The Collapse of Language? Theaetetus 179c-183c

DAVID SEDLEY

University of Cambridge, Christ's College.

UK

Having completed his critique of Protagoras’ general measure doctrine, Socrates returns to the version of it specially developed in order to flesh out Theaetetus’ definition ‘Knowledge is perception’. Universal flux was introduced in the first place, back at 152c-160e, as the necessary condition of perception’s infallibility. Appearances which are the fleeting and private products of instantaneous encounters between ever-changing subjects and objects cannot even in principle conflict with each other: no perceptual appearance, therefore, could ever falsify another, and knowledge can indeed be perception. It is now time for Socrates to ask whether this theory can survive scrutiny.

For despite the now completed refutation of Protagoras’ measure doctrine, the original perceptual theory does for now remain intact, as Socrates admits (179c-d). Moreover, what remains intact is not merely the weakened thesis that every perception is true and a case of knowledge (179c2-7), but also the stronger thesis which, as in Theaetetus’ original definition, fully identifies knowledge with perception (179c7-d1). One might have thought that the stronger thesis, according to which there are no cases of knowledge over and above perception, was no longer tenable, now that Protagoras has been forced to admit that there is expertise about the future which can hardly be thought perceptual. But actually the entire critique of Protagoreanism has limited itself to extracting from him the admission that there are both true and false opinions, including those about the future. Nothing explicit about the existence of expert knowledge has been added.¹ That restraint was methodologically prudent, since defining knowledge is the whole purpose of the dialogue, as yet unfulfilled; but it also serves to leave Theaetetus’ original definition undamaged, and ready for the refutation which it only now faces.

Socrates’ new line of attack, at 179c1-183c7, is to probe the character of the Heraclitean flux posited as underwriting the infallibility of perception. Theodorus remarks that in Ionia (the region which includes Heraclitus’ native city of Ephesus) there are swarms of Heracliteans, and that they behave in a way consistent with their belief in total instability, refusing to pause the discussion at any point, firing off enigmatic locutions all the time, and responding to requests for clarification with more such obscurities and evasions. The portrayal is light-hearted, but we should not for that reason dismiss it as mere fiction. Plato’s own earliest philosophical influence was said to be an Athenian named Cratylus, an extreme Heraclitean who according to Aristotle actually ended up abandoning language altogether, believing things to be changing too fast to be captured by speech.² And we will see below good reason to suspect that actual Heracliteans like Cratylus really did make a practice of radically reforming language to adapt it to their view of reality.

After undertaking to consider the opposite camp as well – those like Parmenides and Melissus who believe in entirely stable being – Socrates launches his scrutiny of the flux theory (183c1-183c7). The question is, just how radical is the flux to which this party is committed? It is very important to bear in mind the purpose of the inquiry. It is sometimes implied in the discussions of this passage that its aim is simply to ensure that the Heracliteans’ flux should be as radical as it can possibly be, and that the upshot will be that when that is achieved an unacceptable consequence follows, the collapse of language, with the result that extreme or total flux is refuted. That, if so, would be a strategy of little utility or relevance to Socrates’ main argument. What he has to do rather, and what he surely does do, is to find out how radical the flux needs to be in order to preserve the definition of knowledge as perception.³ The upshot will be that in order to be radical enough to ensure that result, the flux will also have to be too radical for the definition to admit of being stated.

We need to start by recapitulating and amplifying the perceptual theory, based on 155e3-160e5. Every perception is an interaction between a subject and an object. The subject may be thought of either as the perceiver, or more specifically as the relevant sense organ. In the case of seeing a stone as white, three aspects of the process are described (156d3-e7), although we need not think of them as temporally distinct. (1) The eye and the stone come together to generate both whiteness and the matching (twinned) vision of whiteness. (2) In between the two of them, vision is travelling rapidly from the eye, whiteness from the stone. (3) The eye has been filled with vision and become a seeing eye, while the stone has been filled around with whiteness and become white. The eye and the stone are ‘parents’, and undergo slow change without altering their place. The vision and the whiteness which it perfectly represents are their inseparable twin offspring, and undergo rapid motion involving change of place – a feature no doubt primarily designed to account for the high (if non-uniform) speed at which the distance senses operate, although it is taken to occur even in the contact senses (cf. 159e1-5 on taste).

The parents are themselves bundles of ever-shifting relativities. Specifically, the stone is a bundle of whiteness, hardness, and whatever other properties the subject is now perceiving, while the perceiver is the bundle of perceptions (visual, tactile etc.) now being experienced. These constituents being themselves mere fleeting relativities, the parents consisting of them are inferred to have no endurance through time either, so that neither the perceiving subject nor the stone endures as the same perceptual parent: even this ‘slow change’ is enough to ensure their constant replacement by new perceivers and new objects. From moment to moment the world moves on. Each instantaneous perception incorrigibly reveals its immediate object or ‘twin’, but can never be prolonged or repeated, and therefore can never be brought into conflict with another perception (157e1-160c6).

