I have previously argued that gappy physicalists who embrace Swamp Mary cannot account for her satisfaction of psychosemantic criteria along the lines of either quotational, causal, or nomological accounts.
Let us turn, then, to examine the availability, or lack thereof, to gappy physicalists of a psychosemantics along the lines of Descriptive-isomorphism. It helps to first begin by examining a clear case to which Descriptive-isomorphism is most likely to apply. Suppose that someone, let us call him George, is an otherwise normal human who has never before seen the Japanese flag. Let us suppose that George has normal color vision, and has had a typical range of color experiences including having seen red and white before. Suppose further that George has seen various shapes before, including circles and rectangles. George just has never seen a Japanese flag before. Without showing him one, how can we augment his knowledge to include knowledge of what it would be like to see one? Here the answer seems quite simple. We convey to George a description along the lines of “red circle centered on a white rectangular background.” The adequacy of such a description, according to Descriptive-isomorphism is due chiefly, if not exclusively, to two factors. The first factor is the flag’s being a structured complex (with the comparative simples being redness, circularity, etc. and structural relations being the having of redness by the circle, the centering of the circle on the rectangle, etc.). The second factor is the description’s being a structured complex allowing it to pick out the parts of the complex and their structural relations to each other. It may perhaps ease subsequent exposition to label these two factors “content complexity” and “vehicle complexity,” respectively.
Given how Descriptive-isomorphism works for George, it is quite difficult to see why it wouldn’t work just as well for Mary. One of the key suppositions of physicalism is that qualia are ontological complexes of non-phenomenal relative-simples. Thus are qualia in a position to satisfy the first key factor of Descriptive-isomorphism, the content complexity factor. And Mary is physically omniscient, so there would be no non-phenomenal entity or non-phenomenal relations between entities that would exceed her grasp. Thus are states of Mary in a position to satisfy the second key factor of Descriptive-isomorphism, the vehicle complexity factor. I am aware of no third factor essential to Descriptive-isomorphism, so I’m aware of no basis for the gappy physicalist to embrace Descriptive-isomorphism as the psychosemantics of Swamp Mary’s phenomenal knowledge while maintaining prerelease Mary’s phenomenal ignorance.
It’s natural here for the gappy physicalist to try to claim that Mary cannot actually satisfy the vehicle complexity factor of Descriptive-isomorphism. The maneuver I have in mind is to affirm that while Mary is able to represent the phenomenal facts under one description, there is some other description that prerelease Mary is not able to represent the facts under. One way in which this might be put is to say that prerelease Mary knows the facts under a physical description but not under a phenomenal description. It is hard to see, however, what is to distinguish kinds of description here in a way that is also going to be consistent with key tenets of both Descriptive-isomorphism and gappy physicalism.
Suppose that the relevant difference between the descriptions is thought of as analogous to differences between inter-translatable descriptions from distinct languages. Compare the difference between, for example, the English phrase “white rectangle” and the Pig Latin phrase “itewhay ectangleray”. According to Descriptive-isomorphism, their inter-translatability is due to structural commonalities between the descriptions (or commonalties between structures such as languages or language games that the descriptions are respectively embedded in). What makes them different descriptions isn’t going to be any difference in what they are descriptions of, that is, differences in semantic content, but instead differences in the representational vehicles. (The vehicular differences may, of course, be relational as well as intrinsic.)How can such vehicular differences suffice for prerelease Mary’s phenomenal ignorance? The suggestion along such lines has to be something like that Mary may be able to get into one kind of state, but not be able to get into another. But note now what the key features of this account of Mary’s prerelease omniscience and ignorance amounts to: her omniscience is due to her having a representation, under some description, of every fact, and her ignorance is due to there being states that she is nonetheless unable to get into. But this is just to give a version of the ability response(Lewis, 1990). And the ability response is a defense of physicalism that is not a gappy physicalism. It is crucial to gappy physicalism to hold that what Mary lacks prerelease is a kind of knowing-that, as opposed to a kind of knowing-how.