The Consequences of Intentionalism*
Australian National University
Draft of 19/6/02
for NEH Summer Seminar, Santa Cruz, CA.
Most of the recent discussion in philosophy of mind concerning intentionalism—the thesis, roughly, that the phenomenal character of an experience supervenes on its intentional character—has focused, naturally enough, on whether the thesis is true. I think enough has been said on this topic to demonstrate that, whether or not it is actually true, intentionalism is at the very least plausible, a thesis that pleasingly integrates the idea of phenomenal character into the dominant picture of cognition as a matter of information flow. But less has been said about the further consequences of intentionalism. One suggestion which is common in the literature is that, if it were true, intentionalism would have a significant impact on the ‘hard problem’ (Chalmers 1996) of phenomenal character, the problem of explaining whether and how the phenomenal character of an experience supervenes on its physical character. But does intentionalism have this consequence in fact?
In the first part of this paper, I argue that it does not. In the second, I ask what the consequences of intentionalism might be if the doctrine does not impact on the hard problem in the way it is widely thought to. I suggest that at least one major philosophical import of intentionalism lies, not with the hard problem, but with the secondary qualities, and in particular with the error theory, the doctrine that experiences of color are systematically illusory.
The relation of intentionalism to the hard problem is an abstract issue, even by the standards of philosophy of mind. So, to focus matters, I will concentrate on an argument which tells us that if intentionalism is true, the phenomenal must supervene on the physical—I call it the transitivity argument. Our working assumption is that intentionalism has an impact on the hard problem if, but only if, the transitivity argument is persuasive.
The first premise of the transitivity argument is the supervenience of the phenomenal on the intentional, a thesis implied by intentionalism:
(1) For any subjects S1 and S2, and possible worlds W1 and W2, if S1 in W1 is having an experience and S2 in W2 is having an experience then if the two experiences are identical in all intentional respects, they are identical in all phenomenal respects.
The second premise is the supervenience of the intentional on the physical—physicalism about the intentional, as I will call this view:
(2) For any subjects S1 and S2, and possible worlds W1 and W2, if S1 in W1 is having an experience and S2 in W2 is having an experience then if the two experiences are identical in all physical respects, they are identical in all intentional respects.
If these two supervenience theses are right, then, by the transitivity of supervenience, we have no choice but to believe physicalism about the phenomenal, i.e. the supervenience of the phenomenal on the physical:
(3) For any subjects S1 and S2, and possible worlds W1 and W2, if S1 in W1 is having an experience and S2 in W2 is having an experience then if the two experiences are identical in all physical respects, they are identical in all phenomenal respects.
But this last claim is precisely the claim—or is a version of the claim—that is at issue in discussions of the hard problem. In short, if intentionalism is true, and if physicalism about the intentional is true, there is no choice but to adopt physicalism about the phenomenal. So how could intentionalism fail to have an impact on the hard problem?
In §4, I take up the evaluation of the transitivity argument. I will not be concerned with its conclusion. To answer the question of whether the phenomenal does supervene on the physical would be to start on the road to a solution to the hard problem and not simply to comment on the relation between it and intentionalism. I will do nothing of the sort here. But what I will suggest is that the transitivity argument is unpersuasive in the following sense: there is no interpretation of its central terms according to which both of its premises are true or at least plausible. Given our working assumption, this entails that intentionalism has no impact on the hard problem. Before turning to the transitivity argument proper, however, it will be useful to review three points concerning the formulation of intentionalism.
First, intentionalism is or entails a thesis about the phenomenal character of experience, viz., that the phenomenal character of an experience supervenes on its intentional character. But understood this way, it needs to be distinguished from another thesis, according to which every psychological state or episode has, in some sense, an intentional aspect or character. Somewhat confusingly, this thesis is sometimes also called intentionalism (Crane 2002). To avoid confusion, and to honor the historical background of the thesis, I will call this thesis Brentano’s Thesis
At least on natural assumptions, intentionalism entails Brentano’s Thesis but not vice versa. Brentano’s Thesis does not entail intentionalism because it would be quite possible to hold that every mental state has some intentional character and deny that its phenomenal character supervenes on this intentional character—the phenomenal character and the intentionality might in this sense be only contingently related. That intentionalism entails Brentano’s thesis can be brought out by demonstrating that the falsity of Brentano’s thesis yields the falsity of intentionalism, or at least does so given natural assumptions. Suppose that Brentano’s thesis is false. Then, presumably, there are two possible experiences e and e* such that e and e* differ in phenomenal character but both lack intentional content outright. But if two experiences both lack intentional content outright they are the same with respect to intentional content. So if Brentano’s Thesis is false it is possible that two experiences are distinct with respect to phenomenal character but are the same with respect to intentionality. But that contradicts intentionalism. Hence, if Brentano’s thesis is false, so too is intentionalism.
Second, intentionalism says that the phenomenal character of an experience supervenes on its intentionality. But there are at least two things one might mean when one speaks of the intentionality of an experience. First one might mean only the semantic or informational content of the experience, how it represents the world as being. Second one might mean both the informational content of the experience and in addition various features of its functional role.
To bring out the difference, contrast the case of visual imagination and visual perception. At the moment I am having a certain visual experience, and we might suppose that there is some proposition p which completely characterizes the informational content of the experience. (Perhaps we would not be able to express this proposition in language, but let us set that aside.) If I were adept enough at visual imagination I might be able to conjure up in imagination a counterpart experience, i.e. one that is also characterized completely by p. Now, are these two experiences, the imagination and the perception, intentionally identical or not? If by ‘intentional’ we mean simply the semantic or informational content, the answer is yes, for both represent the world as being a world in which p is true. On the other hand, if by ‘intentional’ we mean not only the informational or semantic content of the experience but also their functional roles, they will not count as intentionally identical. For imagination has a very different functional role from perception having to do with the fact that imagination is an act of will in the way that perception is not—perception, as it sometimes put, forces belief in some contingent fact in a way that imagination does not.
Should an intentionalist operate with the more restricted or the more liberal conception of intentionality? The answer is that they should adopt the more liberal conception, for otherwise their position would be too easily refuted. For consider: it is natural to say that the phenomenal character of visual imagination is distinct from that of visual impression: because of its different functional role, imagination feels different from perception. But if so, the imagination/perception contrast presents a straightforward objection to intentionalism. So it seems that intentionalism in its strongest form is going to make reference not only to informational content but also to functional role. There are in fact two distinct ways in which we might include the notion of the functional role of an experience in a claim such as (1). On the one hand, we might insist that the idea of the intentionality of an experience does not simply include the informational or semantic content of the state, but rather includes both the informational content and it functional role. On the other hand, we might insist that the intentionality of an experience just is its informational content, and simply adjust (1) so that it is true only of experiences of the same functional character. I will operate here with the first option rather than the second, but I take this to be a matter of theoretical decision more than anything else.
