An Empirical Assessment of Comprehension Fostering Features in an Intelligent Tutoring System

This paper describes the design and evaluation of two features in an Intelligent Tutoring System designed to facilitate a deeper conceptual understanding of domain principles in conjunction with the development of procedural skills. The first feature described here relates to the timing of feedback. Some researchers have argued that immediate corrective feedback, as embodied in many cognitive tutors, can block the exercise of activities that may enable students to gain a deeper conceptual understanding of a domain. These include self-monitoring, error detection, and error correction skills. We compare an immediate feedback tutor with a tutor that allows students to reflect on problem solving outcomes, and engage in error detection and correction activities. The other feature reported here is a component of declarative instruction. We assess the use of Ex-ample Walkthroughs as a comprehension-fostering tool. Prior to procedural practice, Example Walkthroughs step students through the study of example problems and guide them to reflect on the reasoning involved in going from a problem statement to a solution. An evaluation has shown that the best learning outcomes were associated with a combination of immediate feedback and Example Walkthroughs. There are indications that a combination of lower cognitive load during procedural practice and a robust and accurate encoding of declarative concepts contributed to the observed outcomes.