1 This paper has benefited from the comments of David Pears, Thomas Hurka, and the referees and editors of this journal. I owe a very special debt to Professor J.L. Ackrill for his help and support.

2 The most extreme kind of akrasia is the kind in which the agent comes closest to believing without ambiguity or self-deception that he should not do what he does. We should agree that there are cases that it is tempting to describe in this way, even if we are convinced by the arguments that claim that the agent cannot actually have this belief.

3 Charles, David Aristotle’s Philosophy of Action (Ithaca, NY: Cornell University Press 1984)Google Scholar. Page references to this book will be included in the text.

4 References to the Nicomachean Ethics and the De Motu Animalium will be given by line number and included in the text. Translations of passages from Aristotle will be my own.

5 Nussbaum thinks that the main point of the analogy with theoretical reasoning is to claim that there is a kind of necessity in the link between the premises and the action in practical reasoning (Nussbaum, Martha Aristotle’s De Motu Animalium [Princeton: Princeton University Press 1978], 175-84Google Scholar). 701a10-13 shows that this implication of the comparison is important to Aristotle. But I think the main point of the analogy is to emphasize the importance of reasons and reasoning in explaining action — just as a new belief can be the result of a process of theoretical reasoning, an action can be the result of a process of practical reasoning.

6 I regard the examples in the De Motu Animalium as having a special status in determining what is and what is not part of Aristotle’s theory of the practical syllogism (a status they share, in my opinion, with the examples at Nicomachean Ethics 1147a5-7 and 1147a29-30, as well as with the sketches of examples at 1144a31-4 and De Anima 434a16-21). There are other samples of practical reasoning in Aristotle’s writings— for example, Nicomachean Ethics 1141b18-21 and Metaphysics 1032b6-10. But they need not be counted as being, as they stand, practical syllogisms, and our view of the practical syllogism does not have to be adjusted to accommodate all of their features.

7 Some writers who make the action the conclusion mean by this view that if the premises are true, or believed by the agent, then the action will be performed —for example, Nussbaum, 184-8; and Anscombe, G.E.M. ‘Thought and Action in Aristotle’ in Bambrough, Renford ed., New Essays in Plato and Aristotle (London: Routledge & Kegan Paul 1965) 143-58Google Scholar; see 151-2. This interpretation ignores the distinction between the practical argument and the explanation of action incorporating the argument. If there is a way of defending the literal reading of Aristotle’s claim, it will be by showing that treating the conclusion as an action is essential for understanding how a practical argument makes the case for acting in a certain way, not by arguing that the agent’s acceptance of the premises guarantees that the action will be performed.

8 I am attributing to Aristotle a demanding view of what is involved in acting rationally. In fact Aristotle thinks that someone can fall short of this standard and still be guided by reason to some extent. In discussing akrasia Aristotle considers the complex case of akrasia with respect to anger, and that is how he would describe the agent in these examples. By identifying the practical syllogism with a rational explanation of action I am not suggesting that there are strict limits on what could in ordinary language be called a ‘rational explanation’ of an action. I want to show that for Aristotle the practical syllogism is a distinctive kind of explanation, that it does not reduce to explanation in terms of desire, and that it captures a clear and central sense in which people can be said to act rationally.

9 The two syllogism interpretation is usually associated with Joachim (Joachim, H. H. The Nicomachean Ethics [Oxford: Clarendon Press 1951] 226-9Google Scholar). But I think the interpretation is also found in Sir Alexander, Grant The Ethics of Aristotle (London: Longman, Green, and Co. 1885) 4th ed., vol. II, 205-6Google Scholar; Stewart, J.A. Notes on the Nicomachean Ethics (Oxford: Clarendon Press 1892) vol. II, 156-60Google Scholar; and Walsh, J.J. Aristotle’s Conception of Moral Weakness (New York: Columbia University Press 1963) 150-8Google Scholar, although usually in a weaker version than the one I describe.

