Horgan and Tienson on Phenomenology and Intentionality
As promised here are some remarks on the content of phenomenal states. Or rather, on Horgan and Tienson's remarks on the content of phenomenal states in their paper "The Intentionality of Phenomenology and the Phenomenology of Intentionality". Dave Chalmers pointed me at that last week. I should say that this post is not meant to be a response to Dave's comment. I can't respond to it yet because I haven't understood it yet. I'll have to read the other papers he mentions.
So. Horgan and Tienson say that phenomenal states have intentional content. For instance, the phenomenal state we're in when we take ourselves to see that a certain picture is hanging crooked represents that the picture (or some picture?) is hanging crooked. I have no objection to this. But if it's true it must be true because the phenomenal state plays the role something must play (by conceptual analysis) in order to represent the crooked picture. So to support their claim, Horgan and Tienson first have to tell us what they take to be that role -- what they mean by "represents" --, and then show that phenomenal states do satisfy that role.
At some places, this seems to be just what they do. In particular, when in section 3 they claim that the phenomenal states of my phenomenal duplicates all represent exactly what they represent in me, they support this view by pointing out that the states of my duplicates and those of me have "the same apparent input conditions [...] and the same apparent effects, involving experiences of apparently acting appropriately" (p.525). If we ignore all the "apparent"s in there, this sounds like my favourite, functionalist analysis of content. But of course we shouldn't ignore all the "apparent"s. I think what Horgan and Tienson want is to separate the representational content from any directly non-phenomenal conditions: for a state to represent a crooked image it needn't cause appropriate behaviour in appropriate circumstances. What matters is only the subject's experience of acting appropriately in what appear to him as appropriate circumstances.
So on this view phenomenal states have their content in virtue of their relation to other phenomenal states. The exact nature of these relations remains unclear to me, but I see no reason why if it is spelled out such a view couldn't deliver some kind of content. Let's call this content 'phenomenal role'. If I understand them correctly, Horgan and Tienson would like to identify the content of a belief with its phenomenal role. I'm not sure if this is a very good idea, mainly because it doesn't work for unconscious and more generally non-occurrent beliefs. I also don't know if it's really conceptually incoherent to suppose that a given phenomenal state could fail to occupy its 'phenomenal role'. If not, it is at best an empirical hypothesis that the state has that content, which I think is not what Horgan and Tienson want. One could also worry that phenomenal role, like conceptual role, leaves content radically underdetermined.
Anyway, at other places, Horgan and Tienson seem to opt for semantic primitivism: "Conscious intentional states are intrinsically, by their very nature, directed toward whatever they are directed toward" (p.530). This is hard to square with the view that phenomenal states are directed toward whatever they are directed toward in virtue of playing a particular phenomenal role. For playing that role is hardly an intrinsic property. It can still be essential and thus built into the bearers' very nature. Maybe Horgan and Tienson merely confuse "intrinsic" with "essential". (From the quoted passage it sounds like they really understand them as synonyms.) More probably, they perhaps regard phenomenal properties as being themselves intrinsic, and conclude that if the intentional properties are necessarily coinstantiated with these intrinsic properties they must be intrinsic too. If that inference is sound, then since semantic properties obviously are not intrinsic it really follows that phenomenal properties can't be intrinsic either.
In sum, I'm not sure how to understand Horgan and Tienson. There is much with which I could agree. Thus it might be acceptable to say that (some) mental states have their content in virtue of a kind of phenomenal role they play. It might also be accpetable to say that some mental states have their content in virtue of their phenomenal properties. But the latter is only accpetable if phenomenal properties themselves can be analysed in terms of other properties, as I think they can. If not this view leads to the kind of primitive semantics I find incoherent: Here we have a state with intrinsic, fundamental, primitive property Q; it also represents a crooked picture; why so? in virtue of having Q; but not because Q plays any further role grounding the content, but just because of Q itself. This I don't understand. It's like saying that your comfy chair represents the Spanish Inquisition simply in virtue of being your comfy chair.
I think the view is something like the following. Some states, e.g. perceptual experiences, have (narrow) content in virtue of their phenomenology. Other states, maybe beliefs, have narrow content partly in virtue of their phenomenology, partly in virtue of connections to other phenomenal states (e.g. perceptual experiences), and maybe partly in virtue of connections to inputs and outputs and other states.
Concentrating at the first stage (which is the purest case, as it doesn't invoke role properties and the like), the idea is that in some cases, having phenomenology of a certain sort suffices to have content of a certain sort. This is pretty intuitive in the case of perceptual experience: it seems prima facie plausible that by virtue of the phenomenology of a perceptual experience, things are presented to me as being a certain way. This case is argued at length in Chapter 7 of Charles Siewert's book "The Significance of Consciousness".
This connection between phenomenology and content is in principle compatible with various degrees of internalism or externalism about both. But given the premise that phenomenal properties are intrinsic (in the strong sense of not depending on environment or on internal role), which H&T accept, it follows that some intentional properties are intrinsic. You say this is obviously false, but don't give a reason. On the H&T view, some states have content not in virtue of their functional or causal role, but in virtue of their intrinsic phenomenology. Of course if we presuppose a causal or functional theory of content, it will follow that the claim is false, but such a theory can't be presupposed here.
What's the "analysis of content" here? I think the relevant conceptual analysis is just something like, a state has content (such-and-such) if it is assessable for accuracy (in such-and-such a way). Then, combined with the plausible premise that perceptual experiences are assessable for accuracy in virtue of their phenomenology (argued extensively by Siewert), it follows that perceptual experiences have content in virtue of their phenomenology. Maybe you want a conceptual analysis of content that's more reductive than this, but of course it's far from obvious that there's such an analysis to be had. Some concepts have to be taken as primitive, and may well be that something in the circle of intentional concepts will be among them.
On this view, both phenomenal concepts and intentional concepts may end up unanalyzable in non-mental terms. They may nevertheless have strong internal connections to each other: e.g. it's plausible that any adequate characterization of perceptual phenomenology must invoke intentional notions. Of course if one wants to use conceptual analysis to naturalize all mental properties in physical/functional terms, this will cause problems. But we already know that the phenomenal causes problems for such a reduction. (If they don't, e.g. because phenomenal concepts are functional concepts, then the issue about intrinsicness won't arise). On this view, the intentional will cause problems too.