¹ Kripke, S., Naming and Necessity, 2nd edn. (Cambridge: Harvard University Press, 1980), p. 117.Google Scholar

² Hawkins, J. M. and Allen, R. (eds.), Oxford Encyclopedic English Dictionary (Oxford: Oxford University Press, 1991), p. 606.Google Scholar

³ For convenience, I cite Kant's works in parentheses. These citations normally include an abbreviation of the English title and the corresponding volume and page numbers in the standard ‘Akademie’ (Ak.) edn. of Kant, 's works: Kants gesammelte Schriften, 29 vols., ed. the Koeniglich Preussischen (now Deutschen) Akademie der Wissenschaften (Berlin: G. Reimer (now de Gruyter), 1902–00)Google Scholar. There are two exceptions to this convention. First, for references to the first Critique I follow the common practice of giving page numbers from the A (1781) and B (1787) editions. I also consulted the following German composite edition: Kritik der reinen Vernunft, ed. Weischedel, W., Immanuel Kant Werkausgabe 3 (Frankfurt: Suhrkamp, 1968)Google Scholar. And secondly for references to Kant's Reflexionen, the 10 vols. of Kants handschriftlicher Nachlass, I give the entry number in addition to the Ak. volume and page nos. When quoting from Kant's works I stay fairly close to the standard English translation(s), but modify or retranslate them whenever it seems appropriate; and the translations of the Reflexionen are my own. Here are the relevant English abbreviations and translations:

BL The Blomberg Logic, in Immanuel Kant: Lectures on Logic, tr. Young, J. M. (Cambridge: Cambridge University Press, 1992), pp. 1–246Google Scholar.

CJ Critique of Judgement, tr. Meredith, J. C. (Oxford: Oxford University Press, 1952)Google Scholar.

CPR Critique of Pure Reason, tr. Smith, N. Kemp (New York: St Martin's, 1965)Google Scholar.

Critique of Pure Reason, tr. Guyer, P. and Wood, A. (Cambridge: Cambridge University Press, 1997)Google Scholar.

GMM Grounding for the Metaphysics of Morals, tr. Ellington, J. (Indianapolis, IN: Hackett, 1981)Google Scholar.

JL The Jäsche Logic, in Immanuel Kant: Lectures on Logic, tr. Young, J. M. (Cambridge: Cambridge University Press, 1992), pp. 519–640Google Scholar.

MFNS Metaphysical Foundations of Natural Science, tr. Ellington, J. (Indianapolis, IN: Bobbs-Merrill, 1970)Google Scholar.

P Prolegomena to Any Future Metaphysics, tr. Ellington, J. (Indianapolis, IN: Hackett, 1977).Google Scholar

R Reflexionen = Kants handschriftlicher Nachlass.

⁴ Nietzsche, , The Portable Nietzsche, tr. Kaufmann, W. (Harmondsworth: Penguin, 1976), p. 9.Google Scholar

⁵ Kripke, , Naming and Necessity, p. 39.Google Scholar

⁶ Ibid., p. 123 n. 63.

⁷ Ibid., p. 138.

⁸ The microphysical and the microscopic should not be confused; see n. 16.

⁹ See Kripke, , Naming and Necessity, p. 48Google Scholar; and Stanley, J., ‘Names and rigid designation’, in Hale, B. and Wright, C. (eds.), A Companion to the Philosophy of Language (Oxford: Blackwell, 1997), pp. 555–85, at 556.Google Scholar

¹⁰ See Kripke, S., ‘Identity and necessity’, in Moore, A. W. (ed.), Meaning and Reference (Oxford: Oxford University Press, 1993), pp. 162–91Google Scholar, at 162–3.

