The new research combines eye tracking with artificial intelligence to identify the exact moments of educational videos that are important for children's learning.
This study also allows children to predict how much they understand from the video based on their eye movements.
The study is preliminary, but it offers promises to some exciting breakthroughs in video education, he said. Jason Coronelassociate professor with the lead author of the study Communication at Ohio State University.
“Our ultimate goal is to build an AI system that allows viewers to know in real time whether or not they understand what they are seeing in educational videos,” Coronel said.
“It would give us the opportunity to dynamically adjust content to help individuals understand what is being taught.”
Coronel conducted research all-around Ohio with Matt Switzer, Alex Bonus, Rebecca Door and Blue Lerner, an interdisciplinary team of eye tracking, machine learning and children's media experts.
It's published today Journal of Communicationinvolved 197 children aged 4 to 8 who watched a four-minute combined video from the popular YouTube series Scishow Kids and Learn Bright.
This video taught children about animal camouflage. Eye tracking allowed researchers to measure their attention to video in real time. This is important for learning, Coronel said.
After watching the video, the children asked a series of questions and decided what they had learned about camouflage. (Before they saw, the children answered questions to assess their baseline knowledge.)
AI analysis of eye tracking results identifies points in the video and relates to whether children were able to answer questions about camouflage correctly.
For example, one important point was near the beginning when the video host asked the kids to help them find her personified sidekick, Squeaks.
“Our machine learning and gaze tracking data show that child's eye movements at this early moment are one of the most powerful predictors of overall understanding of video,” the study author wrote.
“So children who follow the queue with intensive attention (to help find creaks) are more attractive and better prepared to understand the more complex concepts introduced later.”
The analysis identified seven key moments in video where significant changes in child's eye movements are more strongly linked to how well they understand the concept of animal camouflage.
Alex Bonus, one research co-author with children and media expertise, noted that seven points lined up with significant changes in video educational content. These boundaries are when people realize that one meaningful experience is over and a new experience is beginning.
For example, when the narrator began to explicitly define camouflage, one event boundary occurred, combining the description with the visual representation of the word.
Coronel emphasized that these results are preliminary and that they still don't understand much about what happens at key points in the video when learning appears to be enhanced.
“However, this approach can help experts design their messages at the boundaries of events that enhance learning,” he said.
Coronel said the findings are particularly relevant now as vision techniques become cheaper and more common.
It enables a future in which video learning is truly personalized, along with advances in AI. For example, for now, it may take a few days or weeks for a teacher to find out whether a student understands his lesson. In many cases, teachers do not discover the next test or quiz.
“Imagine a future in which eye tracking can instantly convey when a person doesn't understand a concept in a video lesson.
“Video may be able to provide alternative examples and ways to illustrate concepts. This could make instructions more personalized, effective and scalable.”
