¹ I owe this observation, and several points pertinent to it, to Bernard Williams' brilliant survey, ‘The Legacy of Greek Philosophy’, in The Legacy of Greece: A New Appraisal, Finley, M. I. (ed.) (Oxford: Oxford University Press, 1981).Google Scholar

² One reason Berkeley missed this is that he translated Theaetetus' definition as ‘Sense is science’, taking the thesis of Theaetetus and Protagoras to be that sense alone suffices for knowing (understanding) the connections between things. Consequently he thought he could agree with Plato's refutation of Theaetetus and Protagoras without ceasing to approve what he supposed to be Plato's theory of perception and the flux of sensible things (§§253, 304–305). I discuss this aspect of Berkeley's reading of the Theaetetus in ‘Aristotle on Understanding Knowledge’, in Aristotle on Science: The ‘Posterior Analytics’, Berti, E. (ed.) (Padua: Antenore, 1981).Google Scholar

Failure to grasp that the argument is an extended reductio ad absurdum and that the theory of perception is not presented as Platonic doctrine is equally characteristic of modern commentators on the Theaetetus—although they have not Berkeley's excuse of mistranslation to disassociate the definition from the theory which supports it. This is not the place for elaborate exegesis or scholarly controversy, but the case for the reading I present can be summarily set out in three stages, as follows.

(A) We first go through the text picking out the main stage-directions, as it were, by which Plato indicates how, in his view, the three theses under discussion (Theaetetus, Protagoras, Heraclitus) are related. This is best done in abstracto, without delving into the content of the theses themselves.

Most important, because centrally and emphatically placed at a turning point in the discussion, is 160de: the three theses ‘come to the same thing’ [Th↔ Prot↔Her], and more particularly (ei), if Protagoras and Heraclitus are correct, perception is knowledge, as Theaetetus says [(Prot & Her)→Th]. Compare 183a, already cited: Heraclitus was brought in to make Theaetetus' definition hold good [Her→Th]. Now go back to the beginning of the discussion at 151e–152a: Theaetetus and Protagoras say the same thing in different ways [Th↔Prot]— and then follows argument (152a–152c) to show that Protagoras makes Theaetetus' definition come out right [Prot→Th]. Compare 164d: counter-examples to Theaetetus tell equally against Protagoras [Prot→Th]. Next, 152cd: Heraclitus gives the ‘real truth' behind Protagoras’ riddling statements [Prot↔Her]— and soon follows extended argument (153d ff.) to show that Protagoras requires a Heraclitean ontology [Prot→Her]. Compare 166b: Protagoras relies on Heraclitus to defend himself against an objection to Theaetetus' definition [(Prot→ Th) & (Prot→Her)]. Finally 183b: the refutation of Heraclitus demolishes Protagoras [Prot→Her] and disposes of Theaetetus' definition [Th→Her]— unless Theaetetus can find some other method than Heraclitus' to work out his equation of knowledge and perception. This terminal qualification to the mass of evidence just listed shows that there is a difference of status between the two halves of the equivalence we began from.

It is thought to be reasonably clear that

(1) Her→Prot→Th.

The work goes into showing

(2) Th→Prot→Her,

and then that both Protagoras and Heraclitus engender absurdity. (2) is hammered out step by step through the construction of the Protagorean-Heraclitean theory down to 160de. At each step the claim is that Theaetetus has no reasonable alternative if his definition is to be vindicated (for more detail on this characteristic feature of the dialectical method, see my papers ‘Examples in Epistemology: Socrates, Theaetetus and G. E. Moore’, Philosophy 52 (1977), 381–398Google Scholar, and ‘Socratic Midwifery, Platonic Inspiration’, Bulletin of the Institute of Classical Studies 24 (1977), 7–16)Google Scholar. Consequently it remains an abstract possibility at the end that Theaetetus might find some alternative to (2) to avoid the reductio— a possibility which is, however, foreclosed by the direct refutation of the definition which follows at 184b–187a. In sum, the Protagorean-Heraclitean theory states a complete set of sufficient conditions for Theaetetus' definition to hold good, which conditions, it is argued, are (i) necessary conditions for it, (ii) harbingers of absurdity and hence, in the end, its downfall.

