1. Leibniz, G. W., New essays on human understanding, preface 48–9 (trans. Remnant, P. & Bennett, J.)CrossRefGoogle Scholar.

2. Aëtius Plac. 4.1 = Stoicorum Veterum Fragmenta (SVF) II.83, 11.13–23. The Greek text is quoted below on p. 136.

3. Plutarch, , De communibus notitiis 1070c–dGoogle Scholar. For the Greek text see below p. 137. I discuss this translation in n.36 below.

4. Meno 85d9–86b4. For some reasons for taking recollection very literally in the Phaedo see my ‘Platonic anamnesis revisited’ CQ 37 (1987) 355Google Scholar.

5. Meno 82b9–5d4.

6. Phaedo 74a9–5b2.

7. For a statement of this criticism see Gallop, D., Plato's Phaedo (1975) 132Google Scholar.

8. See Jolley, N., Leibniz and Locke (1984) 167Google Scholar; Barnes, J., ‘Mr Locke's darling notion’, PQ 22 (1972) 193CrossRefGoogle Scholar; Aaron, R. I., John Locke ed. 3 (1971) 94Google Scholar.

9. Locke, Essay Bk 1.2 §5.

10. Locke's strategy is clearly set out by Jolley (1984) 170. Recently, Nelson Goodman has exhumed the point of 1.2 §5 in ‘The epistemological argument’ reprinted in Morrick, H. (ed.), Challenges to empiricism (1980) 255–6Google Scholar.

11. That he singled out this portion of Locke's attack and devoted such a large amount of space to it is an indication of its importance in the debate. It is worth noting that Locke's 2 §5 has the longest of the replies in Leibniz's Bk 1.

12. Leibniz, New essays 1.1.78.

13. Leibniz seems to recognize his relationship to Platonic recollection at N.E. Preface, p. 52, 1.1.78–9, and 1.3.106. See also Discourse on Metaphysics 26.

I should point out that Leibniz's theory of implicit knowledge is by no means the only type of innateness he espouses; this is in fact only one strand out of the very complex account of innatism that he gives in the New essays. Some scholars have listed several different types of innateness in Leibniz. (See, for instance, Savile, A., ‘Leibniz's contribution to the theory of innate ideas’, Philosophy 47 (1972) 113–24CrossRefGoogle Scholar; Jolley (n.8) 171–8. Among these is the theory of innate dispositions that I discuss below. See also n.25 below.

14. Among the Cambridge Platonists, More and Cudworth embraced dispositionalism and often took some care to reject claims that we are born with any actual knowledge. See, for instance, More's Antidote against atheisme 1.5 §2. Descartes made similar qualifications to innatism. See next note.

15. Descartes, Notes against a certain programme (trans. Haldane, & Ross, ) vol. i (1984) 442Google Scholar.

16. Locke, Essay 1.3 §9.

17. For this criticism of Locke see Jolley (n.8) 171.

18. One such philosopher seems to have been Locke's most forthright opponent, Bishop Stillingfleet, who is discussed by Yolton, , John Locke and the way of ideas (1956) 36–8Google Scholar.

19. Meno 97a3–4.

20. This would be to follow Meno's example of mouthing the pronouncements of his mentor Gorgias. In the first few pages of the Meno Plato uses a number of references to Gorgias to underline Meno's shortcomings in this respect. At first, Socrates seems to make out that discovering the definition of virtue is a question of recollecting what Gorgias has said on the matter (70c10–d1); he asks Meno to perform this task of recollection, but then generously adds that Meno can say what he himself thinks (as he will no doubt agree with Gorgias). After Meno's first downfall, Socrates asks him again to recollect what Gorgias said (73c6–8), hastily adding a terse καὶ σύ to imply that this will be Meno's own view. At 76b1, however, Meno is getting closer to ἀπορία and Socrates asks him for a third time to remember what Gorgias has said, though now without pretending that Meno has worked things out for himself. Soon, of course, we shall see that the discovery of the definition will involve a rather different kind of recollection. See also 76c4.

21. See above, p. 126.

22. Quine, W. V. O., ‘Natural kinds’, in Ontological relativity and other essays (1969) 123Google Scholar.

23. Quine, W. V. O., Word and object (1960) 83Google Scholar.

24. Quine, W. V. O., ‘Linguistics and philosophy’ in Hook, S. (ed.), Language and philosophy 96Google Scholar.

For Chomsky's reaction to Quine, see ‘Quine's empirical assumptions’ in Davidson, D. & Hintikka, J. (eds.), Words and objections: essays on the work of W. V. O Quine (1969) 53–68Google Scholar, esp. 54–5; Reflections on language 198ff.

