Optimizing Human Performance using Individualized Computational Models of Learning
Proceedings of the Eighth Annual Conference on Advances in Cognitive Systems
2020
Abstract
A key challenge for human performance optimization designers is cost effectively evaluating differ- ent interventions. Typically, A/B experiments are used to evaluate interventions, but running these experiments is costly. We explore how computational models of learning can support designers in causally reasoning about alternative interventions for a fractions tutor. We present an approach for automatically tuning models to specific individuals and show that these individualized models make better predictions than generic models. Next, we apply these individualized models to gen- erate counterfactual predictions for how two students (a high and a low-performing student) will respond to three different fractions training interventions. Our model makes predictions that align with previous human findings as well as testable predictions that might be evaluated with future human experiments.
BibTeX
@inproceedings{maclellan-acs-2020,
title = {Optimizing Human Performance using Individualized Computational Models of Learning},
author = {MacLellan, Christopher J. and Stowers, Kimberly and Brady, Lisa},
booktitle = {Proceedings of the Eighth Annual Conference on Advances in Cognitive Systems},
year = {2020},
}