Senses for Senses

Brad Thompson

Department of Philosophy

Southern Methodist University

 

bthompso@smu.edu

Phenomenal Content

 

When I open my eyes and look at a Rubik’s cube, there is something it is like for me visually in looking at it. Various color qualities are presented to me, and they are arranged in a specific pattern. By having an experience of this particular sort I am thereby visually representing the world as being a certain way. If I experience a blue square to the left of a red square, I represent that in the world outside my experience there is roughly something blue and square-shaped to the left of something red and square-shaped.[1] This experience differs from an experience as of a yellow triangle at the center of my visual field both phenomenologically and representationally. In virtue of the difference in phenomenology there is a corresponding difference in how the world is represented as being.

Moreover, it seems that any two experiences with the same phenomenal character will share a certain sort of intentional content.[2] If two subjects have phenomenally identical experiences, there is an important sense in which the way the world appears to them is precisely the same. I will call this intentional content that supervenes on phenomenal character “phenomenal content”.

But how are we to understand this notion of “ways of appearing”? Most philosophers who have acknowledged the existence of phenomenal content have held that the way something appears is simply a matter of the properties something appears to have.[3] On this view, the way something appears is simply the way something appears to be. This identification is often driven by phenomenological considerations. Experience is often said to be “transparent” or “diaphanous”.[4] According to the transparency thesis, when one introspects on the phenomenal character of one’s perceptual experience one finds only the properties that objects in one’s environment appear to have. One does not find properties of one’s experience or any other perceptual intermediaries.

Perhaps due to this widespread conviction, philosophers have rarely questioned the idea that ways of appearing are simply properties that things appear to have. They have generally ignored an equally intuitive alternative—that ways of appearing are manners or modes of appearing. It is this alternative view that I will explore and defend in the present paper.

One can approach this issue more carefully by considering what kind of content phenomenal content is. Those who think that the way something appears is simply a matter of the way something appears to be adopt a Russellian theory of phenomenal content. According to Russellian representationalism, phenomenal content is purely extensional, exhausted by facts about what specific properties are represented by an experience. The vast majority of theories of phenomenal content that have been offered are Russellian.[5]

Despite its popularity, Russellian representationalism faces numerous difficulties. In what follows, I will briefly raise problems for the most popular form of Russellianism, which I call “Standard Russellianism”. I will then present and motivate a Fregean theory of phenomenal content that handles the problems faced by Standard Russellianism. Fregean representationalism claims that phenomenal content involves, in addition to represented properties and objects, modes of presentation. This view would be attractive to those who think that the way something appears is not simply a matter of the way it appears to be, but instead involves a mode of appearing. I will argue that ways of appearing are modes of presentations of external properties and objects, and present a detailed theory about the nature of the modes of presentation involved in color experience.

 

Standard Russellianism

 

According to Standard Russellianism, phenomenal character is identical to a certain sort of representational content. This content is purely extensional, involving the attribution of specific objective properties to objects. Typically, the objective properties that color experiences are said to represent are simply the colors, which themselves are identified with physical properties of external objects.[6]

Given that sameness of phenomenal character entails sameness of phenomenal content, Standard Russellian theories of phenomenal content are theories that accept the following:

For any experience with phenomenal character r, there is some mind-independent physical property p_r such that necessarily if an experience has phenomenal character r then it attributes p_r.

           

Inverted Spectra Without Illusion

 

A problem for Standard Russellianism that has been discussed by Sydney Shoemaker and others concerns the possibility of spectrum inversion.[7] Two individuals are spectrum inverted with respect to each other if their color experiences are phenomenally inverted relative to one another. For instance, it might be that the color experience Jack has when viewing a red thing is phenomenally identical to the experience Jill has when looking at a green thing.

The Standard Russellian is not threatened simply by the possibility of spectrum inversions. But she cannot say that two individuals whose color experiences are inverted with respect to each other both have veridical experiences in the same perceptual circumstances. According to Standard Russellianism, color experiences represent physical colors. If two individuals have phenomenally distinct color experiences, their color experiences represent different properties. And a physical object cannot be both red and green all over its surface. The Standard Russellian must insist that one of the two individuals perceives the colors of things incorrectly.

            But let's suppose that roughly half the human population is spectrum-inverted relative to the other half. And suppose further that they make the same discriminations among color stimuli and are otherwise behaviorally indistinguishable. The intuition against Standard Russellianism is that there seems to be no principle by which we can say that one portion of the population perceives the true colors of things while the other suffers from massive sensory error. Intuition seems to be on the side of saying that both individuals have veridical experiences upon looking at a ripe tomato, despite the fact that one is enjoying a red quale and the other enjoys a green quale.[8] The way the tomato appears to Jack differs from the way the tomato appears to Jill, but the tomato appears to be red to both Jack and Jill.

 

Color Constancy

           

A similar problem for Standard Russellianism that has received very little discussion, and which doesn’t depend on any assumptions about the possibility of spectrum inversion, concerns the phenomenon of color constancy. [9] Consider the way that a ripe tomato looks at noon in full sun versus at five o’clock in the afternoon. The tomato, both at noon and in the late afternoon, will appear to be the same color. Changes in illumination are adjusted for in such a way that the tomato will appear to be the same color despite differences in the spectral distribution of the light reaching the eye. This kind of adjustment by the visual system is called “color constancy”. But despite the fact that the tomato will look to be the same color, the tomato nonetheless looks different at noon versus five o’clock. If this isn’t immediately obvious, consider what it is like to put on a pair of tinted sunglasses or to view an object first in sunlight and then under incandescent lighting. The visually presented scene changes in its appearance, but the objects you see do not appear to have changed color.

The Standard Russellian must explain these phenomenal facts (both of looking to be the same color and having a different visual appearance) in terms of representational content, where this content consists solely in what properties the tomato is represented as having. She might appeal to the richness of visual phenomenal content. The visual system represents objects as having stable surface color properties through color constancy mechanisms.[10] But the visual system does not discard the information about illumination and the spectral distribution of light reaching the eye. So the Standard Russellian might insist that visual content is layered, consisting of a representation of illumination-invariant color properties as well as representations of illumination and current spectral distribution.[11]

But it is not clear that the Standard Russellian can consistently give this account of phenomenal content while also adhering to the transparency thesis. We are phenomenally presented with one relevant quality, not two. If this quality is to be identified with the illumination-invariant quality (as in Tye 2000), then the representationalist gets the phenomenology wrong. The tomato is represented as having this illumination-invariant quality at both times, but the phenomenology of the experiences differ.

If the represented quality is identified with some illumination-variant property, then the view faces the same problem Tye poses against dispositionalists like Shoemaker (Tye 2000). The color property directly seen doesn’t look like a relational property, but rather looks like a monadic property of the tomato.[12] Tye also objects that the property experienced as belonging to the tomato is one which is represented as being a stable mind-independent property, one which the tomato has or can have over long periods of time and independently of being perceived.

