Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Platonic qua predication

Platonic qua predication Plato is fond of a kind of predication which is distinctive in important ways and which, so far as I know, has no name. As a result, it tends to go unrecognized, and the arguments it appears in are easily misunderstood. My purpose here is to draw attention to it, giving a rough account of how it works and why it matters.My topic is what I will call qua predication in Plato, or Platonic qua. And my interest will be not in any and every sentence involving a qua‐like expression, but more specifically in the form of general sentence in which Fs are claimed to be Gs qua Fs, but with the qua phrase left implicit: so, general implicit redoubled qua predications, or girqps.1Pronounced, I am sorry to say, ‘girk‐p’. (The ‘redoubled’ picks out the fact that the subject term F repeats in the qua phrase.) I will try to show that these sentences are a distinctive kind of talk in Plato, not reducible to anything else. In particular, they are not to be confused with the universal generalizations expressed in the same words – what I will term the corresponding universal generalizations. In fact, qua predications are, I will argue, best understood not as generalizations about F individuals at all, but as claims linking the properties or natures of Fness and Gness themselves: their purpose is to point to a correlation and at least suggest a causal relation between the two, although how clearly and distinctly this causal claim emerges varies greatly case by case (see Section 4). I here and throughout will use the terms ‘causal’, ‘causation’, etc. in the very broad way familiar from Aristotle's system of the four ‘causes’ [aitiai], which cover the full range of answers to a ‘why’ question (not merely ‘efficient’ causality), and formal explanatory relations in particular.2Hence, I use ‘explanatory’ interchangeably with ‘causal’; one might also speak of some types of girqps as setting out relations of grounding.These girqps are for Plato the proper vehicle of philosophical discourse. But nothing in his use of them depends on his distinctive metaphysics; indeed, part of their importance is that he also takes them to be pervasive in everyday, confused non‐philosophical discourse. And it seems to me that Plato is right about that. So, part of my task will be to show that – for us as for the ancient Greeks – girqps are indeed a standard mode of everyday speech – one curiously marginalized both in the intervening philosophical tradition and in modern analyses of predication.3This is not to deny that some forms of qua‐ification have received philosophical attention. For the long history of the Aristotelian ‘reduplicated syllogism’, which played an important role in medieval logic and in the metaphysics of philosophers as late as Spinoza and Leibniz, see Bäck (1996). Qua also figures in a range of recent philosophical studies including Szabó (2003), Fine (1982), and Bäck (2002). But in none of these does the qua‐ification in view seem to me to be quite the kind we find in Platonic girqps.It will help to begin by looking at some Platonic texts in which qua‐ification is explicit. So, in Section 1, I work through a few case studies of explicit Platonic qua; Section 2 sums up with a preliminary analysis of how it works. In Section 3, I step away from Plato to make the case for the pervasiveness of girqps in everyday speech; in Section 4, I set out the three principal ways in which Plato seems to deploy them. Section 5 explores their significance for his metaphysics and his conception of philosophy. By way of conclusion, Section 6 sketches a brief contrast with Aristotle, and brings out some implications for the study of generics, a category with which Platonic girqps substantially overlap.EXPLICIT PLATONIC QUA: SOME TEXTSPlato's Lysis features what looks like a proof by cases that friendship is impossible (214a‐16c). Socrates is trying to discover what sorts of people are friends with each other. He argues by various ingenious sophistic means that bad people cannot be friends with each other (for they harm each other); nor can one good person be friend to another insofar as [kath' hoson] he is like (for those who are like cannot benefit each other) (215a3‐5); nor could the good person be friend to another good person insofar as [kath' hoson] he is good (for the good are self‐sufficient) (215a6‐7); nor can unlike people be friends (for they are enemies). And since the good and the bad and the like and the unlike together seem to pretty much cover the bases, we are left with the alarming result of friendship nihilism. Or are we?A question mark is placed beside this conclusion by the key phrase I noted above: ‘insofar as’ [kath' hoson]. This phrase is explicit only at three points (214e4, 215a5, a6), but implies a qualification and complication which pervades the argument as a whole. For it expresses a landmark philosophical insight: we pick out subjects under different descriptions, which can be used to express different facts about them. Socrates might be friends with Crito insofar as they are both good but not insofar as they are both like, or vice versa; there is a significant distinction between the two facts even if it is by being good that they are like each other.To see what the argument really intends, we need to take a close look at how this ‘insofar as’ talk works. The kath' hoson locution, like its English counterpart ‘insofar as’, often serves to flag gradability of predicates; but here its primary purpose seems to be simply to pick out the different descriptions under which it might be true or false that two people are friends. That makes it roughly equivalent to the Greek hêi, of which the Latin term qua is a translation.4Only roughly because hêi and qua unlike kath' hoson do not particularly suggest gradability. I treat this as a difference of nuance, since as we will see shortly gradability is not part of what a qua predication actually asserts: it asserts only a correlation between two predicates, which may also apply to ungradable ones (See Section 3). Plato uses hêi similarly in an important passage of the Meno, where Socrates insists that a proper definition is one which picks out a unity:But Meno, … if I were asking you what is the nature [ousia] of bees, and you said that they are many and of all kinds, what would you answer if I asked you: “Do you mean that they are many and varied and different from one another in being bees [tôi melittas einai]? Or are they no different in that regard, but in some other respect, in their beauty, for example, or in their size or in some other such way?” Tell me, what would you answer if thus questioned?‐‐ I would say that they do not differ from one another insofar as they are bees [hêi melittai eisin]. (72a8‐b9)5Quotations from Plato are in the translations by many hands in Cooper (1997), sometimes with revisions.In the very late Philebus, by contrast, Socrates argues, against the hedonist Protarchus, that there is no unity to the diverse things, some good and others bad, which we call ‘pleasures’. Protagoras balks:What are you saying, Socrates? Do you think anyone will agree to this who begins by laying it down that pleasure is the good? Do you think he will accept it when you say that some pleasures are good but others are bad?‐‐ But you will grant that they are unlike each other and that some are opposites?Not insofar as [kath' hoson] they are pleasures. (13b3‐c5)Here, we see kath' hoson doing exactly what hêi does in the Meno: it picks out a feature or aspect of the subject in respect of which a further predicate – in both cases, the tricky second‐order property of likeness or sameness – is said to obtain. So, I take it that kath' hoson and hêi are in the relevant constructions interchangeable; I count sentences where either occurs or seems to be implicit as qua predications.6Willie Costello points out to me that redoubled qua predications can be expressed in a multitude of other ways in Greek (as in English), including causal datives and kath' hauto expressions. I focus on hêi and kath' hoson for simplicity's sake, not because they exhaust the possibilities.At this point, we can already note several significant features of Platonic qua. First, a qua predication may be explicit, as in the Meno, Philebus and those three crucial moments of the Lysis argument, or implicit (as in the other related premises of the Lysis argument). And it may or may not be redoubled: that is, it may pick out Fs qua Fs or qua some other feature G.7‘Redoubled’ here is not to be confused with ‘reduplicative’ [epanodiploumenon] in Aristotle. Aristotle's account of the reduplicative syllogism is far from clear, but it seems that any syllogism revolving around explicit qua predication counts as reduplicative, and his focus in Prior Analytics I.38 is on the non‐redoubled case (i.e., ‘Fs qua G's are H's’). There is a long medieval tradition of analysis of the reduplicative syllogism starting from this text; see Bäck (1982) and (1996). In both Meno and Philebus, we encounter redoubled qua: bees qua bees, pleasures insofar as they are pleasures. In the Lysis, we get both kinds: Socrates speaks in turn of good people qua like each other and good people qua good, where only the second is a case of redoubled qua.In English, redoubled qua‐ification can be either implicit or expressed using an impressive variety of locutions: ‘Fs qua Fs are Gs’, ‘Fs are Gs insofar as they are Fs’, ‘Fs as such are Gs’, ‘inasmuch as something is F, it is G’, ‘what is F is to that extent G’, ‘strictly speaking, Fs are Gs’ (see below for this locution in Republic I) and so on.8I do not here include ‘As an F, x is G’, since these are not always explanatory in the way which I will argue is characteristic of Platonic qua. In fact, the discussion of predications of this form in Szabó (2003) brings out that they may do something very different from Platonic girqps. Consider his canonical sentence ‘John as a judge earns $50,000’, which is to my ear problematically ambiguous. If it were understood as instantiating the claim ‘Judges (as such) earn $50,000’, we would be dealing with an application of Platonic qua to a particular case: John earns that sum because he is a judge and that is what judges earn. But for that claim, more conventional and perspicuous phrasing would be ‘As a judge, John earns $50,000’. Alternatively, if the sentence claims only that John's judging work earns him that salary (with nothing implied or suggested about the wages of any other judge), then a better, less ambiguous phrasing would be ‘John earns $50,000 as a judge’. That some as‐phrases do something very different from Platonic qua seems clear. Consider: ‘As a composer, Mozart was a model of delicacy; as a man, he was a boor’. It would be a terrible mistake for a non‐native speaker to protest that there is no correlation, let alone causation, between being a composer and being delicate or being a man and being a boor (Contrast: ‘As a citizen of Vienna, Mozart was liable for taxation by the Austrian crown’, where the as phrase is explanatory.) This sentence and Szabo's seem intended only to metaphysically ‘partition’ their subject so as to restrict the application of the predicate, a form which in medieval philosophy is termed specificative qua: see Bäck (1998) and compare Lear (1982) on Aristotle's qua operator. I cannot here engage with this or the other permutations of qua in medieval logic (and later quatenus in Spinoza), which owe more to Aristotle than to Plato and present many complications of their own: see Bäck (1996). I take it that all of these formulations say roughly the same thing, in the absence of any differentiation by context; and these rich linguistic resources are, I think, telling. They strongly suggest that redoubled qua predication is not just a local linguistic anomaly, a by‐product or epiphenomenon; still less is it peculiarly Greek or dependent on proprietary Platonic philosophical assumptions. It is a way of saying something distinctive that we, like the Greeks, very much want to be able to say.9I discuss more fully in the next section what that something is; I will, however, be leaving open the question of what we should understand the logical form of girqps to be. Syntactically, it would seem that in the sentence ‘Fs qua Fs are Gs’, ‘qua Fs’ should be understood as an adverbial phrase modifying the copula ‘is’, and so belong to the predicate. (As Nicholas Denyer pointed out to me, this seems clear in the non‐redoubled case: ‘As Members of Parliament, Cabinet Ministers already earn a healthy pension’.) But semantically, when, for instance, Thrasymachus says that ‘The doctor qua doctor never errs’, he can equally well be heard as intending ‘qua doctor’ as a determinant of the subject of the sentence, indicating that its reference is to a nature or essence reifiable as ‘the doctor qua doctor’ (see Sections 3–5). I am not sure how to represent girqps in a way that respects this tension or ambiguity. Sandra Peterson has suggested to me (on the basis of the puzzles of scope raised by negations of qua claims, discussed in Section 2) that ‘qua Fs’ might best be understood as a kind of sentential operator (Cf. also Lear (1982) on Aristotle's qua operator). I thank Nicholas Denyer and Ben Morison for pressing me on this question.It is striking that in both the Meno and the Philebus (and, as we will see, in Republic I), it is Socrates' interlocutor who makes the qua construction explicit – as if to show that the move is well understood by any contemporary intellectual. And this can help us to see how philosophical use of qua gets started. For it seems clear that the original home of qua‐ification was eristic or dialectic. This is the form of game‐like intellectual combat for two players, going back to the sophistic movement of the 5th century BCE, which is discussed in Aristotle's Topics and Sophistici Elenchi. In eristic, one speaker propounds a thesis, and the other attempts to refute him by posing a chain of questions demanding yes/no answers; the aim is to lead him into the contradiction of that thesis, or some other absurd result.10On eristic see Castelnérac and Marion (2009, 2013), Nehamas (1990), Brunschwig (1984/5), Moraux (1968), and Chapter 4 of Ryle (1966). One common kind of eristic strategy moves from qua‐ified claims to the corresponding universal generalization. For instance, in the Hippias Major, Hippias puts forth the thesis that the beneficial is the fine. Socrates gets him to agree that the beneficial is the cause of the good; and that it follows that ‘the fine is not good, nor the good fine’, since the cause and the effect cannot be the same (296e‐7c). This crucial move is supported by an analogy: ‘the father is not a son, and the son is not a father’ (297c‐d). This is fair enough as a claim about fathers qua fathers, of course; but it obviously does not license the corresponding universal generalization that no father is a son or vice versa. Another kind of sophistic fallacy seems to have turned on equivocation between different possible qua‐ifications. An example Aristotle mentions in the Sophistic Elenchi is that if I own a work of art painted by someone else, it is both mine (qua property) and not mine (qua creation) (179b4‐6, cf. Euthydemus 298d‐e with Sophistic Elenchi 180a1‐5). Bringing mismatched qua‐ifications to light is thus an important way for the respondent in the eristic game to block or dissolve fallacious moves. The primordial philosophical use of qua constructions, then, must have been for what we might call distinction‐drawing or dialectical qua: cases in which two different possible qua‐ifications are in play and the point of philosophical interest is to distinguish them and to get clear about what can and cannot legitimately be inferred from each.Plato does occasionally deploy distinction‐drawing qua, but it is less common in his works than one might expect.11For example, Ion 540e, and the (possibly inauthentic) Alcibiades I, 115a‐16d. Plato has other tools for drawing qua‐like distinctions in dialectical contexts, including the language of ‘natures’, e.g. at Republic 5.453B‐454C, on which see El Murr (2020). Of much greater interest to him is the locution in which the qua is redoubled but with no particular non‐redoubled alternative in view, so that there is no need for the qua phrase to be explicit: these will be our general implicit redoubled qua predications (girqps), sentences which claim that Fs < qua Fs > are Gs.PLATONIC QUA: A PRELIMINARY SKETCHThese first few cases already reveal the basic workings of Platonic qua. In the Lysis arguments, it seems obvious that the ‘insofar as’ locution points to a correlation between two predicates or descriptions F and G: specifically, between the candidate Fs of being like or good and the target G of being a friend. What I mean by ‘correlation’ is simply that the ‘insofar as’ construction draws our attention to the possibility that the predicates in question are gradable, and claims that their grades in any particular case will match up. It seems further that these claims are meant to affirm not only correlation but causation; for Socrates' inquiry throughout the Lysis is into what makes two people friends – the feature of a friend in virtue of or because of which the friendship arises. So read, the arguments serve to produce the important and quite plausible result that two of the most promising candidates for that causal role – likeness and goodness – fail, as do their opposites.12Cf. Taylor (1960): ‘it is not true either that any and every “likeness” nor yet that every and any “unlikeness” can be the foundation of friendship’ (70 n. 1).The upshot of this reading is, of course, that, so read, Socrates' alarming conclusions are not to be identified with the corresponding universal generalizations, and do not entail them. Again, by the ‘corresponding universal generalization’ I mean the universal generalization which is expressed in the very same form of words as the girqp, so that either might be mistaken for the other. Of course, the whole point is that the two do not ‘correspond’ semantically: hence, the danger of fallacy and the importance of distinguishing them and registering the truth–condition gap between them.13By making the qua phrase of a girqp explicit, we can always produce a sentence which corresponds to it better semantically (‘No son qua son is a father’; ‘The doctor qua doctor acts for the health of the patient’), and if we construe this as a universal generalization it will be entailed by the girqp. But so construed, I will suggest, it still does not fully capture the meaning of the girqp itself. Cf. Sections 2 and 3 and n. 18 below. The gap is particularly tricky to navigate where negations are involved, since these may differ in scope and force so as to allow for a weak and a strong reading of the same claims. On a weak reading of the Lysis arguments, they leave open the possibility that the features specified (the candidate Fs) are simply orthogonal to those which ground friendship so that, for instance, some