¹¹ This glosses over some fairly subtle points. Suppose that it is metaphysically necessary that water is H₂O; still, it seems conceivable and logically possible that there are worlds in which water is something other than H₂O. See Putnam, H., ‘Is water necessarily H₂O?’, in Realism with a Human Face (Cambridge: Harvard University Press, 1990), pp. 54–79.Google Scholar But if this is so, then metaphysical necessity is not logical necessity. See Farrell, R., ‘Metaphysical necessity is not logical necessity’, Philosophical Studies,-39 (1981), 141–53.Google Scholar Orthodox Kripkeans handle this difficulty by rejecting the step from conceivability to possibility; see Yablo, S., ‘Is conceivability a guide to possibility?’, Philosophy and Phenomenological Research, 53 (1993), 1–42.Google Scholar Neo-Kripkeans, by contrast, also insist that necessity is univocal but then distinguish between two different types of intensions of ‘water’. The ‘primary' intension (which expresses narrow or individualist cognitive content) is purely descriptive and takes us from possible worlds ‘considered as actual’ to whatever conforms to the relevant description in those worlds; and the “secondary” intension (which expresses wide or externalist cognitive content) is purely referential and takes us from possible worlds ‘considered as counterfactual’ to whatever is identical to the relevant referent in those worlds. The necessity of a proposition according to the primary intension is ‘deep necessity’ or a priori conceptual necessity, and the necessity of a proposition according to the secondary intension is ‘superficial necessity’ or a posteriori metaphysical necessity. See Davies, M. K. and Humberstone, L., ‘Two notions of necessity’, Philosophical Studies, 38 (1980), 1–30Google Scholar; and Chalmers, D., The Conscious Mind (New York: Oxford University Press, 1996), pp. 52–69.Google Scholar On this two-dimensional modal semantics, ‘Water is the watery stuff’ is deeply necessary according to the primary intension while ‘Water is H₂O’ is deeply contingent; and ‘Water is H₂O’ is superficially necessary according to the secondary intension while ‘Water is the watery stuff’ is superficially contingent.

Kant's modal semantics is significantly different from both orthodox Kripkean and neo-Kripkean modal semantics: it ties apriority tightly to necessity; it accepts the step from conceivability to possibility; and it is dualist rather than two-dimensional. For details, see Hanna, R., Kant and the Foundations of Analytic Philosophy (Oxford: Oxford University PressCrossRefGoogle Scholar, forthcoming), esp.chs. 2, 3 and 5. On the other hand, for the purposes of this article it is important to note a close formal similarity between Kant's notion of analyticity and the neo-Kripkeans' deep or a priori conceptual necessity according to the primary intension.

¹² Hanna, R., ‘A Kantian critique of scientific essentialism’, Philosophy and Phenomenological Research, 58 (Sept. 1998), 497–528.Google Scholar

¹³ On Kant's view as I am construing it, things are directly perceivable or observable if and only if they can at least in principle be made intuitionally present to the conscious human senses without a necessary dependence on concepts, beliefs, inferences, theories or theory-driven detection technology (for example, cloud chambers or cyclotrons); otherwise, they are nothing but ‘entities of the understanding’ (Verstandeswesen) (B306) or purely theoretical entities. See also section 3.

¹⁴ Officially stated, transcendental idealism (TI) is the view ‘that appearances are to be regarded as being, one and all, mere representations and not things in themselves, and accordingly that space and time are only sensible forms of our intuition, but not determinations given for themselves or conditions of objects as things in themselves’ (A369). Most contemporary philosophers find TI implausibly strong. In response, however, a contemporary Kantian could offer a weakened transcendental idealism (WTI): nothing counts as a real thing unless it is possible for humans to represent it veridically as an appearance; and nothing counts as space or time unless it is an a priori form of human sensory intuition. On WTI, neither real things nor space and time are things-in-themselves, because conformity to the structure of the human mind is a necessary condition of their existence; but on the other hand, real things are not ‘merely’ representations, nor are space and time ‘only’ forms of our intuition. For although real things, space and time are the proper satisfiers of our representations of them, they can still exist even if human beings do not actually exist — assuming that human beings are at least possible. It seems to me that every Kantian argument that is valid on TI is also valid on WTI, and also that at least some Kantian arguments that are valid but unsound on TI are sound on WTI.

