1 See Rosenthal, David ‘Two Concepts of Consciousness,’ Philosophical Studies 94 (1986) 329–59CrossRefGoogle Scholar; and Dennett's, Daniel oddly titled Consciousness Explained (New York: little, Brown 1991).Google Scholar

2 The other lengthy development of a representational theory of consciousness is in Tye, Michael Ten Problems of Consciousness: A Representational Theory of the Phenomenal Mind (Cambridge, MA: The MIT Press 1995)CrossRefGoogle Scholar. Tye's treatment will occasionally be contrasted with Dretske's, although, despite their being independently arrived at, they are in very many respects extremely similar. This is interesting, since it suggests that there are some fairly strict constraints upon the development of a representational theory of consciousness which is, I think, a virtue of the view. It is also worth pointing out that the first representational theorist of consciousness was Descartes himself, who maintained that every element of consciousness was or involved an idea and that every idea was a representation. So my consciousness was an awareness of how my ideas are representing the world to me.

3 For a more detailed defense see Tye (especially ch. 4) and Dretske (especially ch. 3).

4 The most vexed question is about the intrinsic nature of qualitative consciousness, or whether consciousness is an intrinsic feature of conscious entities. Here it is the choice of one's underlying theory of representation which forces the representational theory either towards or away from accepting intrinsicness. This will be examined below.

5 See Tye, ch. 3 for a discussion of various problems with the adverbial theory.

6 For example, what is the status of the principles which connect the adverbial properties of experience with the perceptible properties of objects? Some such principle would have to be added to the sub-argument by the adverbialist. Perhaps it would read so: sometimes when I experience red-object-in-front-of-mely there is a red object in front of me. If so, we require an infinitude of principles connecting these predicates to their correlates in the perceived world. And the second premise of the sub-argument is now true only in a Pickwickian sense; in fact, objects are never really red in the sense that they have the property of which I become aware in experience, for I am only ever aware of properties of experiences (this is the other, unfortunate side of the coin which is the very point of the adverbial theory).

7 Just when children can begin to introspect depends upon their conceptual development, and in particular it depends upon when they acquire some understanding of the nature of the mind, especially its representational nature. There is evidence that at least a basic form of such an understanding is acquired around the age of three to four years. See Perner, Josef Understanding the Representational Mind (Cambridge, MA: The MIT Press 1991)Google Scholar and Gopnik, Alison ‘How Do We Know Our Minds: The illusion of First Person Knowledge of Intentionality,’ Behavioral and Brain Sciences 16 (1993) 1–14.CrossRefGoogle Scholar

8 Notwithstanding that it is exactly contrary to a certain powerful philosophical tradition perfectly exemplified by the passage in Locke where ‘consciousness’ is defined: ‘Consciousness is the perception of what passes in a man's own mind’ (Locke, John An Essay Concerning Human Understanding. 1690. Page references are to the Nidditch edition [Oxford: Oxford University Press 1975], bk. 2, ch. 1, at 115)Google Scholar. But surely this is at best an attempt at a definition of self-consciousness or introspective consciousness rather than consciousness itself. Locke's definition would seem to imply that animals, for example, are entirely unconscious, which is extremely implausible (but see Carruthers, Peter ‘Brute Experience,’ Journal of Philosophy 89 [1989] 258–69CrossRefGoogle Scholar for a modem defense of a version of this Cartesian doctrine). It is, however, quite plausible to assert that animals are entirely unselfconscious. The representational theory accounts for this distinction very nicely.

9 Thus Dretske is led to endorse the rather implausible ‘ability’ analysis of knowing what it is like advanced by Nemirow, Laurence in his ‘Review of Thomas Nagel's Mortal Questions Philosophical Review 89 (1980) 473–7Google Scholar and ‘Physicalism and the Cognitive Role of Acquaintance,’ in Lycan, William ed., Mind and Cognition: A Reader (Cambridge, MA: The MIT Press 1990)Google Scholar; and Lewis, David ‘What Experience Teaches,’ in Copley-Coltheart, J. ed., Proceedings of the Russellian Society (Sydney: University of Sydney Press 1988)Google Scholar; reprinted in Lycan. It seems to me preferable to have a unified theory of introspective knowledge. Tye, who takes a line very much like the one offered here, calls the ability analysis ‘very counterintuitive’ (175). In fact, the approach outlined above to ‘knowing what it is like’ offers a substantial objection to the ability analysis, as given below. For more criticism of the ability analysis see Seager, William Metaphysics of Consciousness (London: Routledge 1991), ch. 5.Google Scholar

10 Is it possible to imagine an experience so novel that it fits under no determinable concept? No, because any experience can be indexically specified as that experience (as opposed to the more determinate that color, that sound, etc.)

11 Just what makes a representation into an experience is a lurking question here; it is not simply a matter of the representational function of the systems at issue. Of course, this is nonetheless a crucial problem which we'll examine below.

