Have Byrne and Hilbert Answered Hardin’s Challenge?

I argue that Byrne and Hilbert have not answered Hardin’s objection to physicalism about color concerning the unitary-binary structure of the colors for two reasons. First, their account of unitary-binary structure seems unsatisfactory. Second, pace Byrne and Hilbert, there are no physicalistically acceptable candidates to be the huemagnitudes. I conclude with a question about the justification of physicalism about color. I. The Magnitude-Representation Account of Unitary-Binary Structure Fails In their impressive target article, Byrne and Hilbert (B&H) attempt to answer Hardin’s objection to physicalism about color. In my opinion the attempt doesn’t succeed. First, their account of unitary-binary structure in terms of the representation of four hue-magnitudes (call the account ‘MR’) seems mistaken. Consider a possible world w where, as it happens, everything that is visually represented as circular is visually represented as (having a ratio of) R and Y. In w, circularity satisfies B&H’s formula but is not binary reddish-yellowish. It would not be satisfactory for B&H to reply that the reason why circularity in w is not reddish-yellowish is that it is not a color, that is, on B&H’s view, an SSR-property. Why should an SSR’s satisfying B&H’s formula suffice to make it reddish-yellowish, while a shape’s doing so not suffice? Adding the modal operator ‘necessarily’ to the original account may solve the problem (since in w it only happens to be the case that circularity satisfies B&H’s formula), but it raises another. Since it is necessary that nothing is visually represented as prime, it is necessary that everything that is visually represented as prime is visually represented as R and Y. But primeness is not reddish-yellowish. To rule out such vacuous cases a further proviso must be added to the effect that the property involved is possibly visually represented. Perhaps MR could be amended to avoid counterexamples, but even then I don’t think that it could be right. B&H don’t say what they mean by ‘account’, but I take it that at a minimum for q to be an adequate account of p, q must specify the way the world must be in order for p to be true. 1 But what could make it the case that ‘That shade of orange is reddish-yellowish’, for instance, has the very complicated truth-condition B&H 1 B&H say that their account explains why “a binary hue like orange appears to be a ‘mixture’ of red and yellow” and why “green (and yellow, red, and blue) are said to be ‘unique’ hues” (sect. 3.2.3, para. 3; my italics). Do B&H then deny that orange is reddish-yellowish, and that red is a unique color? B&H’s circumspection here suggests that, despite what they say (sect. 3.2, final para.), they are error theorists about the unitary-binary distinction, and that their goal is to explain the error. For this reason, it is not entirely clear to me what B&H are up to. must assign to it? When we utter such a sentence, we don’t mean to attribute to the color a complicated relation to perceivers. Rather, it seems we mean to say something about its intrinsic nature. And an externalist/natural-kind account of how the sentence could have the required truth-condition seems inapplicable in this case. For this and other reasons (cf. Pautz 2002b), I think that MR does not correctly capture the unitary-binary distinction, and hence cannot be used to answer Hardin’s challenge. II. The Hue-Magnitudes are not Reducible to Extradermal Physical Properties Even aside from these problems there is reason to think that MR cannot provide a solution to Hardin’s challenge. MR appeals to four hue-magnitudes, but it is hard to see how they could be extradermal physical properties. B&H say that an object is reddish to a certain degree if it produces more L-cone activity than Mcone activity (sect. 3.2.3). However, they don’t want to identify a certain degree of reddishness with a disposition involving perceivers, but with a corresponding extradermal physical property not involving perceivers (personal communication). What physical property? They don’t say. Could it be the disjunction of all the SSRs of all actual or possible objects that are (on their view) reddish to that degree? Or maybe the property of having an SSR whose S*, M* and L* components stand in a certain complicated relationship, where S*, M* and L* correspond to the short, medium and long regions of the visible spectrum (Bradely & Tye 2001)? Aside from a priori problems to do with the explanatory gap between qualities and quantities (a problem which B&H don’t address) my main worry is that none of the extradermal candidates stand in the same higher-order relations of congruence and proportion that the degrees of reddishness (etc.) stand in. It doesn’t even make sense to say that the difference between the disjunctions of SSRs D1 and D2 is the same as the difference between the disjunctions of SSRs D2 and D3, or that D1 is twice as great as D2, while it certainly makes sense to say such things of