Idealism and Greek Philosophy: What Natorp Saw and Burnyeat Missed

In his landmark article “Idealism and Greek Philosophy: What Descartes Saw and Berkeley Missed,” Myles Burnyeat purports to establish that:

Idealism, whether we mean by that Berkeley’s own doctrine that esse est percipi or a more vaguely conceived thesis to the effect that everything is in some substantial sense mental or spiritual, is one of the very few major philosophical positions which did not receive its first formulation in antiquity.^[1]

To this effect, he begins by criticizing Berkeley’s interpretation of the first part of the Theaetetus and shows that the position there criticized (and not endorsed, as Berkeley seems to believe) by Socrates is not a reduction of esse to percipi, but rather the thesis of a dependency of the perceived object on the perceiving subject and vice versa, a thesis which does not at all conflate “the two realms of mind and matter” into one. He then extends his investigation to other ancient authors (Parmenides, Gorgias, Metrodorus of Chios, Aristotle, the Neoplatonists) whom he dismisses as potential candidates to the title of idealists as he has characterized it. In so doing he analyzes the famous argument of Parmenides 132b–c “as indicative of the deep hold of realism on Greek Philosophy.”^[2] The second half of his paper proposes an ambitious explanation of this hold, namely that idealism would only become a tempting doctrine after Descartes introduced for the first time (1) the idea that truth can be obtained without going outside subjective experience, (2) the claim to knowledge of one’s own subjective states and (3) the recognition of one’s own body as a part of the external world – all points which Burnyeat argues cannot be found in ancient thought, including the Sceptics and Augustine. From this point of view, the very possibility of idealism would be the sign of a major difference between ancient and modern philosophy.

As this brief summary shows, Burnyeat’s rich paper is extremely wide-ranging and thought-provoking. It had notably a great impact on research on Scepticism by denying that the ancient Sceptics posed the problem of the knowledge of the external world in its full generality, i.e., as concerning not only the nature of things, but their very existence. Burnyeat’s interpretation has stimulated many discussions on this and related points and remains essential reading for anyone interested in these issues.^[3] However, my focus here is strictly on the main thesis of the paper, i.e., that idealism was not, or even could not have been, a philosophical position held by anyone in antiquity.

Let me start by remarking that for someone who has not been educated primarily in the analytic tradition, Burnyeat’s conception of idealism, which probably stems from G. E. Moore and B. Russell’s reaction to a certain trend in British and American philosophy, is quite a bit of surprise.^[4] What Burnyeat calls ‘idealism’ is indeed for the most part a version of Berkeley’s immaterialism. Now Berkeley himself never labelled his doctrine ‘idealism,’ although it was thus labelled by some of its critics. As far as we know, the first occurrence of the word ‘idealist’ can be found in a 1702 text by Leibniz, who precisely applies it to an ancient author, namely Plato (as well, it is true, as to the Cartesians).^[5] In this text, Leibniz seems to imply that idealism is the view according to which “everything happens in the soul as if there was no body,” whereas materialism, which he ascribes to Epicurus and Hobbes, would be the view according to which “everything happens in the body as if there was no soul.”^[6] Such a characterization, which was later systematized by Christian Wolff and Alexander Gottlieb Baumgarten,^[7] fits well with Burnyeat’s formula according to which:

Just as materialism is the monism which asserts that ultimately nothing exists or is real but matter and material things, so idealism is the monism which claims that ultimately all there is is mind and the contents of mind.^[8]

But we should keep in mind that this is not the conception of idealism advocated by most of the major philosophers who claimed the label ‘idealism’ for themselves. This is especially the case for the first self-avowed idealist, namely Kant, who has repeatedly stressed the difference between his own ‘transcendental idealism’ and what he calls the ‘material idealism,’ i.e., “the theory that declares the existence of objects in space outside us to be either merely doubtful and indemonstrable, or else false and impossible,” the first position being identified with Descartes’ “problematic idealism,” the second one with Berkeley’s “dogmatic idealism.”^[9] On the contrary, the Critique of Pure Reason claims to offer the only possible rigorous proof of the objective reality of external spatial objects.^[10] By contrast, Kant’s own ‘transcendental idealism’ is the doctrine that appearances “are all together to be regarded as mere representations and not as things in themselves, and accordingly that space and time are only sensible forms of our intuition, but not determinations given for themselves or conditions of objects as things in themselves.”^[11] It is from this background that the later ‘German Idealists’ should be approached, although they developed various versions of idealism, sometimes very different from each other. And it is also to this tradition that Neo-Kantianism and Husserl’s transcendental phenomenology should be associated, and certainly not to any version of Berkeley’s philosophy.^[12]