Furthermore, suppose that the experience of a portion of a solid colored piece of fabric in shadow is to be understood as being a representation of the fabric in that region as something like “red₂₇ in partial shade” or “red₂₇ with illuminance i”.[13] The appearance of that region of the shirt might be phenomenally identical to the appearance of a portion of an apple which is veridically represented as “red₁₂ in full mid-day sunlight” or “red₁₂ with illuminance i₂”.[14] The Standard Russellian’s only resource for accounting for sameness in phenomenal character is in terms of shared Russellian content. But in this case there seems to be phenomenal identity without identical content. The shirt and the apple might be veridically represented as being two different shades of red. The only level at which the visual system might deploy identical representations in the two cases concerns the spectral distribution of light striking the eye. But contra transparency, this property is neither a property of the object of perception nor represented as such.

 

 

The Way Things Look

 

In considering inverted spectrum cases, we noted that the way things appear to Jack is different from the way things appear to Jill. Nonetheless, neither Jack nor Jill is a systematic misperceiver. From the fact that the way a ripe tomato appears to Jack differs from the way it appears to Jill, plus the fact that neither Jack nor Jill misperceives the tomato, one might conclude that the properties Jack and Jill each represent a ripe tomato as having must be different properties.[15]

Likewise, we noted that due to color constancy, the way a red object appears to a single subject in the morning differs from the way it appears to that very same subject in the evening. Again, it is not plausible to suppose that at most one of those perceptions is veridical, and that the other must be a misperception of color. And so one might conclude that the two color experiences each represent the object as having different properties.[16]

            There is an inference being made here from differences in the way things appear to difference in represented properties. And there is one way of understanding the phrases “the way things appear” and “how things appear” which plausibly supports the inference. The expression “the way things appear” can be taken as elliptical for “the way things appear to be”. And on this reading of the expression, it looks to explicitly denote the properties an experience represents objects as having. If the way the tomato appears to Jack in this sense differs from the way the tomato appears to Jill, this entails that Jack and Jill represent different properties as belonging to the tomato.

            But this is not the only reading of expressions like “the way things appear”. To illustrate this, imagine that you are rearranging the furniture in your living room. You switch the sofa and the recliner and stand back and look at your living room. You might say, “I like the way the room looks.” This cannot plausibly be paraphrased as “I like the way the room looks to be.” Or you might say, “I prefer the way it looks with the recliner over there.” This doesn’t translate as “I prefer the way it looks to be with the recliner over there.” What is being evaluated in these instances of interior decorating is the manner of visual appearance of the living room—not the ontology of the living room itself.[17] Notice that it is possible to like the way the living room looks but not like the properties that the living room is visually represented as having. For instance, a certain arrangement of furniture might be aesthetically pleasing but impractical. In such a case, one might like the way the room looks but dislike the way the room looks to be.

The distinction between two meanings of “the way things look” blocks the inference from phenomenal character’s being a matter of the way things look to its being a matter of what properties things are visually represented as having. That inference trades on the ambiguity of the phrase “the way things look”. The identification of visual phenomenal character with the way things look is plausible on the reading of “the way things look” which concerns the manner of visual presentation rather than on the reading which is elliptical for “the way things look to be”. But the inference from that intuitive claim to the thesis that phenomenal character is exhaustively characterized in terms of what properties objects are visually represented as having is plausible only on the other reading of the phrase.

The intuitive claim that visual experience is a matter of “the way things look” thus fails to support Russellianism representationalism. The reading of “the way things look” on which it is correct to say that visual experience is solely a matter of the way things look is one on which the phrase denotes a visual mode of presentation of a scene. This thesis, plus the claim that neither Jack nor Jill are misperceivers, does not entail that Jack and Jill represent ripe tomatoes as having different properties. Instead, it suggests that Jack and Jill might represent ripe tomatoes as having the very same property—redness—but under different modes of presentation.

One might worry that on this way of understanding the difference between Jack and Jill, the difference is no longer a difference in content. Rather, it might be a difference in “mental paint”—a kind of non-semantic mode of presentation. In what follows I will offer a genuinely semantic way of understanding these modes of presentation.

 

Fregean Representationalism

 

Jack and Jill (who are spectrum inverted with respect to each other) can both have veridical experiences when looking at a ripe tomato, despite their experiences being phenomenally distinct. This along with other considerations counts against Standard Russellianism. Shoemaker’s (1994, 2001) attempted solution, while preserving Russellian representationalism, is to find some set of properties other than the colors that Jack and Jill might veridically represent. The properties he settles on are subjective properties—properties objects have only in relation to perceivers. I argue elsewhere (Thompson 2003a) that this solution fails, in part because of its difficulty in accounting for the possibility of misrepresentation. But more simply, Shoemaker’s solution intuitively gets phenomenal content wrong. Color experiences represent objective colors, properties which both persist when they are not perceived and would have existed even if there never had been perceivers. An alternative kind of theory of phenomenal content, Fregean representationalism, promises to address the problems of both of the other views. According to Fregean representationalism, the difference between Jack and Jill is not in what their experiences represent, but in the way it is represented. Jack and Jill both represent the tomato as being red, but under different modes of presentation.

Frege introduced the idea that difference of meaning can be determined by difference in cognitive significance. A familiar example is the case of Hesperus and Phosphorus. A person can believe that Hesperus is a star while nonetheless rationally believe that Phosphorus is not a star. And though “Hesperus is Hesperus” is not informative or cognitively significant, “Hesperus is Phosphorus” is cognitively significant. If the semantic values of both “Hesperus” and “Phosphorus” are simply their referents—Venus—then it is difficult to see how this could be the case. “Hesperus is Hesperus” and “Hesperus is Phosphorus” each express the very same Russellian propositions. Frege’s notion of sense is intended to capture this element of meaning that is in addition to reference.

The cognitive significance test applies equally, though perhaps less obviously, to phenomenal content. Here's one case of difference in cognitive significance for phenomenal content. Imagine that one's ability to match perceived locations across the sense modalities (such as sight and hearing) becomes impaired. Now suppose that one sees something at a certain spatial location, and that simultaneously one hears something at that location. The content of these two experiences is "x at L₁" and "y at L₂" respectively. "L₁" and "L₂" refer to the same location. According to Standard Russellianism, this location in space is all there is to the content of L₁ and L₂. But whereas “L₁ is L₁” is not cognitively significant, “L₁ is L₂” is cognitively significant. This is illustrated most clearly by the case of a person with our imagined impairment—this person does not know, without doing a lot of work, that L₁ is L₂.

