Semantics for Dualists

Non-reductive (a posteriori, type-B) materialists say that even though phenomenal terms denote physical states or properties, the phenomenal way things are is not a priori entailed by the physical way things are. This means that no amount of physical information can tell us what our phenomenal terms denote. That is, non-reductive materialism implies that the projects of naturalising linguistic and intentional content are doomed. I would say that contraposititvely, since there are good reasons to believe in the project of naturalising linguistic and intentional content, non-reductive materialism is doomed.

Some versions of dualism might escape this problem. For instance, an interactive dualist could accept a causal theory of reference and hold that phenomenal terms denote non-physical entities in virtue of standing in a suitable causal relation to them. On my view, epiphenomenalists and parallelists could tell an equally plausible story:

Consider a world where dualism is true. Presumably inhabitants of such a world can refer to the non-physical qualia in their surroundings. If the qualia are epiphenomenal, the reference is not constituted by causal chains, but by description: the inhabitants denote epiphenomenal qualia by their qualia terms because these are what satisfies their qualia theory. The qualia theory in turn is believed by the inhabitants perhaps because attributing this theory to their belief states is part of the most explanatory theory of their linguistic and non-linguistic behaviour; or perhaps because of causal or counterfactual or teleosemantic relations connecting some of their mental representations to external things (not their qualia representations).

At any rate, dualists need not reject the feasibility of reducing facts about content to other facts. If the other facts are not all physical, this won't count as 'naturalising content'. But none the worse for that. What's really incredible is only that semantic properties should be fundamental, inexplicable and irreducible.

However, dualists may not like my proposals. By appealing to causal relations or satisfaction of theories and interpretation of behaviour I have apparently detached subjects from their qualia. On my proposal, dualism looks like phlogiston theory: Certainly in some worlds phlogiston theory is true, and inhabitants of those worlds can refer to phlogiston by means of descriptive or 'direct' reference. (By "'direct' reference" I mean reference in virtue of something like causal or counterfactual or teleosemantic relations, not reference in virtue of nothing at all.) But it is a matter of scientific investigation to find out whether phlogiston really exists -- whether something satisfies phlogiston theory or stands in the direct reference relation to "phlogiston" -- and if so, what further properties it has. Could this be true for qualia? Is it an empirical possibility that there are no qualia, that all our qualia-beliefs are empty? And could science make substantial discoveries of so far unknown qualia properties? Aren't qualia directly and fully given in experience? Don't I feel pain if and only if it seems to me that I feel pain; so that it doesn't make sense to suppose that science could tell me I'm not really feeling pain because some hidden but essential feature of genuine pain is missing, or because there simply is no pain at all?

Unfortunately, this is hard to square with a sensible account of content. It is very well to say that qualia are somehow parts or constituents of qualia beliefs, and thereby directly and immediately contained in the believed propositions. This tells us something about the structure of the semantic values we should assign to certain belief states. But it tells us nothing about why this is the semantic value of those belief states. Even if mental representations standing for qualia themselves contain these qualia (rather than their content containing the qualia), this doesn't explain why they denote the qualia. After all, representations don't normally denote their parts. But there must be some story to tell. I take it to be a contingent, empirical matter whether there are non-phsyical qualia and therefore fundamental non-physical facts. But it is not a contingent, empirical matter whether besides qualia facts and physical facts, the fundamental facts also include additional and irreducible semantic facts. The idea of semantic facts being primitive doesn't make sense to me. Even a dualist has to offer some account of how qualia belief states and qualia words in public language get their content.

In The Content and Epistemology of Phenomenal Belief, Dave Chalmers begins by making some reasonable suggestions for how phenomenal terms could pick out their referents, e.g. descriptively as "the phenomenal quality typically caused in normal subjects by red things", or more indexically as "that phenomenal quality" (§2.1). He notes that these accounts don't deliver A-intensions that pick out the respective qualia at every possible world. Thus he introduces 'pure phenomenal concepts' for which it is entirely unclear to me how they pick out their referents. Chalmers says they do so "directly" (§2.1), but unless "directly" means "in virtue of something like causal or counterfactual or teleosemantic relations" -- which it can't mean as all this yields the wrong A-intensions -- I really don't understand it. Fortunately, Chalmers agrees that at least for terms in public language this magical direct reference is implausible or inapplicable (§3.2). But this seems to lead to a peculiar position. Doesn't it mean that one has to say things like:

It's true that all my physical duplicates in all possible worlds feel pain, and that the physical facts entail a priori whether somebody feels pain; but that's only true if by 'feeling pain' we mean what that expression means in English. It is not true if by 'feeling pain' we mean something quite different, namely what I mean in my idiolect.

