We investigate how positive, neutral and negative feedback responses from an Intelligent Tutoring System (ITS) influences learners‟ affect and physiology. AutoTutor, an ITS with conversational dialogues, was used by learners (n=16) while their physiological signals (heart signal, facial muscle signal and skin conductivity) were recorded. Learners were asked to self-report the cognitive-affective states they experienced during their interactions with AutoTutor via a retrospective judgment protocol immediately after the tutorial session. Statistical analysis (Chi-square) indicated that tutor feedback and learner affect were related. The results revealed that after receiving positive feedback from AutoTutor, learners mostly experienced „delight‟ while surprise was experienced after negative feedback. We also classified physiological signals based on the tutor‟s feedback (Negative vs. Non-Negative) with a support vector machine (SVM) classifier. The classification accuracy, ranged ...
Payam Aghaei Pour, M. Sazzad Hussain, Omar AlZoubi