How to Individuate the Powers of Knowledge and Opinion in Plato ’ s Republic V

At the end of Republic V Socrates argues that differences between knowledge and opinion justify rule by philosophers in his ideal city; within this argument he gives a theory of powers. The theory contains a logical gap: Socrates mentions two criteria by which he individuates powers, yet assumes in his argument that both criteria will speak with a single voice. I argue that these criteria are two  ways  of  picking  out  a  power’s  manifestation—that is, the change a power is directed toward; I shall call this view the Identity Reading. Since they refer to a single phenomenon, the results of consulting both criteria cannot differ. The Identity Reading both solves the logical problem with the argument and sheds light on four other features of the passage. This reading also provides support for what  Gail  Fine  has  called  the  “contents  analysis”  of  a  power’s  relation  to  its  “objects,”  as  opposed  to the  “objects  analysis.” Toward the end of Republic V Socrates argues that the differences between knowledge and opinion justify rule by philosophers; within this argument he gives a theory of δυνάμεις,   or   “powers.” The theory contains a logical gap: Socrates says he looks at two criteria in order to individuate powers, generating four possible outcomes. Yet he assumes in his argument that both criteria will speak with a single voice, leaving only two possible cases. If this is a logical error, it passes unremarked and unjustified. But if it is not a mistake it has yet to be explained convincingly. I give a new account drawing on contemporary metaphysics: what Socrates looks at to individuate powers are two ways of picking out a power’s  manifestation — the change a power is directed toward. 1 I  use  “opinion”  as  equivalent  to  δόξα,  but  without  endorsing this against the competing translation “belief.”   For   recent   work   translating   δόξα   as   “belief,”   see   C.   J.   Rowe,   Plato and the Art of Philosophical Writing (Cambridge: Cambridge University Press, 2007); Katja Maria Vogt, Belief and Truth  :  A  Skeptic Reading of Plato (New  York  ;;  Oxford:  Oxford  University  Press,  2012);;  Theodore Scaltsas,  “Knowledge  as  ‘True  Belief  Plus  Individuation’  in  Plato,”  Topoi 31.2 (November 17, 2012): 137–49;;  opting  for  “opinion”  are     Grube,  G.  M.  A.  (translator,   revised  by  C.  D. C. Reeve), Plato’s Republic (Hackett  Publishing  Company,  1992);;  Joel  A.  Martinez,“Rethinking  Plato’s  Conception  of Knowledge: The Non-Philosopher   and   the   Forms,”   Apeiron 44.4 (2011): 326–334. I call this a “theory”  of  powers  on  the  grounds  that  Socrates  makes  three  important  and  connected  claims  about powers within a few lines: (1) powers exist; (2) powers are pervasive (given that every ability of every existing thing is said to depend on a power); (3) he gives a method for individuating powers. In short, the claims here are not likely to have arisen without deep reflection on the nature and epistemology of powers. Other texts suggesting that Plato had at least a moderately sophisticated theory of powers (whether or not he meant to endorse this theory) include Phaedrus 270c9–d7 and Sophist 247e3–4. Journal of Ancient Philosophy ISSN 1981-9471 FFLCH/USP www.revistas.usp.br/filosofiaantiga J. anc. philos. (Engl. ed.), São Paulo, v.8, n.2. p. 92-115, 2014. DOI: http://dx.doi.org/10.11606/issn.1981-9471.v8i2p92-115

I begin by explaining the puzzle and its centrality to the argument.I next argue that what the two features refer to is the manifestation of a power.This view raises further questions: how can Socrates' formula "being epi x" be a way of referring to a power's manifestation, and why should Socrates use two phrases if both refer to the same phenomenon?Finally, I show what is gained from adopting this reading.
Readers may be especially interested in the consequences of my argument for the overall argument about knowledge and opinion in this passage.The view I argue for is a version of what Gail Fine calls the "contents analysis" as against an "objects analysis": that is, to say that knowledge is epi x is to say that x is the propositional content of the knowledge in question, rather than to say that knowledge is related to an object in the world. 2 This is because the objects analysis applied to the epi relation says that for a power to be epi x is for it to be related to an entity distinct from the power itself (and distinct from its manifestation).
As I shall argue, the numerical distinctness of a power (and its manifestation) from what it is epi makes the argument in the text question-begging.The contents analysis does not require this kind of distinctness, so the view I shall argue for, the Identity Reading, supports it.
Finally, it is widely recognized that of the three competing ways in which "is" can be read in this passage, the existential and predicative senses are natural allies of the objects analysis, while the veridical sense goes with the contents analysis.It follows that the Identity Reading is committed to the veridical reading of "is" (in most places), and to the contents analysis. 3e Passage Near the end of Book V Socrates proposes that philosophers should rule, explaining that philosophers are those who love knowledge.As a counterexample, Glaucon describes a group of "sight-lovers," who seem to love some forms of knowledge; Socrates counters that philosophers love the truth (475e4) and they must therefore love the forms. 4The sight-lovers do not even acknowledge the existence of the forms, so they fail to qualify as philosophers by this criterion (475e2-475d6).
One might think the argument has now ended successfully.But Socrates embarks on a second argument for the same conclusion, this time addressed to the sight-lovers themselves (476e4-480a13).This argument shows that since knowledge and opinion are different powers, they are necessarily epi different things; therefore what can be known is different from what one can hold an opinion about.In this argument Socrates must eschew claims the sight-lovers would not accept, including appeals to the existence of forms. 5crates interrupts this second argument to propose a theory of powers with two main claims.First, powers are real and are what enable us to do whatever we are able to do.
