¹ SeeAn Essay Concerning Human Understanding, Bk. I.

² The evidence for belief in innate principles in the seventeenth century has been excellently documented by Yolton, J. W. in John Locke and the Way of Ideas (Oxford, 1968), 30–48.Google ScholarThe dramatis personae include clerics from Bishop Stillingfleet to the Cambridge Platonists More, Culverwel and Cudworth, who embraced more subtle theories of innatism. A good example of the association between innatism and religion is More's Antidote Against Atheism, reproduced in Patrides, C. A. (ed.), The Cambridge Platonists (Edward Arnold, 1969). See especially pp. 218ff.Google Scholar

³ Descartes, Notes Directed against a Certain Programme, tr. Haldane, E. S. & Ross, G. R. T. (Cambridge, 1911), 442–3.Google ScholarDescartes also used the term ‘innate’ in a more specialized sense to describe ideas that were not adventitious or fictitious. In the broader sense, however, all ideas are innate, even those of primary qualities. For a discussion of this issue see Adams, R. M., in Stich, S. P. (ed.), Innate Ideas (California, 1979), 77–8.Google Scholar

⁴ Cudworth, R., A Treatise concerning Eternal and Immutable Morality (1731), 148–9.Google Scholar

⁵ Leibniz, in his attack on Locke's polemic, re-asserts the innateness of both ideas and principles (practical and speculative). He points out that although speculative maxims may not be explicitly known to everyone, they are innate in so far as they would be assented to as soon as heard. He adds that they are innate also because they are in us ‘potentially’, suppressed as in enthymemes (see Remnant, P. and Bennett, J., G. W. Leibniz: New Essays on Human Understanding [Cambridge, 1982], 76). He talks on p. 84 of the mind relying on certain principles constantly, these serving as ‘the inner core and mortar of our thoughts’. Some innate principles are in fact the necessary conditions of thought, and in using them we know them ‘fundamentally’. He also espouses the innateness of ideas by making several intellectual ideas - being, unity, substance, duration, change, action - innate to us (see p. 51).Google Scholar

⁶ Locke, Essay, Bk. I.3.§2, and Yolton, op. cit. (n. 2), 39–41.

⁷ Meno 81d4–5.

⁸ Phaedo 74e ff.

⁹ It is fascinating to note, however, that one person who dissociates himself from this ‘Kantian’ view of anamnesis is Kant himself. In the Critique of Pure Reason A 313/B 370 he talks of the laborious process of recollection and identifies it with philosophy. Elsewhere, (Reflexionen zur Metaphysik, Nr. 6050, in Kant's gesammelte Schriften [Berlin and Leipzig, 1928], xviii (5). 434–5), he says that we recollect the ideas only with difficulty. It is thus a very recondite affair and Kant's reason for thinking this was that he saw the ideas not as categories or concepts of pure reason, which combine with sensible intuitions to make experience possible, but as far surpassing these and constituting intellectual intuitions of things as they are in themselves, which is a very different matter. In fact, Kant interpreted anamnesis as amounting to no less than a participation in the Divine Intellect.

¹⁰ For the origin of this fragment, see Plutarch's Moralia (Loeb), ed. F. H. Sandbach, xv. 388–9; Westerink, L. G., The Greek Commentators on Plato's Phaedo (Amsterdam, 1976). ii. 166.Google Scholar

¹¹ VII, 239, 4.

¹² See SVF II 83.

¹³ There was available an empiricist account of the emergence of knowledge from sensation that is mentioned in the Phaedo at 96b, and it has been attributed to Alcmaeon (24 A 11 DK). Its stage-by-stage account - sensation, memory, opinion, knowledge - was echoed by Aristotle (An. Po. B19, 100a3 ff. and Met. Al, 980a27 ff.) and the Stoics (SVF II 83). On my story, Plato's quarrel with Alcmaeon would have been with the final transition from opinion to knowledge, not that from sensation through to opinion.

¹⁴ Notice Plato's lack of interest in Alcmaeon's theory in the Phaedo passage.