The standard Russellian might claim that sight and hearing do not have identical spatial contents. In particular, vision provides finer-grained spatial contents than does hearing. But it seems that we can easily imagine that the subject in question has poorer than normal vision or unusually good hearing in such a way that the subject can indeed see something at L₁ and hear something at L₂ where L₁ and L₂ are the same in reference, rather than merely overlapping.

            Fregean representationalists can describe such cases by claiming that the same physical spatial location is represented under two different modes of presentation, L₁ and L₂. Modes of presentation can be thought of as placing conditions on reference. It might be that L₁ picks out the property that typically causes visual experiences with a certain spatial character in the subject. L₂ picks out the property that typically causes auditory experiences with a certain spatial character. Though L₁ refers to the same property as L₂ in the actual world, there are other possible worlds in which they do not co-refer.[18] This explains the cognitive significance of “L₁ is L₂”; that proposition provides information about what world one is in that goes beyond what can be known on the basis of the visual and auditory experiences alone.

 

Modes of Presentation as Intensions

 

            A first inclination might be to think of these phenomenal modes of presentation as Fregean senses. But if we restrict ourselves to a strict analogy with Frege’s philosophy of language, it is not clear that we can adequately accommodate the lessons from the inverted spectrum and color constancy cases. On a typical way of understanding Fregean senses, sense by itself determines reference. This entails that two representations with the same sense must have the same reference. But inverted spectrum cases and color constancy cases present circumstances in which the same mode of presentation can determine a different reference. For example, the mode of presentation under which Jack visually represents red things is the same as the mode of presentation under which Jill represents green things.

            The two-dimensional framework provides a different way of thinking about the relationship between sense and reference that is applicable to phenomenal content.[19] Modes of presentation can be thought of as corresponding to primary intensions, functions from centered worlds considered as actual to properties. A centered world is a possible way the world could be, with a marked center indicating an individual and a time. The same function, given a different centered world, can return a different property.

            For instance, the primary intension of my concept “water” picks out H2O in the actual world. The same intension might pick out XYZ on Twin Earth.[20] This reflects the fact that if it turns out that the actual world is Twin Earth, and the oceans and lakes are filled with XYZ rather than H2O, then “water” refers to XYZ. The primary intension thus captures the way in which the extension of a concept depends on facts about the world. Two individuals might share a concept that has the same primary intension, but given that they are situated in different worlds, the extensions of their concepts might differ. Oscar’s water concept, since he is on Earth, will have as its extension H2O. Twin-Oscar’s water concept, on Twin Earth, will have as its extension XYZ. Primary intensions thus constitute a kind of narrow content, content that is shared by microphysical or (narrow) functional duplicates.[21]

            The inverted spectrum scenario considered earlier is a form of Twin Earth case, and places a similar constraint on the correct theory of phenomenal content. When Jack and Jill have phenomenally identical color experiences, there is an intentional feature of their experiences that they have in common in virtue of their shared phenomenology. But given that their respective bluish experiences can both be veridical despite being caused by objects with entirely different physical colors, this shared content cannot consist in a common physical color property that is attributed by both experiences. This is much like Oscar and Twin-Oscar, who intuitively are in cognitive states with a form of shared content when they are both thinking “water is wet”. This content cannot consist in the extension of their concepts, since Oscar’s water concept is about H2O whereas Twin-Oscar’s is about XYZ.

            As in the case of conceptual contents, we can identify the common intentional feature between Jack and Jill with a shared primary intension. A phenomenally blue experience picks out, roughly, the property in the environment that typically causes blue experiences in the subject. This same primary intension in Jack picks out one set of physical properties, and in Jill it picks out a different set of physical properties. It is this feature that allows for Jack and Jill to represent the same physical color property under different modes of presentation, and to represent different physical color properties under the same mode of presentation. “Reddish” experiences, given the actual world centered on Jack, return physical redness as their extension. A phenomenally identical experience, given the actual world centered on Jill, returns physical greenness.[22]

By reflecting on various ways the world might be outside our experience, and asking ourselves whether those ways the world might be are ways which our experience “rules out”, we can get at the semantic contours of an experience’s phenomenal content.[23] Even after having a red experience, say of a ripe tomato, there are myriad ways the world (and the tomato) might be which are epistemically open to me. This is so even if I assume that my visual contents are fully veridical, that the world is as it seems to be.

For example, it might be that the tomato has a physically irreducible quality—redness—which is intrinsically just like the perceptual quality I am acquainted with upon looking at the tomato. To elaborate on this scenario, perhaps something like Aristotle’s theory of perception was correct. My experience of redness is caused by my being receptive to an intrinsic quality of the tomato. And the property referred to by my red experiences is this physically irreducible intrinsic quality of redness.

I don’t find this story about color and color perception to be remotely plausible. But is that because it conflicts directly with how the world is presented as being by my red experiences? Surely not. The intentional content of my visual experience is neutral between this scenario and the more familiar causal story involving surface-reflectance properties and light waves. This is to say that the Aristotelian scenario is a way the world might be for all that the phenomenal content of my visual experience tells me. Similarly, it might be that the textbooks are wrong about which properties of external objects correspond to the perception of red. As a result, the particular surface-reflectance property of physical objects that we think is the property which our red experiences refer to, R₁, is not in fact the ones which ground the dispositions of things to appear red. Instead, some different property R₂ of objects grounds that disposition. These are both ways that the world might be which are epistemically open to me, independent of any additional beliefs or experiences. That red things are R2 rather than R1 is a way that the world might be which cannot be ruled out a priori on the basis of the phenomenal content of my red experiences. As discussed below, I think the endpoint of this exercise is the conclusion that color qualia have response-dependent semantic intensions, picking out the property or properties which typically cause experiences of that phenomenal type.

 

The Particular Nature of Phenomenal Content

 

            We’ve seen the promise of a Fregean theory of phenomenal content, given that it introduces a second dimension of content beyond reference. But there are of course a wide range of possible Fregean views. In what follows I will propose a particular theory about the nature of the phenomenal modes of presentation involved in color experience. Phenomenal color content is response-dependent, indexical, and holistically determined.

 

Response-dependence (picks out property that typically causes the experience)

 

            Notions of “response-dependence” or “response-dispositionality” have been employed both at the level of properties and of concepts (Johnston 1989, 1992, 1993; Petitt 1991; Wright 1992). Here I will use the term “response-dependent” as describing a semantic feature of concepts,experiences, or any other content-bearing entity. In particular, a concept C is a response-dependent concept if it is a priori that, for some non-trivial specification of r, s, and k, x is C if and only if x is disposed to produce a response of type r in subjects s under conditions k.[24] For instance, color concepts might plausibly be understood as response-dependent concepts as follows: x is red if and only if x is disposed to produce a phenomenally red sensation in ordinary human observers under standard lighting conditions. Other concepts that have been suggested as being response-dependent—though not always using that term—include the concepts of intrinsic value (Lewis 1989), moral goodness (McDowell 1985, Thompson manuscript), and even all concepts (Pettit 1990).