Second, he explains how to differentiate one power from another as follows: A power (δυνάμεως) has no color for me to see, nor a shape, nor any feature of the sort that many other things have, and that I can consider in order to distinguish them for myself as different from one another.In the case of a power, I look only at (εἰς ἐκεῖνο μόνον βλέπω) [1:] what it deals with (εφ' ᾧ ἔστι) and [2:] what it does (ὃ ἀπεργάζεται), and it is on that basis that I come to call each the power it is: those assigned to deal with the same things and do the same things, I call the same; those that deal with different things and do different things, I call different.(477c6-d5, Reeve's translation modified) 6 The Problem The individuation of powers is central and crucial to the argument.One of Socrates' central claims is that knowledge and opinion are two distinct powers, and hence that they are properly distinguished from one another.He shows that the sight-lovers do not have knowledge by showing that they do have opinion, that knowledge and opinion are distinct powers, and that knowledge has a requirement the sight-lovers cannot meet. 5This reading of the dialectical requirements of the argument is widely accepted, with adherents including J. C. Gosling, "Δόξα and Δύναμις in Plato's Republic," Phronesis 13.2 (1968): 119-130, at 121; Julia Annas, An Introduction to Plato's Republic (Oxford University Press, 1981); Gail Fine, "Knowledge and Belief in Republic V-VII," 217.For disagreement see Vogt,Belief and Truth,[58][59][60][61][62] δυνάμεως γὰρ ἐγὼ οὔτε τινὰ χρόαν ὁρῶ οὔτε σχῆμα οὔτε τι τῶν τοιούτων οἷον καὶ ἄλλων πολλῶν, πρὸς ἃ ἀποβλέπων ἔνια διορίζομαι παρ΄ ἐμαυτῷ τὰ μὲν ἄλλα εἶναι, τὰ δὲ ἄλλα· δυνάμεως δ'εἰς ἐκεῖνο μόνον βλέπω [1:] εφ' ᾧ τε ἔστι καὶ [2:] ὃ ἀπεργάζεται, καὶ ταύτῃ ἑκάστην αὐτῶν δύναμιν ἐκάλεσα, καὶ τὴν μὲν ἐπὶ τῷ αὐτῷ τεταγμένην καὶ τὸ αὐτὸ ἀπεργαζομένην τὴν αὐτὴν καλῶ, τὴν δ'ἐπὶ ἑτέρῳ καὶ ἕτερον ἀπεργαζομένην ἄλλην. Geek quotations are from Plato, Platonis Rempublicam, ed. S. R. Slings (Oxford University Press, 2003).Translations from the Republic are from C. D. C. Reeve,  Republic, (Indianapolis, Hackett, 2004), unless otherwise noted.
Socrates looks at two things to individuate powers: the first is εφ' ᾧ ἐστι ("what a power is epi" or "deals with") and the second is ὃ ἀπεργάζεται ("what a power does").I shall later argue that these two phrases refer to the same entity, but let us begin by assuming that they refer to two distinct phenomena.Looking at two non-identical things gives four possible results: 1.The powers are epi the same thing and do the same thing.
2. The powers are epi different things and do different things.
3. The powers are epi different things but do the same thing.
4. The powers are epi the same thing but do different things.
But in practice Socrates assumes that only (1) and (2) are possible.He is therefore assuming the following Constraint: Constraint: Distinct powers never do the same things or deal with the same things.This Constraint is equivalent to saying that only (1) and ( 2) are possible, and that (3) and (4) cannot occur. 7 know that the Constraint is part of the argument for the following reasons. 8First, Socrates' initial statement that he looks at two things mentions only cases (1) and (2).But second and more importantly, Socrates' argument about knowledge and opinion depends on the Constraint.Socrates first establishes that knowledge and opinion are different powers, by showing that they do different things (first at 477a-b, then again at 477d-478a).He then infers that since they are different powers, they both do different things and are epi different things (478a1-5).Glaucon makes the reliance on the Constraint explicit: Socrates asks him whether what can be known and what can be opined might be the same.He answers that this is impossible (ἀδύνατον), since different powers are always epi different things (478a13-b2).
It is important to note the modal nature of Glaucon's claim.It is not merely the case that 7 One might think that the Constraint implies not that (3) and ( 4) are impossible, but the weaker claim that neither (3) nor (4) ever in fact occur.I shall show below why the stronger version is required. 8The Constraint is recognized by Crombie, An Examination of Plato's Doctrines (New York,  Humanities Press, 1962), 57-8;; Jaakko Hintikka, "Knowledge and Its Objects in Plato," in Patterns in Plato's Thought, J. M. E. Moravcsik (ed.) (Dordrecht, D. Reidel, 1973), 7-18, 15; Gerasimos Santas, "Hintikka on Knowledge and Its Objects in Plato," in Patterns in Plato's Thought, J. M. E. Moravcsik (ed.) (Dordrecht, D. Reidel, 1973), 31-51, 44 knowledge and opinion happen to be epi different things, in his view.Rather, no two powers can be epi the same thing-that is, case four above is ruled out for all powers.Glaucon reasons from the general fact that no two powers are epi the same thing to the particular claim that knowledge and opinion are not epi the same thing.Note that case (3) has not been ruled out, but absent some reason for thinking it is a possibility we are justified in assuming that it is also excluded, since as a mixed case nothing distinguishes it logically from case (4).The Constraint is puzzling, since at first glance the two features Socrates looks at seem to be two distinct entities.As Rachel Barney writes, "Plato seems to offer one sufficient condition too many…"9 Note that Socrates' remarks about the individuation of powers would create no puzzle without the Constraint.Socrates is of course free to look at as many criteria as he might wish to for individuating powers, and he might propose any procedure for making a judgment about whether two (or more) putative powers are in fact the same power.