like or good people might still be friends in virtue of some other feature of their character – as charming conversationalists, say, or birdwatchers. On the strong reading, the conclusion would be that being an F tends pro tanto to exclude being a G: it seems plausible that the argument that the bad are not friends is to be understood in this way. But even when the strong reading of a negative girqp is in view, the corresponding universal generalization does not immediately follow. The most obvious reason for this is the possibility of a defeater: that is, in any given case, another feature of the subject which might cancel or outweigh the qua‐ification under consideration. Even if likeness tends to exclude friendship, two like people might be friends in virtue of some other feature able to outweigh or override that likeness.Platonic qua claims, then, are always not only pro tanto but in principle defeasible in their application to particulars. And which ones can have defeaters, and in which kinds of cases, is a question for substantive philosophical inquiry, not something we can read off from the logical form of qua predication. In the Lysis, Plato likely means us to think that injustice really does trump any other feature of a prospective friend: there can be no defeaters in this case, and so ‘the bad are not friends’ is also true as a universal generalization (cf. Republic 351b‐2a, 580a). But whether the self‐sufficiency of the good person has the same power is not at all clear. Without further inquiry, qua‐ified arguments by themselves do not license inferences to the corresponding universal generalizations: this truth–condition gap vis‐à‐vis the corresponding universal generalization is one of the reasons that it is so important to recognize girqps (and, indeed, possible to do so). And the cheering upshot of reading Platonic arguments in implicitly qua‐ified terms is that it is far harder than we might think to refute them by counterexample. On the other hand, wherever we take Socrates or his author to be encouraging an inference to the corresponding universal generalization, as with sons and fathers at Hippias Major 297c, it becomes far harder to acquit him of deliberate fallacy.That Plato is alert to the truth–condition gap can be seen from a famous and complex argument at Protagoras 351b‐4e. Here, Socrates tries to get a reluctant Protagoras to answer dialectically for ‘the Many’, and to affirm hedonism both on their behalf and his own. Protagoras initially resists, insisting that only honourable pleasures are good. Socrates persists:Surely you don't, like the Many, call some pleasant things bad and some painful things good? I mean, isn't a pleasant thing good just in that [kath' ho] it is pleasant, that is, if it results in nothing other than pleasure; and, on the other hand, aren't painful things bad in the same way, just insofar as [kath' hoson] they are painful?…. Just insofar as [kath' hoson] things are pleasurable are they good? I am asking whether pleasure itself [hêdonê autê] is not a good. (351c2‐e3)Socrates is alleging confusion: the Many, and Protagoras with them, are misrepresenting their own position when they say that some pleasures are bad. Their real view is one demanding expression in qua form: what is pleasant is good insofar as it is pleasant, and what is painful is bad insofar as it is painful. Here, it is obvious that the correlation is symptomatic of a causal or explanatory relation: it is precisely because something is pleasant that it is, in the view imputed to the Many, a good thing. This is hedonism in its purest form, and the Many are quite wrong to think of themselves as something other than hedonists merely on the grounds that they reject the universal generalization ‘All pleasures are good’. Perspicuously expressed, hedonism consists precisely in their position that the pleasant qua pleasant is alone good; and this, of course, allows for the pains attendant on some particular pleasure to defeat its claim to goodness. It is precisely because of their commitment to hedonism that the Many reject the corresponding universal generalization which they took to express it.14Thomas Slabon has objected to me that Protagoras seems to interpret the position pressed on him by Socrates so that it makes purely extensional identity claims about the good and pleasant. I cannot here engage properly with this reading, but would suggest that if Protagoras does so, this may well be the point at which the argument becomes fallacious, as it clearly is when redeployed at 360aff.One final example of explicit qua‐ification is of a rather different sort. This is the star turn of ‘qua’ in Plato: the debate about the ruler ‘qua ruler’ or ‘in the strict sense’ in Republic I (340d‐7e).15I survey these arguments in Barney (2006). Here, it is the sophist Thrasymachus – once again, not Socrates but a dialectically trained interlocutor – who introduces qua‐ification, with vast philosophical consequences. He has stepped into a Socratic refutation by saying that justice is the advantage of the stronger, that is, the rulers (338c‐9b); and that for subjects to do what the rulers say is just (339b); and that the rulers sometimes make mistakes (339c). But it follows that if the rulers mistake their own advantage, obedience to them is both just and unjust. Invited to repair his position, Thrasymachus rather startlingly opts to retract his admission that rulers can make mistakes. ‘Do you think I'd call someone who is in error stronger at the very moment he errs?’ (340c6‐7). He continues:When someone makes an error in the treatment of patients, do you call him a doctor in regard to [kata] that very error? Or when someone make an error in accounting, do you call him an accountant in regard to [kata] that very error in calculation? I think that we express ourselves in words that do say [legomen tôi rhêmati houtôs] that the doctor is in error, and the accountant, and the grammarian. But I think that each of these, insofar as [kath' hoson] he is what we call him, never errs, so that, according to the precise account [kata ton akribên logon] ‐‐ since you too go in for precise accounts ‐‐ no craftsman ever errs. It's when his knowledge fails him that he makes an error, and in regard to that error he is no craftsman. No craftsman, expert or ruler makes an error at the moment when he is ruling, even though everyone will say that a physician or a ruler makes errors. It's in this loose way that you must also take the answer I gave earlier. But the most precise answer is this. A ruler, insofar as [kath' hoson] he is a ruler, never makes errors and unerringly decrees what is best for himself, and this his subject must do. (340d2‐41a3)Thrasymachus' concern here is to vindicate his conception of the ruler as expert practitioner of self‐interested injustice. Although this began life as an empirical claim, about ‘rulers in the cities’ (338d‐e), the really important point to him is normative. And what looks here like a claim about the perfect possessor of the craft really signals a more general qua‐ified stance. Any ruler whom we observe to fail in the practise of his self‐interested craft is, on that occasion and to that extent, defective as such – ‘ruler’ is, it turns out, a gradable predicate, and someone counts as a ruler precisely insofar as he is the expert practitioner of his craft. This conception of the ruler as craftsperson is one which Socrates picks up and makes his own: his whole conception of the philosophical Guardian, developed in the middle Books of the Republic, is his replacement for it. Socrates also, within Book I, offers his own rival conception of craft and the craftsperson, focused on the case of the ‘doctor qua doctor’ (341c‐50d).16On these arguments see Barney, 2006 and 2020.Claims about the doctor qua doctor and the ruler qua ruler are important cases of role specification qua, to be discussed in Section 4. For now, the main point is just to see what they have in common with our other cases, the better to fix the scope of Platonic qua. The claims here look very different: they are about kinds of agent, mostly in the singular (as of a prototype or perfect case) rather than the plural, and predicating actions rather than properties. And the immediate point of the qua‐ification is neither distinction drawing (the point is not to contrast technê with some other hat the agent might wear) nor explanatory, but criterial. That is, Thrasymachus and Socrates alike are concerned to specify conditions for the (precise or strict) ascription of terms like ‘ruler’ and ‘doctor’, by exploring the other predicates they entail; and we are to envisage this as guiding the application of the term to candidate particulars. (This project contrasts with that of the Lysis, where it seems to be presupposed that to predicate the term ‘friend’ in particular cases is unproblematic: the challenge is to explain what grounds it.) This criterial use, as we will see in Sections 4 and 5, is important because of the chasm between loose talk and the properly precise kind: while it is true that no real doctor or doctor strictly speaking ever errs, it is far from being true of those conventionally denoted by the term. I will have more to say about this phenomenon, referential disorder, later on.UN‐PLATONIC INTERLUDEIt should be clear by now that Platonic qua can be expressed in a wide variety of ways, and that it is apt for several different kinds of job. And since in the passages I have cited the qua‐ification fluctuates in and out of explicitness, in keeping with Plato's distaste for repetition and regimented language, it should be plausible that implicit qua predications, to which I now turn, work in just the same way as explicit ones. In particular, girqps – again, general implicit redoubled qua predications, that is, sentences in which Fs are claimed to be Gs qua Fs, but with the qua phrase left implicit – should say just the same thing as the counterpart sentences deploying explicit redoubled qua. Before turning to Plato's use of girqps, it is worth being clear that there is nothing proprietary to Plato about them (or to Greek); they are pervasive in our discourse as well as his. Consider the following quick sample pack, chosen for its diversity and independence of any proprietary Platonic assumptions:Humid days are unpleasant.Gossips are unpopular.Cyclists need carbs.A newspaper reader is an informed voter.A teacher is not a therapist.A gardener is a craftsperson, not a mere manual labourer.The President is the Commander‐in‐Chief.Each of these is chosen to bring out some salient features of girqps and their use. ‘Humid days are unpleasant’ makes the causal character of qua almost explicit: no one would say this in a tone of surprise, as if noting a mysterious statistical coincidence. Likewise, ‘Cyclists need carbs’, which would be wholly uninformative as a universal generalization: all human beings need carbs, so there is no point here other than the pro tanto and implicitly causal one. ‘Gossips are unpopular’ brings out the very real possibility of defeaters: think of Jane Austen's Emma or Cher in Clueless, popular for her high spirits and prettiness despite the trouble caused by her gossiping. With the newspaper reader, we can see what I take to be a different basis for truth–condition gaps, namely threshold variation for applying predicates once the qua‐ification is removed. That is, I might assert the girqp yet not be willing to call Homer Simpson, who admittedly browses the paper in a minimal way, an informed voter simpliciter. (Threshold variation might also complicate the case of Emma: we might apply ‘unpopular’ and ‘gossip’ to different percentiles for those traits.) The teacher–therapist case, like the Lysis arguments, exposes some of the problems and ambiguities of negations involving qua. Is it being asserted only that no teacher is as such qualified to play the role of therapist, or that the two roles actively exclude each other? Most likely the point is merely the former, weak one; so here too we see clearly that the girqp may easily be true while the corresponding universal generalization is false. The gardener sentence is more suggestive of ‘strong’ qua and perhaps relatedly (unlike the ‘therapist’ sentence) it registers as criterial: if many of the people we habitually refer to as gardeners really are just manual labourers, we are misapplying the term. The final sentence is also an instance of role specification qua: and it shows that the terms deployed in a girqp need not be gradable, nor are defeaters always a possibility, nor need the subject always be plural. (That it is nonetheless a girqp is still clear, I take it, from the fact that we could add in ‘as such’ or an ‘insofar as…’ phrase after the subject term with no change in meaning.) It should also help to make vivid how fundamentally different a girqp can be from superficially similar kinds of predication with apparently the same subject. If we juxtapose it with ‘The President is from Delaware’, the difference is intuitively clear, since we determine their truth by looking in different directions: in one case, directly at the object specified by the subject term, in the other, at what the Constitution says about the properties and powers which correlate with the Presidency for any occupant of that role.For all their diversity, these examples should enable us to see that there is a common core to girqps, calling for their recognition as a distinctive kind of sentence. Their key feature has already emerged: they assert a pro tanto correlation between two features and at least suggest that it is due to a causal relation. A qua predication does not just register a coincidence: it at least suggests that Fs are Gs (when and insofar as they are Gs) because they are F or by being F. I say cautiously ‘at least suggests’ because the sharpness with which this causal claim emerges seems to vary considerably case by case, and we might in some cases be tempted to view it as merely a matter of suggestion or implicature. But in other cases, perhaps the majority, the causal claim does seem to be fully asserted and to fix the truth conditions of the sentence. Where the causal claim is merely suggested, we can at least say that the correlation imputed by a girqp is non‐accidental, and prima facie asymmetrical. That is, to say that Fs < qua Fs > are Gs is not at all to say that Gs < qua Gs > are Fs17This is not itself one of the features that differentiates qua predication from quantified universal generalization, since it is also true that ‘Fs are Gs’ is a different fact from ‘Gs are Fs’. But this is a fact about the asymmetry of class inclusion, whereas the asymmetry of qua predication is due to the asymmetry of causal (or explanatory or grounding) relations.; if the claim in the other direction is true, it represents a different fact. (It may be true both that good people as such tend to be self‐sufficient and that self‐sufficient people as such tend to be good; but these are different claims and if they are true, it is for different reasons.) Only prima facie, however, because in asserting a qua predication, we may also mean to at least leave open the possibility that a full and correct theory of Fness and Gness would show them to have a common cause, or a reciprocal causal relation, or even to be ultimately reducible to the same thing. One of Plato's canonical uses of qua predications is to assert that good things as such are fine (Symp. 201c, Lysis 216d, Rep. 457b etc.); but the ultimate relations of the good and the fine in Platonism are a great metaphysical mystery.So understood, a girqp says both less and more than the corresponding universal generalization. More, because of its irreducibly causal character; less, because it does not entail that any particular F is G simpliciter. This is because girqps are not really a way of talking about particular F and G things at all, but a way of talking about Fness and Gness themselves while bracketing all questions about their instantiations. As we will see, they are important to Plato precisely in virtue of this detachment from particulars, which puts us in a position to think and speak clearly about the natures of Fness and Gness as such.18As Willie Costello has pointed out to me, this has an interesting resemblance to Dretske's account of the logical form of a law of nature (Dretske, 1977). As Dretske puts it, ‘To say that it is a law that F's are G is to say that “All Fs are G” is to be understood (insofar as it expresses a law), not as a statement about the extensions of the predicates “F” and “G”, but as a singular statement describing a relationship between the universal properties F‐ness and G‐ness.’ (252). A disanalogy is that Dretske does take a law to imply the corresponding universal generalization; but he emphasizes that the two are still different, in part because universal generalizations have no explanatory pretensions (the fact that every F is G fails to explain why any F is G' (262)). This convergence suggests that we might think of girqps as the broader, garden variety genus of assertion of which laws of nature are a special privileged subset, special not least in that with them the usual truth–condition gap is closed.GIRQPS IN PLATOOnce we learn to spot them, girqps turn out to be everywhere in Plato's work. I will here offer a brief field guide to them, grouped under three very general headings. (We have seen instances of all three already, albeit at exceptional moments where the qua‐ification surges into explicitness.) I do not mean to present these as sharply defined or formally different species: my point is just the rather superficial one that girqps are especially at home in three kinds of Platonic habitats, with a somewhat specialized look and function in each.19If anything, we might see the third kind as a special case of the second (since a technê is a causal power [dunamis]), and the second as a special case of the first (since a causal power can in turn be seen as a kind of property). My point is just that girqps belonging to these somewhat blurry categories tend to appear in different contexts and to be used with different argumentative purposes.One kind of use is what we might call property network qua. A great many theses and arguments about central evaluative concepts belong here, including the Protagoras argument about pleasure which we considered in Section 2. Likewise, Plato argues that the good qua good is fine (Republic 457b); that courage as such is fine and thus advantageous (Alcibiades I 115a‐16c); that the admirable is so in virtue of being either pleasant or beneficial, and conversely that the shameful is so in virtue of being either painful or harmful (Gorgias 474c‐77e); that the appropriate is fine (Hippias Major 290c‐91c); and that the harmed are wretched insofar as they are harmed (Meno 78a). All these are natural candidates for use as theses or as premises in eristic arguments: property network argument is the realm in which the sophistic roots of (attentiveness to) qua‐ification are most evident.20Property‐network arguments are easily formulated in Greek because any adjective can be substantivized by the definite article. So, in these arguments, Plato typically speaks of ta F (neuter plural with the definite article, for all F things), to F (neuter singular, for any F thing – or perhaps for ‘what is F’, with F as a kind of mass term), or, if the property is envisaged as applying particularly to humans, hoi F (masculine plural, ‘those who are F’).A second arena for qua predication puts their explanatory dimension front and centre. These are arguments about causal powers. This would be my preferred bucket to put the Lysis arguments in since, on my reading, these arguments are investigating the grounds or basis for friendship between two people. Plato's conception of causal powers [dunameis, to the extent he has a single preferred term] is a central, woefully underdiscussed, and ill understood topic, and I cannot properly enter into it here.21But see Benson (1997), Sedley (1998), Wolfsdorf (2005), and Lefebvre (2018), particularly Part Two on Plato (183–346). But I take it that Platonic powers are real dispositional properties and that his preferred mode of causal explanation consists in identifying them. It might be too strong to say that Plato is uninterested in necessary and sufficient conditions, but they are at any rate not what he thinks a cause is; a real causal explanation picks out the cause par excellence. Depending on the kind of explanandum in view that will be either an agent (an aitios, literally the guilty party) or a power – roughly, a dispositional property the manifestation of which reliably brings about the effect in question.22All this is of course oversimplified and calls for far more discussion; cf. Sedley (1998).A particularly important kind of causal power for Plato is virtue. The presence of a virtue in an agent – courage, say, or justice – reliably manifests in virtuous actions. So we can test prospective definitions of a virtue by swapping in the proposed candidate, understood as a causal power, and asking whether it invariably brings about behaviour we judge to be virtuous in a relevant way. For instance, in the Laches, Plato considers whether courage could be correctly defined as endurance. Answer: no. For enduring foolishly is not admirable, which courageous behaviour always is (192d). Socrates then turns to the other possibility, wise endurance. (So, the project here is again superficially one of refutation by exhaustion of cases, as in the Lysis arguments.) He considers various kinds of behaviour that we would count as enduring wisely and notes that we would not count them as courageous, and that indeed the person who endures wisely in certain circumstances would be judged less courageous than the person who endures foolishly (193b‐d). So, of the two species of endurance, wise and foolish, neither is reliably the cause of courageous behaviour, and therefore neither is identical to the power of courage. Plato's inference is that endurance was the wrong general power to start with: wisdom taken by itself is a better candidate and so is taken up as the next definition to be considered (194cff.).23Another intriguing but somewhat non‐standard case of explicit qua plausibly belongs here. In Republic IV, Socrates uses the example of thirst to show that generic states or powers such as desire and knowledge are correlative with equally generic or unqualified objects. For thirst, insofar as it is just thirst [kath' hoson, 437d8, 9a9] is not for any particular kind of drink, even good drink, but simply for drink (Republic 437d‐8a). So, the statement ‘Thirst is for drink ‐‐ drink simpliciter’, understood as a girqp, is true. Note the extreme version of the truth–condition gap: the corresponding universal generalization will plausibly be false in every case, since when thirst occurs together with heat it will be for cold drink, when together with coldness it will be for hot drink, and so forth. Here, the use of redoubled qua is evidently to strip away these adventitious and contingent factors so as to leave only what belongs to thirst as such [kath' hauto]. This ‘stripping‐away’ use of qua comes close to Aristotle's principal uses of it (cf. n. 39 below) and to ‘partitioning qua’ (cf. n. 8 above).A third characteristic use of Platonic qua is the one we have already seen in Republic I: role specification qua, used to determine the actions and properties correlated with agents' roles and identities. A particular focus here are the actions and properties correlated with knowledge and with the crafts [technai]. Plato, like his character Socrates, is obsessed with technê, craft or expertise, as a kind of general grounding for the normativity of practical reason and a model for moral virtues.24I discuss the technê model with particular reference to the ethical agenda of Republic I in Barney (2020). But craft terms are just a subspecies of a very broad class of locutions. We get the same kind of argumentation when, for instance, Plato tries to fix the properties of ‘the lover’, ho erastês (i.e. the one who loves, insofar as he loves) in the Symposium and Phaedrus. Likewise in other works when Plato investigates the properties of ‘the democrat’, ‘the poet’ or even ‘the tyrant’ or ‘the sophist’. (Often for role specification qua the preferred construction is the singular with the article, suggesting the inspection of a prototype or paradigm case, and thus perhaps keeping at arm's length the plurality of questionable instantiations around us.) Nor does this kind of qua‐ification apply only to human agents. The same goes for the city strictly speaking, the tool as such, and indeed the word ‘name’ [onoma] itself: for a name, according to the Cratylus, is a tool for dividing being and revealing the natures of things (387a‐9a). The background assumption in all such cases is that the role in question is in at least a weak sense teleological or functional, in the sense that it is part of the ‘job’ or office or identity of Fs to be Gs – or, since predications of actions are particularly common in this sphere, part of the job of Fs to reliably φ.25So, my use of ‘teleological’ and ‘functional’ here should not be thought to import any presumption of normativity into qua‐ification, except insofar as doing what Fs qua Fs do is a condition for something's being a good instance of Fness. In the Republic, Socrates first illustrates the idea of an ergon with the claim that it is the ergon of the hot not to cool but to heat things (335d3): in other words, its natural, characteristic operation and effect. Plato seems to hold that much of ordinary language – far more than you might suspect – is implicitly teleological or functionalist in this way, and that philosophical inquiry can use role specification qua to articulate these background conceptions. Role specification qua is thus at home wherever the salient feature of some nature is its characteristic work or function, and like other forms of Platonic qua it can cause confusion when it goes unrecognized.26Consider, for instance, Republic X, where Plato offers a complex series of arguments intended to articulate just what the imitative artist, the mimêtês, does. Mimêtês is not for Plato a positive role or function; on the contrary, his point is to argue that such people must be excluded from a just city (595a‐608b). But to do so requires figuring out exactly what mimetic art is. His opening argument is that as the creator of an imitation, the mimetic artist does not have any knowledge about things in themselves (595a‐8d). This argument has long been found puzzling and unfair: how can Plato be so sure that no artist ever has had knowledge of what he imitates? But this objection is only in order if we mistakenly suppose the argument to be working with universal generalizations. Instead, we should understand it in terms of role specification qua. The artist qua artist knows nothing about realities: that is, such knowledge is not required for his function of imitative creation. This indeed does nothing to exclude the possibility that some individual artists might have knowledge on some other basis. The argument does not purport to issue directly in a universal generalization that would ground the exclusion of all artists from the city; that is why it is followed by a further, empirical argument that, as a matter of fact, even the greatest imitative artists seem not to have had knowledge about the things they represent (599a‐600e).Role specification is where we can most easily see the ethical stakes of Platonic qua, and how it indeed serves as a vehicle for social critique. Properly speaking, the doctor qua doctor is someone who seeks the health of the patient and knows how to provide it by medical techniques. So agents who seek something other than the health of the patient – the maximization of their billings, say – are to that extent not doctors at all, no matter how much they might know about bloodletting and cautery. This line of argument has revisionist implications relative to the current conventional denotation of the term. Many ‘doctors’ will turn out to be frauds. But, Plato thinks, such arguments leave intact – indeed, they depend upon – the sense we standardly give to ‘doctor’. (Only Socrates' teleological analysis of medicine can explain, for instance, why ‘good doctor’ is used, by all of us, for the one who treats patients correctly – and not, say, for whoever excels at using medical knowledge for his own selfish unmedical ends.) What reflection on qua predication brings out is that our everyday generalizations (the kind which forms the corresponding universal generalizations for girqps, and must be distinguished from them) are subject to a kind of linguistic entropy or degeneration: a truth–condition gap arises in relation to role specification qua because the reference of F ceases to be guided by its sense, and is instead fossilized by social convention and unreflective habit. This is the phenomenon I call referential disorder.27I discuss this in Barney (2001) in terms of ‘the project of the strict sense’ (10–14). When our language is in a state of referential disorder, we use words like ‘ruler’ or ‘doctor’ with a sense given by their teleological or functional role; but we apply them, on the basis of habit and social convention, to people who (as we may or may not be aware) do not in fact perform it. If the disorder is deeply enough embedded in our linguistic practice, so that we predominantly use ‘F’ for objects which lack Fness correctly understood, the phenomenon repeats at the meta‐level: the name ‘F’ will not really do what a name is supposed to, that is, divide the F objects from other things, and so will not really deserve the name ‘name’. (This is, I believe, a major worry being explored in Plato's treatment of naming in the Cratylus.) Social roles such as crafts, offices, and practical identities are the natural habitat for referential disorder, for we need to use these terms in ways that reflect realities on the ground: we can hardly get by without using ‘doctor’, ‘ruler’, ‘lover’, ‘hero’, ‘friend’, etc. for the socially accepted occupants of those roles. But at the same time, such roles typically do have functional specifications built in, often of a moral or epistemic kind; these are easily enough revealed under dialectical scrutiny, and so too is the fact that their socially recognized occupants – the Fs ‘loosely speaking’ – may or may not meet them. As Thrasymachus allows, everyone will say that the ruler errs, including himself; but that is because the conventional denotation of ‘ruler’ is so far from being the correct one that for practical purposes we have to allow for loose talk as well as the precise kind. The best we can hope for is to be aware of which we are employing at any given time.Referential disorder is one of the main reasons redoubled qua, and girqps in particular, are so important to Plato. For a redoubled qua predication invites us, so to speak, to consider what really is F, and only insofar as it is F. The referential openness of this kind of claim, and the invitation to think of Fness in graded terms, mean that it immediately invites reflection and inquiry as to what should really count as F and how far, in a way in which universal generalizations do not.28Cf. the ‘principle of real reference’ in Penner (1991) and Penner and Rowe (2005), and the ‘dominance view’ of desire in their (1994); similarly, Reshotko (1992) and later works. I would not agree that Plato takes intentional psychological states to standardly have this ‘whatever really is F’ form, so as to float free of the agent's own beliefs about what satisfies the descriptions he uses. But qua‐ified attitudes do have this structure, and Penner can be read as working with the insight that far more of Plato (and of everyday thought and speech) is qua‐ified than has usually been recognised. However, Penner himself rejects qua‐ified analysis of the key attitudes in Penner (1991) at 197–8. Qua‐ification thus serves as a very immediate gateway to philosophical inquiry. It expresses an aspiration to think and speak clearly and correctly about how functions, natures, and other properties are related to each other – even if they are only instantiated very defectively by the objects around us, and no matter how misleading our loose talk about them may be.PLATONIC PHILOSOPHY AND THE METAPHYSICS OF QUASo qua predication is at home in the analysis of property networks; in accounts of causal powers; and in arguments about crafts, social roles, and functional categories. That covers quite a lot. Indeed, at this point, it may start to look like pretty nearly any Platonic argument will, by my standards, be a good candidate for reading in girqp terms. And so much the better. For, according to Plato, philosophical dialectic is indeed not about particulars, classes or sets of particulars, but about how various kinds or natures are related. This is made particularly explicit in a passage that deserves more serious attention than it gets as a guiderail for Platonic interpretation, the so‐called ‘Philosophical Digression’ of the Theaetetus (172c‐77c). Here, Socrates draws an elaborate contrast between the life of the philosopher and that of the careerist lawyer. The philosopher is aloof from social hubbub and practical concerns. Uninterested in particular humans like his neighbour, the question he asks is, ‘What is Man? What actions and passions properly belong to human nature and distinguish it from all other beings? This is what he wants to know and concerns himself to investigate’ (174a8‐b6). Likewise on ethical topics, in contrast to the lawyer, he is interested not in who did injustice to whom or whether the king is happy, but in general inquiry into kingship, and into human happiness and misery in general (175b9‐c6).In short, philosophical discourse – strict talk, the kind we should aspire to – is concerned with general truths about the natures of things. The business of philosophy is the production of true, explanatory, defensible, systematically organized qua predications about those natures: ‘precise accounts’, as Thrasymachus would say (Rep. 340e). The particular objects around us, on the other hand, are for Plato literally, not what philosophy is about. Here as elsewhere, seeing that Plato is thinking in qua terms can make a puzzling argument much less so. Interpreters have felt some anxiety about the way in which the digression portrait of the philosopher seems not to describe Socrates himself, and have wondered if some sort of irony or self‐undermining must be in play. But Socrates is just doing what he says philosophers are supposed to do: speaking not about individuals (such as himself) but about philosophers as a kind. And what is distinctive of the kind, on his view, is disinterested inquiry into the general natures of things. Since Socrates himself engages in disinterested inquiry into the nature of virtue, he should indeed count as a philosopher so understood; but so do astronomers, who might seem very different if we do not inquire into the features of the general kind.It might be responded that this much is unsurprising: for what philosophy is about, according to Plato, is Forms, that is, general kinds held to be firmly distinct from and independent of sensible particulars. This brings us to a question I have so far avoided: what relation does qua predication have to the theory of Forms? I am not entirely sure what to say about this, but we can start with a negative: Platonic qua does not depend on the theory of Forms. Qua predication is after all part of everyday linguistic practice: we all can express and understand girqps about all sorts of topics, gossips and cyclists included, with no metaphysics implied or required. So I would suggest that the phenomenon of qua predication comes first, so to speak; Plato's own philosophical uses of it, as charted in Sections 1, 2 and 4, come second; and the theory of Forms comes in third place as, among other things, Plato's reflective attempt to articulate the presuppositions of both ordinary usage and his own practice. For at least from the Phaedo onwards, one of the jobs of Forms is to serve as the primary referents for the terms that pick them out; our words refer only secondarily and indirectly, by ‘eponymy’, to the many particulars which participate in the Form (103b). The Cratylus account of naming as a matter of natural correctness thus analyses names ‐‐ true names, the kind deserving of the name ‘name’ ‐‐ as attempts to express the natures of things. (The Cratylus is an enigmatic dialogue, however, and whether Plato in the end accepts this account is an open question.) This explains how qua predication can be, for Plato, a kind of default: when I use a general term F, I refer first and foremost to the nature which it picks out, Fness, and through its mediation to whatever objects participate in that nature, insofar as and because they do so. This linguistic order mirrors the causal one: it is because beautiful things participate in the Form, Beauty itself, that they are beautiful (Phaedo 100b‐e, 101c). This does not yet entail, however, that there is a Form for every kind of thing to which we can successfully refer, or which can be the subject of qua predication. It might be that we need to postulate only some metaphysically privileged subset of things as having Forms, while others, as Socrates somewhat mysteriously says in the Parmenides, are ‘just as we see’ (130d). The question is difficult to investigate (and I will evade it here) because the scope of the Forms – that is, the question what things there are Forms of – is notoriously unclear, and seems to be rethought from dialogue to dialogue.29For an account of Forms which makes some use of qua predication, see Ademollo (2013).A broader and more useful framework for thinking about the metaphysics of qua is a notion I have already, almost unavoidably, been using: nature [phusis], as in the phrase ‘the natures Fness and Gness’, and more generally the natures [phuseis] of things. According to the theory of the Cratylus, names (i.e. descriptive terms in general) are natural or in accordance with nature when they pick out beings, essences or kinds. Phusis sometimes serves as equivalent to ‘Form’, but it often seems to be much broader in scope; and when it is used in relation to Forms, it tends to refer not exactly to the Form itself but to what the Form and its instances have in common – for example, the kind which some craftsman who looks to a Form grasps and puts into the product he creates (Crat. 389b‐c).30Cf., for example, Phaedr. 