Life-Like Pedagogical Agents in Constructivist Multimedia Environments: Cognitive Consequences of their Interaction

The goal of this study is to test the hypothesis that animated pedagogical agents can promote constructivist learning in a discovery-based learning environment. We do this by first, comparing the learning outcomes of students who learn in the context of social interaction with a computer-mediated character with the learning outcomes of students who learn in a computer text context. Second, we focus on the relative contributions of pedagogical agents' image, voice, and language style in promoting deeper learning. The Case for Software Pedagogical Agents Discovery-based learning environments are intended to facilitate constructivist learning through creative problemsolving experiences. However, the characteristics of discovery-based environments which offer the greatest potential for fostering learning also pose the greatest challenge: the complexity of the learning experience, based on the learner's freedom to explore and design artifacts in microworlds, constantly threatens to overwhelm learners. To combat this complexity, discovery-based learning environments should provide scaffolding in the form of highly contextualized problem-solving advice that is customized to each learner. Perhaps the most intriguing vehicle for providing such dynamically individualized scaffolding is the emerging technology of animated pedagogical agents-lifelike on-screen characters who provide contextualized advice and feedback throughout a learning episode (Lester & Stone 1997, Lester et al. 1997). A Life-Like Pedagogical Agent Example The discovery-based learning environment used in this study is a microworld called "Design-A-Plant", developed by the Multimedia Laboratory at the Department of Computer Science of North Carolina State University (Lester & Stone 1997, Lester et al. 1997). In this program, the student travels to an alien planet that has certain environmental conditions (e.g., low rainfall, light sunlight) and must design a plant that would flourish there (e.g., including designing the characteristics of the leaves, stem, and roots). It includes an animated pedagogic agent called Herman--a lovable alien with appealing facial expressions, human-like movements, and an amusing voice, who offers individualized advice concerning the relation between plant features and environmental features, encouragement when students encounter difficulties, and feedback on the choices that students make in the process of designing plants. Herman's actions are dynamically selected and assembled by a behavior sequencing engine that guides the presentation of problem-solving advice to learners, as described more fully in Lester et al. (1997). The program starts with Herman introducing the student to the first set of environmental conditions. Then, he asks the student to choose the appropriate root from the library of roots' names and graphics shown on the computer screen. After the student had chosen a root, Herman gives two explanations: first, a narrated explanation, and second a QuickTime movie that contains a similar explanation for the correct root. The same procedure applies to the stem and leaves, with Herman first asking the student to make a choice, and giving the student feedback afterwards. Once the student was given the last explanation on the leaves for each environment, he is taken to the next environment. The same procedure follows for the rest of the environments. Issue 1: Agent versus Text-Based Environments Does learning with an animated pedagogical agent in a discovery environment promote constructivist learning? In order to help answer this question, we conducted a preliminary study where the learning outcomes of students who learned about environmental science in the Design-A-Plant microworld (personal agent or PA group) was compared with the learning outcomes of students who received the identical verbal and visual materials in a computer-based text environment (no personal agent or no-PA group).