In addition to variation between subjects, there is also variation within subjects. Color matching depends on visual angle (Stiles 1937). The degeneration of the lens with age makes it yellower, producing a shift in perceived hue, with purple objects looking significantly redder (Fairchild 1998, p. 5). Color perception can vary between the right and left eyes due to differences in the optical density of the macula (Fairchild 1998, p. 7).
These facts give rise to an obvious problem, which C. L. Hardin nicely expresses as follows:
[Imagine that all of the] hue chips manufactured by the Mun- sell Company covering [the] 5 Blue-Green to 2.5 Green range were randomly spread out before you to be separately viewed on a dark gray background in North Daylight. One of them would be your considered choice for unique green. Your col- league might make a different choice. If so, which of the chips is unique green? (Hardin 1993, p. 80, endnote omitted)
According to Hardin, “if this question is to be answered at all, it can be answered only by convention. We might, for example, decree that the most frequently chosen chip is to be unique green. But we could decide otherwise” (p. 80).
Hardin’s answer to his own question is a little odd. Sup- pose a certain chip looks to you to be unique green. Convention has nothing to do with this: what makes it the case that the chip looks unique green are facts about your visual system and its interaction with the chip, and these are not matters of convention or decision. Now consider the ques- tion of whether the chip is as it looks. Convention has noth- ing to do with this either: it is entirely a matter of how things are with the chip. If the chip is unique green, then the an- swer is yes; if not, no.
We suspect that Hardin’s eliminativism is influencing his answer. Even if, as Hardin thinks, nothing is red, blue, yel- low or green (let alone unique green), color terminology has great practical value. For various pragmatic reasons, it would not be a good idea to speak the literal truth and to refuse to apply color expressions – for example, “unique green” – to anything. So how should we use this expression? Obviously the answer to this question is a matter of con- vention: the question calls for a decision, not a statement of fact. But this is not the question that Hardin is officially ask- ing, although the two might be easily confused. By his own lights, what Hardin should have said in answer to his offi- cial question is that – as a plain matter of non-conventional fact – neither chip is unique green.
If this answer – that neither chip is unique green – is cor- rect, then we are in trouble. For, since we may fairly sup- pose that if anything is unique green, one of the chips is unique green, the proper conclusion is that nothing is unique green. And if nothing is unique green, it is hard to see why other shades of green, or of any other color, are any better off. The natural terminus of this line of thought is therefore that nothing has any color, that is, that elimina- tivism is true. So we do not have here a problem solely for physicalism, but rather for any realist theory of color.
But what is the problem, exactly? What the facts about individual differences in color vision show is that, under the twin assumptions (a) that objects do not have many differ- ent colors simultaneously (e.g., if a chip is unique green, it is not also bluish-green), and (b) that if objects really are colored (e.g., a certain chip is unique green), then there iS widespread misperception of the determinate colors: Many people will misperceive a chip that is in fact unique green as slightly bluish-green, for instance. If this can be turned into a good argument against color realism then two things must be established. First, that the conclusion, widespread misperception of the determinate colors, is unacceptable. If this is right, then we have to reject either (a) or (b). Sec- ond, to complete the argument it must be established that (b) is the culprit.
We think this argument fails at the first stage, because the conclusion is not unacceptable. First, note that the conclusion is not especially astonishing or at odds with apparently obvious facts. The conclusion is not that people rarely see objects as having the colors they actually have, but that they rarely see objects as having the determinate colors they actually have. It is consistent with the conclusion that people typically see green objects as green, orange objects as orange, and so forth. Second, note that similar conclusions hold for other perceptible properties, for example, spatial properties. For a concrete case consider aniseikonia; a moderately common opthalmological condition in which the size (or shape) of the retinal image differs between the two eyes.