¹⁵ For a clear and succinct statement of modern noumenal scientific realism, see, for example, Smart's, J. J. C.Philosophy and Scientific Realism (London: Routledge & Kegan Paul, 1963).Google Scholar

¹⁶ Microphysical entities, structures, processes and forces should be carefully distinguished from microscopic entities, structures, processes and forces — although, to be sure, both are small and invisible. Microscopic items are small in an anthropocentric sense and contingently invisible. That is, they are small in relation to us, to the extent that they are not visible to the naked human eye. Yet they are visible to the assisted human eye. So in principle microscopic items are directly perceivable by us and thereby count as members of the manifest or phenomenal world. By contrast, microphysical items are small in a strictly mathematical sense and necessarily invisible. That is, by virtue of some physical theory they are assigned quantities that bear determinate lesser-than relations to quantities assigned by our phenomenally applicable systems of measurement. But microphysical items cannot be seen by us even through the most powerful microscopes and are not in principle directly observable, since they are nothing but ‘entities of the understanding’ (Verstandeswesen) (B306) or purely theoretical entities. See section 3 and also Smart, , Philosophy and Scientific Realism, pp. 37–8.Google Scholar

¹⁷ I borrow this apt term from Johnston, Mark; see his ‘Manifest Kinds’, Journal of Philosophy, 94 (1997), 564–83.CrossRefGoogle Scholar

¹⁸ Non-reductive theories hold that phenomenal properties are not identical to microphysical properties. Strongly non-reductive theories hold that phenomenal properties are neither identical to nor strictly determined by (for example, supervenient on) microphysical properties. If I am correct that Kant is a strongly non-reductive scientific realist, then Margaret Wilson is wrong when she claims that ‘for Kant, what is empirically real is primarily the material world of the science of his time — a world that does not possess colors, tastes, and the like in any literal, irreducible sense'; see ‘The “phenomenalisms” of Kant, and Berkeley, ’, in Wood, A. (ed.), Self and Nature in Kant's Philosophy (Ithaca, NY: Cornell University Press, 1984), pp. 157–73Google Scholar, at 161. But to reject Wilson's influential interpretation is not to say that Kant is a Berkeleyan phenomenalist who thinks that the spatiotemporal world is a private construct from mental qualia. My view is that Kant's empirical scientific realism is both strongly non-reductive and also fully intersubjective (hence non-Berkeleyan). See section 3. In this connection see also Aquila, R., ‘Kant's Phenomenalism’, idealistic Studies, 5 (1975), 108–26.Google Scholar

¹⁹ For the classic statement of this, see Eddington, A., The Nature of the Physical World (Cambridge: Cambridge University Press, 1929).Google Scholar

²⁰ See Sellars, W., ‘Philosophy and the scientific image of man’ and ‘Empiricism and the philosophy of mind’, both in Science, Perception, and Reality (New York: Humanities Press, 1963), pp. 1–40Google Scholar and 127–96, esp. p. 173. See also Aune, B., ‘Sellars's two images of the world’, Journal of Philosophy, 87 (1990), 537–45.CrossRefGoogle Scholar

²¹ I mean that the stuff can occur in separated bits or pieces (gold rings), and also impurely or in alloy (white gold).

²² In the two introductions and the second half of the Critique of Judgement Kant argues that LIFE and ORGANISM are not ordinary empirical concepts, and that they have no direct application to natural beings (see, CJ, Ak. 5:385–401); instead, they are mere ‘regulative’ or ‘hypothetical’ concepts of reason — heuristic concepts for the unification and promotion of natural scientific inquiry (A642–7/B670–5).

²³ Here I am using the A version of this phrase instead of B's less explicit version.

²⁴ Since the Kantian conceptual Inhalt has both a decompositional micro-structure and an immanence in human consciousness, it is both a finegrained and hyper-finegrained intensional entity: two Inhalte sharing the same cross-possible-worlds extension, or Umfang, can still differ in internal structure (B4); and two Inhalte sharing the same internal structure and Umfang can still differ in ‘viewpoints’ (Gesichtspunkte) (JL, Ak. 9:147) or modes of presentation.

²⁵ Identification by means of concepts is not complete determination or individuation: identification picks out things as actual or possible tokens of a type, not as fully determinate or individuated items. Otherwise put, identification does not satisfy Leibniz's Laws because it picks out only necessary features of things, not sufficient features. Indeed, Kant holds that complete determination of a thing can occur only by way of intuition (A581–2/B609–10; see also JL, Ak. 9: 99).

