Cognitive load theorists claim that problem-centered instruction is not an effective instruction because it is not compatible with human cognitive structure. They argue that the nature of problem-centered instruction tends to over-load learner working memory capacity. That is why many problem-centered practices fail. To better support students and reduce learner cognitive load, researchers like Barron and Hmelo-Silver propose the use of embedded teaching or just-in-time information in problem-centered instruction when students are struggling with a problem and are in need of a particular kind of knowledge. A number of research studies done by Schwartz and Bransford suggest that a good time to provide a lecture in problem-centered instruction is after the initial problem evaluation or hypotheses generation and before self-directed learning. This research study tests Schwartz and Bransford's theory by incorporating agent-based mini-lectures in an online problem-centered learning environment (Learning to Teach with Technology Studio, LTTS) to help students develop a scientific understanding of inquiry, belief of inquiry, and the application of inquiry in designing a lesson plan. There were twelve participants in this study. In the experimental condition, students were required to first analyze the problems presented in the course, generate their initial thoughts about the problems, and then listen to the agent-based lectures before proceeding to the resources. There were no mini-lectures in the control condition. The results failed to find a significant difference on students' understanding of inquiry, belief about inquiry, or the ability to apply inquiry between the two groups. However, qualitative analysis on the interview data shows that both groups seem to be able to develop a scientific concept of inquiry. Moreover, click-stream data showed that students in the two conditions used the online resources differently. The experimental group spent less proportion of their course time on resources and made significantly less resource clicks compared to the control group. This seems to suggest that the experimental group was able to select and process relevant information more effectively than the control group, [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com.bibliotheek.ehb.be/en-US/products/dissertations/individuals.shtml.]