¹ Here I follow Aristotle's bad example: he notes that the etymology of ‘choice’ (prohairesis) indicates that its object is ‘selected in preference to other things’ (pro heterún haireton; Nicomachean Ethics, 3.2.1112a16–17), but then focuses his attention, almost exclusively (though note 3.3.1112b16–17), upon means to ends.

² I set aside such reasoning as ‘I promised to do X; so I will do X.’ At least as it stands, no one could suppose that this is logically valid. It presents a reason for which the speaker intends to do X.

³ ‘How Theoretical is Practical Reason?’, in Diamond, C. and Teichman, J. (eds) Intention and Intentionality: Essays in Honour of G. E. M. Anscombe (Brighton: Harvester Press, 1979), 91–108; 98–9Google Scholar; cf. ‘Der dreifache Ziel-Bezug des praktischen Denkens’, in Pleines, J. -E. (ed.), Teleologie: ein philosophisches Problem in Geschichte und Gegenwart (Würzburg: Königshausen & Neumann, 1994), 163–82; 169Google Scholar. In one respect, the second piece may be more judicious than the first: we may agree that the thinking must hope to succeed on the ground of its content, without being persuaded that the teleology of a practical thought is ‘internal to its own content’, so that ‘the distinction between the content and the employment of a thought is of limited validity only’; ‘How Theoretical is Practical Reason?’, 99.

⁴ Yet in its specific form the deliberation can only instance what Müller analyses as ‘unreasoned purposiveness’. For it would be incoherent to reason as follows about what form the deliberation should take: ‘I will do X; a (or the) way to find out how to do this is to reflect that my intention to do X will be best realized by doing Y; so let me reflect that my intention to do X will be best realized by doing Y.’ As Müller comments, ‘This is a problem when the first order reflection is represented by a conception of its very content’; ‘Mental Teleology’, Proceedings of the Aristotelian Society 92 (1991/2), 161–83CrossRefGoogle Scholar; 166. Still peculiar would be this inference: ‘I will do X; a (or the) way to do this is to form an intention to do Y; so, I will form an intention to do Y.’ Intentions are transparent to their objects, and this is only readable as a distortion of the familiar ‘I will do X; a (or the) way to do this is to do Y; so, I will do Y.’

⁵ Intention, 2nd edn (Oxford: Basil Blackwell, 1963), §23Google Scholar.

⁶ The Oxford Dictionary of Philosophy (Oxford & New York: Oxford University Press, 1994)Google Scholar, ad ‘inference’

⁷ ibid., ad ‘deduction’

⁸ However, the distinction between ‘shall’ and ‘will’ is inverted in questions (so that ‘I will do Y’ answers ‘How shall I do X?’), in the second and third persons (so that ‘You shall do X’ is a command or expression of intention, ‘You will do X’ ususally a prediction,) and in American English. So it is only a tendency, not a rule.

⁹ Intention, 56.

¹⁰ Will, Freedom and Power (Oxford, Basil Blackwell, 1978), 39Google Scholar.

¹¹ The Possibility of Practical Reason (Oxford: Clarendon Press, 2000)Google Scholar.

¹² The point is well made, and in this connection, by Vogler, Candace, ‘Anscombe on Practial Inference’, in Millgram, E. (ed.), Varieties of Practical Reasoning (Cambridge Mass.: MIT Press, 2001), 437–64; 460Google Scholar.

¹³ Object of Thought (Oxford: Clarendon Press, 1971), 138Google Scholar.

¹⁴ ‘Practical Inferences’, in his Practical Inferences (London: Macmillan, 1971), 59–73CrossRefGoogle Scholar; 59.

¹⁵ Practical inference that share the form of B2 seen equally secure: given that p, to do X if p is to do X. It may, of course, turn out wiser to recoil from the conclusion than to abide by the first premise. Suppose that this ran, ‘We will destroy your cities if you try to destroy ours.’

¹⁶ ‘Practical Inferences’, 62–4. Very similar, about fiats, is Kenny, , Will, Freedom and Power, 81–2Google Scholar.

¹⁷ Note that the form of words ‘I do X or Y’ describes the performance not of a disjunctive act (there is no such thing), but of one of a pair of acts.

¹⁸ See MacKay, Alfred F., ‘Inferential Validity and Imperative Inference Rules’, Analysis 29 (1969), 145–56CrossRefGoogle Scholar.

¹⁹ However, the inference may be at once non-idle and innocuous if it serves to initiate deliberation about how to do X—if such is needed—that will not be put into effect until it is certain that the agent will do Y. Another example has a different structure which does not invite that qualification. Suppose I decide, ‘I will go to Paris by train.’ This entails ‘I will go to Paris’; but how acceptable as apiece of practical reasoning is ‘I will go to Paris by train; so, I will go to Paris’? Suppose that I can only go to Paris by train or plane; then that, and ‘I will go to Paris’, together entail ‘I will go to Paris by train or plane.’ So we have this inferential sequence: ‘I will go to Paris by train; therefore, I will go to Paris; therefore, I will go to Paris by train or plane.’ But ‘I will go to Paris by train; so, I will go to Paris by train or plane’ could only with great ingenuity be presented as a piece of practical inference.