Comments

# on 03 March 2004, 17:11

"Directly" at that point just comes to a claim about content (primary intensions), not a claim about how this content is grounded. I don't try to give a full metasemantic story in the paper, but obviously I'd appeal to grounding in the phenomenal state itself rather than to causal and
counterfactual relations. More generally, I'm doubtful about prospects for a wholly causal theory of content and am inclined to give phenomenology a central role in an account of intentionality.

Anyway, I certainly don't think the consequence at the end follows. Even if one analyzes the primary intension of "pain" in English as something like "The sort of phenomenal state that typically occurs in circumstances C", there will be no a priori entailment from physical truths to pain truths, since there's no a priori entailment from physical truths to instantiation of a
phenomenal state. To get an a priori entailment (even assuming that physical truths entail C), one will also need to add a lot of truths about the existence and distribution of phenomenal states, in order to determine whether someone is having the sort of phenomenal state that is normal for C.

# on 03 March 2004, 18:38

Have you sketched somewhere how this account of intentionality might go? Is it a primitive fact about (all?) phenomenal states that they represent themselves, or that whatever contains them as a component denotes them? Or is this due to other facts, perhaps facts about the inferential roles in which the representation is embedded? And do the same facts also determine the semantic properties of other, non-phenomenal mental representations?

You're right that the consequence at the end doesn't follow. I've chosen bad examples. What follows is perhaps that it is a priori given what "pain" means in English that pain typically occurs in C, so that there could be no world considered as actual where pain doesn't typically occur in C.

# on 04 March 2004, 03:02

Which things are related by the causal relations that figure in determining content? On the one end... the supposed referents. But what about the other end?

Here's an example from Brian Loar ("Self-interpretation and the Constitution of Reference"). Unbeknownst to me, a scientist is stimulating my brain to cause an experience as of a fly. I think "That fly is huuge." There happens to be fly in front of me, but behind some screen or other.

If afterwards I discover the true situation, I will not think that my thought-token was about the fly in question. Presumably, because it didn't stand in the right causal relation to my sensation.

Now did I also think about my sensation in some causal-descriptive way? It is hard to see how.

The problem is basically what Devitt calls 'The Passing the Buck objection' (against Descriptivism). How can there be any constraints on reference if the reference of certain thoughts is explained in terms of others, whose reference is explained in terms of others... etc. 'The buck must stop somewhere!'

# on 04 March 2004, 11:13

I'm also quite doubtful about a wholly or even largely causal theory of content. Causal accounts might work for perceptual states, but they hardly work for intentional states in general.

However, I don't think the "passing the buck" objection affects causal theories. It only affects causal descriptivism, but that'a quite different theory. A causal theory as I understand it says that representational states denote whatever stands in the right causal relation to the state, irrespective of any other states or descriptive beliefs the subject might have. This doesn't even create the "passing the buck" problem. Moreover, I don't think it has special problems with intentional states about other intentional states: Surely an intentional state can stand in the right causal connection to another such state.

# on 05 March 2004, 18:17

No, I haven't given a theory of how phenomenology grounds intentionality. I'm sympathetic with many of the things that Horgan and Tienson say in their piece on the phenomenology of intentionality (in my philosophy of mind anthology), but that isn't really a theory. There's some preliminary discussion of the relationship between phenomenology and intentionality in the case of perceptual content in my papers "The Representational Character of Experience" and "Perception and the Fall from Eden" (on my website).

I don't think that phenomenal states represent themselves, though I do think that they represent certain other properties (e.g. the primitive properties discussed in the second paper mentioned above). Phenomenal states are represented by phenomenal concepts, which have phenomenal states as a constituent but which require significantly more in the way of cognitive structure, including the potential for being embedded in judgments, and so on. The two jointly determine the representational features of the phenomenal concept.

I agree with Brandt that the buck has to stop somewhere. On the two-dimensional view of content that I favor, which has a lot in common with causal descriptivism (when causal connections are relevant to determination of reference, this is usually because there's a causal element in a primary intension), this manifests itself in the need for an underlying semantically neutral vocabulary. On my view the buck stops with certain key semantically neutral concepts, which don't get their referents via causal connections. These include phenomenal concepts (as discussed), and also other key concepts such as logical concepts and the concept of causation itself.

# from on 08 March 2004, 21:03

Suppose we want to know whether some thing A has the property of representing B. The first thing to do is to ask what exactly is meant by "representing" in this context. That is, we must inquire into the general conditions under which it would be true t

Add a comment

Please leave these fields blank (spam trap):

No HTML please.
You can edit this comment until 30 minutes after posting.