NTNU

News

    Font Size:
  • L
  • M
  • S
2022.1226

Look at Me!The application of facial recognition technology in science education

 With the rapid development of facial recognition technology, the relationship between chang-es in our facial expressions and science learning is gradually being revealed and proven. Utilizing a facial expression analysis software (FaceReader) from the Netherlands, Distinguished Professor Mei-Hung Chiu of the Graduate Institute of Science Education at National Taiwan Normal Uni-versity and her collaborators have attempted to analyze the facial expressions high school and uni-versity students show as they undergo conceptual change during science learning. The exploration will allow teachers to provide timely adjustments to their instructional approach to better suit stu-dents’ needs.

Facial expression is one of the most direct ways for us to know how other people are feeling. They offer important clues for us in deciphering what others are feeling, whether it is a smile, a frown, or even just a glance. As most people show their facial expressions subconsciously and without intention, they are also a window for us to glean into people’s most honest and unfiltered feelings. Consequently, the analyses of students’ facial expression states at the moments of their learning could be a way for us to better understand how they learn and the process of learning it-self. 

Professor Mei-Hung Chiu and her team have collected and analyzed facial expressions changes high school and university students in Taiwan showed as they react to different scientific phenomena. In particular, Professor Chiu studied the reactions students showed when they saw scientific experiment outcomes that contradicted their expectations. Students’ facial states were analyzed using facial expression analysis software, and compared against the facial states students exhibited during the subsequent science learning process. It was found that there are unique pat-terns in students’ microexpression states (i.e., facial expressions that last no longer than 0.5 sec-onds), meaning that these fleeting microexpression states may be used to identify students’ level of understanding, which in turn may inform teachers on when and how to provide customized inter-vention strategies for different students as they plan for their classes. 

In their study, Professor Chiu’s research team adopted a counterintuitive science demonstra-tion as a way to trigger conceptual change. In the demonstration, water in a flask was brought to a boil. The flask was then removed from the heat source, sealed, and inverted. The air pressure in-side the flask was then lowered by the placement of a pack of ice on the inverted flask, causing the water inside to boil again. The resultant heat-free re-boiling phenomenon triggered students’ ques-tioning of their existing understanding of pressure and boiling point. Consequently, it was found that students exhibited many microexpressions as they watched the demonstration, and there is a strong relationship between students’ microexpression states and conceptual change. This finding was repeated in demonstrations of other counterintuitive scientific phenomena, indicating that the finding of the study is not limited to a single phenomenon or a single scientific concept.  

A statistical decision-tree analysis also revealed that among the six universal facial expressions (namely, happy, surprise, disgust, sad, angry, and scared), anger, sadness, and disgust are the three expression states that may predict if a student would undergo conceptual change. In contrast with past studies where positive facial expressions are found to be associated with student learning, it was shown that it was the negative facial expression states that are the key indicators of scientific conceptual change among students. Further analyses also revealed that the way scientific materials are presented also have an impact on the prediction of student learning. For example, using anima-tion to explain scientific concepts is a better predicting indicator than using text or experiment vid-eo. Moreover, textual explanation is better at conveying conceptual concepts than experiment videos.