¹ Presumably it had been handed down from medieval philosophy (I do not know by what route), for Aquinas makes a similar threefold distinction in his Summa Theologica, II/II, Q. 1, art. 2 and Q. 2, art. 1. St Thomas said that faith is mid-way between science and opinion: fides is a propositional attitude which differs from scientia (rational knowledge derived from first principles) and from opinio (uncertain or probabilistic belief about matters of fact). Faith is the acceptance of propositions which can only be revealed; the assent of faith is voluntary; yet faith involves firm commitment, unlike opinion. Aquinas's treatment of the epistemological trio itself derives from Hugh of St Victor, who worked in Paris in the preceding twelfth century: he wrote ‘Faith is a form of mental certitude about distant realities that is greater than opinion and less then science’ (De sacramentis I. 10. 2) – see the footnote on p. 11 and appendix 4 by O'Brien, T. C. in Aquinas, St Thomas, Summa Theologiae, vol. 31 (London: Eyre & Spottiswoode, 1974)Google Scholar.

² Kant continued to use the trio of meinen, glauben and wissen as a framework in which to organize his thoughts on epistemology. See the Blomberg Logic (24: 148ff, 228ff.), the Vienna Logic (24: 850ff.), the Dohna-Wundlacken Logic (24: 732ff.), and the Jasche Logic (9: 66ff.) – all in Kant's Lectures on Logic, translated by Young, J. Michael (Cambridge: Cambridge University Press, 1992)Google Scholar . These lecture-notes by various hands can hardly be accorded the same authority as Kant's published works, but they provide useful supplementary evidence of the development of his thinking. The Jasche Logic was prepared for publication by Jasche with the approval of the ageing Kant, so it can be taken with more confidence as representing how he treated these topics in later years. ‘Logic’ in eighteenth-century usage covered much more than (Aristotelean) formal logic; it included topics in what we now call epistemology, theory of meaning and philosophy of mind. See Kitcher, Patricia, Kant's Transcendental Psychology (Oxford: Oxford University Press, 1990)Google Scholar, ch. 1; and the Preface to Logic and the Workings of the Mind, ed. Easton, Patricia A. (Atascadero, CA: Ridgeview, 1997)Google Scholar.

³ Rendered as ‘opining’ by Kemp Smith in his translation of the Critique of Pure Reason (London: Macmillan, 1929)Google Scholar, and by J. M. Young in translating the logic lecture-notes.

⁴ In everyday usage, people tend to think of knowledge and belief as mutually exclusive - for example, when Jung said ‘I don't believe that God exists, I know it’. But such cases can be treated in terms of Gricean conversational implication: in most contexts, when someone says ‘I believe p’, he conversationally implies that he does not know that p, although that is not logically entailed. For to say one believes p in a situation in which one knows p (or thinks one does) is to mislead one's hearers into inferring that one thinks one has insufficient grounds for p.

⁵ , Kant'sCritique of Pure Reason, trans. Guyer, Paul and Wood, Allen (Cambridge: Cambridge University Press, 1998), p. 687 n.Google Scholar

⁶ , Kant'sCritique of Judgment, trans. Pluhar, Werner (Indianapolis: Hackett Publishing Company, 1987), p. 360, n.75.Google Scholar

⁷ Kant was not very careful about the distinction between a disposition and its actualizations. ‘Here the talk is not of truth but of holding-to-be-true… This is judgment in relation to the subject' (Dohna-Wundlacken Logic, 24: 731-2); ‘the judgment through which something is represented as true … is, subjectively, holding-to-be-true'' (Jasche Logic, 9: 65-6).

⁸ Descartes allows that the will can be inclined by ‘reasons of truth or goodness’, or ‘a divinely-produced disposition of my inmost thoughts'; yet he maintains that ‘neither divine grace nor natural knowledge ever diminishes freedom; on the contrary, they increase and strengthen it’ (Fourth Meditation, Adam and Tannery, VII: 58).

⁹ ‘No truth appears to me more evident, than that beasts are endow'd with thought and reason as well as men… ‘Tis … by means of custom alone, that experience operates upon [the animals]. All this was sufficiently evident with respect to man’ (A Treatise of Human Nature, I. III. xvi). Hume allows however that reflection, as well as mere custom or habit, can influence the formation and strength of our beliefs. For a fuller treatment of this topic see my paper ‘Freedom of judgment in Descartes, Spinoza, Hume and Kant’, forthcoming in British Journal for the History of Philosophy.

¹⁰ See the Blomberg Logic (24: 156ff) - where Kant is reported as saying ‘a procedure of giving our approval, or withdrawing it, or holding it back, does not rest at all on our free choice, but rather is necessitated through and by the laws of our understanding and our reason’: see also the Vienna Logic, 24: 859ff., the Dohna-Wundlacken Logic, 24: 736ff. – ‘In suspensio judicii there lies some freedom’ – and the Jasche Logic, 9: 74ff.

¹¹ Groundwork of the Metaphysic of Morals (4: 448), trans. H. J. Paton.

¹² This topic is touched on in the first Critique at A 446/B 474, A 534/B 562ff., and A802/B830.

¹³ See, for example, A 51/B 75, A 68/B 93, B 130, B 132.

¹⁴ From the translation by J. M. Young, in Kant's Lectures on Logic.

¹⁵ Good starting points are Wilfrid Sellars, ‘this I or he or it (the thing) which thinks’ (Presidential address to the American Philosophical Association in December 1970), esp. the closing pages; and Pippin, R., ‘Kant on the spontaneity of mind’, Canadian Journal of Philosophy, 17 (1987), 449–76CrossRefGoogle Scholar . See my paper referred to in n. 9, and ‘Six levels of mentality’ , Philosophical Explorations, 51.2 (2002), 105–24Google Scholar.