(B) If commentators have almost to a man been unwilling to take at face value these manifold indications by Plato as to the intended structure of his argument (and there are many lesser confirmatory signs which I have not mentioned), that is because they have not seen the underlying philosophical connections which make (2) intelligible and plausible. So the next task is to outline the connections in a manner which will enable us to take Plato at his word: see text below (it turns out that on this aspect Berkeley's philosophical acumen scores better).

(C) Even so, even supposing that the account I shall sketch is found satisfactory, one major stumbling block will remain. To carry through the reductio it is necessary to remove the impression many readers have formed (most recently, John McDowell, Plato-Theaetetus (Oxford: Clarendon Press, 1973), 179–184Google Scholar, Bolton, Robert, ‘Plato's Distinction Between Being and Becoming’, Review of Metaphysics 29 (1975), 66–95)Google Scholar that what finally gets refuted at 179c–184b is a different and more extreme Heraclitean theory than that elaborated in the earlier section to 160e. The answer here is that Plato furnishes an argument (181de) to show that there is no escaping the further developments which are to be the theory's undoing. And here too an appreciation of the underlying connections between the three theses is indispensable. For, unlike most commentators, I believe that this argument is to be taken seriously. But that is a large project which I must leave to another occasion. Our present need is to understand the theory, not destroy it.

³ For a defence of the claim that Plato is serious about taking Protagoras' Measure doctrine as a theory of truth and derives from it the above rule and its converse, see my ‘Protagoras and Self-Refutation in Plato's Theaetetus’, Philosophical Review 85 (1976), 172–195.Google Scholar

⁴ That all perception is knowledge is explicitly and validly argued at 152ac from the rule that every appearance is true for the person who has it and the premise that perception is the same as having an appearance. That all knowledge is perception is not explicitly argued there, but it can be derived by application of the converse rule and it needs to be derived if Socrates is to prove that the whole equation of knowledge and perception follows from Protagoras' philosophy. The reason for Plato's silence here about the latter half of the package is, I think, the following: it is only where sensible qualities are concerned that one would venture to equate the having of appearances with perception—Socrates says as much at 152C1–2—and Plato wants later (161b ff.) to discuss Protagorean relativism in its most general form, as the view that things are for each person as they appear to him, whether the ‘appearing’ is appearing to sense or to thought. The completely general relativism will preserve the thesis that all perception is knowledge, while allowing for knowledge (veridical appearance) that is not perception. There is more to be said on this aspect (for some of it, see paper cited in the preceding note), but for present purposes it will be enough to amend our previous overall description of the argument: it is a reductio ad absurdum with asides, namely, those asides which treat of Protagorean relativism in a more general form than is required for sustaining Theaetetus' definition. That said, we can from now on confine attention to sensible appearances and to one half of Theaetetus and Protagoras: the thesis that all perception is knowledge and the rule that, whatever sensible appearances a person has, they are true for him.

⁵ References to Berkeley's Three Dialogues are by page numbers of Jessop's edition: The Works of George Berkeley, II, Luce, A. A. and Jessop, T. E. (eds) (London: T. Nelson, 1949).Google Scholar

⁶ The above summarizes a two-stage argument in the Theaetetus. Socrates first establishes the relativity of sensible qualities (153d–154b) and then develops its implications for the identity of objects (156a ff.). Already at stage one there are complications into which we should not enter here, but interestingly they are complications paralleled in Berkeley's Three Dialogues, as I try to show in ‘Conflicting Appearances’, Proc. Br. Acad. 65 (1979), 69–111.Google Scholar

⁷ If ‘represent’ here has to be construed differently for Berkeley and for Protagoras, that is part of the ontological divergence we are coming to.

⁸ This is the correct translation of gignesthai, which Berkeley (above, p. 20) rendered ‘is made’.

