Relational Agents are computer agents designed to form long-term, social-emotional relationships with their users. We are investigating the use of these agents in task domains in which human-agent relationships actually improve task outcomes, such as in coaching, counseling, psychotherapy and healthcare. Current research is focused on the design of empathetic dialogues for these agents and the development of agent dialogue and behavior that is effective at promoting long-term engagement with these agents and adherence to their recommendations.

Relational Agents represent an especially effective modality for educating and motivating older adults and other individuals with low reading, health or computer literacy, given the agents’ intuitive emulation of face-to-face conversation and the engagement and social bonding afforded by their use of relational behavior such as social dialog, empathy and humor. In a recently completed study, participants recruited from the Geriatric Ambulatory Practice at Boston Medical Center who used a Relational Agent exercise coach daily for two months performed significantly more walking compared to a non-intervention control group drawn from the same practice.

The goal of this two-year NIH-funded project is to investigate the efficacy of real-time motivational messages on exercise adoption, compared to motivational messages delivered retrospectively as is typically done with human or computer-based coaches. The real-time messages will be delivered by a PDA equipped with an integrated accelerometer that can sense whether users are performing their exercise or not. The project, performed in conjunction with researchers at MIT and Harvard Medical School, will also evaluate the efficacy of motivational messages delivered by a Relational Agent compared with the same content delivered via a text modality only.