Of course, one can reply that Burnyeat’s target is strictly idealism as it was (and to a large extent still is) understood in the Anglophone world, namely Berkeley’s immaterialism. But even with this much more limited agenda, it is not clear that Burnyeat succeeds. On the one hand, it can be doubted that his interpretation of Berkeley as rejecting the kind of realism he finds in Greek thought is correct;^[13] and on the other, his interpretation of the ancient material is not immune to objections. For example, as Richard Sorabji has shown, some version of immaterialism can actually be found in Gregory of Nyssa, and it was later developed in a most impressive manner by the great Latin philosopher John Scottus Eriugena.^[14] Of course, this is a relatively marginal trend, but it speaks against Burnyeat’s bold claim according to which ‘idealism’ as he defines it could not have arisen before Descartes. More generally, it has been argued that Burnyeat’s rather superficial take on Neoplatonism does not resist a closer examination of the texts.^[15] Similarly, Burnyeat’s assertion that the special cognitive status of one’s own subjective states had not been recognized before Descartes is doubtful, and many texts have been adduced (or reinterpreted) against it.^[16] To add but one passage which does not seem to have been mentioned in this context, when Socrates introduces the theme of false pleasures in the Philebus, he encounters fierce resistance on the part of Protarchus, who claims that “we all assume” that “there is no one who, either in a dream or awake, either in madness or any other delusion, sometimes believes he is enjoying himself, while in reality he is not doing so, or believes he is in pain while he is not” (36e, trans. Frede 1993), and he suggests that the same holds for fears and expectations (36c). Of course, Socrates will try to prove otherwise, but what Protarchus ascribes to “all of us” is strikingly similar to what we can find in Descartes on this point.^[17]

More fundamentally, is it legitimate to restrict idealism to versions of Berkeley’s immaterialism? As I have already suggested, idealism has a long and complex history and can refer to a variety of things. In the introduction of their collection entitled Eriugena, Berkeley, and the Idealist Tradition, Stephen Gersh and Dermot Moran distinguish four kinds of idealism:

First, Platonic or Neoplatonic idealism; second, Berkeleian immaterialism or mind-dependence of physical objects; third, Kantian and neo-Kantian transcendental idealism, with its a priori correlation of objectivity with subjectivity (e. g., in Edmund Husserl) and its claim that space and time are conditions of sensibility rather than intrinsic properties of mind-external objects (Kant); and, finally, Hegelian absolute idealism, with its conception of the cosmos as the self-evolution and coming to self-awareness of absolute spirit (versions of which can also be found in Bradley and the British Idealists generally).^[18]

Now of course, one can use historical categories in a variety of ways; but it is rather paradoxical to pick between these meanings, as the only one which would be “of any historical use at all,”^[19] one which was actually never claimed for itself by the doctrine in question, namely immaterialism, while apparently denying it to others who did make such claims; and it becomes frankly problematic when it can be shown that the understanding of this doctrine, ultimately stemming from G. E. Moore, is confused in a variety of ways.^[20] This is not merely a question of vocabulary. For it can be argued that this dominant conception of idealism in the Anglophone world has determined the reception of all other philosophers which could be labelled ‘idealists’ in various ways and hindered access to their proper understanding. This is especially the case of Kant and the great German idealists, who are still commonly understood through Berkeleian lenses.^[21] And the same can be said of all attempts at ‘idealist’ interpretations of ancient philosophy, which are regularly disqualified on the basis of a (historically problematic) conception of idealism that they do not share at all. We shall soon see an example of this.

For this is exactly what happens in Burnyeat’s article, although this does not seem to have been noticed previously. Indeed, contrary to what is usually assumed,^[22] it is not the case that Burnyeat completely reduces idealism to immaterialism. At one point in his article, he introduces another kind of idealism:

The idealism which has been most influential in modern times is the idealism which asserts, in one version or another, that the world is essentially structured by the categories of our thought.^[23]

I take this as an attempt at characterizing a Kantian type of “transcendental idealism.”^[24] Since Burnyeat earlier wrote that the only historically legitimate use of the label ‘idealism’ indicates a form of monism, i.e., as we have seen, “the monism which claims that ultimately all there is is mind and the contents of mind,” one has to conclude that according to Burnyeat, such a variety of idealism should also be conceived as a monism of this kind.^[25] As far as Kant is concerned, this is simply wrong;^[26] but this shows quite clearly that Burnyeat cannot but envisage critical philosophy from the viewpoint of Berkeleian immaterialism. In the remainder of this paper, I would like to show how this perspective prevents him from detecting the presence of another kind of idealism in ancient thought, one that is more akin to the genuine Kantian version.

Despite acknowledging that the idealism just defined is “the idealism which has been most influential in modern times,” Burnyeat has surprisingly very little to say about it. He is content with analyzing the argument of Plato’s Parmenides 132b–c, which in his eyes is a direct refutation of this kind of idealism.