25. Leibniz is here showing his dispositionalist tendencies as an innatist. Dispositionalism seems to co-exist alongside his theory of implicit or unconscious knowledge in Bk 1 of the New essays. In such dispositionalist moods Leibniz invokes the analogy of a veined block of marble: learning is not like carving a Hercules out of any old block of marble, but out of one whose veins naturally follow the shape of Hercules. Leibniz dismisses the tabula rasa because it allows anything to be written on it. His tabula would be such as to favour a certain message. See Preface 52, and Bk 1.1.80, where he talks of the mind having special affinities, dispositions, aptitudes and pre-formations, and 86, where he mentions potentialities and tendencies. All these he opposes to a bare passive faculty. For an interpretation which claims that for Leibniz dispositinalism in fact reduces to implicit or unconscious knowledge see Broad, C.D., Leibniz: an introduction (1975) 134–5Google Scholar, and Jolley, N., ‘Leibniz and Malebranche on innate ideas’, PR 97 No. 1 (1988) 86–7Google Scholar.

26. See Barnes, (n.8) 211–12; Hacking, I., Why does linguistics matter to philosophy? (1975) 57CrossRefGoogle Scholar; Jolley (n.8) 167 and 179, where he connects the dispute about innatism with one about materialism.

27. Leibniz, , New essays, preface 49Google Scholar; see Hacking (n.26) 63.

28. See Descartes' third Meditation and the first book of More's Antidote against atheisme.

29. For a discussion of the strong moral motivation behind the doctrine of innate ideas see Yolton (n.18) 28–35.

30. This objection was made in 1666 by Parker, Samuel in A free and impartial censure of the Platonick philosophy, 55Google Scholar.

31. This is More's approach at the beginning of his Antidote against atheisme. In a later edition of the same work he included an appendix in which he expressly denied that he had argued from the innateness of our ideas to their reliability; rather, it is the naturalness of certain ideas, whether they are innate or not. See More, H., A Collection of several philosophical writings (1662)Google Scholar, Appendix to the Antidote against atheisme II §§1–3Google Scholar.

32. For this approach see, for instance, Herbert of Cherbury De Veritate (Locke's only named opponent on the issue of innate ideas) (trans. Carré, Meyrick H.) 126Google Scholar.

33. To revert to Platonic apologetics: the epistemological impotence of psychological innatism rules it out as a reasonable alternative to anamnesis. Because Plato was attempting to solve an epistemological problem when he first proposed anamnesis, he could have no interest in dispositional innatism as a purely psychological thesis: he would have wanted to see how it could account for full-blooded knowledge. So merely to say that he should have taken up dispositional innatism misses the point. In Locke's day the obvious way to supplement the psychological thesis was to appeal to God. Plato does not avail himself of God, and so without Him would ask where the innate knowledge comes from. The answer is that it cannot come into being (any more than God's knowledge does); so an eternally pre-existent soul is a perfectly logical solution, and one necessary to solve the problem that he has raised (see Meno 85d12ff). If the charge is that Plato should have invoked God, at least he can no longer be accused of greater metaphysical extravagance than some of his successors. At any rate, it is amply clear by now that Plato was a wishful thinker, not in the sense that he wished the soul to be immortal but that he wished it to know. (I am grateful to Myles Burnyeat for suggesting this line of argument.)

34. The debate over whether the Stoics believed in innate moral notions is long-lived. Among those who advocate an innatist interpretation are Bonhöffer, A. (Epictet und die Stoa (1890) 199 207)Google Scholar and Grumach, E. (‘Physis und Agathon in der alten Stoa’, Problemata 6 (1932) 72–6Google Scholar). Bonhöffer argued that Stoic prolepses were seminal moral concepts whose origin owed nothing to sensation; he was then taken to task by Sandbach, F. H. in his influential article ‘Ennoia and prolepsis’ (reprinted in Long, A. A. (ed.), Problems in Stoicism (1971) 23–37)Google Scholar. Rist, J. M. (Stoic Philosophy (1969) 134)Google Scholar, Gould, J. B. (The philosophy of Chrysippus (1970) 64 n.2)CrossRefGoogle Scholar and Todd, R. B. (‘The Stoic common notions’, Symbolae Osloenses 48 (1973) 53ff.CrossRefGoogle Scholar) have sided with Sandbach's strongly empiricist interpretation, while earlier M. Pohlenz had advocated a middle path, though still rejecting any theory of innate a priori knowledge (Grundfragen der stoischen Philosophie (1940) 82–99Google Scholar, esp. 96, and Die Stoa ed. 2 (1969) I 56–9Google Scholar, II 33–5); he was followed by Luschnat, O., ‘Das Problem der προκοπή in der alten Stoa’, Philologus (1958) 191–2)Google Scholar. Other scholars who are more sympathetic to the innatist cause, or who at least feel uneasy with Sandbach's position are Watson, G. (review of Gould in PQ 22 (1972) 268–9CrossRefGoogle Scholar) and Irwin, T. (‘Aristotelian and Stoic conceptions of happiness’ in Schofield, M. & Striker, G. (eds.), The norms of nature (1986) 208–9Google Scholar with n.5).

