The Quotational Model of Phenomenal Concepts

For a while, I planned to write something of the quotational account of phenomenal concepts (suggested in Papineau's "Thinking about Consciousness") and on why I think that it is less convincing than Loar's recognitional account. Today I have been working on it, and now I am not sure about what I think.

One the one hand, I had this point against the quotational model:

Papineau (2002) has put forward a “quotational” model of phenomenal concepts, which, I think, cannot provide a successful account of the reference-fixing mechanisms at work in the case of phenomenal concepts. According to this model, phenomenal concepts are formed by entering an experience “into the frame provided by a general experience operator ‘the experience: ---’. For example, we might apply this experience operator to a state of visually classifying something as red (…) and thereby form a term which refers to the phenomenal experience of seeing something red. (…) Very roughly speaking, we refer to a certain experience by producing an example of it” (2002: 116). I think that this quotational model of phenomenal concepts is a plausible account of how we acquire new phenomenal concepts, but as an account of the reference-fixing of phenomenal concepts it is insufficient, for the following reason: if a phenomenal concept of kind K refers to such a kind K just by virtue of producing a token of K, how could we distinguish between concepts of kind K and concepts of a sub-kind K*? Plausibly, both concepts could be formed by entering a token of K* into the operator ‘the experience: ---’. So, according to the quotational account, both concepts would be identical: they would refer to concepts that resemble that token of K*. But this seems false: we have two different concepts, one refers to kind K and the other refers to the more specific kind K*. On my view, we can easily explain this by appealing to the recognitional dispositions that are associated with our concepts: the phenomenal concept of kind K is associated with a more general recognitional disposition, whereas the phenomenal concept of kind K* is associated with a finer-grained recognitional disposition.

So far so good. But Papineau adds that the quotational character of phenomenal concepts is not enough for explaining the reference-fixing of such concepts. He argues that we would need to supplement such an account with an additional semantic theory which explains how a given phenomenal concept can refer to the corresponding phenomenal property (a token of which is incorporated into the concept). Papineau suggests that we can appeal to theories of content such as Fodor's causal account or Millikan's teleological account .

I was wondering whether the problem I explain above still affects this or not. According to Papineau, “the phenomenal concept will refer to a type of experience whose instances bear a certain resemblance to the ‘quoted’ exemplar. (…) Phenomenal concepts refer to items that resemble their ‘fillings’ because applications of these concepts are typically caused by those items, or because it is the function of such concepts to track those items” (2002: 119-21).

How should we understand this? If phenomenal concepts are supposed to fix their referents by virtue of some (more or less complex) relation between the concept and a property-exemplar, then I think we will have the same problem again. But if the point is that the concept fixes the referent by means of a relation between the concept and the corresponding property, then we could solve the problem, because then phenomenal concepts K and K* will be connected with different properties, K and K*.

I was also wondering whether this is really different from the recognitional account. On the recognitional account, what fixes the reference is our disposition to recognise tokens of a certain property. This sounds like a causal theory: the concept refers to that property tokens of which causally trigger the concept to be applied. I guess that we will need to supplement this view with a teleological account, if we want to allow for misrepresentation.

So maybe the recognitional and the quotational models are not so different after all, and maybe they both need a bit of biosemantics to keep going!