Microworlds based on linear equation systems: A new approach to complex problem solving and experimental results

Abstract

The method of computer-simulated scenarios has recently been introduced to study how people solve complex problems. This paper describes a special approach to constructing such microworlds by means of linear structural equation systems. Subjects' task in the experimental situation is to first identify in a knowledge acquisition phase the causal structure of an hitherto unknown system. In a later knowledge application phase they have to control this system with respect to a given goal state. Knowledge that was acquired on the task is assessed both by means of causal diagrams - a method developed within this project and proven to be very useful - as well as by the degree of successful control performance. Three experiments on special attributes of such systems (active interventions versus observations only, effects of different degrees of Eigendynamik, the influence of different degrees of side effects) illustrate the approach. The menioned factors have considerable influence on identification and control of the system SINUS. The conclusion deals with the advantages of an experimental approach in this area.