Thus the perceptual theory has clearly introduced two kinds of change: the locomotive change of the twins themselves, and the qualitative change undergone by the parents. In embarking on his criticism at 181c-d, Socrates starts by making this explicit, calling the two kinds of change ‘motion’ (phora) and ‘alteration’ (alloiôsis). The question is put (181d8-e2): will the Heracliteans say that everything is undergoing both these kinds of change? The answer is affirmative: otherwise they will be introducing as much stability as change. This may at first sight look like a doxographical question about what Heracliteans happen to believe, but, although it is not made explicit we must assume the main point of the question, and of the answer, to be to ensure that the flux is sufficient to guarantee the incorrigibility of perception which the flux doctrine is meant to be underwriting. And that means specifying more change than has previously been made explicit.

We already knew that the twins, which combine to constitute both perceiver and object, are in rapid locomotion between the two, and that the perceiver and object thus constituted are themselves undergoing constant, albeit slow, qualitative change (as we are carefully now reminded at 181c9-d3). But are the twins also undergoing constant qualitative change, and are the parents in constant locomotion? In both cases, they must be. Nothing is specifically said about the locomotion of the parents, but it is not hard to work out that they must be constantly undergoing some degree of locomotion in relation to each other in order to ensure the total instability of the relativities in which their constitutive properties consist. (It is here important to notice that locomotion, as defined at 181c6-7, includes not just translocation but also turning on the spot, which will allow observers to change their perspective without necessarily always shifting location.) What is made rather more explicit – because it is the case on which Socrates’ refutation will hinge – is that the twins must be undergoing qualitative change as well as locomotion: the whiteness that flows between an eye and an object cannot remain the same shade for any length of time whatsoever (182c6-d4).⁴ This is the first time that the constant qualitative change of the twin offspring has been recognised, but it is only what was to be expected: if, for example, a parent such as a stone is, as we already knew it must be, in constant qualitative change, and indeed consists of the sum of all such changes, including change of colour, it follows directly that its offspring, for instance this or that shade of white generated by the stone in interaction with some seeing eye, must have no duration either.

This explicit recognition of the twins’ constant qualitative (as well as locomotive) change leads Socrates into a new question (182c9-e7): given the kinds of change that must be perpetually going on, what can we succeed in saying about anything? The answer turns out to be ‘Nothing’, because the object of your discourse will be vanishing before you can get the words out.

We will turn shortly to the details of this finding, and to the lesson drawn from it. But first I must pause to emphasise a key point that seems to have been missed in the modern discussions of this passage. It is regularly assumed that the eventual upshot is a ‘collapse of language’, which is being ingeniously forced upon the Heracliteans by Socrates, and which they are expected to recognise as too high a price to pay for their flux doctrine. This misses the point of the passage.

It is the Heracliteans themselves who have, from the outset, been presented as voluntarily self-denying about the use of language. There are no determinate subjects or predicates, they told us, in that whatever you speak of as one thing (i.e. as a subject), or of one kind (i.e. as bearing a predicate), will also turn out to be the opposite (152d2-6); words such as ‘be’, ‘something’, ‘something’s’, ‘mine’, ‘this’ and ‘that’ must be banned (157a7-b7); and in conformity with these strictures they themselves make a practice of talking in riddles (179e3-180b3). Even allowing for a degree of humour in this last description, the fact remains that the move to outlaw determinate uses of subjects and predicates is presented as one made by the Heracliteans on their own behalf, and not by Socrates against them.

We must remember, once again, Plato’s early mentor Cratylus, whose commitment to flux became so extreme that he ended up abandoning language altogether. The radical reform of language described cannot credibly be considered the speaker Socrates' own ad hoc invention, and must to some extent reflect the actual linguistic practices of Heracliteans like Cratylus.

Nor, indeed, is there anything historically implausible about it if they did develop their master’s ideas in this direction. Recall once more their cardinal principle: you should not say that anything ‘is’ anything – either a determinate subject or a determinate predicate - because, whatever you call it, it will equally turn out to be the opposite. Now compare, by way of example, two classic Heraclitean sayings about god: ‘One thing, the wise, wants and does not want to be called the name Zên’ (B32); ‘The god: day, night; winter, summer; war, peace; satiety, hunger. He varies just as fire, when mixed with perfumes, receives names according to the savour of each’ (B67). Both the god’s name (‘Zên’, a variant for ‘Zeus’ that also means ‘life’) and his predicates are radically unstable, in that you cannot correctly apply one of them without its opposite following in its wake.⁵ Notice even the characteristic Heraclitean avoidance of ‘be’ in the opening of the second quotation.