Let me first recall this well-known, but not so often studied, argument. In order to escape the Largeness regress, Socrates submits that each form or idea is a kind of “thought” (νόημα) that occurs only “in souls” (ἐν ψυχαῖς). According to most readings at least from Proclus onwards, Parmenides offers two distinct refutations of this suggestion.^[27] The first (132b7-132c8) purports to show that a thought must have an object, and that it is this object which corresponds to the idea rather than the thought itself. The second one (132c9–12) sets before Socrates a dilemma: if each idea is a thought and each thing is what it is insofar as it participates in the corresponding idea, then each thing will be composed of thoughts (ἐκ νοημάτων … εἶναι), and either all things will think or there will be thoughts that are unthinking (ἀνόητα).

It is not immediately clear why this argument should be directed specifically against a Kantian type of idealism. Most interpreters construe it instead as a refutation of conceptualism, the position according to which general terms correspond to concepts (ennoiai) or mental dispositions. Burnyeat criticizes this reading and argues that the target of the argument is not “conceptualism about universals,” but rather “the suggestion that it is thought which explains the way things are in the world,”^[28] that “one might seek to explain the nature of the world by reference to the categories of thought.”^[29] His main argument is that only such a reading can make sense of the “final twist in Parmenides’ refutation,” i.e., the idea that “[i]f Forms are thoughts, […] then the things which participate in them will themselves consist of thoughts, so that either they all think, or (an alternative absurdity) they are thoughts which nevertheless do not think.”^[30]

We can substantiate this interpretation a little. Throughout this text, Socrates and Parmenides use the word νόημα. This is a very rare word in Plato. According to Brandwood, apart from this passage, it only occurs in quotations (of Simonides at Meno 95e5; of Parmenides at Sophist 237a9 and 258d3), in Agathon’s speech in the Symposium (197e5) and at Statesman 260d8.^[31] By contrast, it is relatively frequent in Parmenides’ poem (see fr. 7.2, 8.34, 8.50, 16.4), and its presence here may be a deliberate allusion to this.^[32] In any case, both in Parmenides’s poem and here, it quite clearly means an act of thinking rather than its result: it is a thought that thinks (cf. τὸ νόημα … νοεῖ, 132c3), not a thought that is thought.^[33] Now what is this act exactly? Given the previous argument (131e8–132b2), it seems to be the act by which the soul looks at a certain single character (ἰδέα) common to a plurality of things – the common root between ἰδέα (or εἶδος, for that matter) and the verbal forms used in this context (ἰδόντι, 132a3; ἴδῃς, 132a7) offering a direct stimulus in that direction. If that is right, Burnyeat’s interpretation of the meaning of the argument seems warranted: rather than a form of ‘conceptualism’ treating forms as the results or products of the thinking process, it is their identification with the thinking process itself, or more precisely with the categories according to which it occurs, that seems to be the target.

Now what are Parmenides’ objections against this version of ‘idealism’? The second one has already been mentioned: given Socrates’ assumption that sensible things are what they are by participation in the Forms, everything would partake in thought processes, and either everything would be thinking or there would be thinking processes which would not think, and both options are considered absurd (132b9–12). Contrary to what David Keyt has claimed, there is no fallacy here, no more than in arguing that “since Forms are intelligible entities (νοητά) and things share in Forms, each thing is an intelligible entity”^[34]; for it is the case that participation in Forms should give to their participants a share in the “formal” or “ideal” attributes which Forms possess simply by being Forms: being, stability, determinacy, and also intelligibility. So if Forms are thinking processes, their participants should also be so, at least in part.

Anyway, for Burnyeat’s purpose, the important objection is rather the first one. It develops in four steps:

1. each one thought (i.e., thinking act) must be of something (τινός) (132b7–12);

2. this something is (ὄντος);

3. this something that is is some single character (μίαν τινά … ἰδέαν) which the thought thinks over all the cases (132c3–5);

4. it is this one thing that is thought (τοῦτο τὸ νοούμενον ἕν) which is the Form (εἶδος), being always the same over all the cases (132c6–8).

Burnyeat summarizes this argument by the following formula: “Thought must have an object, a really existing object independent of itself, and that object will be the Form”; and he comments: “What is remarkable about this argument is its swiftness and the brutality of its realism.”^[35]

Indeed, if this is the meaning of the argument, one cannot help finding it disappointing, to say the least. Of course thinking of an object is not sufficient to make it “really existing independently of thought.” Is Plato really making such a crude fallacy?^[36] It is clear that the argument purports to show that the Form is not the act of thinking itself, but its object; not τὸ νόημα (in this active sense) but τὸ νοούμενον. But all that is said about this object is that it must be, and be one. Why interpret this being as that of “a really existing object independent of thought”? First, why “independent of thought”? The fact that it is νοούμενον does not clearly speak in favour of this view, to say the least. But secondly, why interpret “being” as “real existence”? And what does it mean?