35. See above p. 123.

36. When I translated this sentence on p. 124 above I took ταûτα (‘the former’) to refer not to the good and bad things themselves, but to the notions of them. This makes a better parallel with ἐκείνων μὲν… αἱ φαντασίαι and means that we do not have to attribute to the Stoa the view that both good and evil proceed from within the soul; this would contradict the point of Galen's De placitis Hippocratis et Platonis 5.5 and Diogenes Laërtius 7.89 (among others) which I discuss below. See pp. 142ff.

Also, I am accepting ἀρχῶν as an emendation of ἀγαθῶν, which makes very little sense in the context. This emendation has been proposed by Kronenberg, A. J. (‘Ad Plutarchis Moralia’, Mnemosyne (1924) 105–6)Google Scholar. As an alternative emendation of ἀγαθῶν, Pohlenz has suggested ὰφορμῶν (Die Stoa II 34Google Scholar). This would be very tempting in the light of what I say below.

37. The word ἐπεισόδιος is used in a context very relevant to ours in a fragment of Plutarch describing Plato's theory of recollection (Plutarch, Moralia vol. 15 (Loeb) 388–9Google Scholar). In contrast to the innate knowledge within us, Plutarch talks of τῶν ἄλλων ἐπεισοδίων, presumably referring to the incoming data of sense-perception. For a discussion of this fragment see my ‘Platonic anamnesis revisited’ (n.4) 349 with n. 10.

38. Sandbach ((n.34) 28) has difficulties with this piece of evidence, and ends up describing Chrysippus' use of the word ἔμφυτος as a ‘temporary aberration’: all he meant was that moral notions come to us without training, not that they pre-exist experience. As Watson ((n.34) 268–9) has pointed out, this state of affairs is unsatisfactory. In effect, Sandbach's case rests in part on discarding one piece of evidence, which is exactly what he accuses Bonhöffer of doing with Aëtius. Anyway, why should we assume that Plutarch, or worse still Chrysippus, is at fault?

39. A note of warning should be sounded about using this text as a checklist of prolepses, as a number of scholars have done. What is under discussion is the formation of conceptions (νοουμένων), which the Loeb translation happily renders as ‘general notions’. This is fine for the first items on the list, notions of sensible things, but the next example, thinking of Socrates, is obviously concerned with a particular, so the term νοουμένων should be taken just as conceptions, of which general conceptions are a species. Another reason this chapter cannot be about prolepses alone is that most of the examples – pygmies, handless men and the like – could hardly be criteria of truth, as prolepses are meant to be. So the chapter is about conceptions in general, some of which are scientific, such as the example of analogy (the centre of the earth).

When we come to the formation of the conceptions good and just, the qualification τι, which has been missing from all the other examples, is added. This qualification could be interpreted in two ways. Either he means that a particular thing is thought to be good, in which case this example joins the Socrates case, or he is talking about the formation of a hazy general notion. Though these may seem very different interpretations, I do not think that it matters too much which one we take, because involved in the formation of a notion of a good thing will be some sort of general notion of goodness, which is not very developed. For a discussion of the qualification τι see Bonhöffer (n.34) 200 and Sandbach, (n.34) 33–4.

40. Bonhöffer (n.34) 204.

41. Sandbach ((n.34) 29) recognizes the importance of this task and originally attempted to explain φυσικῶς by reference to Cicero's De Finibus 3.33–4, where it is claimed that we gain an understanding of the good by a process of analogy – the mind ascends by inference from things in accordance with nature, and so reaches the good: but, in the end, it does so not by comparison, but recognizes the good ‘propria vi sua’. Sandbach translates this as ‘by the force of its own nature’, saying that this explains the adverb in Diogenes Laertius 53. But, as he himself pointed out in the postscript to his article, this explanation is unsatisfactory: Cicero is not making a point about the naturalness of our learning, but about the nature of what is learnt ((n.34) 33). Another problem with Sandbach's explanation is that it does not show how the notion of justice is formed: he appeals to no text parallel to the one by Cicero.

42. Pohlenz, , Grundfragen 96Google Scholar, see also 88; Die Stoa 58 (see n.34).

43. Luschnat, O., Philologus (1958) 191–2Google Scholar.

44. A third problem is one that has been pointed out by Sandbach who, in the post-script to his article ‘Ennoia and prolepsis’ (n.34), criticizes Pohlenz's case because he only tries to show how the concept of good comes naturally and neglects to do the same for justice which D.L. 7.53 also mentions.