254b, Rep. 476b, Crat. 389b, Rep. 501b, Phaed. 103b, Rep. 597b4, c2 (for en phusei, ‘in nature’, as apparently a reference to the realm of Forms; cf. Adam's (1902) notes ad loc.), Parm. 132d. Social and functional roles, causal powers, and properties related to others in qua‐ified networks are all good candidates for phuseis. And Forms, whatever (else) they are, should perhaps be conceived as the privileged subset of natures which we must postulate as somehow metaphysically primary, in order to explain how things, in general, are – including all the other things for which there is no Form.In sum, I suggest that we see qua predication as being prior to metaphysics in the trajectory of Plato's thinking, and the metaphysics of natures as in turn prior to and broader than the metaphysics of Forms. And certainly, no metaphysics is required to see why qua predications would be important and pervasive in Plato's discourse – or our own. As assertions of causally asymmetrical correlations between kinds, they are naturally central to explanatory talk; and a great deal of our talk aspires to be explanatory. The point is easiest to see when put in terms of pragmatic considerations. With the possible exception of the Parmenidean One, every object of interest to us bears indefinitely many descriptions; whenever we set out to predicate a property of a subject, we have to choose among indefinitely many possibilities for picking that subject out. One obvious way to make that choice is on the basis of relevance, and one obvious kind of relevance is a causal relation to the possession of the property in question. Presumptions of relevance thus create a standing pressure towards speaking and interpreting speech in qua terms. This can explain why, for instance, it is deprecated for a newspaper headline to identify a criminal by a racial, ethnic or religious descriptor. (Here, we start to converge with recent discussions both of generics and of relevance in communication.)31On relevance theory, the seminal work is Sperber and Wilson (1986); on generics see Leslie (2017), and cf. also the uses of ‘Aristotelian categoricals’ in Thompson (2008). If my account of Plato is right, and if Plato is right about how we all think and talk, the real ‘original sin’ behind generics is an uncritical, system A impulse to interpret each other's assertions as explanatory. If a headline says ‘Muslim arrested in robbery’, presumptions of relevance suggest that the suspect's religion, rather than say his poverty or opioid addiction, is being singled out as causally relevant to his crime: Platonic qua is sufficiently our own default mode that, all else being equal, we hear an implicit ‘qua Muslim’. Likewise with a general claim like ‘F's have a high rate of street crime’ for some ethnic descriptor ‘F’. We hear this by default not as ‘F's qua‐who‐knows‐what‐independent‐variable’ or ‘The set of Fs, for no particular reason’, but as a claim about Fs qua Fs – perhaps even, Platonically, about the causal power of their Fness.32What about the kind of annoyance people sometimes feel at headlines like ‘Mother of two wins Nobel Prize’? Here, the descriptor has been chosen precisely as non‐relevant and surprising, hence presumptively intriguing to the reader. (Relevance is obviously not the only value we select for; or perhaps we should say that causal salience is not the only form of relevance.) The annoyance is at this decision not to focus on the scientist qua scientist, which seems undermining of that identity.Plato seems to be aware of this and indeed to operate with a particularly strong version of it. In a remarkable moment of the Phaedo, Socrates distinguishes between attributing a property to a sensible particular directly and via a Form. Only the latter, apparently, ever properly captures the way the world is:But, he said, do you agree that, as for “Simmias is taller than Socrates,” that way of putting it is not how the truth actually holds [oukh hôs tois rhêmasi legetai houtô kai to alêthes echein]?33Taking to alêthês as the subject of echein: ‘how the truth holds’, that is, how things really are, cf. Crat. 384c1, Phaedr. 278c5, Gorg. 484c4. I am indebted for discussion of this passage to Gail Fine and Terry Irwin. It is not, surely, that Simmias has come to be [pephukenai] taller than Socrates for this reason, that he is Simmias [tôi Simmian einai],34Cf. ‘by being bees’, at Phaedo 72b. but rather because of the tallness he happens to have? Nor is he taller than Socrates because [hoti] Socrates is Socrates, but because Socrates has smallness compared with the tallness of the other?‐‐ True. (102b8‐c4)35On this passage cf. Dancy (2004), 306ff with references.Note that the failure here is not a failure of the sentences in question to be true in any standard modern sense: Simmias is in fact taller than Socrates. The objection is that saying so still fails in some way to capture the way the world is. It is a kind of ‘loose talk’. Apparently, when I use ‘F’ to name some subject in order to predicate Gness of it, it fully captures the truth only if Gness belongs to that object by virtue of its nature as F. In order to capture the truth in full, without being misleading, our choice of terms must perspicuously track salient causal or explanatory relations, and these have to do with the natures of things. And although it is not the point directly being made here, this does have implications for truth in a more familiar sense: for if we presume that our interlocutors are attempting to conform to this demand, our default will be to hear an implicit ‘qua Simmias’ after the subject term, which does render false the sentence, ‘Simmias is taller than Socrates’.36Cf. ‘some pleasures are bad’, as voiced by the Many in the Protagoras (as discussed in Section I above), to which Socrates also objects (misleading since it is not qua pleasures that they are bad), and ‘the doctor errs’, which Thrasymachus deprecates (since it is not qua doctor that he does so). That ‘Simmias’ is a proper name does not exempt the sentence from this expectation, but points to why it cannot be met. For the names of individuals do not correspond to natures in the right way: it is not at all clear what ‘Simmias‐qua‐Simmias’ or ‘Simmias insofar as he is Simmias’ would pick out, and what if anything could be predicated of him (‘A human being’ picks out his nature; but this is not a very promising answer if we are interested in the differences between him and Socrates). Simmias seems to be little more than a kind of nexus, a locus for the various prior and more general natures – such as Tallness – which compete and interact to determine his properties. That is one of the reasons philosophical dialectic is about things like them, and only ever indirectly and unsatisfactorily about things like him.For Plato, it is the central job of the philosopher to sort out patterns and relationships among kinds or natures, while keeping a critical eye on their putative instantiations around us. That is what it is to ‘carve nature at the joints’ (Phaedrus 265d‐6b, cf. Statesman 262c‐3a). And it is by performing this dialectical work that we gain the resources necessary to see where everyday conventional reference has come to be disordered, and to rectify it in light of a grasp of the true natures of things. It is, I think, fair to refer to this as a project of analysis: an analysis of Fnesses and Gnesses as natures, which means the determination of their true causal powers and relations. For Plato, then, philosophy is inherently analytical; and philosophical analysis is inherently a critical, even a political, enterprise.CONCLUSIONSI close with two reflections that gesture towards further questions worth exploring. First, so far as I can see, Plato is the first and last philosopher to have taken qua predication to be our default mode of expression: both as the correct vehicle for strict, properly philosophical discourse and as pervasively embedded and implicit in ordinary talk. If he is right, then many sentences that might look to us like universal generalizations are rather to be understood as girqps; and this will include many of those now classified as generic generalizations. And if a large and important class of generics are best understood as girqps, then it turns out that they are not really generalizations at all.37The test for whether a generic does express a girqp is straightforward: can we add ‘qua Fs’ without change to its truth conditions? They do not claim that all Fs are Gs – not even that most of them are, or the most characteristic or most salient ones. Rather they are irreducibly explanatory, and deal in natures or causal powers rather than sets of individuals: what they bring out is that the Gness of those Fs which are Gs is due to their Fness. This diagnosis fits with some of the most striking features of generics: the truth–condition gap, our tendency to derive them from very small samples and their default status and apparent cognitive priority to universal claims.38For a useful overview, see Lerner and Leslie (2016). That causal considerations have some kind of bearing on generics is not a new thought: but the Platonic proposal that it is the defining feature of a large class of them seems to me to be both radical and worth consideration.Second (and although I cannot here argue properly for the ‘first and last’ claim above), qua predication in Aristotle and later Western philosophy seems to me very different. Clearly qua predication is important to Aristotle – he provides qua with its most memorable star turn in philosophy, after all, in designating metaphysics ‘the study of being qua being’. Aristotle also offers much more than Plato in the way of explicit analysis of qua predication, in several different contexts.39A survey of qua in Aristotle would have to consider: (1) its role in his diagnosis of sophistic fallacy (cf. Sophistici Elenchi 168a34‐b10, and see Section I above on the original philosophical uses of qua), and the related puzzles generated by accidental unities (cf. Cohen, 2013); (2) the reduplicative syllogism (Prior Analytics I.38, cf. Bäck, 1996); (3) the role of qua‐ification in demonstrative science, which studies some class of substances only qua Fs (Posterior Analytics I.6 and passim; cf. Malink, 2009), so that, for instance, mathematics deals with objects which are sensible but not qua sensible (Metaphysics XIII.3, 1077b17‐8a4, cf. 1078a1‐4, 23–6), while metaphysics deals with beings qua beings; and (4) miscellaneous deployments of distinction‐drawing qua scattered across the corpus. Particularly important instances of this last are his claims (i) that when we see Callias, we perceive him not qua Callias but qua human being (An. Post. II.19, 100a17‐b3); and (ii) that a failure to distinguish qua‐ifications in relation to not‐being is what led the early philosophers into confusions about ‘what is not’ (Physics I.8, 191b1‐9). But this amounts to a frustrating fragmentation of the phenomenon, and whether qua talk remains as central to his philosophical discourse as it is to Plato's is controversial.40See Malink (2009), Malink (2013) on the ‘orthodox’ and ‘heterodox’ readings; for non‐extensionalist Aristotelian semantics cf. also Castagnoli (2016). Granted, Aristotle's own understanding of scientific discourse can be construed in terms of Platonic qua; for scientific demonstration [apodeixis], as Aristotle understands it, is precisely an articulation of the properties which belong to Fs qua Fs [kath' hauta, per se]. But whether Aristotle himself understands such discourse in terms of Platonic qua is another question: on a traditional reading, he is instead committed to conceiving of demonstration in terms of universal generalizations – despite the notorious difficulty in reconciling this with his admission that natural science deals in claims which are true only ‘for the most part’.41Cf. Mann (2000): on Plato see especially 77ff. And whether the transformation is rightly attributed to Aristotle himself or not, universal generalizations about more or less Aristotelian substances have since then come to be seen as the default mode of scientific discourse. This transformation has helped to make qua predication hard for us to see, even as it continues to be deeply embedded in our own everyday speech and thought. And so, it seems to me worth trying to recover and recognize this more ancient, and perhaps more basic, way of talking and thinking.42In addition to the Annual Analytic Philosophy Symposium at the University of Texas in 2018, this paper owes a great deal to discussion with the Stanford Metaphysics and Epistemology Conference; the Oxymorons group; the Institute of Classical Studies in London; the B Club at Cambridge University; the Philosophy Department at the University of Pittsburgh; the University of Oklahoma conference honouring Hugh Benson; the Notre Dame University Ancient Philosophy Workshop; and Rachana Kamtekar's seminar at Cornell. I have been particularly helped by comments and questions from Francesco Ademollo, Andreas Anagnostopoulos, Merrick Anderson, Jonathan Beere, Hugh Benson, Tad Brennan, Nick Denyer, David Ebrey, Gail Fine, Benj Hellie, Terence Irwin, Fiona Leigh, Rachana Kamtekar, Sean Kelsey, Martin Lin, Ben Morison, Gurpreet Rattan, Gretchen Reydam‐Schils, Kenneth Sayre, David Sosa, Jan Szaif, Michael Thompson, Jennifer Whiting, Jessica Wilson, and Raphael Woolf. I am deeply grateful to James Allen, Willie Costello, Nick Denyer, Mark Gatten, Rory Harder, Marko Malink, Wolfgang Mann, Stephen Menn, Jennifer Nagel, Michail Peramatzis, David Sedley, Thomas Slabon and Rob Stainton, who generously gave comments on the paper as a whole which have saved me from a multitude of blunders. I am sure that many remain; I am also sure that I have forgotten the names of others from whom I have learned important things, for which I apologize.REFERENCESAdam, J. (1902). The republic of Plato, 2 volumes. Cambridge University Press.Ademollo, F. (2013). Plato's conception of the forms: Some remarks. In R. Chiaradonna & G. Galluzzo (Eds.), Universals in ancient philosophy (pp. 41–86). Edizione della Normale.Bäck, A. (1982). Syllogisms with reduplication in Aristotle. Notre Dame Journal of Formal Logic, 23(4), 453–458.Bäck, A. (1996). On reduplication: Logical theories of qualification. Brill.Bäck, A. (2002). The role of qualification. Journal of Philosophical Research, 27, 159–171.Barney, R. (2001). Names and nature in Plato's Cratylus. Routledge.Barney, R. (2006). Socrates' refutation of Thrasymachus. In G. Santas (Ed.), The Blackwell guide to Plato's republic (pp. 44–62). Blackwell.Barney, R. (2020). Technê as a model for virtue in Plato. In T. Johansen (Ed.), Productive knowledge in ancient philosophy (pp. 62–85). Cambridge University Press.Benson, H. (1997). Socratic dynamic theory: A sketch. Apeiron, 30, 79–93.Brunschwig, J. (1984/5). Aristotle on arguments without winners or losers. In Jahrbuch Des Wissenschaftskolleg Zu Berlin (pp. 31–40). Siedler Verlag. https://www.wiko‐berlin.de/wikothek/jahrbuchCastagnoli, L. (2016). Aristotle on the non‐cause fallacy. History and Philosophy of Logic, 37(1), 9–32.Castelnérac, B., & Marion, M. (2009). Arguing for inconsistency: Dialectical games in the academy. In G. Primiero et & S. Rahman (Eds.), Acts of knowledge: History, philosophy and logic: Essays dedicated to Göran Sundholm (pp. 37–76). College Publications.Castelnérac, B., & Marion, M. (2013). Antilogic. The Baltic International Yearbook of, Antilogic Cognition, Logic and Communication, 8, 1–31. https://doi.org/10.4148/1944‐3676.1079Cohen, S. M. (2013). Accidental beings in Aristotle's ontology. In G. Anagnostopoulos & F. Miller (Eds.), Reason and analysis in ancient Greek philosophy (pp. 231–242). Springer.Cooper, J. M. (1997). Plato: Complete works. Hackett.Dancy, R. M. (2004). Plato's introduction of forms. Cambridge University Press.Dretske, F. (1977). Laws of nature. Philosophy of Science, 44, 248–268.El Murr, D. (2020). Eristic, antilogy, and the equal disposition of men and women (Plato, resp. 5.453B‐454C). The Classical Quarterly, 70(1), 85–100.Fine, K. (1982). Acts, events, and things. In W. Leinfellner, E. Kraemer, & J. Schank (Eds.), Sprache und Ontologie (pp. 97–105). Holder‐Pichler‐Tempsky.Lear, J. (1982). Aristotle's philosophy of mathematics. Philosophical Review, 91(2), 161–192.Lefebvre, D. (2018). Dynamis: Sens et genèse de la notion aristotélicienne de la puissance. J. Vrin.Lerner, A., & Leslie, S. J. (2016). Generics and experimental philosophy. In W. Buckwalter & J. Sytsma (Eds.), The Blackwell companion to experimental philosophy (pp. 404–417). Blackwell.Leslie, S. J. (2017). The original sin of cognition: Fear, prejudice, and generalization. Journal of Philosophy, 114, 1–29.Malink, M. (2009). A non‐extensional notion of conversion in the organon. Oxford Studies in Ancient Philosophy, 37, 105–141.Malink, M. (2013). Aristotle's modal syllogistic. Harvard University Press.Mann, W.‐R. (2000). Aristotle and the discovery of things. Princeton University Press.Moraux, P. (1968). La joute dialectique d'apres le huitieme livre des Topiques. In G. E. L. Owen (Ed.), Aristotle on dialectic (pp. 277–311). Oxford University Press.Nehamas, A. (1990). Antilogic, sophistic, dialectic: Plato's demarcation of philosophy from sophistry. History of Philosophy Quarterly, 7(1), 3–16.Penner, T. (1991). Desire and power in Socrates: The argument of “Gorgias” 466A‐468E that orators and tyrants have No power in the City. Apeiron, 24(3), 147–202.Penner, T., & Rowe, C. (1994). Desire for the good: Is the Meno inconsistent with the Gorgias? Phronesis, 39(1), 1–25.Penner, T., & Rowe, C. (2005). Plato's Lysis. Cambridge University Press.Reshotko, N. (1992). The Socratic theory of motivation. Apeiron, 25(3), 145–169.Ryle, G. (1966). Plato's progress. Cambridge University Press.Sedley, D. (1998). Platonic causes. Phronesis, 43, 114–132.Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Harvard University Press.Szabó, Z. G. (2003). On qualification. Philosophical Perspectives, 17, 385–414.Taylor, A. E. (1960). Plato (7th ed.. (1st ed 1926)). Methuen.Thompson, M. (2008). Life and action: Elementary structures of practice and practical thought. Harvard University Press.Wolfsdorf, D. (2005). Δυναμις in Laches. Phoenix, 59, 324–347. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Analytic Philosophy Wiley

Platonic qua predication

Barney, Rachel
Analytic Philosophy , Volume Early View – Jun 8, 2023
Download PDF
20 pages

Loading next page...
 
/lp/wiley/platonic-qua-predication-R8VHEEmlfU

References (42)

Publisher
Wiley
Copyright
© 2023 John Wiley & Sons Ltd.