Finally, intentionalism entails that the phenomenal character of an experience supervenes on its intentionality. But does the reverse entailment hold? That is, can we say that intentionalism is true only if the supervenience of the phenomenal on the intentional is true? In part of course the question is terminological: if someone wants to use the phase ‘intentionalism’ to denote the supervenience thesis, there is in one sense no point criticizing them. However, there are a number of positions which, while agreeing with the supervenience thesis, are most plausibly distinguished from intentionalism, or at least from intentionalism in its purest form.
A position in this general class is what Shoemaker (1996) calls literal projectivism. Literal projectivism starts by assuming that experiences have qualia, construed as non-intentional and intrinsic properties of experiences of which we are directly aware. It then says that experiences represent objects in the world as instantiating these qualia, and also that an experience represents an object as instantiating a particular quale Q just in case the experience itself instantiates Q. This sort of view is quite consistent with the supervenience of the phenomenal on the intentional—indeed, it entails such a supervenience thesis—and yet it is difficult to classify this position as a bona fide version of intentionalism.
What justifies the claim that literal projectivism and related views are not bona fide versions of intentionalism? The distinction here is elusive, but I think the main point might be brought out by noting that intentionalism is usually developed within the context of what is in effect a certain sort of reductionism, according to which the difference between psychological states is a matter of variation along two dimensions, informational content and functional role. Every psychological state is to be distinguished on this view, depending on its informational content and functional role. So, for example, on this view, the difference between belief and desire, and between one belief and another, is a difference that is capturable somehow or other in a difference between informational content and functional role. Similarly, the difference between experience and belief, and between one experience and another, is a difference that is capturable somehow or other in a difference between informational content and functional role. The problem with literal projectivism is that it is most naturally understood as appealing to a third dimension of difference, namely, whether there are qualia or not, and this justifies us in regarding it as not a version of intentionalism.
In setting aside some positions as ‘not bona fide’ versions of intentionalism, I am not suggesting that the assessment of these views is simple or straightforward. Nor do I mean to suggest that they are in any sense less plausible than intentionalism. It might certainly be that adopting one or other of these compromise views can be justified by argument. Still, in investigating the consequences of intentionalism I mean to be investigating the consequences, not of these compromise positions, but of bona fide intentionalism. So it follows that one possible response to our inquiry is to abandon intentionalism in favor of one or another of these compromise views.
So far I have been concerned simply to review some central ideas associated with intentionalism. But, returning to our main theme, what of the transitivity argument and the impact of intentionalism on the hard problem? I said earlier that my own approach will be to suggest that the argument involves a fallacy of equivocation. But we will have a deeper understanding of the issue if we appreciate the different possibilities before us.
The first possibility is to deny the first premise. One version of this option is to adopt the view that as a matter of fact every experience has both intentional and phenomenal aspects, but it is not the case that the former supervenes on the latter, or at least does not do so with metaphysical necessity. Another version explores the possibility of weakening the supervenience thesis that intentionalism invokes. If the supervenience thesis at issue were a ‘within a world’ thesis or a natural supervenience thesis, then we would not be able to derive the troubling conclusion. Notice that both positions are consistent with Brentano’s Thesis which says only that every psychological state has some intentionality or other.
The problem with denying the first premise is that there is considerable theoretical reason to adopt intentionalism (or at least the supervenience thesis it implies) in some form or other. Suppose we agree that there are two features of an experience, its intentionality and its phenomenal character. Now the question to be asked is, what is the relation between them? Considerations of theoretical economy suggest that we should suppose that there is a dependency relationship here.
It is sometimes said that there are reasons beyond theoretical economy for supposing that intentionalism is true. For example, one argument here is the so-called argument from diaphanousness. The starting point of this argument is the thesis of diaphanousness, the phenomenological point that one attends to the phenomenal character of one’s experience by attending the objects and properties presented in that experience. For example, when I focus on the phenomenal character of my experience of looking at the gray filing cabinet, I apparently do so by focusing on the gray filing cabinet. From this it is concluded that a state with phenomenal character must also exhibit intentionality—experiences with phenomenal character seem by their nature something that we have introspective access to, and if one has access to experiences by concentrating on their intentionality, it is reasonable to say that every experience has intentionality. The final step of the argument is to draw the conclusion that intentionalism is true.
However, the problem with this argument is that it does not establish the supervenience thesis in the strong form that is required by intentionalism. All it establishes is that every phenomenal state or event is a state or event with intentional content—all that is established, in other words, is a version of Brentano’s thesis. But in order to move from Brentano’s thesis to supervenience one will need a consideration such as theoretical economy. I am not suggesting that this consideration is absent or implausible—quite the contrary. But I am suggesting that it is necessary component in any case for intentionalism.
The second possibility is to deny the second premise and deny that the intentional supervenes on the physical. From this point of view, the supervenience of the intentional on the physical is as problematic as the supervenience of the phenomenal on the physical and for the same reasons.
One thing to say about this possibility is that, in the form that I have been discussing it, the supervenience of the intentional on the physical is certainly questionable. For in the form that I have been discussing it, the thesis is true of any two experiences in any two worlds. While this is a reasonable assumption when the focus is on intentionalism, it is much less reasonable when the focus is on physicalism, as it is in the second premise and the conclusion of the argument. However, as I noted previously (fn. 4) this is only a simplification. Horgan, Lewis and others have developed proposals about how to transform a supervenience thesis which applies to any two worlds into a supervenience thesis that applies to our world in particular, and there is no reason why those proposals could not be implemented in the present case. Of course, in order for the argument to remain valid, the supervenience thesis in the conclusion will likewise need to be adjusted but there appears to be no special problem in doing so.
Apart from the issue of the formulation of the supervenience thesis, however, the general idea that we should abandon premise two is surely an overreaction to the argument. At any rate we should certainly not adopt the position that in the case of intentionality that has no connection to the standard examples that motivate the hard problem—mere intentionality, as I will call it—we have reason to suspect a failure of supervenience on the physical. For it is simply not the case that the relation between mere intentionality and the physical presents us with the same sort of problem as the relation between the phenomenal and the physical. In the latter case, we have a series of arguments, such as the knowledge argument and the conceivability argument, which make it plausible that the relation between the phenomenal and the physical is contingent, and so not a matter of supervenience. But we do not have analogues of these arguments in the case of intentionality as such.
