A cognitive framework called REflective Agent Learning environment (REAL) is developed in this study. REAL is a reusable framework that allows researchers to develop a simulation-based learning environment where users can learn through passing their thoughts to some computer-based agents and observe how the agents embodying their knowledge behave as the result of their instruction. Our research benefits from the research in Intelligent Tutoring Systems, game based learning systems, and agent technologies, stressing reflection as part of the thinking processes. It focuses on the design of the framework and the testing of its usability. The external evaluation of specific implementations serves as the guidance for the future design of the REAL applications. We hope, by grounding themselves in the needs of local practice, the REAL applications can give us opportunities to understand how theoretical claims about teaching and learning can be effectively transformed into meaningful learning.
TopIntroduction
Using computers to provide personalized instruction as an alternative to human tutors has drawn the attention of researchers in the fields of education, psychology, computer science, and cognitive science. Thanks to their cooperative efforts, we have Intelligent Tutoring Systems (ITSs), or Intelligent Computer-Assisted Instruction (ICAI), embodying the computer-as-tutor paradigm. In this sense, ITSs/ICAI can be considered as Pedagogical Agents, which have a set of normative teaching goals and plans for achieving these goals (e.g., teaching strategies), and associated resources in the learning environment (Thalmann, 1997). The agent paradigm now allows researchers to collaborate effectively in an effort to develop other efficient user-centered learning environments. Examples include the computer as a collaborator (Blandford, 1994; Dillenbourg & Self, 1992), the computer as a learning companion (Chan & Baskin, 1990), and the computer as a teachable agent (Biswas, 2005), to name a few.
However, the design and development of computer-based instruction systems are costly (Murray, 1999; Anderson, 1993). Adaptive and broadly applicable cognitive tools are needed to reduce the development cycle time and the level of the required expertise. This will allow for computer-based intelligent tutoring systems to become affordable learning environments in traditional classroom settings.
The primary goal of this research is to design and prototype an intelligent reflective agent situated in an educational gaming environment based upon a cognitive framework, called REAL (REflective Agent Learning Environment). Attempts were made to encourage reflective thinking through having users explicitly externalize their internal knowledge representations by instructing the agent, which, ideally, generates a sequence of behaviors analogous to those generated by the users’ imaginary worlds. It is our hope that when users begin to recognize relationships between their prior knowledge and the newly presented meanings, learning occurs, thus making the new information accessible as part of the learners’ active reservoirs of knowledge. The tangible results of this research are some prototypes of the REAL applications.