It follows from the
His view is only puzzling because he seems to posit two separate criteria, but then treats them as if they were in fact a single criterion.
Since the use Socrates actually makes of the criteria is clear, the only scope for solving the puzzle lies in thinking about what the criteria refer to.Crombie, Gosling, and Hintikka all saw that the text requires that what a power is epi and what it does be related in some intimate way that prevents them from varying independently. 10Crombie proposes that what a power is epi is an "internal accusative," that is, the power's own manifestation as named by a verbal formula similar to the power's own name.Thus, for example, the power of opinion is epi opinions, and the power of sight would be epi things seen.On Gosling's view, the two criteria refer to more or less the same phenomenon, although his description of this referent is not entirely clear.According to these readings the two criteria are either redundant or unclear: redundant if they refer to exactly the same thing, unclear if there is some difference we cannot discern.I will defend an account similar to those of Crombie and Gosling, but I shall both argue more carefully for the view and be more definite about what the two criteria refer to.

Why we cannot avoid the Constraint
Two initially plausible thoughts about the two criteria, both of which would deny the Constraint, either fail to fit the text or render the argument question-begging.
First, overlooking the Constraint, one might suppose that for a power to be epi x is just for it to exercise its power on x, much as one might use a tool on an object.11Just as it is possible to hit a wide range of things with a hammer, then, one might suppose that a typical power can be brought to bear on a large range of objects, so that the nature of the power in question plays little or no role in determining what these objects are.To say that a power is epi some object just means that this power is being brought to bear on that object, without implying any closer connection between power and object.This reading renders the argument question-begging.For if a power can be epi various objects, then Socrates has no reason for thinking that knowledge and opinion are epi different things just because they are different powers-yet this inference is central to his argument.I am not aware of anyone who holds this view in the strong form I am describing, but it is worth setting out as an extreme against which the next, more plausible view can be compared.Further, many authors seem to read the epi relation in ways that fall somewhere between this view and the next to be described: these authors probably would not endorse the extreme liberality of the epi relation given by this first view, but neither do their proposals restrict the relation between powers and objects to a one-to-one relation, as the following view does.
The second view is as follows.It recognizes that what a power is epi and what it does cannot be completely independent of one another.But, this line of thought goes, we do not need a relation as strong as identity between the two criteria; what we need instead is some natural or typical connection between what the power does and what it is epi. 12I am unaware of a published version of precisely this idea.But many scholars say in passing that the two things Socrates looks at are related in a natural way, without making clear exactly what connection they see or how they think this renders the argument valid.These views fall somewhere between these two positions; these authors probably do not think the two things are entirely independent of one another, but in most cases they also do not claim explicitly that a natural connection between them will deliver the one-to-one pairing for all powers that is required by the argument.

For example, Stokes proposes a natural pairing of powers and "provinces" (his term
for what a power is epi): Now sight, hearing and the other senses are capacities each distinguished from the others by both province and effect.Notoriously you cannot see sounds, smell colours, hear tastes, touch smells; each sense has in that way its own province. 13anting Stokes' point about the physical senses, nothing at all follows from this about the behavior of knowledge and opinion.If Socrates' argument rests on hoping the sight-lovers will accept this analogy between knowledge, opinion, and the physical senses, it is a very bad argument. 14ca Boylu makes it a matter of definition that what a power is epi is something able to have done to it what that power does. 15She thus wants to preserve the claim that powers are epi things that are distinct from the powers themselves.But this makes the Constraint a mystery, and it is unclear why Glaucon, let alone the sight-lovers, should or would accept this claim from Socrates.It is better to read the epi relation in way that allows us to understand why it seems immediately obvious to the sight-lovers.
Katja Maria Vogt proposes that knowledge and opinion are naturally directed toward "that which fully fits and reflects their natures." 16On her view, a power can also be directed toward things it is not fully fitted for, but it is only said to be epi what is most appropriate for it.This view, too, fails to give Socrates what his argument needs.Socrates and Glaucon argue for the claim that opinion and knowledge do different things, and are therefore different 12 I am grateful to an anonymous referee for this example, and to more than one referee for suggesting variations of this idea. 13Stokes, "Plato and the Sight-Lovers of the Republic," 119-20. 14Gonzalez follows Stokes on this point ("Propositions or Objects," 264). 15Boylu, "The Powers Argument in Plato's Republic," 120. 16Vogt, Belief and Truth, 64.powers.But they say nothing to show that different powers must always be epi different things, except in giving the individuation conditions for powers.So the individuation conditions themselves must provide some grounds for accepting the claim; if they do not, the argument is question-begging.To claim that knowledge and opinion are naturally fitted to different things cannot show that this must be the case; yet it is the stronger claim that Glaucon relies on (478a13). 17 On this second type of view, then, there is at least a typical, or better, a natural relation between each power and what it is epi, and this is enough to secure a one-to-one relation between powers and their objects.One might say, for example, "I care only about this, how many assets she has and what her purchasing power is." 18 Assets and purchasing power are distinct entities, but naturally correlated.If these were someone's criteria for individuating purchasing power, one could reasonably expect that the two things mentioned would give the same guidance in all or nearly all situations.