10 There are similarities between the two syllogism interpretation of Aristotle and Donald Davidson’s account of akrasia (Davidson, Donald ‘How Is Weakness of the Will Possible?’ in Essays on Action and Events [Oxford: Clarendon Press 1980]21-42Google Scholar). Davidson thinks the akrates is confronted with two practical arguments. His reasoning against the action does not reach its conclusion but stops at the merely conditional judgment that, with reference to all the reasons that he has considered, the action should not be performed. The reasoning in favor of the action reaches the unconditional conclusion that the action should be performed. The agent acts on the conclusion of the second argument. However Davidson, unlike the two syllogism interpretation, thinks that the akrates is irrational in acting in the way he does, even though his action is explained by practical reasoning (41-2). Davidson himself is aware of the similarities. His own explanation of akrasia respects principles that give a unified account of acting intentionally, acting on a desire, and acting because of practical reasoning, and Davidson thinks that Aristotle’s explanation is constrained by similar assumptions (31-2; the similarities are concealed by Davidson’s criticism of Aristotle’s view of the logic of practical reasoning on 33-4). I will argue that the two syllogism view is mistaken as an interpretation of Aristotle, and that Aristotle’s account of akrasia rejects the assumption that acting because of practical reasoning and acting because of a desire amount to the same thing.

11 Charles gives many different arguments for his view of VII3. I cannot consider the points that he makes, but it is significant that his conclusion depends on some other parts of his view. He thinks that the second part of 1147b11-12 (‘this a man either does not have when he is feeling passion, or has it in the sense in which having does not mean knowing but speaking, as a drunken person may quote Empedocles’) applies to the agent’s grasp of the conclusion of the syllogism against the action (120-1). Even if we agree that Aristotle is talking about the conclusion and not the final premise of the syllogism, Charles must still argue that this passage does not qualify the sense in which the agent believes the conclusion. Appealing to his view that Aristotle regards desire as a way of accepting propositions about the good, he suggests that 1147b11-12 says only that the agent’s desire does not decisively accept the conclusion, not that his intellectual comprehension of it is in any way weakened (167,191). If we disagree with Charles’s account of desire then we will also disagree with his claim that 1147b11-12 does not describe an impairment in the agent’s grasp of the reasoning against the action.

12 This does not mean that Aristotle’s account of akrasia is ‘belief-based’ in the sense explained by Charles (161-8). In a belief-based theory the agent’s evaluative beliefs explain the state of his desires. Aristotle thinks that the intellectual failure in akrasia is caused by the strength of the agent’s desire.

13 This interpretation is defended in Anthony, Kenny ‘The Practical Syllogism and Incontinence,’ Phronesis 11 (1966) 163-84Google Scholar; see esp. 182-3. However, it is obscured by Kenny’s view that all of the reasoning described at 1147a31-4 is part of the practical syllogism against the action. Some remarks of Burnet (Burnet, John The Ethics of Aristotle [London: Methuen 1900]Google Scholar; see 303) point towards this interpretation in a more ambiguous way.

14 If we look outside Nicomachean Ethics VII3 itself, I think there is very strong evidence that the akrates does not believe that the particular action he is performing should be performed or is the best thing to do: Nicomachean Ethics 1136b6-9, 1151a20-4, 1152a4-6, Eudemian Ethics 1223b7-9, 1224b1-2, 1224b19-21.

15 There is a more fundamental objection to my argument. At De Motu Animalium 700b28-30 Aristotle says: ‘We should believe that an apparent good may take the place of an actual good, and so may the pleasant, since it is an apparent good.’ It might be claimed that this means that a practical syllogism can begin from a premise about the pleasant instead of a premise about the good. If so, the reasoning at 1147a32-3 would, as it stands, be in order as an example of a practical syllogism. However, I think the passage should be understood in a different way. Aristotle is saying that we tend to have mistaken evaluative beliefs about the goodness of pleasure or of particular kinds of pleasure, and that these beliefs explain some of our actions. In these cases the practical syllogism will begin with a principle about what is good, so it remains significant that the reasoning at 1147a32-3 does not start with an explicitly evaluative belief.

It is a difficulty for my interpretation that the De Motu Animalium uses practical syllogisms to explain the behavior of animals as well as human actions. This might seem easier to understand if a practical syllogism could begin with a simple desire for pleasure rather than an evaluative belief about pleasure. But given Aristotle’s strongly teleological view of nature, it is possible that he thinks that we can explain the movements of animals as though they did reason about the good.