²⁶ In other words, analyticity for Kant is determined not solely by concept-decomposition, but also by comprehensional inclusion. In most cases, these are equivalent. But in a few crucial cases, propositions are analytic through comprehensional inclusion, but not through concept-decomposition: for example, ‘Triangulars are trilaterals’ (A716/B744; see also BL, Ak. 24:115 and JL, Ak. 9:60–1 ). In turn, both concept-decomposition and comprehensional inclusion fall under Kant's general criterion of analyticity, according to which a proposition is analytic if and only if its denial leads to a contradiction (A151/B190–1). See Hanna, , Kant and the Foundations of Analytic Philosophy, section 3.1.Google Scholar

²⁷ See Kroon, F. and Nola, R., ‘Kant, Kripke, and gold’, Kant-Studien, 78 (1987), 442–58Google Scholar, at 449–51.

²⁸ See Smith, E., ‘Concepts and thought’, in Sternberg, R. and Smith, E. (eds.), Psychology of Human Thought (Cambridge: Cambridge University Press, 1988), pp. 19–49.Google Scholar

²⁹ See Putnam, H., ‘The meaning of “meaning”’, in Mind, Language, and Reality: Philosophical Papers, Vol. ii (Cambridge: Cambridge University Press, 1975), pp. 215–71CrossRefGoogle Scholar, at 249–52 and 269. If by the word ‘concept’ one intends Putnam's notion of a ‘property’ (that is, a strictly rough-grained intension), then a stereotype or prototype is not a concept; on the other hand, if by ‘concept’ one means a Kantian empirical concept (or its close relative, Frege's notion of the ‘sense’ or Sinn of an ordinary predicate expression), then a stereotype is at least concept-like. In fact, however, the nearest analogue to a stereotype in Kant's theory is the ‘empirical schema’ or ‘representation of a universal procedure of imagination in providing an image for a concept’ that must be added to the intension of an empirical concept in order to apply it to actual cases (A140–2/B179–81).

³⁰ Kroon and Nola, ‘Kant, Kripke, and gold’, 451–6. Propositions reflecting the decompositional structure of a stereotype or prototype cannot be genuinely analytic. For if they were, then they would be necessarily true and strictly a priori. On the Kroon—Nola account, and also on Putnam's account, they are at best contingent and contextually a priori.

³¹ See n. 12.

³² As I mentioned in n. 29, Kant holds that the intension of an empirical concept must be supplemented by an empirical schema in order to be applicable to actual objects. So for Kant every objectively real empirical concept contains (i) a logical essence or conceptual core made up of analytic characteristics, (ii) a conceptual periphery made up of synthetic empirical characteristics and (iii) a schema, stereotype or prototype.

³³ Kant is not, however, committed to the view that it is going to be easy — even relatively easy — to tell whether a given characteristic is analytic or synthetic. That is an epistemic issue which must be worked out case-by-case, and there will be tricky or borderline cases. Is ADULT an analytic or synthetic characteristic of the concept BACHELOR? Are little boys bachelors or not? What Kant is committed to is that every ordinary empirical concept has some absolutely necessary analytic characteristics and also some contingent synthetic empirical characteristics.

³⁴ Locke, J., An Essay Concerning Human Understanding, ed. Nidditch, P. H. (Oxford: Clarendon Press, 1975), book II, chs. 22–3, pp. 288–317Google Scholar, and book III, chs. 5–6, pp. 428–71.

³⁵ There is a small interpretive puzzle concerning Kant's views as to precisely which characteristics count as analytic characteristics of GOLD. He says that ‘one man can think in the concept of gold, in addition to its weight [that is, HEAVY or HIGH DENSITY], color [that is, YELLOW], malleability, also its property of resisting rust, while another man can perhaps scientifically know nothing of this quality’ (A728/B756). What he seems to be saying is that the concept GOLD analytically contains the characteristics HEAVY/HIGH DENSITY, YELLOW, and MALLEABLE/DUCTILE, while RUST-RESISTANT remains a merely synthetic and contingent characteristic of GOLD. Does this make any sense? Why would ‘low density gold’, ‘colorless gold’ and ‘unmalleable gold’ be analytic contradictions, while ‘rusty gold’ is analytically consistent? I am strongly inclined to think that Kant believes that RUST-RESISTANT is an analytic characteristic of GOLD. Here we must remember that Kant's theory of concepts is not solipsistic: the content of a given empirical concept is determined for an indefinitely large intersubjective community of human cognizers spread out over time and space, each of whom shares the same formal unity of apperceptive consciousness and the same set of cognitive capacities — not merely for an actual individual cognizer at a time. This is at least part of what Kant means when he says that every concept is a conceptus communis (B133–4). So my reading of this passage is that Kant is saying that heaviness/high density, yellow, malleability/ductility and rust-resistance all equally belong as analytic characteristics to GOLD: but just which of those characteristics is or are self-consciously taken to be important or notable in fact varies across individual concept-users and situations. The second concept-user in Kant's example is ignorant or forgetful of some basic definitional knowledge about gold (that it is rust-resistant), which shows that his personal grasp of the concept GOLD is somewhat loose.

