Perceived Realism of Virtual Environments Depends on Authenticity

While the perception of a virtual environment (VE) is usually described in terms of its level of immersion and users’ sense of presence, the construct of authenticity might be more useful. The authenticity of a VE depends on whether the affordances and simulations chosen in its implementation support (1) users’ expectations based on their Bayesian priors for regularities in the real world and (2) the users’ intentions in the VE. With the growing number of consumer-grade virtual reality (VR) headsets, there is increasing interest in what makes a good virtual environment (VE). While there are several useful constructs for discussing users’ perception of a virtual environment, that is, fidelity, immersion, and presence, the term ‘‘authenticity’’ refers to whether the virtual environment provides the experience expected by the user, both consciously and unconsciously. The concept of fidelity is frequently tied to the goodness of a VE; how realistic is it? However, depending on the context, different types of fidelity become relevant. Visual fidelity (Does it look realistic?) might be important for the windshield portion of a driver training simulator (McMahan, Bowman, Zielinski, & Brady, 2012; Riener, 2010), but auditory fidelity (Does it sound realistic?) is critical for simulators of heavy equipment operation (Lu & Davis, 2016), where operators make decisions based on engine sounds. In a surgical simulator (Triantafyllou, Lazaridis, & Dimitriadis, 2014), surgeons need to feel the tension of tissue through a high-fidelity haptic interface (Does it feel realistic?). This list of fidelities is by no means exhaustive, but enumerating a few argues that measuring fidelity depends on the context and tasks at hand. Also, the concept of fidelity offers no real guidelines for establishing target thresholds. How much fidelity is good enough? Bowman and McMahan (2007) use the terms ‘‘immersion’’ and ‘‘presence’’ to discuss the realism of a virtual environment, framing immersion as an objective measure of the simulation equipment provided. What field of view is provided by the VR headset or surround screen? What is the reaction latency? A VE with higher immersion, so the argument goes, should lead to higher fidelity, and generate a greater sense of presence, the subjective experience felt by the user. But if I put you in a highly immersive environment and give you badly designed content to experience, will you perceive the VE as realistic and experience presence? Probably not. What’s missing from this dichotomy of immersion (objective, system-focused) and presence (subjective, user-focused) is a computational theory about the extent to which the VE reflects the expected regularities of world that it is attempting to represent—its authenticity. Authenticity draws on two streams of thought: expectations and motivations. The expectations stream arises from natural computation (Richards, 1988) and Bayesian inference (Tenenbaum & Griffiths, 2001). These researchers focus on the fact that perception is an underconstrained computational problem. Without observers’ probabilistic assumptions about the external world (Bayesian priors), the mathematics of perception are not possible to resolve based solely on the limited inputs to our senses. Thus, as observers, we build mental probabilistic models of the regularities of the world around us. Yang and Purves (2003), for example, described how the visual space we perceive is based on Bayesian inferences. A virtual environment, in its attempt to offer a useful proxy of one portion of the world, offers an even further reduced set of sensory information than the real world. However, a clever VE designer who understood observ-