Affect-Aware Adaptive Tutoring Based on Human–Automation Etiquette Strategies. (June 2018)

• Record Type: Journal Article
• Authors: Yang, Euijung; Dorneich, Michael C.
• Is Part Of: Human factors. Volume 60:Number 4(2018)
• Page Start: 510
• Page End: 526
• Publication Date: 2018-06
• Subjects: adaptive automation -- affective factors -- intelligent tutors -- human–computer interaction -- etiquette; Human engineering -- Periodicals; 620.82
• DOI: 10.1177/0018720818765266
• Languages: English
• ISSNs: 0018-7208

Abstract

Objective: We investigated adapting the interaction style of intelligent tutoring system (ITS) feedback based on human–automation etiquette strategies. Background: Most ITSs adapt the content difficulty level, adapt the feedback timing, or provide extra content when they detect cognitive or affective decrements. Our previous work demonstrated that changing the interaction style via different feedback etiquette strategies has differential effects on students' motivation, confidence, satisfaction, and performance. The best etiquette strategy was also determined by user frustration. Method: Based on these findings, a rule set was developed that systemically selected the proper etiquette strategy to address one of four learning factors (motivation, confidence, satisfaction, and performance) under two different levels of user frustration. We explored whether etiquette strategy selection based on this rule set (systematic) or random changes in etiquette strategy for a given level of frustration affected the four learning factors. Participants solved mathematics problems under different frustration conditions with feedback that adapted dynamic changes in etiquette strategies either systematically or randomly. Results: The results demonstrated that feedback with etiquette strategies chosen systematically via the rule set could selectively target and improve motivation, confidence, satisfaction, and performance more than changing etiquette strategies randomly. The systematic adaptation was effective no matter the level of frustration for the participant. Conclusion: If computer tutors can vary the interaction style to effectively mitigate negative emotions, then ITS designers would have one more mechanism in which to design affect-aware adaptations that provide the proper responses in situations where human emotions affect the ability to learn.