| OCR Text |
Show Chapter III. The Utility-Maximizing Model of Rationality: Informal Interpretations 102 This makes (U) and its alternative reformulations a special case of a more general, Kantian theory of rationality that situates the principle of noncontradiction at its foundation. However, by this I do not mean what Allais means. For him, action that is "non-self-contradictory" and so rational satisfies two criteria: first, its ends are "logically consistent;" and second, its means are appropriate to them. "Logically consistent ends" for Allais are those which constitute a set ordered by his axiom of absolute preference, i.e. such 9 that its elements are ordered by the relation "<<". I agree with Allais' insistence that consistency is the only criterion for the rationality of ends, which are otherwise arbitrary, and with his sensitivity to legislating any more substantive, "politically correct" criteria for the 10 rationality of ends. But his account would not suffice for logical consistency as philosophers ordinarily use that term; nor would the more general account of non-self-contradiction in which his notion of logical consistency figures. In the philosophical context, two sentences are consistent if and only if one does not contradict the other. We ascertain this by quantifying them using the conventions of predicate logic, and relating them by means of the traditional Boolean connectives ".", "v", "", and "~". However, the ends that constitute elements in Allais' ordered set are not quantifiable using the symbolic resources of predicate logic, and "<<" is not one of the Boolean connectives. Below in Chapter IV, Sections 2 and 3, I argue that no notion of consistent choice that does not meet these two basic philosophical desiderata can do the job even on its own turf; and in Volume II, Chapters II and III of this project I try to develop one that does. 2. The Single End Interpretation of (U) I describe as the single end interpretation of (U) the commonsense notion that a rational agent maximizes utility if he acts efficiently to achieve a particular goal, i.e. by minimizing the expenditure of resources in its service (this is efficiency in the pedestrian rather than the Pareto sense). Can any particular action fail to achieve its particular end efficiently in this sense? I conclude that either no action can, in which case (U) is vacuous; or else the concept of efficiency is inconsistent. In the single end interpretation of (U), we rely on an implicit ceteris paribus clause, by evaluating the rationality of an action in the service of one particular end, assuming all others to be fixed. Conventionally, this interpretation finds expression in questions as to whether a particular action is the most efficient way to achieve a given end. According to the pedestrian version of the concept of efficiency, we achieve such an end efficiently when 9 Allais, op. cit. Note 2, 40. Ibid., 70 and footnote 52. 10 © Adrian Piper Research Archive Foundation Berlin |