³⁶ My attention was drawn to the presence of the indexicai component in Kant's analysis of natural kind concepts by Anderson, Erik. See also his ‘Kant, natural kind terms, and scientific essentialism’, History of Philosophy Quarterly, 11 (1994), 355–73Google Scholar, in which he gives an account of its semantic role that is different from the one I work out here.

³⁷ See Kaplan, D., ‘Demonstratives’ and ‘Afterthoughts’, both in Almog, J. et al. (eds.), Themes from Kaplan (New York: Oxford University Press, 1989), pp. 481–563Google Scholar, and 565–614; and Perry, J., ‘Indexicals and demonstratives’, in Hale, and Wright, (eds.), Philosophy of Language, pp. 586–612.Google Scholar

³⁸ Strawson, , The Bounds of Sense: An Essay on Kant's Critique of Pure Reason (London: Methuen, 1966), pp. 48–9Google Scholar and p. 49 n. 1.

³⁹ I do not mean to imply that I accept Strawson's interpretation of the transcendental aesthetic in toto.

⁴⁰ See Heal, J., ‘Indexicai predicates and their uses’, Mind, 106 (1997), 619–40.Google Scholar

⁴¹ Here is another small interpretive puzzle. In Metaphysical Foundations of Natural Science, Kant argues that infinite divisibility is a synthetic feature of actual matter (Ak. 4: 503–4); elsewhere, of course, he holds that divisibility is an analytic characteristic of the concept BODY. Assuming charitably that Kant is not simply contradicting himself, then either BODY and MATTER are different concepts, or else he is implicitly drawing a distinction between infinite divisibility and divisibility per se. Since the former seems unlikely, I opt for the latter. Hence matter's being divisible is analytic; but its being infinitely divisible is synthetic. That makes sense, because quantitative properties for Kant always depend upon pure forms of intuition, and of course there are possible worlds in which matter is only finitely divisible.

⁴² See Westphal, K., ‘Kant's proof of the law of inertia’, in Robinson, H. (ed.), Proceedings of the Eighth International Kant Congress, vol. ii, part 1, sections 1–9 (Milwaukee, WI: Marquette University Press, 1995), pp. 413–24.Google Scholar Since Kant identifies life with spontaneity, ‘vital gold’ could not be accurately described by causal natural laws, or support true law-expressing counterfactuals.

⁴³ Interestingly, Kripke fudges on this point; see Naming and Necessity, pp. 117–18, and 123. Putnam on the other hand assigns METAL to what he calls the ‘semantic marker’ of the word ‘gold’ ('The meaning of “meaning”’, pp. 266–8). Putnam says that the semantic marker is a central feature of the speaker's linguistic competence for that word; but even so, according to him in principle something could be gold but fail to be a metal.

⁴⁴ For Kant, causal laws are synthetically necessary and a priori, but their necessity is constrained by something actual: ‘That which in its connection with the actual is determined in accordance with universal conditions of experience is…necessary’ (A218/B266, first emphasis added). My reading of this text is that for Kant causal laws are true in all and only those possible worlds of human experience that also contain our actual world totality of lifeless matter. So the strong modality of causal laws is constrained by the empirical concept BODY or MATTER (MFNS: Ak. 4:472). It must be admitted, however, that Kant's theory of causal laws is very controversial; see Hanna, , ‘Kantian critique’, p. 525Google Scholar, n. 58, for some references to the secondary literature.

⁴⁵ Kripke, , Naming and Necessity, p. 138.Google Scholar

⁴⁶ Idib., pp. 139–40.

