| OCR Text |
Show Rationality and the Structure of the Self, Volume I: The Humean Conception 161 Since the appeal to the principle of charity is intended to interpret a temporally sequential series of pairwise comparisons of alternatives, and since this appeal is required in order to preserve the transitivity of those comparisons in the face of (C), the vacuity of the concept of a simple ordering 32 arises, in part, from its implicit time-dependence. Then even when (Ct) obtains, (U) is satisfied; i.e. (U) is vacuous. Seemingly intransitive behavior explained by systematic changes in taste resulting from learning or sequential effects over repeated trials is one obvious application of the principle of charity. But (T) may be vacuously preserved by appeal to this principle even in cases otherwise explained as a momentary lapse or glitch in the evaluative process. In these cases, too, no 33 genuinely intransitive behavior can be identified. Thus the probabilistic contrary, the utility-maximizing theory of choice may meaningfully survive if its explanatory scope is reduced. If not, I think it is not "this quibble" that should be "rejected at once." Davidson, McKinsey, and Suppes (op. cit. Note 26) conceive the problem as consequent on a revealed preference interpretation of choice. They think that if preference is taken to be equivalent to selection behavior, then each timedependent selection may spring from a "momentarily rational preference ranking," and so we cannot prove what a person's preference ranking is over more than two alternatives. Their solution is to interpret particular selections as evidence for preference interpreted as a disposition to select. A cyclical ranking is then "evidence, so far as it goes, that [a person's] preference ranking is not rational; but we would reconsider this verdict if we learned he had changed his mind about the relative ranking of a and c after his first two choices" (147). But to change one's mind after each selection just is to have a series of "momentarily rational preference rankings." So the dispositional interpretation of preference is of no help here; I address some further arguments for this interpretation in Note 34, below. Also see Donald Davidson, Sidney Siegel, and Patrick Suppes, "Some Experiments and Related Theory on the Measurement of Utility and Subjective Probability," Applied Mathematics and Statistics Laboratory, Technical Report 1, Stanford University, Stanford, Cal., August 15, 1955. 32 Notice that it will not solve the problem to devise a way for Cyril to select a simultaneous linear ordering of F, G, and H - say, by requiring him to choose among pressing buttons for an F-G-H, an H-G-F, and an F-G-H-F scale. If he chooses the last, the time-dependent appeal to the principle of charity can be made for scanning the scale just as for pairwise-ranking alternatives. The implications for the use of the principle of charity of "changing one's mind" under such circumstances are discussed further in Section 3.1, below. In any case, a simultaneous linear ordering is a plausible substitute for pairwise comparisons only where the number of alternatives to be ordered is sufficiently restricted to those which are simultaneously comparable in practice. Even three or four such alternatives, simultaneously presented, is a stretch. 33 Again I mean to be referring only to unidimensional criteria for ranking alternatives. In "Intransitivity of Preferences" (op. cit. Note 23), Tversky demonstrates that genuinely intransitive behavior can be identified when multidimensional criteria are used. He acknowledges, however, that in the absence of replication, "one can always attribute intransitivities to a change in taste that took place between choices" (45). For this reason © Adrian Piper Research Archive Foundation Berlin |