            The notion of response-dependence can be generalized to apply to phenomenal content as discussed here. Response-dependence can be understood as a feature of the primary intension of a concept or experience, articulating the nature of the condition on reference. The apriority of the bi-conditional follows straightforwardly from the connection between primary intensions and apriority. We can say that a primary intension is response-dependent if it is a function that is properly characterized as picking out at a world only the properties or objects that are disposed to produce a response r in subjects s under conditions k.[25]

            The evidence that guides us in determining the nature of the conditions of satisfaction of color experience is our judgments about the veridicality of an experience under various scenarios considered as actual. The arguments that lead us to reject Standard Russellianism show that there is no particular physical property that an object must have in order for a red experience, for example, to be veridical. Is there nonetheless some condition that an object must satisfy in order for a red experience to veridically represent that object? Though there is no particular physical property that the object must have, it seems that it minimally must have some property or other that typically causes experiences of that type under those lighting conditions. Color experiences can thus be seen as having a particular kind of response-dependent primary intension, one in which the vehicle of content is an instance of the relevant response type. Further, a color experience picks out not just any property that is disposed to cause an experience of that type, but only those properties that have typically in the past caused such experiences.

This can be seen easily by considering a brain-in-a-vat scenario. Suppose that I have lived the majority of my life as a fully-embodied human being whose visual experiences are caused in the ordinary manner. But at some point in adulthood, an evil scientist removes my brain from my body while I am sleeping and places it in a vat of chemicals. She keeps my brain alive and electrochemically stimulates it in such a way that I have visual experiences that are phenomenally just like the kinds of experiences that I normally have, and like those that I would have had if I had not been newly envatted. I have an experience as if I am looking at tangerines in a supermarket. The orange phenomenal properties of my visual experience are not caused by those properties of objects that normally cause such experiences in me; instead, all of my visual experiences are due to direct brain stimulation. It seems clear that in this scenario, I am subject to systematic misperception with regard to the colors and spatial features of the world around me. And it seems as though the reason for this is that my orange experiences during envatment are not caused by those properties that have normally in the past caused orange experiences in me.[26]

Ned Block’s “Inverted Earth” thought experiment provides further intuitive data for the idea that part of the conditions of satisfaction of a color experience is that it be caused by properties that typically cause such experiences (Block 1990). In the thought experiment, a normal perceiver on Earth is transported to another planet which is like Earth except that the colors of things are inverted relative to the Earth. The grass is physically red, the sun is blue, etc. But the individual Fred has color-inverting lenses placed in his eyes, such that his color experiences are indistinguishable from the ones that he would have had if he had remained on Earth. That is, even though the grass is red on Inverted Earth, it still phenomenally appears green just as it did on Earth.

Intuitively, when Fred first opens his eyes on Inverted Earth, he is a misperceiver of color. After all, he represents the grass in just the same way as he did a day before—as being green. The grass is in fact red, and so this counts as a misperception. But it also seems plausible that after some amount of time—and saying precisely how long is difficult—Fred’s color experiences are no longer mistaken. There comes a point at which we are inclined to say that Fred represents objects as being physically red by having phenomenally green experiences.

            This intuition is explicable on the present view about the primary intensions of color experiences. Before leaving Earth, phenomenally green experiences were typically caused by physically green things. Immediately upon arriving on Inverted Earth, this remains the case. When Fred has a green experience in response to the red grass, his experience is not caused by a property that typically causes green experiences. The initial experience on Inverted Earth thus counts as a misperception. But after some amount of time on Inverted Earth, green experiences are typically caused in Fred by red things. And so after some amount of time, the present view entails that Fred becomes a veridical perceiver of color on Inverted Earth.[27]

            A difficult question concerns how much time must pass before Fred becomes a veridical perceiver. And in general, it is difficult to give a precise analysis of what counts as a “typical” cause. I will not offer a precise analysis here. Instead, two things deserve noting. First, the natures of the primary intensions involved in color experiences are characterized or approximated, but not “analyzed”, by the descriptions being offered here. It may be that there are no precise necessary and sufficient conditions for the satisfaction of a color experience’s content that can be articulated with an English language description. On the present view, there need not be an analysis in the form of necessary and sufficient conditions for when something satisfies the phenomenal content of a red experience. But what is being claimed (and what is usually tacitly assumed among those trying to provide such an analysis in this and other domains) is that we as subjects who have experiences with a certain phenomenal content are in a position to judge about a given epistemic possibility whether that possibility would satisfy the experience’s phenomenal content if it were actual.[28]

Second, the vagueness of “typicality” might be seen as a virtue in this context. For our intuitions about various scenarios are fuzzy in roughly the same places as are our intuitions about what counts as typical. The Inverted Earth thought experiment discussed above provides one example. Insofar as it is difficult to say what the typical cause of Fred’s red experiences is, we have difficulty deciding whether his experience is veridical. After a long period of time, after which it becomes clear that Fred’s experience has as its typical cause the statistically normal cause on Inverted Earth, we find it easer to decide that Fred’s red experiences are veridical. If there are scenarios of which we have great difficulty deciding whether an experience would be veridical, we should not expect a characterization of the experience’s phenomenal content to deliver a clear verdict about those cases. This would, I think, put the cart before the horse.

            One worry that might be raised about the requirement of a typical cause in the primary intension of color experience comes from considering a creature’s very first visual experience. The notion of a typical cause is a backwards-looking, statistical notion. A first experience thus lacks a typical cause. But it might seem that such an experience could be a veridical one. And at a minimum, it surely is the case that such experiences have phenomenal content. This in fact seems to be required by the notion of phenomenal content, since a first visual experience could plausibly be phenomenally just like a veteran perceiver’s experience (at least with respect to color).

But we have already encountered reasons for thinking that a first color experience might lack a truth-value. The inverted spectrum thought experiment showed that two phenomenally identical experiences could have different truth values, even if they are had in response to the very same circumstances. For suppose that Jack has a red experience while looking at a cherry. If Jill were to have an experience just like Jack’s in response to the very same object under the very same conditions, her experience would be mistaken with respect to its color content. A red experience in Jill would falsely represent the cherry as being green.

Now consider Jane who has a red experience for the first time, perhaps while viewing the same cherry. On what grounds should we say that Jane’s experience is made true by the cherry, like Jack’s, rather than made false by the cherry, like Jill’s? There is nothing about the experience, nor about Jane’s history, that could choose between these possible assignments of truth-value. We might appeal to the causes of Jane’s future experiences, but it isn’t plausible that the conditions of satisfaction of color experiences are forward-looking in this way. It might be that Jane will in the immediate future have inverting lenses placed in her eyes, or become a brain-in-a-vat. But this fact about the future surely cannot have a bearing on whether Jane’s present or past experiences are veridical. I might now be a veridical perceiver of color, despite the fact that from tomorrow and the next fifty years, I might be a brain-in-a-vat whose color experiences are typically caused by entirely different properties.