This problem can be alleviated somewhat by a passage from Porphyry's De Abstinentia in which we are told that the followers of Zeno made οἰκείωσις the beginning of justice. Unfortunately, however, this rather vague statement is not further explained. As Pembroke, S. G. points out (‘Oikeiosis’ in Long, A. A. (ed.), (n.34) 114–49Google Scholar), it is not clear what οἰκείωσις means here: is it οἰκείωσις to oneself or to one's children, or what? Is it a peculiarly human form of οἰκείωσις or does it apply to all animals? Because Porphyry is arguing for vegetarianism, he himself holds that justice and injustice apply to animals as well as men, but he could be perverting a Stoic doctrine for his own ends. In the next chapter. Porphyry launches into a bitterly sarcastic attack on Chrysippus, so his representations of Stoic doctrine may be none too careful. Also, we do not know who exactly held the doctrine – perhaps not Zeno himself, though it is reasonable to think that Chrysippus can be included among his followers.

Nevertheless, this passage does suggest a case of an innate disposition giving rise to justice. What exactly that innate disposition is does not affect our issue which is whether or not the formation of concepts is entirely dependent upon sense experience. On this see also the Anonymous commentator on Plato's Theaetetus col. 5–6 where he discusses the view – presumbaly Stoic – that we have οἰκείωσις towards members of the same species, and that justice could be derived from this οἰκείωσις. Also relevant is perhaps Seneca, De Ira 1.5. §2: ‘Homo in adiutorium mutuum genitus est …’.

45. T. Irwin (n.34) 208–9 with n.5.

46. Galen, De Plac. Hip. et Plat. 5.5.1–15. See also SVF III. 230–6.

47. . (Diogenes Laërtius 7.89)

48. See s.v. esp. 3, 4.

49. Diogenes Laërtius 7.76.

50. . (Stob. Ecl. 2.65.7 = SVF 1.566). I am assuming that both halves of the sentence ( and ) are to be attributed to Cleanthes: without the first half, the second would be Delphic in the extreme.

51. .

One scholar, Grilli, A. (Il problema della vita contemplativa (1953) 116 n.1)Google Scholar, has attempted to argue that neither from this evidence nor from Stob. 2.65.7 nor from D. L. 7.89 can we claim that the early Stoa used the word ἀφορμὴ to mean ‘starting-point’ or ‘inclination’; this use only seeped into the Stoa with Panaetius who advocated a life lived ‘in accordance with the ἀφορμαὶ given to us by Nature’ (fr. 96 van Straaten). In the case of the Cleanthes citation Grilli adopts a punctuation that attributes only the words to Cleanthes; as I have argued in the previous note, this is implausible. As for Diogenes Laërtius 7.89, Grilli accepts that the thought expessed is Chrysippus' (because of Galen's testimony in Plac. Hip. et Plat. 5.5) but he still denies that the term ἀφορμή can be attributed to anyone earlier than Panaetius. His argument is that as both Stobaeus and Diogenes are influenced by Panaetius (via Hecato) their use of the word ἀφορμή could easily be a product of this influence. But this is made implausible by the fact that, as he himself admits, ‘starting-point’ was the normal meaning of ἀφορμή, and was no less available to Zeno, Cleanthes, or Chrysippus, than it was to Panaetius.

52. The evidence for this comes from Plut., Comm. not. 1057aGoogle Scholar.

53. My example is culled from Diogenes Laërtius 7.108.

54. See above p. 133.

55. This passage is quoted by Barnes ((n.8) 207) who points out that the innatism espoused here seems to have by-passed Berkeley's best-known works.

56. See above Varieties of innatism, 125–35.

57. See above pp. 135–6.

58. Nature's providence in endowing us with sense-perception is implied by Sext. Adv. math. 7.259–60 and Cic. Ac. 1.42. An interesting parallel can be found in Sext. Adv. math. 8.276, where the Stoics are said to have argued that the existence of signs can be inferred from the fact that the notion of the sign is part of the natural constitution of man.

59. Cic. De Nat. Deorum 1.39.

60. Epictetus talks of God-given prolepses in Diss. 3.5.8. Interestingly enough, in the same passage he talks of these prolepses as ἀφορμαὶ. But for Epictetus, ἀφορμή, when used in these contexts, has a rather general meaning: it applies not only to prolepses but also to perception (see also 4.10.14).

61. SVF III. 229–36.

62. D.L. 7.87 and 143.

63. Stob. Ecl. 2.75.11–76, 8; D.L. 7.87–9.

64. I am very grateful to Margaret Atkins, Myles Burnyeat, Malcolm Schofield, David Sedley and Robert Wardy for their comments on various drafts of this paper. I have also benefited from a characteristically spikey interrogation by the Cambridge B Club and, more recently, from the editorial hustling of M. M. Mackenzie.