Maik Friedrich (Master-Colloquium): Implementing diagrammatic reasoning strategies in a high level language: Extending and testing the existing model results by gathering additional data and creating additional strategies
This thesis builds upon a cognitive model and a study by Ritter and Bibby (2008). This existing learning model called Diag was implemented in Soar 6 and is based on the idea that learning consists of procedural, declarative, and episodic learning. Despite the high correlation between observed and predicted performance there were participants whose behavior could not be predicted. Therefore, the main goal of this thesis is to understand the strategies humans use while doing a task of reasoning and learning, and to use this knowledge to predict human behavior. For this purpose, the model was reimplemented in a high level language, Herbal, and compiled into Soar 8. An additional user study was run to gather enough data for the analysis of different strategies and strategy shifts. It could be shown that participants created different strategies to solve the fault-finding task. Based on the observations, four additional strategies for the diagrammatic task were identified and implemented in strategy models. These strategy models were validated by comparing their predictions to the observations. As a result, it could be shown that participants not only created different strategies but some also shifted between strategies when solving the fault-finding task.