⁹ The caveat is necessary on account of a surprising group of entries in Berkeley's unpublished Philosophical Commentaries headed by No. 577 (obelized for rejection): ‘The very existence of ideas constitutes the soul’. This belongs with the early stages of Berkeley's philosophizing studied by Luce, A. A., The Dialectic of Immaterialism (London: Hodder and Stoughton, 1963)Google Scholar. It is explicitly denied at Dial. 233.Google Scholar

¹⁰ No doubt the mind must have ideas to operate upon (cf. Phil. Comm. 478Google Scholar; Princ. §§27, 139Google Scholar; Dial. 231–234Google Scholar), but that is not enough for ontological dependence. Cf. Phil. Comm. 878Google Scholar: ‘Extension tho it exist only in the Mind, yet is no Property of the Mind, The Mind can exist without it tho it cannot without the Mind’.

¹¹ Thus the ‘mental containment’ view of sense-qualities which Findlay, J. N., Plato: The Written and Unwritten Doctrines (London and New York: Routledge and Kegan Paul, 1974), 362Google Scholar, attributes to Plato in the Theaetetus is neither Plato nor in the Theaetetus.

¹² On this point the comparison is with Russell's position in The Problems of Philosophy (Oxford: Oxford University Press, 1912), 19Google Scholar, rather than with Berkeley's mature philosophy. See further my ‘Plato on the Grammar of Perceiving’, Classical Quarterly 26 (1976), 29–51Google Scholar. Of course, having got thus far it is only a short step, as Russell found, to abolishing the subject-object distinction altogether. But the point is that the step is not taken in the Theaetetus. In any case, if one detects an idealist slant in the ostensibly neutral monism of modern philosophers who have pushed further, that has a lot to do with one's knowledge that, historically, the monism is reached by way of Berkeley and presupposes Berkeley's elimination of matter. It is Berkeley's previous elimination of matter which ensured that, when Hume cut out Berkeley's substantial mind, all he could be left with was perceptions.

¹³ See De Anima II, 5 and III, 2. Berkeley overlooks Aristotle's central claim (418a3–6) that the sensible object must already be in actuallity what, prior to the act of perception, the sentient subject is potentially. If the red apple's redness is a potentiality as well as an actuality, this is a second potentiality, on a par with the potential knowledge of a man who has actually learned something but is not currently using his knowledge, not with the potentiality which precedes the learning. Berkeley also draws on Aristotle's doctrine that actual knowledge and the thing known are one: ‘Whence it follows that the things are where the knowledge is, that is to say, in the mind’ (Siris §310). Aristotle's own conclusion is, of course, not that at all: ‘It is not the stone which is in the soul but its form’ (De An. 431b29). At Met. 1010b30–35 (which Berkeley should have seized on) the aisthēta that are conceded to depend for their existence on being perceived must be actualized sensible qualites (so Christopher Kirwan, Aristotle's Metaphysics Γ, Δ, Ε (Oxford: Clarendon Press, 1971)Google Scholar, ad loc), or else Aristotle will be slipping into the Megarian account of possibility which he disputes on this very issue in Met. H 3.

¹⁴ In stricter moods Aristotle would not allow that an apple, as opposed to an apple tree, is a proper substance. But this hardly affects the issue, and I choose the example as being Berkeley's own (Princ. §1).

¹⁵ Diels-Kranz, , Die Fragmente der Vorsokratiker, 6th edn (Berlin: Weidmannsche Buchhandlung, 1951), 70Google Scholar A 6. Cf. 67 A 8, 68 A 38. More on Metrodorus in n. 42 below.

¹⁶ Siris §§262, 266–269, 274–279, 290–291, 328, 352–353, 362–364.

¹⁷ Cf. Armstrong, A. H., The Architecture of the Intelligible Universe in the Philosophy of Plotinus (Cambridge: Cambridge University Press, 1940), 87, 113.Google Scholar

¹⁸ ontōs ouk on (II, 5.5.25), to be contrasted with that which is not at all (to pantē/panteles mē on, VI, 9.11.36–38).

¹⁹ This inconsistency between Siris and the earlier Berkeley is well noted by Baladi, Naguib, ‘Plotin et l'Immatérialisme de Berkeley. Témoignage de la Siris’, in Atti del Convegno Internazionale sul tema: Plotino e il Neoplatonismo in Oriente e in Occidente (Rome: Accademia dei Lincei, Quaderno No 198, 1974). 597–602.Google Scholar

²⁰ For a recent discussion and advocacy of the view, see Rist, J. M., Plotinus: The Road to Reality (Cambridge: Cambridge University Press, 1967), chap. 4.Google Scholar

²¹ Authorities who are firmly against a Plotinian idealism include Inge, W. R., The Philosophy of Plotinus II, 3rd edn (London, New York, Toronto: Longmans Green & Co., 1929), 37ffGoogle Scholar., and Armstrong, , op. cit., chap. 8Google Scholar. According to Wallis, R. T., Neoplatonism (London: Duckworth, 1972), 124Google Scholar, later Neoplatonists make sure that any ambiguities in Plotinus' position are resolved in the direction of realism.

