ケインズとラムジー : 確率と合理性をめぐって

Abstract

It is generally agreed that Frank Ramsey's paper "Truth and Probability" (1926) was the true forerunner of both the subjective theory of probability and the utility theory, that is, the two pillars of modern Decision Theory. But it was originally written as a thorough criticism of John Meynard Keynes's A Treatise on Probability (1921). The present paper is a historical and analytical reconsideration of this exchange concerning probability and rationality between Keynes and Ramsey. In this paper, I not only try to scrutinize the minute points of the arguments and threads of reasoning on both sides by reading these works together with some other materials (such as Ramsey's early criticism of Keynes and Keyns's access ment of Ramsey's theory after the latter's untimely death), but also try to make sense of this exchange in the light of the present theoretical situation in Decision Theory. I would like to argue the following points in this paper; (1) The usual textbook characterisation of Keynes's and Ramsey's positions in probability theory as, respectively, "logical" and "subjective" theories is misleading. Keynes's so called "logical" theory has only superficial affinity to Carnapean logicist analysis of probability. Rather, the true precursor of Carnap was Wittgenstein's Tractatus. On the other hand, Ramsey indeed succeeded in formulating a subjective theory of probability. But his position was not exclusively subjectivist in the way those of many modern subjectivists are. Rather, he was proposing the possibility of compativility between subjective and objective theories. In this regard, neither Keynes nor Ramsey misunderstood the other's position. (2) Keynes's "logical" theory is a curious amalgamation of a Platonistic conception of logic and that of ordinary language logicality, something like the later Wittgensteinean theory of logicality. Ramsey rejected both concepts of logic as mysterious. He replaced them with a combination of an operationalistic conception of mind and an axiomatic treatment of utility and probability. With these ideas, Ramsey saved an important element of Keynes's theory, that is, the idea of probability as the degree of partial belief. (3) However, in this process, Ramsey came very close to the opposite pitfall : He almost identified rationality with the utility of cognition. Keynes criticized this aspect of Ramsey's theory. The validity of this criticism lies in the fact that present-day Ramsey-type Decision Theory faces many kinds of paradoxes. For example, the many paradoxical experimental results concerning decision making lead psychologists Tversky and Kerneman to their "Prospect Theory". which regards the "editing phase" in the decision process as an important element. But this phase is also what Keynes tried to identify in his analysis of rationality. Again, the philosopher Robert Nozick's recent theory of rational decision tries to avoid dilemmas of standard theory by incorporating the element of the "symbolic utility" of an action. Here we find another echo of Keynes-later Wittgenstein conception of rationality in ordinary language. These examples show that the controversy between Keynes and Ramsey is still alive.