out the whole sequence of actions in the representation is the desire to smoke, but he previously cited Gorgias 467C5-E1 in support of this claim.36 Let us see whether this helps. At Gorgias 467D6-E1, Socrates claims, “If a person does anything for the sake of something [ean tis ti prattêi heneka tou], he does not want this thing that he is doing, but the thing for the sake of which he is doing it.”37 The principle that Socrates endorses here is that when S does X for the sake of Y, S wants Y and S does not want X. But note this is precisely a case of instrumental reasoning and Lorenz cannot avail himself of the principle without allowing the Appetitive part to engage in instrumental reasoning after all. But in any case, this principle is modified at Gorg. 468C2-5 to the effect that “If S does X for the sake of Y, S wants X only for the sake of Y.” The modified principle, besides being a case of instrumental reasoning, holds that the means are wanted. The Gorgias passage would thus not help Lorenz’s position, and we have not seen how the mere desire to smoke along with the representation can explain action in accordance with it.

In sum, I think that we have seen good reason to think that the Appetitive part of the soul can have means-end beliefs and desires.38

The Appetitive Part of the Soul and the Good

Recent years have seen a lively debate over the sorts of motivations open to the Appetitive part. One issue concerns whether it has good-independent desires, that is, desires for objects that are not desired under the guise of the good. In particular, discussion has focused on whether thirst in Book 4 is such a desire. Indeed, on some accounts of thirst, it is a desire for its object (i.e., drink), that does not involve seeing this object under any desirability characterization at all.39 Although I cannot enter into this debate here, two points are worth noting. First, even if thirst is good-independent, this does not entail that all appetitive desires are. Second, there is convincing evidence that at least many appetitive desires aim at things insofar as they are pleasant. In virtuous people, the desires in the soul’s lower parts “pursue” (diôkousai) and “take” (lambanôsi) the pleasures that reason approves (585D5-7), whereas in others these desires “pursue” (diôkein, 587A6) alien pleasures. The Appetitive part is identified in the partitioning argument as the part with which “we desire the pleasures of nutrition and generation and their kind” (436A10-B2) and later as the “companion of certain replenishments and pleasures” (439D8). 436AB is especially important because it is part of the exact question that the partitioning argument tries to answer. It and 439D8 flank the discussion of thirst. So we cannot see the idea that appetitive desires are good-independent as simply correcting an earlier view, but we should allow that at least many appetitive desires aim at pleasure.

Our question follows on the point just discussed about means-end reasoning. If the Appetitive part is capable of means-end reasoning, then it is capable of having some conception of an end, at least the end of an action on a particular occasion. A passage from Plato’s discussion of the oligarch goes further. The fact that this person commits injustices when unnoticed shows that

in those other contractual obligations, where he has a good reputation and is thought to be just, he is forcibly holding his other bad desires in check by means of some decent part of himself. He holds them in check, not by persuading [them] that it is better not, nor by taming [them] with reason [ou peithôn hoti ouk ameinon, oud’ hêmerôn logôi], but by compulsion and fear, trembling for his other possessions. (554C11-D3)

Whatever else may be uncertain here, it seems clear that Plato is criticizing the oligarch for not persuading or taming with reason his appetitive desires or part and that he is being unfavorably contrasted with someone who does exactly that.

Charles Kahn, however, objects that “[I]t is only if the oligarchic man attempted such an argument, and if it succeeded, that something might follow about the Appetitive part agreeing to a judgment of goodness. But since the first condition is not satisfied, nothing is asserted here about the appetites making a judgment of goodness.”40 It would, however, be pointless to criticize the oligarch for failing to do something that simply cannot be done. Thus, Plato must have thought that such persuasion was, in some way, possible. Joshua Wilburn suggests that although we can reproach the oligarch for his appetites’ content, Plato is not objecting to his methods: even in the virtuous person, persuasion of the appetites is not possible and she can only force or frighten them.41 For Wilburn, the implicit contrast in the earlier passage is not with the virtuous person’s better way of relating to her Appetitive part, but with the better way that she can relate to her Reasoning and Spirited parts. But this just changes the topic, and the oligarch is no longer being criticized for the relation in which he stands to his Appetitive part at all.

The passage’s context should help clarify the intended contrast. Both just before and after 554CD, Plato contrasts the oligarch and the virtuous person. The oligarch has bad desires because he has ignored the right education (554B4). So he suffers from internal stasis and is a “double” man (554D9-E2) lacking “the true virtue of a like-minded [homonoêtikês] and harmonious soul” (554E5-6). Plato’s language here echoes his description of moderation, and the role of the Appetitive part is especially important in moderation. Thus, the most natural way of understanding the criticism is as contrasting the oligarch’s relation to his Appetitive part with the virtuous person’s relation to hers.

We can make some progress on the virtuous person’s sort of persuasion by seeing why the oligarch cannot persuade his bad desires. First, we need to note Plato’s distinction between necessary and unnecessary desires. Necessary desires are those that either (1) we cannot get rid of or (2) benefit us (558D11-E3). There is a natural necessity for us to seek to satisfy both (558E2-3). Unnecessary desires are those that (a) we can get rid of and (b) harm us or at least do us no good (559A3-6). The oligarch’s better part, his Reasoning part, and his better desires aim at making money and perhaps the satisfaction of some necessary desires such as those for health or its corresponding pleasures. His bad desires are his unnecessary desires and these aim, paradigmatically, at the unrestrained pleasures of food, sex, and drink. Little or no persuasion of unnecessary desires seems possible because there is scant overlap between the actions recommended by the oligarch’s Reasoning part and those that could satisfy his unnecessary desires, and these latter desires are unchangeable in the relevant respects (they remain fixed on unbeneficial pleasures). Nor would just people be any better able to persuade their unnecessary desires, but their training from childhood has left them with few such desires (559A3-6, compare 571B2-C1).

What sort of persuasion, then, does the just person engage in with respect to her Appetitive part? We can, I think, make some sense of the idea of persuading a necessary desire. To begin, suppose that I learn that the bread I am about to drink is spoiled. In this case, I shall typically lose my occurrent necessary desire for it, although the necessary desire for bread remains as a disposition. This loss of a derivative desire via new information is one way that a necessary desire may be “persuaded.”42 But how does the Appetitive part know that the bread is spoiled? It might be that it is perceptibly evident. Yet there could be cases in which it took more than the Appetitive part was capable of to determine this, and at least in some of them the Appetitive part might take over the Reasoning part’s belief that the bread is spoiled and thus lose its occurrent desire. This is, we should note, a relatively weak form of persuasion; the necessary desire is left exactly as it was; it is only pointed out that an object that purports to satisfy it does not in fact do so. But if the parts of the soul are agent-like, the virtuous person may be able to engage in persuasion in the richer sense of persuading her Appetitive part itself.

