Investigation of the relation between task engagement and eye gaze

In human-computer interaction, it is important that the users sustain their engagement during cognitively demanding tasks. To that end, the system needs first to estimate the level of engagement and detect the declines and then to trigger a support mechanism to recover engagement. In that respect, many studies in the literature propose estimating engagement from facial images. Although these studies have proved that facial landmarks, especially those relating to the eyes (ocular landmarks), serve useful in the estimation of task engagement, they are subject to criticism, since it may make the user uncomfortable to be constantly watched. In addition, the video data needs to be handled very carefully due to privacy issues. In that respect, this study investigates whether it is possible to estimate the level of engagement from anonymous data, specifically from eye gaze. This sort of data is considered to be closely related to the changes in ocular landmark locations and therefore has the potential to involve similar qualities to those of ocular landmarks. In addition, since it is not possible to identify the individuals from gaze fixations, it is not privacy sensitive. Moreover, it is also expected to help relieve users' discomfort due to constant observation and also prevent modifications in their behavior due to their awareness of being observed.