This reading seems to have some merits in fitting the text.On one hand, the two apparently distinct criteria Socrates mentions are kept distinct by this reading: Socrates really is talking about two things, what a power does and what it is epi.On the other hand, the reading can explain why the two things Socrates looks at speak with one voice in deciding whether powers are the same or distinct: if the two things are naturally associated with one another, looking at either is likely to bring us to the other as well.This explains why Socrates pays no attention to cases (3) and (4) in his argument.This view, however, is less successful than it first appears, for neither typical nor natural association are strong enough relations to justify excluding cases (3) and ( 4) in an argument about knowledge and opinion.The very least that Socrates needs is that knowledge and opinion are necessarily different from one another; that is, there must be no possibility that knowledge and opinion might be epi the same things.And even this requirement is probably too lenient, for it is more plausible that Glaucon and Socrates are appealing to a general necessity covering all powers, not one restricted to the special case of knowledge and  2012): 51-73.Smith's articles provide an order of magnitude greater sophistication than any of the others, in the following sense.While most views can be classified as either objects or contents analyses, Smith attempts to combine both types of view in a single account.But the increased sophistication does not change the basic point here. 18I am grateful to an anonymous referee for this example.opinion (478a13-b2).Socrates needs it to be the case that each power is necessarily epi something different from every other power, and a typical or natural connection cannot deliver this.Absent any further considerations that would show why the two features are necessarily paired as they are, the argument is open to easy, even obvious counterexamples.
Animal husbandry and butchery both seem to deal with animals, but to do different things with them. 19Medicine and gymnastics both deal with bodies (Gorgias 464b2-5a7). 20If this is the correct understanding of Socrates' two criteria, the argument is question-begging.For the counterexamples show that many distinct powers seem to deal with the same things, and Socrates provides no reason for thinking that knowledge and opinion do not form another case like husbandry and butchery.No merely natural pairing of what a power does with what it is epi will guarantee that knowledge and opinion are not epi the very same thing(s).And of course if Socrates is merely assuming this claim, the argument is question-begging.One might read the argument as question-begging, of course, but this is to be avoided if possible.Such a reading is available, as Crombie, Gosling, and Hintikka all realized.Their proposals gave a valid argument, but failed to explain how and why Socrates' two criteria could be read as referring to the same target.This difficulty can be solved by paying more attention to the fact that Socrates is discussing powers.

Socrates on Powers
In contemporary philosophy a power is a property whose nature is to enable whatever 19 Fine, "Knowledge and Belief in Republic V," 73. 20While medicine and gymnastics make an intuitively plausible example, note that Plato himself is careful in the Gorgias to avoid saying that medicine and athletics are two powers.Instead he describes them as a single craft with two parts.The same passage describes other practices that also deal with the body, that is pastry-baking and cosmetics, but these are called empirical "knacks" that do not qualify as crafts-and so, one suspects, they are also not genuine powers.This treatment is consistent with the view that Plato avoided commitment to distinct powers that would be epi the same things.
has it to bring about some change or to do something; the change it is directed toward is its manifestation. 21For example, the power of heating is the property enabling its bearer to raise the temperature of things in the vicinity; its manifestation is the increase in temperature. 22crates' δυνάμεις in this passage are recognizable as powers in the contemporary sense by their dispositionality: they are what enable their bearers to do whatever they are able to do. 23In the same way, the second "thing" he looks at to individuate powers, "what it does", seems to describe the power's manifestation. 24 the Identity Reading, saying that a power is epi x means that x is its manifestation.
Thus, when Socrates apparently gives two criteria for individuating powers, in fact both criteria refer to the same phenomenon.This is the reason that cases (3) and (4) cannot arise.
This solves the logical problem, but to make the reading convincing we need to show that the text supports this approach.In the following section I give three further arguments for this reading, and then consider three objections.

Three Reasons, Three Objections
The first reason favoring the Identity Reading is that Socrates introduces the two phrases by saying, "In the case of a power, I look only at that thing (ἐκεῖνο), what it deals with and what it does…" (477d1-2, Reeve's translation modified).The singular ἐκεῖνο refers to both phrases together, and so may indicate that they refer to a single thing.Further, Socrates' expression has the same structure as a hendiadys, the rhetorical trope by which a 21 See Vogt, Belief and Truth, 63.Of course, not all powers need bring about changes as opposed to properties.But since we are discussing Plato's metaphysics, and change (γένεσις) is for Plato the mark of becoming as opposed to the realm of Forms, it seems likely that Plato does conceive of all powers as bringing about changes.Contemporary thinkers are exploring a number of ways of thinking about powers in detail (for several of these views see Anna Marmodoro (ed.),The Metaphysics of Powers: Their Grounding and Their Manifestations, New York: Routledge, 2010); I am not assuming the correctness of any of these.