ISSN
2153-9596
eISSN
2153-960X
DOI
10.1111/phib.12306
Publisher site
See Article on Publisher Site

Abstract

Plato is fond of a kind of predication which is distinctive in important ways and which, so far as I know, has no name. As a result, it tends to go unrecognized, and the arguments it appears in are easily misunderstood. My purpose here is to draw attention to it, giving a rough account of how it works and why it matters.My topic is what I will call qua predication in Plato, or Platonic qua. And my interest will be not in any and every sentence involving a qua‐like expression, but more specifically in the form of general sentence in which Fs are claimed to be Gs qua Fs, but with the qua phrase left implicit: so, general implicit redoubled qua predications, or girqps.1Pronounced, I am sorry to say, ‘girk‐p’. (The ‘redoubled’ picks out the fact that the subject term F repeats in the qua phrase.) I will try to show that these sentences are a distinctive kind of talk in Plato, not reducible to anything else. In particular, they are not to be confused with the universal generalizations expressed in the same words – what I will term the corresponding universal generalizations. In fact, qua predications are, I will argue, best understood not as generalizations about F individuals at all, but as claims linking the properties or natures of Fness and Gness themselves: their purpose is to point to a correlation and at least suggest a causal relation between the two, although how clearly and distinctly this causal claim emerges varies greatly case by case (see Section 4). I here and throughout will use the terms ‘causal’, ‘causation’, etc. in the very broad way familiar from Aristotle's system of the four ‘causes’ [aitiai], which cover the full range of answers to a ‘why’ question (not merely ‘efficient’ causality), and formal explanatory relations in particular.2Hence, I use ‘explanatory’ interchangeably with ‘causal’; one might also speak of some types of girqps as setting out relations of grounding.These girqps are for Plato the proper vehicle of philosophical discourse. But nothing in his use of them depends on his distinctive metaphysics; indeed, part of their importance is that he also takes them to be pervasive in everyday, confused non‐philosophical discourse. And it seems to me that Plato is right about that. So, part of my task will be to show that – for us as for the ancient Greeks – girqps are indeed a standard mode of everyday speech – one curiously marginalized both in the intervening philosophical tradition and in modern analyses of predication.3This is not to deny that some forms of qua‐ification have received philosophical attention. For the long history of the Aristotelian ‘reduplicated syllogism’, which played an important role in medieval logic and in the metaphysics of philosophers as late as Spinoza and Leibniz, see Bäck (1996). Qua also figures in a range of recent philosophical studies including Szabó (2003), Fine (1982), and Bäck (2002). But in none of these does the qua‐ification in view seem to me to be quite the kind we find in Platonic girqps.It will help to begin by looking at some Platonic texts in which qua‐ification is explicit. So, in Section 1, I work through a few case studies of explicit Platonic qua; Section 2 sums up with a preliminary analysis of how it works. In Section 3, I step away from Plato to make the case for the pervasiveness of girqps in everyday speech; in Section 4, I set out the three principal ways in which Plato seems to deploy them. Section 5 explores their significance for his metaphysics and his conception of philosophy. By way of conclusion, Section 6 sketches a brief contrast with Aristotle, and brings out some implications for the study of generics, a category with which Platonic girqps substantially overlap.EXPLICIT PLATONIC QUA: SOME TEXTSPlato's Lysis features what looks like a proof by cases that friendship is impossible (214a‐16c). Socrates is trying to discover what sorts of people are friends with each other. He argues by various ingenious sophistic means that bad people cannot be friends with each other (for they harm each other); nor can one good person be friend to another insofar as [kath' hoson] he is like (for those who are like cannot benefit each other) (215a3‐5); nor could the good person be friend to another good person insofar as [kath' hoson] he is good (for the good are self‐sufficient) (215a6‐7); nor can unlike people be friends (for they are enemies). And since the good and the bad and the like and the unlike together seem to pretty much cover the bases, we are left with the alarming result of friendship nihilism. Or are we?A question mark is placed beside this conclusion by the key phrase I noted above: ‘insofar as’ [kath' hoson]. This phrase is explicit only at three points (214e4, 215a5, a6), but implies a qualification and complication which pervades the argument as a whole. For it expresses a landmark philosophical insight: we pick out subjects under different descriptions, which can be used to express different facts about them. Socrates might be friends with Crito insofar as they are both good but not insofar as they are both like, or vice versa; there is a significant distinction between the two facts even if it is by being good that they are like each other.To see what the argument really intends, we need to take a close look at how this ‘insofar as’ talk works. The kath' hoson locution, like its English counterpart ‘insofar as’, often serves to flag gradability of predicates; but here its primary purpose seems to be simply to pick out the different descriptions under which it might be true or false that two people are friends. That makes it roughly equivalent to the Greek hêi, of which the Latin term qua is a translation.4Only roughly because hêi and qua unlike kath' hoson do not particularly suggest gradability. I treat this as a difference of nuance, since as we will see shortly gradability is not part of what a qua predication actually asserts: it asserts only a correlation between two predicates, which may also apply to ungradable ones (See Section 3). Plato uses hêi similarly in an important passage of the Meno, where Socrates insists that a proper definition is one which picks out a unity:But Meno, … if I were asking you what is the nature [ousia] of bees, and you said that they are many and of all kinds, what would you answer if I asked you: “Do you mean that they are many and varied and different from one another in being bees [tôi melittas einai]? Or are they no different in that regard, but in some other respect, in their beauty, for example, or in their size or in some other such way?” Tell me, what would you answer if thus questioned?‐‐ I would say that they do not differ from one another insofar as they are bees [hêi melittai eisin]. (72a8‐b9)5Quotations from Plato are in the translations by many hands in Cooper (1997), sometimes with revisions.In the very late Philebus, by contrast, Socrates argues, against the hedonist Protarchus, that there is no unity to the diverse things, some good and others bad, which we call ‘pleasures’. Protagoras balks:What are you saying, Socrates? Do you think anyone will agree to this who begins by laying it down that pleasure is the good? Do you think he will accept it when you say that some pleasures are good but others are bad?‐‐ But you will grant that they are unlike each other and that some are opposites?Not insofar as [kath' hoson] they are pleasures. (13b3‐c5)Here, we see kath' hoson doing exactly what hêi does in the Meno: it picks out a feature or aspect of the subject in respect of which a further predicate – in both cases, the tricky second‐order property of likeness or sameness – is said to obtain. So, I take it that kath' hoson and hêi are in the relevant constructions interchangeable; I count sentences where either occurs or seems to be implicit as qua predications.6Willie Costello points out to me that redoubled qua predications can be expressed in a multitude of other ways in Greek (as in English), including causal datives and kath' hauto expressions. I focus on hêi and kath' hoson for simplicity's sake, not because they exhaust the possibilities.At this point, we can already note several significant features of Platonic qua. First, a qua predication may be explicit, as in the Meno, Philebus and those three crucial moments of the Lysis argument, or implicit (as in the other related premises of the Lysis argument). And it may or may not be redoubled: that is, it may pick out Fs qua Fs or qua some other feature G.7‘Redoubled’ here is not to be confused with ‘reduplicative’ [epanodiploumenon] in Aristotle. Aristotle's account of the reduplicative syllogism is far from clear, but it seems that any syllogism revolving around explicit qua predication counts as reduplicative, and his focus in Prior Analytics I.38 is on the non‐redoubled case (i.e., ‘Fs qua G's are H's’). There is a long medieval tradition of analysis of the reduplicative syllogism starting from this text; see Bäck (1982) and (1996). In both Meno and Philebus, we encounter redoubled qua: bees qua bees, pleasures insofar as they are pleasures. In the Lysis, we get both kinds: Socrates speaks in turn of good people qua like each other and good people qua good, where only the second is a case of redoubled qua.In English, redoubled qua‐ification can be either implicit or expressed using an impressive variety of locutions: ‘Fs qua Fs are Gs’, ‘Fs are Gs insofar as they are Fs’, ‘Fs as such are Gs’, ‘inasmuch as something is F, it is G’, ‘what is F is to that extent G’, ‘strictly speaking, Fs are Gs’ (see below for this locution in Republic I) and so on.8I do not here include ‘As an F, x is G’, since these are not always explanatory in the way which I will argue is characteristic of Platonic qua. In fact, the discussion of predications of this form in Szabó (2003) brings out that they may do something very different from Platonic girqps. Consider his canonical sentence ‘John as a judge earns $50,000’, which is to my ear problematically ambiguous. If it were understood as instantiating the claim ‘Judges (as such) earn $50,000’, we would be dealing with an application of Platonic qua to a particular case: John earns that sum because he is a judge and that is what judges earn. But for that claim, more conventional and perspicuous phrasing would be ‘As a judge, John earns $50,000’. Alternatively, if the sentence claims only that John's judging work earns him that salary (with nothing implied or suggested about the wages of any other judge), then a better, less ambiguous phrasing would be ‘John earns $50,000 as a judge’. That some as‐phrases do something very different from Platonic qua seems clear. Consider: ‘As a composer, Mozart was a model of delicacy; as a man, he was a boor’. It would be a terrible mistake for a non‐native speaker to protest that there is no correlation, let alone causation, between being a composer and being delicate or being a man and being a boor (Contrast: ‘As a citizen of Vienna, Mozart was liable for taxation by the Austrian crown’, where the as phrase is explanatory.) This sentence and Szabo's seem intended only to metaphysically ‘partition’ their subject so as to restrict the application of the predicate, a form which in medieval philosophy is termed specificative qua: see Bäck (1998) and compare Lear (1982) on Aristotle's qua operator. I cannot here engage with this or the other permutations of qua in medieval logic (and later quatenus in Spinoza), which owe more to Aristotle than to Plato and present many complications of their own: see Bäck (1996). I take it that all of these formulations say roughly the same thing, in the absence of any differentiation by context; and these rich linguistic resources are, I think, telling. They strongly suggest that redoubled qua predication is not just a local linguistic anomaly, a by‐product or epiphenomenon; still less is it peculiarly Greek or dependent on proprietary Platonic philosophical assumptions. It is a way of saying something distinctive that we, like the Greeks, very much want to be able to say.9I discuss more fully in the next section what that something is; I will, however, be leaving open the question of what we should understand the logical form of girqps to be. Syntactically, it would seem that in the sentence ‘Fs qua Fs are Gs’, ‘qua Fs’ should be understood as an adverbial phrase modifying the copula ‘is’, and so belong to the predicate. (As Nicholas Denyer pointed out to me, this seems clear in the non‐redoubled case: ‘As Members of Parliament, Cabinet Ministers already earn a healthy pension’.) But semantically, when, for instance, Thrasymachus says that ‘The doctor qua doctor never errs’, he can equally well be heard as intending ‘qua doctor’ as a determinant of the subject of the sentence, indicating that its reference is to a nature or essence reifiable as ‘the doctor qua doctor’ (see Sections 3–5). I am not sure how to represent girqps in a way that respects this tension or ambiguity. Sandra Peterson has suggested to me (on the basis of the puzzles of scope raised by negations of qua claims, discussed in Section 2) that ‘qua Fs’ might best be understood as a kind of sentential operator (Cf. also Lear (1982) on Aristotle's qua operator). I thank Nicholas Denyer and Ben Morison for pressing me on this question.It is striking that in both the Meno and the Philebus (and, as we will see, in Republic I), it is Socrates' interlocutor who makes the qua construction explicit – as if to show that the move is well understood by any contemporary intellectual. And this can help us to see how philosophical use of qua gets started. For it seems clear that the original home of qua‐ification was eristic or dialectic. This is the form of game‐like intellectual combat for two players, going back to the sophistic movement of the 5th century BCE, which is discussed in Aristotle's Topics and Sophistici Elenchi. In eristic, one speaker propounds a thesis, and the other attempts to refute him by posing a chain of questions demanding yes/no answers; the aim is to lead him into the contradiction of that thesis, or some other absurd result.10On eristic see Castelnérac and Marion (2009, 2013), Nehamas (1990), Brunschwig (1984/5), Moraux (1968), and Chapter 4 of Ryle (1966). One common kind of eristic strategy moves from qua‐ified claims to the corresponding universal generalization. For instance, in the Hippias Major, Hippias puts forth the thesis that the beneficial is the fine. Socrates gets him to agree that the beneficial is the cause of the good; and that it follows that ‘the fine is not good, nor the good fine’, since the cause and the effect cannot be the same (296e‐7c). This crucial move is supported by an analogy: ‘the father is not a son, and the son is not a father’ (297c‐d). This is fair enough as a claim about fathers qua fathers, of course; but it obviously does not license the corresponding universal generalization that no father is a son or vice versa. Another kind of sophistic fallacy seems to have turned on equivocation between different possible qua‐ifications. An example Aristotle mentions in the Sophistic Elenchi is that if I own a work of art painted by someone else, it is both mine (qua property) and not mine (qua creation) (179b4‐6, cf. Euthydemus 298d‐e with Sophistic Elenchi 180a1‐5). Bringing mismatched qua‐ifications to light is thus an important way for the respondent in the eristic game to block or dissolve fallacious moves. The primordial philosophical use of qua constructions, then, must have been for what we might call distinction‐drawing or dialectical qua: cases in which two different possible qua‐ifications are in play and the point of philosophical interest is to distinguish them and to get clear about what can and cannot legitimately be inferred from each.Plato does occasionally deploy distinction‐drawing qua, but it is less common in his works than one might expect.11For example, Ion 540e, and the (possibly inauthentic) Alcibiades I, 115a‐16d. Plato has other tools for drawing qua‐like distinctions in dialectical contexts, including the language of ‘natures’, e.g. at Republic 5.453B‐454C, on which see El Murr (2020). Of much greater interest to him is the locution in which the qua is redoubled but with no particular non‐redoubled alternative in view, so that there is no need for the qua phrase to be explicit: these will be our general implicit redoubled qua predications (girqps), sentences which claim that Fs < qua Fs > are Gs.PLATONIC QUA: A PRELIMINARY SKETCHThese first few cases already reveal the basic workings of Platonic qua. In the Lysis arguments, it seems obvious that the ‘insofar as’ locution points to a correlation between two predicates or descriptions F and G: specifically, between the candidate Fs of being like or good and the target G of being a friend. What I mean by ‘correlation’ is simply that the ‘insofar as’ construction draws our attention to the possibility that the predicates in question are gradable, and claims that their grades in any particular case will match up. It seems further that these claims are meant to affirm not only correlation but causation; for Socrates' inquiry throughout the Lysis is into what makes two people friends – the feature of a friend in virtue of or because of which the friendship arises. So read, the arguments serve to produce the important and quite plausible result that two of the most promising candidates for that causal role – likeness and goodness – fail, as do their opposites.12Cf. Taylor (1960): ‘it is not true either that any and every “likeness” nor yet that every and any “unlikeness” can be the foundation of friendship’ (70 n. 1).The upshot of this reading is, of course, that, so read, Socrates' alarming conclusions are not to be identified with the corresponding universal generalizations, and do not entail them. Again, by the ‘corresponding universal generalization’ I mean the universal generalization which is expressed in the very same form of words as the girqp, so that either might be mistaken for the other. Of course, the whole point is that the two do not ‘correspond’ semantically: hence, the danger of fallacy and the importance of distinguishing them and registering the truth–condition gap between them.13By making the qua phrase of a girqp explicit, we can always produce a sentence which corresponds to it better semantically (‘No son qua son is a father’; ‘The doctor qua doctor acts for the health of the patient’), and if we construe this as a universal generalization it will be entailed by the girqp. But so construed, I will suggest, it still does not fully capture the meaning of the girqp itself. Cf. Sections 2 and 3 and n. 18 below. The gap is particularly tricky to navigate where negations are involved, since these may differ in scope and force so as to allow for a weak and a strong reading of the same claims. On a weak reading of the Lysis arguments, they leave open the possibility that the features specified (the candidate Fs) are simply