¹ Especially in the following: Aischylos und das Handeln in Drama (Philologus, Supp. 20, 1928)Google Scholar, to be read with the review by Wolff, E. in Gnomon 5 (1929), 386–400Google Scholar; ‘Das Bewuβtsein von eigenen Entscheidung im früheren Griechentum’ (repr. in his Gesammelte Schriften [Göttingen, 1966], 18–31)Google Scholar; ‘Göttliche und menschliche Motivation im homerischen Epos’ (repr. in Ges. Schriften, 55–61)Google Scholar; Die Entdeckung des Geistes (Hamburg, 1948)Google Scholar; Szenen aus griechischen Dramen (Berlin, 1971).Google Scholar See also Fränkel, H., Dichtung und Philosophie des frühen Griechentums (Munich, 1962), pp. 83–103.Google Scholar

² Snell, , Ges. Schr., op. cit. (n. 1), p. 61.Google Scholar

³ Voigt, C., Ueberlegung und Entscheidung: Studien zur Selbstauffassung des Menschen bei Homer (Beiträge zur Klass. Phil., 48; Meisenhaim am Glan, 1972), p. 106.Google Scholar

⁴ Reasons and purposes engage with one another: there is no general antecedence of one to the Other. An Aristotelian – as opposed to a Humean – model of action allows for situations in which overall purposes are shaped by occurrent perceptions of features of the world as constituting reasons for action. See here Wiggins, D., ‘Deliberation and Practical Reason’ reprinted in Rorty, A. (ed.), Essays on Aristotle's Ethics (Berkeley, 1980), pp. 221–40.Google Scholar

⁵ Die Entdeckung des Geistes, ch.1. Snell relies on the earlier work by Böhme, J., Die Seele und das Ich im homerischen Epos (Göttingen, 1929).Google Scholar

⁶ That is, he has no one word for the Gesamtgemü, tas Böhme, op. cit. (n. 5), showed. In particular, the word ψνχ⋯ does not discharge this function, since it denotes no more than the life or consciousness of a man, in the sense of that which is taken away when he swoons or dies. Otto Regenbogen showed (in his ‘Δαιμóνιον ψνχc φc’, repr. in his Kleine Schriften [Munich, 1961], 1ff.) that the ψνχ⋯ is not to be thought of as materialising at the moment of death, but rather as accompanying a man throughout his life and deserting him at the moment of death. I think Regenbogen succeeds in showing that the ψνχ⋯ is the ‘im Lebenden conditio sine qua non aller körperlichen, geistigen und emotionalen Regungen’ (p. 20). But there is a difference between the life-principle of a man, which ψνχ⋯ may well be taken to represent, and the unitary nature of his self, which ψνχ⋯ cannot, as such, be taken to represent.

⁷ cf. Sharpies, R., ‘“But why has my spirit spoken with me thus?”: Homeric Decisionmaking’, Greece and Rome 30 (1983), 1–7CrossRefGoogle Scholar who correctly points out that the occurrence of the first-person pronoun is itself enough to equip Homer with a concept of selfhood, and that Snell's picture of Homeric man as an assemblage of various lobbying groups better fits Plato's model of the soul in Republic 4, or the myth of the charioteer in the Phaedrus. Sharples also alludes to a similarity between Homer and Aristotle – and the divergence of both of them from Plato – on the subject of akrasia, which I discuss below.

⁸ See the interesting piece by Gill, Christopher (‘Did Chrysippus Understand Medea?’, Phronesis 28 [1983], 136–49)CrossRefGoogle Scholar, in which he argues that the Medea of Euripides’ Medea 1078–80 must be thought of as a unitary agent. That was the line taken by Chrysippus, who argued that Medea's impulse to kill her children was contrary to reason in the sense that it was unreasonable, not in the sense that it was arational, as a Platonic model of the soul would have it. For Chrysippus, ⋯ρμα⋯, π⋯θη etc. are rational in the sense that they are conceptual mental states: they involve judgements. He apparently used Homeric examples to justify his view that the soul functions in a unified way (Galen, , De Placitis Hippocratis et Platonis, ed. De Lacy, P. [Berlin, 1978–1984], III.2.10–20).Google Scholar Galen takes a Platonic line on Medea's struggle (III.3.13–22), describing Medea's deliberation as an inner dialogue between logismos and thumos. But he introduces – as anyone who wishes to make sense of Medea must do – a self (αủτ⋯) separate from these parts of the soul, and alternately under their hegemony.