I suggest that first color experiences lack truth-values but do not lack phenomenal content. They have conditions of satisfaction that are just like those of all other experiences that are phenomenally identical to them. But those conditions of satisfaction involve a presupposition that is not satisfied—that there is a typical cause of experiences of that phenomenal type. This makes them infelicitous, but not false. And phenomenally different experiences will have different phenomenal content, even without truth values. An area of space that looks red will look to be different from an area of space that looks blue. But this can be made sense of without endorsing the idea that those experiences have truth-values.

            It might seem that the causal condition is too strong. After all, there seems to be the possibility of “veridical hallucination” (Lewis 1980). You might, for example, have a veridical perception as of there being a green lamp three feet in front of you. But that experience might not have been caused by the green lamp, or indeed by anything that is green or three feet away. The experience might instead have been caused by direct brain stimulation by a neurosurgeon. Such an experience would plausibly fail to be a genuine perception of the lamp. Cases of veridical hallucination have often been used to illustrate the fact that to see that x is F, one must not only visually represent that x is F, but one’s experience must also causally depend on the fact that x is F (Grice 1961, Lewis 1980, Strawson 1974, Noe 2003). But it might seem that though causal dependence is a necessary requirement for veridical perception, it is not required in order for an experience to have veridical phenomenal content.

            But cases of veridical hallucination are not precluded by the claim that phenomenal color content involves a causal condition on reference in the way suggested. Suppose that I am an ordinary perceiver in a world like we take this one to be. My green experiences have as their usual causes the kinds of surface physical properties that contemporary science suggests. On a particular occasion, I have a hallucination induced by a neurosurgeon. I hallucinate that there is a green lamp three feet away from me. There is in fact a green lamp three feet away from me. On the present proposal, the greenness of my experience attributes the property that typically causes green experiences in me to the region of space three feet away from me. By hypothesis there is an object with this property in that region of space. And so on the present proposal this is indeed a case of veridical hallucination—and not a case of false phenomenal content. My experience accurately represents certain color properties as being instantiated in a particular region of space.

            As the above illustrates, the causal condition involved in phenomenal color content is not a condition that requires that an instance of the property attributed by a green experience is a cause of the green experience on that occasion. All that is required for veridicality is that the object represented has the property that typically causes green experiences in me. And so all that this condition requires is that there is a typical cause of green experiences, not that the green experience is caused by its typical cause on that occasion.

            Perhaps the causal condition might seem too strong at even this point. For isn’t it possible that green experiences never be caused, or never be caused by instances of the property that they attribute? Occasionalists such as Malebranche believed that God is the only cause. According to this idea, it might be that all green experiences are caused by God. But my green experiences need not thereby represent God or some aspect of God. Instead, God causes my green experiences when I am looking at green objects, and my red experiences when I am looking at red objects. My color experiences can plausibly be understood to represent the properties of objects that, when I am suitably related to them, inspire God to cause a corresponding color experience in me.

            Even if one is not attracted to occasionalism, the general point remains. Is it plausible that phenomenal color experiences involve, as part of their conditions on satisfaction, a notion arcane as causation? Isn’t it conceivable that color experiences like our own could be veridical even in a world in which there is no causal relation between them and the properties they represent? Perhaps there are other “matching” relations between phenomenal color properties and colors that would be sufficient.

            No stand is being taken here on the metaphysics of causation. Depending on how that issue is resolved, it may be that the causal condition would need re-examining. But what experience minimally seems to require is counterfactual dependence. Notice that counterfactual dependence is maintained under the occasionalist view. For we can suppose, at least as safely as seems possible on such theological matters, that worlds in which the colors of object or my position relative to them are different from the actual world are much closer possible worlds than worlds in which God has different habits of eliciting color experiences in me. By analogy, worlds in which the distribution of matter is different are nearer than worlds in which the laws of nature differ.

In general, I find it suitable to describe the conditions of satisfaction involved in color experience as “causal” because of the phenomenology of perceptual experience. Visual experiences are, phenomenologically, modes of “receptivity” (Kant 1781). Visual experiences are experienced as being due to something outside oneself, as ways of being affected by external things. This is in sharp contrast to the way thoughts and actions are experienced (what Kant called “spontaneity”). It is in this intuitive and non-technical sense that visual experiences purport to represent features of the external world that cause them. It should be admitted that science or metaphysics might arrive at a theory of causation, and that what causation turns out to be is not demanded by the phenomenal content of experience.

 

The indexicality of phenomenal color content

 

To allow for the possibility of spectrum inversion without illusion, it must be the case that Jack and Jill can both have veridical phenomenally green experiences. It is definitional of the notion of phenomenal content that phenomenally identical experiences have the same phenomenal content. Thus Jack and Jill's green experiences must have the same phenomenal content. And so Russellianism requires that there be some property that Jack and Jill can veridically represent objects as having by their green experiences. Shoemaker (2000) seeks, as a candidate for what is represented by both Jack and Jill, a dispositional property of causing green experiences under certain circumstances in some sort or other of observer. But as I argue elsewhere (Thompson 2003a), there is no satisfactory way to specify the relevant observers. If Jack represents the object as causing a green experience in observers like him, and unlike Jill, then his experience does not share content with Jill's experience. If Jack represents the object as causing a green experience in observers like him and in observers like Jill, his experience is a misperception (since the object is causing a red rather than a green experience in Jill). If Jack represents the object as causing a green experience in observers like him or in observers like Jill, his experience will be veridical. But then it becomes difficult for Jack to ever misrepresent the colors of things. For example, if he had a green experience while looking at a red thing, his experience would represent the red object as having the disposition to cause a green experience in observers like him or in observers like Jill. Red things do in fact cause green experiences in observers like Jill. Jack's red experience, which is clearly a case of misperception, would on this account be counted as a veridical perception.

The Fregean approach to phenomenal content has available to it a solution to this kind of problem. The dispositional "element" of phenomenal color content is, on this view, a feature of the modes of presentation involved in phenomenal content. The Fregean does not need to find a dispositional property that is shared by the objects that cause green experiences in Jack and the objects that cause green experiences in Jill. Instead, there need only be some condition on reference that is shared by Jack and Jill. And that condition on reference can involve indexicality. An utterance of the English word "I" refers to the speaker who tokens it. All instances of "I" thus share a condition on reference, the condition of being the person who uttered it. Uses of the word "I" need not share a reference. Uttered by Jack, "I" refers to Jack. Uttered by Jill, "I" refers to Jill. Likewise, it is available to the Fregean about phenomenal color content to hold that a green experience picks out the property that typically causes green experiences in the subject. Had by Jack, a green experience would pick out the property that typically causes experiences like that in Jack. Had by Jill, a green experience would pick out the property that typically causes experiences like that in Jill. Jack and Jill's green experience still share phenomenal content, understood as Fregean modes of presentation. But their experiences refer to different properties in their respective environments.