This is not to say that there are no puzzles about the relation between the merely intentional and the physical. For example, Kripke’s (1992) Wittgenstein provides precisely such a puzzle. But discussions of problems such as those posed by Kripke’s Wittgenstein standardly proceed on the assumption that the intentional does supervene on the physical. These problems raise what are sometimes called vindication questions—i.e., questions about explaining or justifying a supervenience thesis which is presumed to be true. But in the phenomenal case we are not faced with a vindication project, or at least are not in the first instance. We are not worried about vindicating a supervenience thesis which we agree to be true. We are rather attempting to defend the truth of a supervenience thesis in the first place.
The final possibility is to accept that the transitivity argument is sound, and therefore that its conclusion is true. I have already said that I will not be concerned with the truth of the conclusion, but I do think that we should be suspicious of any attempt to endorse the soundness of the transitivity argument.
One problem is that the transitivity argument presents a route to materialism about the phenomenal which is altogether too quick. One may see this if one notices that neither of the premises of the transitivity argument have by themselves any connection to the arguments that people typically discuss in the course of discussing the hard problem, such as the knowledge argument or the conceivability argument. It is easy enough to imagine philosophers who suppose that the knowledge argument (for example) is sound, and yet accept the first premise the transitivity argument. And then again it is easy enough to imagine philosophers who suppose that the knowledge argument is sound in the case of phenomenal properties, think that nothing of the sort is true in the case of the intentional properties, and so accept the second premise of the argument.
Of course, one might say that the transitivity argument is simply another argument for the conclusion that one should support the supervenience of the phenomenal on the physical, an argument that is quite independent of the knowledge argument. In this sense it is fruitfully compared to causal arguments which suggest that, regardless of the persuasiveness of the knowledge and conceivability arguments, one should accept that the phenomenal supervenes on the physical, the reason being that the supervenience of the phenomenal on the physical is the best way of legitimating the causal efficacy of the phenomenal.
However, there is a difference between any version of the causal argument and the transitivity argument. It is true that causal arguments of one or another variety present the possibility of a completely independent line of argument in favor of the supervenience of the phenomenal on the physical. But—as I will argue in more detail in a moment—it is not clear that the considerations lying behind the transitivity argument are in this sense independent. Suppose that one has reasons to deny the conclusion of the transitivity argument, say on the basis of the knowledge argument. It seems very plausible to suppose that those very reasons or at least closely related reasons are going to give you reasons to deny at least one of the premises of the transitivity argument. But if so, it is mistaken to invoke the transitivity argument as an independent route to the denial of the conclusion of the knowledge argument.
If we are faced with a formally valid argument which looks too good to be sound, and we cannot see a straightforward way to deny either of its premises, the only other option is to suggest that there is a fallacy of equivocation in the reasoning, and so that the argument is in that sense unpersuasive. In this section I want to argue that this is in fact the case.
(i) The objection stated.
The reason for supposing that the transitivity argument is unpersuasive can be stated in outline very simply. Suppose first that there is a certain sort of information which is the informational content of an experience—call it, without prejudice, experiential information. If experiential information has the resources to ground the phenomenal character of experience, then—the objection goes—it is very likely that the hard problem of phenomenal character will simply be transposed into the hard problem of experiential information, a problem which is the direct counterpart of the original hard problem. On the other hand, if experiential information is such that it does not on its own give rise to the hard problem, or to a counterpart problem, then it is very plausible to suppose that experiential information does not ground phenomenal character, and that intentionalism is false.
If this objection can be made good, it follows that the transitivity argument is unpersuasive. For that argument implicitly invites us to consider a notion of experiential information which, first, supports the phenomenal character of experience, but, second, does not raise the problems of phenomenal character. On the other hand, if the objection we just stated is correct, it follows that there is no such notion: the notion of experiential information presupposed in the first premise is distinct from that presupposed in the second.
In order to bring out this objection in further detail, we may concentrate on a particular example. Thus, suppose that (4) reports a particular visual experience.
(4) It appears to John that the book on the table is red.
Of course, the suggestion involves a considerable amount of idealization. For one thing, not everyone accepts the propositional form of (4) as appropriate to the nature of experience. In addition, any genuine experience will be considerably richer in respect of informational content that one would predict from (4). Nevertheless, it is reasonable to abstract away from these issues for purposes of discussion. We may think of (4) a microcosm of the debate, and therefore a useful example with which to work.
Now, the question I want to ask about (4) is this: how is the occurrence of ‘red’ here to be interpreted? The informational content of the experience reported by (4) is presumably encoded in the proposition denoted by its constituent ‘that’-clause. So—to put our question differently—what is that proposition? What I want to suggest is that however one answers the question, the transitivity argument fails.
(ii)The physicalist interpretation.
One answer to our question is suggested by the physicalist account of color. According to this account, the property red is to be identified with some physical property, say a spectral reflectance property of some sort—to fix ideas, let us say that, according to physicalism, red = ref, where ‘ref’ is some spectral reflectance property. If we answer our question in the light of this account, we might say that (4) is to be understood as (4-phys):
(4-phys) It appears to John that the book is ref.
In saying that this is an interpretation suggested by physicalism about color, I do not mean to suggest that physicalists about color would necessarily endorse it. In fact, one cannot derive (4-phys) from (4) without helping oneself to significant and controversial premises from the philosophy of language. It is not at all clear that a physicalist about color need or should endorse those premises. So the point at this stage is not that this is the interpretation of the occurrence of ‘red’ favored by the physicalist, nor that this interpretation is particularly plausible. The point is only the weaker one that this is a possible interpretation of that occurrence.
Suppose then that (4) is in interpreted as (4-phys). How does this suggestion bear on the premises of the transitivity argument? Well, on this interpretation what the first premise of the argument says or implies is that if two experiences are the same in representing physical properties such as ref, then they are the same in terms of phenomenal character. And what the second premise says or implies is that if two experiences are the same in terms of physical character, they are same in terms of whether they represent the world as instantiating these physical properties.
But on this version of the argument, while it seems plausible to say that the second premise is true, the first premise is false.
