The Physical Presence of a Robot Tutor Increases Cognitive Learning Gains

The Physical Presence of a Robot Tutor Increases Cognitive Learning Gains Daniel Leyzberg (daniel.leyzberg@yale.edu) Samuel Spaulding (samuel.spaulding@yale.edu) Mariya Toneva (mariya.toneva@yale.edu) Brian Scassellati (scaz@cs.yale.edu) Department of Computer Science, Yale University 51 Prospect St., New Haven, CT 06511, USA Abstract We present the results of a 100 participant study on the role of a robot’s physical presence in a robot tutoring task. Partic- ipants were asked to solve a set of puzzles while being pro- vided occasional gameplay advice by a robot tutor. Each par- ticipant was assigned one of five conditions: (1) no advice, (2) robot providing randomized advice, (3) voice of the robot providing personalized advice, (4) video representation of the robot providing personalized advice, or (5) physically-present robot providing personalized advice. We assess the tutor’s ef- fectiveness by the time it takes participants to complete the puzzles. Participants in the robot providing personalized ad- vice group solved most puzzles faster on average and improved their same-puzzle solving time significantly more than partic- ipants in any other group. Our study is the first to assess the effect of the physical presence of a robot in an automated tu- toring interaction. We conclude that physical embodiment can produce measurable learning gains. Keywords: Robotics; Computer Science; Tutoring Introduction What kinds of human-robot interactions benefit from the physical embodiment of a robot? For human-robot interac- tions that require manipulating the physical world, a physical robot is a necessity, but for those interactions where physi- cal embodiment is optional, when is an embodied robot more useful than an on-screen agent? In this study, we explore the differences in task perfor- mance of participants engaged in a cognitive learning task in which a robot acts as a tutor. Participants were asked to play a puzzle game while receiving strategy advice from either: a physically-present robot, a video of the same robot, its disem- bodied voice, a robot giving randomized advice, or no agent at all. We use the resulting data to draw conclusions about the effect of embodiment in robot tutoring tasks. Previous work has investigated the social influence of a robot’s embodiment. Does a robot engender more trust, more compliance, more engagement, or more motivation by its physical presence, more so than an on-screen agent or a video representation of a robot would? Such questions have been explored via two methodologies: self-report mea- sures and task-performance measures. Using self-report mea- sures, Kidd and Breazeal (2004) found that a physically- present robot was perceived as more enjoyable, more cred- ible, and more informative than an on-screen character in a block-moving task. In Wainer, Feil-Seifer, Shell, and Mataric (2007), an embodied robot was rated as more attentive and more helpful than both a video representation of the robot and a simulated on-screen robot-like character. Tapus, Tapus, and Mataric (2009) found that individuals suffering from cog- nitive impairment and/or Alzheimer’s disease reported being more engaged with a robot treatment than a similar on-screen agent treatment. Kiesler, Powers, Fussell, and Torrey (2008) used task- performance measures to find that participants who received health advice from a physically-present robot were more likely to choose a healthy snack than participants who re- ceived the same information in robot-video or on-screen agent conditions. In Bainbridge, Hart, Kim, and Scassellati (2008), a physically-present robot yielded significantly more compliance to its commands than a video representation of the same robot. No previous work has investigated whether learning out- comes are affected by a robot’s physical presence. The closest related work is in Intelligent Tutoring Systems (ITS´s), which are educational computer programs that produce individual- ized lessons, advice, and questions usually in a workbook- style or quiz-style environment (Nkambou, Bourdeau, & Psych´e, 2010). A parallel notion of embodiment called “the persona effect” exists in ITS research. (See Dehn and Van Mulken (2000) for an overview.) The persona effect is the impact, if any, that an on-screen character has on students using an ITS. The majority of research on the persona ef- fect has shown no significant learning gains produced by on- screen agents, although many studies note that students find an ITS with an on-screen more engaging than one without (Moundridou & Virvou, 2002). Our study is the first to assess the effect of the physical presence of a robot in an automated tutoring interaction. We use the task-performance measure of puzzle solving time in this work as well as several self-report measures. Methodology Participants There were 100 participants in this study, between 18 and 40 years of age. The study was conducted in New Haven, Con- necticut. Most participants were undergraduate and graduate students of Yale University. Each participant was assigned to one of five groups: (1) no lessons, (2) randomized lessons from a physically-present robot, (3) personalized lessons from a disembodied voice, (4) personalized lessons from a video representation of the robot, and (5) personalized lessons from