So far we have considered cases in which an occurrent necessary desire is lost, but reason can also help specify general necessary desires. Consider the necessary desire “to eat to the point of health and good condition” (mechri hugieias te kai euexias, 559A11-B1). (At 559AB the other necessary desires seem to be specified in terms of their intentional object. The intentional object of this necessary desire is not food, but something like a healthful quantity of food. Indeed, it is very hard to see how this necessary desire’s intentional object could be food while it plays the role that it does. This may be in tension with Book 4’s account of thirst if that account holds that all appetitive desires are specified in terms of unqualified generic intentional objects similar to that of the thirst-drink case.43) The Reasoning part could persuade the Appetitive part that a smaller than expected quantity of food is more healthful. This belief of the Appetitive part, along with the necessary desire, would then produce a more specific desire for the right amount of food. Plato describes a related process in Book 9:

those desires of even the gain-loving and the victory-loving parts that follow [hepomenai] knowledge and argument [logôi] and pursue in conjunction with [meta] them those pleasures that reason [to phronimon] approves [exêgêtai] will attain the truest pleasures possible for them, because they follow truth, and the ones that are most their own, if indeed what is best for each thing is most its own.44 (586D4-E2)

This echoes an earlier passage about moderation in the city:

But you meet with the desires that are simple, measured, and directed by [agontai] calculation with the aid of understanding and true belief only in the few people who are born with the best natures and receive the best education. (431C5-7)

In the first passage, reason acts as an “exegete” insofar as it points out to the lower parts which pleasures are to be pursued as best without trying to convey to these parts what would be beyond them, that is, a full explanation of why such pleasures are good and true. As the context makes clear, Plato, with respect to the Appetitive part, has in mind the pleasures of drink, sex, and especially food (compare with 586A1-B4). So we can take this passage to apply to eating to the point of health. Reason here is clearly intervening in order to shift the objects of the appetitive desire from what they would otherwise be, and this is to specify the necessary desire for eating up to the point of health. This is persuasion about what is most pleasant, but I shall shortly consider the relation between pleasure and the good for the Appetitive part.

If the foregoing analysis is right, we have seen that even necessary desires and the Appetitive part itself are open to a kind of persuasion. But before turning to the last two topics in the section, there is a point of clarification to make. In focusing on the example of the oligarch (554CD), it is easy to think that all persuasion takes place wholly within the context of a single decision-making episode. But this is not necessary and we should see the auxiliaries’ musical education in Books 2 and 3 as a kind of long-run persuasion of their nonrational motivations, including their necessary desires.45 The kinds of poetry they are to hear will affect the contents of their Appetitive parts’ nonrational motivations, as evidenced by Plato’s concern with not only poetry’s rhythms and harmonies but also its words.46

Let us consider the conception of the Appetitive part that emerges from our discussion. To begin, what cognitive abilities does it need to be persuaded by the Reasoning part? First, it needs an ability to revise its beliefs in response to new information. Second, this ability would be pointless if the Appetitive part never actually revised its beliefs or if it revised them at random. Thus, it needs some grounds for belief change, although these need not be explicit and conscious. The Republic does not provide an analysis of belief, so any answer to this question will be somewhat speculative. It is natural to think that the Appetitive part’s belief revision is governed by some norm of aiming at the truth, but, as we shall see in the next section, this is controversial. For our purposes, what is important is that it can, and sometimes does, revise in the direction of truth in response to the Reasoning part.

On the practical side, we saw in the last section that the Appetitive part was capable of some means-end reasoning. Being such requires some conception of oneself as temporally extended (the means, the end, and the thought of them are not all simultaneous). Means-end reasoning also brings with it a concern for efficiency and involves some concern for one’s long-run interests.47 We have also seen in this section that the Appetitive part insofar as it receives input from the Reasoning part can take more than one consideration into account in forming desires.

Does the Appetitive part also have a conception of the good? Let me briefly note two issues before taking up this question. First, some of the reluctance to take seriously the attribution of a notion of “good” to the Appetitive part may dissipate once we see some of the other notions attributed to it. At 442D2, the Appetitive part shares the belief that the Reasoning part “should” (dein) rule, an attributive use of “good” occurs in the necessary desire “to eat to the point of health and good condition [euexias],” and the Appetitive part contains “false and boastful words and beliefs” (560C2) about, among other things, the virtues. 580D6-7, in attributing a “ruling principle” (archê) to each soul part, seems to attribute to it a prominent ultimate end, for example, knowledge for the Reasoning part. “Prominent” is an important qualification: this is not the Reasoning part’s only ultimate end, because it also aims at the good of the whole soul as an ultimate end. There is more variability in the Appetitive part’s ultimate ends, but 580D-581C seems to take for granted that it does have some such prominent ultimate end. These passages do not force us to read 554CD literally, but the more passages we have to read away, the less plausible it is to do so.

The second point concerns the Appetitive part’s substantive conception of the good. As the definition of wisdom (442C4-7) shows, each part of the soul has its own good.48 The passages about pleasure noted at the beginning of this section are strong enough that it is tempting to think that pleasure is the good for the Appetitive part.49 I cannot try to settle here whether this is the case, although it seems reasonable to think that pleasure will be a major component of the good for the Appetitive part.50 But this raises several concerns. First, we might worry whether the Appetitive part is capable of developing and maintaining some grasp of a single ultimate end which is such that all other objects are chosen for its sake and it is chosen for its own sake and not for the sake of anything else. Pleasure, of course, might still be the good for the Appetitive part even if it could not recognize this. Second, good-independent desires do not fit into such a scheme. Third, allowing a part to have an ultimate end threatens to reintroduce akrasia at the level of the part and thus split the Appetitive part if there is a conflict between the Appetitive part’s desire for its overall good and some appropriately conflicting desire.

With respect to the first worry, the Appetitive part need not have an articulate and worked-out conception of the good or be able to rank all possible options. What is important is that in the virtuous person, at any rate, it is able to get right a standard range of cases, for example, prefer true and satiable pleasures to false and insatiable ones. Second, as we noted earlier, even thirst may not be good-independent. But in any case, the threat of further splitting is not a problem unique to this interpretation; there is no interpretation of the parts of the soul and the Principle of Contraries that I think successfully avoids the danger of splitting within the parts in all cases.51 Finally, it is reasonable to assume that the Appetitive part does not infer the goodness of pleasure from independent criteria of goodness. The relation of priority runs in the opposite direction: x is found good because it is pleasant, x is not found pleasant because it is good. In finding x good because it is pleasant, the Appetitive part takes x as an ultimate end because it is pleasant.52

Before turning to some larger issues, let us return to 554CD now that we have a fuller understanding what “it is better” involves: it involves the claim that it would be more pleasant in the long run to refrain.53 Is this something that either a necessary or an unnecessary desire could be responsive to? Both necessary and unnecessary desires (at any rate some of them) aim at pleasure, so there is overlap with the admonition to that extent. But the types of pleasure that these two kinds of desire aim at are quite different: unnecessary desires aim at pleasures that are typically insatiable and false, while necessary desires aim at pleasures that are (relatively) satiable and true. So in the case of the oligarch, if “it is better that” entails “it is more genuinely pleasant in the long run,” then there would be little overlap with what would be sought by the unnecessary desires and thus little possibility of persuasion. Necessary desires, however, do benefit us and so must be directed toward satiable and true pleasures. So there would be overlap with the admonition and to that extent necessary desires are persuadable. This fits well with the two passages we have seen in which Plato claims that reason can lead desires for pleasure, including appetitive desires for pleasure, to the right sorts of objects.