The reason for rejecting the first premise has been set out by Ned Block (1990, 2000). One example he uses is the inverted, acclimatized, amnesiac, Fred (as we may call him). Poor Fred proceeds through four stages. At stage 1, Fred is normal: he sees green grass, his experience has a phenomenal character of the green variety—i.e., the phenomenal character normally associated with experiences as of grass—and his verbal behavior is of a kind that is usual for a subject presented with green grass, i.e., he is disposed to say ‘that’s green’, &c. At stage 2, Fred is fitted (never mind how) with an inverter: he sees green grass, his experience has a phenomenal character of the red variety—i.e., the phenomenal character normally associated with experiences as of tomatoes—and he says `that's red'. At stage 3, thirty years later, Fred is acclimatized: he sees green grass, his experience has phenomenal character of the red variety, he is disposed to say ‘that’s green’, but he is also disposed to say ‘I remember when green things produced an experience with a different character’. At stage 4, Fred loses his memory: he sees green grass, his experience has phenomenal character of the red variety, he is disposed to say ‘that's green’. What Block points out is that the experience that Fred has at stage 4 would seem to be different in the relevant respects from the experience he is having at stage 1. Nevertheless, the informational content of the experience is the same, so long as one interprets that content in the way suggested by the physicalist account of color. To put matters in our terms, what Block argues is that that if one interprets (4) as (4-phys) then it is very plausible to deny intentionalism on the basis of examples such as Fred. So Fred constitutes a decisive counterexample to intentionalism, so long as experiential information is understood in this way.
Now, Block himself wants to draw the more general moral from this conclusion that intentionalism is false. In light of the fact that there are other interpretations of the occurrence of ‘red’ in (4), and so other versions of intentionalism, that conclusion looks premature. But the important point for us is that on the first interpretation we have considered, intentionalism, and so the first premise of the transitivity argument, is false.
(iii) The dispositionalist interpretation.
The main rival to the physicalist account of color is the dispositionalist or Lockean account. According to this account, being red is to be identified with the disposition to look red to normal observers in normal contexts—to fix ideas, let us suppose that on this interpretation, red = disp, where is the disposition on the part of an object to look red in the right circumstances. If we answer our question in the light of this account, we can say that (4) is to understood as (4-disp):
(4-disp) It appears to John that the book is disp.
Again, in saying that this is an interpretation suggested by dispositionalism about color, I am not suggesting that dispositionalists about color would necessarily endorse it. Once again, one cannot derive (4-disp) from (4) without helping oneself to significant and controversial premises from the philosophy of language. It is not at all clear that a dispositionalist about color need or should endorse those premises. So again the point at this stage is not that this is the interpretation of the occurrence of ‘red’ favored by the dispositionalist, nor that this interpretation is particularly plausible. The point is only the weaker one that this is a possible interpretation of that occurrence.
Suppose then that (4) is understood as (4-disp). How does that bear on the transitivity argument? Well, on this interpretation, the first premise of the argument says or implies that if two experiences are identical in terms of representing objects as being disp, then they are identical in terms of phenomenal character. And what the second premise says or implies is that if two experiences are identical in physical respects, they are identical in respect of representing objects as being disp.
But interpreted this way the situation is the reverse of that which obtained in the case of the physicalist interpretation: while the first premise seems plausible, the second premise is implausible, or rather is as implausible as the conclusion of the transitivity argument, and for the very same reasons.
The reason for the implausibility of the second premise comes out when we notice that, unlike (4-phys), (4-disp) includes an occurrence of ‘red’—at least it does so when we recall that ‘disp’ simply abbreviates ‘disposed to look red in the right circumstances’. There is notoriously a question for the dispositionalist about how to interpret this occurrence, but on any way of interpreting it, what the dispositionalist is saying is that objects which are red are disposed to produce experiences which have a certain phenomenal character. Understood this way, the second premise says that if two experiences are identical in physical respects they are identical in respect of representing objects as being disposed to produce this phenomenal character. But if that is what the second premise says it is vulnerable to the very same lines of argument that challenge the conclusion.
One of these lines of argument, for example, is the knowledge argument, which begins by pointing out that one could know all the physical truths about an experience without knowing about its phenomenal character. The conclusion is then drawn that the phenomenal truths cannot be wholly a matter of the physical truths concerning an experience. Of course, there are a number of questions about whether such a conclusion could be drawn from this premise. However, whatever one thinks of the details of the argument, surely the very same line of argument, or at least a counterpart of the argument, could now be used to threaten the second premise of the transitivity argument in the sense that we are intending it. For consider: to the extent that it seems plausible to say that one might know all the physical truths concerning an experience and not know the phenomenal truths, it is equally plausible that one might know all the physical truths about an experience without knowing if or whether it represents objects as being disposed to produce this phenomenal character. To make the case vivid, imagine someone who has forgotten what the phenomenal character of an experience of red is. According to the knowledge argument, such a person could not work out for himself what the experience is like, no matter how much physical information he had at his disposal. Similarly, imagine someone who has forgotten whether an experience of red represents objects has producing this phenomenal character—again, that person could not work this out from physical information alone. In short, if (4) is interpreted as (4-disp), the second premise of the transitivity argument is as controversial as its conclusion, and for precisely the same reasons.
It is important to note that in saying that, on this interpretation, the second premise of the transitivity argument is implausible in the way that the conclusion of the argument already is, I am not saying that the premise is implausible sans phrase, still less that it is false. What I am saying is that, on this interpretation, the argument could not be persuasive because the reasons for doubting the conclusion of the argument are the very same as the reasons for doubting one of its premises. It is in this sense that the transitivity argument is implausible.
(iv) The primitivist interpretation.
The final of interpretation I will consider is that suggested by primitivism about color. It is often argued against dispositionalism that colors do not appear to be dispositional properties of an object (e.g. McGinn 1996). Rather, the color of an object appears to be an intrinsic or non-relational feature of an object, rather like its size or shape, a property that is smoothly and homogeneously distributed across the surface of objects, and one that is irreducible to other physical properties that qualify objects. According to primitivism, this appearance should be taken at face value: colors are irreducible and intrinsic properties of physical objects—to fix ideas, let us say that, on this view, red = prim, where ‘prim’ is some primitive property. On this view (4) is interpreted as follows:
(4-prim) It appears to John that the book on the table is prim.
Now, while primitivism is attractive from a phenomenological point of view, according to many philosophers it cannot be the truth about what colors actually are. We know enough about the physics of the situation, they say, to know that objects do not have primitive homogenous intrinsic properties. So on this interpretation (4-prim) attributes an erroneous experience to John. Once again, however, my interest here is not so much with (4-prim) as what it tells us about the transitivity argument
Suppose then that (4) is interpreted as (4-prim); how does this bear on the argument? On this interpretation, the first premise of the transitivity argument says that two experiences identical in representing an object as prim are phenomenally identical. And the second premise says that two experiences which are physically identical are identical in respect of whether they represent objects as prim.
