Date of Award
Spring 5-20-2022
Degree Name
Master of Science (MS)
Degree Type
Thesis
Abstract
Inferences about executive functions are commonly drawn through serial administration of various individual assessments that often take a long time to complete and cannot capture complex trends across multiple variables. In an attempt to improve upon current methods used to estimate latent brain constructs, this thesis makes two primary contributions to the field of behavioral modeling. First, it brings attention to sequential designs for more efficient diagnostic testing of fluctuations in executive functions with respect to a baseline level. It was shown that a sequential framework was successfully capable of detecting significant differences in cognitive performance more rapidly than conventional fixed approaches. Second, it introduces a scalable Gaussian Process estimator that can build individual psychometric models of task performance without requiring prohibitive amounts of data. This probabilistic machine learning classifier was capable of obtaining fully predictive models of working memory capacity person by person with high confidence.
Language
English (en)
Chair
Dennis Barbour
Committee Members
Dennis Barbour, Jacob Gardner, Jason Hassenstab