Publication Date


Document Type


Committee Members

Ion Juvina, Ph.D. (Advisor); Valerie Shalin, Ph.D. (Committee Member); Herbert Colle, Ph.D. (Committee Member); Tiffany Jastrzembski, Ph.D. (Committee Member)

Degree Name

Doctor of Philosophy (PhD)


Mathematical models of learning and retention have long been developed in psychology for both basic and applied research. For basic research, models of learning and retention attempt to explain how individuals acquire and retain information over time. While for applied research, models of learning and retention are used to inform education and training decisions. In both of these applications, the primary purpose of using a model is to fit and predict the performance of individuals. However, little attention has been paid to the interpretation of a model’s free parameters (i.e., learning and decay rates) and the effect that a model’s formulation has on the inferences that can be made from a model’s estimated parameter values. Particular model formulations, such as having units on different terms or nested equations, can lead to correlations between a model’s free parameters, which limit their theoretical interpretation. These limitations have been identified in one model of learning and retention in particular, the Predictive Performance Equation (PPE). In this dissertation, I propose and evaluate an alternative version of the PPE (Modified PPE), which simplifies and decreases the correlation of its free parameters compared to the Standard PPE. A large-scale model comparison was conducted across nine different historical datasets, comparing both models’ ability to fit, predict behavior, and make inferences about particular characteristics of an individual's learning behavior from each model’s estimated parameters. The results of this dissertation show that the Modified PPE when compared to the Standard PPE can both calibrate and predict to a similar degree as the Standard PPE. However, the Modified PPE decreases the correlation between its free parameters, allowing parameters to map onto particular aspects of performance across a majority of datasets. These results show how models of learning and retention can be improved to allow for a more meaningful interpretation of an individual’s performance, having implications for both psychological theory and applied application.

Page Count


Department or Program

Department of Psychology

Year Degree Awarded