Here I think we have reached the heart of the matter. For one should take note of the fact that the opposite of idealism has now turned out to be, not materialism, as at the beginning of Burnyeat’s paper, but ‘realism,’ a very vague and never defined label (at least under Burnyeat’s pen). Moreover, Burnyeat adds a little later that the realism he ascribes to Plato and Greek philosophy in general is not an explicit philosophical thesis, which, like idealism, would only have become possible after the question “Is there anything other than mind?” had been brought to the center of philosophical attention, but “an unquestioned, unquestioning assumption of realism,” which does not make the matter clearer.^[37] In any case, I would like to stress that it is far from obvious that there is any opposition between idealism and realism. As Bracken has suggested, “in the sense in which [Burnyeat] finds Greek thought to be realist, Descartes and Berkeley are realists,” too^[38]; and it is worth remembering that Kant insisted that his “transcendental idealism” is an “empirical realism,” in the sense that it maintains the existence of external things in the space – although it adds that space itself is “in us.”^[39] Actually, when Burnyeat initially opposed materialism and idealism, he himself implicitly characterized both as forms of ‘realism’: materialism as a realism of matter and material things, and idealism as a realism of mind and the contents of mind.^[40] The shift seems to occur when he first suggests that Greek philosophy is dominated by this “unquestioned, unquestioning assumption of realism,” which he develops by saying that for Greek philosophers, “[t]here is a reality of some sort confronting us; we are in touch with something, even if this something, reality, is not at all what we think it to be.”^[41] Here, ‘reality’ has turned out to be what is external to, and different from, mind; and this, according to the previous dichotomy between materialism and idealism, can only be the material world. This is confirmed by Burnyeat’s following argument that “the monism which comes most naturally to a Greek philosopher is materialism”, i.e., the reduction of “mind to matter,” whereas “a monism leaning in the other direction, from reality to mind, would be repellent to Greek thought.”^[42] This shows that in this context at least, Burnyeat (like so many others) primarily thinks of reality on the model of material existence.^[43] Of course, he knows that Plato’s ideas cannot be material; but he cannot help but think them, insofar as they are supposed to be ‘real,’ with this model in mind. And this also holds for the way he understands ‘being’ at Parmenides 132a–b, as well as in ancient philosophy in general, when he characterizes it by its ‘brutal realism.’

Now such an understanding of being is far from obvious, and it is precisely what ‘idealist’ interpretations of Plato have constantly denied. To show this, let us now turn to someone conspicuously absent from Burnyeat’s article, namely Paul Natorp.^[44]

As is well known, Paul Natorp offered a powerful interpretation of Plato in his 1903 (second edition 1921) Platos Ideenlehre.^[45] The subtitle of this book is “Eine Einführung in den Idealismus” (“An Introduction to Idealism”). As he explains in the foreword to the first edition, the main philosophical aim of his book is to restore the right understanding of idealism – that is of philosophy, since for him, true philosophy is identical with idealism properly understood. In his eyes, the best way to do this is to follow its birth and growth in Plato.^[46] This shows, and Natorp insists on this point, that he understands ‘idealism’ as deriving its name “not from Berkeley’s but from Plato’s ideas.”^[47] That is to say: at this point, calling Plato’s philosophy ‘idealism’ is an innocuous truism. But of course, Natorp has a very definite conception of idealism that results from an interpretation of both Kant and Plato – and not from a projection of Kant upon Plato, as is often, but wrongly, assumed.

Natorp belonged to the so-called Marburg Neo-Kantian School founded by Hermann Cohen.^[48] Cohen’s interpretation of Kant, shared by the other members of the school, was characterized by its radical anti-subjectivism: it fought any reduction of Kant’s a priori to innatism and any confusion between transcendental philosophy and psychology. According to this interpretation, Kant’s Critique of Pure Reason should not be understood as a critique of human reason, but rather as an investigation into the principles of the sciences as they appeared in history. “Transcendental method” therefore consists in starting with facts – most notably scientific facts, but also all cultural productions in general – and showing their conditions of possibility, that is to say their “lawfulness.”^[49] The idea underlying this conception is that objects are never ‘given’ to us, as the dogmatic point of view uncritically assumes, but are always culturally ‘produced’ and continually remain to be produced.^[50] In order to produce anything objective at all, such a production has to respect some laws, which are what the transcendental method purports to unveil in all areas of culture. Of course, ‘production’ should not be understood literally here; it really means that any object is grounded in laws that explain its objectivity, as something consisting of more than merely random associations.^[51] What is a law in this sense? Natorp defines it as “the permanent unity into which the changing manifold of the phenomenon is ideally unified and fixed.”^[52] In other words, it corresponds to a concept in the Kantian sense, i.e., a function of unity that determines the manifold of intuition and transforms it into an object. But it is worth stressing that such concepts, at least according to the Marburg interpretation of Kant, are in no way ‘subjective’ or ‘psychological’ entities: on the contrary, they are objective, or more precisely ‘objectivizing,’ insofar as they are that according to which objects can be produced in the sense explained.^[53]