22 Alexander Bird, Nature's Metaphysics: Laws and Properties (Oxford University Press, 2007), 3. 23 Socrates says that powers are αἷς δὴ καὶ ἡμεῖς δυνάμεθα ἃ δυνάμεθα καὶ ἄλλο πᾶν ὅτι περ ἂν δύνηται, "that by means of which we are able to do what we are able to do, as well as everything else whatever it can do" (477c1-2, my translation). 24Is Socrates really thinking of δυνάμεις as properties?"Property" here means a way something is which gives it the ability to bring about some change: in this sense Socrates is discussing properties.Socrates later speaks of the power of sight coming to be in the eye, corroborating the view that he is thinking of powers as properties possessed by things (508a-b).Second, in what follows, "potential" and related terms are based on Socrates' heavy use of δύναμις and related words, combined with contemporary thinking about powers; these are not attempts to read Aristotelian concepts back into Plato.I am grateful to anonymous reviewers for raising these issues.
speaker uses two words or phrases to indicate a single thing, as in, "by length of time and siege," meaning simply "a long siege." 25Hendiadys typically places the more specific of the two items last, and Socrates' two phrases observe this pattern as well, suggesting that the best guide to the meaning of "what a power is epi" is "what a power does."These reasons are suggestive, but not decisive. 26e second reason favoring the Identity Reading depends on noting where each phrase occurs in the passage.Socrates speaks several times of knowledge, opinion, and even ignorance as epi this or that. 27He sometimes adds additional words beyond the epi + dative phrase, which seem to say what the powers in question do (ὃ ἀπεργάζεται).For example, "Now, doesn't knowledge naturally deal with what is (ἐπὶ τῷ ὄντι), to know how what is is? (γνῶναι ὡς ἔστι τὸ ὄν)" (477b11-12).Here "to know how what is is", seems to describe what the power of knowledge does. 28e first time Socrates says that knowledge is epi what is, he says, "Then, since knowledge dealt with (ἦν) what is, ignorance must deal with what is not…" (477a10).This is slightly puzzling: although this is the first time Socrates has used the epi + dative construction about knowledge, he is saying that he and Glaucon have already agreed that knowledge is epi what is.The puzzle is easily solved: the earlier statement is a few lines above, immediately after Socrates has proposed that they find something to say to the sightlovers.That exchange runs: 25 Herbert Weir Smyth, Greek Grammar, (Cambridge: Harvard University Press, 1920) citing Demosthenes 19.123, χρόνῳ καὶ πολιορκίᾳ.Traditionally, "hendiadys" refers to pairs of individual words, not entire phrases.See David Sansone, "On Hendiadys in Greek," Glotta 62.1-2(1984): 16-25; T. M. Robinson notes two instances of hendiadys proper in the Timaeus, at 29d7-e1 and 29e4 (Cosmos as Art Object, Binghampton, New York, Global Academic Publishing, 2004, 5 and 47 note 9). 26They are not decisive because the two phrases might also refer to two distinct phenomena: see further below.Stokes argues that ἐκεῖνο is singular in order to contrast with the colors, shapes, etc. mentioned in the previous sentence ("Plato and the Sightlovers of the Republic," 122 n.14), but this is inconclusive, since it does not show that its grammatical number fails to reflect the number of what it refers to. 27Socrates uses the epi + dative construction at 477a11, b8, b11, d2, d3, and d5, 478a4, a7, and 480a1, and Glaucon uses it at 478a13.For a defence of the claim that ignorance is a power in this passage see Smith, "Plato on the Power of Ignorance." 28Gosling agrees ("Δόξα and Δύναμις in Plato's Republic," 125).Two further examples of this pattern: ᾿Εφ' ἑτέρῳ ἄρα ἕτερόν τι δυναμένη ἑκατέρα αὐτῶν πέφυκεν;; ("Each of them, then, since it has a different power, deals by nature with something different?"478a4-5) and ᾿Επιστήμη μὲν γὲ που ἐπὶ τῷ ὄντι, τὸ ὄν γνῶναι ὡς ἔχει;; ("Surely knowledge deals with what is, to know what is as it is?" 478a7).Here ἕτερόν τι δυναμένη and τὸ ὂν γνῶναι ὡς ἔχει express what these powers do.Does someone who knows know something or nothing?…[H]e knows something.-Something that is or something that is not?(ὂν ἢ οὐκ ὄν) -That is.(476e6-77a1) These must be the lines that justify Socrates' statement a few lines below that knowledge "dealt with what is," for there is no earlier mention of "what is" or "what is not" in relation to knowledge.
Glaucon's answer, that one who knows must know something that is (476e6-477a1), is prompted by the verb "knows" and nothing else.Since what the power of knowledge does is knowing, the remainder of the phrase-"knowing something [that is]"-must be the part that designates what the power is epi.So knowing what the power of knowledge does (i.e.knowing) suffices for knowing what it is epi (i.e.something that is, or knowing something that is).Glaucon has therefore inferred from what knowledge does to what it is epi without the aid of any further information or prompting.Now if Glaucon understands a power's doing x and being epi y as referring to the same entity or activity, his inference is understandable and justified.If he thinks of the two criteria as referring to different entities, however, his reasoning is unjustified.
Note that this is a different point from the one made above, where I argued that according to Glaucon no two powers do or are epi the same things (478a13-b2).The former point was about a general claim applying to all powers.But the current argument concerns the inference Glaucon draws specifically from what knowledge does to what it is epi.
There are two ways Glaucon might draw this inference: he might have views specifically about knowledge which he uses to answer Socrates' questions, or he might be drawing an inference based on his general views about all powers.Suppose, first, that he is answering based on his prior beliefs about knowledge, so that no general views about powers are involved. 29In this case Glaucon's inference from "knowledge knows" to "knowledge is epi what is" is justified by this prior belief, but he has no justification for going on to infer that opinion must be epi something different from what knowledge is epi (477b8-10).On this reading, the sight-lovers should probably not accept Glaucon's answers as representing their views, for he has no principled reason for his answers, merely some particular beliefs that play directly into Socrates' hands.