²⁰ Will, Freedom and Power, 91.

²¹ ibid., 94.

²² Ethics in Aristotle (New York: Oxford University Press, 1991), 236Google Scholar.

²³ For ‘quasi-Ziele’, see Müller, ‘Der dreifache Ziel-Bezug praktischen Denkens’, 164; for ‘einschränkende Ziele’, see his ‘Wie notwendig ist das Gute? Zur Struktur des sittlichen Urteils’, in Honnefelder, L. (ed.), Sittliche Lebensform und praktische Vernunft (Paderborn: Ferdinand Schöningh, 1992), 27–57; 44Google Scholar. To quasi-ends that are both negative and stringent one may apply the term ‘side-constraint’; see Nozick, Robert, Anarchy, State, and Utopia (Oxford: Basil Blackwell, 1974), 28–35Google Scholar.

²⁴ Verbally, the agent may simply add to his judgement that doing Y is sufficient for doing X the protecting clause ‘And there is nothing that excludes my doing Y.’

²⁵ The point is made by Hare (with acknowledgement to R. F. Stalley); ‘Practical Inferences’, 65–6.

²⁶ Will, Freedom and Power, 89.

²⁷ ‘Practical Inferences’, 60–61. Strictly, however, there are no circumstances in which I can count as fulfilling the requirement.

²⁸ Will, Freedom and Power, 74–9.

²⁹ ibid., 78.

³⁰ ‘Whatever Happened to Deontic Logic?’, in Geach, (ed.), Logic and Ethics (Dordrecht: Kluwer, 1991), 33–48CrossRefGoogle Scholar.

³¹ A degenerate exception is ‘It ought never to have happened’ (loosely derivative from ‘It ought never to have been allowed to happen’). Yet I detect no ambiguity of scope within ‘We ought to do X.’ When I speak of practical or theoretical ‘ought’s or ‘must’s, I do not mean that the terms themselves are ambiguous.

³² The same point arises with telling that. I can tell you that p, or that it is the case that p because q. But within the context ‘q; so p’ I cannot both be telling you that q, and that p. If I tell you that p by saying ‘p’, I standardly intend to cause you to believe that p just through my saying ‘p’; this does not hold if I present p as an inference from ‘q; so p’.

³³ ‘The Justification of Commands’, British Journal of Educational Studies 16 (1968), 258–70; 260CrossRefGoogle Scholar. I owe the reference to Wiggins.

³⁴ To insist that a reason or ground must be a fact is not to settle the interesting question, which is what makes a certain fact a reason or ground, for an or any agent, for him or her to act in a certain way.

³⁵ I owe point and example to Wiggins.

³⁶ Note that talk of disobedience implies—what needs to be understood— that it is up to you whether you do X or not. To tell someone to die nobly is not to tell him by implication to die.

³⁷ Michael Martin puts the point to me as follows: ‘Assume that the assertibility of imperatives is tied to the desirability of what is requested; then it is easy to set up a circumstance in which “X and Y” is highly desirable but “X” is not, because “X and not-Y” is very undesirable, and is a much more likely outcome among X-situations than “X and Y”.’

³⁸ I owe the first of these to Wiggins (Augustine, In epistulam Iohannis ad Parthos tractatus, 7.8), the second to Bob Hale.

³⁹ Thus Anscombe, Intention, 3: ‘An imperative will be a description of some future action, addressed to the prospective agent, and cast in a form whose point in the language is to make the person do what is described. I say that this is its point in the language, rather than that it is the purpose of the speaker, partly because the speaker might of course given an order with some purpose quite other than that it should be executed (e.g. so that it should not be executed).’ In this way, a conventional device such the imperative mood has a function analogous to that ascribed to living things by unquantifiable ‘Aristotelian categoricals’ such as ‘Rabbits are herbivores’, which is true of the rabbit, but maybe not of every rabbit; Foot, Philippa, Natural Goodness (Oxford: Clarendon Press, 2001), 27–9CrossRefGoogle Scholar, takes both phrase and example from Michael Thompson.

⁴⁰ Though written long before, this is most accessibly printed in Hursthouse, R., Lawrence, G., and Quinn, W (eds) Virtues and Reasons: Philippa Foot and Moral Theory (Oxford: Clarendon Press, 1995), 1–34;Google Scholar the relevant pages there are 12–13, and 20–21.

⁴¹ ‘Practical Inference’, 13. Kenny came to agree: ‘Her point shows that the logic of satisfactoriness concerns merely the relations between states of affairs qua want-satisfactions: in order to be applied to the bringing about of states of affairs—and thus to become a genuinely practical logic, rather than a wish-fulfilment logic ensnaring Midases and useful only to fairy godmothers—it needs supplementing with a logic of the description of action’; Will, Freedom and Power, 84 n. 11.

⁴² ‘Practical Inference’, 13.

⁴³ ibid., 21.

⁴⁴ This should hardly surprise: purposive action is not a prerogative of homo sapiens.

⁴⁵ For help in revising this paper substantially after the lecture I am indebted to Dorothy Edgington, Stephen Everson, Michael Martin, Anselm Müller, Tom Pink, Joseph Raz, Ian Rumfitt, and David Wiggins.