16 David Charles thinks that Book VII chapter 6 denies that a syllogism is present in akrasia with respect to epithumia because the agent acts without explicit deliberation (96). His view contradicts 1149a35, and it is difficult to see this as a difference between akrasia with respect to thumos and akrasia with respect to epithumia. The person whose motive is anger is just as likely to act without stopping to think things through. Aristotle does not explain what his distinction rests on, but he probably thinks that an evaluative belief is part of anger in a way in which it is not part of a simple desire for pleasure (at least, not if the desire is epithumia). If an action is explained by anger then it is also explained, at least partially, by the evaluative belief, but we cannot say the same when the action is explained by a desire for pleasure.

17 In this respect Aristotle’s account of akrasia seems to me more reasonable than Davidson’s. According to Davidson although the akrates believes that the reason for the akratic action is outweighed by the reasons against it, he nevertheless concludes on the basis of that reason that the action would be the best thing to do (Davidson, 39). Consequently when the akrates acts he can be properly described as acting for that reason. But if the akrates sees clearly that the reason is outweighed by the opposing reasons, then it cannot be in virtue of its importance as a reason that this consideration motivates the action. So the action should not be explained by practical reasoning that starts from that reason and reaches a strong evaluative conclusion in favor of acting. The problem is not solved by adding, as Davidson does (42), that the akrates recognizes his own irrationality in reasoning in this way. The agent cannot both acknowledge that the reason is outweighed and conclude from its strength as a reason that the action would be best.

18 We can make this concession and still preserve the distinction between thumos and epithumia explained in note 15. The difference need not consist in the presence of an evaluative belief in the one case but not the other; it might be a matter of the relationship between the evaluative belief and the desire.

19 As far as the De Motu Animalium itself is concerned, I think that ordinary belief is the mode of acceptance of the premises and conclusion in Aristotle’s examples. He uses the expression ‘he thinks’ (‘noesei;’ 701a14) to introduce one of the premises, and in comparing theoretical reasoning with practical reasoning it would be strange for him to switch from ‘he thinks that …’ in the theoretical case to ‘his desire accepts that …’ in the practical case without saying something to mark the change. Charles thinks that the example at 701a31-3 supports his view (94). He suggests that when Aristotle says ‘desire takes the place of questioning or thinking’ he means that in practical reasoning desire takes over the role played by belief in theoretical reasoning, the role of accepting the propositions that constitute the practical syllogism. I think the example makes a point that only applies to actions performed without explicit deliberation (701a29-30). In these cases desire provides one of the premises of the syllogism in the sense that we attribute the belief that he should drink to the thirsty man because he wants to drink and not because he has thought to himself that he should drink. Even in this example I think that Aristotle regards belief as the mode of acceptance of the judgments in the syllogism.

20 On 194-6 Charles suggests that the goal that the akrates irrationally frustrates is supplied by the nature of practical reasoning rather than the nature of desire. He does not attribute this view to Aristotle, but if he did it would face a similar objection. Someone might refuse to engage in practical reasoning, so that he did not accept its goal as his own goal. If Charles were right this person would not be irrational. But I think that Aristotle would say that the person was irrational if he knowingly acted against the good.

21 The discussion is difficult to follow because Charles supposes that the alternative to the view that desire explains the action is the view that the propositions in the reasoning themselves explain the action. A more reasonable alternative view that he does not consider would say that the agent’s belief in those propositions explains the action.

22 With respect to the connection between premises and conclusion inside the syllogism, Charles suggests in Appendix 3 (263) that the best way to understand the notion of validity appropriate to means-end practical reasoning is to explain it in terms of the fact that a desire for an end will typically generate a desire for what is seen as a means to that end.

23 Charles thinks that explanation by a practical syllogism can be understood as a kind of teleological explanation, and that it is compatible with there being a causal explanation of the same action in terms of desires (198-202). He thinks that his view that desire is the acceptance of an evaluative proposition helps to establish the compatibility (200). I would agree that Aristotle thinks the two explanations amount to the same thing in the case of actions that can be rationally explained. It is not necessary to understand desire in Charles’s way to make this claim. But I think that Aristotle’s view is that the two explanations come apart over akrasia. Charles applies his schema for teleological/rational explanation to the akratic action, but only by including as an extra condition for the application of the schema the fact that the agent’s desire decisively accepts the weak evaluative judgment that the akratic action has a good feature (201). I think Aristotle would use a schema for teleological explanation in which the evaluative beliefs were themselves strong enough to explain the action without the addition of Charles’s extra condition. He would draw the conclusion that the akratic action cannot be explained in this way.