MOODS: A prototype tutoring system that detects students' motivation

Previous work concerned with motivation in Intelligent Tutoring Systems (ITS) has mainly focused on the strategies that an ITS could use to motivate the student. In this paper we focus on the prior task of detecting the student’s motivational state, on which such strategies could be used. Detecting the student’s motivational state is not straightforward, since a lot of information about the student is available to a tutoring system during an instructional interaction. In this paper we discuss the advantages and disadvantages of different methods of motivation detection. We also present a couple of empirical studies that we performed in order to inform the motivation detection strategies of MOODS, a simple ITS prototype, which we also present in this paper. The approach taken for the development of MOODS is based on three main aspects: 1. The separation of the student model into two parts: one updated by the student and one updated internally by the system. 2. The separation of the motivation model into: (a) A ‘motivation state’ part, (b) A ‘motivation trait’ part. 3. The use of language as a communicator of affect. While a formal evaluation of MOODS is yet to be done, we believe that the approach presented in this paper for motivation detection allows for the development of systems which interfere very little with student’s interaction, but that at the same time are able to detect a number of situations in which motivational problems arise.