orthogonal to those which ground friendship so that, for instance, some like or good people might still be friends in virtue of some other feature of their character – as charming conversationalists, say, or birdwatchers. On the strong reading, the conclusion would be that being an F tends pro tanto to exclude being a G: it seems plausible that the argument that the bad are not friends is to be understood in this way. But even when the strong reading of a negative girqp is in view, the corresponding universal generalization does not immediately follow. The most obvious reason for this is the possibility of a defeater: that is, in any given case, another feature of the subject which might cancel or outweigh the qua‐ification under consideration. Even if likeness tends to exclude friendship, two like people might be friends in virtue of some other feature able to outweigh or override that likeness.Platonic qua claims, then, are always not only pro tanto but in principle defeasible in their application to particulars. And which ones can have defeaters, and in which kinds of cases, is a question for substantive philosophical inquiry, not something we can read off from the logical form of qua predication. In the Lysis, Plato likely means us to think that injustice really does trump any other feature of a prospective friend: there can be no defeaters in this case, and so ‘the bad are not friends’ is also true as a universal generalization (cf. Republic 351b‐2a, 580a). But whether the self‐sufficiency of the good person has the same power is not at all clear. Without further inquiry, qua‐ified arguments by themselves do not license inferences to the corresponding universal generalizations: this truth–condition gap vis‐à‐vis the corresponding universal generalization is one of the reasons that it is so important to recognize girqps (and, indeed, possible to do so). And the cheering upshot of reading Platonic arguments in implicitly qua‐ified terms is that it is far harder than we might think to refute them by counterexample. On the other hand, wherever we take Socrates or his author to be encouraging an inference to the corresponding universal generalization, as with sons and fathers at Hippias Major 297c, it becomes far harder to acquit him of deliberate fallacy.That Plato is alert to the truth–condition gap can be seen from a famous and complex argument at Protagoras 351b‐4e. Here, Socrates tries to get a reluctant Protagoras to answer dialectically for ‘the Many’, and to affirm hedonism both on their behalf and his own. Protagoras initially resists, insisting that only honourable pleasures are good. Socrates persists:Surely you don't, like the Many, call some pleasant things bad and some painful things good? I mean, isn't a pleasant thing good just in that [kath' ho] it is pleasant, that is, if it results in nothing other than pleasure; and, on the other hand, aren't painful things bad in the same way, just insofar as [kath' hoson] they are painful?…. Just insofar as [kath' hoson] things are pleasurable are they good? I am asking whether pleasure itself [hêdonê autê] is not a good. (351c2‐e3)Socrates is alleging confusion: the Many, and Protagoras with them, are misrepresenting their own position when they say that some pleasures are bad. Their real view is one demanding expression in qua form: what is pleasant is good insofar as it is pleasant, and what is painful is bad insofar as it is painful. Here, it is obvious that the correlation is symptomatic of a causal or explanatory relation: it is precisely because something is pleasant that it is, in the view imputed to the Many, a good thing. This is hedonism in its purest form, and the Many are quite wrong to think of themselves as something other than hedonists merely on the grounds that they reject the universal generalization ‘All pleasures are good’. Perspicuously expressed, hedonism consists precisely in their position that the pleasant qua pleasant is alone good; and this, of course, allows for the pains attendant on some particular pleasure to defeat its claim to goodness. It is precisely because of their commitment to hedonism that the Many reject the corresponding universal generalization which they took to express it.14Thomas Slabon has objected to me that Protagoras seems to interpret the position pressed on him by Socrates so that it makes purely extensional identity claims about the good and pleasant. I cannot here engage properly with this reading, but would suggest that if Protagoras does so, this may well be the point at which the argument becomes fallacious, as it clearly is when redeployed at 360aff.One final example of explicit qua‐ification is of a rather different sort. This is the star turn of ‘qua’ in Plato: the debate about the ruler ‘qua ruler’ or ‘in the strict sense’ in Republic I (340d‐7e).15I survey these arguments in Barney (2006). Here, it is the sophist Thrasymachus – once again, not Socrates but a dialectically trained interlocutor – who introduces qua‐ification, with vast philosophical consequences. He has stepped into a Socratic refutation by saying that justice is the advantage of the stronger, that is, the rulers (338c‐9b); and that for subjects to do what the rulers say is just (339b); and that the rulers sometimes make mistakes (339c). But it follows that if the rulers mistake their own advantage, obedience to them is both just and unjust. Invited to repair his position, Thrasymachus rather startlingly opts to retract his admission that rulers can make mistakes. ‘Do you think I'd call someone who is in error stronger at the very moment he errs?’ (340c6‐7). He continues:When someone makes an error in the treatment of patients, do you call him a doctor in regard to [kata] that very error? Or when someone make an error in accounting, do you call him an accountant in regard to [kata] that very error in calculation? I think that we express ourselves in words that do say [legomen tôi rhêmati houtôs] that the doctor is in error, and the accountant, and the grammarian. But I think that each of these, insofar as [kath' hoson] he is what we call him, never errs, so that, according to the precise account [kata ton akribên logon] ‐‐ since you too go in for precise accounts ‐‐ no craftsman ever errs. It's when his knowledge fails him that he makes an error, and in regard to that error he is no craftsman. No craftsman, expert or ruler makes an error at the moment when he is ruling, even though everyone will say that a physician or a ruler makes errors. It's in this loose way that you must also take the answer I gave earlier. But the most precise answer is this. A ruler, insofar as [kath' hoson] he is a ruler, never makes errors and unerringly decrees what is best for himself, and this his subject must do. (340d2‐41a3)Thrasymachus' concern here is to vindicate his conception of the ruler as expert practitioner of self‐interested injustice. Although this began life as an empirical claim, about ‘rulers in the cities’ (338d‐e), the really important point to him is normative. And what looks here like a claim about the perfect possessor of the craft really signals a more general qua‐ified stance. Any ruler whom we observe to fail in the practise of his self‐interested craft is, on that occasion and to that extent, defective as such – ‘ruler’ is, it turns out, a gradable predicate, and someone counts as a ruler precisely insofar as he is the expert practitioner of his craft. This conception of the ruler as craftsperson is one which Socrates picks up and makes his own: his whole conception of the philosophical Guardian, developed in the middle Books of the Republic, is his replacement for it. Socrates also, within Book I, offers his own rival conception of craft and the craftsperson, focused on the case of the ‘doctor qua doctor’ (341c‐50d).16On these arguments see Barney, 2006 and 2020.Claims about the doctor qua doctor and the ruler qua ruler are important cases of role specification qua, to be discussed in Section 4. For now, the main point is just to see what they have in common with our other cases, the better to fix the scope of Platonic qua. The claims here look very different: they are about kinds of agent, mostly in the singular (as of a prototype or perfect case) rather than the plural, and predicating actions rather than properties. And the immediate point of the qua‐ification is neither distinction drawing (the point is not to contrast technê with some other hat the agent might wear) nor explanatory, but criterial. That is, Thrasymachus and Socrates alike are concerned to specify conditions for the (precise or strict) ascription of terms like ‘ruler’ and ‘doctor’, by exploring the other predicates they entail; and we are to envisage this as guiding the application of the term to candidate particulars. (This project contrasts with that of the Lysis, where it seems to be presupposed that to predicate the term ‘friend’ in particular cases is unproblematic: the challenge is to explain what grounds it.) This criterial use, as we will see in Sections 4 and 5, is important because of the chasm between loose talk and the properly precise kind: while it is true that no real doctor or doctor strictly speaking ever errs, it is far from being true of those conventionally denoted by the term. I will have more to say about this phenomenon, referential disorder, later on.UN‐PLATONIC INTERLUDEIt should be clear by now that Platonic qua can be expressed in a wide variety of ways, and that it is apt for several different kinds of job. And since in the passages I have cited the qua‐ification fluctuates in and out of explicitness, in keeping with Plato's distaste for repetition and regimented language, it should be plausible that implicit qua predications, to which I now turn, work in just the same way as explicit ones. In particular, girqps – again, general implicit redoubled qua predications, that is, sentences in which Fs are claimed to be Gs qua Fs, but with the qua phrase left implicit – should say just the same thing as the counterpart sentences deploying explicit redoubled qua. Before turning to Plato's use of girqps, it is worth being clear that there is nothing proprietary to Plato about them (or to Greek); they are pervasive in our discourse as well as his. Consider the following quick sample pack, chosen for its diversity and independence of any proprietary Platonic assumptions:Humid days are unpleasant.Gossips are unpopular.Cyclists need carbs.A newspaper reader is an informed voter.A teacher is not a therapist.A gardener is a craftsperson, not a mere manual labourer.The President is the Commander‐in‐Chief.Each of these is chosen to bring out some salient features of girqps and their use. ‘Humid days are unpleasant’ makes the causal character of qua almost explicit: no one would say this in a tone of surprise, as if noting a mysterious statistical coincidence. Likewise, ‘Cyclists need carbs’, which would be wholly uninformative as a universal generalization: all human beings need carbs, so there is no point here other than the pro tanto and implicitly causal one. ‘Gossips are unpopular’ brings out the very real possibility of defeaters: think of Jane Austen's Emma or Cher in Clueless, popular for her high spirits and prettiness despite the trouble caused by her gossiping. With the newspaper reader, we can see what I take to be a different basis for truth–condition gaps, namely threshold variation for applying predicates once the qua‐ification is removed. That is, I might assert the girqp yet not be willing to call Homer Simpson, who admittedly browses the paper in a minimal way, an informed voter simpliciter. (Threshold variation might also complicate the case of Emma: we might apply ‘unpopular’ and ‘gossip’ to different percentiles for those traits.) The teacher–therapist case, like the Lysis arguments, exposes some of the problems and ambiguities of negations involving qua. Is it being asserted only that no teacher is as such qualified to play the role of therapist, or that the two roles actively exclude each other? Most likely the point is merely the former, weak one; so here too we see clearly that the girqp may easily be true while the corresponding universal generalization is false. The gardener sentence is more suggestive of ‘strong’ qua and perhaps relatedly (unlike the ‘therapist’ sentence) it registers as criterial: if many of the people we habitually refer to as gardeners really are just manual labourers, we are misapplying the term. The final sentence is also an instance of role specification qua: and it shows that the terms deployed in a girqp need not be gradable, nor are defeaters always a possibility, nor need the subject always be plural. (That it is nonetheless a girqp is still clear, I take it, from the fact that we could add in ‘as such’ or an ‘insofar as…’ phrase after the subject term with no change in meaning.) It should also help to make vivid how fundamentally different a girqp can be from superficially similar kinds of predication with apparently the same subject. If we juxtapose it with ‘The President is from Delaware’, the difference is intuitively clear, since we determine their truth by looking in different directions: in one case, directly at the object specified by the subject term, in the other, at what the Constitution says about the properties and powers which correlate with the Presidency for any occupant of that role.For all their diversity, these examples should enable us to see that there is a common core to girqps, calling for their recognition as a distinctive kind of sentence. Their key feature has already emerged: they assert a pro tanto correlation between two features and at least suggest that it is due to a causal relation. A qua predication does not just register a coincidence: it at least suggests that Fs are Gs (when and insofar as they are Gs) because they are F or by being F. I say cautiously ‘at least suggests’ because the sharpness with which this causal claim emerges seems to vary considerably case by case, and we might in some cases be tempted to view it as merely a matter of suggestion or implicature. But in other cases, perhaps the majority, the causal claim does seem to be fully asserted and to fix the truth conditions of the sentence. Where the causal claim is merely suggested, we can at least say that the correlation imputed by a girqp is non‐accidental, and prima facie asymmetrical. That is, to say that Fs < qua Fs > are Gs is not at all to say that Gs < qua Gs > are Fs17This is not itself one of the features that differentiates qua predication from quantified universal generalization, since it is also true that ‘Fs are Gs’ is a different fact from ‘Gs are Fs’. But this is a fact about the asymmetry of class inclusion, whereas the asymmetry of qua predication is due to the asymmetry of causal (or explanatory or grounding) relations.; if the claim in the other direction is true, it represents a different fact. (It may be true both that good people as such tend to be self‐sufficient and that self‐sufficient people as such tend to be good; but these are different claims and if they are true, it is for different reasons.) Only prima facie, however, because in asserting a qua predication, we may also mean to at least leave open the possibility that a full and correct theory of Fness and Gness would show them to have a common cause, or a reciprocal causal relation, or even to be ultimately reducible to the same thing. One of Plato's canonical uses of qua predications is to assert that good things as such are fine (Symp. 201c, Lysis 216d, Rep. 457b etc.); but the ultimate relations of the good and the fine in Platonism are a great metaphysical mystery.So understood, a girqp says both less and more than the corresponding universal generalization. More, because of its irreducibly causal character; less, because it does not entail that any particular F is G simpliciter. This is because girqps are not really a way of talking about particular F and G things at all, but a way of talking about Fness and Gness themselves while bracketing all questions about their instantiations. As we will see, they are important to Plato precisely in virtue of this detachment from particulars, which puts us in a position to think and speak clearly about the natures of Fness and Gness as such.18As Willie Costello has pointed out to me, this has an interesting resemblance to Dretske's account of the logical form of a law of nature (Dretske, 1977). As Dretske puts it, ‘To say that it is a law that F's are G is to say that “All Fs are G” is to be understood (insofar as it expresses a law), not as a statement about the extensions of the predicates “F” and “G”, but as a singular statement describing a relationship between the universal properties F‐ness and G‐ness.’ (252). A disanalogy is that Dretske does take a law to imply the corresponding universal generalization; but he emphasizes that the two are still different, in part because universal generalizations have no explanatory pretensions (the fact that every F is G fails to explain why any F is G' (262)). This convergence suggests that we might think of girqps as the broader, garden variety genus of assertion of which laws of nature are a special privileged subset, special not least in that with them the usual truth–condition gap is closed.GIRQPS IN PLATOOnce we learn to spot them, girqps turn out to be everywhere in Plato's work. I will here offer a brief field guide to them, grouped under three very general headings. (We have seen instances of all three already, albeit at exceptional moments where the qua‐ification surges into explicitness.) I do not mean to present these as sharply defined or formally different species: my point is just the rather superficial one that girqps are especially at home in three kinds of Platonic habitats, with a somewhat specialized look and function in each.19If anything, we might see the third kind as a special case of the second (since a technê is a causal power [dunamis]), and the second as a special case of the first (since a causal power can in turn be seen as a kind of property). My point is just that girqps belonging to these somewhat blurry categories tend to appear in different contexts and to be used with different argumentative purposes.One kind of use is what we might call property network qua. A great many theses and arguments about central evaluative concepts belong here, including the Protagoras argument about pleasure which we considered in Section 2. Likewise, Plato argues that the good qua good is fine (Republic 457b); that courage as such is fine and thus advantageous (Alcibiades I 115a‐16c); that the admirable is so in virtue of being either pleasant or beneficial, and conversely that the shameful is so in virtue of being either painful or harmful (Gorgias 474c‐77e); that the appropriate is fine (Hippias Major 290c‐91c); and that the harmed are wretched insofar as they are harmed (Meno 78a). All these are natural candidates for use as theses or as premises in eristic arguments: property network argument is the realm in which the sophistic roots of (attentiveness to) qua‐ification are most evident.20Property‐network arguments are easily formulated in Greek because any adjective can be substantivized by the definite article. So, in these arguments, Plato typically speaks of ta F (neuter plural with the definite article, for all F things), to F (neuter singular, for any F thing – or perhaps for ‘what is F’, with F as a kind of mass term), or, if the property is envisaged as applying particularly to humans, hoi F (masculine plural, ‘those who are F’).A second arena for qua predication puts their explanatory dimension front and centre. These are arguments about causal powers. This would be my preferred bucket to put the Lysis arguments in since, on my reading, these arguments are investigating the grounds or basis for friendship between two people. Plato's conception of causal powers [dunameis, to the extent he has a single preferred term] is a central, woefully underdiscussed, and ill understood topic, and I cannot properly enter into it here.21But see Benson (1997), Sedley (1998), Wolfsdorf (2005), and Lefebvre (2018), particularly Part Two on Plato (183–346). But I take it that Platonic powers are real dispositional properties and that his preferred mode of causal explanation consists in identifying them. It might be too strong to say that Plato is uninterested in necessary and sufficient conditions, but they are at any rate not what he thinks a cause is; a real causal explanation picks out the cause par excellence. Depending on the kind of explanandum in view that will be either an agent (an aitios, literally the guilty party) or a power – roughly, a dispositional property the manifestation of which reliably brings about the effect in question.22All this is of course oversimplified and calls for far more discussion; cf. Sedley (1998).A particularly important kind of causal power for Plato is virtue. The presence of a virtue in an agent – courage, say, or justice – reliably manifests in virtuous actions. So we can test prospective definitions of a virtue by swapping in the proposed candidate, understood as a causal power, and asking whether it invariably brings about behaviour we judge to be virtuous in a relevant way. For instance, in the Laches, Plato considers whether courage could be correctly defined as endurance. Answer: no. For enduring foolishly is not admirable, which courageous behaviour always is (192d). Socrates then turns to the other possibility, wise endurance. (So, the project here is again superficially one of refutation by exhaustion of cases, as in the Lysis arguments.) He considers various kinds of behaviour that we would count as enduring wisely and notes that we would not count them as courageous, and that indeed the person who endures wisely in certain circumstances would be judged less courageous than the person who endures foolishly (193b‐d). So, of the two species of endurance, wise and foolish, neither is reliably the cause of courageous behaviour, and therefore neither is identical to the power of courage. Plato's inference is that endurance was the wrong general power to start with: wisdom taken by itself is a better candidate and so is taken up as the next definition to be considered (194cff.).23Another intriguing but somewhat non‐standard case of explicit qua plausibly belongs here. In Republic IV, Socrates uses the example of thirst to show that generic states or powers such as desire and knowledge are correlative with equally generic or unqualified objects. For thirst, insofar as it is just thirst [kath' hoson, 437d8, 9a9] is not for any particular kind of drink, even good drink, but simply for drink (Republic 437d‐8a). So, the statement ‘Thirst is for drink ‐‐ drink simpliciter’, understood as a girqp, is true. Note the extreme version of the truth–condition gap: the corresponding universal generalization will plausibly be false in every case, since when thirst occurs together with heat it will be for cold drink, when together with coldness it will be for hot drink, and so forth. Here, the use of redoubled qua is evidently to strip away these adventitious and contingent factors so as to leave only what belongs to thirst as such [kath' hauto]. This ‘stripping‐away’ use of qua comes close to Aristotle's principal uses of it (cf. n. 39 below) and to ‘partitioning qua’ (cf. n. 8 above).A third characteristic use of Platonic qua is the one we have already seen in Republic I: role specification qua, used to determine the actions and properties correlated with agents' roles and identities. A particular focus here are the actions and properties correlated with knowledge and with the crafts [technai]. Plato, like his character Socrates, is obsessed with technê, craft or expertise, as a kind of general grounding for the normativity of practical reason and a model for moral virtues.24I discuss the technê model with particular reference to the ethical agenda of Republic I in Barney (2020). But craft terms are just a subspecies of a very broad class of locutions. We get the same kind of argumentation when, for instance, Plato tries to fix the properties of ‘the lover’, ho erastês (i.e. the one who loves, insofar as he loves) in the Symposium and Phaedrus. Likewise in other works when Plato investigates the properties of ‘the democrat’, ‘the poet’ or even ‘the tyrant’ or ‘the sophist’. (Often for role specification qua the preferred construction is the singular with the article, suggesting the inspection of a prototype or paradigm case, and thus perhaps keeping at arm's length the plurality of questionable instantiations around us.) Nor does this kind of qua‐ification apply only to human agents. The same goes for the city strictly speaking, the tool as such, and indeed the word ‘name’ [onoma] itself: for a name, according to the Cratylus, is a tool for dividing being and revealing the natures of things (387a‐9a). The background assumption in all such cases is that the role in question is in at least a weak sense teleological or functional, in the sense that it is part of the ‘job’ or office or identity of Fs to be Gs – or, since predications of actions are particularly common in this sphere, part of the job of Fs to reliably φ.25So, my use of ‘teleological’ and ‘functional’ here should not be thought to import any presumption of normativity into qua‐ification, except insofar as doing what Fs qua Fs do is a condition for something's being a good instance of Fness. In the Republic, Socrates first illustrates the idea of an ergon with the claim that it is the ergon of the hot not to cool but to heat things (335d3): in other words, its natural, characteristic operation and effect. Plato seems to hold that much of ordinary language – far more than you might suspect – is implicitly teleological or functionalist in this way, and that philosophical inquiry can use role specification qua to articulate these background conceptions. Role specification qua is thus at home wherever the salient feature of some nature is its characteristic work or function, and like other forms of Platonic qua it can cause confusion when it goes unrecognized.26Consider, for instance, Republic X, where Plato offers a complex series of arguments intended to articulate just what the imitative artist, the mimêtês, does. Mimêtês is not for Plato a positive role or function; on the contrary, his point is to argue that such people must be excluded from a just city (595a‐608b). But to do so requires figuring out exactly what mimetic art is. His opening argument is that as the creator of an imitation, the mimetic artist does not have any knowledge about things in themselves (595a‐8d). This argument has long been found puzzling and unfair: how can Plato be so sure that no artist ever has had knowledge of what he imitates? But this objection is only in order if we mistakenly suppose the argument to be working with universal generalizations. Instead, we should understand it in terms of role specification qua. The artist qua artist knows nothing about realities: that is, such knowledge is not required for his function of imitative creation. This indeed does nothing to exclude the possibility that some individual artists might have knowledge on some other basis. The argument does not purport to issue directly in a universal generalization that would ground the exclusion of all artists from the city; that is why it is followed by a further, empirical argument that, as a matter of fact, even the greatest imitative artists seem not to have had knowledge about the things they represent (599a‐600e).Role specification is where we can most easily see the ethical stakes of Platonic qua, and how it indeed serves as a vehicle for social critique. Properly speaking, the doctor qua doctor is someone who seeks the health of the patient and knows how to provide it by medical techniques. So agents who seek something other than the health of the patient – the maximization of their billings, say – are to that extent not doctors at all, no matter how much they might know about bloodletting and cautery. This line of argument has revisionist implications relative to the current conventional denotation of the term. Many ‘doctors’ will turn out to be frauds. But, Plato thinks, such arguments leave intact – indeed, they depend upon – the sense we standardly give to ‘doctor’. (Only Socrates' teleological analysis of medicine can explain, for instance, why ‘good doctor’ is used, by all of us, for the one who treats patients correctly – and not, say, for whoever excels at using medical knowledge for his own selfish unmedical ends.) What reflection on qua predication brings out is that our everyday generalizations (the kind which forms the corresponding universal generalizations for girqps, and must be distinguished from them) are subject to a kind of linguistic entropy or degeneration: a truth–condition gap arises in relation to role specification qua because the reference of F ceases to be guided by its sense, and is instead fossilized by social convention and unreflective habit. This is the phenomenon I call referential disorder.27I discuss this in Barney (2001) in terms of ‘the project of the strict sense’ (10–14). When our language is in a state of referential disorder, we use words like ‘ruler’ or ‘doctor’ with a sense given by their teleological or functional role; but we apply them, on the basis of habit and social convention, to people who (as we may or may not be aware) do not in fact perform it. If the disorder is deeply enough embedded in our linguistic practice, so that we predominantly use ‘F’ for objects which lack Fness correctly understood, the phenomenon repeats at the meta‐level: the name ‘F’ will not really do what a name is supposed to, that is, divide the F objects from other things, and so will not really deserve the name ‘name’. (This is, I believe, a major worry being explored in Plato's treatment of naming in the Cratylus.) Social roles such as crafts, offices, and practical identities are the natural habitat for referential disorder, for we need to use these terms in ways that reflect realities on the ground: we can hardly get by without using ‘doctor’, ‘ruler’, ‘lover’, ‘hero’, ‘friend’, etc. for the socially accepted occupants of those roles. But at the same time, such roles typically do have functional specifications built in, often of a moral or epistemic kind; these are easily enough revealed under dialectical scrutiny, and so too is the fact that their socially recognized occupants – the Fs ‘loosely speaking’ – may or may not meet them. As Thrasymachus allows, everyone will say that the ruler errs, including himself; but that is because the conventional denotation of ‘ruler’ is so far from being the correct one that for practical purposes we have to allow for loose talk as well as the precise kind. The best we can hope for is to be aware of which we are employing at any given time.Referential disorder is one of the main reasons redoubled qua, and girqps in particular, are so important to Plato. For a redoubled qua predication invites us, so to speak, to consider what really is F, and only insofar as it is F. The referential openness of this kind of claim, and the invitation to think of Fness in graded terms, mean that it immediately invites reflection and inquiry as to what should really count as F and how far, in a way in which universal generalizations do not.28Cf. the ‘principle of real reference’ in Penner (1991) and Penner and Rowe (2005), and the ‘dominance view’ of desire in their (1994); similarly, Reshotko (1992) and later works. I would not agree that Plato takes intentional psychological states to standardly have this ‘whatever really is F’ form, so as to float free of the agent's own beliefs about what satisfies the descriptions he uses. But qua‐ified attitudes do have this structure, and Penner can be read as working with the insight that far more of Plato (and of everyday thought and speech) is qua‐ified than has usually been recognised. However, Penner himself rejects qua‐ified analysis of the key attitudes in Penner (1991) at 197–8. Qua‐ification thus serves as a very immediate gateway to philosophical inquiry. It expresses an aspiration to think and speak clearly and correctly about how functions, natures, and other properties are related to each other – even if they are only instantiated very defectively by the objects around us, and no matter how misleading our loose talk about them may be.PLATONIC PHILOSOPHY AND THE METAPHYSICS OF QUASo qua predication is at home in the analysis of property networks; in accounts of causal powers; and in arguments about crafts, social roles, and functional categories. That covers quite a lot. Indeed, at this point, it may start to look like pretty nearly any Platonic argument will, by my standards, be a good candidate for reading in girqp terms. And so much the better. For, according to Plato, philosophical dialectic is indeed not about particulars, classes or sets of particulars, but about how various kinds or natures are related. This is made particularly explicit in a passage that deserves more serious attention than it gets as a guiderail for Platonic interpretation, the so‐called ‘Philosophical Digression’ of the Theaetetus (172c‐77c). Here, Socrates draws an elaborate contrast between the life of the philosopher and that of the careerist lawyer. The philosopher is aloof from social hubbub and practical concerns. Uninterested in particular humans like his neighbour, the question he asks is, ‘What is Man? What actions and passions properly belong to human nature and distinguish it from all other beings? This is what he wants to know and concerns himself to investigate’ (174a8‐b6). Likewise on ethical topics, in contrast to the lawyer, he is interested not in who did injustice to whom or whether the king is happy, but in general inquiry into kingship, and into human happiness and misery in general (175b9‐c6).In short, philosophical discourse – strict talk, the kind we should aspire to – is concerned with general truths about the natures of things. The business of philosophy is the production of true, explanatory, defensible, systematically organized qua predications about those natures: ‘precise accounts’, as Thrasymachus would say (Rep. 340e). The particular objects around us, on the other hand, are for Plato literally, not what philosophy is about. Here as elsewhere, seeing that Plato is thinking in qua terms can make a puzzling argument much less so. Interpreters have felt some anxiety about the way in which the digression portrait of the philosopher seems not to describe Socrates himself, and have wondered if some sort of irony or self‐undermining must be in play. But Socrates is just doing what he says philosophers are supposed to do: speaking not about individuals (such as himself) but about philosophers as a kind. And what is distinctive of the kind, on his view, is disinterested inquiry into the general natures of things. Since Socrates himself engages in disinterested inquiry into the nature of virtue, he should indeed count as a philosopher so understood; but so do astronomers, who might seem very different if we do not inquire into the features of the general kind.It might be responded that this much is unsurprising: for what philosophy is about, according to Plato, is Forms, that is, general kinds held to be firmly distinct from and independent of sensible particulars. This brings us to a question I have so far avoided: what relation does qua predication have to the theory of Forms? I am not entirely sure what to say about this, but we can start with a negative: Platonic qua does not depend on the theory of Forms. Qua predication is after all part of everyday linguistic practice: we all can express and understand girqps about all sorts of topics, gossips and cyclists included, with no metaphysics implied or required. So I would suggest that the phenomenon of qua predication comes first, so to speak; Plato's own philosophical uses of it, as charted in Sections 1, 2 and 4, come second; and the theory of Forms comes in third place as, among other things, Plato's reflective attempt to articulate the presuppositions of both ordinary usage and his own practice. For at least from the Phaedo onwards, one of the jobs of Forms is to serve as the primary referents for the terms that pick them out; our words refer only secondarily and indirectly, by ‘eponymy’, to the many particulars which participate in the Form (103b). The Cratylus account of naming as a matter of natural correctness thus analyses names ‐‐ true names, the kind deserving of the name ‘name’ ‐‐ as attempts to express the natures of things. (The Cratylus is an enigmatic dialogue, however, and whether Plato in the end accepts this account is an open question.) This explains how qua predication can be, for Plato, a kind of default: when I use a general term F, I refer first and foremost to the nature which it picks out, Fness, and through its mediation to whatever objects participate in that nature, insofar as and because they do so. This linguistic order mirrors the causal one: it is because beautiful things participate in the Form, Beauty itself, that they are beautiful (Phaedo 100b‐e, 101c). This does not yet entail, however, that there is a Form for every kind of thing to which we can successfully refer, or which can be the subject of qua predication. It might be that we need to postulate only some metaphysically privileged subset of things as having Forms, while others, as Socrates somewhat mysteriously says in the Parmenides, are ‘just as we see’ (130d). The question is difficult to investigate (and I will evade it here) because the scope of the Forms – that is, the question what things there are Forms of – is notoriously unclear, and seems to be rethought from dialogue to dialogue.29For an account of Forms which makes some use of qua predication, see Ademollo (2013).A broader and more useful framework for thinking about the metaphysics of qua is a notion I have already, almost unavoidably, been using: nature [phusis], as in the phrase ‘the natures Fness and Gness’, and more generally the natures [phuseis] of things. According to the theory of the Cratylus, names (i.e. descriptive terms in general) are natural or in accordance with nature when they pick out beings, essences or kinds. Phusis sometimes serves as equivalent to ‘Form’, but it often seems to be much broader in scope; and when it is used in relation to Forms, it tends to refer not exactly to the Form itself but to what the Form and its instances have in common – for example, the kind which some craftsman who looks to a Form grasps and puts into the product he creates (Crat. 389b‐c).30Cf., for example, Phaedr. 