Persuading the Appetitive part itself in the way envisaged at 586DE will result in that part’s having the true belief that certain moderate and restrained pleasures are the truest, most pleasant, and best pleasures for it. Such a belief should allow the Appetitive part to specify, for example, the general necessary desire for healthful eating. If the suggestion that the intended contrast case with the oligarch is the virtuous person and her relation to her Appetitive part is right, then persuasion seems possible both with respect to her necessary desires and her Appetitive part itself. We have also seen that there is a reasonable case for thinking that “good,” if properly understood, is not outside the cognitive abilities of the Appetitive part, and so we do not have to dismiss or read away 554CD.

Jessica Moss goes further and suggests that the Appetitive part finds pleasure “fine and good” and “most worthy of pursuit.”54 She appeals to the parallel with the democratic city, which finds that license “is the finest [kalliston] thing it has, so that this is the only city worth [axion] living in” (562B10-C2).55 Moss rejects as “needlessly indirect” the idea that the Reasoning part would make such a judgment.56 But in the democratic person it is the Reasoning part, not the Appetitive part, that makes the judgment of worth that all desires are to be honored equally (561B8-C4).57 So it seems better to understand the notion of good involved here in the thin sense sketched earlier.

Such an interpretation not only allows us to avoid dismissing 554CD as a mere rhetorical exaggeration, but it also fits together well with a number of Plato’s other views. Moderation requires more than agreement among the three parts of the soul: it requires friendship among them (philia, 442C9-D2). In a virtuous person, the Reasoning part knows and seeks not only what is best for the whole soul but what is best for each part (442C5-7) and leads each part to its truest and best pleasures (586D4-587A2, compare with 588E4-589B6). This helps to make sense of the requirement of friendship, because friendship is most likely to obtain where there is a coincidence of interests between the parties (412D4-7). Although the Appetitive part is unlikely to have any specific beliefs about what benefits the Reasoning part, it seems that it would especially feel friendship toward the Reasoning part when it believes that the Reasoning part’s rule is in its own best interests. The passages at 586DE and 431C quoted earlier stress that the desires in both the Spirited and the Appetitive part are supposed to follow and be sensitive to reason’s judgment of what their best and truest pleasures are. It is plausible that in doing so they are intended to be aware of doing what is best for themselves.

Finally, this possibility of feelings of friendship in the Appetitive part gives Plato a further interesting way to distinguish moderation from something like Aristotelian self-control (enkrateia). Plato seems to think that even in good people, some unnecessary and lawless desires remain (571BC) and perhaps some unnecessary, but lawful ones. Having friendship in the Appetitive part, along with agreement about which part should rule, would distinguish genuine Platonic moderation from a less valuable state in which it was only the case that the appetitive desires for what reason approves were stronger than those that were contrary to it.

The Rational Versus Nonrational Distinction

Finally, let us consider what distinguishes the rational from the nonrational parts of the soul. For Lorenz, it is the capacity for means-end reasoning.58 I have argued in the second section, however, that the Appetitive part engages in means-end reasoning and that the Spirited part is also capable of it (590B). Nor does the capacity for means-end reasoning seem to help to explain the sense in which the nonrational parts are nonrational in virtue of being subject to bad emotions in response to imitative poetry or being prone to believe the wrong side in cases of perceptual illusion.

Todd Ganson suggests two related ways in which the nonrational parts differ from the Reasoning part. Nonrational parts

  1. (1) are incapable of measuring, weighing or calculating,59 and

  2. (2) cannot draw a distinction between their representational states and how things are, that is, they cannot distinguish between appearance and reality.60

With respect to (1), in Book 10 Plato distinguishes the Reasoning part, which “puts its trust in measurement and calculation” [metrôi … logismôi], from the nonrational parts, which hold beliefs opposed to measurement (603A4-7).61 Socrates adds that “measuring, calculating, and weighing [metrein kai arithmein kai histanai] … are the function [ergon] of the Reasoning part of the soul” (602D6-E2). This is clearly an important claim for understanding the rational/nonrational distinction. But how, exactly, are we to understand measuring, calculating, and weighing? For example, does denying them to the lower parts preclude these parts from making comparative judgments?62 The lower parts do, however, seem capable of various comparative judgments. The Spirited part can judge that one thing is more shameful than another and that it is sometimes better to wait to exact vengeance. One might object that these are not genuinely quantitative judgments. But cannot the Appetitive part judge that one outcome involves making more money than another?63 In order to avoid the possibility of splitting within the Reasoning part (compare the earlier discussion in the cigarette example), we cannot always ascribe the desire for two talents of gold rather than one or for two honors as opposed to one to the Reasoning part. Although Plato does not go into details, he seems to recognize perceptual or sensible versions of some number concepts, that is, number concepts that have only sensible content. If this is right, then not all forms of measuring, calculating, and weighing may be beyond the capacities of the lower parts.64

Similar caution is needed with respect to (2). On a straightforward reading of the Book 10 passages, we might take Plato to mean that in all cases, the content of the lower parts’ beliefs is simply provided by the relevant sense-perception so that in cases involving Müller-Lyer lines, the lower parts always believe that the lines are unequal; in cases involving sticks in water, they always believe that the sticks are bent, and so on.65 It also seems to be part of this interpretation that the lower parts are incapable of learning from experience that such judgments are false. This is a view shared by Moss: “the non-rational part of the soul is the part that fails to question appearances … with respect to size or shape.”66 The difficulty here is that such an interpretation gives Plato a wildly implausible view of people’s nonrational motivations. Could Plato have really supposed that Greek soldiers on seeing hordes of Persian troops at a great distance suddenly found that their fear vanished and their Spirited parts gloated in the belief that the Persians were literally the size of ants? The problem is that once we allow that the nonrational parts literally believe what seems to be the case to perception, these beliefs can then be incorporated into nonrational motivations with unacceptably bizarre results. Such difficulties give us good reason to avoid attributing to Plato the strong form of the thesis found in Ganson and Moss that nonrational parts of the soul cannot criticize appearances.