Frege's notions of sense and reference had difficulty accounting for indexicality.[29] But many of the insights of Frege's semantic theory can be captured in a way that allows for indexicality. One problem for Frege's theory is that sense by itself determines reference, and so there does not seem to be any room for shared senses with different references. The two-dimensional framework handles indexicality via the notion of a centered possible world. A centered possible world is a possible world from a point of view, with a space-time point (or an individual at a time) marked as the “center”.[30] The primary intension of a concept is a function from centered possible worlds to an extension. For instance, the primary intension of “I” can be understood as picking out the individual at the center of a given possible world. Centered on George Bush, “I” returns George Bush as its referent. The same primary intension, centered on Sandra Day O’Connor, picks out Sandra Day O’Connor.

The phenomenal content of a red experience can now be understood as a function from a centered world that picks out the property that typically causes red experiences in the individual at that center. Centered on Jack, this content will pick out one property, such as physical redness. Centered on Jill, the same intension picks out a different property, physical greenness.

The indexicality of phenomenal color content, understood in this way, also explains our intuitions about the veridicality of our color experiences when considering various epistemic possibilities about how the world is outside our experience. Earlier I noted that if a roughly Aristotelian theory of color were the right one, this would not lead us to conclude that our color experiences are mistaken. And there are various other possibilities about the physical nature of color that seem to be compatible with the phenomenal content of color experience. This can be illuminated by thinking about color content indexically as a function from centered worlds considered as actual to properties. If I am in an Aristotelian world, my red experiences represent the form of redness. That is to say that my experience, centered on the Aristotelian world, has the form of redness as its extension. If I am in a world in which my red experiences are typically caused by some particular physical property, then my red experiences has that physical property in its extension. In each case, what all red experiences have in common with regard to content is the same indexical mode of presentation.

It might seem that Shoemaker could have made a similar move in defending the idea that color experiences have Russellian content that represents appearance properties. But Russellianism requires that there be some property in common that is represented by Jack and Jill’s red experiences. And it is not clear that there could be such a thing as an indexical property. What would this indexical property be, such that both Jack and Jill veridically represent it when they have red experiences?

 

The Holism of Phenomenal Color Content

 

The problem posed by color constancy for Russellian theories of phenomenal content is that color constancy demonstrates that a single phenomenal color property can represent different physical colors under different conditions within the same individual. According to Russellianism, for any phenomenal color property there is a single property attributed by an experience with that property. In the abstract, Fregean theories of phenomenal content can accommodate color constancy because they do not entail that there is some single property that all phenomenally identical color experiences must represent. What a Fregean theory does require is that there be some condition on reference that all phenomenally identical color experiences share. The phenomenon of color constancy thus places a constraint on the particular nature of the modes of presentation involved in the phenomenal content of color experience.

One feature that seemed to be relevant to the determination of color content involving color constancy was lighting conditions. For instance, consider two phenomenally identical color experiences experienced under two different lighting conditions. One is the experience had while viewing a white sheet of paper in the yellowish sunlight of dawn. The other is the experience had while viewing a slightly yellow sheet of paper under fluorescent lighting. The white sheet of paper will look to be white, despite presenting a slightly yellowish appearance. The yellow sheet of paper will not look to be white, despite phenomenally appearing just like the white sheet of paper viewed at dawn.[31] A significant difference between the two appearances is the lighting conditions under which the visual experiences are elicited. And so it might be that color constancy can be accommodated within a Fregean theory of phenomenal content by including lighting conditions as part of the conditions of satisfaction of color experiences. For instance, a red₂₇ experience had under lighting conditions p represents the property that typically causes red₂₇ experiences under conditions p. A red₂₇ experience had under lighting conditions q represents the property that typically causes red₂₇ experiences under conditions q. More generally, the experience represents the property that typically causes red₂₇ experiences under these conditions, whatever those conditions might be.

A problem arises with this suggestion. Suppose that two objects are viewed under different lighting conditions, but where the differences in lighting conditions was undetectable to the subject. For instance, two objects might be viewed on a stage. The lighting throughout the room and on the stage is ordinary incandescent lighting. One of the objects is red. The other object is white, but it is illuminated via a spotlight with a red light. There is no dust in the air that would reflect the red light. And the light is projected in such a way that it does not reflect off of anything other than one of the objects, not even the floor of the stage surrounding the object. Other adjustments might need to be made so that there is no detectable shadow from the red light source, which is hidden from view.

In such a circumstance, it seems that an ordinary observer would have a visual experience that misrepresents the color of the object that is cast with red light. The phenomenal content of the visual experience had while viewing the scene will be such that both the white and the red object appear to be red. But on the suggestion above for accommodating color constancy, the extensions of phenomenal color content vary by lighting conditions. The two objects are viewed under different lighting conditions, and so the respective experiences would attribute different properties to the objects. A red experience had under lighting conditions in which red light is projected onto a white object is typically caused by a white object. And so it seems that the present suggestion would entail that an observer of the unusual scene would veridically represent the white object as being white and the red object as being red.

The view might be revised in such a way that “these conditions” picks out the same condition for both objects in the example. But this would make many other, and more ordinary, visual experiences misperceptions. We almost always view the world under uneven illumination conditions. Part of the visual scene might be in shadow, or there might be multiple light sources that emit different qualities of light. In order for these experiences to be veridical on the present suggestion, it would be necessary for the demonstrative “these conditions” to pick out different lighting conditions for different portions of the field of view.

One solution might be to make phenomenal color content depend on what lighting conditions a subject represents as being instantiated in a scene. After all, the scenario is an unusual one that is specifically designed to make it appear to the perceiver as though there is uniform standard lighting on both objects. So perhaps the phenomenal content of color experience picks out properties relative to the lighting conditions that the subject represents as obtaining. But this solution is unattractive for two reasons. First, it is questionable whether subjects do in fact represent lighting conditions, or at least to a degree of specificity that would be required. Certainly subjects do not consciously represent lighting conditions. We are consciously aware of differences in lighting over time, but this is most plausibly itself due to differences in the phenomenal character of our color experiences. Remember that the kinds of variations in illumination conditions that would need to be represented do not only include changes in brightness (which we do seem to be aware of, as when a bright light reflexively causes one to squint). Color constancy also concerns changes in the quality of light, such as the ratios of various wavelengths of light that are present. We seem to be aware of the latter only derivatively in virtue of our being aware of the phenomenal character of our color experiences.[32] And even then, it does not seem that we usually consciously form representations of lighting conditions, and certainly not with much precision.

