Tuesday, July 29, 2008

Rockwell's Anti-Zombie Argument

Over at The Splintered Mind, Teed Rockwell has put up a brief defense of an argument that zombies are impossible. The argument is this:

(1) Zombies are possible if and only if subjective experiences are epiphenomenal.
(2) Subjective experiences are epiphenomenal if and only if we have direct awareness of them.
(3) There is no such thing as direct awareness.

Therefore, etc.

My comments:

Ad (1). This is not true. The closest truth in this vicinity is:

(1`) If it is possible that subjective experiences are epiphenomenal, then zombies are possible.

I can't see why subjective experiences would have to be epiphenomenal in the actual world in order for zombies to be exist in some other world. Why couldn't it be the case that qualia have some functional role in our cognitive architecture in the actual world, but the non-qualitative aspects of cognition subserve that functional role in another world? Perhaps the idea is that properties are individuated by the causal powers they grant when instantiated, so that P = Q only if, necessarily, Px causes p iff Qx causes p. Then a property is actually epiphenomenal iff necessarily epiphenomenal. This is implausible to me, though. Surely some property bestows some causal powers in the actual world that it does not in other worlds. But if that is the case, then qualia could be properties of this sort. They could bestow functional cognitive properties on conscious thinkers in the actual world that they do not in other worlds. Anyway, all this is OK for Rockwell's argument so far, since we only need something of the strength of (1`) for the argument to go through, if (2) and (3) are true.

The real problem is with (2). I suppose what would make (2) plausible is an argument to the effect of:

(4) We are either directly aware of a property's being instantiated, or we infer it.
(5) We infer some fact p on the basis of q only if p causes q.
(6) The instantiations of epiphenomenal properties do not cause anything else.
Therefore, (2).

Then, from (3), we get Rockwell's desired conclusion. As things stand, I lean towards saying that epiphenomenalism is actually false, but might have been true. By (1`), this immediately gets us that zombies are possible. But let's assume that epiphenomenalism is true, and even that (1) is true. In this sort of position, my beef has always been with (5). I always thought that, if epiphenomenalism is true, then we brought qualia into our ontology to explain the justification we have for believing self-ascriptions of certain mental states. The qualitative character of the state of, say, seeing an apple from ten feet away in decent, neutral light is supposed to be what explains how I generally know, or am justified in believing, that I see an apple when I do. This is not to say that the instantiation of a certain quale causes me to believe that I see the apple. Rather, the instantiation of the quale (perhaps given some other conditions about my proper mental functioning) is what provides that belief with its positive epistemic status. But it would be strange to say that the conferral of that epistemic status is a causal matter. I want to say that the instantiation of the quale does not causally affect my belief's epistemic status because the epistemic status of a belief is not the sort of thing that can be caused to be one way or another. I do infer the existence of qualia, in general, but not from their causal roles. I infer them to explain the epistemic status that I confer on mental state self-ascriptions. So qualia are a counterexample to (5) because they are an instance in which inference to the best explanation is not inference from cause to effect.


Pete said...

I'm not sure that (1') is the closest truth in the vicinity of (1). How about(1*)?

(1*) C & Z iff E


C = the causal closure of the physical

Z = it is possible for there to be a phenomenally vacant being that is my physical/causal duplicate


E = qualia epiphenomenalism

Re (5), you don't need controversial stuff like qualia to come up with counterexamples. Plug in some stuff about the future for P and some stuff about the past or present for Q. In other words, we infer stuff about the future based on stuff that's happened so far. But its relatively noncontroversial that future events don't cause past events. The moral I draw from this is that since it's so easy to shoot down (5), whatever plausibility (2) is going to have, it's not going to be via some argument that has (5) as one of its premises.

iolasov said...

Right. (1*) is also an apparent truth in the vicinity of (1). I would change it around a little bit, though, to clear up the modal scope of C and E:

(1**) For all worlds w, (C is true at w and zombies exist at w) iff epiphenomenalism is true at w.

I say this only because, as it stands, (1*) sorta looks like the causal closure conjoined with the possibility of zombies rests on the necessity of epiphenomenalism, but that's not the case.

You're right about (5) - it's easier to show that it's false than I made it out to be. That's important. I think what I had to say about how the epiphenomenalist infers the existence of qualia is still useful, though. Even though it's probably controversial, it shows how even a restricted version of (5) that doesn't run aground on the sort of example you gave still won't be able to make a mystery out of how qualia got into the epiphenomenalist's ontology.


Pete said...

(Pete Mandik here, by the way. I'm to dopey to figure out how to make it say so up above)

Re: (1**). That is clearer than what I said. I like.

Re: (5), I've always assumed that it was epiphenomenalists and non-inferential direct awareness types on the one side and physicalist inferentialists on the other. So, I'd be pretty interested to see if you can successfully stake out this third position whereby the existence of qualia are infered yet affirmed to be epiphenomenal.