For Moss:

appetite and spirit desire pleasures and honour, and feel grief or anger, for the same reason that they perceive the stick as bent and are responsive to perceptibles in general: because they are cognitively limited to the perception and acceptance of appearances … The rational part’s ability to calculate, meanwhile, allows it to criticize and transcend appearances both in the sensory realm and in the ethical.67

Here, as on Ganson’s view, “appetite and spirit believe optical illusions.”68 More generally, the lower parts accept the contents of their experiences; they are unable to criticize their appearances in any way. There are several points to make about Moss’s interpretation.

  1. (1) Insofar as the lower parts literally believe optical and other illusions, such content will infect their nonrational motivations. Since the lower parts cannot criticize appearances, the corresponding beliefs should be persistent and thus, for example, the thumotic confidence inspired by believing that the Persians are ant-sized should persist despite the Reasoning part’s opposed judgment. Appetitive desire for another apple should persist through trying an entire bowl of fakes. This leaves Plato with a bizarre theory of nonrational motivations that is inconsistent with our experience of them. (Mutatis mutandis, for actions that the Appetitive part originates)

  2. (2) Moss thinks that in order to explain our ordinary desire for things as fine (kalon) we must attribute to Plato belief in a “special mode of perception, evaluative perception, distinct from but in the same psychological category as seeing and smelling.”69 Such perception takes the fine as its object.70 It is important to note that Moss attributes this mode of perception not to the Reasoning part, but rather to the Spirited part. Thus, it is wrong to see it as a kind of rational intuition of the value of the fine. The fine, on Moss’s view, is not accessible to the five senses but is accessible to a special mode of perception of the Spirited part.

But I do not think that we should accept the existence of such a special faculty. First, there is no textual evidence for it in the Republic. We would expect that such a major innovation in Plato’s epistemology would receive some mention. It has no place, for example, on the Divided Line. Second, in the Phaedrus and the Timaeus, Plato takes pains to make it clear that the lower parts never have any contact with the Forms; thus, they cannot have any concepts that come via Recollection.71 In the Phaedo, the lower parts’ desires are attributed to the body along with sense-perception, and the body has no share in Recollection. Thus, it is reasonable to think that in the Republic also the lower parts have no concepts that come via Recollection and do not cognize Forms. But then it is very hard to see how the Spirited part could perceive the fine. It is much more reasonable to think that the Spirited part, like the lovers of sights and sounds in Book 5, to the extent that it has a conception of the fine, really has a concept that has purely sensible content. Moss objects that honor is not a sensible property.72 But, first, the Republic has a broad conception of perception: we can, for example, perceive that something is a finger (523CD). Second, we have good evidence that Plato was not troubled by the idea that honor was a perceptual property. In the Phaedo, the love of honor is attributed to the body, and the body’s only epistemic resource is perception.73

So, how, then, should we understand the rational/nonrational distinction? To begin, let us consider the original equipment of the parts. We can divide the original equipment into a practical side and a cognitive side. On the practical side, each lower part comes equipped with self-concern. The Spirited part also has desires and emotions connected with self-assertion; the Appetitive part has desires for food, sex, drink, and money. On the cognitive side, the fundamental resource for each of the lower parts is sense-perception, broadly construed, and belief. Along with this, they have various abilities, including those of using a language and engaging in simple forms of reasoning, including means-end reasoning.74

On the practical side, the Reasoning part has both a sense of self-concern and a concern for the whole soul. It naturally loves knowledge and truth and seeks to express this love of order in its actions. On the cognitive side, it also has, in some way, sense-perception along with belief and has to a much higher degree the abilities had by the lower parts. But this is not merely a matter of degree, because the Reasoning part can use concepts that have nonsensible content, including number concepts. The Reasoning part’s fundamental and unique resources are thought (dianoia) and understanding (epistêmê) which take nonsensible reality as their object.75

The fundamental dividing line between the nonrational and the rational is that between sense-perception and nonsensible cognition. This roughly coincides, as we saw at 602D-603A, with the distinction between noncalculative and calculative thought (measuring, counting, and weighing), but the coincidence is not perfect because a kind of calculation might be performed using sensible units (525D5-526B3, compare with 522D1-7, 523A1-3, Phil. 56D4-57A6), for example, counting up to four cows by adding two groups of two cows. On the other hand, it seems that engaging in the sort of discussion that leads to the accounts of the virtues in Book 4 does not require any measuring, counting, or weighing.76

What is necessary and sufficient for rationality on this account on the cognitive side would be engaging in some mathematical or philosophical thinking that employs, and is understood to employ, nonsensible concepts.77 This statement is only a rough first approximation because, for example, it contains no success or correctness constraint: any line of thinking, no matter how error-filled or confused, would count as rational if it employed nonsensible concepts. Because this seems unsatisfactory, we might make the initial condition a necessary condition and add a success or truthfulness constraint. But there are at least two worries about such a move. First, it is not clear that rationality requires truth even for Plato: consider, for example, the likely account (eikôs muthos) of the Timaeus. At any rate, an appropriate truthfulness constraint may not require completely true accounts. Second, one’s cognitions could satisfy even a strong truth condition and still be irrational: they might, for example, be the product of lucky guessing. So the condition imposed will have to go beyond truth. What seems to be required here is the proper way of thinking about the things with which one is dealing: for example, the elenchus, mathematical reasoning, the method of hypothesis, or dialectic.78 I cannot go in more detail here, but I hope that the basic idea is clear enough for our present purposes.

On the practical side, what would be required is love of the knowledge of nonsensible reality enabled by the use of such concepts as well as genuine self-concern and concern for the other parts based on a (more or less) correct grasp of what is best for each part and for the whole soul.79 Being rational, on this account, although it falls short of philosophical wisdom, is still quite a demanding state. Is it too demanding to be an appropriate conception of rationality? Hendrik Lorenz, for example, rejects one account of the nature of rationality on the grounds that it restricts it to “few” people.80 Also on Moss’s and Ganson’s accounts, insofar as rationality is displayed in questioning the kind of appearances found in sensory illusions, it would seem to be found in at least all normal adults (as well as many children). But there are textual indications that rationality may not be so widespread. In the Book 4 argument distinguishing the Spirited part from the Reasoning part, Glaucon remarks “as for reason [logismou], some people seem to me never to share [oudepote metalambanein] in it, while most do so quite late” (441A9-B1) and Socrates enthusiastically agrees (441B2). Because both Glaucon and the reader have yet to hear anything of the Republic’s epistemology and metaphysics, it seems reasonable to take this as an upper bound on those who will count as rational. It may be the case that every human soul has the potential to be rational; in most people, however, the distinctively rational capacities of their soul remain dormant (533D1-6, compare with 523A10-524C14) insofar as they do not receive an appropriate mathematical education.81 In the late dialogue, the Timaeus, wild land animals originate from the reincarnations of

men who had no tincture of philosophy and who made no study of the universe whatsoever, because they no longer made use of the revolutions in their heads but instead followed the lead of the parts of the soul that reside in the chest … The tops of their heads became elongated and took all sorts of shapes depending on the particular way the revolutions were squeezed together from lack of use.

