Additional research in virtual reality is required to develop and refine the instructional practices and technological tools that serve both a learning and evaluative function. For this work in particular, several directions have been identified for future exploration.
With respect to the design of the educational environment, a number of modifications may be implemented. An immediate modification to NICE would be the enhancement of the central behavior model to support a more complete life cycle. This means adding appropriate feedback cues to indicate the end of the plant's life cycle or other characteristics that follow a more realistic environmental model. Additional enhancements include some features that were originally planned but not implemented, such as the relationship between the growth of a plant and its proximity to the plants surrounding it. The NICE project is a continuing effort with many possibilities for expansion. If our goal with NICE is to provide emerging learning experiences, then the learner should take a more active role. It is necessary to create a model of an environment that will place students more in charge of their own learning and thinking, in a way that can engage more than one learners (in our case more than just the driver). An important step in this direction is the development of an end-user authoring system that will allow children to construct their own ecological models. It seems that this should initially take place outside the VR environment and eventually move in, when technology permits. Other immediate needs include the development of a more familiar and natural interface for use by children, that strives to resolve the technical difficulties observed during this evaluation.
With respect to the evaluation framework, a more detailed and focused framework is needed on the conceptual side. The cognitive and pedagogical aspects of VR learning need to be examined further, as technical factors are currently being resolved by the industry. Specifically, the framework can be expanded to account for the evaluation of deep ideas and concepts that children have a hard time with, and support ways in which their acquisition can be evaluated.
With respect to the evaluation itself, substantial amounts of data have still to be gathered on both the initial questions and the questions that emerge during observation. Future research demands intensive observational studies of children, as well as longitudinal studies on the role of virtual reality in shaping attitudes and learning. To gain a deeper understanding of the issues involved, a more directed evaluation should be conducted with multiple case studies, studied in depth, for longer and multiple times. Case studies of a small number of subjects make prolonged engagement and persistent observation much easier to accomplish, as the subjects can stay longer, come more than once, and be observed under calmer, more relaxed conditions. They also allow to deal with variables that are hard to quantify, such as affective behavior, but which are important in the study of virtual reality for learning.