So interpreted, however, our response to the argument should parallel our response to the argument on its dispositionalist interpretation: while the first premise now looks plausible, the second is implausible, and for the very same reasons that the conclusion of the argument looks implausible. Once again, suppose someone rejects the conclusion of the argument on the basis of the knowledge argument; such a person would likewise reject the second premise of the argument interpreted in the light of the primitivist account of color. If physical knowledge about an experience does not suffice for knowledge of what the phenomenal character of an experience is, then it likewise does not suffice for knowledge of whether the experience represents objects as being prim, as opposed to anything else.
As before, I am not saying here that the knowledge argument presents a good reason to reject the conclusion of the transitivity argument. I am simply saying that, if one thought that it did, one would likewise think one had a good reason to reject the second premise of the argument. As before, therefore, the argument is unpersuasive.
(v) Objections and replies.
My overall objection to the transitivity argument is that there is no way to interpret its two premises so that they are both plausible. If the physicalist interpretation is adopted, the first premise of the argument is false. If the dispositionalist or primitivist interpretation is adopted, then while the first premise is true, or seems true, the second premise looks no more plausible than the conclusion of the argument. On no interpretation, therefore, is the argument plausible.
How might one respond to this line of argument? First, one might wonder whether the argument has covered all the cases. Perhaps there are theories of color which I have not considered? I grant that this is a possibility. But philosophers have been discussing color for a long time, and these three accounts seem to be the ones they have converged on. So it is plausible to suppose that these three accounts exhaust our options, or at least our relevant options.
Second, one might object that the argument is being distorted by the formal mode in which I have put the question. The semantic question of how one understands the word ‘red’ as it occurs in (1) seems quite unlike the questions about the intentionality of the experience that are my main concern.
I myself am somewhat skeptical of the methodology lying behind this objection, but nevertheless I think there is a way of putting the point that abstracts away from the semantic issues. For suppose we introduce the notion of a visual world, i.e. the world (or, better, set of worlds) as it is presented in visual experience. We can now ask: what is the theory of color that is true at the visual world? This is of course not the question of what theory of color is true at the actual world. It is open to someone to say that the theory of color true at the actual world is not the theory that is true at the visual world. Typically, when people describe themselves as (e.g.) physicalists about color, they are saying that physicalism about color is true at the actual world. However, that commitment leaves open what theory of color is true at the visual world. What I have been saying is that if physicalism is the theory of color that is true at the visual world, then intentionalism is false; if either dispositionalism or primitivism is true at the visual world, then intentionalism is true but the hard problem remains untouched.
Third, one might object that the versions of intentionalism suggested by the interpretation of (4) as either (4-disp) and (4-prim) are not bona fide version of intentionalism. We noted in §3 that some positions, while agreeing with the supervenience of the phenomenal on the intentional should not be thought of as genuine versions of intentionalism on the grounds that they do not comfortably sit with the reductionist approach to mental states according to which the difference between any two psychological state is to be accounted for as a difference between informational content and functional role. Perhaps it might be thought that to construe (4) as (4-prim), say, is to take up one of those positions.
However, in considering the possibility that (4) be rendered as (4-prim), we have not departed from the root idea of intentionalism. The primitivist conception provides us with a particular picture of what the world is like, in fact it presents an account of colors that are is wholly separable from one’s experience. We might believe or want the world to be that way, and we might represent the world to be that way in experience. But none of this suggests that the difference between mental states is not to be explained ultimately in terms of a differences along the dimensions of informational content and functional role.
Finally, one might object that my suggestion in the case of the dispositionalist or primitivist interpretation of the argument is too quick. The intentionality of a mental state is often understood in relational terms—to be in an intentional state is a matter of being related in such and such a way to a property or a proposition. However, we noted previously that, the knowledge argument, for example, could not be developed against the mere intentionality of a mental state in the same way as it could be developed against its phenomenal character. This suggests that, so far as physicalism is concerned, there is no problem about my merely being related in the intentional way to a property or a proposition consistently. However, that conclusion is apparently independent of what precisely are the relata of the relation, i.e. of what sort of properties or propositions constitute the relata of the relation. And one might infer from this, that just as we apparently have no problem as far as physicalism goes when the relata of the relation are properties such as ref, we should likewise have no problem when the relata of the relation are properties such as prim.
However, the problem with this suggestion is the inference from ‘there is no problem about standing intentional relation to some property as such’ to ‘there is no problem about standing in an intentional relation to some property where doing so involves some phenomenal character’. In general, we know that there were certain aspects of the psychological realm that are the subject matter of the hard problem. The observation that that these aspects are intentional and that there is no problem in general about intentional states, does not entail that there is no problem about phenomenal character.
My interest in the transitivity argument is driven by a background interest in the relation between intentionalism and the hard problem. My assumption has been that if the transitivity argument is persuasive then intentionalism would certainly have an impact on the hard problem. But I have argued that it is not persuasive – there is no interpretation of its central terms which allow for the joint truth of both premises. To that extent, I think it is reasonable to conclude that intentionalism is going to have a minimal impact on the hard problem.
At this point, there are two distinct directions in which the discussion might proceed. First, taking to heart the idea that intentionalism will have a minimal role in a solution to the hard problem, one might go on to ask what would play a role in such a solution. Second, one might go on to ask what the consequences of intentionalism are, if the doctrine does not impact on the hard problem. To discuss the first issue would be to take up the question of what the solution to the hard problem is. I have already sworn off any such ambitions for this paper. So what I want to do in the remainder of the paper is explore a suggestion about what the other consequences of intentionalism might be. For it seems to me that, while intentionalism has a minimal impact on questions of phenomenal character, what we have said so far suggests that its impact is surprisingly large in an area which is to some extent a close cousin of the issue of phenomenal character, namely, the issue of the secondary qualities. In particular, I want to suggest that intentionalism implies the truth of the error theory, the doctrine that experience of color is systematically false.
The relation of intentionalism to the question of color realism is an abstract issue, even by the standards of philosophy of mind. So, to focus matters, I will concentrate on an argument which tells us that if intentionalism is true, then projectivism must be true also—I will call it the dual role argument.
The dual role argument starts with the observation that concepts of color play a dual role in our thought and talk. On the one hand, color concepts can be used to attribute properties or qualify physical objects (including volumes and lights), as when one says that the table is red. Let us call this the qualifying role of concepts of color. On the other hand, color concepts can be used to characterize the content of an experience, as when one says that one is having an experience as of red. Let us call this the characterizing role of concepts of color. Of course, to some extent the distinction is a general one. One could say the same thing about concepts of furniture or trees.