254b, Rep. 476b, Crat. 389b, Rep. 501b, Phaed. 103b, Rep. 597b4, c2 (for en phusei, ‘in nature’, as apparently a reference to the realm of Forms; cf. Adam's (1902) notes ad loc.), Parm. 132d. Social and functional roles, causal powers, and properties related to others in qua‐ified networks are all good candidates for phuseis. And Forms, whatever (else) they are, should perhaps be conceived as the privileged subset of natures which we must postulate as somehow metaphysically primary, in order to explain how things, in general, are – including all the other things for which there is no Form.In sum, I suggest that we see qua predication as being prior to metaphysics in the trajectory of Plato's thinking, and the metaphysics of natures as in turn prior to and broader than the metaphysics of Forms. And certainly, no metaphysics is required to see why qua predications would be important and pervasive in Plato's discourse – or our own. As assertions of causally asymmetrical correlations between kinds, they are naturally central to explanatory talk; and a great deal of our talk aspires to be explanatory. The point is easiest to see when put in terms of pragmatic considerations. With the possible exception of the Parmenidean One, every object of interest to us bears indefinitely many descriptions; whenever we set out to predicate a property of a subject, we have to choose among indefinitely many possibilities for picking that subject out. One obvious way to make that choice is on the basis of relevance, and one obvious kind of relevance is a causal relation to the possession of the property in question. Presumptions of relevance thus create a standing pressure towards speaking and interpreting speech in qua terms. This can explain why, for instance, it is deprecated for a newspaper headline to identify a criminal by a racial, ethnic or religious descriptor. (Here, we start to converge with recent discussions both of generics and of relevance in communication.)31On relevance theory, the seminal work is Sperber and Wilson (1986); on generics see Leslie (2017), and cf. also the uses of ‘Aristotelian categoricals’ in Thompson (2008). If my account of Plato is right, and if Plato is right about how we all think and talk, the real ‘original sin’ behind generics is an uncritical, system A impulse to interpret each other's assertions as explanatory. If a headline says ‘Muslim arrested in robbery’, presumptions of relevance suggest that the suspect's religion, rather than say his poverty or opioid addiction, is being singled out as causally relevant to his crime: Platonic qua is sufficiently our own default mode that, all else being equal, we hear an implicit ‘qua Muslim’. Likewise with a general claim like ‘F's have a high rate of street crime’ for some ethnic descriptor ‘F’. We hear this by default not as ‘F's qua‐who‐knows‐what‐independent‐variable’ or ‘The set of Fs, for no particular reason’, but as a claim about Fs qua Fs – perhaps even, Platonically, about the causal power of their Fness.32What about the kind of annoyance people sometimes feel at headlines like ‘Mother of two wins Nobel Prize’? Here, the descriptor has been chosen precisely as non‐relevant and surprising, hence presumptively intriguing to the reader. (Relevance is obviously not the only value we select for; or perhaps we should say that causal salience is not the only form of relevance.) The annoyance is at this decision not to focus on the scientist qua scientist, which seems undermining of that identity.Plato seems to be aware of this and indeed to operate with a particularly strong version of it. In a remarkable moment of the Phaedo, Socrates distinguishes between attributing a property to a sensible particular directly and via a Form. Only the latter, apparently, ever properly captures the way the world is:But, he said, do you agree that, as for “Simmias is taller than Socrates,” that way of putting it is not how the truth actually holds [oukh hôs tois rhêmasi legetai houtô kai to alêthes echein]?33Taking to alêthês as the subject of echein: ‘how the truth holds’, that is, how things really are, cf. Crat. 384c1, Phaedr. 278c5, Gorg. 484c4. I am indebted for discussion of this passage to Gail Fine and Terry Irwin. It is not, surely, that Simmias has come to be [pephukenai] taller than Socrates for this reason, that he is Simmias [tôi Simmian einai],34Cf. ‘by being bees’, at Phaedo 72b. but rather because of the tallness he happens to have? Nor is he taller than Socrates because [hoti] Socrates is Socrates, but because Socrates has smallness compared with the tallness of the other?‐‐ True. (102b8‐c4)35On this passage cf. Dancy (2004), 306ff with references.Note that the failure here is not a failure of the sentences in question to be true in any standard modern sense: Simmias is in fact taller than Socrates. The objection is that saying so still fails in some way to capture the way the world is. It is a kind of ‘loose talk’. Apparently, when I use ‘F’ to name some subject in order to predicate Gness of it, it fully captures the truth only if Gness belongs to that object by virtue of its nature as F. In order to capture the truth in full, without being misleading, our choice of terms must perspicuously track salient causal or explanatory relations, and these have to do with the natures of things. And although it is not the point directly being made here, this does have implications for truth in a more familiar sense: for if we presume that our interlocutors are attempting to conform to this demand, our default will be to hear an implicit ‘qua Simmias’ after the subject term, which does render false the sentence, ‘Simmias is taller than Socrates’.36Cf. ‘some pleasures are bad’, as voiced by the Many in the Protagoras (as discussed in Section I above), to which Socrates also objects (misleading since it is not qua pleasures that they are bad), and ‘the doctor errs’, which Thrasymachus deprecates (since it is not qua doctor that he does so). That ‘Simmias’ is a proper name does not exempt the sentence from this expectation, but points to why it cannot be met. For the names of individuals do not correspond to natures in the right way: it is not at all clear what ‘Simmias‐qua‐Simmias’ or ‘Simmias insofar as he is Simmias’ would pick out, and what if anything could be predicated of him (‘A human being’ picks out his nature; but this is not a very promising answer if we are interested in the differences between him and Socrates). Simmias seems to be little more than a kind of nexus, a locus for the various prior and more general natures – such as Tallness – which compete and interact to determine his properties. That is one of the reasons philosophical dialectic is about things like them, and only ever indirectly and unsatisfactorily about things like him.For Plato, it is the central job of the philosopher to sort out patterns and relationships among kinds or natures, while keeping a critical eye on their putative instantiations around us. That is what it is to ‘carve nature at the joints’ (Phaedrus 265d‐6b, cf. Statesman 262c‐3a). And it is by performing this dialectical work that we gain the resources necessary to see where everyday conventional reference has come to be disordered, and to rectify it in light of a grasp of the true natures of things. It is, I think, fair to refer to this as a project of analysis: an analysis of Fnesses and Gnesses as natures, which means the determination of their true causal powers and relations. For Plato, then, philosophy is inherently analytical; and philosophical analysis is inherently a critical, even a political, enterprise.CONCLUSIONSI close with two reflections that gesture towards further questions worth exploring. First, so far as I can see, Plato is the first and last philosopher to have taken qua predication to be our default mode of expression: both as the correct vehicle for strict, properly philosophical discourse and as pervasively embedded and implicit in ordinary talk. If he is right, then many sentences that might look to us like universal generalizations are rather to be understood as girqps; and this will include many of those now classified as generic generalizations. And if a large and important class of generics are best understood as girqps, then it turns out that they are not really generalizations at all.37The test for whether a generic does express a girqp is straightforward: can we add ‘qua Fs’ without change to its truth conditions? They do not claim that all Fs are Gs – not even that most of them are, or the most characteristic or most salient ones. Rather they are irreducibly explanatory, and deal in natures or causal powers rather than sets of individuals: what they bring out is that the Gness of those Fs which are Gs is due to their Fness. This diagnosis fits with some of the most striking features of generics: the truth–condition gap, our tendency to derive them from very small samples and their default status and apparent cognitive priority to universal claims.38For a useful overview, see Lerner and Leslie (2016). That causal considerations have some kind of bearing on generics is not a new thought: but the Platonic proposal that it is the defining feature of a large class of them seems to me to be both radical and worth consideration.Second (and although I cannot here argue properly for the ‘first and last’ claim above), qua predication in Aristotle and later Western philosophy seems to me very different. Clearly qua predication is important to Aristotle – he provides qua with its most memorable star turn in philosophy, after all, in designating metaphysics ‘the study of being qua being’. Aristotle also offers much more than Plato in the way of explicit analysis of qua predication, in several different contexts.39A survey of qua in Aristotle would have to consider: (1) its role in his diagnosis of sophistic fallacy (cf. Sophistici Elenchi 168a34‐b10, and see Section I above on the original philosophical uses of qua), and the related puzzles generated by accidental unities (cf. Cohen, 2013); (2) the reduplicative syllogism (Prior Analytics I.38, cf. Bäck, 1996); (3) the role of qua‐ification in demonstrative science, which studies some class of substances only qua Fs (Posterior Analytics I.6 and passim; cf. Malink, 2009), so that, for instance, mathematics deals with objects which are sensible but not qua sensible (Metaphysics XIII.3, 1077b17‐8a4, cf. 1078a1‐4, 23–6), while metaphysics deals with beings qua beings; and (4) miscellaneous deployments of distinction‐drawing qua scattered across the corpus. Particularly important instances of this last are his claims (i) that when we see Callias, we perceive him not qua Callias but qua human being (An. Post. II.19, 100a17‐b3); and (ii) that a failure to distinguish qua‐ifications in relation to not‐being is what led the early philosophers into confusions about ‘what is not’ (Physics I.8, 191b1‐9). But this amounts to a frustrating fragmentation of the phenomenon, and whether qua talk remains as central to his philosophical discourse as it is to Plato's is controversial.40See Malink (2009), Malink (2013) on the ‘orthodox’ and ‘heterodox’ readings; for non‐extensionalist Aristotelian semantics cf. also Castagnoli (2016). Granted, Aristotle's own understanding of scientific discourse can be construed in terms of Platonic qua; for scientific demonstration [apodeixis], as Aristotle understands it, is precisely an articulation of the properties which belong to Fs qua Fs [kath' hauta, per se]. But whether Aristotle himself understands such discourse in terms of Platonic qua is another question: on a traditional reading, he is instead committed to conceiving of demonstration in terms of universal generalizations – despite the notorious difficulty in reconciling this with his admission that natural science deals in claims which are true only ‘for the most part’.41Cf. Mann (2000): on Plato see especially 77ff. And whether the transformation is rightly attributed to Aristotle himself or not, universal generalizations about more or less Aristotelian substances have since then come to be seen as the default mode of scientific discourse. This transformation has helped to make qua predication hard for us to see, even as it continues to be deeply embedded in our own everyday speech and thought. And so, it seems to me worth trying to recover and recognize this more ancient, and perhaps more basic, way of talking and thinking.42In addition to the Annual Analytic Philosophy Symposium at the University of Texas in 2018, this paper owes a great deal to discussion with the Stanford Metaphysics and Epistemology Conference; the Oxymorons group; the Institute of Classical Studies in London; the B Club at Cambridge University; the Philosophy Department at the University of Pittsburgh; the University of Oklahoma conference honouring Hugh Benson; the Notre Dame University Ancient Philosophy Workshop; and Rachana Kamtekar's seminar at Cornell. I have been particularly helped by comments and questions from Francesco Ademollo, Andreas Anagnostopoulos, Merrick Anderson, Jonathan Beere, Hugh Benson, Tad Brennan, Nick Denyer, David Ebrey, Gail Fine, Benj Hellie, Terence Irwin, Fiona Leigh, Rachana Kamtekar, Sean Kelsey, Martin Lin, Ben Morison, Gurpreet Rattan, Gretchen Reydam‐Schils, Kenneth Sayre, David Sosa, Jan Szaif, Michael Thompson, Jennifer Whiting, Jessica Wilson, and Raphael Woolf. I am deeply grateful to James Allen, Willie Costello, Nick Denyer, Mark Gatten, Rory Harder, Marko Malink, Wolfgang Mann, Stephen Menn, Jennifer Nagel, Michail Peramatzis, David Sedley, Thomas Slabon and Rob Stainton, who generously gave comments on the paper as a whole which have saved me from a multitude of blunders. I am sure that many remain; I am also sure that I have forgotten the names of others from whom I have learned important things, for which I apologize.REFERENCESAdam, J. (1902). The republic of Plato, 2 volumes. Cambridge University Press.Ademollo, F. (2013). Plato's conception of the forms: Some remarks. In R. Chiaradonna & G. Galluzzo (Eds.), Universals in ancient philosophy (pp. 41–86). Edizione della Normale.Bäck, A. (1982). Syllogisms with reduplication in Aristotle. Notre Dame Journal of Formal Logic, 23(4), 453–458.Bäck, A. (1996). On reduplication: Logical theories of qualification. Brill.Bäck, A. (2002). The role of qualification. Journal of Philosophical Research, 27, 159–171.Barney, R. (2001). Names and nature in Plato's Cratylus. Routledge.Barney, R. (2006). Socrates' refutation of Thrasymachus. In G. Santas (Ed.), The Blackwell guide to Plato's republic (pp. 44–62). Blackwell.Barney, R. (2020). Technê as a model for virtue in Plato. In T. Johansen (Ed.), Productive knowledge in ancient philosophy (pp. 62–85). Cambridge University Press.Benson, H. (1997). Socratic dynamic theory: A sketch. Apeiron, 30, 79–93.Brunschwig, J. (1984/5). Aristotle on arguments without winners or losers. In Jahrbuch Des Wissenschaftskolleg Zu Berlin (pp. 31–40). Siedler Verlag. https://www.wiko‐berlin.de/wikothek/jahrbuchCastagnoli, L. (2016). Aristotle on the non‐cause fallacy. History and Philosophy of Logic, 37(1), 9–32.Castelnérac, B., & Marion, M. (2009). Arguing for inconsistency: Dialectical games in the academy. In G. Primiero et & S. Rahman (Eds.), Acts of knowledge: History, philosophy and logic: Essays dedicated to Göran Sundholm (pp. 37–76). College Publications.Castelnérac, B., & Marion, M. (2013). Antilogic. The Baltic International Yearbook of, Antilogic Cognition, Logic and Communication, 8, 1–31. https://doi.org/10.4148/1944‐3676.1079Cohen, S. M. (2013). Accidental beings in Aristotle's ontology. In G. Anagnostopoulos & F. Miller (Eds.), Reason and analysis in ancient Greek philosophy (pp. 231–242). Springer.Cooper, J. M. (1997). Plato: Complete works. Hackett.Dancy, R. M. (2004). Plato's introduction of forms. Cambridge University Press.Dretske, F. (1977). Laws of nature. Philosophy of Science, 44, 248–268.El Murr, D. (2020). Eristic, antilogy, and the equal disposition of men and women (Plato, resp. 5.453B‐454C). The Classical Quarterly, 70(1), 85–100.Fine, K. (1982). Acts, events, and things. In W. Leinfellner, E. Kraemer, & J. Schank (Eds.), Sprache und Ontologie (pp. 97–105). Holder‐Pichler‐Tempsky.Lear, J. (1982). Aristotle's philosophy of mathematics. Philosophical Review, 91(2), 161–192.Lefebvre, D. (2018). Dynamis: Sens et genèse de la notion aristotélicienne de la puissance. J. Vrin.Lerner, A., & Leslie, S. J. (2016). Generics and experimental philosophy. In W. Buckwalter & J. Sytsma (Eds.), The Blackwell companion to experimental philosophy (pp. 404–417). Blackwell.Leslie, S. J. (2017). The original sin of cognition: Fear, prejudice, and generalization. Journal of Philosophy, 114, 1–29.Malink, M. (2009). A non‐extensional notion of conversion in the organon. Oxford Studies in Ancient Philosophy, 37, 105–141.Malink, M. (2013). Aristotle's modal syllogistic. Harvard University Press.Mann, W.‐R. (2000). Aristotle and the discovery of things. Princeton University Press.Moraux, P. (1968). La joute dialectique d'apres le huitieme livre des Topiques. In G. E. L. Owen (Ed.), Aristotle on dialectic (pp. 277–311). Oxford University Press.Nehamas, A. (1990). Antilogic, sophistic, dialectic: Plato's demarcation of philosophy from sophistry. History of Philosophy Quarterly, 7(1), 3–16.Penner, T. (1991). Desire and power in Socrates: The argument of “Gorgias” 466A‐468E that orators and tyrants have No power in the City. Apeiron, 24(3), 147–202.Penner, T., & Rowe, C. (1994). Desire for the good: Is the Meno inconsistent with the Gorgias? Phronesis, 39(1), 1–25.Penner, T., & Rowe, C. (2005). Plato's Lysis. Cambridge University Press.Reshotko, N. (1992). The Socratic theory of motivation. Apeiron, 25(3), 145–169.Ryle, G. (1966). Plato's progress. Cambridge University Press.Sedley, D. (1998). Platonic causes. Phronesis, 43, 114–132.Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Harvard University Press.Szabó, Z. G. (2003). On qualification. Philosophical Perspectives, 17, 385–414.Taylor, A. E. (1960). Plato (7th ed.. (1st ed 1926)). Methuen.Thompson, M. (2008). Life and action: Elementary structures of practice and practical thought. Harvard University Press.Wolfsdorf, D. (2005). Δυναμις in Laches. Phoenix, 59, 324–347.

Journal

Analytic Philosophy – Wiley

Published: Jun 8, 2023

There are no references for this article.