Perhaps we unconsciously represent lighting conditions with enough precision to accommodate color constancy. The property represented by a red experience on a particular occasion would depend partly on what lighting conditions the subject represents himself and the objects of perception to be in. Two phenomenally identical red experiences could attribute different properties to their objects if the subject represented lighting conditions as being different in the two situations. But it doesn't seem that a subject need represent lighting conditions in order to have determinate phenomenal color content. We can imagine a subject that does not represent lighting conditions in any way, but who has a visual experience with phenomenal character just like the one I am having now. Remember that on the present suggestion, these representations of lighting conditions are not elements within the phenomenal character of visual experience.

A further problem with making phenomenal color content dependent on lighting conditions, or on the representation of lighting conditions, is that the very existence of lighting conditions does not seem to be essential for veridical phenomenal color content. We can imagine a creature who has color experiences like ours, but for whom those experiences are typically caused by other physical phenomena that bear no interesting relationship to lighting conditions. Perhaps color experiences in it are caused by the radiation of heat from nearby objects. Perhaps they are caused by a sensory system like a bat's echolocation. It even seems that a creature could have veridical color experiences in a world with a different physics in which there is no light at all. And thus it does not seem plausible that the phenomenal modes of presentation involved in color experience invoke or are relative to illumination conditions in their conditions of satisfaction. For even in a creature for whom color experiences are caused by radically different sensory systems and environmental conditions, the way the world is presented to the subject as being is just as it is for creatures like the ones we take ourselves to be. This shared content is what we are trying to capture in the notion of phenomenal content.

But perhaps there is another way to accommodate color constancy within a Fregean theory of phenomenal content. Changes in phenomenal appearance that do not result in changes in represented color are often cases in which there is a wholesale change in the visual field. A cloud that passes to block the sun will cause the entire field of view to receive less light. The world around the subject will look different in phenomenal color appearance--but the objects in view will not look to change in their color. The previous suggestion was that phenomenal color properties represent physical colors relative to lighting conditions or to represented lighting conditions. This idea was inspired by the observation that the color represented by a particular phenomenal color property largely varies with changes in illumination.

But there is another possible explanation of this covariance that does not appeal to objective lighting conditions or depend on representations of lighting conditions. Up until now, it has been assumed that the phenomenal content of color experiences is atomistic. That is, it was assumed that phenomenal color properties represent the properties that they represent independently of other phenomenal properties (such as spatial phenomenal properties or other phenomenal color properties instantiated in the visual field). But there is another possibility. Perhaps what a particular instance of a phenomenal color property represents is determined by what other phenomenal color properties are instantiated in the visual field. Rather than invoking lighting conditions, phenomenal modes of presentation for color experiences might be sensitive to properties of the entire visual field in their determination of reference.

For any difference in illumination conditions that might be relevant to the phenomenal content of a phenomenal color property at a particular location in the visual field, there is a corresponding phenomenal difference elsewhere in the visual field. For example, consider two phenomenally identical red experiences caused by viewing objects that are in different illumination conditions (such as mid-day sunlight versus a red light). Insofar as it is correct to say that a white object under red lighting is represented as being white rather than red, there will be a difference in the phenomenal colors or spatial properties instantiated in surrounding regions of the visual field that distinguish the experience from one in which the object is represented as being red. For instance, one would likely veridically represent an object as white if it were cast in red light and the areas surrounding the object also presented a more reddish appearance than other areas (that are not cast in red light). By contrast, if the white object presented a reddish appearance but the surrounding area did not, the object would likely appear to be red. This was the case in the example discussed earlier, in which we imagined that all evidence of a red light was made unavailable to the perceiver. Part of what is required to create that type of deceptive scenario is that the “hidden” lighting conditions leave no phenomenal trace in other regions of the visual field. The red light, for example, should not reflect off any dust particles in the air between the source and the illuminant. The red light should not reflect off the region surrounding the illuminant, nor should it leave evidence in the form of a shadow. Any of these would result in changes in the phenomenal character of the total visual field of the perceiver.

The key observation here is that although color constancy shows that phenomenally identical color experiences within a single subject can differ in color represented, it does not seem to be the case that phenomenally identical total visual experiences within a single subject can differ in color represented. This means that for any two phenomenally identical color experiences that differ in color represented, there must be some additional phenomenal difference between the total experiences within which they are embedded. This allows for the possibility that what property is represented by a phenomenal color property depends on these other features of the phenomenal character of the total visual experience (including spatial features and what other phenomenal color properties are instantiated in the experience).

On this proposal, what property a color experience attributes to an object depends as a matter of fact on illumination conditions. But this fact about the actual world is derivative on the fact that what property a particular color experience attributes depends on how the phenomenal color property is experienced spatially and on what other phenomenal color properties are present in the visual field, and where they are experienced spatially.[33]

           

The Transparency of Experience

 

            I come now to what is likely to be the strongest source of disagreement with a Fregean theory of phenomenal content.[34] That is, that it purportedly conflicts with the phenomenology of experience. According to the transparency intuition, what we are aware of when we introspect on our color experiences are properties that we experience as belonging to the external objects of perception. We are not aware of modes of presentation of external colors. Or if we are, then visual experience is systematically delusive; the properties that we are aware of seem to be properties of external objects. The Fregean view at best leads to an error theory of color perception.

            These claims about phenomenology seem plausible enough at first glance. But I think this plausibility stems from several confusions. First, there is the matter of what we are “aware” of in perception, and in introspection on perceptual experience. The Fregean theory of phenomenal content need not deny, and I would certainly accept, that in having a color experience we are made aware of external colors. And under the sense of “awareness” that makes this true, we are not aware of any sensory intermediaries. We have visual experiences with a certain phenomenal character. These experiences have Fregean phenomenal content. By having these experiences, we are made aware of external properties of external objects. Our awareness of these properties is not mediated by awareness of purely mental properties of the experience.

            But there is perhaps another sense of “awareness” according to which what we are most directly aware of is a property of the experience, rather than a property of the external object of the experience. Certain qualities are phenomenologically manifest to us in color experience, and the Fregean theory that I have defended here claims that these properties are not properties of the external objects of perception. And so perhaps here the transparency intuition is meant to gain a foothold against the present view. For don’t these phenomenologically salient properties seem to be “out there” on the surfaces of objects, rather than “in the mind”?

            It does appear to be a common belief that this is the case. But I am not convinced that this is a phenomenological datum. The genuine phenomenological datum that this intuition appears to focus on is the fact that the color features of experience are spatially articulated. They occupy volumes of space. We can experience redness to the left of greenness, or a yellow thing that is further away from us than a purple thing. The phenomenology of visual experience is far from exhausted by the experience of color. Visual experience has spatial phenomenology as well. And like color, the spatial features of experience seem to be intentional. Differences in spatial phenomenology correspond to differences in how the world appears to the subject.

