Discrete field-of-view primes reinstate holistic representations of 360° space

Abstract

We experience our visual environment in discrete glances. How do we build a unified representation of a 360º visual environment? One possibility is that neighboring scene views become associated through concurrent activation during learning, enabling a single view to subsequently prompt reactivation of the surrounding environment. In this study, we used Virtual Reality (VR) to ask whether a single view in an immersive panorama primes perceptual judgements of adjacent views in 360º space. In the Study Phase, participants (N=6) learned 18 real-world panoramas using head-mounted VR. In the Priming Test, on each trial, participants made a perceptual (open/closed) judgement on a target image taken from a studied panorama (110º field-of-view). Participants turned their head to view the target, which appeared either on their left/right in 360º space. Before target onset, a brief (300ms) prime image was presented directly ahead, either depicting: 1) an adjacent snapshot from the same panorama as the target (Valid prime), 2) a snapshot from a different panorama (Invalid prime), or 3) a grey box (Neutral prime). In an Explicit memory test, on each trial, participants reported whether a snapshot had appeared on the left/right/front in the panorama during learning. We found a significant effect of priming condition on reaction time (p<.01): participants were faster to make open/closed judgements when primed with an image from the same panorama (Valid prime condition) as compared to the Invalid (p<.01) and Neutral prime conditions (p=0.1). Explicit memory performance was high (>85%) for all participants, indicating accurate learning of the panoramas. We show that discrete views of a panorama prime each other in 360º space. These results suggest that spontaneous reinstatement of unseen, adjacent views may play an important role in facilitating ongoing perception and maintaining a seamless sense of our 360º visual environment.