Emotions are present in any form of learning and social interaction. The realistic use of the emerging educational technologies has drawn attention to the consideration of emotions in learning environments. Towards this direction, Affective Computing is offering remarkable system implementations that detect and recognise students’ emotional states with high accuracy using machine learning algorithms, and provide feedback aiming at both students’ cognitive performance and their emotional regulation.

However, these technologies often employ expensive sensors and complex computer intelligence that require special expertise or extra resources, while they introduce obtrusiveness and invasiveness in the learning process. Moreover, research on emotions in e-learning is still rather limited while the enrichment of learning environments with emotion awareness capabilities it is still in its infancy. There is still a need for more realistic, in-context studies to investigate successful affective learning sequences that propel students’ self-motivation and engagement.

The main objective of this thesis is to investigate the importance of emotion awareness in e-learning environments. To this end, a conceptual model has been developed, including affective states and moods of interest that usually appear in e-learning settings with emphasis in Computer Supported Collaborative Learning (CSCL) scenarios. In the basis of this model, a computational model has been implemented consisting of a usable, expressive and effective multimedia interface for students to report their affective state and an affective virtual agent, employed with expressive faces to provide affectuve and task-based feedback in response to the students’ reported emotions. The integration was also enhanced with effective visualisations of students’ individual and group affective states.

Both models have been tested in real education settings by conducting experiments with university students in Virtual Learning Environments. A qualitative and quantitative analysis of the results suggests 12 final affective states and 5 moods of interest in e-learning environments, raising, however, a speculation considering the adequacy and the potential of verbal or pictorial labels, to express the respondent’s exact feelings. Patterns of affective sequences have been also identified, which that can be exploited by computer intelligence to provide effective emotional scaffolds.

With respect to self-reporting of emotions, it was found that students are willing to participate and express their affective state, once a tool provides an easy and usable way for them to do it, improving in that way their engagement. In response to their emotions’ sharing, students need to see an adaptation from the system immediately or after a very short period of time. The provision of affective feedback, enriched with task-oriented scaffolds, can improve student cognitive performance and emotion regulation, at least when it is human guided. Group emotion awareness also appears to improve social interaction.