⁴⁷ See, for example, Quine, W. V. O., ‘Natural kinds’, in Ontological Relativity (New York: Columbia University Press, 1969), pp. 114–38.Google Scholar

⁴⁸ See Brown, J., ‘Natural kind terms and recognitional capacities’, Mind, 107 (1998), 275–302.Google Scholar Brown associates human recognitional capacities for natural kinds with an ability to pick out microstructural kinds. But in fact her account would work just as well for manifest kinds.

⁴⁹ For a start, however, see Bennett, J., Locke, Berkeley, Hume: Central Themes (Oxford: Oxford University Press, 1971)Google Scholar; Hacker, P. M. S., Appearance and Reality (Oxford: Blackwell, 1987)Google Scholar; Hirst, R. J., ‘Primary and secondary qualities’, in Edwards, P. (ed.), Encyclopedia of Philosophy (New York: Macmillan, 1967), vol. vi, pp. 455–7Google Scholar; and McGinn, C., The Subjective View (Oxford: Oxford University Press, 1983).Google Scholar

⁵⁰ This needs a further comment. Kripkean scientific essentialists are non-reductionist about secondary qualities in general and phenomenal consciousness in particular. See Kripke, , Naming and Necessity, pp. 144–55Google Scholar; and Chalmers, , The Conscious Mind, chs. 3–4.Google Scholar It may seem possible to defend both SE and either physicalist reductionism (the thesis that mental properties are identical to physical properties) or eliminativism (the thesis that mental properties are nothing but mythic constructs of folk psychology that will eventually wither away – like phlogiston and the ether). See Smart, , Philosophy and Scientific Realism, chs. 4–5Google Scholar; and Churchland, P. M., Matter and Consciousness (Cambridge, MA: MIT Press, 1985), pp. 43–9.Google Scholar But on the one hand physicalist reductionism and SE are inconsistent, since the former holds that true psychophysical identity statements are merely contingent while the latter holds that all true identity statements are necessary. And on the other hand, eliminativism reduces to a speculative wager on the epistemic and cultural efficacy of future neuroscience. I should add that I do not mean to assert that SE cannot be consistently combined with other forms of reductionism (for example, reductive functionalism). But the logical status of that combination is very much a moot point at present; see Chalmers, , Conscious Mind, chs. 6–7.Google Scholar

⁵¹ More precisely, on the Boyle-Locke view, primary qualities produce ideas exactly resembling their causes, while secondary qualities produce non-resembling and privately individuated ideas.

⁵² There are some important similarities here between my conception of Kant's empirical realism and that offered by Langton, Rae in Kantian Humility (Oxford: Oxford University Press, 1998).Google Scholar We both agree that for Kant all properties of noumenal objects are intrinsic, that all properties of phenomenal objects are relational, and that Kant rejects the primary versus secondary quality distinction by assimilating both kinds of qualities to relational phenomenal properties. I disagree only with her construal of Kant's phenomenalism, which is quite close to Wilson's; see n. 18.

⁵³ See McDowell, , Mind and World (Cambridge, MA: Harvard University Press, 1994)Google Scholar, lecture 4.

⁵⁴ To say that this level of explanation is irreducible to physical causal explanation is to say, like Davidson, that human actions are anomalous; see Meerbote, R., ‘Kant on the nondeterminate character of human actions’, in Harper, W. A. and Meerbote, R. (eds.), Kant on Causality, Freedom, and Objectivity (Minneapolis: University of Minnesota Press, 1984), pp. 138–63.Google Scholar It does not follow, however, that Kant is committed to the Davidsonian token-token identity of mental events and physical events, as Meerbote asserts, since Kant is not committed to the thesis that something is the cause of a physical effect only if it instantiates a strict law of nature. Rather Kant is committed only to the weaker thesis that something is the physical cause of a physical effect only if it instantiates a strict law of nature. But this is consistent with holding that there are non-physical causes of physical effects. See n. 55.