Following Cohen’s lead, Natorp identifies Plato’s ideas to such laws.^[54] This interpretation has often been dismissed or even ridiculed, but it is often misunderstood. In order to see how Cohen and Natorp are led to it, it is useful to remark, with A. Kim, that Gesetz, the German word for “law,” is formed from the verb setzen, “to set,” “to place”, “to set down”, so that “the literal sense of ‘law’ for Natorp is ‘that which has been set or laid down’” and that “law” “signifies any rational posit whatsoever.”^[55] These rational posits are a priori: they cannot be found in experience, but they are themselves the conditions of possibility of experience. They are methods according to which thinking generates objectivities out of the sensible manifold.^[56]

Natorp’s (and Cohen’s) main ground for interpreting Platonic ideas as Gesetze thus understood is the ‘second sailing’ passage of the Phaedo (99d–100b). In that passage, Socrates begins by explaining that he has realized that the truth of that which is is only to be discerned in λόγοι, and he adds (100a, in an English version of Natorp’s translation and paraphrase):

By always making whichever logos, or logical proposition, I adjudge (κρίνω) to be the strongest (that stands up best logically) into the basis (ὑποθέμενος, I posit as foundation), I posit (τίθημι) as true whatever seems to me to agree with it.^[57]

Natorp comments:

Hence the ‘facts’ themselves can only be justified as true by being grounded in fundamental positings or logical ‘principles’ [or ‘fundamental propositions,’ ‘Grund-Sätze.’] And hence the λόγοι, in which truth of the ὄντα is to be founded, are the propositions, the positings peculiar to thought.^[58]

According to Natorp, this is “the principle of idealism” – but he immediately adds: “meaning (this caution may be necessary) critical or, as we would prefer to say, method-oriented idealism [methodischen Idealismus].”^[59]

Now the next step of this crucial passage of the Phaedo is equally important; for in order to explain what he has just said, Socrates rephrases it the following way:

Well, this is what I mean: it’s nothing new, but what I’ve spoken of incessantly in our earlier discussion as well as at other times […] I’ll go back to those much harped-on entities, and start from them, hypothesizing that a beautiful, itself by itself, is something, and so are a good and a large and all the rest. If you grant me that and agree that those things exist, I hope that from them I shall display to you the reason, and find out that soul is immortal. (Phaedo 100b, trans. Gallop 1975)

According to Natorp, what happens here is a full reduction of the meaning of ideas to what was previously called ‘hypotheses,’ i.e., positings in thought. And this occurs at two levels: “Plato’s first hypothesis is not only that there are ideas, but the ideas themselves are hypotheses.”^[60] Now this implies that they are laws, since hypothesis means exactly the same as Gesetz, i.e., what is posited or set under, as a principle for interpreting the sensible manifold.^[61] They can also be understood as concepts, and Natorp indeed states at the very beginning of his Plato book that eidos can be rendered as ‘concept’ (Begriff).^[62] But it must be stressed again that such concepts, at least according to the Marburg interpretation of Kant, are in no way ‘subjective’ or ‘mental’ entities, but are rather methods of objectivation. Hence Natorp insists that his interpretation of Plato’s ideas as concepts or laws is not at all subjective; and in the face of so many misunderstandings due to the dominating Berkeleian understanding of ‘idealism,’ he shows himself ready to drop the very word ‘idealism’ if it now inevitably brings the suspicion of subjectivism.^[63] As we have seen, however, there is no reason to do so given the historical roots of the word.

Natorp develops and defends his interpretation by a close scrutiny of most of Plato’s dialogues, culminating in an impressive interpretation of the Parmenides in which he finds the best proof that Plato conceived of ideas not as objects, but as objectivizing laws of thought in the sense explained. I cannot go into the details of his complex and fascinating interpretation of this dialogue here.^[64] In the present context, however, one question should be asked: does not Parmenides 132a–b, as understood by Burnyeat, precisely exclude such an interpretation of Plato’s ideas? To answer that question, let us turn to Natorp’s own reading of this argument.

