Abstract
The conditional analysis of phenomenal concepts purports to give physicalists a way of understanding phenomenal concepts that will allow them to (1) accept the zombie intuition, (2) accept that conceivability is generally a good guide to possibility, and yet (3) reject the conclusion that zombies are metaphysically possible. It does this by positing that whether phenomenal concepts refer to physical or nonphysical states depends on what the actual world is like. In this paper, I offer support for the Chalmers/Alter objection that the conditional analysis fails to accommodate the true zombie intuition, and develop a new and far more powerful argument against the conditional analysis. I argue that, as stated, the conditional analysis is radically incomplete. But when fully fleshed out, the analysis becomes viciously circular. The only way to avoid this circularity is to adopt a species of analytic functionalism, on which it’s a priori that phenomenal concepts refer to the state (perhaps physical, perhaps nonphysical) that actually plays so-and-so functional role. While this rigidified analytic functionalism is coherent, it is highly unattractive, running contrary to both the intuitions that motivate functionalism and the intuitions that motivated the conditional analysis.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
A minimal physical duplicate of our world is a world physically identical to our own that contains nothing more than it needs to be a physical duplicate. This formulation of physicalism is due to Braddon-Mitchell and Frank Jackson (The Philosophy of Mind and Cognition). There are other ways of formulating the thesis, but this won’t matter for our purposes.
This clause might be too simple: Presumably, not just any nonphysical state is a phenomenal state, just as not just any physical state is a brain state. For this reason, one might want to rephrase the antecedent to be about the actual world containing nonphysical states of the right sort. Along these lines, Hawthorne includes the restriction that the nonphysical state “typically [be] caused in the way that folk think pain [or whatever other phenomenal experience] is caused.” (Hawthorne 2002, p. 25) Braddon-Mitchell does not include such a restriction. My argument against the conditional analysis, given in section three, hinges on the feasibility of spelling out this of-the-right-sort clause.
Braddon-Mitchell offers an alternative formulation on which, if the actual world is merely physical, then in the actual world our phenomenal concepts refer to the physical states that actually play the right functional roles, and in each counterfactual world, our phenomenal concepts refer to whatever physical states play the right functional role in that world. I’ll ignore this for simplicity. It won’t affect my argument against the conditional analysis.
I’ve avoided using any technical terminology. But for those unfamiliar with two-dimensional semantics, it may be helpful to refer to Chalmers’ introduction to the subject: Chalmers, “Two-Dimensional Semantics”, as the basic ideas are essential to understanding the conditional analysis and my objections to it.
It’s not clear that these intuitions are by themselves sufficient to push one towards a conditional semantic analysis for phenomenal concepts. For example, Papineau and Balog’s “constitutional theory” of phenomenal concepts, satisfies the motivation for the conditional analysis, but doesn’t seem to offer a conditional semantic analysis. On this theory, phenomenal concepts are partially constituted by the phenomenal experiences they refer to. Nothing in this analysis of phenomenal concepts specifies whether the experiences that partially constitute phenomenal concepts are physical or nonphysical. So it is not analytic that phenomenal experiences refer to either physical or nonphysical states. Thus, when we come to believe in or abandon dualism, this does not require our phenomenal concepts to change. It just requires our beliefs about the nature of the states composing the concepts to change. While the constitutional theory may be conditional at a metasemantic level, insofar as the analysis of our analysis of phenomenal concepts depends on the nature of reality, it is not conditional at a semantic level.
One might deny that this was conceivable, on grounds that we should be talking not merely about “nonphysical states”, but about “nonphysical states of the right sort”. If you favor adding an “of the right sort” clause, you can read this as a call to flesh out this clause in a meaningful way. As will become clear, I think fleshing this clause out will either require reliance on phenomenal concepts, making the resulting analysis circular, or will collapse into a form of analytic functionalism. Further, such a response seems to rule out as inconceivable the possibility of overdetermination.
As Braddon-Mitchell implies, while a physicalist, he is not an arrogant one, and hence “[does] not give zero credence to physicalism being false” (Braddon-Mitchell 2003, p. 127).
This implicit shifting in theory-of-consciousness when we consider worlds in different ways is quite strange. When we ask whether a world is a zombie world or a world with consciousness, we’re asking whether there’s a what-it’s-like in that world. Intuitively, whether or not there’s a what-it’s-like in a world doesn’t depend on whether that world is being thought of as actual or counterfactual. It depends on what there is in that world. Words may be able to shift their referents depending on the attitude we take to the world they’re being used in. But surely whether there’s something that things feel like doesn’t depend on the attitude you take towards a world! On the conditional analysis whether there’s a what-it’s-like in a world or not depends not just on intrinsic features of that world, but on whether we are taking an attitude towards the world whereby we “hold fixed” features of our world (considering the world as counterfactual) or whether we mentally submerse ourselves in the world (considering the world as actual).
