I wish to clarify my thoughts on the zombie argument for dualism. There are two issues I've puzzled over lately: 1. Chalmers's defense of epiphenomenalism, and 2. Whether the proper interpretation of the zombie argument is the disjunctive: Type-D (interactionist dualism), Type E (epiphenomenalism), or Type-F (panprotopsychism) as Chalmers claims, or--as his opponents claim--the only option, if the zombie argument goes through, is Type-E dualism, i.e. epiphenomenalism.
I discuss the latter first.
To recount, the zombie argument goes like this: If we can conceive of physically identical versions of ourselves that nonetheless lack consciousness, then such beings are logically possible. If such beings are logically possible, consciousness is not logically entailed by physical facts alone. Hence, physicalism is false and some form of dualism is true.
The forms of this true dualism on the table are as follows: 1. Type-D dualism: interactionist dualism. Consciousness, though non-physical, can have effects on the physical world. 2. Type-E dualism is epiphenomenalism, in which consciousness has no effect on the physical world. 3. Type-F dualism is panprotopsychism, in which the essence of the physical is in fact consciousness, or protoconscious properties. I'm not familiar with Type-F dualism, and will not discuss it in depth.
NB: I am not sure where Leibniz's parallelism would fit. It is closest to, but not, Type-E, and should probably be classed alone as Type-P.
Type-E dualism is the easiest target, because it leads to the odd question of how we're able to talk about consciousness if it has no causal power. Assuming this criticism is good—as later I will deny—the opponent of the zombie argument would be much advantaged if he could prove the only possible outcome of the argument is Type-E dualism, and not the more defensible full disjunctive of Types D through F.
The argument to show that only Type-E dualism is an option to the zombie proponent goes like this: the zombie argument requires a mirror of the physical world without consciousness. If consciousness has physical effects, we cannot mirror the physical world without it. Ergo, the only option left to the zombie proponent is the brand that contains a consciousness that has no effect on the physical world--Type-E dualism.
Chalmers, in correspondence, denies the second line in the argument: "If consciousness has physical effects, we cannot mirror the physical world without it." This is wrong, says Chalmers, because we can mirror the physical world sans consciousness even if consciousness has physical effects: the result will be a world where some physical events occur, but without causes (those events that would have been caused by consciousness were consciousness not removed from the world). Odd, yes, but it remains logically possible--which is all the zombie argument requires to go through.
When Chalmers made this point to me, I was unsure if it worked, because I wasn't sure that effects without causes were logically possible. The more I reflect on it, however, the more I think Professor Chalmers is correct. I can conceive of effects without causes. Indeed, as I pointed out to Chalmers, if the universe we have had a beginning, it might very well have been a causeless effect.
The result is that the dualist can retain casually efficacious consciousness, even through the zombie argument. He has all options on his plate: Types D through F.
As to the second issue, that of Type-E dualism and Chalmers's defense, remember the question is "How can we talk about consciousness if consciousness has no effect?"
Chalmers's move is to deny a causal theory of reference: to wit, he denies that we need to have a causal connection to things we reference. Chalmers's zombie twin says: "I am conscious" and is wrong. Chalmers says "I am conscious" and is right. Even if his own consciousness has no connection to what Chalmers is saying, what Chalmers says is nonetheless true. His consciousness has not caused the belief, it is the belief. Ergo, says Chalmers, epiphenomenalism remains viable.
I see nothing wrong with the argument. The only objections I've found have been ones insisting that a causal theory of reference must be true. Richard Chappell, on his blog and in correspondence, points out that it surely is not—we can speak of, for instance, the dead space outside the boundaries of our light cone, or things that don't exist, like unicorns. Neither of these can have any causal influence on us, and yet we refer to them easily.
I know of no counter.
If these defenses succeed, the metaphysical landscape is as follows: the zombie argument, if its premises are correct, proves that physicalism is false and some form of dualism is true. Type-D, Type-E, and Type-F dualisms are all possible candidates. Type-E dualism, though superficially flawed, remains viable.
It should be noted that Chalmers admits the zombie argument is weaker than the inverted spectrum argument, et al, so we are dealing with one of the more easily criticized arguments for dualism. Nonetheless, I see no flaw.
Chalmers's extensive analysis of the zombie argument, among others, occurs in his The Conscious Mind. This is also where he presents his defense of epiphenomenalism. One can also find a more specific presentation in his paper, The Content and Epistemology of Phenomenal Belief, available here. Richard Chappell has an easily understood summary of that argument on his own blog, here.
For a more thorough explanation of varieties of dualism and materialism, see Chalmers's paper Consciousness and its Place in Nature, available online here.
Chappell's argument against a causal theory of reference is on his blog, here.