Now, so far my concentration has been on the characterizing side of characterizing/qualifying distinction. Earlier, when we considered the idea that ‘red’ in (4) might be understood in accordance with the physicalist, dispositionalist or primitivist conception of color, this was a suggestion about whether these conceptions of color could play the characterizing role. In effect our suggestion was that only some of these conceptions of color could plausibly be drawn on to play the characterizing role, compatibly with the assumption of intentionalism.
But this point raises another more basic issue, namely, is there a concept of color that can play both roles? Reflection on this question seems to confront us with a dilemma, a dilemma which comes out most clearly if we concentrate initially on the contrast between physicalism and primitivism. Resolving this dilemma will lead, I think, to the conclusion that intentionalism and error theory go together.
Suppose it is thought that the primitivist conception of color is the one that can play these two roles. This suggestion is plausible if one is concentrating on the characterizing role of the concept of color. But it is not plausible if one is concentrating on the qualifying role. For it is not clear that physical objects have any of the properties attributed to them by the primitivist account, so an error theory of color seems to be the only answer on this conception. In short, on this horn of the dilemma, we get the result that intentionalism is true, and yet error theory is also true.
Suppose on the other hand it is thought that the physicalist conception of color is the one that plays both roles. This suggestion is plausible if one is concentrating on the qualifying role of the concept of color—for it seems clear that physical objects do have the properties attributed to them by the physicalist conception, and so an error theory is avoided. But it is not plausible if one is concentrating on the characterizing role. For if the physicalist conception plays the characterizing role we lose the plausibility of intentionalism. In short, on this horn of the dilemma, we get the result that projectivism is false, and yet intentionalism is false also.
The contrast between physicalism and primitivism is certainly suggestive of the conclusion of the dual role argument, namely intentionalism and error theory go together. However before this conclusion is established we need to consider two obvious responses. The first response is dispositionalism. The second response is to give up a presupposition of the argument, namely that there is a single conception of color which plays both roles. I will argue that the second response to the argument is the best one, but that this second response leaves an important grain of truth in the error theory. But first we need to consider dispositionalism
The most straightforward way to respond to the dual role argument is to propose that a dispositionalist conception of color can play both the qualifying role and the characterizing role. There seems to be no particular question about physical objects having the dispositional properties discussed in the dispositionalist conception, and so there seems no problem in its playing the qualifying role. However, there are at least two objections to the idea that the dispositionalist account plays the characterizing role, or at least can do so without a revision of the notion of experience.
The first objection is the phenomenological objection that, as it is sometimes put, ‘colors do not look like dispositions’. In some cases this objection is presented as showing directly that colors are not dispositions, and so cannot play the qualifying role. So understood, the objection is mistaken (cf. Byrne 2001a). The mere fact that colors do not look like dispositions does not entail that they are not dispositions. However, there is another way to state the objection, according to which it is much more powerful and much more pertinent to our concern. On this interpretation, the phenomenological objection shows, not that colors might not be dispositions, but that they are not represented as such in the content of visual experience. This would show that the dispositional conception of color cannot play the characterizing role, for if it did then colors should look like dispositions.
The second objection is the circularity objection (cf. Boghossian and Velleman 1989). We noted earlier that there is famously a question for the dispositionalist, namely, how to account for the occurrence of red in ‘It appears to John that the book is disposed to look red’. One possibility here is to construe ‘red’ in this context as denoting a sensory property (Peacocke 1983). But this position is of course unavailable to the intentionalist. Another possibility is to refuse to say what the occurrence means at all, and simply reiterate that ‘red’ means red. But now it is unclear what the difference is between it appearing to John that the book is red and its appearing to him that the book is green. An intentionalist needs to say what the difference between these two states, and the problem is that dispositionalism seems to be incapable of explaining this difference.
Lying behind these objections is the fact that dispositionalism is usually developed against the background of theory of perceptual experience other than intentionalism. If one can appeal to a notion of sensation which is theoretically in good standing, then it is easy to define up the property of being caused to produce that sensation. But if one is operating with an intentionalist conception of experience, it is not so clear that dispositionalism can be developed, or at least can be developed as a conception which is intended to play the characterizing role.
It might perhaps be that dispositionalism can be developed in such a way that it avoids the phenomenological and the circularity objections. But I think that taken together these objections strongly motivate the search for an alternative response to the dual role argument. And in fact, there is a response available. The alternative is to reject a presupposition of the argument, namely that a single concept can play both the qualifying role and the characterizing role. If one adopts the hypothesis that the primitivists conception plays the characterizing role, and the physicalist conception (or the dispositionalist conception) plays the qualifying role, one may avoid the dual role argument.
Motivation for this sort of ambiguity-based account can be brought out as follows. A lesson of much recent literature on color is that not everything that we spontaneously or naturally believe about color can be true simultaneously. Mark Johnston brings out the point vividly by suggesting that, when it comes to color, we need to speak less inclusively in order to speak truly. The problem that we are facing here is similar in structure to that faced by Johnston, namely, there is a conflict between two things we would like to say about color experience, first that intentionalism is true, and second that color experience is in general veridical, and so that there are in fact properties of physical objects which deserve the name ‘color. What the ambiguity theory reminds us is that the conflict arises because we tacitly believe something further, namely that the conception of color that plays the qualifying role is the same as that which plays the characterizing role. If that belief is overcome or resisted, then the conflict is effectively quarantined, since the conception of color appropriate to characterizing experience is distinct from the conception appropriate to qualifying objects.
Of course the ambiguity view does not completely defeat the dual role argument. For it remains true that color experience remains illusory, and moreover does so in a systematic way. Instead of saying that the ambiguity view defeats the dual role argument, therefore, it is better to say that the view provides us a way to re-interpret our talk about color in such a way that it comes out true, at least most of the time. In Johnston’s terms it provides a strategy for speaking ever less inclusively about the colors, in such a way that color talk comes out generally as true.
I have been considering two lines of argument designed to bring out the consequences of intentionalism. The first line of argument, summarized as the transitivity argument, purported to show that intentionalism would directly have an impact on the hard problem. I argued that the transitivity argument was unpersuasive, and drew the more general conclusion that intentionalism has no impact on the hard problem. The second line of argument, summarized as the dual role argument, purported to show that intentionalism would directly have an impact on the issue of the secondary qualities, and in particular that it entails that the error theory is true. I suggested that the dual role argument is persuasive, but sought to minimize the impact of the argument by providing an ambiguity view.
