Hi!

I’m a second year Master’s in Computer Science student at University of Michigan. I work as a research assistant in the Cognitive and Data Science Lab at Rutgers Newark. My interests lie in computational modeling of cognition and creative domains including music.

Before joining the lab, I did my undergrad at Michigan, studying cognitive science and french horn in the music school. There I worked with Professor Taraz Lee in the Cognition, Control, and Action Lab and wrote my honors thesis on music and language.

publications/past projects

Inferring knowledge from behavior in search-and-rescue tasks

Yang, S. C.-H., Anderson, S., Wang, P., Rank, C., Folke, T., Shafto, P. (2021)

In DARPA's ASIST project, we designed a model of machine theory-of-mind in a complex Minecraft search-and-rescue task. I led software development of our utility model. The model inferred subjects' prior map knowledge from movement.

Explainable AI for medical imaging: explaining pneumothorax diagnoses with Bayesian teaching

Folke, T., Yang, S. C.-H., Anderson, S., & Shafto, P. (2021) doi

We generated explanations for a neural classifier's predictions in a medical domain. Our method selects explanatory examples of predictions via a model of pedagogical reasoning. I developed an human experiment interface for evaluating explanations in JavaScript.

A linguistic model of minimalist syntax composes Tebe poem

Anderson, S.P. (2020) doi

For my undergrad honors thesis, I reformulated a simplified model of Minimalist syntax to write Western Tonal chord progressions. This resulted in a better parse of Bortniansky's Tebe Poem. It was the first (to our knowledge) programmatic simulation of Minimalist ideas in music composition.

Rewards interact with explicit knowledge to enhance skilled motor performance

Anderson, S. P., Adkins, T. J., Gary, B. S., & Lee, T. G. (2020) doi

When we learn movement skills (such as riding a bike), we develop both explicit “conscious” knowledge of our movements (the sequence of letters in a password) as well as implicit, “unconscious” knowledge (quickly typing the password with muscle memory). When we’re more motivated, we perform these movements faster and more accurately. In our experiments we demonstrate that increased motivation additionally enhances the ability to plan movements ahead, enabled by explicit knowledge.

ongoing projects (coming soon)