¹⁵ Bluck, R. S., Plato's Meno (Cambridge, 1961), 9–10, for instance, argues against including experiences of a previous life into the matter of recollection.Google ScholarVlastos, G., ‘Anamnesis in the Meno’, Dialogue 4 (1965), 143ff., excludes empirical knowledge from Plato's programme. For an extremely severe restriction on the meaning of , seeGoogle ScholarNehamas, A., ‘Meno's Paradox and Socrates as a Teacher’, Oxford Studies in Ancient Philosophy 3 (1985), 1–30: on his view, the slave boy does not recollect at all, and would only do so if he attained knowledge, not just true opinion.Google Scholar

¹⁶ It is now time to unmask some of the adherents of K. The most articulate versions come from Gulley, N., ‘Plato's Theory of Recollection’, CQ 4 (1954), 197ff., and Plato's Theory of Knowledge (London, 1962), 31ff;CrossRefGoogle ScholarJ. L. Ackrill, ‘Anamnesis in the Phaedo: Remarks on 73c–75c’, Exegesis and Argument, Phronesis, suppl., vol. I(1973), 177–97, especially p. 192; andGoogle ScholarBostock, David, Plato's Phaedo (Oxford, 1986), 66ff. See alsoGoogle ScholarCornford, F. M., Plato's Theory of Knowledge (London, 1935), 108. I have said that K interprets anamnesis as explaining concept formation, but just what is meant by ‘concept formation’ varies depending on how much of our conceptual apparatus different versions of K think is to be explained by anamnesis. The most careful claims are made by Bostock, who argues that recollection accounts for our ordinary and everyday grasp of meanings of those words, such as ‘equal’, of which there are no paradigm examples provided by sense-perception; it should also be pointed out that Bostock gives a more linguistic slant to the issue than other commentators by talking about ‘meanings of terms’ rather than ‘concepts’. At the other extreme, Gulley (CQ 4 [1954], 198 n. 2) thinks that the form of the argument of the Phaedo ‘almost’ implies an unlimited range of forms. This approach is more typical of commentators on the Phaedrus where Plato is thought to be talking of the use of universals in language without implying any restriction whatever (see below, n. 44). Despite the differences between versions of K, I shall mount my attack on them as one body, because I am refuting interpretations which require anamnesis to explain any of our ordinary conceptual apparatus, however limited the range of concepts concerned. Another point that should be clarified is that it is essentially concept formation rather than concept use that anamnesis is meant to explain on any version of K. As a theory of learning, it attempts to show how we came to form the concepts that we use, or how we came to understand the meaning of certain terms; inasmuch as the formation of concepts is a necessary condition for their use it contributes to an account of concept use. But I take it that adherents of K are not concerned with the issue of concept use in the sense of how we apply the concepts which we have already formed correctly to objects in the world.Google Scholar

¹⁷ One might think that at 74a9–b3 Socrates is drawing attention to the fact that we all have a humdrum knowledge of equality, i.e. we know what ‘equal’ means. But the contrast made in 74a9–12 jars with this: ‘We say that there is something equal, I don't mean a stone to a stone, or a stick to a stick…" The contrast is between our acknowledging the equality of the form and the equality of the sticks and stones; in 74alO there is an ellipsis, and that part of the sentence, if filled out, would run: ‘…I don't mean that we say that a stick is equal to a stick…’. Now it is precisely in such statements as, ‘this stick is equal to that one’ that our humdrum grasp of the concepts and meanings is manifested, yet Socrates dismisses these as irrelevant to his argument. So these lines cannot be used to show that recollection is to be invoked to solve mysteries surrounding our ordinary grasp of ‘equal’.

¹⁸ Gulley (CQ 4 [1954] 197–8) paints himself into a corner by saying ‘What appears to be envisaged is an immediate transition from the sensible to the intelligible world, the argument relying on a contrast between sensation and a conceptual level of apprehension. Plato is apparently saying that the fact that we attain this conceptual level in describing what is given in sense-experience constitutes recollection of forms.’ Gulley goes on to consider the claim at 74c that in being reminded of the form we are gaining knowledge of it, which he takes to imply that the immediate transition referred to above is one from mere sensation to philosophical knowledge of the forms, but, as he goes on to say: ‘Plato never assumes elsewhere that the fact that we employ concepts to describe what we see is either a mark of knowledge of the forms, or in itself any reliable pointer.’

¹⁹ Ackrill, op. cit. (n. 16), 183: ‘There may be a lurking problem for Plato's programme. For if recollection is to explain concept-formation, can a pre-condition for reminding be recognition or something akin to it?’

²⁰ 65dl 1 ff. and 82d9 ff. It is particularly at 75e3 ff. that Plato seems to contradict these other passages.

