SFB 303 Discussion Paper No. B - 363

Author: Tang, Fang-Fang
Title: Anticipatory Learning in Two-Person Games: An Experimental Study Part II. Learning
Abstract: "Learning" can be viewed as any systematic change of behavior due to experience accumulation. A learning model, following the probabilistic approach of Bush and Mosteller (1955), is a mathematical system which predicts the probabilities of available choices or feasible actions at the next occurrence, based on all or part of the historical information. Mathematically some stochastical process must be defined explicitly, which we will do so for a number of models in the subsequent sections of this part. Nevertheless, our main interest concentrates on the empirical side of these models, viz. the extent of their predictive success for our experimental data. This paper is organized as follows. In Section 2.2, we will formally lay out all the learning models we have tested, and discuss some inter-relationships among them. In Section 2.3, numerical results on the predictive success of these learning models under the quadratic deviation measurement will be shown and interpretations or implications of these results will be discussed. The main findings of an analogous investigation under the absolute deviation measurement will be briefly reported as well. We conclude in Section 2.4.
Creation-Date: March 1996
Unfortunately this paper is not available online. Please contact us to order a hardcopy.

SFB 303 Homepage

26.10.1999, Webmaster