Wednesday, October 12, 2005

Chalmers Still on the Case

David Chalmers (home page, blog) has a new paper (The Two-Dimensional Argument Against Materialism) which provides a definitive and comprehensive update of his work on the “conceivability argument” and related issues. Defense of this argument over the last decade has led him to broaden the turf: the bulk of the paper goes beyond the mind/body issue to a general defense of the premise that conceivability implies metaphysical possibility; in other words, an epistemic premise (about what one can conceive) can lead to a modal conclusion (about what is possible).

I won’t try to summarize the paper in this post (other than a bare-bones sketch of the original conceivability and 2-dimensional arguments below). It is very thorough and I found the cumulative impact of Chalmer’s responses to various critics to be powerful.

I also think a section tucked in late in the paper is thought provoking(Section 10 – Modal Rationalism). Here he steps back and wonders about the implications of his defense of the premise that conceivability implies possibility. He talks about the crucial role modality plays in our capacity for rational thought. We need modal concepts to analyze various phenomena rationally -- he says you might call this logical modality and make use of logically possible worlds. And given Chalmer’s analysis in the body of the paper, there is no reason to think logically possible worlds and metaphysically possible worlds are different animals. Interesting stuff. I wonder, as always, about the ontological implications of this also. By virtue of what outside of ourselves is our capacity for rational modal thought based?

Sketch of conceivability and 2-dimensional arguments:

The conceivability argument has 3 premises (my paraphrase of the basic argument – the argument is presented in iterations of increasing precision over the course of the new paper): 1. A world physically just like ours with no first-person conscious experience is conceivable. 2. If such a world is conceivable, it is metaphysically possible. 3. If it is metaphysically possible, than materialism is false, since materialism is a metaphysical thesis which says physical facts necessarily entail all other facts, including experiential ones. The paper responds to a wide variety of objections posed, but most of the attention is on the second premise.

Let me just mention that the two-dimensional argument of the title (a version of which was already included in Chalmer’s 1996 book) was constructed to respond to what was the most common sort of this objection to the second premise. It is one of this sort (exemplified by using an example of Kripke’s): we can conceive of a world where water is not H2O, but it isn’t metaphysically possible -- we know that water is H2O. Chalmers breaks down the idea that {“water is not H2O” is conceivable} into two senses: the first (primary) is that there is a possible world where there is watery stuff (doing the job of water) which is not H2O. This is clearly conceivable. The secondary sense is that the stuff in the possible world isn’t watery stuff – it really is water! – then it is not conceivable that this water is not H2O. It is the primary sense of conceivability which we utilize in Chalmer’s second premise.


Richard Jordan said...

I have been itching to comment on this, and keep waiting till I get some time to write something coherent. But I guess I won't get that time so here goes anyway.

I think there is a funamental problem in this approach. Arguing that you cannot imagine a world where water is not H2O because water is H2O is fine. Okay, I can buy that. Only ignorance of the fact that water is H2O allows one to make the statement that it is possible to imagine a world where it is not.

Problem is the same argument can be made for the bold statement that it is possible to imagine a world exactly like ours without consciousness.

Given that we still don't have a handle on what consciousness is, how first person experience arises, then the statement doesn't seem credible to me, that "A world physically just like ours with no first-person conscious experience is conceivable". At least it has the risk of being just as wrong, and for the same reasons, as the statement about a world where water is not H2O.

What am I missing here?

Steve said...

Here's a couple of thoughts.
First, on the water example, remember there is a primary sense of the example where we can conceive of a world where water is not H20 - where there is something watery which plays the role of water. In this sense, the fact that we gain the knowledge that water really is H20 (a posteriori) only impacts the secondary sense of the example.

In the case of consciousness, we are surely in a state of ignorance. But the primary sense of the conceivability of zombies (something physically just like you or me with no first-person consciousness on the inside) will survive further knowledge gains on the subject according to the argument.

