Online cognitive training platforms such as Lumosity support the investigation of skill learning and cognition at an unprecedented scale, as part of a trend to use naturally occurring large-scale data sets to develop and test theories of cognition. Steyvers will discuss two projects that use a combination of cognitive modeling and basic machine learning approaches to understand learning across tasks as well as learning within particular tasks. In the first project, he analyzed the practice effects of 36,000 users across 51 tasks (involving over 50,000,000 gameplays). To establish baseline results, he applied probabilistic principal component analysis to infer latent learning factors that capture not only covariation in performance across tasks but also across different levels of practice. He assessed model performance in a task where the goal is to predict how well a user will learn a novel task (at various levels of practice) on the basis of performance on other tasks. In the second project, he developed a cognitive model of practice effects in a specific task switching game that requires participants to flexibly and efficiently adapt behavior in response to continuously changing contextual demands. The model of task switching includes latent measures of activating the relevant task, deactivating the irrelevant task, and making a decision. While long-term practice improves performance across all age groups, it has a greater effect on older adults. Indeed, extensive task practice can make older individuals functionally similar to less practiced younger individuals, especially for cognitive measures that focus on the rate at which task relevant information becomes available.