²² Objection: But Parmenides himself famously declares that the words mortals use, words like ‘come into being’, ‘changing place’, etc., are mere names (frag. 8, 38), i.e. empty names to which nothing corresponds. Answer: It is his editors who have made him say this. The authentic text, with much the best authority in the manuscripts, is not toi pant' onom' estai (‘Wherefore all these are [sc. mere] names which mortals have laid down believing them to be true’), but toi pant' onomastai. The latter was vigorously defended by Woodbury, Leonard, ‘Parmenides on Names’, Harvard Studies in Classical Philology 63 (1958), 145–160CrossRefGoogle Scholar, with the translation ‘With reference to it [sc. that-which-is] are all the names given that mortal men have instituted, in the belief that they were true’. But his defence has remained open to a technical grammatical objection (see Owen, G. E. L. in Studies in Presocratic Philosophy, II, Allen, R. E. and Furley, D. J. (eds) (London: Routledge and Kegan Paul, 1975), 69)Google Scholar. A simple solution will circumvent the difficulty: keep toi as ‘wherefore’ and take the subject of the verb from to eon in the previous line (construction as in frag. 9, 1): ‘Wherefore it (the one being) is named all the names which mortals have laid down in the (mistaken) belief that they are true (of it)’. On this construal mortals continue to talk about something, viz. the only thing there is to talk about, but what they say about it is wrong and contradictory.

²³ Cf. Oehler, Klaus, Die Lehre vom Noetischen und Dianoetischen Denken bei Platon und Aristoteles (Munich: C. H. Beck, 1962), 103ffGoogle Scholar., who sees the connection between the non-idealism of the Parmenides and some of the differences between ancient and modern scepticism discussed below. In the Sophist Plato defuses the Eleatic principle precisely by showing that to think or speak of what is not is not to think or speak of nothing. That the something thought is independent of thought is the point which, in its several ramifications, also falsifies Hegel's claim to find anticipations of his own brand of idealism in Plato and Aristotle: Lectures on the History of Philosophy II, trans. Haldane, E. S. and Simson, Frances H. (London and New York: Routledge and Kegan Paul, 1955), 1, 43, 188, 196.Google Scholar

²⁴ For an assessment which answers that it does play fair, see Keyt, David, ‘The Mad Craftsman of the Timaeus’, Philosophical Review 80 (1971), 230–235.CrossRefGoogle Scholar

²⁵ For documentation and defence of the interpretation to be given of Pyrrhonian scepticism, see my ‘Can the Sceptic Live His Scepticism’, in Doubt and Dogmatism: Studies in Hellenistic Epistemology, Schofield, Malcolm, Burnyeat, Myles, and Barnes, Jonathan (eds) (Oxford: Clarendon Press, 1980)Google Scholar. The main text is the first book of Sextus' Outlines of Pyrrhonism (PH).

²⁶ The connection was pointed out to me by David Owen.

²⁷ Cf. Stough, Charlotte, Greek Skepticism (Berkeley & Los Angeles: University of California Press, 1969), 1426ff.Google Scholar

²⁸ PH I, 215Google Scholar, is an apparent exception where Sextus is in fact reporting, and resisting, someone else's attempt to assimilate Pyrrhonism to Cyrenaic scepticism, for which see below.

²⁹ Galen, , De pulsuum differentiīs VIII, 711, 1–3Google Scholar Kühn, available also in Deichgräber, Karl, Die Griechische Empirikerschule (Berlin: Weidmann, 1930), frag. 75, p. 135, 28–30.Google Scholar

³⁰ Aristocles apud Eusebius, , Praeparatio Evangelica XIV, 19Google Scholar, 1 = frag. 212 in Mannebach, Erich, Aristippiet Cyrenaicorum Fragmenta (Leiden and Cologne: E. J. Brill, 1961)Google Scholar; cf. frag. 214. The Cyrenaic theory of knowledge was developed by Aristippus the Younger in the second half of the fourth century B.C.