As to whether we all share this intuition that zombies are conceivable in any sense to begin with: try thinking about a robot I build. It's a pretty good robot I make with wires, chips,etc. Does that have first-person experience? Now build better and better models until you approach zombie specifications. Does consciousness pop in somewhere along the line? It seems as I can conceive of a world where getting all the physical specs right (with things like our neurons and proteins, etc) to create a counterpart of a human would succeed without the light of interior experience ever going on inside the counterpart.

There are many people who don't see the conceivability of this. Chalmers undoubtably does a better job pumping the intuition than I can.

Ellis Seagh said...

It seems as I can conceive of a world where getting all the physical specs right (with things like our neurons and proteins, etc) to create a counterpart of a human would succeed without the light of interior experience ever going on inside the counterpart.

Only if you already believe that "the light of interior experience" is not a function of the "physical specs", no? Otherwise you would no more be able to conceive of such a thing than you would be able to conceive of a "zombie" car that had identical physical specs as a regular car, but in which its engine was without effect.

It seems to me, in other words, that these kinds of arguments (the "Knowledge Argument" being another example) are essentially begging the question.

Steve said...

That's exactly what the thesis is: first-person experience is not a function of the physical structure. Physical description does not entail experience.

The conceivability and knowledge arguments are hard ones to carry off: they try to move from an epistemological premise to a ontological conclusion. A better argument might be this (related to Russell's monism proposal): physical descriptions are about extrinsic relations and structure. The descriptions are invariant as to whether the intrinsic basis of reality on which the structure adheres is one that gives rise to experience or not.

Steve said...

Also, if you don't like zombies, remember you can construct the same arguments with "inverted spectra". I conceive of a possible world where the same physical make-up makes me see apples as blue (or something like that).

Ellis Seagh said...

I understand that that's what the thesis is here, Steve -- my point was that if you need to assume that thesis in order to make the conceivability argument for it work (as I think you do, and as my example of the zombie car, by being absurd, was intended to illustrate), then that's a basic form of error. I think something similar is going on with your description of the monism proposal: if the question is whether or not conscious experience is amenable to physical description/explanation, then you can't just assume from the start -- or define from the start -- that it's not.

Re: inverted spectra -- I think this simply underlines how the content of experience, while real and unavoidable, is almost entirely arbitrary. What's doing the causal work is the difference between qualia, or their qualitative distinctiveness, which only need whatever content they have (blue, green, loud, soft, etc.) in order to exist at all.

Steve said...

Thanks for your comments, Ellis. Here's a couple of questions.

First, if the content of experience is arbitrary, doesn't that support the conclusion that it isn't entailed by a specified physical structure?

Second, the physical descriptions we're familiar with involve quantitiative relations; how can they specify differences which are (as you say) qualitative?

Ellis Seagh said...

And thanks for the questions. Here's my take:

First, I meant that the content is "arbitrary" only in the sense that it's not the content "red", say, in itself that constitutes the signal, but simply its difference from other such contents. But, certainly, there would have to be a specific physical structure that "entails" the content, whatever it is.

The second question relates to the first, of course, but maybe gets into some deeper waters too. At a fundamental level, the level at which information-bearing signals operate, all that I think really pertains is difference as such -- "qualitative" being redundant, and "quantitative" being just a relatively sophisticated kind of difference that's only apparent to, and meaningful for, observers with number systems, measuring devices, and the like. That is, a system designed for signal input (which is what I'm asserting consciousness is) would only be able to distinguish difference, and would have no idea of "quantitative" difference. External observers of such a system can then come along and apply various measuring devices, related to a system of units and of counting, and derive quantitative differences for such signals related to their specific physical structure. But in doing so they will have lost sight of the notion of information that makes the system run. To my mind, in other words, "physical descriptions" can or should be able to accommodate information-bearing difference, in which the quantitative difference is no doubt present, but is largely irrelevant.

(I'm hoping to write a post on this soon, by the way -- and it looks like I need it.)