Exploring the Sequences of Synthetic Facial Expressions and Type of Problems Solved in a Personal Instructing Agent using Lag Sequential Analysis

This study attempted to determine and visualize the sequences of synthetic facial expressions (SFE) and problems solved in a personal instructing agent named PIA of 75 grade 7 students. PIA is a pedagogical agent that exhibits SFE and provides hints as feedback while students solve an algebra problem. PIA exhibited 8,879 SFE and 83% of which was a happy SFE. Students mostly solved easy problems (63%). Lag-sequential analysis revealed that students tend to stick on the level of difficulty they are comfortable with. When PIA exhibited a happy SFE, it is more likely that the students will answer the problem correctly. However, on the other hand, when PIA exhibited a sad SFE, there is a high probability that PIA will exhibit a cycle of sad SFE, which may eventually lead to the abandonment of the problem. It can be concluded that sequences of SFEs can be utilized to detect students persistence (or lack thereof) in a PIA and can be utilized to improve the software. Design implications to personalization of a learning environment are recommended.