theories of color perception
 
 
Theories of color perception propose to explain how it is that colors are perceived as properties of physical objects. What proposal one makes depends in turn on what proposal one makes about the nature of color (see theories of color). See theories of color.
 

Details:
 
Introduction
 
By contrast with theories of color, which address the problem of the nature of color, theories of color perception address the problem: how is it that colors (whatever their nature) are perceived as properties of physical objects?
 
However, the answer one gives to this question about color perception does depend on one's theory of color. If one proposes that the colors we perceive physical objects as having are mental qualitative properties of visual states themselves (the proposal about the nature of color called subjectivism), then one must explain how it is that mental qualitative properties are perceived as properties of mind-independent objects. The subjectivist's options seem to be limited to sense datum theories and projectivist theories.
 
If one proposes that the colors we perceive physical objects as having are properties of physical objects (as is claimed by physicalist and dispositionalist proposals about the constituting nature of color), then an explanation of how it is colors are perceived as properties of physical objects is relatively straightforward. However, the question remains whether this access is mediated by visual states with mental qualitative properties.
 
Intentionalist theories of perception claim that visual states of color are representational states which have no qualitative properties apart from those that they represent. Thus, it holds that the qualitative aspects of color experience aren't determined by mental qualitative properties of experience itself. Most intentionalists hold physicalism about the nature of color, which claims that the colors we attribute to physical objects in visual states are physical properties of physical objects. Also, most physicalists (including Smart, Armstrong, Hilbert, Byrne and Hilbert) are intentionalists.
 
Intentionalism is rejected by most dispositionalists about the nature of color. According to dispositionalism, the colors we perceive physical objects as having are dispositions of physical objects to produce perceptual responses. The most common version of dispositionalism holds that perceptual responses are visual states of color characterized in terms of mental qualitative properties, in particular, color qualia (currently, Peacocke and McDowell; formerly, McGinn and Johnston).
 
However, even though most dispositionalists reject intentionalism, these theorists do not thereby claim that the colors we perceive physical objects as having are color qualia themselves. (If they were to claim this, they would have to explain how qualia are perceived as properties of mind- independent objects--the options would be some version of sense datum theory or projectivism). Rather, Peacocke, for example, claims that the colors we perceive physical objects as having are dispositional properties of physical objects. Color qualia determine what it's like to experience physical properties of physical objects as colors.
 
Thompson's answer to the problem of color perception cannot be easily categorized as intentionalist or nonintentionalist. He addresses the problem from the standpoint of the evolution of color perception in different species. His answer takes an ecological approach to characterizing color perception.
 
 
Sense datum theories
 
According to sense datum theories, perceptual states are relations between perceivers and sense data, mental objects that have mental colors and mental shapes. According to Jackson's sense datum theory (which Jackson has by now rejected--see Jackson and Pargetter), physical objects seem to be colored because colored sense data are caused by physical objects and the spatial properties of sense data vary as a function of the spatial properties of physical objects.
 
Sense datum theories face difficulties in explaining how sense data, as mental objects, are related to the physical states and in particular the neural states of perceivers. Furthermore, according to Jackson's theory, sense data, which are mental objects, are located in three-dimensional physical space. Jackson's theory faces difficulties in explaining the relation between perceivers and sense data located outside of their bodies in physical space.
 
By contrast with sense datum theories, adverbialist theories claim that perceptual states are relations between perceivers and (actual or nonactual) physical objects rather than sense data, and that perceptual states themselves have mental properties such as mental colors. Mental colors are then held to be identified with, or supervenient on, properties of neural states of perceivers. Most current theories of color perception which claim that perceptual states have mental colors assume an adverbialist theory of perceptual states of color.
 
 
Projectivist theories
 
According to projectivist theories, in color perception mental colors, characterized as properties of perceptual states rather than sense data, are experienced as properties of mind- independent objects.
 
There are two versions of projectivism, which provide different explanations for how it is that mental colors are experienced as properties of mind-independent objects. According to one, proposed by Boghossian and Velleman, mental colors are experienced as properties of physical objects.
 
But since mental colors are properties of visual states, experiencing mental colors as properties of physical objects involves experiencing properties of mental states as properties of physical objects. Since states are fundamentally different sorts of things than objects, it's not clear that it even makes sense to hold that we experience properties of mental states as properties of physical objects.
 
According to the other version of projectivism, proposed by McGilvray, we are not aware of physical objects or any of their properties in perception. McGilvray doesn't deny that there are physical objects. Rather, he claims that the spatial properties as well as the colors we're aware of in visual perception are mental qualitative properties of visual states themselves. Thus what we're aware of in visual perception are mind-dependent patches with mental colors and mental shapes. (Although this claim suggests a sense datum theory, McGilvray explicitly accepts an adverbialist theory of perceptual states.) McGilvray claims that projective representation of color involves experiencing such mind-dependent color patches as external to our minds.
 