The first surprise when one looks at the page where Natorp tackles this argument is that it does not trouble him at all.^[65] Not only does Natorp not see it as a threat to his interpretation, but, on the contrary, he considers it as a confirmation of his view of Plato’s idealism. Parmenides had just reminded Socrates, rightly, that the unity of ideas signified “the unity of consciousness” which “comes into being for us when we, looking at what is many, recognize it as something determinate from this one particular ‘point of view’”^[66] (Natorp’s take on Parmenides 132a). Yet Socrates misunderstands him and transforms his suggestion in a way “making ideas completely subjective.” This is “a great advance” in that it does not treat “a unity which is that of thought alone” as just another external thing, but rather as that for which alone there are objects in general. The problem, however, is that it considers this thought as subjective, as the plural ἐν ψυχαῖς at 132b5 shows, i.e., as belonging to the many “consciousnesses” rather than to “consciousness in general.” Hence if this view of ideas can be labelled ‘idealism,’ it is only in the sense of Berkeley’s psychological idealism, which has nothing to do with Kant’s (or Plato’s) critical or transcendental idealism.^[67] According to idealism in the latter sense, consciousness in general is not subjective: it rather corresponds to “the method of unifying a manifold” of which ideas are different modes. Hence as we have seen, ideas are neither subjective nor strictly objective: they are rather “objectivizing.” As Natorp stresses, notably with reference to the ‘second sailing’ of the Phaedo, “[m]ethod is sovereign”: it is not preceded by either a subject or an object; it is rather what first posits both an object and a subject for which this object is an object.^[68]

Here it will be useful to pause and look at Natorp’s interpretation of the first part of the Theaetetus. Contrary to Berkeley (as read by Burnyeat), Natorp is well aware that the relativism expounded and developed there is not Socrates’ own thesis, but “his spelling out of the meaning and the presuppositions of the initial thesis that knowledge is perception”^[69]; as a result, Natorp very clearly sees that this relativism implies the dependency both of the perceived object on the perceiving subject and of the perceiving subject on the perceived object. Here is how Natorp formulates this:

I am only perceiving in relation to this or that, which I happen to be perceiving at any moment; and it is this or that for me only, who is perceiving in this way. So it is only in relationship to one another that we both are, viz., the object and the subject of sensation, when we are, or that we both become (this or that), when we become. For in this way ‘necessity’ links our beings, but not the being (of any one of us) with the being of any other. Thus ‘we,’ I and my object, exist only in indivisible mutual connectedness with one another. There is no isolated being or becoming, but only a being or becoming in relation to something: that of the subject in relation to an object, of the object in relation to a subject (160ab).^[70]

One will notice an important difference between Natorp and Burnyeat’s accounts, namely that contrary to Burnyeat, Natorp does not interpret this relationship as a relationship between mind and matter – a move unwarranted in the text. He is content with the vocabulary of ‘subject’ and ‘object,’ which he leaves undetermined.

Now according to Natorp, although this is supposed to be Protagoras’ position rather than Socrates’, Plato recognizes that there is some truth in it:

The character of thoroughgoing relativity and subjectivity is really the character of pure sensibility. But it requires, indeed includes, the opposite, namely a being to which the diametrically opposed predicates of identical determinacy and hence object-likeness belong.^[71]

This is what Socrates expresses in characterizing the sensible as ἄπειρον (183b), which does not only mean that it is in itself purely undetermined, but also that it is to be determined – a determination which can only be the achievement of thought.^[72] According to Natorp, this determining function of thought is what is explored in the concluding argument against the first definition of knowledge (184–187). There Socrates states that our many sensations are synoptically viewed in “some one idea” (μίαν τινὰ ἰδέαν), “whether one calls it ψυχή or what one will” (184d). For Natorp, ψυχή should be translated here by ‘consciousness’ and refers not to a substance, but to a function: it is Kant’s “consciousness in general,” and not the many “consciousnesses” of Parmenides 132b.^[73] This function is “the general function of ‘synthetic unity,’” which is later called “judgment” (κρίνειν, 186b; δοξάζειν, 187a; according to Natorp, who notably relies on 189e–190a, δόξα in this context turns out to be the result of the dialectical activity of thought when it is no longer in conflict with itself).^[74] Now for Plato as for Kant, concepts are “predicates of possible judgements”^[75]; hence they are types of synthesis that can be traced back to fundamental types of synthesis, i.e., to fundamental functions of judging.^[76] Such are the κοινά discussed at Theaetetus 184–187: being and not-being, identity and difference, qualitative similarity and dissimilarity, unity and number, even and odd, “and everything which follows from this” – which for that reason are all concepts of relationship.^[77]

It is interesting to note that Natorp’s interpretation is not without similarities to Burnyeat’s own interpretation of this passage – including the translation of ψυχή by ‘consciousness’ – as Burnyeat himself recognizes in another article.^[78] But Burnyeat does not seem to see that such an interpretation can lead to a kind of idealism that is entirely compatible with the argument of Parmenides 132b–c. For in this context, being (οὐσία) plays a central role; as Natorp notices, it is the principal class in which all theorical predicates are included (whereas ὠφέλια is the principal class in which all practical predicates are included) (see Theaetetus 186c). Now how is this being to be understood? Certainly not as (material) existence – this is precisely one of the points on which Burnyeat explicitly agrees with Natorp.^[79] According to Natorp, the understanding of being as existence is derived from Aristotle, who in his objections to Plato presupposes that “the only sense of being is the being that is represented in discrete and particular existences.”^[80] This presupposition is also palpable in his system of categories, which places the category of a thing, or substance, at the pinnacle, “for no deeper reason, ultimately, than the fact that things, precisely as things, are given” – the primary mistake of all forms of dogmatism.^[81] By contrast, for Plato, “being signifies quite generally: determinacy in thought”^[82]; or, as Natorp puts it in the context of the Theaetetus:

“Being” means here, as it always does in Plato’s more rigorous philosophical language, the positing in thought, the unity of determination, and, therefore, of predication.^[83]

It is precisely for this reason that for Plato, ideas – that is, laws, according to Natorp’s interpretation – are the only things that truly are:

A law is the only thing that provides for determination and is the only thing that is completely determinate in itself; and in Plato’s sense this means that only a law can in the strict sense be said to be.^[84]

All that remains to be done is to import this understanding of being in the argument of Parmenides 132b–c according to which an ἰδέα is not a νόημα, in the sense of a (subjective) act of thought or a (subjective) content of thought, but rather a νοούμενον, i.e., what is thought by such an act or through such a content. From this point of view, the fact that the object of the νόημα is does not mean that it is something ‘given from outside,’ understood on the model of material existence, but simply that it is posited by thought, which after all is perfectly compatible with its designation as a νοούμενον. As Natorp explains in a later text with reference to Parmenides – and he thinks Parmenides and Plato are in agreement on this point – the being that is the object of thought can be considered as the “intentional object” of thought, but only on the condition that, by ‘intentional object,’ one understands not something that would be absolutely “outside” or “beyond” the realm of what is thought, but what is thought (das Gedachte) itself. This is the meaning of the Platonic identification between pure being and pure thought (in the sense of what is thought and not of what thinks):

This [identification] means as little subjectivization as objectivization, in the sense of a going beyond the whole realm of what is thought [des Gedachten]. Such a going beyond could in general in no way be executed in thinking [im Denken]; the realm beyond would in its turn be – thought [gedacht], posited in a new thinking [Denken].^[85]

As Natorp insists, there is no subjectivism here, since being is identified not with the act of thinking (das Denken), but with what is thought (das Gedachte).^[86] But this identification between being and what is posited in thought is the core idea of idealism properly understood.

This understanding of being is anything but arbitrary. Before turning to philosophy, Natorp had a solid education in philology with Hermann Usener in Bonn^[87]; and his first philological works remain landmarks in the field.^[88] Besides his philological skills, his remarkable knowledge of Plato’s texts is evident on every page of Platos Ideenlehre. The book itself ends with a thirty-six page index of key terms for Natorp’s interpretation of Plato with references to thousands of passages from the dialogues meant to justify his understanding of them.^[89] As Natorp explains in his foreword to the first edition, this index is supposed to provide the philological basis on which his exegesis relies.^[90] If one looks at the entry “Sein, εἶναι, ὄν, οὐσία,” one finds four columns of about two hundreds references to passages in the dialogues classified under the following eight headings (from the most important and general to the least frequent):

1. the content of the predication or of the (particular) concept;

2. the concept, the conceptual (predicative) being in general;

3. τὸ ὄντως ὄν or ὄντως οὖσα οὐσία in the same sense;

4. truth, being-so, especially ὄν, οὐσία and μὴ ὄν in the sense of particular affirmative or negative predication;

5. being as opposed to appearence;

6. being as opposed to γίγνεσθαι, γιγνόμενον, γένεσις;

7. being as the existence of sensible things, in contrast with ideas;

8. empirical being, insofar as it is grounded in the idea or in the determination of the undetermined.^[91]

As one can see, the first four headings, which include by far the most numerous references, are all connected with predication, and so, too, with concepts, since, as we have seen, concepts are predicates of possible judgments in the Kantian sense. As Natorp explains in Über Platos Ideenlehre, this does not at all reflect a specific or novel use in Plato’s philosophical language, but rather the usual and basic meaning of εἶναι in the Greek language.^[92] He recalls that ‘being’ is first and foremost a word of the human language. In ordinary Greek, the mere ἔστιν signifies “it is so” (es ist so), “it behaves this way” (es verhält sich so), i.e., the way it was expressed. According to Natorp, it is this being as it is in fact understood in an obvious way by any individual who expresses and understands thoughts in general that Parmenides thematizes in his poem; his use of ἔστιν does not inaugurate a new and revolutionary meaning of being. This is what he is getting at when he states that being is “identical” with thought (fr. 3), or rather with what is thought (fr. 8.34), or still more precisely with what thought is expressed in (fr. 8.35–6). Natorp comments:

Thinking, any thinking expresses itself in a “it is”; consequently, thinking is, according to its complete signification, in all cases positing of a being.^[93]

For Natorp, the same is true for Protagoras, although whereas Parmenides ascribes an eternal and immutable unity to this being, Protagoras makes it something purely subjective and always changing.^[94] And it also holds for Plato, for whom being is the correlate of the logos, and more precisely of the dialegesthai, and who demonstrates that this being must be eternally unchanging, one and identical, if thinking and knowledge are to be possible in general:

The being of ideas, this αὐτὴ οὐσία, is nothing else than precisely such a being which is posited at the same time in everything which is purely thought as such, by anyone who thinks it and for anyone who understands something that is thought this way in general. If we think, i.e., if we do not merely dream, but if we take seriously what we are spiritually concerned with, we think: it is, or it is not, and this being and not-being does not flow away with this fugitive thought event of ours in the dreamlike stream of “coming-to-be,” but it stands still, incontrovertible, eternally and immutably one and the same. This is what Parmenides states, what Plato re-establishes, especially against Protagoras’ objection, against the whole disruption of all firmly held concepts and truths which comes to expression with the sophists; this is what he maintains with a deeper, more essential justification, and what he takes from now on as a basis for all philosophical or scientific ideas.^[95]

Now as we have seen, according to Natorp, this idea that truth is to be found in λόγοι is “the principle of idealism,” which Plato formulates most definitely at Phaedo 99e.^[96] It therefore appears that for Natorp, there is a kind of “underlying idealism” in the whole Greek philosophy, i.e., the assumption that being is intrinsically connected to thought and to language, as what is thought and what is expressed in judgment. In order to think and speak, we have to fix things in determinations, to assume that they are steady and will not change immediately with no reason – in other words, that they are. This conception is common to all Greek philosophers, including Parmenides, Protagoras and Plato – and also Aristotle, even if Natorp does not mention him here. They differ in the way they interpret this being which constitutes the meaning of what is thought and the features they ascribe to it.^[97] As for Aristotle, Natorp suggests that he was not faithful to this underlying idealism, since he attributes the being that is the condition of thought and language not to thought and its activity, but to the things themselves, as if thought had nothing to do but passively receive its objects, which would be given with all their determinations. In other words, Aristotle turned this underlying idealism into dogmatism.^[98]

Such an interpretation of Greek philosophy seems to be the exact opposite of Burnyeat’s, which on the contrary ascribes to antiquity “an unquestioned, unquestioning assumption of realism.” Who is right on this point? I think there is much to be said in favour of Natorp, even if we do not follow him in all the details of his interpretation of Plato and other ancient philosophers. For Natorp provides a solid philological basis for his interpretation by analyzing numerous occurrences of the verb ‘to be’ in and outside Plato. Now it is remarkable that the results of this analysis anticipate important aspects of later investigations, most notably by G. E. L. Owen, Charles Kahn and Lesley Brown, who have independently argued, in my eyes convincingly, for the centrality of predication in the Greek use of εἶναι and its relation to truth^[99] – what Kahn refers to as his “version of the Copernican Revolution,” but which might have appeared less revolutionary had Natorp’s works not been ignored.^[100] Burnyeat himself is actually sympathetic to this trend, especially when it comes to interpreting Theaetetus 184–187, as is evident in a paper that demonstrates some knowledge of precisely the text in which Natorp develops this interpretation of the Greek εἶναι in its fullest form.^[101] What Burnyeat did not see, however, and what Natorp did, is that such an understanding of being is intrinsically connected to a kind of idealism – an idealism which has admittedly nothing to do with Berkeley’s immaterialism, but which is nevertheless certainly incompatible Burnyeat’s “unquestioned, unquestioning assumption” in his assessment of idealism in Greek philosophy that being primarily means existence, understood in a way ultimately derived from material existence, an assumption which it is hard to reconcile with his much more nuanced interpretation of Theaetetus 184–187.

In replying to his critics, Natorp is keen to recall that his first training was in classical philology and that his “moral consciousness as a philologist” required him to ask: where is the evidence?^[102] It is indeed revelatory that the great French historian Léon Robin could feel authorized to write that Natorp’s interpretation of ideas as laws “is contradicted less by the texts than by the realistic tendencies of Greek thought.”^[103] How could “the realistic tendencies of Greek thought” be known, if not from texts? Just as in Burnyeat’s case, one gets the feeling that one is here facing a mere postulate on the interpreter’s part, so that it is not clear that the bias is really to be found on Natorp’s side.^[104] At the very least, I hope to have helped show that the issue of ‘idealism in Greek philosophy’ is far from settled by Burnyeat’s paper, and that any advance on this question must begin, very Platonically, by clarifying the terms of the debate – but also by paying attention to alternative accounts, including those coming from different traditions.