²¹ For a statement of this problem, see Gallop, D., Plato's Phaedo (Oxford, 1975), 121.Google Scholar

²² Socrates is perhaps referring to the necessary role of sense-perception at 83a6–7.

²³ This has been discussed by Irwin, T., Plato's Moral Theory (Oxford, 1977), 315 n. 13.Google Scholar

²⁴ The absurdity of the claim that all men compare equal particulars unfavourably with the equal itself undermines Bostock's (op. cit. n. 16) first argument against an interpretation which limits those who recollect to philosophers. On p. 67 he says that if ‘we’ is limited to Platonists, then one of the leading premises of the whole argument would be unacceptable to most people, whereas on his view, the argument is supported by a sensible and uncontroversial point that we understand the meaning of the word ‘equal’ (pp. 70, 72). But when he discusses our humdrum cognitive achievements that Plato is allegedly interested in, he confines himself to 74a–b, and treats 74d4 ff. in a separate section, dealing with it primarily as a metaphysical thesis about the contrast between forms and particulars; he does not explain what humdrum cognitive achievement is being referred to there.

²⁵ Gallop (op. cit. (n. 21), 120) assumes that if the ‘we’ who actually recollect is limited to a few, then the entire theory is likewise thus limited, as if Plato could not generalize the results of his argument: ‘Recollection is not a philosopher's privilege, but, as in the Meno, is possible for human beings generally.’ But how does Plato in the Meno argue that recollection is possible for everyone? He takes one slave boy, shows him actually recollecting, and then assumes that if he can recollect, so can everyone. He has no qualms about generalizing from one case, and hardly expects us to respond ‘What a clever and interesting slave, I wonder if anyone else can do this.’ Exactly the same strategy is followed in the Phaedo: in the Meno Socrates' argument depended upon the true opinions that the slave boy acquired during the interview and the certainty that these had not already been learnt in this lifetime; the Phaedo parallels this with the Platonists' knowledge of the equal and the certainty that this was not derived purely from perception. In both dialogues these premises are used jointly to prove recollection for one or a small number of cases, from which Socrates then makes a tacit generalization. (Of course, if, in the Meno, Socrates made use of an argument to the effect that since someone of such humble origins can recollect so can everyone, then there would be a considerable difference in strategy between the two dialogues. But nowhere in the Meno does he appeal to such considerations. What he does make use of is the fact that because the slave boy has always been in Meno's household, they know that he cannot have already learnt geometry (83e3–5). It is not so much that he is a slave boy but that he is Meno's slave boy that matters, as it is this that ensures that the experiment is a controlled one.)

²⁶ See, for example, 83d4 ff.

²⁷ I am following Hackforth here (Hackforth, R., Plato's Phaedo [Cambridge, 1955], 76). Gallop (op. cit. (n. 21), 133) objects to this view because ‘moral and mathematical forms are expressly said to be on a par’ (75clO–d2), but the only way in which all the forms are put on a par at 75c 10 is by being objects of dialectical argument, not of knowledge.Google Scholar

²⁸ Meno 85c6–7.

²⁹ I am translating here as ‘being reminded’ rather than ‘are reminded’ as the latter would create a needless contradiction with an earlier passage. If Socrates and Simmias are now concluding that all men recollect, they are contradicting what they have just decided, viz. that not all men know the forms: at 75e5–6 it has been stated that to recollect is to regain knowledge, so if all men recollect, all men know, and this is just what has been denied.

³⁰ Interestingly enough, Hackforth (op. cit. (n. 27), 72) translates 76c4 as ‘Can they then recollect what they once learnt?’.

³¹ In what follows I shall be using the Phaedrus as a testing ground for D and K in much the same way as I used the other two dialogues. The whole passage is, of course, a myth and so it may be objected that the text requires different treatment and cannot be used ‘straightforwardly’ as evidence for a particular interpretation of anamnesis. I emphasize, however, that my specific purpose is to reject K, which means arguing against those who have used the Phaedrus myth as evidence. I shall therefore be giving K the benefit of the doubt on this issue, and shall be claiming in effect that if one were to use the myth in this way, then it would be D, not K, that emerged as the most convincing interpretation of anamnesis.

