^{page 351 note 1} This task I reserve for a further article on Kant's Distinction between Categorical and Hypothetical Imperatives.

^{page 351 note 2} Hume, , Treatise of Human Nature (Selby-Bigge), p. 415Google Scholar.

^{page 351 note 3} P. 459.

^{page 352 note 1} Treatise of Human Nature, pp. 459–60Google Scholar.

^{page 352 note 2} P. 460.

^{page 353 note 1} Treatise of Human Nature, p. 458Google Scholar.

^{page 354 note 1} Treatise of Human Nature, p. 458Google Scholar.

^{page 354 note 2} Bk. II, Pt. Ill, Sect. III.

^{page 354 note 3} P. 399.

^{page 354 note 4} P. 413.

^{page 354 note 5} P. 4I3.

^{page 355 note 1} Treatise of Human Nature, p. 414Google Scholar.

^{page 355 note 2} P. 414.

^{page 355 note 3} Bk. II, Pt. Ill, Sect. I.

^{page 355 note 4} p. 416.

^{page 355 note 5} P. 416.

^{page 356 note 1} This compliment to Hume presupposes the neglect of irreducibly practical reason not only by Hume but also by the moralists whom Hume attacks. And, while Hume has been allowed to speak for himself, the moralists whom Hume attacks have been convicted without a hearing. But Hume attacks “the greatest part of moral philosophy, ancient and modern” (p. 413). We cannot hear all this. Most ethical rationalism, however, is, like that of Socrates and Plato, unashamedly based on ethical intellectualism. Of “those who affirm that virtue is nothing but a conformity to reason” (p. 456) most, instead of trying to exhibit reason as practical, are content to exhibit virtue as theoretical. Aristotle's distinction of moral, from intellectual, virtue is a protest against the intellectualism of his predecessors. But Aristotle, although he tries, does not succeed in disengaging rationalism from intellectualism. I have examined his position in an article on Rationalism and Intellectualism in the Ethics of Aristotle now ready for publication.

^{page 358 note 1} A course excluding x.

^{page 360 note 1} Discursive reason is opposed to intuitive reason as a different specification of theoretical reason. Discursive reasoning is inferring. Accordingly, I do not say that choosing is discursive reasoning. But I do say that choosing is reasoning in exactly the same sense of “reasoning.”

^{page 360 note 2} As its ⋯ποτελεύτησις (Sophist, 264b.).

^{page 361 note 1} E. N., 1142a, 31–32. “ζήτησις” would be a good enough name for the genus of which inquiry and deliberation are the species. But that is not what Aristotle means.

^{page 361 note 2} Signor Mussolini is reported to have attempted by decree to settle, without finding out, the answer to the question of the identity of the Rubicon.

^{page 361 note 3} Unfortunately, we are allowed to say even that a man should know whether to do this or not, instead of saying that he should be able reasonably to decide whether to do this or not.

^{page 362 note 1} To translate Kant's “Sollen” by “ought” is to miss the distinction between theoretical and practical reason.

^{page 363 note 1} Not necessarily other in kind.

^{page 365 note 1} I am not suggesting that this principle of valid choice is true. Nor am I here claiming that the syllogistic principle of valid inference is true. It suffices for my purpose that these principles, even if false, are sufficiently plausible to serve as illustrations.

But why alongside of an obviously true principle of valid inference do I place an obviously false principle of valid choice? Why not, rather than the less plausible principle of the fulfilment of the agent's strongest desire, select, say, the more plausible principle of the fulfilment of the agent's promise? Because I am not sure that either this or any equally plausible principle is a fundamental principle. Now, if it is a derivative principle, it should be formulated, not: x can be validly chosen on the ground that x would fulfil the agent's promise, but: x can be validly chosen on a ground validly inferable from the premiss that x would fulfil the agent's promise. And this, however true, is not a genuine principle of choice. It is a principle whose adoption involves not only practical, but also theoretical, reason. The principle of the fulfilment of the agent's strongest desire, on the other hand, could hardly be considered, whether favourably or favourably, as anything but a fundamental principle. Granted that we do not ask whether, we do ask why, the agent ought to fulfil his promise. Granted that we do ask whether, we do not ask why, the agent ought to fulfil his strongest desire.

Perhaps I ought in candour to add two confessions, (I) That x would fulfil the agent's strongest desire, even that x would fulfil any of the agent's desires, is, I am convinced, a good reason for choosing x, although it may be outmatched by a better reason against choosing x. (2) I am inclined to think that there is only one true principle of valid choice and that this is an elaboration of the principle of the fulfilment of the agent's strongest desire. Against the claims of actual desires the agent must, if his choice is to be valid, weigh the claims of potential desires, desires which he would experience if he fully entertained their objects. At least this measure of elaboration is indispensable. I am not convinced of its sufficiency.

^{page 366 note 1} To analyse or define moral tightness is not to analyse or define tightness. But what is usually meant by saying in an ethical context that rightness is unanalysable or indefinable is that moral rightness is so. Taking the adjective “moral” to distinguish not a species of lightness but a sense of “rightness,” philosophers hold themselves at liberty to drop the adjective and to use the unqualified word “rightness” in this sense. Thus Professor Broad says that “the kind of appropriateness and inappropriateness which is implied in the notions of ‘right’ and ‘wrong’ is, so far as I can see, specific and unanalysable” (Five Types of Ethical Theory, p. 165)Google Scholar. And I do not deny what Professor Broad here says. But I do deny what I think Professor Broad here means. For I think that by “the notions of ‘right’ and ‘wrong’ ” he here means those of moral right and moral wrong. He would not otherwise proceed to the question “Can the term ‘right’ be analysed into a combination of other, and not specifically ethical, terms?” (p. 166). Cf. Ross, , Foundations of Ethics, pp. 42–3, 52, 315–16Google Scholar; Sidgwick, , Methods of Ethics (7th ed.), pp. 31–33Google Scholar; Rash-dall, , Ethics, pp. 12–13Google Scholar.

Since neither the term “validity” nor the term “choice” is an ethical term, my definition of moral rightness is what the Provost of Oriel would call “naturalistic” (Foundations of Ethics, p. 6). But my definition is very different from any that he examines. And I do not think he would classify my definition either as an “attitude theory” or as a “consequence theory.” To the conclusion that moral rightness is indefinable the Provost says that he is “led by the break-down of the attempts to define rightness which we have considered.” He is careful to add: “That-they have broken down does not prove that every attempt must break down, but it creates a strong presumption that it will.” Dare I hope that my attempt may survive the searching examination to which the Provost has subjected others?

^{page 367 note 1} But the relevant possibility is not that of knowingly, but that of wittingly, doing what is wrong. An inference necessitates, upon raising the question, either the judgement that the inference is valid or the abandonment of the inference. The question is whether, a choice is similarly related, not to the knowledge, but to the judgement that the choice would be right. To this question I now answer: No. Unlike inference, choice is compatible with doubt whether it is right and even with full conviction that it is wrong. Once convinced, moreover, of the possibility of wittingly doing what is wrong, we may reduce the possibility of knowingly doing what is wrong to the possibility of knowing what is wrong. It is only qua judgement that knowledge can influence conduct.