COMPUTER LEARNING 1 Running head: COMPUTER LEARNING Computer Learning Environments with Agents that Support Deep Comprehension and Collaborative Learning

During the last decade, learning scientists have developed technologies with animated pedagogical agents that interact with the student in natural language and other communication channels, such as facial expressions and gestures. These pedagogical agents model good learning strategies and coach the students in actively applying their knowledge. This chapter focuses on agent-based learning environments that attempt to facilitate deep comprehension (e.g., causal explanations, plans, logical justifications), reasoning in natural language, and inquiry (i.e., question asking, question answering, hypothesis testing). These agent-based learning environments have targeted high school and college students who learn about topics in science and technology. Tests of these systems have exhibited both successes and failures with respect to learning and generalization. One of these projects on AutoTutor has analyzed transfer at both coursegrain and fine-grain levels. The course-gain assessments have examined whether working on physics problems facilitate the solutions for similar physics problems with different surface characteristics. The fine-grain assessments have tracked the mastery and application of particular principles (e.g., net force equals mass times acceleration) throughout the history of pretest, training, and posttest. In all of these projects with agents, learning and generalization were assessed with multiple tests, tasks, and criteria, as opposed to relying on a single measure or goal standard.