Designing Affective Support to Foster Learning, Motivation and Attribution

Our work involves the design, implementation and evaluation of Affective Learning Companions, real-time computational agents that infer students’ emotions and related states and leverage this knowledge to improve educational outcomes. In this paper, we focus on describing the affective support delivered by these learning companions. To date, unfortunately, there do not exist prescriptive theories on how to design the structure of affective support or how it should be tailored to suit individual student needs, even though these factors are critical in shaping learning and affect-related outcomes, making the design of affective support challenging. Our approach in addressing this challenge has been to involve human tutor experts in the affective support design process. Here, we present the wide range of affective scaffolding that we designed, which includes both verbal and non-verbal interventions, the stochastic algorithm for combining various affective interventions into complex messages aimed at fostering motivation, attribution and positive affect, and the evaluations we are currently conducting.