⁵⁵ Even if a given physical cause and a given physical effect are related by their instantiating a strict natural law, this is consistent with the counter-factual claim that if the cause had not also had a further non-physical property (transcendental, practical or moral freedom of the will) then the physical effect would not have happened. And counterfactual dependency is arguably a criterion of causation of any sort, mental or physical. See Lewis, D., ‘Causation’ and ‘Postscripts A-F to “Causation”’, both in Philosophical Papers, vol. ii (New York: Oxford University Press, 1986), pp. 159–72Google Scholar, 172–213, and ‘Causation as Influence’, Journal of Philosophy, 97 (2000), 182-97. The counterfactual dependency of physical effects on non-physical causes in turn is possible only if physical events can also have non-physical properties, which is ruled out by micro-physical noumenal realism. Phenomena can have both physical and non-physical properties, but something that is essentially physical cannot have non-physical properties.

⁵⁶ The soundness of the Refutation is of course much disputed. But I think that on at least one charitable interpretation, it is sound; see Hanna, R., ‘The inner and the outer: Kant's “Refutation” reconstructed’, Ratio, 13 (2000), 146–74.Google Scholar

⁵⁷ For a critique of the thesis that perceptual qualities are how things appear to normal observers under normal conditions, see Hacker, , Appearance and Reality, ch. 3.Google Scholar

⁵⁸ This is of course Thomas Nagel's influential characterization of phenomenal consciousness; see ‘What is it like to be a bat?’, in Mortal Questions (Cambridge: Cambridge University Press, 1979), pp. 165–80Google Scholar.

⁵⁹ For a different although also non-Lockean reading of Kant's theory of sensation, see Falkenstein, L., ‘Kant's account of sensation’, Canadian Journal of Philosophy, 20 (1990), 63–88.CrossRefGoogle Scholar

⁶⁰ It must be admitted that the entire text at B69–70n. is fiendishly difficult to interpret, since in it Kant seems to say that redness is both a real and an illusory property — that to ascribe redness to a rose is both unlike and like the mere fallacy of ascribing ‘handles’ to Saturn. And of course Kant may simply be confused here. On the other hand, interpreting him charitably, I think that if we sharply distinguish between the Object itself’ (Objekte selbst) or real empirical object (the ‘empirical thing in itself’), and the ‘object in itself’ (Objekte an sich selbst) or noumenal object (the ‘transcendental thing-in-itself’), then it is quite correct and intelligible to say that colour properties are real properties of the object itself (for an idealized human sensory constitution) yet only merely illusory properties of the object-in-itself (the noumenon). The same would of course be true of spatial and temporal properties, as Kant explicitly points out: ‘on the other hand, if I ascribe redness to the rose in itself, handles to Saturn, or extension to all outer objects in themselves, without paying any regard to the determinate relation of these objects to the subject, and without limiting my judgement to that relation, illusion then first arises’ (B70n.). The tendency to regard all manifest properties as illusory belongs to noumenal realism, which is the flip side of ‘empirical idealism’ (A369) or ‘problematic idealism’ (B274), namely Cartesian external world scepticism.

⁶¹ According to physical monadology, the elements of matter are unextended dynamical atoms, or centres of force. See Laywine, A., Kant's Early Metaphysics and the Origins of the Critical Philosophy (Atascadero, CA: Ridgeview, 1993), ch. 3.Google Scholar

⁶² According to the standard conception of supervenience, X supervenes on Y if and only if Y is sufficient for X and there can be no changes in X without corresponding changes in Y. That is, X supervenes on Y if and only if Y strictly determines X even though X may not be literally identical to Y. Many microphysical noumenal scientific realists hold that the phenomenal world supervenes on the physical world. If I am right, Kant's view is the exact converse of this. It is therefore not only strongly non-reductive: it also offers a radically anti-microphysical metaphysics of nature.

⁶³ This is what I call the ‘concept-independence’ of Kantian intuition; see Hanna, , Kant and the Foundations of Analytic Philosophy, section 4.2.Google Scholar

⁶⁴ See n. 16. There was considerable confusion about the microscopic versus microphysical distinction in early modern philosophy, and this has carried over to some extent into contemporary philosophy. See Wilson, C., The Invisible World: Early Modern Philosophy and the Invention of the Microscope (Princeton, NJ: Princeton University Press, 1995), esp. chs. 1, 6–8.Google Scholar

⁶⁵ Directed by Roger Corman, 1963. Also known as X.

⁶⁶ I am not talking about magic or miracles, for example, clairvoyance and mental telepathy (A222/B270). The sensory developments I am talking about must at least be consistent with causal laws of nature.