But if the locations we're aware of in visual experience are never physical locations, the problem now is that it's difficult to characterize these mental locations that McGilvray claims we are aware of. He holds that we can describe mental locations as external to our minds by way of describing them as points in a three-dimensional visual field. However, McGilvray provides no explanation of the relation between perceivers, located in physical space, and the three-dimensional visual field comprised of mental locations, rendering mental locations mysterious.
 
 
Intentionalist theories
 
Intentionalist theories of perception claim that visual states of color are representational states which have no qualitative properties apart from those that they represent. Any proposal about color perception that claims that visual states of color have mental qualitative properties, for example, color qualia, rejects intentionalism.
 
An objection to intentionalist theories asserts that such theories cannot account for the colors of afterimages. In a recent formulation of this objection, Boghossian and Velleman claim that when we represent an afterimage as an afterimage, and not a physical object, what's represented as colored can't be a physical object, but rather must be a part of the visual field itself. (Although this claim suggests a sense datum theory, Boghossian and Velleman state that their claim is consistent with adverbialism.)
 
Boghossian and Velleman assume that when we represent an afterimage as colored, then something is represented as colored. However, it isn't clear that their assumption is correct. We may make the inference that if we represent an afterimage as colored, then we represent something as colored. But we can't make the inference that there is something that is represented as colored, namely, the visual field.
 
Rather, intentionalists would argue that when we represent an afterimage as colored, we attribute a physical qualitative property to a merely intentional--not an actual--object. What intentional objects are and how physical qualitative properties relate to them is indicated by a theory of intentionality.
 
However, those who accept that visual experiences have mental colors, for example, qualia, may claim that afterimage colors are mental colors. And this may seem to be a more plausible way of accounting for afterimage colors.
 
 
An ecological approach
 
Thompson addresses the question of how colors are perceived as properties of physical objects from the standpoint of the evolution of color perception in different species. On the basis of comparative studies of the visual systems of different species, Thompson claims that different species' color vision has evolved to perform biological functions which must be described in ecological terms.
 
But how we are to characterize the biological function of color vision is controversial. (The notion of biological function used in this context is Wright's notion, or some variation on it.)
 
Hilbert addresses the problem of color perception from the standpoint of computational color vision, according to which vision involves the processing of information contained in representations that encode physical properties of objects in the world. A focus of research in computational color vision is the problem of how a visual system achieves what's called color constancy. Color constancy is a visual effect in which the color a physical object looks remains fairly constant despite changes in the light illuminating the object.
 
As it turns out, the constant colors that we perceive physical surfaces as having are correlated with surface reflectance. (Surface reflectance is the ratio of reflected to incident light for a surface, where light is measured by its wavelength composition and its intensity at each wavelength.) Hilbert argues that color constancy shows that the biological function of color vision is to detect physical object surfaces by way of surface reflectance, and thus characterizes the biological function of color vision in physical terms.
 
By contrast, Thompson offers an alternative characterization of the biological function of color vision in ecological terms. Thompson appeals to Mollon's proposal that the biological function of primate color vision includes promoting the detection of objects in contexts described in ecological terms, for example, the detection of fruit against a background of foliage, or the detection of objects against backgrounds which may include things without surfaces, such as volumes of water or the sky. According to Thompson, comparative color vision indicates that there are nonhuman species with color vision, and that for these species the biological function of color vision cannot be described merely in terms of detecting object surfaces by way of surface reflectances.
 
Hatfield provides further considerations in favor of characterizing the biological function of color vision in ecological terms. Psychophysics indicates that objects with different surface reflectances can produce perceptions of the same determinate color (such as teal). Such physically distinct objects are called metamers. If we describe the biological function of color vision as the detection of object surfaces by way of reflectance properties, then metamerism marks a failure of the visual system.
 
Hatfield points out that if the biological function of color vision isn't best described as the detection of object surfaces by way of surface reflectance, but rather as, for example, the detection of objects against backgrounds, then it doesn't follow that metamerism marks a failure--the task of detecting fruit against a background of foliage isn't hampered by metamerism. In fact, the detection of objects against backgrounds is better served by a relatively small number of broad color categories which include starkly contrasting pairs, such as red and green, and yellow and blue.
 
Thompson argues that if the biological function of color vision is to be described in ecological terms, then colors are relations between perceivers and objects. Thompson supports this claim by citing studies that indicate that species which have evolved in different ecological niches have different categories by which colors are classified as qualitatively identical or similar.
 
But it may be that whereas the color categories of a species are best explained in terms of ecological relations between members of the species and objects in their environments (over the course of evolution), the colors themselves are nonrelational properties of objects.
 
 
Peter Ross