Physically and virtually embodied agents offer great potential due to their capacity to afford interaction using the full range of human communicative behavior. To know when to best utilize these behaviors, these agents must be able to perceive subtle shifts in users’ emotional and mental states.
Contributing to the development of agents with such capabilities, Dan Szafir and Bilge Mutlu presented their work in implementing a robotic agent capable of sensing and responding to decreasing engagement in humans in their recent paper, Pay Attention! Designing Adaptive Agents that Monitor and Improve User Engagement.
In any learning situation, teachers communicate most effectively to learners when they sense that learners are beginning to lose attention and focus, and re-engage their attention through verbal and non-verbal immediacy cues. These cues–for example, changes in speech patterns, gaze, and gestures–create a greater sense of immediacy in the relationship between the teacher and learner, drawing the focus of the learner back to the topic at hand. Szafir and Mutlu posit that equipping robotic agents with the ability to monitor learners’ brain wave patterns through electroencephalography (EEG), recognize declining attention, and respond with such immediacy cues, the efficacy of learning can be improved. They also argue that such agents will promote a stronger sense of rapport between learner and agent, as well as a greater motivation to learn in learners.