| OCR Text |
Show Chapter IV. The Utility-Maximizing Model of Rationality: Formal Interpretations 172 This is a condition about which conventional decision-theoretic interpretations of (T) have nothing to say. (T)'s silence on the concept of a genuine preference results not only from conflating genuine preference with selection behavior, as Sen has shown. It therefore results from failing to incorporate the fact that, like any human behavior, the selection behavior of making pairwise comparisons is physically discrete and time-dependent, whereas the logic and concepts that inform the preferences thereby expressed are not. The process of making transitive pairwise comparisons can be represented as selection behavior of the following sort, where "s" means "is selected over": (Tt) t1: FsG t2: GsH t3: FsH Now ordinarily, when Horace is asked to rank F, G, and H, we assume (b) above, i.e. that he remembers at t2, while ranking G and H, his previous ranking of F and G at t1. So if he has selected F over G at t1 and G over H at t2, we do not therefore suppose, as we did provisionally for Wallace, that Horace has changed his mind about G from t1 to t2, first ranking it as inferior and then as superior. We do not conclude this because we understand at t2 that whereas earlier Horace was ranking G relative to F, he is now ranking it relative to H; and these two rankings are obviously consistent. Ordinarily, then, we also assume condition (a) above, i.e. that there is a most selected alternative common to both trials, namely F. These commonsense assumptions ensure the interpretation of Horace's preferences as psychologically consistent. Canonical preference language gives no indication of this psychological consistency, and fails to incorporate the conditions that ensure it. We have already seen that this is what makes (T) vacuous. Instead, it collapses preferences that satisfy (a) and (b) into a time-dependent series of discrete physical selection behaviors. Thus literally interpreted, the first two pairwise comparisons are psychologically unconnected and temporally sequential, such that G is ranked lowest in the first trial and highest in the second (recall that this absence of a most highly ranked alternative common to t1 and t2 accounted for Wallace's learning problem). But of course this literal interpretation is not the one we make, for it misrepresents the rank we assume Horace has in fact assigned to G. We do not think that he prefers G most at one moment and least the next. This would be logically inconsistent, in the familiar, time-dependent sense in which we say of someone with unstable or shifting opinions that they are "constantly contradicting themselves". Describe this as intertemporal logical inconsistency. If Horace were simply to © Adrian Piper Research Archive Foundation Berlin |
