Defending the Phenomenal Concept Strategy

Abstract

One of the main strategies against conceivability arguments is the so-called phenomenal concept strategy, which aims to explain the epistemic gap between physical and phenomenal truths in terms of the special features of phenomenal concepts. Daniel Stoljar has recently argued that the phenomenal concept strategy has failed to provide a successful explanation of this epistemic gap. In this paper my aim is to defend the phenomenal concept strategy from his criticisms. I argue that Stoljar has misrepresented the resources of the strategy, which can indeed accomplish the required explanatory task, once it is properly understood.

Notes

¹I am indebted to the Arts and Humanities Research Council, the University of Sheffield and the Royal Institute of Philosophy, for financial support during the writing of this article. I have presented earlier versions of this material at the University of Girona, the University of Leeds and the University of Sheffield, and I am very grateful to the audiences in all those occasions, for very useful feedback. I am especially grateful to the following, for very helpful comments and/or discussion: Alex Buckley, Manuel García-Carpintero, Dominic Gregory, Luca Incurvati, Rosanna Keefe, Stephen Laurence, Dan López de Sa, Marta Moreno, David Pineda, Karol Polcyn, Jennifer Saul, Richard Woodward and Elia Zardini. Extra thanks are due to Stephen Laurence, for very helpful comments and suggestions to many earlier versions of this paper. I am also indebted to three anonymous referees for this journal for very useful comments.

²This label is introduced in [Stoljar 2005.

³Some recent formulations of the phenomenal concept strategy include [Loar 1997, [Loar 1999, [Hill and McLaughlin 1999, [Papineau 2002, [Sturgeon 2000 and [Tye 2000.

⁴I follow Stoljar in using this rough formulation of the consequences of physicalism, for ease of exposition. A more precise formulation is provided in [Jackson 1998, among others.

⁵He also discusses and criticizes other versions of the strategy, such as concept-acquisition versions, but I will focus on the two mentioned above because they seem to me to be the most promising.

⁶Stoljar also discusses the prospective of the phenomenal concept strategy as a response not only to the conceivability argument but also to the so-called knowledge argument against physicalism (prominently defended by Jackson 1986). The knowledge argument, in a nutshell, claims that someone could know all the physical truths about the world without knowing all the phenomenal truths, and therefore physicalism is false. To this, the phenomenal concept strategy replies that this fact (if it is a fact) can be explained in terms of the possibility of someone knowing all physical truths but not possessing certain phenomenal concepts, and therefore, not knowing certain phenomenal truths (since you need to possess the required phenomenal concepts in order to entertain the relevant phenomenal truths). Stoljar claims that this is a good response against this formulation of the knowledge argument, but not against other formulations that bring it closer to the conceivability argument [Stoljar 2005: 474–5, 486–8]. Therefore, we will focus here on the discussion concerning the conceivability argument.

⁷Stoljar 2005: 479] says that ‘the central question is whether it is plausible to suppose that the experience thesis [or whatever thesis about phenomenal concepts is invoked] entails or otherwise suggests that [P→P*] is not a priori.’ However, it is not clear to me how Stoljar interprets the ‘otherwise suggests’ criterion, if it is intended to be different from entailment. In the rest of the paper, Stoljar just seems to assume that the relevant theory of phenomenal concepts must entail that the conditional is a posteriori, and he evaluates the different versions with respect to this criterion. For instance, he says: ‘to say that [P→P*] is not a priori synthesizable leaves open the possibility that it is a priori … So there seems to be a logical gap in the suggestion that the experience thesis tells us that [P→P*] is a posteriori’ [2005: 479]. Here, Stoljar assumes that if the experience thesis leaves open the possibility that the conditional is a priori, then the experience thesis is not a good version of the phenomenal concept strategy. Since in practice Stoljar assumes that the corresponding phenomenal concept thesis must entail that the conditional is a posteriori, I will ignore his (unelaborated) suggestion that something less might do. (For my own elaboration and further discussion see the remarks at the end of section III.)

⁸It is important to notice that Loar's defence of the possibility that the conditional is both necessary and a posteriori proceeds by defending first the possibility that psychophysical identities (involving phenomenal concepts and physical-functional concepts) such as ‘pain = C-fibre firing’ are both necessary and a posteriori. If psychophysical identities of this sort are necessary and a posteriori, then P→P* will be necessary and a posteriori too.

⁹This element of my defence of the phenomenal concept strategy makes use of a very important idea from Loar, namely, that apriority and aposteriority are (at least in part) psychological or cognitive features. That is, the fact that a certain inference or a certain thought is a priori or a posteriori can be explained in terms of certain psychological-cognitive connections holding or not between the constituent concepts. This idea has clear consequences for the question at issue here, namely, whether we can offer an alternative explanation of the aposteriority of the psychophysical conditional which does not appeal to such a conditional's being contingent. In my view, the psychological conception of the a priori/a posteriori distinction is a very promising and fruitful idea: it could have important consequences regarding the connection between conceivability and metaphysics, going beyond the main topic of this essay, namely the specific question of whether the concept-possession versions of the phenomenal concept strategy are tenable or not. Unfortunately I do not have the space here to explore these further questions, but I hope to do it elsewhere. Thanks to an anonymous referee for this journal for suggestions on this issue.

¹⁰It could be argued, at this point, that the fact that certain concepts are psychologically distinct does not suffice to explain the fact that they are not a priori connected, since there seem to be many cases of concepts that are psychologically distinct but are a priori connected nonetheless. In the following section we will discuss some putative counterexamples discussed by Stoljar, but here I want to emphasize that the crucial aspect of the psychological distinction thesis is not merely the psychological distinctness of the relevant concepts but also the fact that those concepts lack the appropriate cognitive ties and therefore are not a priori connected.

¹¹For possessing NOT A RED SENSATION requires possessing RED SENSATION.

¹²I am assuming that these sentences are a priori, for the sake of discussion. The denial of this assumption would constitute another way of replying to Stoljar, which I will not pursue here.

¹³This is why: ‘x is a number’ a priori entails ‘x is an abstract entity’, by (Number) and Modus Ponens; ‘x is an abstract entity’ a priori entails ‘x is not a mental state’, by (Mental) and Modus Tollens; ‘x is not a mental state’ a priori entails ‘x is not a red sensation’, by (Sensation) and Modus Tollens. So ‘x is a number’ a priori entails ‘x is not a red sensation’. That is, (2) is a priori true.