- 著者
-
池田 信夫
- 出版者
- 文教大学
- 雑誌
- 情報研究 = Information and Communication Studies (ISSN:03893367)
- 巻号頁・発行日
- vol.15, pp.153-163, 1994-01-01
The concept of rationality is often quoted in the literature of social sciences, but its implication is not fully unederstood until recently. Traditionally, Hume, Smith, and many economists after the "Marginalist Revolution" has built their theories upon the instrumental rationality, i.e. utility maximization independent of other social contexts. It was an attempt at imitating classical mechanics that is based on the deterministic differential equations describing the motion of independent matters. But human behaviour cannot be exactly described by the mechanical model of neoclassical economics. Hayek rejected the concept of Walrasian equilibrium as erroneous "scientism" and Keynes refered to fallacy of composition due to the psycological interaction, which was confirmed logically by the game theory. In contrast to the neoclassical premise that adding up the individual best lead to the sosical best, game-theoretic solution concept, Nash equilibrium, does not necessarily coincide with the Pareto optimum (social efficiency). The most famous example is the Prisoner's Dilemma in which cooperation is better than defection for each player but cannot be attained spontaneously. Ironically, the players cooperate if and only if they are not rational. This paradox is rooted in the logical structure of deductive decision-making, because player A's startegy depends on B's that depends on A's ・・・ and so on. As a corollary of G?del’s famous theorem, this kind of self-referential reasoning cannot constitute self-contained system of "right" decision. In 1980's many scientific enterprise such as artificial intelligence, generative grammar, and Bayesian decision theory tried to formulate human intellect as a logically consistent system, but failed at last. Now many social scientists are re-examining Simon's concept of bounded rationality that take into account the complexity of the interaction among people. It corresponds to the new findings in the natural sciences such as deterministic chaos, self-organization and complex adaptive systems.