One might object that I can’t conceive of a world in which my phenomenal concepts don’t refer to anything: Surely “zombies” would refer to something with their uses of ‘pain’. But this is beside the point, much as it would be beside the point that inhabitants of some world might use ‘pain’ to refer to puppies. What’s relevant is what our term would refer to in such a world, and it seems clear that it wouldn’t refer.
This would be a case where the physical states aren’t phenomenal states, and though there are nonphysical states, these states are intuitively not phenomenal states either. (Surely we can imagine that there are nonphysical states that are just some weird extra, and have nothing to do with our mental lives!) So in such a world, phenomenal experiences, although there are both physical and nonphysical states.
This is a strategy Haukioja takes in responding to Alter’s objection that an oracle could tell us that the world is purely physical and nothing plays the role of qualia.
This is something of a simplification, as we have not included a clause to accommodate the possibility of overdetermination. For simplicity—and because it will not pose any problems to the argument I go on to give—I will omit this clause. Note that, if anything, the need to accommodate such a clause poses even more difficulty for the conditional analysis.
Dualists who endorse the constitutional theory will deny that it’s epistemically possible that conscious states be physical states.
Thanks to Richard Yetter Chappell for his helpful discussion of the aims of the constitutional theory and the conditional analysis, and for the descriptors ‘experience first’ and ‘concept first’.
This strategy was suggested to me in conversation by John Hawthorne.
Another thought one might have for avoiding the problems that I’ve suggested is to replace uses of PAIN with “state that actually stands in so-and-so reference relation to the concept of PAIN.” There are two ways to read this suggestion, neither of which is helpful. Let’s look at how this phrase would be embedded in one of the conditional analysis’s clauses. (1) becomes (1').
(1') If the actual world is merely physical and we have states that in the actual world stand in so-and-so reference relation to the concept ‘PAIN’, then ‘PAIN’ refers rigidly to the physical states that actually play the right functional roles.
(1') can be read in one of two ways. On the first, both ‘actual world’s refer to the same world, so that the second reference to the actual world' is redundant. On the second way of reading it, they refer to different worlds, giving us something like this: If the world we’re considering as actual is merely physical and, in this world, we have states that are the very same states which in our world — the “really” actual world—stand in so-and-so reference relation to the concept ‘PAIN’, then ‘PAIN’ refers rigidly to the physical states that, in the world we’re considering as actual, plays the right functional roles. On this reading, the second ‘actual’ isn’t trivial. But the resulting analysis is of no help to the conditional analysis. For according to the conditional analysis, pain might have a very different nature in some other world-considered-as-actual from the nature it has in our world. What we emphatically do not want is to build into the antecedent that pain in some other world we’re considering as actual is the state that in our world is the referent of ‘PAIN’.
If we allow for the possibility of overdetermination, we reintroduce the need for clause (4), and reintroduce the circularity.
This could be thought of as a condensed way of putting the following conditional:
-
1.
If the actual world contains nonphysical states that play the relevant functional roles, then phenomenal concepts refer rigidly to these nonphysical states.
-
2.
If the actual world is merely physical and has physical states that play the right functional roles, then phenomenal concepts refer rigidly to these physical states.
-
1.
Thanks to Richard Yetter Chappell for this insight.
Bibliography
Alter, T. (2007). On the conditional analysis of phenomenal concepts. Philosophical Studies, 131(3), 777–778.
Balog, K. (2012). Acquaintance and the mind-body problem. In C. Hill & S. Gozzano (Eds.), The mental, the physical. Cambridge: Cambridge University Press.
Braddon-Mitchell, D. (2003). Qualia and analytic conditionals. Journal of Philosophy, 100(3), 111–135.
Braddon-Mitchell, D., & Jackson, F. (2007). The philosophy of mind and cognition. Malden: Blackwell Publishing.
Chalmers, D. (2006). Two-dimensional semantics. In E. Lepore & B. C. Smith (Eds.), The Oxford handbook to the philosophy of language. New York: Oxford University Press.
Chalmers, D. Responses to Articles on My Work. http://consc.net/responses.html#braddon. Accessed 30 April 2012.
Haukioja, J. (2008). A defense of the conditional analysis of phenomenal concepts. Philosophical Studies, 139(1), 145–151.
Hawthorne, J. (2002). Advice for physicalists. Philosophical Studies, 109(1), 17–52.
Papineau, D. (2002). Thinking about consciousness. New York: Oxford University Press.
Perry, J. (2001). Knowledge, possibility, and consciousness. Cambridge: MIT University Press.
Acknowledgments
Thanks to Richard Yetter Chappell, John Hawthorne, Frank Jackson, Sarah-Jane Leslie, David Nowakowski and Jack Spencer for many thoughtful discussions and comments on the conditional analysis and various drafts of this paper.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Yetter-Chappell, H. Circularity in the conditional analysis of phenomenal concepts. Philos Stud 165, 553–572 (2013). https://doi.org/10.1007/s11098-012-9958-8
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11098-012-9958-8