How real is virtual? A test of episodic memory with a shelf task

Immersive virtual reality (IVR) has been proposed as a bridge between laboratory studies and real-world behavior, offering both high experimental control and ecological validity. However, studies directly comparing virtual and real-world replicas remain scarce. The present study investigated three experimental conditions—IVR, non-immersive VR (N-IVR), and Reality—focusing on subjective experiences and episodic memory performance. Participants (N = 156) were randomly assigned to one of the three conditions and completed a Shelf Task involving two encoding sessions in which they placed 8 objects on a shelf per session, followed by two recall sessions 10 minutes later. After encoding, participants rated subjective experiences, including spatial presence (self-location and action possibilities), similarity to daily life, enjoyment, cybersickness, and game mechanics. Overall, IVR approximated real-world experience more closely than N-IVR. Ratings of action possibilities and similarity to daily life in IVR matched Reality, and IVR elicited higher enjoyment than both N-IVR and Reality. Some differences remained: self-location and game mechanics favored Reality, and cybersickness was unique to IVR, though at low levels. Memory performance revealed robust session × condition effects. In Session 1, IVR recall lagged behind N-IVR and Reality, suggesting initial challenges with IVR encoding. By Session 2, IVR performance improved markedly, reaching levels comparable to N-IVR for free recall and to both Reality and N-IVR for spatial recall. These findings highlight both the promise and limitations of IVR: it can closely approximate real-world experience and memory performance over repeated exposure, but subjective and embodied aspects of reality are not fully replicated. The results have implications for experimental and applied cognitive assessment, as well as clinical and training contexts.