I want now to close the paper by considering an objection that brings together the two lines of argument I have examined. The objection is that there is an asymmetry in our response to the two arguments. When responding to the transitivity argument I never once took seriously the possibility that one might provide a revisionary or eliminativist account of experience. On the other hand, when discussing the dual role argument, I endorsed an eliminativist or revisionist conception of color. What justifies this asymmetry in attitude? Why is eliminativism about color a view to be taken seriously while eliminativism about color experience is not?
One response says that eliminativism about experience is an incoherent position while eliminativism about color is not. This line of argument goes roughly as follows. The question of whether experience exists raises a question concerning a contingent matter of fact. But in raising a question of contingent matter of fact we have no recourse but to involve ourselves in empirical enquiry. But empirical enquiry by its nature relies on experience. It follows—or so the argument goes—that there is no possibility that the outcome of an empirical enquiry could be that there are no experiences. We would be involved in a pragmatic or perhaps methodological contradiction, if not a formal contradiction, if one thought that one could have an empirical argument for the conclusion that experiences don’t exist.
The problem with this argument is that it does not allow the eliminativists a distinction that is available to them both here and elsewhere, a distinction between various different conceptions of experience. On the one hand, there is what I will call the target conception—the conception of experience according to which, there are no experiences, if eliminativism is true. Then there is what I will call the replacement conception—the conception according to which there are experiences. When confronted with the paradoxical aspect of their position, eliminativists are quick to point out that they are not denying experience in every sense of the word. They emphasize that from their point of view, experience does exist in the only sense possible: according to the replacement conception. In short, while it would be certainly convenient for my position to say that it is incoherent to eliminate experience, while not incoherent to eliminate color, to say that also seems to be indefensible in general.
If one should not respond to the objection with the response that eliminativism about experience is incoherent how should one respond? In my view, the crucial difference between eliminativism about color and eliminativism about experience emerges when we notice that any argument to eliminativism must at some point appeal to an empirical premise. In the case of eliminativism about color, for example, the premise in question has to do with the empirical facts concerning the objects, light and volumes that are colored. It seems reasonable in those cases to say that we know enough about that situation to know that such things do not have the properties attributed to them by experience. In the case of eliminativism about experience, on the other hand, no parallel claim is plausible. We simply do not know enough about the physical facts underlying experience to be able to draw the inference that eliminativism about experience is true.
A common metaphor when thinking about the secondary qualities and the mind, at least as these were thought of in the seventeenth century, is ‘the dustbin of the mind’—the general idea is that various aspects of the world are simply projections of our minds rather than real aspects of the world. This metaphor went along with a conception of the mind that was metaphysically dualist, and so I think should be rejected. Nevertheless, the dustbin metaphor is in many ways apt, for it reminds us that there is our ignorance of these matters is so profound, so profound that it can sometimes seem like a metaphysical mystery. What I have been suggesting is that, while the issue is mysterious alright, there is no reason to view it as metaphysical.
Armstrong, David. 1968. A Materialist Theory of the Mind. Routledge and Kegan Paul.
Boghossian, P. and Velleman, D. 1989. ‘Colour as a Secondary Quality’, Mind, 98 (1989) pp. 81-103.
Block, N. 1990. ‘Inverted Earth’, in Tomberlin, J. (ed.) 1990 Philosophical Perspectives, Vol. 4, Ridgeview, pp.52-79. Reprinted in Block et alia, (eds.) 1997 The Nature of Consciousness: Philosophical Debates, MIT Press.
Block, N. 2000: ‘Mental Paint’. In Hahn, M and Ramberg, B (eds.) Essays in Honor of Tyler Burge, MIT Press. References are to the PDF version.
Byrne, Alex and Hilbert, D. 1997. ‘Introduction’. In Byrne, A and Hilbert, D. (eds.) 1997. Readings On Color Volume I: The Philosophy of Color. MIT Press.
Byrne, Alex. 2001. "Intentionalism Defended", Philosophical Review 110, 2001
Byrne, Alex. 2001a. "Do Colours Look Like Dispositions? Reply to Langsam and Others." Philosophical Quarterly 51 (2001)
Chalmers, David. 1996. The Conscious Mind. Oxford.
Crane, Tim. 2002. Elements of Mind. Oxford.
Davies, M. 1997. ‘Externalism and Experience’. In Block et alia, (eds.) 1997 The Nature of Consciousness: Philosophical Debates, MIT Press.
Dretske, Fred. 1995. Naturalizing the Mind. MIT Press.
Evans, Gareth. REF. ‘A Commentary on Chapter One of Strawson’s Individuals’. In Evans, Gareth. 1985. Collected Papers. Oxford.
Harman, G. 1990: ‘The Intrinsic Quality of Experience’, Philosophy of Mind and Action Theory: Philosophical Perspectives. 4:31-52.
Harman, G. 1996: ‘Explaining Objective Color in terms of Subjective Reactions’. In Villanueva, E. (ed) 1996 , Atascadero, CA: Ridgeview.
Horgan, T. 1982. ‘Supervenience and Microphysics’. Pacific Philosophical Quarterly, 63, 1982, 29-43.
Johnston, M. 1992. ‘How to Speak of the Colors’. Philosophical Studies. 68 (1992) pp. 221-63.
Kripke, S. 1982. Wittgenstein on Rules and Private Language. Harvard.
Lewis, D. 1983. New Work for a Theory of Universals’, Australasian Journal of Philosophy, 61, 1983. Pp. 343-77.
Lycan, William. 1996. Consciousness and Experience. MIT Press.
McGinn, Colin. 1996. ‘Another Look at Color’, Journal of Philosophy, 93: 537-53.
Peacocke, C. 1983 . Sense and Content. Oxford: Oxford University Press.
Tye, M. 1995. Ten Problems of Consciousness, Cambridge, MA: MIT Press.
Tye, M 2000. Color, Content and Consciousness, Cambridge, MA: MIT Press.
Shoemaker, S. 1994. 'Phenomenal Character', Nous, 28 (1994), pp.21-38
Shoemaker, S. 1996: The First Person Perspective and Other Essays, Cambridge University Press.
Shoemaker, S. 2001: ‘Introspection and Phenomenal Character’. MS.
Stoljar, D. Forthcoming. ‘The Argument from Diaphanousness’. In Stainton, R. and Escurdia, M. Language, Mind and World. Oxford??
 For very good recent discussions of intentionalism, see Byrne 2001, Crane 2002 and the references therein. It is important to note that that the statement of intentionalism in the first sentence of the paper is rough—it is clearly true that intentionalism entails the supervenience of the phenomenal on the intentional, but the reverse entailment is more controversial. See fn. 4 and §3.2 for more discussion.
