Currently, I am a PhD student in the Psychology department at Carnegie Mellon University where I am advised by Dr. John Anderson. I am a member of the Program for Interdisciplinary Education Research as well as the Pittsburgh Science of Learning Center.

Before coming to CMU, I worked at Stanford University in the Cognitive Systems Neuroscience Lab with Vinod Menon. As a research assistant at SCSNL, I led a large-scale intervention study investigating the effects of tutoring on children with developmental dyscalculia.

I completed my B.A. at University of Texas, Austin where I majored in Psychology and Plan II Honors. Here I worked with Dr. David Schnyer.

As the U.S. fights to raise student's math scores, an industry in educational technology has emerged with the promise of a tutoring tailored to the individual student. Most of these systems only look at student's correctness, and could be improved by instead focusing on the strategies students form to solve math problems.

[maybe put a graphic here of an efficiently solved problem and an inefficiently solved problem]

My research focuses on ways to accurately detect what strategies students use to solve various types of math problems. To do this I use both student's behavior and the signals from their brains to model and predict the strategies used to solve math problems. (See my Projects for more detail)

Using brain imaging provides us valuable insight into the process of learning. While it is unlikely this technology be introduced to classrooms (at least in the near future), brain imaging can provide a solid foundation on which to build models of student learning that can inform how computerized tutoring systems assess student understanding.

Currently much of the educational technologies available have cool designs and great graphics, but questionable educational effectiveness. As we begin to take a critical perspective of these tools, neuroscience and data mining can help provide a more accurate measure of student learning which in turn can improve these technologies.

Study 1

This study explores the potential for using classification analysis along with modeling to assess the strategy used to solve various math problems in the scanner. Using three imperfect indicators of strategy use (latency, retrospective self report, fMRI classification data) we built a model that predicts strategy use.

Study 2

In this study I test a previously developed model by weaving concurrent assessments into the fMRI task. I extend the model to explore how successfully I can predict the change in strategies as people learn a set of math problems. In particular I am exploring several means of optimizing machine learning algorithms to better detect strategies.

Interdisciplinary Independent Project

This research is focused on extending my strategy detection work from simple single step problems to more complex multi-step problems. This fall I have focused on my PIER IIP project, collecting linear problem solving data from over 150 middle and high school students. My PIER IIP project focuses on using behavior data to develop a technique for identifying strategies in multi-step linear equation solving problems.

  • Tenison C., Fincham J., Anderson J. (2014). Detecting math problem solving strategies. An investigation into the use of retrospective self-reports, latency and fMRI data. Neuropsychologia, 54, 41-52.
  • Supekar K., Swigart A., Tenison C., Jolles D., Rosenberg-Lee M., Fuchs L., Menon V., (2013). Neural predictors of individual differences in response to math tutoring in primary-grade school children. PNAS, 110(20), 8230-8235.
  • Ashkenazi S., Rosenberg-Lee M., Tenison C., Menon V. (2011). Weak task-related modulation and stimulus representations during arithmetic problem solving in children with developmental dyscalculia. Developmental Cognitive Neuroscience, 2, 152-156.