Machine Learning and Human Learning: 
                 Learning by Observing and Explaining

                          Michael Redmond
                         Computer Science
                        Rutgers University
                         Camden, NJ 08102
                  E-mail: redmond@crab.rutgers.edu
                 
                             Abstract


Cognitive-based Artificial Intelligence attempts to create "intelligent" 
behaving programs through analysis of how people carry out the task in
question. A successful model may suggest testable predictions about human 
cognition or thinking. In earlier work, we created a model, called CELIA, 
which is a computational model of how a novice student can quickly become 
competent at a procedural task. It accomplishes this through observing and 
understanding an expert's problem solving. This model  was inspired by 
human protocol studies, and was implemented in a computer program. This 
model of a student's effective learning suggests some implications for
teaching novices in a new domain; the same strategies that allow a cognitive
-based program to learn effectively may also be useful for people. These  
implications may be relevant for both human teaching and intelligent
tutoring. The main implications include: encourage the student to predict 
prior to observation, present example steps in an interactive step-by-step 
manner, encourage self-explanation by the student, and allow flexible 
interaction with the student. These implications represent hypotheses that 
follow from the learning model; they suggest further research. We have
developed a computer-based tutor, that is based on some of the techniques 
from CELIA, including prediction, iterative observation, and self-
explanation. Initial pilot testing suggests that this is an effective 
tutoring approach.