How can an educational chatbot's feedback influence human attention?

Abstract

Educational chatbots have been shown to be useful assistants in computer-supported learning settings. However, how does feedback of an educational chatbot impact on the learner’s attention? Thus, this paper proposes a study to measure changes in human attention when learning by using an educational chatbot Liza that is intended to improve human reasoning ability. In total, 18 participants participated in the study and had a conversation with Liza. During the interaction with Liza, the attention of the study participants was measured using a mobile electroencephalogram (EEG) device. Three findings have been determined based on statistics methods. First, it was found that there was a significant attention effect occurring in 54% of the times, after the educational chatbot showed feedback and the attention measurement took place over the length of a task. Second, when differentiating the type of feedback, positive feedback had a significant effect in 71 of these 137 cases (51.82%) and negative feedback had a significant effect in 66 of the 137 cases (48.18%). Third, statistical results showed that there was no significant difference in attention at the significant level of 0.05 during the 10 seconds before, and 10 seconds after positive feedback is received. Similar is the case for negative feedback.