¹⁶ This is the first, brief appearance of Kant's moral theology in the Critical philosophy; it is much further developed in the Dialectic of the Critique of Practical Reason, in §§86–91 of the Critique of Judgment, and in Religion within the Boundaries of Mere Reason.

¹⁷ From the Guyer and Wood translation, reproducing their emphasis for the words that were set in larger and thicker type in the first German editions of the first Critique.

¹⁸ See, for example, the Jasche Logic 9: 66.

¹⁹ I have tried to treat differences between first and third person judgements more systematically in ‘First Person Epistemology’, Philosophy, 74 (1999), 475–97CrossRefGoogle Scholar.

²⁰ See the Blomberg Logic, 24: 148ff.; the Vienna Logic, 24: 850ff.; the Dohna-Wundlacken Logic, 24: 732ff.; and the Jasche Logic, 9: 66ff.

²¹ Many subtle varieties of justification have been distinguished in recent epistemology, and in the light of the internal/external distinction it has been questioned whether there is any single conception of justification; but we need not bring in these modern distinctions here, unless they turn out to be necessary to make sense of Kant. It is clear that by ‘objective sufficiency’ he means more than just having some minimal amount of evidence (enough, perhaps, to make it not blameworthy to believe the proposition); and for Wissen he requires certainty – but it remains to be discussed what such ‘certainty’ can amount to.

²² Rather than burden the page with letters representing the subject and the proposition, as in ‘Hap ‘for person A holds proposition p to be true, I permit myself the obvious abbreviations.

²³ In his Commentary on Kant's Critique of Practical Reason (Chicago: University of Chicago Press, 1960)Google Scholar, L. W. Beck remarks that ‘subjective’ in Kant does no t mean arbitrary and contingent; it just means ‘dependent upon the nature of the subject’ (p. 256). (He goes on to say, obscurely, that this can be interpreted either a priori or posteriori: perhaps he was alluding to the distinction between species-dependence and individual-dependence.)

²⁴ Allison, H. E., Kant's Transcendental Idealism (New Haven: Yale University Press, 1983), ch. 1.Google Scholar

²⁵ Wood, Allen, Kant's Moral Religion (Ithaca, NY: Cornell University Press, 1970), pp. 14–16.Google Scholar

²⁶ Ibid., footnote on p. 34.

²⁷ From the translation by Allen Wood in , Kant'sReligion and Rational Theology (Cambridge: Cambridge University Press, 1996)Google Scholar.

²⁸ From the translation by J. M. Young, in Kant's Lectures on Logic.

²⁹ From the translation of the Critique of Judgment by Pluhar. However, that Kant's distinctions here are not primarily ontological is suggested by the preceding sentence: ‘The question whether something is a cognizable being or no t is no t a question concerning the possibility of things in themselves but concerning the possibility of our cognising them.’

³⁰ Kant ha d touched on the topic of testimony in earlier lectures, implying that it can, in favourable cases, be a source of well-justified beliefs which amoun t to empirical knowledge. See the Blomberg Logic, 24: 30–1, 245, and the Dohna-Wundlacken Logic, 24: 749–50.

³¹ From the translation by J. M. Young, in Kant's Lectures on Logic.

³² Kant is closer to Locke (in time, and in doctrine), but Kant is prepared to say that we can have knowledge of the material world by induction and testimony, where Locke would say we can only have a lesser grade of probability called belief, assent, or opinion (Essay Concerning Human Understanding, IV xv. 3). And Kant has a distinctively practical account of religious faith, whereas Locke still treats it a special kind of theoretical belief, based on the alleged testimony of God Himself (Essay, IV. xvi. 14; IV xviii. 2).

³³ From the translation by J. M. Young, in Kant's Lectures on Logic.

³⁴ See the Blomberg Logic, 9: 228, where the middle degree of holding to be true is characterized as ‘to believe, or to hold something to be true to such a degree that it is sufficient for action and for deciding to act'; and 9: 241–2, where it is said ‘one can opine something without believing… Here I hold something to be true without its having an influence on our actions.’

³⁵ But there may be no uncontroversial way to assign numerical values to degrees of belief in general, aside from the easy cases of finite numbers of equiprobable outcomes, as in the fall of a fair dice. (Not everyone will be prepared to bet about anything!)

³⁶ The degree of someone's belief may also depend on their aversion to risk, for example, someone may be unwilling to bet his last ducat, even on very favourable odds. Not all the values involved may be measurable in monetary terms.

³⁷ Allen Wood compares Kant with Kierkegaard and Pascal on the personal nature of faith – see Kant's Moral Religion, p. 16 and p. 252.

³⁸ For further comment on the use of meinen, glauben, wissen and other words in the eighteenth- and nineteenth-century German epistemic vocabulary, especially for the association of Glaube with religious faith, see the entries on ‘belief, faith and opinion’ and ‘knowledge, cognition and certainty’ in Inwood, M., A Hegel Dictionary (Oxford: Blackwell 1992)CrossRefGoogle Scholar . Inwood makes some connections between Kant and Hegel on these matters.

³⁹ I make a first shot at assessing Kant's philosophy of religion in ‘Is there any hope for Kant's account of religion?’, in Proceedings of the Ninth International Kant Kongress (Berlin: Walter de Gruyter, 2000)Google Scholar.

⁴⁰ I am grateful to three anonymous referees for the Kantian Review, whose comments enabled me substantially to improve this article.