Modeling Bounded Rationality And Researches On Some Satisficing Decision Theory Problems

Simon’s "bounded rationality" and "satisficing decision" broadened the research field of classical decision theory, and laid the foundation of modern decision theory. Bounded rationality theory held that complete rationality in the rational decision making models did not exist in reality. Decision makers’behaviors were not only confined by external factors such as time, information and technology, but also restricted by psychological factors of decision makers themselves. Therefore, the optimal criterion should be replaced by the satisficing criterion. Many researchers tried to modify and improve the rational decision making models by axiomatic modeling, experiments of behavioral economics and so on. They expected to establish a set of theories and methods which can be used to describe the decision maker’s actual behaviors in reality.In this thesis, the concepts of bounded rationalities are defined, and individual internal rationalites are modeled by mathematical logic, fuzzy mathematical, etc. on the one hand. The bases of rationalities of satisficing decisions which are compatible with "complete rationalities" and "optimization" in classical decision theory are discussed on the other hand. Specifically, the works of Tyson and Rubinstein are reviewed and the relevant model is expanded in this thesis. On the one hand, the actual decision problem such as "The more alternatives, the more difficult the choice" in reality is explained and illustrated. On the other hand, the concepts about internal preferences and revealed preferences are presented, and individual internal rationalities are modeled based on the logical consistency relations of internal preferences and revealed preferences. The existence and features of compatible decisions of ordinal preferences are analysed by some conditions such as revised Weak Expansion, Auxiliary Expansion and Strong Expansion. The "satisficing heuristic" which Simon proposed is analysed and modeled. A satisficing decision function is defined, and its existence and rationalities are discussed. A class of irrational choice behaviors based on the satisficing decision function are explained and illustrated. In addition, aiming at the fuzziness of individual preferences and choices in reality, the complete rationality and bounded rationality of a fuzzy decision function are defined. And especially, the bounded rational characteristics of the fuzzy decision function are discussed.In application, the thesis gives two application cases. One is, aiming at the features of the consumer with different rationalities showed in the process of purchase decision-making, the quality strategies are analyzed and discussed in quality competition of power industry materials. Moreover, there is an assumption that the purchaser with complete rationality is in pursuit of power industry materials with the high quality, and the purchaser with bounded rationality is in pursuit of the satisficing ones. The other is that aiming at the non-optimal behaviors of a decision maker in reality, the rationality hypotheses of secretary problem are weakened and a kind of bounded rational ideas and satisficing decision methods are applied to study the secretary problem. Moreover, the possibility degree of the decision-maker to obtain satisfactory alternatives is given.The results of this thesis can be applied to the purchase decision of a consumer, a class of sequential observation and selection etc. This thesis can provide them with a theoretical reference and guidance in product quality or price competition when firms face different kinds of consumers. It can also provide the decision maker with a satisficing decision method in case of time pressure, limited information, analysis and judgment by experience and intuition, etc. The results of this thesis are expected to obtain a base of the rationality of satisficing decisions which are compatible with classical decision theory. Moreover, they can be applied to explain and analyse some paradox phenomena combined with bounded rational models. They have certain significance and academic value to understand economic phenomena and deal with decision problems effectively in reality. Moreover, the decision making with incomplete information, fuzzy decision-making, etc can be inspired by this thesis.