⁹ Böhme's comment:‘… nicht mehrere Seelenteile treten nebeneinander, sondern was vorher als Anrede des Ich an den θυμóc beschrieben wurde, wird nun, wo Odysseus die Verantwortung für die ausgesprochenen Befürchtungen von sich abwälzen will, für ein Selbstgesprach seines θυμóc ausgegeben.’ (op. cit. [n. 5], p. 80). And note that the soliloquy is closed with the words: oc ⋯ ταθ᾽ ὥρμαινε κατ⋯ φρ⋯να κα⋯ κατ⋯ θυμ⋯ν (411). Here the soliloquy is represented as having been conducted by Odysseus in his thumos and phren.

¹⁰ Further criticisms of the lexical method – different from mine – are to be found in Lloyd-Jones, H., The Justice of Zeus, revised edition (Berkeley, 1983), passim.Google Scholar

¹¹ The Greeks and the Irrational (Berkeley, 1951), p. 7.Google Scholar

¹² A. Lesky, Göttliche und menschliche Motivation im homerischen Epos (Sitzungsberichte der Heidelberger Akadamie der Wissenschaften, 1961). The point is that the contributions of Apollo and Hector to Patroclus' death are not thought of as independent inputs, either of which would have been sufficient on its own to kill Patroclus. The gods work with (sometimes through) men. The term ‘overdetermination’ misses the essential unity of such action, which can be looked at now as the action of a god, and now as the action of a man. Rather than possessing two, independent components, such action possesses two interdependent aspects. Compare Achilles' anger as seen through the eyes of Ajax (Il. 9.624–42): first as Achilles' own doing (628–9), then as inspired by the gods (636–7). Lesky gives further examples of this ‘double aspect’.

¹³ For an elucidation of the Fregean framework employed here, see especially Evans, G., Tha Varieties of Reference (Oxford, 1982), ch. 1Google Scholar, along with McDowell, J., ‘On the Sense and Reference of a Proper Name’, Mind 86 (1977), 159–85CrossRefGoogle Scholar; Wiggins, D., ‘The Sense and Reference of Predicates’, in Wright, C. (ed.), Frege: Tradition and Influence (Oxford, 1984), 126–43.Google Scholar

¹⁴ See now Cessi, V., Erkennen und Handeln in der Theorie des Tragischen bei Aristoteles (Frankfurt/Main, 1987), passimGoogle Scholar, esp. the discussion of relevant Aristotelian passages, e.g. De Anitna 414b1–6 at pp. 137ff.Google Scholar; note also one of her main conclusions (p. 248): ‘Für Aristoteles ist Wahrnehmen ein unterscheidender, aktiver und spontan werdender Erkenntmsakt, aus dem durch Betätigung des auf ihm beruhenden Vorstellungsvermögens unmittelbar ein Streben (⋯ρεξιс) zur Handlung entspringt.’

¹⁵ On this topic in general see most helpfully Dihle, A., The Theory of the Will in Classical Antiquity (Berkeley, 1982), ch. 2Google Scholar, although Dihle is in my view unduly pessimistic about the competence of a purely cognitivist system of thought to construct a life-like theory of action. For further Homeric examples, see Fränkel, , op. cit. (n. 1), pp. 90–1.Google Scholar

¹⁶ See especially Dodds, op. cit. (n. 11), ch. 1; Lesky, op. cit. (n. 12); Wiist, E., ‘Von den Anfängen des Problems der Willensfreiheit’, Rhein. Museum 101 (1958), 75–91Google Scholar; Schwabl, H., ‘Zur Selbstständigkeit des Menschen bei Homer’, Wiener Studien 67 (1954), 46–64Google Scholar; recently Schmitt, A., ‘Athenes Umgang mit den Menschen bei Homer’, Die Alten Sprachen im Unterricht 29 (1982), 6–23.Google Scholar Lesky shows that we should think of human and divine realms as each possessing its own integrity, but as capable also of working together; so that actions can acquire a ‘double aspect’. This conception of the divine and human does not change markedly from Homer to the tragedians (although Sophocles' dramas which also deploy the device of miraculous divine intervention seem to be an exception). Schmitt argues that divine interventions rarely affect agents in an arbitrary way, but tend to match capabilities and propensities autonomously present in the agents affected. Such divine intervention cannot therefore remove responsibility, because it prompts an agent to do what he might, or would, otherwise have done. An interesting early defence of the thesis that divine intervention in Homer does not detract from human freedom is afforded by Plutarch, Coriolanus, ch. 32.

¹⁷ Aeschylos und das Handeln in Drama, op. cit. (n. 1), and elsewhere in opp. citt., n. 1.