(Tim. 91E2-92A2; compare with Rep. 528E-530C)

In conclusion, I would like to discuss one implication of this conception of rationality. One common worry about our relation to our nonrational desires is that we seem to be passive with respect to them. First, they are not the product of deliberation because we do not choose them; they simply come over us. Second, as we did not acquire them at will, we cannot rid ourselves of them at will. Third, we do not hold ourselves responsible for our desires in the same way that we do hold ourselves responsible for our actions (consider Plato’s remarks about lawless desires even in good people). Fourth, as we saw earlier, our actions seem to be determined by whichever desire happens to be strongest.82

But there is also a concern distinct from that of passivity. Given the Republic’s conception of rationality, reason alone can grasp nonsensible value properties such as good and fine. The lower parts can never grasp any nonsensible properties and so can never grasp genuine value properties. Thus, what moves the virtuous person to act can never move the lower parts or be accorded any value by them. The problem now is not that I am passive with respect to the lower parts’ motivations, but rather that the lower parts because of this motivational gulf are profoundly alien to me (the I who identifies with the best part within me).83 They cannot be brought to share in the motivational viewpoint of the Reasoning part. Moreover, from the point of view of the Reasoning part, the good condition of the lower parts is at most of instrumental value, because such states are not ways of grasping the truth about reality.

We see this alienation from the lower parts in the metaphors that Plato uses for them in the Republic’s closing books. In the great image of Book 9, the Spirited part is a lion and the Appetitive part a many-headed beast; both of these animals are joined together with the human being who represents the Reasoning part (588CE). In the Book 10 passage comparing the soul to the sea god Glaucus, the lower parts are compared to rocks, seaweed, and barnacles clinging to the god’s true form, which is the metaphor for the Reasoning part (611D8-612A6). In both of these images, the reader is invited to identify himself with the human being or the god and distance himself as much as possible from the lower parts.

Finally, it is important to see that this view of the lower parts as alien to me holds not only when the lower parts are in a bad condition. Plato contrasts the nonwisdom virtues with the virtue of thought (hê tou phronêsai aretê, 518D11-E1), which turns the soul from what is “coming into being” (ek tou gignomenou, 518C8) to “that which is and the brightest thing that is” (to on kai tou ontos to phanotaton, 518C9):

It looks, then, as though the other so-called virtues of the soul are akin to those of the body, for they really are not there beforehand but are added later by habit and practice [ethesi kai askêsesin]. However, the virtue of reason seems to belong above all to something more divine, which never loses its power but is either useful and beneficial or useless and harmful, depending on the way it is turned. (518D9-519A1)

Even the virtuous condition of the lower parts is only a so-called virtue; it is like the good condition of the body, which is something I have rather than something that I am.

It has often been recognized that Plato portrays the nonrational parts of the soul as animals. This has sometimes led scholars to deny to them some capacities that we take to be distinctive of human beings, such as language use, means-end reasoning, or the ability to distinguish appearance from reality. The tacit assumption that rationality is something that is displayed by all normal adult humans is, however, doubtful for the Republic. A more demanding conception of rationality leaves the nonrational parts of the soul capable of more but, at the same time, leaves us estranged from a greater part of our own psychology.


Anagnostopoulos, M. “The Divided Soul and the Desire for the Good in Plato’s Republic.” In The Blackwell Guide to Plato’s Republic, edited by G. Santas, pp. 166–188. Oxford: Blackwell, 2006.Find this resource:

Barney, R., T. Brennan, and C. Brittain (eds.). Plato and the Divided Self. Cambridge: Cambridge University Press, 2012a.Find this resource:

Barney, R., T. Brennan, and C. Brittain. “Introduction.” In Plato and the Divided Self, edited by R. Barney et al., pp. 1–5. Cambridge: Cambridge University Press, 2012b.Find this resource:

Benson, H. Clitophon’s Challenge: Dialectic in Plato’s Meno, Phaedo, and Republic Oxford: Oxford University Press, 2015.Find this resource:

Bobonich, C. Plato’s Utopia Recast: His Later Ethics and Politics. Oxford: Oxford University Press, 2002.Find this resource:

Bobonich, C. “Images of Irrationality.” In Plato’s Laws: A Critical Guide, edited by C. Bobonich, pp. 149–171. Cambridge: Cambridge University Press, 2010b.Find this resource:

Bobonich, C. “The Puzzles of Moderation.” In Platon Gesetze Nomoi, edited by C. Horn, pp. 23–43. Berlin: Akademie Verlag, 2013.Find this resource:

Bobonich, C. (ed.) Plato’s Laws: A Critical Guide. Cambridge: Cambridge University Press, 2010a.Find this resource:

Bobonich, C., and P. Destrée (eds.). Akrasia in Greek Philosophy: From Socrates to Plotinus. Leiden: E. J. Brill, 2007.Find this resource:

Bratman, M. “Intention, Practical Rationality, and Self-Governance.” Ethics 119 (2009): 411–443.Find this resource:

Brickhouse, T. C., and N. D. Smith. Socratic Moral Psychology. Cambridge: Cambridge University Press, 2010.Find this resource:

Brickhouse, T. C., and N. D. Smith. “Socratic and Platonic Moral Psychology.” In Ancient Ethics, edited by J. Hardy and G. Rudebusch, pp. 121–140. Goettingen: V&R Unipress, 2014.Find this resource:

Burnyeat, M. “The Truth of Tripartition.” Proceedings of the Aristotelian Society 106 (2006): 1–23.Find this resource:

Carone, G. “Akrasia in the Republic: Does Plato Change His Mind?” Oxford Studies in Ancient Philosophy 20 (2001): 107–148.Find this resource:

Cooper, J. (ed.) Plato: Complete Works. Indianapolis: Hackett, 1997.Find this resource:

Erginel, M. “How Smart Is the Appetitive Part of the Soul?” In Dialogues on Plato’s Politeia (Republic), Selected Papers from the Ninth Symposium Platonicum, edited by N. Notomi and L. Brisson, pp. 204–208. Sankt Augustin: Academia Verlag, 2013.Find this resource:

Ganson, T. “The Rational/Non-Rational Distinction in Plato’s Republic.Oxford Studies in Ancient Philosophy 36 (2009): 179–197.Find this resource:

Gopnik, A., C. Glymour, D. M. Sobel, L. E. Schulz, T. Kushnir, and D. Danks. “A Theory of Causal Learning in Children: Causal Maps and Bayes Nets.” Psychological Review 111, no. 1 (2004): 3–32.Find this resource:

