Detection and Control of Engagement through emotional behavior
- Claude Frasson (U. de Montréal, Theme 2)
- Ramla Ghali (U. de Montréal)
- Asma Benkhedher (U. de Montréal)
Mental engagement is related to the level of mental vigilance and alertness. It gives also a wide indication about the level of attention and motivation. Developing EEG indexes for workload assessment is an important field especially in laboratory contexts. A variety of linear and non-linear classification and regression methods were used to determine mental workload in different kinds of cognitive tasks such as memorization, language processing, visual, or auditory tasks. These methods used mainly EEG Power Spectral Density (PSD) bands combined with machine learning techniques [Berka et al, 2004, Wilson 2004, Stevens 2007]. In our previous work, we developed an EEG workload index based on Gaussian Process Regression (GPR) using data gathered from strict laboratory conditions [Chaouachi et al, 2011]. The index showed to precisely reflect users ‘workload variation in several cognitive task.
Pope and colleagues [Pope et al, 1995] at NASA developed an EEG-engagement index based on brainwave band power and applied it in a closed-loop system to modulate task allocation. Performance in a vigilance task improved when this index was used as a criterion for switching between manual and automated piloting. Performance improvement was reported using this engagement index for task allocation mode (manual or automated). In this paper, we propose to explore the behavior of these mental metrics in a learning environment with regards to a self-reported emotion. The major contribution of this project is to present the different trends in engagement and workload with regards to emotional state [Forbes-Riley et al, 2008]within an educational context.