Creative Commons

Peoples’ behavior is of significant interest in the field of Human-Computer Interaction because it could reveal information about how human behave when they interact with computers. Mouse tracking, according to previous HCI research, can offer a full overview of human behavior under advanced intellectual loads such as a decision or the development of an activity. However, there is a scarcity of learning research. In this paper we look at the possible correlations among mouse movement metrics and users’ emotions when they are interacting remotely with a game-based learning (GBL) task. Towards this goal, we conducted an experiment on 33 participants who completed a GBL quiz task after watched a video course about physics. A JavaScript mouse monitoring mechanism was embedded in the game to track the mouse movements events in real time and store them in JSON files. A set of behavioral and dynamic mouse features was extracted to measure i) average mouse speed, ii) average acceleration, iii) average time between movements, iv) total count of movements, v) total count of time between movements short pauses, vi) total count of time between movements long pauses, vii) count speed number of pauses, viii) task completion, ix) time count movements/task completion time, x) count speed=0/task completion, xi) count pauses>2/task completion, xii) count pauses>5/task completion time. A self-reported questionnaire was used to measure the participants’ perceived emotions of i) self-efficacy, ii) engagement, iii) immersion, iv) enjoyment, v) confusion, vi) frustration, vii) stress, vii) dissatisfaction during playing the GBL task. Τhe results of the experiment revealed the existence of some significant relationships between users’ emotions and mouse features. In particular, the following significant correlations was found: i) the variance of time between movements is significantly correlated with frustration, ii) engagement of users during the game-based learning task significantly associated with mean acceleration, iii) age has a significant correlation with total count of movements v) age has a significant correlation with the count speed=0 (number of pauses), vi) age has a significant correlation with total count of time between movements (short pauses), vii) a significant correlation was founded between level of familiarity and self-efficacy, vii) a negative significant correlation was founded between level of familiarity and the count speed=0(number of pauses). The findings of this paper can reveal an interesting new research direction and may motivate the HCI and GBL fields of study to search further the user’s cursor movement behaviors when interacting with a game-based learning environment, since this method has recently been widely adopted in the education filed, due to the COVID-19 situation.