Against the Two Worlds

What the Two Worlds Theory says

She states the target precisely before touching it — and note that on the strong version the claim is not merely that knowledge and belief have different objects, but that they exclude one another.

It is often thought that in the middle dialogues Plato favours the Two Worlds Theory… On one familiar version, knowledge and belief have disjoint objects — forms and sensibles, respectively… Otherwise put, the contents of knowledge and belief — the propositions one can know or believe — are disjoint… On this view, knowledge, so far from implying belief, excludes it.Introduction §II, p. 8

The Meno already rules it out

Her first move is not to attack the Republic but to establish that the theory cannot be Plato's general view, because another dialogue plainly denies it.

The Two Worlds Theory is at odds with the epistemology of the Meno. For there, as we saw, Plato says that knowledge is true belief bound by an aitias logismos; on this view, knowledge, so far from excluding belief, implies it… He says, for example, that although the slave has a mere true belief about the answer to the geometrical question Socrates poses, he can acquire knowledge of it. Plato also says that one may know, or have a mere true belief about, the way to Larissa (97 a ff.). The Meno also seems to allow knowledge of sensible particulars as well as beliefs about forms.§II, p. 8

So the question narrows to a single one: is Republic V–VII — "Plato's lengthiest and most systematic discussion of knowledge and belief in the middle dialogues" — committed to the theory?

The dialectical requirement

The interpretive key, and the constraint that does most of the work in what follows. The argument at the end of Republic V is addressed to a specific audience.

At the end of Republic V Plato seeks to persuade those he calls sightlovers, on grounds acceptable to them, that they lack knowledge. The argument is therefore governed by what I call the requirement of noncontroversiality, or the dialectical requirement (cf. Meno 75 d): it should use premisses that are acceptable to the sightlovers.§II, p. 9

Everything turns on one Greek word

The two central premisses are that knowledge is set over what is and belief over what is and is not. But esti — "is" — can be used in several ways, and which one Plato means decides the whole dispute.

If esti is used veridically here, then the point is that knowledge but not belief implies truth. I call this a contents analysis, since it distinguishes knowledge and belief by their propositional contents. If esti is used existentially or predicatively, then knowledge and belief are distinguished by their nonpropositional objects; hence I call this an objects analysis.§II, pp. 9–10

On the predicative reading — "probably the most popular one" — the objects of knowledge are restricted to what is F and not also not-F, the objects of belief to what is both F and not F. That reading gives you the Two Worlds Theory. Her whole case is that it is the wrong reading.

Her reading: veridical first, then predicative — and the shift is deliberate

I argue, however, that for most of the argument (from the beginning through 478 b 12), esti is used veridically, to make the point that knowledge but not belief implies truth. One advantage of this reading is that it, unlike its rivals, satisfies the dialectical requirement.§II, p. 10

And in the footnote, a second and colder advantage:

A further advantage of my reading of the argument is that on it, the argument is valid. On at least some versions of the existential and predicative readings, by contrast, the argument is invalid.§II, n. 24

At 479 a ff. Plato does turn to the predicative use — to establish that the many Fs are both F and not F. Her illustration is characteristically concrete:

Bright colour, for example, is beautiful in this Matisse painting; but it is merely garish, and so ugly, in that other painting.§II, p. 10

And this too respects the dialectical requirement, because it is the sightlovers' own reason for saying the beautiful is many rather than one. Plato is not smuggling; he is using their premise against them.

I argue… that he shifts from the veridical to the predicative use, without confusion, in order to persuade the sightlovers that if they define beauty in terms of the many beautifuls, which are beautiful and ugly, they will inevitably have both true and false beliefs about beauty: in which case they lack knowledge of beauty.§II, p. 10

She corrects her own earlier papers

This is the moment the Introduction earns its place. In the two Republic essays reprinted later in the same volume, she had said Plato assumes that beauty is one. She now withdraws it, on her own dialectical-requirement grounds.

In Chapters 3 and 4 I say he assumes that beauty is one. This, however, can't (or, at any rate, had better not) be right: for that assumption is the central bone of contention between Plato and the sightlovers. If Plato simply assumes that beauty is one, there is an especially egregious violation of the dialectical requirement. But in fact, contrary to Chapters 3 and 4, we need not accuse Plato of reasoning in this way.§II, p. 11

The replacement premise is weaker and uncontroversial — which is exactly why it works:

He seems to rely, not on the assumption that beauty is one, but on the assumption that reliance on a satisfactory definition shouldn't inevitably issue in false beliefs about what things do, or don't, fall under the definition: this explains his emphasis on the truth condition for knowledge.§II, p. 11

Only then does Plato argue that beauty is one

Plato therefore in effect eventually argues that beauty is one; but he doesn't assume that it is in arguing against the sightlovers. Nor, in any case, is this last stage of the argument governed by the dialectical requirement. That requirement governs only the argument for the claim that the sightlovers lack knowledge.§II, p. 12

Two stages, two standards of proof. The negative stage must be fought on the opponent's ground; the positive stage need not be.

