Cognitive Tutor

Improvements in cognitive modeling of a computer agent through direct observation of user interaction and data mining
Cognition
Process Data
R
Oracle
Minerva
Cygwin
NFS
IES
Author
Affiliation

Carnegie Mellon University

Published

June 9, 2014

Investigating the Effect of Meta-Cognitive Scaffolding for Learning by Teaching

Matsuda, N., Griger, C. L., Barbalios, N., Stylianides, G., Cohen, W. W., & Koedinger, K. R. (2014)

This paper investigates the effect of meta-cognitive help in the context of learning by teaching. Students learned to solve algebraic equations by tutoring a teachable agent, called SimStudent, using an online learning environment, called APLUS. A version of APLUS was developed to provide meta-cognitive help on what problems students should teach, as well as when to quiz SimStudent. A classroom study comparing APLUS with and without the meta-cognitive help was conducted with 173 seventh to ninth grade students. The data showed that students with the meta-cognitive help showed better problem selection and scored higher on the post-test than those who tutored SimStudent without the meta-cognitive help. These results suggest that, when carefully de-signed, learning by teaching can support students to not only learn cognitive skills but also employ meta-cognitive skills for effective tutoring.

My primary responsibilities on the SimStudent project concerned improvements in cognitive modeling of a computer agent through direct observation of user interaction and data mining.

  • Pre-process output data from Oracle database to discover patterns and bugs within the production files

  • Conduct descriptive and inferential analyses, and produce data visualizations for PI team on monthly basis

  • Analyze reliability on measures of human student knowledge gains

  • Examine cognitive fidelity by disentangling computer agent errors that impact human student’s learning and validating production rules using Minerva via Cygwin

  • Manage projects through recruiting stakeholders, training research assistants, create and maintain IRB documentation, scheduling and managing classroom and lab studies, data analysis, journaling all procedures 

  • Worked closely with programmer to fine tune production rules and interface based on analyses