When we read a didactic text it is natural to expect the teaching to become more demanding as we approach the end. Scholars who work on this kind of literature sometimes argue that an author expects more and more of his readers. It is easier to look for evidence in support of this when the author comments directly on the relationship between himself and the reader or addressee, as Seneca does in his letters to Lucilius: we can compare the representation of Lucilius and the demands placed on him at different stages of the correspondence. C.'s book applies this approach to Plato. Far from focusing on a single dialogue, she treats the entire Platonic corpus as an educational (she rejects the term ‘didactic’) unit. Her discussion is evidently informed by knowledge of ancient literature, Latin as well as Greek, but comparisons with didactic authors, such as Lucretius, are treated with a light touch. She keeps Plato and the scholarship on Plato at the centre. Her book will be of most interest to scholars working on characterisation in Plato, but she recognises that the dialogues have an ‘argument-focused texture’ (p. 124).

C. is careful to avoid suggesting that when Plato started his writing he already had in mind a fully formed programme. But she still gives Plato a programme: for newcomers to philosophy he wrote accessible dialogues, where Socrates' interlocutors are at a basic philosophical level, and he expected his readers to move on to more demanding dialogues, where the interlocutors are more advanced. As C. puts it, Plato did not develop, but he expected his readers to develop and so to match the increasing sophistication of the dialogues' interlocutors. (C. recognises important contrasts between readers and interlocutors, and argues that readers should not merely imitate the most advanced interlocutors.) Unsurprisingly, C. presents her view of the dialogues as a challenge to ‘developmentalism’, as that term is usually understood. But much of the developmental picture is retained: when C. says which dialogues are accessible or advanced she reuses the distinction between ‘early’, ‘middle’ and ‘late’ dialogues, including even the separation of Republic 1 from the other nine books. ‘Early’ dialogues, such as the first book of the Republic, were not necessarily written at an early date (C. distances herself from the view that Republic 1 existed as a separate work); ‘early’ means ‘appropriate to a reader at an early stage of his or her philosophical development’ and ‘containing interlocutors at an early stage of their development’. So too for ‘middle’ and ‘late’.

C. could not have considered every dialogue, let alone every interlocutor, but none the less she needs to consider more broadly whether her characterisation of the interlocutors – philosophically unsophisticated in the ‘early’ dialogues, philosophically advanced in the ‘late’ dialogues – is quite right. There are obvious exceptions in the ‘late’ dialogues: the interlocutors of the Laws, Clinias and Megillus, are advanced in age but have little familiarity with philosophy. More fundamentally, placing interlocutors on a single line from ‘beginner’ to ‘advanced’ often misses out what is most interesting about them. The speakers of the Timaeus–Critias are not merely advanced: they are experts on different subjects, cosmology (Timaeus) and political theory (Socrates). This sequence of dialogues is clearly exploring what different specialisms have to say to one another, whereas C.'s account of philosophical development leaves no room for specialism. (The nature and value of Critias' expertise is more controversial.) What are we to make of a single dialogue that contains different interlocutors, such as the Gorgias? Perhaps we could show Callicles to be more, or less, ‘advanced’ than Gorgias, but there would still be a good deal left to say about the contrasts between them.

There are at least two reasons why we might hesitate to liken Plato to a didactic author, such as Lucretius. (1) In Plato's Phaedrus Socrates criticises texts and argues specifically that they cannot teach. (2) Saying that Plato wishes to convey doctrines is much more controversial than saying that Lucretius wishes to convey (Epicurean) doctrines. C. does not neglect the Phaedrus but, like others, she represents its critique of writing as more vaguely targeted than it really is; on her account, the Phaedrus questions ‘the philosophical value’ of written texts (p. 7) – or rather it appears to. (She argues that the Phaedrus suggests merely that texts have ‘limitations’, p. 17.) She offers much more in response to (2). On her account, Plato aims to make his readers active and independent philosophers, and (or ‘and so’ – I will return to this connection) is not a doctrinal writer. Instead Plato provokes an independent response from his readers by means of puzzling characterisation, unresolved ‘plots’ and disorientating transitions. Myths too are considered, or at least one myth (the myth of the true earth near the end of the Phaedo). Considering myth, characterisation or structure through selected examples is quite reasonable in a book of this length, but there is not always enough care when moving to generalisations. For example, the myth of Er makes far greater use of narrative than the myth of the true earth, and so C. should not take the latter to show that ‘Plato's myths are not narratives as we often think of them’ (p. 279).

C. uses the image of the Cave, including in particular the emphasis in that image on the learner's own ‘vision’ and independence, to illustrate how Plato aims to stimulate independent learning. C. frequently calls Plato's readers ‘receivers’, which suggests a more passive role for them (and looks incompatible with Symposium 175d and Republic 518d) but, unless I misunderstand her, she means simply that the readers receive the text, not that they receive from Plato wisdom or understanding. It is no accident that she talks of ‘education’ rather than ‘didacticism’, and she devotes a chapter to arguing that Plato's ‘late’ dialogues are not more ‘didactic’. It is striking to see the Cave presented not as a problem for interpreters but as the key to interpretation. But it would be hard for anyone to deny that the Cave is (among other things) about the independence of the learner, and so in her discussion of the passage C. shows a good eye for the uncontroversial.

The question for a project like C.'s is as follows: what degree of vagueness, or ambiguity, is required if a text is to avoid constraining its readers' independence? Two philosophical writers may both want independent responses to their writing and yet may offer utterly different answers to that question. Plato, C. seems to assume, thought that pervasive vagueness and open-endedness were required to prevent readers from simply accepting his ideas. But Plato may have thought it possible to be simultaneously specific, decisive and non-constraining. To return to a dialogue that gets little attention in the book, when Timaeus considers the principles of earth, air, fire and water, he expects other philosophers to suggest alternatives to his account – and hails as a ‘friend’ anyone who suggests a better alternative (54a). Clearly he expects competition and criticism. But Timaeus is none the less extremely specific in his own account about which triangles he treats as the principles. I need not assume in this context that Timaeus speaks for Plato, but it is extremely unlikely that Timaeus is supposed to be confused. C., by contrast, makes Plato assume that a clear, well-defined account would not be considered critically and independently. Given what we know about Plato's colleagues and pupils – his first ‘receivers’, as C. would call them – would Plato really have worried about getting an uncritical reception?