³¹ I am nevertheless grateful to Keith McCullough for drawing my attention to the relevance of the fact.

³² Sextus, adv. Math. VII, 190ff., Plutarch, Against Colotes 1120b ff.= Mannebach frags 217–218.

³³ This seems to be the earliest attested citing of the familiar philosophical myth about jaundice: see further my ‘Conflicting Appearances’, op. cit.

³⁴ Cf. n. 13 above. The suggested derivation is due to Keith McCullough.

³⁵ E.g. De An. 425b22–24: ‘That which sees has in a certain way become coloured; for in each case the sense organ is capable of receiving the sensible object without its matter’.

³⁶ The significance of this innovation of Augustine's as a step towards a Cartesian conception of the mind is well brought out by Gareth B. Matthews in a paper offering much that is relevant to our investigation: ‘Consciousness and Life’, Philosophy 52 (1977), 13–26.Google Scholar

³⁷ Texts and historical discussion in Deichgraber, op. cit.

³⁸ On Hume's objection and Sextus' reply, see paper cited n. 25 above.

³⁹ Caution: Platonic soul-body dualism is not to the point here, since it puts no epistemological barrier between soul and body. The body is part of the material or sensible world, which is not at all the same as being part of ‘the external world’ in the modern sense. That is one reason why Plato can vacillate over which ‘mental’ functions belong properly to the soul and which to the body.

⁴⁰ The reference is to Williams, Bernard, Descartes: The Project of Pure Enquiry (Harmondsworth: Penguin Books, 1978), esp. chap. 2.Google Scholar

⁴¹ De dignoscendis pulsibus VIII, 781, 16–783Google Scholar, 5 Kühn = Deichgräber, op. cit., frag. 74, p. 133, 19–p. 134, 6.

⁴² A couple of further texts should be mentioned here, if only that they may be discounted. (1) Xeniades of Corinth, a little-known figure of the fifth century B.C. said that everthing, i.e. every appearance and opinion, is false (Sext. adv. Math. VII, 53–54).Google Scholar Possibly what he meant by this was that nothing meets the Eleatic conditions for true being, since he also remarked that everything that comes to be comes to be from not being and everything that perishes perishes into not being. If so, the effective content of Xeniades' claim would be that none of the things that come to be are really and truly what they appear to be (cf. Melissus frag. 8). But we have no knowledge of the wider context of these assertions. (2) Metrodorus of Chios was famous for having capped the sceptical denial that we know anything by further denying that we know whether we know anything or not (frag. 1), and then on top of that he says (in one source only: Cic. Acad. II, 73)Google Scholar that we do not know whether anything exists or nothing. This, however, was the exordium to a work On Nature containing inter alia an atomist meteorology. It would seem, therefore, that for Metrodorus what we do not know we may none the less theorize about, and perhaps we may explain in terms of the atomic theory why we do not know it. I think we may be confident that if Xeniades, Metrodorus, or anyone else had come at all close to a genuinely Cartesian doubt, the sceptic doxography would have picked it up and told us loud and clear.

⁴³ On Augustine and the Cogito, see Gilson, Etienne, Etudes sur le rôle de la pensée médiévale dans la formation du système cartésien (Paris: J. Vrin, 1930), 191–201.Google Scholar

⁴⁴ References to Descartes are by volume and page number in Haldane, E. S. and Ross, G. R. T., The Philosophical Works of Descartes, corrected edn (Cambridge: Cambridge University Press, 1934).Google Scholar

⁴⁵ For historical information on the transmission to modern times of the two main streams of ancient scepticism, see (for Cicero) Schmitt, Charles, Cicero Scepticus: A Study of the Influence of the ‘Academica’ in the Renaissance (Hague: Martinus Nijhoff, 1972)CrossRefGoogle Scholar and (for Sextus) Popkin, Richard H., The History of Scepticism, from Erasmus to Spinoza (Berkeley, Los Angeles and London: University of California Press, 1979).Google Scholar