            If we could assume that a Russellian theory of phenomenal content was correct for these spatial features of experience, then perhaps we could also correctly draw the conclusion that the color features of experience are experienced as being located in external physical space. Color phenomenology and spatial phenomenology are intertwined in such a way that, if the spatial features of experience are genuine external spatial locations, it would certainly be true to say that the color features of experience are experienced as being “out there” in external space. But unless this Russellianism about spatial phenomenal content was independently supported by phenomenology alone, this conclusion would be in large part theoretical. Its denial would not conflict with phenomenology.

            And we cannot, in fact, simply assume that a Russellian theory of spatial phenomenal content is true. Given the intimate relationship between color and spatial phenomenology and content, this would beg the question against the Fregean theory. Elsewhere I directly argue against Russellianism and for Fregeanism about spatial phenomenal content as well (Thompson 2003c).

 

Conclusion

 

Earlier it was noted that one motivation for Russellian representationalism was the idea that experience is transparent. But the only sense in which experience is rightly said to be transparent is one which is not only compatible with Fregean representationalism but actually supports it. The phenomenal character of a sensory experience is a matter of the way things appear, understood as the mode under which the world is represented. This is a fully representational notion. In agreement with Russellian representationalists who have espoused the transparency thesis, sensory experience is not merely a matter of non-intentional “raw feels”. But it also does not consist simply in how the world outside the experience is represented as being. Rather, it concerns the way in which the world is represented.

I have argued that the nature of these modes of presentation is usefully modeled within the two-dimensional framework as a kind of primary intension, a function from centered worlds or epistemic possibilities to extensions. In particular, color experiences have indexical response-dependent Fregean content, picking out the properties that typically cause such experiences in the subject. But the phenomenon of color constancy shows that phenomenal color content is not atomistic, but holistically determined by other features of the visual field, including its spatial content and facts about what other phenomenal color properties are instantiated and where.

 

           

References

 

Block, N. 1990. Inverted earth. Philosophical Perspectives, 4.

 

Byrne, A. 2001. Do colours look like dispositions? Reply to Langsam and others. The Philosophical Quarterly 51.

Byrne, A. and Hilbert, D. 1997. Colors and reflectances. In A. Byrne and D. Hilbert, eds., Readings on Color, vol. 1. Mit Press.

Chalmers, D. 1996. The Conscious Mind. Oxford University Press.

Chalmers, D. 2002a. The components of content. In D. Chalmers (Ed.), Philosophy of Mind: Classical and Contemporary Readings. Oxford University Press.

Chalmers, D. 2002b. On sense and intension. In J. Tomberlin (Ed.), Philosophical Perspectives 16: Language and Mind. Blackwell.

Chalmers, D. 2004. The foundations of two-dimensional semantics. In M. Garcia-Caprintero and J. Macia (Eds.), Two-Dimensional Semantics: Foundations and Applications. Oxford University Press.

Clark, A. 2000. A Theory of Sentience. Oxford University Press.

Cohen, J. unpublished manuscript. Color constancy as counterfactual.

Dretske, F. 1995. Naturalizing the Mind. MIT Press.

Frege, G. 1892. On sense and reference. Trans. M. Black. In Translations from the Philosophical Writings of Gottlob Frege. P. Geach and M. Black (Eds.), Basil Blackwell.

Harman, G. 1990. The intrinsic quality of experience. Philosophical Perspectives 4:31-52.

Horgan, T. and Tienson, J. 2002. The intentionality of phenomenology and the phenomenology of intentionality. In D. Chalmers, ed., Philosophy of Mind: Classical and Contemporary Readings. Oxford University Press.

Johnston, M. manuscript. The Manifest.

Land, E. 1977. The retinex theory of color vision. Scientific American 237, No. 6:108-128.

Langsam, H. 2000. Why colors do look like dispositions. The Philosophical Quarterly 50:68-75.

Lewis, D. 1980. Veridical hallucination and prosthetic vision. Australasian Journal of Philosophy 58: 239-49.

Lewis, D. 1989. Dispositional theories of value. Proceedings of the Aristotelian Society Suppl. Vol. 63:113-37.

Lycan, W. 1996a. Consciousness and Experience. MIT Press.

Lycan, W. 1996b. Layered perceptual representation. Philosophical Issues 7.

Malebranche, N. [1674]. 1992. Philosophical Selections: From the Search After Truth. S. Nadler, ed. Hackett.

Martin, M. 2002. The transparency of experience. Mind and Language 4.

Moore, G. E. 1903. The refutation of idealism. Mind 12.

Noe, A. 2003. Causation and perception: The puzzle unravelled. Analysis 63.

Perry, J. 1977. Frege on demonstratives. Philosophical Review 86:474-97.

Petitt, P. 1991. Realism and response-dependence. Mind.

Putnam, H. 1974. The meaning of “meaning”. In Mind, Language, and Reality. Cambridge University Press.

Shoemaker, S. 1996. Intrasubjective/intersubjective. In The First-Person Perspective and Other Essays. Cambridge University Press.

Shoemaker, S. 1994. Self-knowledge and “inner-sense”. Philosophy and Phenomenological Research LIV:249-314.

Shoemaker, S. 2001. Introspection and phenomenal character. Philosophical Topics.

Shoemaker, S. 2002. Whitehead lectures. Manuscript.

Shoemaker, S. forthcoming. On the way things appear.

Siewart, C. 1998. The Significance of Consciousness. Princeton University Press.

Stalnaker, R. 1999.  Comparing qualia across persons. Philosophical Topics 26:385-404.

Stoljar, D. 2002. The argument from diaphanousness. Manuscript. http://humanities.ucsc.edu/NEH/stoljar2.htm

Strawson, P. 1974. Causation in perception. In Freedom and Resentment and Other Essays. Methuen.

Thompson, B. 2002. Railton on rigidified response-dependence. Manuscript. http://faculty.smu.edu/bthompso/papers.html .

Thompson, B. 2003a. Shoemaker on phenomenal content. Manuscript. http://faculty.smu.edu/bthompso/papers.html .

Thompson, B. 2003b. Color Constancy and Russellian Representationalism. Manuscript. http://faculty.smu.edu/bthompso/papers.html .

Thompson, B. 2003c. The spatial content of experience. Manuscript. http://faculty.smu.edu/bthompso/papers.html .

Thompson, B. 2004. Inverted spectra without illusion. Manuscript. http://faculty.smu.edu/bthompso/papers.html .

Tye, M. 1992. Visual qualia and visual content. In T. Crane, ed., The Contents of Experience. Cambridge University Press.

Tye, M. 1995. Ten Problems of Consciousness. MIT Press.

Tye, M. 2000. Consciousness, Color, and Content. MIT Press.

Walsh, V. and Kulikowski, J. 1998. Perceptual Constancy. Cambridge University Press.