Contingent interaction with a television character promotes children's science learning and engagement

Abstract

While educational television or video programs are important and accessible learning resources for young children, the lack of contingent interaction afforded within this type of programming may limit how much children learn from them. In this project, we leveraged natural language processing technologies to enable contingent interaction between children and a television character, with the goal of enhancing children's learning and engagement. Our randomized controlled study involving 77 four to six year old children suggested that incorporating contingent interaction within video programs helped children better understand the science concepts introduced in the video. Children who had contingent interaction also showed a higher level of vocalization and more positive affect during video watching and developed a more positive perception of the media character, though they spent less time looking at the screen. These findings shed light on the potential of contingent interaction with on-screen media characters enabled by artificial intelligence to promote learning and engagement.