It is therefore better to understand Glaucon as basing his answers on a general view about the nature of powers, in spite of the fact that Socrates has not yet articulated his method 29 Since Glaucon is answering for the sight-lovers, any beliefs he expresses about knowledge should also be attributed to them at this point.
for individuating powers.If Glaucon's answers are based on a general view about powers, it is plausible that the sight-lovers would accept his answers on the grounds that they express a common and plausible understanding of powers.This supposition is strengthened by the end of Socrates' speech about individuating powers: after explaining his method, he asks Glaucon, "What about you?How do you do it?" 30The present tense suggests that Glaucon already has a view about individuating powers, and Glaucon answers that he individuates them the same way. 31Finally, if Glaucon is reasoning based on a general view, we have already seen that the identity relation is needed to justify the move between what a power does and what it is epi.
The third reason favoring the Identity Reading involves a difference between states and powers.One may be in a state of happiness about various things, and conversely one may be in various states about the fact that today is Tuesday.Most states are not correlated one-toone with objects, but on many conceptions of powers a power and its manifestation are correlated one-to-one.This is because powers are typically individuated by nothing but their manifestations. 32Note that one might plausibly read Socrates as making just this point when he says that powers cannot be individuated by looking at colors or shapes, as one does with other entities (477c6-d1).The power of sight, for example, is the power that produces seeing as its manifestation; therefore any power producing this manifestation is the same power, and any power producing another manifestation is a different power.
If one starts by thinking of knowledge and opinion as states, it is easy to assume that they deal with external, independent objects.This thought leads naturally to the objects analysis, treating what a power does as numerically distinct from what the same power is epi.
Recognizing that knowledge and opinion are powers, on the other hand, we should expect that what each power is epi is not independent from what the power does.The third reason favoring the Identity Reading, then, is the suspicion that competing readings fail to take 30 τί δὲ σύ;; πῶς ποιεῖς;; (478d5-6, my translation).
31 This claim, of course, is defeasible.Socrates might be ironically inviting Glaucon or the sightlovers to claim a view that Glaucon has never thought about before.But Glaucon does not express surprise or confusion, as he often does in other parts of the Republic.I see no positive reason for thinking that Glaucon and the sight-lovers could not recognize in Socrates' words a view that they already hold.
sufficient notice that knowledge and opinion are powers and not merely states.
Three objections to the Identity Reading have been raised in recent scholarship, but none present serious difficulties.
First, Boylu rejects the Identity Reading on the grounds that it would commit Socrates to the existence of something that both is and is not as soon as he introduces the claim that opinion is a power that opines, while later passages show him still intending to argue for this claim (478e). 33On my reading, however, there is no such commitment.Instead, Socrates begins by identifying the power of opinion by what it does, producing opinion or opining.He then looks for what this power is epi, that is, its manifestation under a different description.
Second, Santas objects that at 477c6-d6, in describing his method of individuation, Socrates repeats his two phrases three times within one sentence.Santas takes this to show that two distinct referents must be intended. 34At most, however, the repetition shows that Socrates finds something significant or useful about using both phrases rather than just one; this falls far short of showing that they must refer to two distinct things.
Third, Ian Crystal argues that the use of τε…καί shows that there are two distinct referents, writing that this connective "…is standard Greek for expressing two distinct things.
It emphasizes their difference". 35This is one way an author might use τε…καί, but even granting that the phrase emphasizes some difference between the things connected, we still need to ask what difference this is.Since this is precisely the question at issue, Crystal's suggestion fails to rule out my reading.
Further, a nearby passage, 474d1-5e1, provides at least one -and probably more than one -counterexample to Crystal's claim. 36This passage pairs several verbal formulae, and in a few cases the two clearly refer to the same thing.Plato, then, uses these connectives to join terms referring to identical items as well as terms referring to numerically distinct items.It is a further question whether expressions thus connected refer to the same or distinct objects.
The objections to the Identity Reading are therefore weak.But we also need solutions to two further problems.
The first asks how saying that a power is epi x can be a way of referring to that power's manifestation.That is, it is far from obvious that "what a power is epi" is a way of referring to that power's manifestation;; we need to show that understanding it this way is plausible.The second problem is closely related: if both phrases do refer to a power's manifestation, why does Socrates use two phrases -why not just talk about what a power does, and omit the phrase with epi? 37 The objects analysis may seem to have an advantage on this question, since one thing's being epi another seems to indicate a relation between two different items."x is epi y" can be translated "x is upon, over, or against y" -so one does not expect x and y to be related as part to whole or as source to product.Since a power's manifestation is easily thought of as part of or a product of the power itself, the Identity Reading seems to conflict with the epi phrase.
Since the Identity Reading claims that what a power does is the same as what it is epi, this could sound as if the power of knowledge produces the forms -clearly an absurd result. 38But this would only follow if combined with an objects analysis; on the contents analysis this does not follow.The next step in solving this problem is to show that the preposition epi can support this reading.The epi + dative construction occurs many times in this passage without ἐστι, so it should be possible to understand the sense of the preposition without pinning down the sense of the verb. 39th a Revised Text and English Notes, and a Digest of Platonic Idioms (Oxford University Press, 1867), § §204-30, but especially § §216, 205 B note. 37"An adequate interpretation…must explain both (1) how the two phrases for distinguishing between powers are related and (2) why they are introduced as distinct phrases" (Gonzalez, "Propositions or Objects," 267).Cf.Gosling,"Δόξα and Δύναμις in Plato's Republic," 124. 38 I am grateful to an anonymous referee for raising this objection.