 Philosophers such Harman (1990, 1996), Dretske (1995), Tye (2000) and Lycan (1996) defend intentionalism in the course of defending various approaches to the hard problem. Ned Block (2000), a central critic of intentionalism, notoriously says that the division between intentionalists and their opponents constitutes “the greatest chasm in philosophy of mind”. Byrne 2001 is more circumspect about this consequence, noting that one easily enough divorce the issue of intentionalism from the hard problem, or at least from questions about ‘naturalizing the mind’. But even Byrne says that when considering intentionalism, ‘it is easy to see that the stakes are high’, and intentionalism “holds out the promise of a ‘representational theory of consciousness’—it is difficult to put these remarks in context without assuming that Byrne thinks intentionalism has some impact on the hard problem.
 The idea of formulating these issues in terms of the transitivity of various supervenience theses I take from Davies (1998), who discusses a similar argument in a related context.
 My presentation of the transitivity argument simplifies matters in at least two directions. First, as I have already noted (fn.1) and will return to in § 3, intentionalism is better viewed as entailing (1) rather than being identified with it. Second both (2) and (3) are logically necessary whereas, as noted in §4, I take it that the most plausible versions of materialism is contingent. Neither simplification affects the main argument, however, and so for the time being they can be ignored.
 The assumption in question is that there are more than two experiences which are different in phenomenal character. The reason for the assumption can be brought out as follows. Brentano’s thesis is false if only one experience has phenomenal character and yet no intentionality. But no trouble for supervenience follows if only one experience has phenomenal character. However, I will set aside this complication, for it is difficult to see what would motivate the restriction to the single case here.
 It is worth stress at this point that the impact of these compromise views on the hard problem seems to be the same as the impact of bona fide intentionalism. However, the consequences for the secondary qualities that I will discuss in the second part of the paper are different. (Question: Is that right?)
 For a much more comprehensive discussion of the argument from diaphanousness, see Stoljar 2002 and the references therein. (Question: Does the argument in Byrne 2001 suffer similarly?)
 In brief, what is required is that from ‘It appears to S that the book is F’ and ‘F = G’ one may derive ‘It appears to S that the book is G’. In view of the fact that ‘It appears…’ contexts are plausibly hyperintensional, it is not clear at all that this inference is valid. I will not make a judgement on the validity of the inference however.
It should perhaps be noted in addition that there are some problems with (4-phys) which are extracurricular in the present context. The problem derives from the fact that reports of the general form of (4)—appearance reports—may report beliefs or experiences depending on how the basic notion of appearance is interpreted. On the phenomenal understanding of ‘appearance’, (4) reports a experience; on the epistemic understanding, (4) reports a belief. Now, it is an assumption of our discussion that the occurrence of ‘appearance’ in (4) and its ilk should be interpreted phenomenally. But while it is certainly possible that John believes that the book is ref, it is not clear that it is so much as possible, as (4-phys) says, that it phenomenally appears to him that the book is ref. To put the point differently, it is not clear that the proposition that the book is ref is a proposition that can characterize states of experience as opposed to belief.
 The premise at issue here is the same mentioned in fn. 8. It should be noted that it is more plausible to suppose that the dispositionalist conception plausibly provides the sense of the expression ‘red’, that it is that the physicalist conception, which affects the issue of substitution. (Question: can this be made more precise?)
 The issue of whether the sorts of properties that the primitivist talks about are instantiated in the physical world is a complicated issue. Many primitivists do not hold an error theory, opting instead for a supervenience based account. I think that this sort of view is implausible but I cannot argue for that here.
It should also be noted that some philosophers argue that the primitivist account is mistaken even from a phenomenological point of view. Gareth Evans 1980, for example, objects, first that the primitivist is committed to the idea that it cannot explain the colors of objects in the dark, and, second, that it is guilty of the same mistake as those who would think, in Wittgenstein’s example, that portions of the world are in pain. I think that both objections rest on a misunderstanding. On the issue of objects in the dark, a primitivist is entitled to say that it a priori that red is the color disposed to look red, and so there seems to be no trouble for objects in the dark as there is for anybody else. On the issue of the pain patches, the reason it is mistaken to apply pain to objects in the world is that it is an a priori truth that pains are bodily sensations, so just as things which are red are apples and not numbers, so thinks which are in pain are feet and not apples. But primitivists are allowed to say this as much as anyone else.
 It might perhaps be that Armstrong’s account is a different account here. He says (1968, p. xx) that the concept of color s a ‘blank or gap’ in it. But I think this account is going to face the same problem that the physicalist account, namely that it gets the phenomenology wrong.
 It is important not to overstate this conclusion. The hard problem, as I understand it, is focused on the apparently contingent relation between conscious experience an anything physical. But there are other aspects of the overall question of consciousness—e.g. epistemological aspects—and intentionalism might well make those aspects easier to solve.
 There is also a third objection that might be mentioned. Some experiences are non-conceptual in the sense that one can have an experience as of an F without knowing what an F is (in the sense of ‘knowing what’ demanded by concept possession). But it is unclear that this is true in the case of experiences as of a disposition to F (cf. Byrne 2001a). On the other hand, one can clearly have an experience as of red, without knowing what red is (in that relevant sense). So an experience as of red is simply not an experience as of disposition to look red. This does not strictly entail that they are phenomenologically different, for it is possible that the phenomenal character of an experience is multiply realized in its intentionality.
 This is a suggestion made by Boghossian and Velleman 1989 but they do not develop it.
 It is important to notice that the ambiguity view really does indeed import an ambiguity in the picture. One might miss this for one of two reasons. First, one might take the ambiguity theory as suggesting only that ‘red experience’ does not quite mean what ‘red ball’—‘red ball’ attributes redness to a particular ball, but ‘red experience does not attribute redness to anything. However, the difference between ‘red ball’ and ‘red experience’ is not owing to an ambiguity in ‘red’, it is owing the structural ambiguity between the two expressions. What the ambiguity theory is suggesting on the other hand is that ‘red’ in ‘the ball is red’ and ‘red’ in ‘It appears to John that the book is red’ do not quite mean the same thing. Second, one might take the ambiguity theory as suggesting only ‘red’ is associated with two semantic values, a sense and a reference, and that in (1) it is should be understood as expressing its sense rather than its referent. This suggestion turns on whether the primitivist account of color can be thought of as a proposal about the sense of the expression ‘red’, while the physicalist or dispositionalist account can be thought of a proposal about the reference. The problem is that, on the usual interpretation of the sense/reference distinction, the sense of an expression formulates or is associated with a condition which is then satisfied by the reference. But colors-as-the-physicalist construes them precisely do not satisfy the condition associated with the primitivist conception.