Irwin, T. Plato’s Ethics. Oxford: Oxford University Press, 1995.Find this resource:

Kahn, C. “From Republic to Laws: A Discussion of Christopher Bobonich, Plato’s Utopia Recast.Oxford Studies in Ancient Philosophy 26 (2004): 337–362.Find this resource:

Kahn, C. Plato and the Socratic Dialogue: The Philosophical Use of a Literary Form. Cambridge: Cambridge University Press, 1996.Find this resource:

Kolodny, N. “Why Be Rational?” Mind 114 (2005): 509–563.Find this resource:

Kolodny, N., and J. Brunero. “Instrumental Rationality.” In The Stanford Encyclopedia of Philosophy (Spring 2016 edition), edited by E. N. Zalta. this resource:

Liebert, R. “Pity and Disgust in Plato’s Republic: The Case of Leontius.” Classical Philology 108 (2013): 179–201.Find this resource:

Lorenz, H. The Brute Within: Appetitive Desire in Plato and Aristotle. Oxford: Oxford University Press, 2006.Find this resource:

Lorenz, H. “Desire and Reason in Plato’s Republic.Oxford Studies in Ancient Philosophy 27 (2004): 83–116.Find this resource:

Moss, J. “Appearances and Calculations: Plato’s Division of the Soul,” Oxford Studies in Ancient Philosophy 34 (2008): 35–68.Find this resource:

Notomi, N., and L. Brisson (eds.). Dialogues on Plato’s Politeia (Republic), Selected Papers from the Ninth Symposium Platonicum. Sankt Augustin: Academia Verlag, 2013.Find this resource:

Penner, T. “Plato and Davidson: Parts of the Soul and Weakness of Will.” Canadian Journal of Philosophy Supplement 16 (1990): 35–74.Find this resource:

Penner, T., and C. J. Rowe. Plato’s Lysis. Cambridge: Cambridge University Press, 2005.Find this resource:

Price, A. W. Mental Conflict. London: Routledge, 1995.Find this resource:

Price, A. W. Virtue and Reason in Plato and Aristotle. Oxford: Oxford University Press, 2011.Find this resource:

Schapiro, T. “The Nature of Inclination.” Ethics 119 (2009): 229–256.Find this resource:

Shorey, P. Plato. The Republic. Cambridge, Mass.: Harvard University Press, 1978.Find this resource:

Slings, S. R. Platonis Rempublicam. Oxford: Oxford University Press, 2003.Find this resource:

Smith, M. “Instrumental Desires, Instrumental Rationality.” Proceedings of the Aristotelian Society Supplement 78 (2004): 93–109.Find this resource:

Smith, N. “Return to the Cave.” In Plato’s Republic: A Critical Guide, edited by M. McPherson, pp. 83–102. Cambridge: Cambridge University Press, 2010.Find this resource:

Stalley, R. “Plato’s Argument for the Division of the Reasoning and Appetitive Elements in the Soul.” Phronesis 20 (1975): 110–128.Find this resource:

Todd, S. C. “How to Execute People in Fourth-Century Athens.” In Law and Social Status in Classical Athens, edited by V. Hunter and J. Edmondson, pp. 31–51. Oxford: Oxford University Press, 2001.Find this resource:

Whiting, J. “Psychic Contingency in the Republic.” In Plato and the Divided Self, edited by R. Barney et al., pp. 174–208. Cambridge: Cambridge University Press, 2012.Find this resource:

Wilberding, J. “Curbing One’s Appetites in Plato’s Republic.” In Plato and the Divided Self, edited by R. Barney et al., pp. 128–149. Cambridge: Cambridge University Press, 2012a.Find this resource:

Wilburn, J. “Akrasia and the Rule of Appetite in Plato’s Protagoras and Republic.” Journal of Ancient Philosophy 8 (2014a): 57–91.Find this resource:

Wilburn, J. “Is Appetite Ever Persuaded? An Alternative Reading of Republic 554c-d.” History of Philosophy Quarterly 31 (2014b): 195–208.Find this resource:

Wilburn, J. “Courage and the Spirited Part of the Soul.” Philosophers’ Imprint 15 (2015): 1–21.Find this resource:


(1.) Carone (2001). Wilburn (2014a) holds that the Republic neither clearly accepts nor rejects strict akratic action nor sees the issue as important. A major motivation for Carone’s view is 505D5-506A2, especially 505E1-2, which she reads as saying that all people always act on their judgment of what is overall best (2001, 133–134). Price’s (2011, 274–277) arguments against this reading are persuasive.

(2.) By “deliberately” I mean, as a first approximation, that S realizes that she is doing Y and that her doing Y is not accidental.

(3.) For others, see Bobonich (2002, 267–273).

(4.) NE 1145b22-35, 1147b13-17, MM 1200b25-32.

(5.) For Aristotle’s interest in the differences between Socrates’ and Plato’s views about nonrational motivations, see, for example, EE 1216b2-10, NE 1144b17-30, MM 1182a15-30.

(6.) Hereafter Stephanus numbers without a textual reference refer to Slings’s (2003) edition of the Republic.

(7.) It occurs only three other times (Meno 78A3, A5, Symp. 173D1) and always means “unhappy” rather than being a term of purely ethical condemnation.

(8.) On Leontius, see Liebert (2013) and Todd (2001, 48–49).

(9.) It is too implausible that the Reasoning part would only make an overall judgment about what is best for the Appetitive part and not the rest of the soul.

(10.) Wilburn (2014a, 71–73).

(11.) It is unclear whether Plato thinks that the perfected philosopher (who would be an extreme case of hydraulic channeling) is still liable to akratic conflict and action. At any rate, philosophers who have seen some Forms, although they may not have grasped the Form of the Good fully, still seem to be liable to akratic conflict and action (Rep. 539E5-540A2). (Smith [2010] even holds that the philosophers of Rep. 539E-540A have already directly apprehended the Form of the Good.)

(12.) Bobonich (2002, 273–282).

(13.) As has long been recognized by some scholars, for example, Price (1995).

(14.) Many different stylometric studies converge in placing the Phaedrus relatively close to the Republic, see Kahn (2004, 47–48).

(15.) Brickhouse and Smith (2010, 137–138) cite Phdr. 253E-254C, but anagkazei at 254A5 and anagkazomenô at 254B1 may be conative so that we do not have strict akratic action. A better passage is 254C3-E5; ênagkasen at 254D5 does imply strict akratic action.

(16.) See Bobonich (2002, 1–88). The Phaedo, at least explicitly, does not partition the soul, but rather partitions the agent into body and soul; see Bobonich (2002, 27–28).