The most plausible candidates for the best reading of επί are domain and purpose. 40e phrase may indicate a domain, the set of objects a power acts on or interacts with in order to produce its manifestations.One can take this to mean that sight, for example, is epi the domain of visible objects. 41But the idea of purpose deserves a hearing, since Luraghi observes that this meaning "is mostly triggered by the occurrence of abstract nouns" , and the phrase ἐφ' ᾧ ἐστι is an abstraction par excellence. 42On this approach one might translate ἐφ' ᾧ ἐστι as "the [purpose] for which [a power] is," or "a power's purpose." The reading of ἐπί as indicating purpose shows that the Identity Reading's solution is viable: the phrase ἐφ' ᾧ ἐστι is plausibly read as describing a power's purpose, that is, its manifestation.Further, on this reading we can also see why the domain reading (or objects analysis) has seemed obviously right.If powers are conceived as distinct from their own manifestations, then it follows that what a power is epi is also distinct from the power itself. 43 is undeniably easy to form the impression, when reading this part of the Respublica, that what powers are epi is something distinct from the powers themselves.This may motivate some of the plausibility of the objects analysis.But it now turns out that the contents analysis can explain this impression as well: what knowledge and opinion do, and are epi, are the same entity, and this entity is distinct from the power that gives rise to it.Insofar as we have merely an impression that in discussing the epi relation, Socrates is referring to entities numerically distinct from the powers that produce them, this impression should be counted in favor neither of the objects nor of the contents analysis, since it is equally explained by both views.
Finally, if both phrases refer to the same thing, why not just use one phrase?I answer that the epi expressions refer to a power's manifestation prospectively, that is, as something the power is directed toward but is not at the moment actually producing.This gives the epi expression a force in the same neighborhood as, but not necessarily the same as, Aristotle's "potential." 4440 Silvia Luraghi, On the Meaning of Prepositions and Cases: the Expression of Semantic Roles in Ancient Greek (Amsterdam, J. Benjamins, 2003), 298-302 and LSJ s.v.ἐπί,  B.I.1.bfor the idea of a domain; B.III.2 for purpose.Of course ἐπί is capable of more shades of meaning than those I list here, but these are the most plausible choices. 41Reeve, Blindness and Reorientation, 135-6. 42Luraghi, On the Meaning of Prepositions and Cases,302. 43 This view is at least suggested in Book I at 346e. 44I have avoided the word "potential," except here, so as not to imply that Plato's sense of "being epi x" is the same as Aristotle's sense of "potential."Such a claim would prejudge many questions and is One sentence in particular reveals the semantic difference between Socrates' two phrases.Socrates asks, Each of them, then, since it has a different power, is epi something different?᾿Εφ' ἑτέρῳ ἄρα ἕτερον τι δυναμένη ἑκατέρα αὐτῶν πέφυκεν;; (478a4, Reeve's translation modified).
The phrases "᾿Εφ' ἑτέρῳ" and "ἕτερον τι δυναμένη" seem to express the same meaning: to be epi something is the same as to be able (δυναμένη) to do something.In order to make the two phrases semantically equivalent, Plato has added the word δυναμένη to the second.That is, the wording shows that being epi something is not equivalent to doing something, but rather to being able to do something.This supports the view that ἐφ' ᾧ expresses an ability or tendency to do something, while "what a power does" expresses an actual doing.
The two phrases therefore refer to the same phenomenon, but pick it out in different ways.The epi phrase refers to a power's manifestation prospectively, leaving aside whether the power is actually producing the it."What a power does" refers to the manifestation as actually produced, and therefore as something tangible and observable: the prefix ἀπο-in ἀπεργάζεται emphasizes the finished, completed aspect of the manifestation (LSJ s.v.ἀπό

D.2).
The difference can also be illustrated in English.Consider a radiator whose power to heat produces the manifestation of heating. 45We can think about either what its power is for or what it does: both refer to its power to heat, but the first is not committed to thinking of the power as actually manifesting.If one points to a cold radiator and asks, "What is this for?"one can answer that it is for heating.On the other hand, if one asks , "What is this radiator doing?" one can only answer that it is heating if it is actually heating.Yet both ways of talking refer to the same manifestation, heating.Socrates' use of both phrases may be a way of emphasizing that powers are real regardless of whether they are currently producing their manifestations.

Further Results
Four additional aspects of the text are also explained better by the Identity Reading.
First, this approach makes Socrates' standard for having knowledge somewhat less almost certainly false. 45This is an illustration of the view, not an argument for it.Such an argument would be faulty, since the example depends to some extent on the English present progressive tense ("is doing").
demanding than it would otherwise be, and this makes it easier for him to claim that there would be philosophers available to rule the ideal city.Socrates has claimed that philosophers should rule in answer to a challenge by Glaucon that he should show that his ideal city was possible, and how it was possible (471c6-7, e3-4); soon after our passage Socrates recognizes that the nature required for philosophers is rare (502a-503d). 46Socrates makes significant progress toward this goal by analyzing knowledge as a power.For someone has the power of knowledge if she is able to know what is; therefore having knowledge does not require that one has already actually come to know what is. 47Since actually knowing certain forms (e.g.The Good) may be very difficult, treating knowledge as a power may amount to a significant softening of the requirement. 48This also helps explain why Glaucon acquiesces so easily when Socrates offers his views about powers: it is not just that he is friendly to Socrates' argument, but he may also see that the bar for achieving knowledge is being lowered.