12 It is not just through sustained, specialized use that a system's meta-function of altering its function may be invoked. Another possible mechanism is suggested by an hypothesis of Profet, Margie (Protecting Your Baby-to-be: Preventing Birth Defects in the First Trimester [Reading, MA: Addison-Wesley 1995)Google Scholar) that food aversions developed by pregnant women are adaptations to protect the fetus from various toxins in common foods. Women's olfactory systems, on this hypothesis, become hypersensitive during pregnancy — a beautiful analogue to the transition from a K to J device. It is a virtue of the bio-functional version of the representational theory that such hypotheses fit so naturally into its framework. However, the idea that no taste qualia are shared by a woman and her pre-pregnant self seems rather extreme.

13 The answer is to be found in the explanation of the discrimination device's current capabilities. For example, on Millikan's account (roughly) the function of a device D is to perform A iff ancestors of D actually did A and the explanation of why D does A is because these ancestors did A. So a device, D, would have the function of altering its functional precision in way cp iff the ancestors did alter their discriminatory precision in way cp and this explains D's precisional alteration. It seems entirely reasonable to imagine that it would be more efficient to build a device that could alter its functional precision rather than build in all possible levels of precision, especially if the device is one whose detailed sensory requirements cannot easily be specified in advance. That is, devices like us.

14 But note that as Dretske explains it (68), for something, x, to phenomenally look cp to S requires that (1) x looks the way ϕs normally look to Sand (2) x looks different from non-ϕs to S (i.e. Scan discriminate ϕs from non-ϕs). Dreske notes that this definition is not circular so long as we take ‘looks the same’ and ‘looks different’ as primitive (see 176, n. 1). The legitimacy of this move is doubtful since there is clearly a similar distinction at work within these terms. X and y doxastically look the same to S just in case there is a cp such that S takes x and y both to be ϕ (S might say ‘they both look like dogs to me’). Dretske must have in mind as the primitive notion ‘phenomenally looks the same (different).’ To me, this looks suspiciously like ‘indiscriminable on the basis of the way the objects look,’ where the ‘look’ here must, unfortunately, be ‘phenomenally look’ (mere indiscriminability will obviously not ground the notion of looking the same or different since two objects can look exactly alike while one is beeping, the other silent).

15 I was drawn to this example by a reference to it in Dennett, but note that Dennett gets the taster/non-taster proportions exactly backwards. Dennett's original source, Jonathan Bennett, gets it right, though Bennett makes the taster/non-taster proportion appear to be more precise and less variable than it actually is (see Bennett, Jonathan ‘Substance, Reality and Primary Qualities,’ American Philosophical Quarterly 2 [1965] 1–17).Google Scholar

16 There is a difficulty right here, which I will not go into beyond this note. The notion of implicit representation presupposes that the represented properties fall into a ‘family’ of coherently related elements. The appropriate notion of a family of properties is not very clear. The examples used suppose that the represented property comes from a set that possesses a clear mathematical ordering but it is far from clear that all ‘qualia properties’ meet this condition.

17 A representationalist might reply that the ptu tasters are actually misrepresenting the taste of ptu — misrepresenting it as having the property which more paradigmatically bitter substances possess and which both tasters and non-tasters of ptu can taste. Ptu tasting would then be likened to widespread visual illusions such as the Milller-Lyer. This response depends upon ptu (and the family of substances tasters can taste) being indiscriminable from other bitter substances. I don't know if this is so. In any case, the response is not very plausible; we do not, I think, want to say that it is an illusion that saccharine or aspartame are sweet.

18 Dretske says that he leans towards Godfrey-Smith's account of function. The point remains true with Godfrey-Smith's ‘modem historical’ account of functions.

19 I take it that the generation problem is also what David Chalmers calls the ‘hard problem of consciousness’ (Chalmers, David ‘Explaining Consciousness: The Hard Problem,’ Journal of Consciousness Studies 2 [1995] 200–19).Google Scholar

20 Swampman can engage in a huge range of ‘behaviors’ that, to say the least, threaten the idea that he is entirely unconscious and unthinking. See Brown, Deborah ‘Swampman of La Mancha,’ Canadian Journal of Philosophy 23 (1994) 327–48CrossRefGoogle Scholar for an interesting discussion.

21 There is also the possibility that I am an artifact, in which case the fact that I have functioning representations within me will, I suppose, depend upon the intentions of my creators. I leave, as an exercise for the reader, the development of this line of thought into an argument for theism.

22 Note that a champion of the bio-functional approach, Ruth Millikan, explicitly endorses the claim that this cannot be evident when she says: ‘we do not have … certain knowledge via Cartesian reflection, even of the fact that we mean, let alone knowledge of what we mean or knowledge that what we mean is true’ (Language, Thought, 93) and ‘absolutely nothing is guaranteed directly from within an act of consciousness’ (ibid., 92).

23 Furthermore, on the bio-functional version of the theory it is only such assemblages of matter that also have certain sorts of (evolutionary) histories that possess this remarkable and presumably causal power.