Seeing RedRed is the prototype of a raw sensory quality. How could seeing red possibly involve a sensorimotor skill? If it does, can the resultant new view of color provide new insight into classic philosophers' conundrums such as the problems of the inverted spectrum and Mary the color-blind color-scientist?
A step towards a sensorimotor theory of color sensation can be made by making use of a result discovered by Philipona & O'Regan. They showed that when changes occur in the spectral composition of the light that illuminates a colored surface, as compared to what happens for achromatic surfaces, the resultant changes in the reflected light measured by a human's three photoreceptor types obey precise laws that are characteristic of different surface reflectances. What is meant by a surface color would be the particular law involved.
Thus for example, Philipona & O'Regan show that surfaces which are classified by humans as being red in color turn out to have the peculiar property that when the illumination changes, the information available to the visual system about light reflected off these surfaces varies in a very particular, and much simpler way than for other surface colors. This, they suggest, might be the basis for red being universally considered a very special and salient color. Analogous singular reflectance properties can be shown for certain very particular surfaces generally labelled as blue, yellow and green. Several well-known and up till now unexplained results on so-called unique hues and hue cancellation correlate with surprising accuracy with the analyses presented by Philipona & O'Regan.
A sensorimotor approach to color would now consist in saying that even though perceiving a color generally does not involve changes in incident light or any action on the part of the observer, the sensation of redness resides in the fact that a perceiver knows, implicitly, that if the incident light were to change, perhaps by the observer moving the surface, the resultant changes in photoreceptor activation would obey the laws characteristic of that particular color. (see later for the importance of the observer's own action in this concept of color) Perceiving the color of a surface is thus, under the sensorimotor approach, knowing that certain laws of change would apply if the surface were to be moved, or if the light were to change.
Bompas & O'Regan have successfully tested predictions that follow from this sensorimotor approach by showing that color sensations do indeed change when the sensorimotor laws involved in color sensation are artificially modified by using half-field tinted goggles or eye-contingent computer displays to modify the normal link between incoming light and changes in eye position.
The sensorimotor approach to color thus provides a way of thinking about color in which color sensation is linked to the objective laws that describe the changes that an observer with particular photoreceptors observes when he manipulates surfaces or when illumination changes. The similarities and differences between perceived colors should, under this view, be limited by the similarities and differences in these objective laws. The successful empirical predictions of the approach lends it credance.
Why does this approach to color solve any problems? The main advantage of the approach is that the relation between activation of brain mechanisms and resultant color sensation is no longer an arbitrary relation. There is therefore no room for questions such as: why does this color channel provoke this particular color sensation rather than that one? Just as the feel of driving a car is constituted by knowing that you are engaged in the particular things you do when you drive the car (you cannot by definition have the Porsche driving feeling when you are currently engaged in driving a Volkswagen), the sensation of different colors requires one to have cognitive access to the fact that one is engaged in the sensorimotor skill corresponding to those colors. A consequence of this is that the sensorimotor approach has a principled way of explaining the structure of perceptual color space: why for example red is more similar to pink than to green. On the other hand a neural correlate approach has to rely on postulating a particular link between neural excitation and sensation: and explaining why there is that link rather than another possible link will then require yet another explanatory mechanism.
The inverted spectrum and Mary the color scientistWhat does this sensorimotor approach to color have to say about the inverted spectrum problem? The inverted spectrum problem is a problem that makes sense if the experienced quality of a stimulation is assumed to have an arbitrary link with the characteristics of the interaction an agent has with the stimulation — thus the problem makes sense under the classical approach in which it is supposed that feel is generated by special neural mechanisms.
But under the sensorimotor approach, the quality of perceived color is predominately determined by objective facts about an agent's interaction with colored surfaces. Thus under this view it will generally make no sense to say that one might have the green feeling when one is looking at red, since when one is looking at red, the laws of interaction that apply must be the red ones and not the green ones. Some provisos to this scheme could however arise if a modification has occurred in the observer's sensory apparatus which changes the laws of interaction. It is worth expanding on this, taking as example a more obvious case than color, namely the case of adaptation to spatial distortions in the visual field.
Consider wearing distorting spectacles that compress the vertical dimension. While initially an observer wearing such spectacles would see circles as flattened and squares squashed into rectangles, over time, if the perceptual system is sufficiently adaptable, experience with the world would ultimately, at least logically, have to result in the observer correctly understanding the shapes of objects. This is because any verification that the observer might wish to undertake would involve tests and comparisons with other figures which themselves undergo the distortions. For example, to check whether a figure is square, one would take a ruler and measure the height and the width. Because the ruler itself suffers the same compression as the figure itself, estimation of squareness would be veridical despite the optical compression. Another example of a means to test whether a figure is square would be to rotate it through 90 degrees and check whether it looked the same. Indeed, the projected image (though not itself square because of the compression), will look the same after rotation. Nevertheless, because of extensive prior experience with undistorted sensory input, adaptation of the observer's sensory systems may take time to adequately interiorize the new laws of sensorimotor interaction. This will thus give rise to perceptual effects which may take a long time to dissipate, or which may never dissipate.
What does the sensorimotor approach have to say about the problem of Mary the color scientist who cognitively knows everything there is to know about color, but has been confined all her life to an achromatic room? Does Mary learn anything new when she is finally released from her monochrome prison? The sensorimotor approach claims that experiencing color involves having cognitive access to the fact that one is currently exercising the laws of sensorimotor contingency that are typical of color. Mary the color scientist, in her achromatic room, has never actually exercised the sensorimotor skills associated with color, so has never had color sensation. Mere cognitive access to the laws of color does not constitute cognitive access to the fact that one is actually engaging in the sensorimotor interaction necessary to experience the feel of color.