⁶⁷ The term Kant actually uses is ‘logic’. But it is important to recognize that pure general logic for him includes not only the extensional (monadic) logic of truth and consequence but also an intensional logic of irreducible modal operators and fine-grained decomposable concepts with cross-possible-worlds extensions — and this amounts to conceptual analysis in our sense.

⁶⁸ This is to allow for mixtures and alloys. For example, water is a clear liquid, but we also learn by empirical investigation that there is a lot of water in tea and human blood. What we learn is that tea and blood are mostly composed of what would otherwise be pure water. Moreover we call lake water ‘water’ because it is a (relatively) clear liquid, but do not call tea or blood ‘water’ because it is opaque — that is, dark brown or dark red — even though there is much more water in tea or blood than any other element. (Weak tea is not a counter-example, only a borderline case; indeed, we generally speak of ‘watery tea’ in such cases.) These points provide a rather clear, crisp Kantian solution to a recent debate as to whether the everyday application of a natural kind term tracks microphysical science or is relativized to current human interests. See Chomsky, N., ‘Language and nature’, Mind, 104 (1995), 1–61Google Scholar, at 22-23; Abbott, B., ‘A note on the nature of “Water”’, Mind, 106 (1997), 311–19Google Scholar; and LaPorte, J., ‘Living water’, Mind, 107 (1998), 451–5.Google Scholar The Kantian answer is ‘neither': the application of the concept follows the phenomenological cross-world identification conditions built decompositionally into the concept expressed by the natural kind term.

⁶⁹ It would be a mistake to think that conceptual analysis historically precedes scientific knowledge of natural kinds: in fact, they arise simultaneously. That is, natural kind concept-formation and the scientific investigation of natural kinds are two sides of the same epistemic process. See LaPorte, J., ‘Chemical kind term reference and the discovery of essence’, Nous, 30 (1996), 112–32.Google Scholar

⁷⁰ See Burge, T., ‘Reason and the first person’, in Wright, C. et al. (eds.), Knowing our own Minds (Oxford: Clarendon Press, 1998), pp. 243–70.Google Scholar Oddly enough, although Burge's account of first-person epistemic authority is clearly indebted to Kant's theory of apperception, he takes a rather dim view of conceptual analysis in Kant's sense (p. 261 n. 11). But I think Burge overestimates both the cogency and the scope of Quine's critique of analyticity. For a defence of Kant's theory and a critique of Quine, see Hanna, , Kant and the Foundations of Analytic Philosophy, sections 3.1 and 3.5.Google Scholar

⁷¹ See Hanna, , ‘How do we know necessary truths? Kant's answer’, European Journal of Philosophy, 6 (1998), 115–45.CrossRefGoogle Scholar

⁷² See, for example, Bloor, D., Knowledge and Social Imagery, 2nd edn. (Chicago: University of Chicago Press, 1991).Google Scholar

⁷³ See Williams, B., Descartes: The Project of Pure Inquiry (Atlantic High lands, NJ: Humanities Press, 1978), pp. 65–8.Google Scholar

⁷⁴ The sceptic says: if the objects of physics are merely theoretically posited or inferred entities, then nothing will conclusively show that they exist in the external world (A368). And the relativist says: if the sensory given is a myth, that is, if all theories are underdetermined and ultimately unconstrained by empirical evidence, then anything goes (A236–8/B295–98).

⁷⁵ Karsten Harries drew my attention to this important point. He also argued that, according to the Kantian conception of science, natural science must adopt a methodological absolute conception of nature in order to have an adequate theory of truth. But it seems to me that, according to Kant, an adequate theory of truth requires only the weaker idea of an externally actual world — a ‘truth-maker’ — that provides cognitive orientation. Indeed, the concept of truth must ultimately answer not to pure theoretical reason but instead to the demands of practical reason. See Hanna, , ‘Kant, truth, and human nature’, British Journal for the History of Philosophy, 8 (2000), 225–50.CrossRefGoogle Scholar

⁷⁶ I am grateful to Jeremy Buxbaum, Andy Egan, Christopher Shields and a referee at Kantian Review for comments on earlier versions of this paper; to Bryan Hall and Daniel Stoljar for discussions on or around this topic; and to Adam Vinueza for a pertinent secondary source reference.