Minecraft is a commercially available game that uses a procedurally generated 3D world of simple independent block-shaped entities to represent different materials that can create almost anything the user can imagine (e.g., cities, computers, farms), making it well-suited to creating virtual learning environments for students in many subject areas. However, the literature surrounding the use of Minecraft in the classroom tends to investigate how much students like using the game and how the game can be implemented in project-based learning situations to facilitate interest and collaboration among peers, but these studies typically have poorly defined goals or learning outcomes and are typically not designed to test theoretically derived predictions. The result is a fragmented body of literature that offers no direction to educators about how to design or structure their game lessons to facilitate meaningful learning and make Minecraft a successful instructional medium. This dissertation attempts to address these problems by implementing a quantitative-focused mixed-method approach to test two theoretically derived and empirically supported principles of multimedia design (the guided discovery principle and the pretraining principle) in a custom Minecraft lesson covering five basic logic gates that are used in electrical engineering and computer science (NOT, OR, AND, NOR, NAND) to determine their effect on cognitive load and posttest outcome performance after a week delay; and compares the Minecraft lesson to a static PowerPoint lesson. Qualitative data (game recordings and interviews) were collected to gain a deeper understanding of how students interact with the learning material. In Experiment 1, a Minecraft lesson varied the amount of discovery learning by either limiting the explicit instruction given to students (Pure Discovery), giving robust explanations about how and why the logic gates work (Guided Discovery), or presenting students with a screen capture recording of the Guided Discovery game lesson (Direct Instruction). Analyses indicated no significant difference among the lessons in performance on delayed posttest learning outcomes but reported extraneous cognitive load was significantly higher for the Direct Instruction condition, which indirectly affected performance on the posttest. The qualitative data show these null results were likely due to a weak manipulation between the guided and pure discovery and the lack of narration in the video lesson, which students found distracting. Experiment 2 compared learning outcomes and cognitive load when using a pretraining infographic to no pretraining learning before experiencing either a Minecraft game lesson or an equivalent PowerPoint lesson. Analyses found that the Minecraft groups could accurately recreate the logic gates they learned better and put in less effort to learn than those in the PowerPoint groups. There were mixed results regarding whether pretraining fostered students understanding of the underlying concepts of logic gates and logical thinking better than no pretraining, with patterns trending in favor of pretraining. Those who received the pretraining would be more likely to engage in this sort of lesson again. There were no significant interactions, although the Minecraft + pretraining group performed best on 6 of 7 delayed posttest outcome measures. Overall, the learning outcome data did not support the use of the guided discovery principle and pretraining principle in a Minecraft lesson about logic gates. However, there were major limitations – weak manipulations used in Experiment 1 and low experimental power due to small sample sizes in Experiment 2 – that restrict our ability to make decisive conclusions about whether these principles can positively or negatively affect Minecraft as an instructional medium. This dissertation shows the value of using qualitative data to help explain quantitative results and that designing game-based learning environments should focus on reducing features that cause extraneous load and implementing features that manage essential load.