Propositional knowledge and knowledge of things — not a confusion

The standard charge against Plato is that he slides between knowing that and knowing a thing. She denies the slide by showing the two are connected, not conflated.

For in his view, to know a form requires knowing what it is; and that, in turn, is to know its definition, which is a matter of propositional knowledge.§II, p. 12

Later she puts the same point as a compliment: Plato "has the deep insight that knowing things requires knowing truths about them."

The verdict on the Republic

As I interpret the argument, it doesn't imply the Two Worlds Theory. Rather, Plato defends only the weaker view that to have any knowledge at all, one must know forms. This leaves open the possibility that once one has such knowledge, one can know more than forms; it also leaves open the possibility that one can have mere belief (whether true or false) about forms.§II, pp. 12–13

And then the stronger claim, which is the sentence to remember from the whole Introduction:

Forms are universals, full stop

On the view I defend, forms are just universals, conceived as real, mind-independent, objective explanatory properties; they are not also or instead particulars. As in his epistemology, so in his metaphysics: Plato's views, this time about forms, have much to recommend them still.§VII, pp. 30–31

She grants that Plato calls forms everlasting, unchanging, unperceivable, paradigms — and that he accepts self-predication. Her claim is that none of that stops them being universals, or being immanent in particulars, and that his resulting account of property possession "is no more implausible than Aristotle's allegedly superior account."

She agrees with Aristotle about the origin — and refuses his conclusion

On why there are forms she sides with Aristotle: knowledge requires definition, and definition requires real universals (Metaph. A 6, M 4, M 9). What she rejects is his inference from separation to particularity.

From this point of view, the puzzling question isn't why Plato takes forms to be separate, if he does, but why Aristotle thinks that separation makes forms particulars. I suggest that Aristotle has the deep-seated view that only particulars can be separate… universals, in his view, are not the sorts of things that can exist uninstantiated.§VIII, pp. 32–33

Separation, defined precisely

Separation is a relational notion: to be separate is to be separate from something… Separation is also a modal notion: to say that forms are separate is to say that the form of F can exist whether or not any sensible F things exist… to say that they are separate is to say they can exist uninstantiated by sensible particulars.§VIII, pp. 32–33

Note what this definition does not say. Nothing about location, nothing about another realm, nothing about copies. "Separate" means: can exist uninstantiated. The whole ontological content of "two worlds," on her account, reduces to that modal claim.

The striking fact: no argument of Plato's proves separation

For one thing, it's a striking fact that none of Plato's arguments for the existence of forms implies that they are separate.… This shows that forms must be different from sensible particulars (and from sensible properties). But to say that forms are different from sensibles falls short of saying that they are separate from them: A can be different from B, without being able to exist without it.§VIII, p. 33

She applies this to the flux argument specifically — the argument most people take to establish two worlds:

He certainly argues that the flux of sensibles requires the existence of stable, nonsensible forms that escape flux. But this again shows only that forms are different from sensibles; it doesn't show that forms are separate from sensibles.§VIII, p. 33

And she declines the easy conclusion that Plato therefore blundered:

Nor does Plato invalidly infer separation from the 'flux argument'… He takes such arguments to establish no more than that forms are different from sensibles. Nor does Aristotle accuse Plato of arguing invalidly for separation; he thinks separation rests on other grounds, which give Plato a valid (but, in Aristotle's view, unsound) argument for separation.§VIII, p. 33

He believes it anyway — as an assumption, exactly as Aristotle denies it

Nor, in the middle dialogues, does he even explicitly say that forms exist separately (chōris). None the less, I think Plato accepts separation. Just as Aristotle seems to assume without argument that universals cannot be uninstantiated, so Plato seems to assume without argument that forms are separate, that is, can exist uninstantiated.§VIII, p. 34

Her evidence is oblique and, for that reason, persuasive — artefact forms. In the Cratylus and Republic X Plato countenances a form of shuttle; and a form of shuttle can exist whether or not any shuttle does. The commitment "emerges in unexpected ways and in unexpected cases."