¹⁸ op. cit. (n. 3), p. 41. Here Voigt suggests that Homer's frequent use of such formulae as δο⋯ссατο κ⋯ρδιον εναι indicates that the agent is not really choosing the course of action which seems better. But the distinction is spurious, as I argue in the text. (Cf. also Harrison, E., ‘Notes on Homeric Psychology’, Phoenix 14 [1960], 63–80, at p. 79CrossRefGoogle Scholar: ‘Here the poet naturally uses a formula which focuses attention on the content of the decision. We know already who is involved on each occasion: all we wish to be told at this point is what he decides to do.’) Voigt also claims (loc. cit.) that the description of the better course of action as κ⋯ρδιον removes it from the moral domain. But the line between the moral and the non-moral is not so easily drawn. A counterexample to the claim is raised by Voigt himself (p. 44): at Il. 16.652 Zeus decides that it would be κ⋯ρδιον to give Patroclus a bit longer before he is slain by Hector. There is no gain (in Voigt's sense) in this for Zeus: his deliberation and decision relate to what he ought to do. Consider too Hector's deliberations at Il. 22.99ff. In outline Hector says: it would have been κ⋯ρδιον if I had followed Poulydamas’ advice (103), but since I didn't do so, it is now κ⋯ρδιον to face Achilles in single combat (108). A moral ‘ought’ must be implicit in the second κ⋯ρδιον (Hector later replaces it by β⋯λτερον, 129); so that it is senseless to deny it to the first κ⋯ρδιον. Only someone who thought that prudential considerations always excluded moral considerations could fail to see this. But a course of action may be moral precisely because it is prudent.

¹⁹ op. cit. (n. 3), pp. 87ff. Cf. Snell, , Gesam. Schrift., op. cit. (n. 1), p. 21.Google Scholar

²⁰ Contra Voigt, ibid. (cf. Snell, , Szenen, op. cit. [n. 1], p. 18Google Scholar), who claims that such a questioning would not be open to a Homeric agent. But to question the standards of conduct in which he has been brought up, and which his social position demands of him, is precisely what Achilles does in his speech to the embassy at Il. 9.397ff.

²¹ cf. Lloyd-Jones, , op. cit. (n. 10), p. 240Google Scholar: ‘…all the time in ordinary life we automatically act in accordance with principles whose observance, by what Aristotle calls ethismos, has become a matter of habit; this does not mean that if called upon to justify those principles we would prove unable to do so.’ Lloyd-Jones could have written this last clause as: ‘this does not mean that such action is not genuine action, performed by the agent as a matter of his autonomous and self-conscious choice.’ What Lloyd-Jones wrote is equivalent to my version, since the type of action alluded to in my version is the very type of action we can be called upon to justify.

²² For an elucidation of the distinction as it applies to Phaedra, see Barrett's commentary ad loc.

²³ Compare Odysseus' deliberation in Polyphemus' cave (Od. 9.295–306). He is initially tempted to stab Polyphemus – as one would expect of any hero in his circumstance – but he is restrained by a ἕτεροс θυμ⋯с (303) which suggests a more intelligent policy. And as with Menelaus, second thoughts are best. Contrast Phaedra, for whom second thoughts (436) are worse. Again, the fact that Odysseus' deliberation is described in terms of lobbying θυμο⋯ does not impugn the unitary status of his self. Rather, it is a vivid way of conveying the fact that ideas just do occur to one (out of nowhere, as it seems). But they occur to one, i.e. to the unitary, acting self. Homer can represent decisions which, as it seems, simply ‘happen’ to one as owing to the intervention of a god (e.g. Phoenix's decision not to murder his father, Il. 9.458ff.). There is nothing especially primitive in this: it is one way of portraying a type of process which is not perspicuous to the agent himself.

²⁴ Not surprisingly, Voigt, op. cit. (n. 3), pp. 92ff. cannot cope with the sophistication and intelligence of Menelaus' reasoning; he has to suggest - –wholly implausibly – that Menelaus does select a cowardly course of action.

²⁵ Voigt, , op. cit. (n. 3), pp. 17–18.Google Scholar

²⁶ ‘Die Entscheidung des Achilleus’, in Von Homers Werk und Welt (Leipzig, 1944), pp. 162–9Google Scholar

²⁷ For this general line, see Wittgenstein, L., Philosophische Untersuchungen, e.g. 650Google Scholar, ‘Wir sagen, der Hund fürchtet, sein Herr werde ihn schlagen; aber nicht: er furchte, sein Herr werde ihn morgen schlagen. Warum nicht?’, and passim. The dog would need to be able to talk about tomorrow to be able to think about it. See also some of the papers of Donald Davidson collected in his Inquiries into Truth and Interpretation (Oxford, 1984)Google Scholar, especially the paper entitled ‘Thought and Talk’.

²⁸ Since the difference between a dog and any human being is not incommensurably great. After all, a dog can fear that his master will beat him (that is already a significant achievement, denied to lower animals), even if he cannot fear that his master will beat him tomorrow.

