In ‘The conscious mind’, David Chalmers (1996) develops a multifaceted argument to show that the phenomenal nature of conscious experience is not reducible to functional properties of a physical substance. For Chalmers, this conclusion is inevitable if one is to ‘take seriously’ the justified belief that we are phenomenally conscious (what he calls ‘phenomenal realism’). The argument turns, in particular, on the logical possibility of zombies that share their physical/functional relational properties (hereafter P/F properties) with us, but experience no phenomenal properties (hereafter Φ-properties). It can be summarized as:
1 Zombies are conceivable (‘epistemic gap’)
2 If zombies are conceivable, they are (metaphysically) possible
3 Zombies are possible (‘ontological gap’)
4 Φ-properties do not logically supervene upon the P/F realm.
5 There are non-P/F properties.
This conclusion is generally taken to mean that physicalism is false.
(2) has been the most debated premise (e.g., Tye, 1995, Levine, 1993, Loar, 1997, Papineau, 2002)2, but the more fundamental divide is with those philosophers who reject premise (1) (e.g., Dennett, 1991, 2005, Churchland, 1996). The grounds for rejecting premise (1) amount to the claim that it is a misunderstanding of the phenomenal which gives rise to the belief that zombies are possible. And the error, it is argued, is analogous to that made by vitalists who claim that a system could have all the appropriate biochemical properties without being alive. The mistake is to assume there is something ineffable (Dennett, 1988) that remains unaccounted for, once a description of what it is to be conscious has been given in terms of P/F properties. The claim is therefore that the phenomenal can and must be conceptualized in terms of functional features of a physical system. I shall not examine these various criticisms, but take premises (1) and (2) are true, and assume that the argument is valid as it stands.
Note that Chalmers’s PD includes the following additional claim. If the property of psychological awareness (hereafter Ψ-awareness) characterizes ‘a state wherein we have access to some information, and can use that information in the control of behavior’ (Chalmers, 1996, p. 28), properties of Ψ-awareness are assumed to logically supervene upon P/F properties. They are therefore independent of Φ-properties. This supervenience claim is however not fundamental to PD; I shall indicate where I make use of it.
A key feature of PD is that it rests upon the assumption that we have a justified belief that we are phenomenally conscious. This has implications for the relation between Ψ-properties and Φ-properties. For it requires that we have cognitive access to, i.e., that we be Ψ-aware of the contents of our Φ experience (Chalmers, 1996, p. 221). This does not imply that ‘to have an experience is automatically to know about it’ (ibid., p. 197). Rather, ‘we have the ability to notice our experiences’ (ibid., p. 221), i.e., to make the contents of our Φ experience into the object of a cognitive belief. Such a belief corresponds to what Chalmers calls a second-order phenomenal judgment, such as the judgment that my Φ experience is currently that of a red object. This is a judgment about the content of the first-order phenomenal judgment ‘It’s red’.
Such phenomenal judgments and their verbal expression lead to difficulties that Chalmers discusses at some length. One of these concerns the issue of self-knowledge. Φ-properties are irrelevant to the causal explanation of behavior, on the assumption of the closure of the physical realm3: this is the quasi-epiphenomenalism of Chalmers’s position.4 And Chalmers notes the difficulties raised by this quasi-epiphenomenalist position: ‘for second- and third-order phenomenal judgments (…), explanatory irrelevance seems to raise real problems’ (ibid., p. 182). Indeed, given the zombie has identical behavior to mine, he therefore makes the same claims about being Φ-conscious (causal closure of the physical realm), and, on the assumption of logical supervenience of Ψ upon P/F properties, he forms the same phenomenal judgments5 as I do. How do I know, therefore, that I am not a zombie? Chalmers answers that it is precisely because I haveΦ experience. This provides the epistemic warrant for my belief. Such a warrant characterizes the epistemology of conscious experience. Below, I shall identify a logically possible situation which raises problems for this epistemology.