Understanding Individual Differences in Learning Strategies, Cognitive Characteristics and Task Demands: One-Size Does Not Fit All but Tailoring is Difficult

Abstract

This dissertation examined the fundamental aspects of skill learning, exploring three key facets: the use of different strategies by learners, the stability of these strategies over time and with varying task requirements, and relating latent cognitive characteristics to learning success or strategy. Idiographic computational models and parameter estimation based on the ACT-R cognitive architecture were used to identify strategies and estimate latent cognitive characteristics. Results showed that participants fit diverse memory strategies, with idiographic modeling proving to be crucial in uncovering these differences. However, a declarative long-term memory (LTM) strategy best described most participants. Taking into account that strategies might change across time or in response to changing task demands, a second experiment investigated individual dynamics and learner-task interactions. Here, the learning task was split into 2 time epochs and fit individually, to test if different models explained behavior at different time points. For the simple stimulus-response task that was used, most participants were best fit by the LTM strategy throughout the task, but increasing task difficulty did not effect consistent changes in strategies. Lastly, a third experiment sought to robustly estimate model parameters that were historically related to cognitive characteristics vital to learning: memory decay rate and working memory capacity. The goal after estimation was to test how predictive these are of learning outcomes in multiple cognitive tasks and strategies. The study concluded that parameter values serve as reliable measures of individual cognitive characteristics only within specific models or contexts and not across tasks.