²⁹ Gesam. Schrift., op. cit. (n. 1), p. 59.Google Scholar Cf. Fränkel, , op. cit. (n. 1), p. 87Google Scholar, for the suggestion that Hamlet represents a type inaccessible to Homer. Hamlet's akrasia is marked not by the impact of a desire to act in a way contrary to what, all things considered, has been judged to be best, but rather by the absence of a desire to realise in action the preferred practical judgement. This type is not only inaccessible to Homer, but to classical thought in general, which would not have been able to make sense of an agent who has preponderating reasons for action, but nevertheless does not act; or indeed of an agent – such as Camus's Stranger – who has no appropriate reason for action, but nevertheless acts.

³⁰ Aristotle also makes the point that animals are not capable of akrasia, although his reason is that they are not capable of forming universal concepts (NE 1147b4–6).

³¹ She represents, as Voigt correctly observes (op. cit. [n. 3], p. 67), Helen's ‘Wesen und Vergangenheit’.

³² Aristotle's model of akrasia is a cognitivist one: the failure is characterised as one of knowledge rather than of will. Democritus' treatment of akrasia is also strongly cognitivist (B53, 53a); Theognis (631) and Euripides (Medea 1078–80, Hipp. 373–87) give less explicitly cognitivist descriptions, but are not couched in anything like the terms of a modern notion of the will. Again, I do not see any disadvantage in the cognitivist approach: its critics (e.g. Dihle, op. cit. [n. 15]), have difficulty expressing exactly what, other than a piece of terminology, it misses out.

³³ See here McDowell, J., ‘Virtue and Reason’, Monist 62 (1979), 331–50CrossRefGoogle Scholar; Wiggins, D., ‘Weakness of Will, Commensurability and the Objects of Deliberation and Desire’, repr. in Rorty, (ed.), op. cit. (n. 4), pp. 241–65Google Scholar; Cessi, , op. cit. (n. 14), esp. pp. 228ff.Google Scholar

³⁴ This is not quite how Aristotle presents the matter. He locates the impact of the desire at the minor premiss – rather than at the mental putting together of major and minor premisses – but this conflates akrasia with the sort of mistake with which NE 3.1 deals, and is anyway implausible. He cannot locate the impact at the conclusion (contra Kenny, A., Aristotle's Theory of the Will [London, 1979], pp. 161 f.Google Scholar), since for Aristotle (in keeping with Greek cognitivism) there is no logical gap between conclusion and action (NE 1147a25–31, De Motu 701a8–25). Wiggins (art. cit., previous note, pp. 249–50) suggests that unless the gap is present, no room is left for syllogisms to compete with one another, and so for the akratic man to struggle. But the akratic man struggles – one might so put it – to draw the conclusion. Desire, backing the worse syllogism, interferes with his ability to hold on to the combination of major and minor premiss of the better syllogism. The syllogisms themselves are in conflict not in the sense that their conclusions actually do compete with each other – since not both conclusions are reached – but in the sense that they potentially generate incompatible conclusions.

³⁵ It may of course be the case, as Walter Nicolai points out to me (and see his ‘Wirkungsabsichten des Iliasdichters’ in Kurz, Müller, Nicolai [edd.], Gnomosyne, Festschrift für Walter Marg [Munich, 1981], at p. 99Google Scholar) that Hector's reflections at 22.111–21 hit on a policy – handing back Helen and making financial reparations – which it would have been politically intelligent to pursue; and again that his flight represents a course of action – avoidance of combat with Achilles – which has until this point justified itself as the most practical way of defending Troy. But what is decisive for Hector's akrasia is not what is the case, but what he thinks is the case; and his judgement, whether right or wrong, is that neither negotiation nor avoidance of combat is here and now morally possible for him. Hence his flight is an act of moral weakness.

³⁶ See, e.g., Sappho 102, 130 L-P, Archilochus 7, 67d, with Snell, Szenen, op. cit. (n. 1), ch. 2.

³⁷ As Snell brings out in his discussion of the play in Aischylos und das Handeln, op. cit. (n. 1).

³⁸ I mean to oppose the positions taken up by Wiggins, David in his ‘Truth, Invention and the Meaning of Life’, Proc. Brit. Acad. (1976), 331–78Google Scholar, and Williams, Bernard in his Ethics and the Limits of Philosophy (London, 1985).Google Scholar Their respective recipes for bridging the gulf between unforthcoming world and beleaguered self-invention, confidence – only succeed in making terribly clear how unbridgeable the gulf is. If your decision is not constrained, then to the extent that it is not constrained it will be (contra Wiggins, p. 373Google Scholar) arbitrary. Equally, if you are not Certain, it is no use pretending that you are.

³⁹ The paradox consists in this: when we reflect philosophically on the prerequisites for freedom, we are inclined to think that the agent must be confronted with a range of incommensurable possible courses of action, among which he is unconstrainedly free to choose, without prejudice to his rationality; but the fact remains that when we reflect on our actual practice, actions which are selected for preponderating reasons (actions which are better than their alternatives would have been) are our central cases of free, rational action.