The linguistic strategy that seems to have been developed by the anonymous Heracliteans reported in the Theaetetus is a thoroughly intelligent one, making real philosophical capital out of Heraclitus’ notoriously enigmatic modes of expression, and doing his philosophical writings greater justice than the mainstream doxographical tradition was ever to achieve. By locating flux primarily in the instability of subjects and predicates, due to each one’s entailment of its own opposite, it constructs a Heraclitean theory of flux which harmonises impressively with Heraclitus’ own Delphic utterances and seeks to capture the deep-seated truth about the way the world is.

What Socrates is seeking to do, then, in this final phase of his argument against the Protagorean-Heraclitean theory is not to confront the Heracliteans with unexpected and embarrassing implications for the use of language, but to find out exactly how self-denying about language they must themselves be setting out to be in order to maintain their position.

At 182c9-d7 Socrates raises the questions in the following terms:

SOCR. If then things were only moving and not altering, we would presumably be able to say with what particular qualities the moving things are flowing. Or what do we say?

THEOD. Just that.

SOCR. But since not even this remains, the flowing object’s flowing white, but it changes, so that there is flux of this very thing, whiteness, and change to another colour, so that it should not be convicted of remaining in this state, is it ever possible to speak of some colour in such a way as to refer to it correctly?

THEOD. How could there be, Socrates – or anything else of the kind – given that as one speaks it is constantly slipping away because of its flux?

(This envisaged threat of sentences becoming out of date before they have even been fully uttered will be familiar to anyone who has listened to the techniques of commentators on fast-moving sports.) Note how Socrates’ question is applied, not to things generally, but specifically to the items which feature in the ‘twins’ ontology. The whiteness in the visual encounter with a stone (156c7-157a4) must be not only in fast change, i.e. locomotion, between stone and eye, but also in slow change, i.e. qualitative change, so as to guarantee the perception’s unrevisability and incorrigibility. So far, as I said earlier, Socrates has merely made explicit what we already either knew or could easily work out.

It is only thereafter that the flux theory is pushed to an unexpected new extreme (182d8-e7). The other ‘twin’ in the encounter, sight, must also be radically unstable, not merely in the sense that the content of sight is constantly changing, e.g. from white to grey, which we would of course expect, but in respect of the very fact of seeing:

SOCR. And what are we going to say about any kind of perception, for example that of seeing or hearing? Shall we say that it ever stays in the condition of seeing or hearing?

THEOD. We shouldn’t, given that everything is changing.

SOCR. In that case we should not say that we are seeing something any more than not seeing it, nor speak of any other sense rather than its negation, if at any rate everything is changing in every way.

Why so? Not just because Heracliteanism is antecedently committed to leaving nothing stable. There must, as I have insisted all along, be a motive connected with the goal of preserving the Protagorean-Heraclitean theory of knowledge as perception. Presumably – although this is not made explicit – the point is that one of the things that we perceive is the actual fact that we are seeing, hearing, smelling etc. Consequently, if you remained in one of these perceptual states for any duration whatsoever, there would be time to revise the judgement that you are seeing, or hearing, or smelling, and to conclude that on reflection your current experience is merely a hallucination or dream. If it were a hallucination or dream, it would still of course be true (as explained at 157e1-160c6); but it is the second-order perception about one’s current perceptual modality that is now at issue. It is to avoid any element of corrigibility in this that not only the seeing of white, but even the very fact of seeing, must be no more than instantaneous.

Socrates is now ready for the killer blow (182e8-183a1):

SOCR. And yet knowledge is perception, according to what we and Theaetetus said.

THEOD. We did say that.

SOCR. So when asked what knowledge is, we answered with something that is no more knowledge than not knowledge.

THEOD. It looks that way.

The formal point of the refutation is as follows. The definition offered by Theaetetus, ‘Knowledge is perception’, can survive only if it interprets the definiens, perception, as something so unstable that at every moment it is turning into its own opposite. But that concession renders the definition itself as unstable as the definiens, no more true than false.

It is not immediately clear how we got from ‘Seeing (etc.) should no more be called seeing than not-seeing (etc.)’ to ‘Perception should no more be called perception than non-perception’. One possibility is that Socrates has omitted a step: we must not only be saved from ever revising our judgement as to whether we are (e.g.) seeing or merely hallucinating, but also as to whether we are perceiving at all or merely hallucinating.

