| OCR Text |
Show Rationality and the Structure of the Self, Volume I: The Humean Conception 169 38 straightforwardly logical inconsistency: H both is and is not preferred to F. Just because we cannot symbolize a logical contradiction within the constraints of standard decision-theoretic notation does not mean we do not know it when we see it, nor that we cannot describe it in natural language. We have just done so. So although Rex has thereby produced a cyclical ordering, 39 he nevertheless cannot be said to have ranked F, G, and H at all. Rex's case shows that, as Kant might have put it, concepts without memory come up empty. Without a recollection (or "synthesis") of previous ranking occasions to supply continuity with this one, the concept of a thing's ranking superiority is applied so indiscriminately that it has no proper application at all. But now contrast Winifred with Wallace, whose selection behavior violates (a) but satisfies (b). Wallace remembers on each trial the relation of the pair of alternatives he is ranking to the third he is not, but lacks the concept of ranking superiority to apply to his behavior. If he never forms this concept, he cannot be described as preferring any one alternative to any other. But could Wallace develop and consistently apply that concept to the ordering he produces? - Yes, assuming he does not thoroughly change his mind about his previous orderings. That is, he can form the concept of ranking superiority only if he is minimally psychologically consistent. Minimal psychological consistency may provide a kind of foundational structural support that determines the transitive consistency of future pairwise comparisons with past ones, and so enables the concept of ranking superiority to develop. In the absence of this support, it is hard to imagine how it might. Take Wallace's first pairwise comparison, the choice of F over G at t1. This alone would not enable him to form the concept of F's ranking superiority. For this would be to treat F as an instance of something's ranking superiority. But 38 Essentially this is Davidson, McKinsey and Suppes' defense of transitivity (op. cit. Note 26, 145-6), although they do not distinguish unidimensional from multidimensional orderings. They recognize that the irrationality of a cyclical ordering consists not simply in a violation of transitivity, but of logical consistency in the application of the concept of rational choice. That concept plays the same role in their argument that the concept of ranking superiority plays in mine in this volume; in Volume II, Chapter III, I demand greater formal rigor from the notion of logical consistency. Here Schwartz (ibid.) seems patently mistaken in supposing that under these circumstances, "it is a matter of indifference" which alternative is chosen - unless he means it is a matter of indifference to an agent who is unable to comprehend what is involved in ranking alternatives at all. 39 Of course if Winifred forgets the relation of the pair she is ranking to the third alternative she is not, she is free to apply the concept of ranking superiority locally, to either of the two alternatives presented on each of the three trials sequentially, without regard to what has occurred on either of the others. But it will still be true in fact that, at t3, when she produces a cyclical ordering of F, G, and H, we will have applied that concept inconsistently, if we continue to insist on describing Winifred's behavior as an instance of intentionally "ranking F, G, and H." © Adrian Piper Research Archive Foundation Berlin |
