アイテム詳細

要旨

“Circular vection” refers to the illusion of self-motion induced by rotating visual or auditory stimuli. Visually induced vection can be quite compelling, and the illusion has been investigated extensively for over a century. Rotating auditory cues can also induce vection, but only in about 25-60% of blindfolded participants (Lackner, 1977; Larsson et al., 2004). Furthermore, auditory vection is much weaker and far less compelling than visual vection, which can be indistinguishable from real motion. Here, we investigated whether an additional auditory cue (the sound of a fountain that is also visible in the visual stimulus) can be utilized to enhance visually induced self-motion perception. To the best of our knowledge, this is the first study directly addressing audio-visual contributions to vection. Twenty observers viewed rotating photorealistic pictures of a natural scene projected onto a curved projection screen (FOV: 54x45). Three conditions were randomized in a repeated mea-
sures within-subject design: No sound, mono sound, and spatialized sound using a generic head-related transfer
function (HRTF). Adding mono sound to the visual vection stimulus increased convincingness ratings marginally,
but did not affect vection onset time, vection buildup time, vection intensity, or rated presence. Spatializing the
fountain sound such that it moved in accordance with the fountain in the visual scene, however, improved vection
significantly in terms of convincingness, vection buildup time, and presence ratings. The effect size for the vection
measures was, however, rather small (<16%). This might be related to a ceiling effect, as visually induced vection
was already quite strong without the spatialized sound (10s vection onset time). Despite the small effect size, this
study shows that HRTF-based auralization using headphones can be employed to improve visual VR simulations both in terms of self-motion perception and overall presence. Note that facilitation was found even though the visual stimulus was of high quality and realism, and known to be quite powerful in inducing vection. These findings have important implications both for the understanding of cross-modal cue integration and for optimizing VR simulations.