However, I doubt this. In the passage on illusion at 157e1-160c6 all delusions, even dreams, were implicitly brought under the heading of veridical perceptions, the Protagorean-Heraclitean theory not in fact having been designed to allow for any states of consciousness other than perception.⁶ It therefore seems safer to respect the letter of the text and to let the argument stop one step earlier, with the move which saved the judgement that we are using this or that specific sense from ever becoming subject to revision.

In that case, the final failure of Theaetetus’ definition will rest on the consideration that, while defining knowledge as perception, it has to regard every actual case of perceiving, through whichever of the senses it may be, as one in which the perception through that sense has ceased to be a perception before one has finished referring to it as a ‘perception’. The only way that the perception as such could have endured would have been if it were constantly changing from, say, seeing to hearing to smelling etc., but it is not clear what it would even mean for a single perception to endure in that way; instead, not unreasonably, it is inferred that no single act of perceiving can have any duration whatsoever. If there are no enduring specific perceptions, the generic term ‘perception’ can never be applied to anything actual. One might be tempted to respond that the term could still apply generically to perception as such: we are all enduringly perceiving, albeit not enduringly seeing, hearing, etc. But to introduce any such universal is already to go beyond the Heraclitean ontology of fleeting particulars.

We must now turn to the broader lesson which Socrates draws from the refutation (183a-b). It is here that, as regularly read, the passage is taken to conclude that radical flux leads to the collapse of language. That is surely not the conclusion. Rather, the final upshot is that language will need even more restrictions than the Heracliteans already aim to impose. That it must get by without determinate subject and predicate expressions has been previously established; I have suggested that such modes of discourse are intended as fully in the spirit of Heraclitus himself, and arguably, far from being an abandonment of truth, as representing the one way in which language can capture the truth about the world. The new turn of the screw is to show us that this mode of discourse must after all give up any claim to secure determinate truth values: the subject of a sentence cannot hold its reference long enough for a predicate to be attached to it. So the Heraclitean language must in addition get by without locutions implying determinate truth values: that things are ‘thus’ and ‘not thus’, as Socrates puts it. And even this is presented by Socrates, not as a complete abandonment of language (see especially 183b4-5),⁷ but as involving the fatal concession that every ‘answer’ on any subject is equally correct (183a4-6).

One might think that to say this is to invalidate language as a whole, and certainly that further consequence could have been argued, but in fact it is not. Both the immediate context and the regular use of ‘answer’ in the dialogue confirm that Socrates means something more restricted: this is a collapse, not of language, but of dialectic. The Socratic method par excellence, dialectic, is centred on the investigation of universal truths, especially definitions, by means of question and answer, and every one of the 43 occurrences of ‘answer’ in the dialogue outside this passage refers to dialectical answering, in most cases (including the passage just quoted) the proffering of a definition.

Hence the upshot of Socrates’ refutation of the flux thesis is that, if as the perceptual theory requires there is literally nothing stable, there is nothing about which one’s dialectical answer could hold true, and hence in particular there are no definitions. The Heracliteans are expected to insist that discourse of their own favoured kind can continue. The indeterminacy of its referring terms, and even (as they must now agree) of its truth values, accurately captures the flux of the actual world. If, that is, this now means that their assertions are, taken as a whole, no more true than false, there is no reason to assume that they will not welcome the consequence; indeed, it could even be to real-life Heracliteans that Aristotle is referring when he reports that, according to some people, Heraclitus himself makes assertions which openly violate the Principle of Non-Contradiction (Met. Gamma 3, 1005b23-6). Socrates’ objection is not that such a view of things cannot be adequately expressed in language, but simply that it postulates a world in which there can be no dialectic, and, more specifically, no definitions. Consequently Theaetetus’ definition of knowledge really does undermine itself: it is a definition that presupposes a world in which there can be no definitions.⁸

Bibliography

Burnyeat, M.F., introduction to M.J. Levett, trans., The Theaetetus of Plato, revised by M.F. Burnyeat Indianapolis/Cambridge, 1990.

Cherniss, H.F., ‘The Philosophical Economy of the Theory of Ideas’, American Journal of Philology 57, 1936, pp. 445-56; repr. in Studies in Plato’s Metaphysics, ed. R.E.Allen, London/New York, 1965, 1-12.

Cooper, J.M. (1967), Plato’s Theaetetus, (reprint of 1967 Harvard Ph.D. dissertation; New York/London, 1990)

Denyer, N., Language, Thought and Falsehood in Ancient Greek Philosophy, London, 1991.

McDowell, J. Plato Theaetetus trans. with introduction and notes, Oxford. 1991.

Sedley, D., Plato’s Cratylus, Cambridge, forthcoming (2003).

Silverman, A., ‘Flux and Language in the Theaetetus', Oxford Studies in Ancient Philosophy 18, 2000, pp. 109-52,