(17.) On puzzles connected with moderation, see Bobonich (2013). Courage is characterized as when the Spirited part “preserves through pains and pleasures the order handed down [paraggelthen] by reason as to what is to be feared and not” (442C1-2). Wilburn (2015) thinks that what Spirit preserves is not its own possession of the order handed down by reason, but reason’s own possession of true rational accounts.

  1. ((1)) This produces a complete disanalogy with courage as the virtue of a city: the city is courageous because the auxiliaries preserve the beliefs handed down (parêggeilen, 429C2) by the lawgiver about what is to be feared. The auxiliaries most decidedly are not responsible for ensuring that the philosopher-rulers maintain their knowledge. Wilburn’s reply (2015, 9n29) that the city-soul analogy has limitations is ad hoc.

  2. ((2)) It is reasonable for the Spirited part to preserve in itself reason’s orders; it is much less satisfactory for it to try to do this for the Reasoning part. A virtuous Reasoning part has a much wider range of knowledge of good and bad than it hands down to the Spirited part, and this knowledge is not held in the form of simple orders but includes explanation, which is not part of the order. A virtuous Reasoning part needing help preserving the relatively simple commands given to the Spirited part would be in extremis and the Spirited part, limited to preserving reason’s commands, could not help preserve what is most important for, and distinctive of, the Reasoning part.

(18.) Whiting (2012) denies this; for criticism of Whiting, see Wilburn (2014a, 65n18).

(19.) Philosophers from Aristotle to Davidson have been puzzled by it and have worked on it. Separating a desire’s causal force from the evaluative ranking of its object is a deep idea and is central to the best explanation of akratic action we have. We do not need partitioning for this, but I do not find that theory as implausible as some scholars, and the earlier explanation is presented without partitioning in the Laws. Compare Burnyeat (2006). On partitioning’s explanatory value, see Bobonich (2002, 239–247).

(20.) For the idea that we have reason to be instrumentally coherent as such, see Bratman (2009); for skepticism about the normativity of instrumental coherence, see Kolodny (2005). For an overview of instrumental rationality, see Kolodny and Brunero (2016).

(21.) Negative: Kahn (2004), Lorenz (2006, 44–52), Penner and Rowe (103n.16), and Wilberding (2012a). Positive: Bobonich (2002, 244), Brickhouse and Smith (2010, 203), Irwin (1995, 214–222), Moss (2008, 65), and Price (2011, 166–174). Kahn (2004) and Wilberding (2012a) deny propositional contents to the Appetitive part in the Republic; Bobonich (2002), Bobonich (2010b), Brickhouse and Smith (2010), Irwin (1995), Lorenz (2006, 59–73), Moss (2008), and Price (2011) do not.

(22.) For the Appetitive part desiring money, see 553C, 575A, 581A (compare with 436A), 581CD.

(23.) Lorenz (2006, 44, 47–48) takes this to be equivalent to their lacking logismos; on logismos, see 603A4-5, 604D4. In his introduction, Lorenz asserts that if a soul part can engage in means-end reasoning, then it will be liable to conflict that licenses the application of the Principle of Contraries and thus forces its subdivision (2006, 11). But Lorenz gives a very weak interpretation of the Principle of Contraries: he endorses “the Simple Picture” on which the requisite contrariety is “the simultaneous occurrence of a desire and an aversion towards one and the same object” (2006, 41). Even leaving aside means-end reasoning, this will license employing the Principle of Contraries when, for example, I have an appetitive desire for some Scotch and an appetitive aversion to its taste (compare with Bobonich [2010b, 161n23]). It is not clear, however, that stronger interpretations of the Principle of Contraries inevitably lead to subdividing parts that are capable of means-end reasoning. (If means-end reasoning in a part results in its subdivision, how do we avoid subdivision of the Reasoning part?) If, for example, the Principle of Contraries requires opposition between a first-order desire and a higher order desire, it is not at all clear that even means-end reasoning about final ends will inevitably produce such conflicts. (This is not to deny that the higher order desire interpretation of the Principle of Contraries has its own problems, compare with n51). In any case, Lorenz does not return to this objection in his more extended discussion of means-end reasoning (2006, 48–52) and in the earlier text I consider these other objections.

(24.) Lorenz (2006, 44, 47–48).

(25.) Wilberding (2012a, 134–135); compare with Lorenz (2006, 47).

(26.) Price (2011, 170).

(27.) It may also generate appetitive desires for the democrat’s objects.

(28.) Lorenz (2006, 59–73, 90, and especially 93).

(29.) How much falls under the notion of a means to an end? What is necessary for it, what is sufficient for it, what constitutes it, what facilitates it, what prevents its prevention? Is the Appetitive part going to lack concepts of all these notions as well?

(30.) For example, Gopnik et al. (2004).

(31.) Lorenz (2006, 47–51).

(32.) Lorenz (2006, 50).

(33.) Lorenz argues (2006, 99–110) that something like these representations can be found at Phil. 32B-36C, 38E-40C, and Tim. 71AE. I cannot discuss these passages here, but it is not at all clear that any of these representations has the function of specifying a particular plan or series of actions to be taken as do the representations in the cigarette example.

(35.) Perhaps we should allow a person to act on novel representations generated by reason or imagination (phantasia). Leaving aside the worries of why reason would share such representations with the Appetitive part or whether phantasia is capable (especially in the Republic) of generating such representations, the other problems remain, for example, why the Appetitive part would not pursue very inefficient means.

(36.) Lorenz’s position (2004, 114n51) differs from the more familiar idea that instrumental desires are nothing more than noninstrumental desires and the means-end beliefs that explain them; see Smith (2004, 96).

(37.) I have changed a question into an assertion.

(38.) The Spirited part seems capable of means-end reasoning (590B7-8).

(39.) For appetitive desires as “blind drives,” see Anagnostopoulos (2006, 173) and Penner (1990, 52, 59); as good-independent, see Irwin (1995, 205–211).

(40.) Kahn (2004, 355).

(41.) Wilburn thinks that the recommended management of the appetites is the same in the Republic as it is in the Timaeus and the Phaedrus (2014b, 200). In the Phaedrus, the charioteer who represents the Reasoning part “employs aggressive tactics of violence and force against the bad horse [who represents the Appetitive part], and, when the bad horse finally submits, it does so out of ‘fear.’ Socrates’ comments even seem to rule out the possibility of persuasion … It is also important to note that this is not a characterization of how vicious or inferior souls control their appetites… [it is a characterization of] the philosopher himself” (2014b, 201). In the Republic, according to Wilburn, Plato “depicts appetite being dragged, pulled, forced, or frightened away from an undesirable course of action, but never simply persuaded” (Wilburn 2014b, 202).

(42.) Wilburn (2014b) does not consider this possibility.

(43.) One way to avoid the tension is with Carone (2001, 118–121) to see the claim that unqualified thirst is just for drink as specifying the proper or formal object of any thirst, including thirst for hot drink, much drink, or even good drink.

