As virtual reality (VR) technologies have been increasingly adopted in the educational field, careful consideration and evaluation of instructional modes to support learning with VR are needed. This study focuses on evaluation of two instructional modes: direct interaction and manipulation of the VR learning environment; and the vicarious approach of observing how a novice peer and an expert learn with VR. An experimental design was utilized with two comparison conditions: direct VR interaction (n = 72) and the vicarious approach (n = 69). Participants' knowledge and emotional engagement were evaluated via self-report questionnaires. Results showed that though learning with the direct VR condition was associated with more intense emotional engagement, there was no significant difference in knowledge gains between the two study conditions. The findings suggest that each of the two instructional modes has specific instructional affordances and that the cognitive and emotional components of learning via VR should be further explored. Since learning vicariously through observing the dialogue between an expert and a peer can have immediate knowledge gains comparable to the direct VR interaction, there is an effective and more accessible opportunity for VR-based learning. Future research should evaluate this instructional effect on knowledge retention.