9

FREGEAN, REALIZATION-THEORETIC THEORIES Russellian and identity-theoretic versions of physicalism fail to cope with the epistemology of color because they must portray visual experience as represent ing color without a characterization that denotes it necessarily. Such visual representations would denote properties only contingently, and would therefore fail to provide the appropriate introspective knowledge of the properties denoted. This problem does not affect Fregean, realization-theoretic theories. A realization-physicalist can concede, in response to the naive objection, that red objects aren’t visually characterized in microphysical terms, and yet hold a version of Fregeanism according to which they are characterized in terms that express what it is to be red; for he doesn’t believe that to be red is to have a microphysical property. His theory of visual representation may then enable him to account for the epistemology of color experience. For if visual experience represented red by means of a characterization that represented what it is to be red, then introspection on the content of such an experience would leave no doubt whether there was such a property, introspection on experiences containing the same characterization would leave no doubt whether they represented the same property, and introspection on experiences containing characterizations of various colors would reveal the relations of similarity among them-all because the introspectible content of each experience would reveal what it is to have the property therein represented. We therefore turn to a consideration of Fregean, realization-theoretic versions of physicalism. One such theory was already introduced, in our initial survey of physicalist theories. Before returning to that theory, however, we shall briefly introduce a new proposal, which is motivated by epistemological arguments of the sort considered above. This proposal has little intuitive appeal of its own; indeed, it would hardly have been intelligible before our epistemological arguments against the other proposals had been aired. As a response to those arguments, however, it has some apparent plausibility. A New Proposal The new proposal is an attempt to kill two birds with one stone.34 It purports to explain at a stroke how colors are visually represented and how their similarities and differences are known. The explanation is that colors are visually characterized precisely as those properties which bear the appropriate similarities and differences to one another. How could all of the similarities and differences among colors be included in their visual characterization? Here is how. Let a pigmentation be any property of extended things that stands with its codeterminates in relations of similarity and difference representable by a spheroid space in which distance around the circumference, distance from the ends, and distance from the interior correspond to differences in three different respects (to be called, for our purposes, hue, lightness, and saturation). Then let coordinates be defined so that any determinate pigmentation can be labelled by three numbers specifying its longitude, latitude, and depth in the property space. The description “pigmentation xvz” will then have as its condition of satisfaction the presence of a determinate whose relation to its co-determinates corresponds to position x,vz in a property space of this structure. Now suppose that visual experience characterized surfaces as having pigmentations, specified by their coordinates in pigmentation-space.35 Under the terms of Fregeanism, such experiences would represent the surfaces as having some appropriately related determinates of some appropriately structured determinable. Under the terms of realization-physicalism, colors would be the second-order properties expressed by such characterizations-that is, the properties of having appropriately related determinates of an appropriately structured determinable. We believe that this version of the proposal can account for all of the knowledge claimed in our epistemological intuitions. Reflection on the visual characterization of objects as having pigmentations xyz and qrs would yield the appropriate knowledge about the higher-order properties that the objects were thereby seen as having. That is, it would reveal that the experience represented its objects as having genuine, co-determinate properties, properties identical to those represented by internally similar experiences and differing from one another in degrees proportionate to x – q, y – r, and z – s. One would therefore know when one was seeing things as red or orange, and one would be able to tell their similarities and differences. Unfortunately, this remedy for earlier epistemological problems only creates new ones. Once all of the requisite information has been encoded into the proposed visual characterization of colors, the resulting proposal-in any version-credits the subject of that experience with too much knowledge rather than too little. For it implies that the characterization of any one color encompasses that color’s relations to all of the others, by locating it in a fully conceived color space. If color experience conformed to this proposal, the difference between red and orange would not only be evident from the experiences of seeing red and orange; it would be evident from the experience of seeing red alone, since that experience, by representing red as located in a property space of a particular shape, would already intimate the locations of co-determinate properties. The characterization of something as having a property located at longitude x, latitude y, and depth z in a space of co-determinate properties would already suggest the location of properties to the north or south, properties to the east or west, and properties above or below. Yet the experience of seeing something as red does not by itself reveal that the property now in view has a yellower neighbor (orange) and a bluer neighbor (violet), nor that it has more or less bright and more or less saturated neighbors, either. The current proposal has the unfortunate consequence that to see one color is, in a sense, to see them all.36 The current proposal thus continues to get the epistemology of color wrong. The Initial Fregean, Realization-Theoretic Proposal We therefore return to the initial candidate for a Fregean, realization-theoretic version of physicalism. This was the theory that objects are visually characterized as having properties that normally cause color sensations, and that colors are the higher-order properties expressed by these characterizations. 37 The content that this theory assigns to visual experience, say, of red would be introspectively recognizable as representing a genuine property; for even if there is no property that’s predominantly responsible for sensations of red, the property of having such a property is undoubtedly genuine. Furthermore, any internally similar experience would be introspectively recognizable as representing the same (higher-order) property, by virtue of containing the same characterization. And color properties, so defined, would stand in relations of similarity and difference generated by similarities and differences among the associated sensations. If one visual sensation differed from another in various respects, then the properties of being equipped to cause those sensations would differ isomorphically, by differing as to the sensations caused. Reflection on how it feels to see things as red and as orange would therefore be sufficient to reveal similarities and differences among those colors. This Fregean, realization-theoretic version of physicalism can therefore account for our knowledge of colors. Unfortunately, it does so at the expense of misrepresenting the phenomenology of color experience. The present theory implies that the content of visual experience alludes to color qualia as properties distinct from the perceived colors of objects.38 In order for one to see an object as having the property that causes visual experiences with a particular feel, one’s experience would have to represent that feel as well as the property causing it. And one’s experience would then lack the naivete characteristic of vision. Visual experience is naive in the sense that it doesn’t distinguish between the perceived properties of objects and the properties of perceptions. Whereas the experience of pain, for example, distinguishes between an external cause (a pin’s sharpness) and its sensory effect (a finger’s pain), visual experience does not distinguish between color as it is in the object and as it feels to the eye: one feels sharp points as causing pains but one doesn’t see colored surfaces as causing visual feels. The normal experience of seeing an object as red no more alludes to a sensation as distinct from the object’s redness than it does to tomatoes or fire engines. Thus, the only version of physicalism that gets the epistemology of color experience right gets the phenomenology wrong. In our opinion, any version of physicalism that acknowledges color qualia will commit the same phenomenological error, since it will imply that visual experience always has introspectible color qualities over and above the color properties that it attributes to objects. But this general thesis need not be defended here, since the only version of the second proposal that has survived our epistemological arguments is the Fregean, realization-theoretic version, which portrays visual experience not only as having introspectible qualities but also as alluding to them in its representational content. This version of the proposal implies that visual experience not only involves color sensations but is also about those sensations, in addition to color properties-which is clearly mistaken. CONCLUSION We do not pretend to have proved that any physicalist theory of color must be inadequate, since we have not canvassed every possible theory. We think of our arguments as posing a challenge to any aspiring physicalist. We challenge the physicalist to explain how the physical properties that constitute colors, in his view, are represented in visual experience, and to explain it in a way that meets reasonable epistemological and phenomenological constraints. One might think that the constraints that we have applied cannot be met by any theory of color, and hence that they must be unreasonable. The solution to the problems we have raised, one might conclude, is not to reject physicalism but rather to relax our epistemological and phenomenological constraints. In our view, however, there is a theory that satisfies these constraints, and it is one of the oldest and most familiar. It is the theory that colors are qualitative properties of visual experiences that are mistakenly projected onto material objects. A defense of that theory must be deferred, however, to another occasion.39