Jun Fang

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)




Lael Schooler


ACT-R;Decision Making;Heuristics;Knowledge transfer;Machine Learning

Subject Categories

Psychology | Social and Behavioral Sciences


People often must make inferences in domains with limited information. In such cases, they can leverage their knowledge from other domains to make these inferences. This knowledge transfer process is quite common, but what are the underlying mechanisms that allow us to accomplish it? Analogical reasoning may be one such mechanism. This dissertation explores the role of analogy in influencing decision-making performance when faced with a new domain. We delve into the knowledge transferred between tasks and how this influences decision-making in novel tasks. Experiment I has two conditions, and each condition has two tasks. In one condition, the two task domains are analogically related, where for example, participants make inferences first about water flow and then about heat flow. In the second condition, the domains do not share obvious similarities. For example, car efficiency and water flow. Experiment I shows that participants presented with an analogy demonstrated better performance than those without. We hypothesize that this knowledge transfer occurs in two ways: firstly, analogical mapping enhances comprehension of cue utilization in a new task; secondly, the strategy employed is transferred. In Chapter 3, we developed a machine learning technique to uncover the strategies used by participants. Our findings reveal that the best-performing strategy from the old task is typically carried over to the new task. In Chapter 4, we developed a model of analogical transfer in multi-attribute decision making. We use the ACT-R theory of cognition as a framework to model knowledge transfer by integrating a reinforcement learning model of strategy selection with a model of analogy. The simulation results showcase a similar trend of both accuracy and strategy use to the behavioral data. Finally, we critically analyze our study's limitations and outline promising directions for future research, thereby paving the way for a deeper understanding of knowledge transfer mechanisms.


Open Access