One might object that Socrates' goal is, on the contrary, to make knowledge more difficult, in order to exclude the sight-lovers.This is one of his goals, but if he achieves it by making knowledge too difficult, knowledge will be such a rare achievement that this will become an objection against the possibility of his city.Socrates in fact excludes the sightlovers not by setting a degree of difficulty they cannot meet, but by the absolute criterion that they lack knowledge -and they lack knowledge because they lack the power of knowing the forms.Given this method of excluding them, it is to Socrates' advantage to make the criterion for having knowledge less demanding, in order to make the availability of rulers for the city more plausible.Defining knowledge modally, as the ability to know the forms, gives Socrates 46 There is a debate over exactly how to understand this challenge and Socrates' response.But this much is clear: Socrates has a motivation for making the possibility requirement easier to meet, and on my reading his argument does this.See Mason Marshall, "The Possibility Requirement in Plato's Republic," Ancient Philosophy 28.1(2008): 71-85. 47This claim is contrary to what is usually assumed, namely that Socrates thinks knowledge involves actually grasping or coming into contact with its object.For example, "Only philosophers have knowledge…because only they know forms…" (Fine, "Knowledge and Belief in Republic V," 67);; "In short, knowledge is here understanding and acquaintance, understanding achievable only in direct acquaintance with certain objects…" (Gonzalez, "Propositions or Objects?" 258);; "…the work of episteme is to make perfect cognitive contact with that which admits of only perfect cognitive contact…" (Boylu, "The Powers Argument in Plato's Republic," 120). 48How can Socrates reduce his argumentative burden by adding the epi phrase, if both phrases refer to the same referent?The first phrase allows us to identify the power of knowledge if we can ascertain that it is aimed at the right kind of manifestation, just as we can identify a seed as an onion seed if we can ascertain that it is aimed at becoming an onion plant.The seed need not be an onion plant now in order to be so identified.I am grateful to an anonymous reviewer for raising this question.
Had he not described these people as denying the existence of the forms, it would have remained unclear whether they were capable of seeing forms.Instead, the sight-lovers are defined largely by their denial that the forms exist: Socrates mentions this denial repeatedly throughout our passage.Now it is natural to think that if someone does not believe in a nonphysical thing, then she is incapable of "seeing" it. 51Since Socrates has defined knowledge as being able to see the forms, he has a clear way of excluding the sight-lovers from knowledge.
For they would have to acknowledge that the forms exist in order to have any chance of "seeing" them;; and in taking this first step, they would no longer be the kind of people Socrates and Glaucon are addressing here.This point, too, depends on the Identity Reading of the epi phrase.On this reading, we know that philosophers are capable of seeing the forms because they have a power, knowledge, which is epi what is, and to be epi what is just means that the power is directed toward truth even if it is not at a given moment actually producing truths.On the objects analysis, by contrast, for knowledge to be epi what is means that it is applied to, or works in conjunction with the forms.On this reading, Socrates' use of epi is unrelated to the issue of belief in or acknowledgement of the forms.The objects analysis is therefore unable to shed any light on the modal way Socrates defines the sight-lovers and the philosophers.
Fourth, Socrates needs to introduce the epi phrase because of the logic of his dispute with the sight-lovers.Since these people are angry because he has denied that they have knowledge, they themselves must think that they do or might have knowledge.Thus, any putative example of knowledge Socrates might appeal to would likely be disputed, since in Socrates' view a form will be involved, but the sight-lovers deny that there are such things.
So if Socrates had defined the power of knowledge only by what it does, he would be restricted to discussing what some given power had actually managed to do.This is the position he must be in according to the objects analysis.This would mean finding cases in which the power of knowledge has actually managed to know something: but what this power has accomplished, or in which cases it has accomplished this, is precisely what is under dispute.Socrates will not agree that the power exercised by the sight-lovers is knowledge, someone who is "awake" (476c7);; the sight-lover is someone μήτε νομίζων (476c1-2); the nonphilosophers' disbelief is brought in twice, with heavy rhetorical emphasis, at the end of the argument (479a1-2 and a3-4).Stokes notes that it is unclear why the sight-lovers are both unable and unwilling to recognize the forms (Stokes, "Plato and the Sightlovers of the Republic," 125-6). 51Rowe, Philosophical Writing, 210.Of course "seeing" here is a metaphor for knowing, or possibly for knowing that something exists.I am claiming only that this is an understandable assumption, not that it is justified in this or any other case.

17
For similar treatments of the epi relation, see Santas, "Hintikka on Knowledge and Its Objects in Plato," 43-4;; Nicholas D. Smith, "Plato on Knowledge as a Power," Journal of the History of Philosophy 38.2(2000): 145-68;; Nicholas D. Smith, "Plato on the Power of Ignorance," Oxford Studies in Ancient Philosophy Supplementary Volume ( Further, on this reading the Constraint becomes an extraordinary coincidence, for prima facie there is no reason that two powers should not happen to work on the same sets of objects.If what a power does and what it is epi are really distinct from one another, it is extraordinary that no two powers ever do the same thing or are epi the same thing.It is even more surprising that Socrates and Glaucon both happen to know this extraordinary fact about powers without discussion.We should prefer an account able to explain why powers never overlap in what they do or in what they are epi, and why Socrates and Glaucon find this claim obviously true.
Constraint that the referent of either phrase-what a power is epi or what it does-suffices to individuate a power, regardless of what the other phrase refers to for the same power.That is, if we want to know whether two powers are the same or different, and we establish that they do different things, this shows that they are different powers without needing to check what they are epi.Likewise, if we find two powers that are epi different things, we again know they are different from one another without any need to find out what they do.