⁴⁶ E.g. Rescher, Nicholas, ‘The Legitimacy of Doubt’, Review of Metaphysics 13 (1959/1960), 226–234.Google Scholar

⁴⁷ Further comparisons with Carneades in Couissin, Pierre, ‘Caméade et Descartes’, in Travaux du IX^e Congrès International de Philosophie: Congrès Descartes (1937)Google Scholar, IIIme Partie, 9–16; and with his Stoic opponents in Brochard, V., ‘Descartes Sto'icien’, in his Études de philosophie ancienne et philosophie moderne (Paris: F. Alcan, 1954), 320–326.Google Scholar

⁴⁸ Descartes' Conversation with Burman, trans. Cottingham, John (Oxford: Clarendon Press, 1976), 4.Google Scholar It has been said that ‘it is not true … that Cartesian doubt is more radical than ancient scepticism’ (Caton, Hiram, The Origin of Subjectivity: An Essay on Descartes (New Haven and London: Yale University Press, 1973), 29)Google Scholar—this on the basis of a survey of some of our Protagorean and Pyrrhonian material. The Burman text shows that Descartes himself had a better understanding of his relation to the sceptical tradition.

⁴⁹ This vital difference is overlooked by Robin, Léon, Pyrrhon et le scepticisme grec (Paris: Presses Universitaires de France, 1944), 89–90Google Scholar, who claims that Carneades got as far as Descartes' evil demon doubt. Whether the evil demon was prompted, directly or indirectly, by the Academica is a separate issue about which I make no hypothesis. Curley, E. M., Descartes Against the Skeptics (Oxford: Blackwell, 1978)CrossRefGoogle Scholar, is well informed about the relations between Descartes and ancient scepticism, and he prefers to postulate a source in Montaigne, (pp. 38–40, 68–69).Google Scholar

⁵⁰ See Descartes' letter to Reneri for Pollot, April 1638, in Descartes: Philosophical Letters, translated and edited by Kenny, Anthony (Oxford: Clarendon Press, 1970), 53Google Scholar: ‘Although the Pyrrhonians reached no certain conclusion from their doubts, it does not follow that no one can’.

⁵¹ Interestingly, this contrast was well appreciated by Hegel, , op. cit., 347.Google Scholar

⁵² Cf. Conversation with Burman, 3Google Scholar: ‘Here we are dealing primarily with the question of whether anything has real existence’.

⁵³ Distinguish this account from the story recently told by Rorty, Richard, Philosophy and the Mirror of Nature (Princeton: Princeton University Press, 1979)Google Scholar, esp. chaps. I–II, the moral of which is that indubitability or incorrigibility—the idea that we have incorrigible knowledge of our subjective states—was the innovation by which Descartes created the modern philosophical notion of the mind. Incorrigibility was there before in Hellenistic philosophy, in the shape of Sextus' description of appearance-statements as azētētos, immune to question or inquiry (p. 38 above). The addition of truth is what opens up a new realm for substantial knowledge, and it is knowledge not just because it is incorrigible truth but because of what Descartes will build upon it.

⁵⁴ Cf. Popkin, op. cit., chap. IX.

⁵⁵ HR I, 143, 148, 219–220: II, 44, 206, Letter to Hyperaspistes, August 1641, Kenny, , op. cit., 110.Google Scholar

⁵⁶ Diogenes Laertius IX, 62.

⁵⁷ I have illustrated this by reference to Pyrrhonian happiness, but it is manifest also in the Academic defence of the possibility of sceptical action.

⁵⁸ I am grateful for discussion at the Royal Institute of Philosophy and at the Moral Science Club in Cambridge, and for suggestions from Andreas Berriger, Henry Blumenthal, Ted Honderich, Keith McCullough, David Owen, Hans Sluga, and Richard Sorabji. From Bernard Williams have come both discussion and suggestions, and a great deal more: my starting point in the present essay was an observation of his (n. 1 above), the account in section I of Plato's argumentative strategy in Part I of the Theaetetus is basically his, and I am conscious of more important debts accumulated over the years which cannot be measured in terms of this idea or that. For such gifts the only proper return is the endeavour to make worthy use of what one has learned.