I'm not sure, exactly, how you would want to spell out this inference. Is it supposed to go like this?

1. I am justified in ascribing to myself the having of a tangy quale right now.
2. I could only be justified in self-ascribing a tangy quale if I actually have a tangy quale.
Therefore, I have a tangy quale right now.

Or is it more like this?

1. I am justfied in my percptual belief that I'm sucking a lemon right now.
2. I can only have a justified belief that I'm sucking a lemon right now if I have a tangy quale.
Therefore, I have a tangy quale.

Or is it more like something else entirely?

iolasov said...

Well, I haven't thought very seriously about this stuff in a little while. The last time I did think of this way of developing an ontology of qualia, my approach was a little bit more holistic - "more like something else entirely". Here, without footnotes, is the piece of a seminar paper I wrote a couple of years back sketching how this would work. (I should note, also, that the one time we met, after your last talk at the CUNY Cog Sci group, I tried to explain this view and failed.)

Chalmers writes, "[a]ny plausible epistemological view must find a central role for experience in the justification of beliefs about the world." ("The Content and Epistemology of Phenomenal Belief", Section 5) We can add to this observation that different experiences justify beliefs about the world in finely discriminable ways. It is probably the case that a good portion of the discriminatory power of experience derives from monadic properties of the experience itself, rather than, say, the neural events that caused it. On the basis of these observations we might develop a semantics for the science of consciousness (very roughly) as follows. First, isolate all of the basic sources of evidence. Then set aside those sources of evidence which can be described without reference to the phenomenal properties of experience, e.g. testimony and coherence. For a sizable number of propositions, consider how much warrant you have to believe the proposition at a variety of times. Add up the amount of warrant for each proposition contributed by the non-experiential sources of evidence. The remaining amount of warrant that the proposition has is presumably derived from the phenomenal properties of experience. Now, think of phenomenal properties as operations on the degree of warrant of each of the propositions at a particular time. To simplify things a great deal, we could say that, for each proposition and time, a phenomenal property either adds a specified amount to, subtracts a specified amount from, or does not affect the warrant of the proposition at that time. Then we come up with the smallest number of phenomenal properties needed to account for the different levels of warrant the propositions have at different times that is not attributable to non-phenomenal sources. If a number of individuals carry out this process and compare their results, they can arrive at an intersubjective ontology and semantics of phenomenal properties.

Pete said...

Consider the following scenario:

Pete, Inverted Pete, and Zombie Pete are all at the grocery store in the produce section and they've each formed a perceptual judgment they would express by uttering "there is a red apple before me".

If, as you say, what ever warrant attaches to such judgments via coherence and testimony comes from "non-experiential" sources, there ought not be any differences in the amount of warrant due to coherence and testimony in Pete's, Inverted Pete's, and Zombie Pete's, judgments, since I hereby stipulate that those three dudes differ only phenomenally--where I have a phenomenal redness, Invert has a phenomenal greenness and zombie has a phenomenal nothingness. I guess you'd want to say that I have more warrant than my zombie counterpart. But how do I stack up against Invert, warrant-wise? Better? The same?

Or would you instead say that we could use the sort of transcendental argument you are developing to rule out the possibility of Invert?

iolasov said...

My intuition is that you have the same warrant for your judgment as Inverted Pete and more warrant than Zombie Pete. This brings us right to the limits of the theory. The explication of "quale" that it affords is only good for some purposes. You might call the explicated concept a concept of "diet qualia". It is not good for the purposes of exploring spectrum inversion and its epistemology. (I take the possibility of spectrum inversion to be pre-theoretically obvious, given the language we have at hand for talking about experiences.) My hope is that it would be good for the purposes of scientific consciousness researchers, who aren't generally skeptical enough to worry about spectrum inversion.

I understand this is sort of a cop out, given the stress placed on spectrum inversion in the philosophical literature. But I do think it is important to bear in mind what we want to do with one or another attempt to clear up the conceptual space surrounding "consciousness". One more-or-less isolable and important task is to clear the way for non-confused empirical research. My hope (and it is just a hope - I don't really ooze confidence on these matters) is that the theory can do something like that.

Teed Rockwell said...

HI Guys,

Appreciate the detailed interest in my argument. For the moment, I'll respond to Iolasov's arguments with the following comments.

1) The first premise in my argument is necessarily true in all possible worlds because it is part of the definition of what it means to be a zombie. There's no reason to unpack intuitions of any sort, because if there is something you are calling "qualia" which are not epiphenomenal, they are not qualia as defined by the terms of the zombie thought experiment.

2) It seems that the point you are making with your proposition #5 is that there is a difference between reasons and causes. However, seeing as no one really knows what that difference is, and many claim that it is illusory, to simply assert that it is inuitively obvious seems to me to beg the question.