Emotional Rollercoasters: Day Differences in Affect Incidence during Learning.
Sidney K. D’Mello · The Florida AI Research Society · 2014
This study investigated between-session variability in affective state incidence during learning with an intelligent tutoring system (ITS). We tracked 12 affective states (plus neutral) while three learners completed their statistics coursework in eight 45-minute weekly sessions with the ALEKS ITS. The results indicated that engagement/flow was the most frequent affective state that demonstrated considerable stability across sessions. Nine other affective states (anger, anxiety, boredom, contempt, curiosity, disgust, eureka, happiness, surprise) were rarely observed both within and across sessions. Confusion and frustration occurred with some frequency but there was considerable between-session variability in the incidence of these states. Followup analyses indicated that some of the between-session variability could be explained by initial (before session) positive and negative mood and initial physiological arousal (electrodermal activity). Implications of the findings for basic research on affect and learning and for affect-aware ITSs are discussed.