(44.) I have changed a question into an affirmation.

(45.) One might object that necessary desires cannot be eliminated, but I am not suggesting that, for example, the desire for healthful eating be eliminated, but only some of the desires derivative from it.

(46.) This is not to suggest that harmony and rhythm do not affect the content of people’s desires, but only that this effect is more obvious in the case of the words. According to Wilburn, both unnecessary and necessary desires can only be managed by force and fear (2014b, 202). But we should not accept this claim.

  1. ((1)) This would involve a radical, and unremarked on, disanalogy with the ideal city, since the producers are certainly not held down by force and fear.

  2. ((2)) Wilburn notes how well his account fits with the Timaeus, which denies belief to the Appetitive part (2014b, 200). But this should be a cause for concern since the Republic attributes belief to the Appetitive part and is thus open to affecting it and its desires by changing belief. Such a process seems to be exactly what is intended by several passages in the Republic, for example, 431C quoted earlier. Simple and moderate desires that are directed by calculation (logismôi agontai, 431C6) are restricted to the few and best educated. The desires of the many and the bad (phauloi) are “mastered” (kratoumenas, 431C10).

  3. ((3)) Musical education is primarily, but not exclusively, directed toward the Spirited part. At 389DE, Socrates commends certain poetry for conducing to making people rulers over their own appetitive pleasures. As other poetry, via its content, has good effects on the Spirited part, this sort of poetry, via its content, has good effects on the Appetitive part. Wilburn’s view cannot make good sense of Plato’s worries about the effect that the words of tragedy or Homer have upon the Appetitive part and Plato’s efforts to provide better propositional content in the reformed poetry allowed.

(47.) On this link between means-end reasoning and efficiency, see Irwin (1995, 219).

(48.) See Bobonich (2002, 191–194).

(49.) The Appetitive part cannot possess the Protagoras’ measuring art nor is it likely to have a fully temporally neutral view of future pleasures and pains, but neither is necessary—pace Price (1995, 64)—for judging one course of action more pleasant than another taking into account as much of the future as it can as impartially as it can.

(50.) We should also include in the good of the Appetitive part whatever condition it must be in for the person to possess the virtues, for example, agreeing that the Reasoning part should rule.

(51.) Irwin persuasively argues that previous attempts to understand the Principle of Contraries in terms of (1) non-co-satisfiable desires, (2) desires such that one includes an aversion to the object of the other, and (3) desires such that one includes a higher order aversion to the other desire itself all fail (1995, 205–206). I argue that his own account (1995, 205–217) also fails (2002, 248–254).

(52.) Irwin denies that the Appetitive part conceptualizes desires as for the good but sees it as ranking them in terms of strength (1995, 220). This attempts to avoid conflict between desires for the good and other desires in the Appetitive part that could lead to splitting. We have, however, textual evidence for pleasure as the object of appetitive desire, but not for a ranking in terms of strength, and I have argued that it is unclear exactly how the Principle of Contraries is to be understood and that Irwin’s way does not avoid splitting in the lower parts. If drink is also taken as a good-independent ultimate end, pleasure may differ in that it is not only seen as an ultimate end but also as immediately appealing and attractive rather than, for example, pressing and needful.

(53.) To simplify matters, I focus on pleasure, compare with n50.

(54.) Moss (2008, 63).

(55.) Moss (2008, 63).

(56.) Moss (2008, 63).

(57.) Literally, the person judges, but along with many commentators I take Plato to be referring to the Reasoning part. The more that is built into the Appetitive part’s notion of the good, the more likely it is to face an analogue of akrasia and thus the danger of splitting. On Spirit’s value judgments, see, for example, Irwin (1995, 211–213).

(58.) Lorenz (2006, 44–52).

(59.) Ganson (2009, 188–191).

(60.) Ganson (2009, 191).

(61.) I accept the interpretation of 602E4-6 first proposed by James Adam and defended at length by Lorenz (2006, 55–71).

(62.) Ganson seems to suggest this with respect to judgments of betterness (2009, 191).

(63.) For the Spirited part, see 390D. At 553D, the Reasoning part is set to calculate which option maximizes money, not that, for example, two talents are greater than one.

(64.) For sensible versions of number concepts, see 525D5-E3, 522D1-523A3; compare with Phil. 56D1-57A2.

(65.) Ganson (2009, 195–196).

(66.) Moss (2008, 57).

(67.) Moss (2008, 65).

(68.) Moss (2008, 39).

(69.) Moss (2008, 58).

(70.) But not, perhaps, the good, Moss (2008, 65).

(71.) Bobonich (2002, 298–326).

(72.) Moss (2008, 59n50).

(73.) Phd. 68B8-C3, 79C2-8, 81B1-C2, 82B10-C8, 83D4-E3, 93C9-E6.

(74.) Their concepts can only have sensible content, although the reference of their concepts may not be determined by what is “in their heads,” compare with Bobonich (2002, 310–311). The lower parts seem able to think with some degree of coherence; for example, they do not endorse contradictions as unproblematic. It fits with this that they can engage in some simple calculation using sensible concepts. The Spirited part is likely to be more sophisticated and, for example, better able to deal with long-run considerations.

(75.) I remain neutral on whether the Republic allows belief about nonsensibles and thought or understanding about sensibles insofar as they have nonsensible properties. If we accept these claims, then the unique resources of the Reasoning part will also include them.

(76.) This remains true even if the correct ultimate account of the virtues is mathematical. On the other hand, certain formal features of inquiry may rely on a kind of counting. Inquiry might involve, for example, establishing that its subject is “one”; compare with Rep. 524B and Tht. 185B.

(77.) First, I must leave aside here the origin of such concepts in the Republic and in particular whether it accepts the theory of Recollection and what the relation is between philosophic and mathematical knowledge. Second, is it necessary that one appreciate that one is dealing with nonsensible concepts and properties? A crucial moment in the development of rationality in the Republic is the recognition that perception is insufficient to answer some questions: 523A10-524C14, especially 523D8-9, 524B3-5, and 523C10-11. So perhaps what is necessary is a recognition of the insufficiency of perception in certain cases.

(78.) For a recent account of these and their relations in some early and middle dialogues, see Benson (2015).

(79.) It may also require some concern with other people; compare with Bobonich (2002, 450–473).

(80.) Lorenz (2006, 44n9).

(81.) The myth of metals may suggest that this potential is not actualizable in every incarnation.

(82.) I draw here on Schapiro (2009, 232–233).

(83.) I leave aside here puzzles about how this identification is supposed to work. I would like to thank Alan Code, Grant Dowling, Huw Duffy, Amos Espeland, Landon Hobbs, Rachana Kamtekar, Roy Lee, Katy Meadows, and two anonymous referees for their comments on an earlier version of this article.