PACT Center

    Current Research Projects | Completed Research Projects

 
 

Pittsburgh Science of Learning Center (PSLC)

 

The Pittsburgh Science of Learning Center (PSLC) will address the widely recognized problem that education research has not yet produced research results that unquestionably work to produce robust student learning and that survive the transition from the lab to the classroom and from one domain, researcher, or developer to other domains, researchers, and developers. Although learning science has no shortage of rigorous laboratory results and associated principles and no shortage of realistic classroom design experiments and associated course innovations, there is a shortage of experimental results that are both rigorous and realistic. Our goal is to produce results that survive rigorous experimentation with laboratory-quality methods in real classroom settings.

 

 


Ken Koedinger

Kurt VanLehn (Pitt)

And over 31 other faculty and researchers from 10 different departments at Carnegie Mellon University (CMU), University of Pittsburgh (Pitt), and Carnegie Learning, Inc.

         
 

Rapid Development of Cognitive Models and Tutors

 

Cognitive Tutors have been successful in raising students' math test scores in high school and middle-school classrooms, but their development requires considerable time and expertise. The Cognitive Tutor Authoring Tools (CTAT) project focuses on developing a set of authoring tools to make modeling both easier for experts and possible for novices in cognitive science. The tools draw on ideas of programming by demonstration, structured editing, and others. Careful application of HCI methods is key.

 

 


Ken Koedinger

Vincent Aleven

Neil Heffernan

         
 

Tutoring at the Explanation Level

 

The Geometry Explanation Tutor, a 3rd-generation computer tutor, not only helps students to solve problems, as 2nd-generation intelligent tutoring systems do. It also engages students in natural language dialogue to help them state general explanations of their problem-solving steps. We will evaluate whether this helps students learn with greater understanding

 

 

Vincent Aleven
Octav Popescu
Ken Koedinger

         
 

SmartLab Statistics Tutor

 

An important challenge in education is improving students' ability to transfer their knowledge and skills outside the original learning context. This project will develop components of an innovative, intelligent learning environment aimed directly at facilitating students' transfer.

 

 

Marsha Lovett
Ken Koedinger

 

     
 

MERGE-AI

(Mastery in Early Representation Generation And Interpretation)

 

MERGE-AI, an intelligent tutor designed for use in home-schools and other non-traditional intelligent tutoring contexts, teaches students to generate and interpret scatterplots. Within this project, we are investigating how to deal with inappropriate transfer of pre-existing knowledge, and how to focus student time on the most difficult sub-skills of a multi-skill process without sacrificing overall problem structure.

 

 

Ryan Baker

Albert Corbett

Ken Koedinger

Mike Schneider

 

     
 

Statistical Techniques For Computational Models of Cognition

 

The PACT Center has worked with computational models of cognition since the very beginning of our efforts with cognitive tutors. We are designing methods to allow researchers to use statistical techniques to analyze cognitive models, using both automatic derivation and approximation of the mathematical equations underlying the computational models.

 

 

Ryan Baker

Ken Koedinger

Ben MacLaren