A key goal in virtual reality (VR) is to provide the user with a first-person perspective of a virtual environment that feels every bit as real as the world around them. Achieving this goal is of great interest to fields such as architecture, where it would allow both the architect and their client to explore a space they are designing before it is built, and use its realistic presentation to aid in the decision making process. Careful attention has to be paid, however, to ensure that the virtual representation is faithfully portraying the same sense of space as the real world. Unfortunately, research has uncovered that using an egocentric point of view within virtual environments results in the users experiencing errors in their spatial judgment [Henry & Furness 93]. Specifically, from a user's perspective, objects in the virtual environment will appear closer than they really are. This poses a serious problem for using virtual environments as veridical representations, as the mismatches occurring between perceived virtual spaces and finalized designs would compromise any conclusions being made within the virtual reality system.
This dissertation presents work that explores this misjudgment problem from a cognitive sense, one where we evaluate the effect that a user's sense of presence has in the virtual environment. Our experiments are grounded with an initial study, which finds that placing a user within a virtual environment that matches the real, physical space they occupy does not generate the same errors found in previous studies. Further experiments build off this discovery by exploring the beneficial aspects of co-location of the virtual space with the physical, and delving into the possible cues used in high-fidelity replicas that aid in a user's increased accuracy. In addition, we present work on how to increase the user's spatial accuracy in remote locations, which is the true goal of virtual reality. This research focuses on giving users additional feedback in the virtual world, through the use of full-bodied virtual avatars. With these additional cues, we grant the user the ability to form a better cognitive model of the virtual space, and start making significant headway into eliminating the distance compression problem.