degrees of reddishness. As for the second candidate, equal differences between degrees of yellowishness, for instance, do not map onto equal differences between the corresponding values of [(L* + M*) – S*], because the relationship between these variables is nonlinear (Werner & Wooten 1979). Since degrees of yellowishness and the corresponding values of [(L* + M*) – S*] stand in different objective congruence relations, they cannot be identical. (Compare pitch and frequency.) B&H might reply by adding co-efficients, exponents, and so on (following Bradely & Tye 2001). But these operations don’t apply to properties (it makes no sense to square a property), but to the numbers by which we index them. So this maneuver doesn’t yield a new set of physical properties which do correlate with degrees of yellowishness, but only a new way of assigning numbers to objects, that is, a new set of relations between objects and numbers. And degrees of yellowishness certainly aren’t relations to numbers. III. There is no Argument for Reflectance Physialism MR also appeals to visual representation, but there is very good reason to think that it cannot be reduced. (I understand reduction broadly here to include identification with physical properties or physically-realized functional properties.) Visual representation is a relation between people and extradermal properties such as colors (on B&H’s view, SSRs). (Strictly speaking, it is supposed to be a relation between peoples’ experiences or brain states and propositions, but such niceties will not matter here.) So on B&H’s view, if visual representation is identical with a physical/functional relation, it is identical with a physical/functional relation between people and (inter alia) SSR properties. Call this the ‘Relationality Constraint’. But what physical/functional relations obtain between people and SSR properties to which visual representation might be reduced? It seems that the only candidates are extrinsic, causal/teleological relations (Tye 2000, Dretske 1995). But there are good reasons to think that visual representation cannot be such a relation. First, there are serious problems of detail (Loewer 1997). Concerning causal theories, B&H themselves say “we do not actually find any of these theories convincing” (sect. 2.6). Second, there appears to be a very simple argument from the opponent process theory of color vision (OPT) and representationism to the failure of all such externalist accounts. (B&H appear to accept both premises. See Byrne & Hilbert 1997 and sect. 3.5 of the target article.) Let w be the closest possible world where, owing to evolved differences in our postreceptoral processing, our opponent channel states are regularly different, but where our receptor systems are the same, so that the states of our visual systems, though different, are optimally causally corrected with, and designed by evolution to indicate, the very same extradermal properties. By OPT, we have different color experiences in w, and so, given standard representationism, represent different color properties (on B&H’s view, different hue-magnitudes or different ratios of the same hue-magnitudes), under the same circumstances. But the states of our visual systems are optimally causally correlated with and designed to indicate the very same properties (e. g. the very same SSR properties). So our representing different color properties in w cannot be accounted for in causal/teleological terms (for full details, see Pautz 2002a). Could visually representing a certain color then be reduced to a neurobiological (e. g. opponent channel) property or to a forward-looking narrow functional property concerning what other inner states and behaviors a given inner state is apt cause? No, because (artificial tricks aside) these ‘internal’ properties don’t satisfy the Relationality Constraint: none is a relation to a color. At most, visually representing a certain color (and hence, given standard representationism, color phenomenology) supervenes on or is constituted by such an internal physical/functional property, without being reducible to it. (On supervenience/constitution without sameness of logical form, see Horwich 1998 and McGinn 1996.) For these reasons it appears that something like Primitivism is the right view of visual representation. Many would argue that this is not an isolated case, and that reduction (as opposed to the weaker relation of supervenience/constitution) is in general an unattainable aim. This raises a question for B&H. Either they must convince us, as against these arguments, that there is a physical/functional relation between people and SSRs to which visual representation might be reduced, 2 and more generally that reduction is the rule, or else they must explain why, if the ‘plausibility’ arguments for reduction (avoiding brute emergence, causal considerations) don’t work in general, we should think that they work in the case of colors, notwithstanding the considerable a priori and empirical obstacles standing in the way.