³² Hackforth, R., Plato's Phaedrus (Cambridge, 1952).Google Scholar

³³ Thompson, W. D., The Phaedrus of Plato (London, 1868), 55, for instance, says ‘It is a law of human understanding that it can only act by way of generic notions… sensibles are per se unintelligible’. One scholar who does not follow this line, remarkably enough, is Gulley, who, despite his reading of the Phaedo, takes the Phaedrus passage to refer only to the philosopher: ‘Thus whereas the Phaedo argued that the presence of [the possibility of reasoning from sensation to conceptual apprehension] was explicable as recollection of forms, the Phaedrus can be consistently interpreted only as a description of a process of inductive reasoning from a number of instances of sense-perception.’ (CQ 4 [1954], 201). See also f. Irwin, op. cit. (n. 23), 173.Google Scholar

³⁴ 249c7.

³⁵ 251a ff.

³⁶ 250e4.

³⁷ Though see n. 14 above.

³⁸ Heindorf inserts Τò.

³⁹ Badham, followed by Thompson, op. cit. (n. 33), 55.

⁴⁰ Plato's Phaedrus 86 n. 1.

⁴¹ Verdenius, W. J., ‘Notes on Plato's Phaedrus’, Mnemosyne (Series IV), 8 (1955), 280.Google Scholar

⁴² I am resisting the temptation to read ‘collection’ into this passage (contra Gulley, op. cit. (n. 33), 200–1). This new tum in Plato's dialectic has yet to be announced (265d). Furthermore, is not part of Plato's terminology for this procedure - it is not in fact used anywhere else in his works.

⁴³ δεî is used of our epistemological duty to inquire after what we do not know at Meno 86b7.

⁴⁴ A further problem that some versions of K would have to tackle is that of the range of forms. If the Phaedrus is meant to explain the use of universals in language, then we do not need only forms produced by an argument from opposites, but also forms corresponding to all universal terms - hair, mud and dirt included. One advantage of embracing D is that we avoid tying anamnesis down to this particular crux. We should note that the only parallel text that K could appeal to for its range of forms in the Phaedrus is the notorious, but ambiguous, sentence at Republic 596a6–8: Smith, J. A. (‘General Relative Clauses in Greek’, CR 31 [1917], 69–71)Google Scholarpoints out two linguistic difficulties involved in taking this sentence as advocating forms of all universal terms. In the first place, the relative clause that ends the sentence is very unlikely to be a general relative clause: this would have a subjunctive or an optative, or, if an indicative, őοοίς for οΐς (see Goodwin, W. W., Syntax of Greek Moods and Tenses [London, 1889], §§ 532, 534). Second, the is ambiguous, meaning particulars having the same name as each other or having the same name as the form. Smith suggests that would be a much more natural expression for the former possibility. More work needs to be done on these problems, but as they stand they are sufficient to cast considerable doubt upon the usual claims made for this sentence.Google Scholar

⁴⁵ See above, pp. 346f.

⁴⁶ The Cambridge Platonist Cudworth (op. cit. (n. 2), 100), in the course of arguing for innate ideas, praises Theaet. 184ff. as an accurate assessment of the limitations of sense-perception.

⁴⁷ Plato's volte-face on this issue has been discussed by Burnyeat, M. F. in ‘Plato on the Grammar of Perceiving’, CQ 26 (1976), 29–52.CrossRefGoogle Scholar

⁴⁸ Phaedo 65a9–66alO, 79a, 83a–b.

⁴⁹ 83bl–2:…

⁵⁰ Diogenes Laertius 10.33; Cic. De Natura Deorum 1.45.

⁵¹ For the Stoics on prolëpsis, see SVFII 83. There is a dispute as to whether the Stoics believed in innate moral prolepseis, and whether Epictetus is really departing from the earlier Stoa on this point, but I shall not attempt to tackle that issue here.

⁵² The Anonymous Commentator on the Theaetetus describes recollection as a process of articulating ‘common notions’ (47.1–18.11, 53).

⁵³ See Patrides, op. cit. (n. 2), 132 n. 21, and Yolton, op. cit. (n. 2), 36, who points out that Leibniz included the Stoics in the innatist tradition.

⁵⁴ Plutarch, Comm. Not. 1060a.

⁵⁵ I have profited greatly from discussions with Gail Fine and Geoffrey Lloyd, and from the comments of the editors, as well as from the reserved agreement of Robert Wardy and the unreserved disagreement of M. M. Mackenzie. My greatest debt, however, is to the maieutic powers of Myles Burnyeat.