Datensatz

Zusammenfassung

There is a long tradition of investigating the self-motion illusion induced by rotating visual stimuli ("circular vection"). Recently, Larsson et al. (2004)[1] showed that up to 50 of participants could also get some vection from rotating sound sources while blindfolded, replicating findings from Lackner (1977)[2]. Compared to the compelling visual illusion, though, auditory vection is rather weak and much less convincing.
Here, we tested whether adding an acoustic landmark to a rotating visual photorealistic stimulus of a natural scene can improve vection. Twenty observers viewed rotating stimuli that were projected onto a curved projection screen (FOV: 54°x40.5°). The visual scene rotated around the earth-vertical axis at 30°/s. Three conditions were randomized in a repeated measures within-subject design: No-sound, mono-sound, and 3D-sound using a generic head-related transfer function (HRTF).
Adding mono-sound showed only minimal tendencies towards increased vection and did not affect presence-ratings at all, as assessed using the Schubert et al. (2001) presence questionnaire [3]. Vection was, however, slightly but significantly improved by adding a rotating 3D-sound source that moved in accordance with the visual scene: Convincingness ratings increased from 60.2 (mono-sound) to 69.6 (3D-sound) (t(19)=-2.84, p=.01), and vection buildup-times decreased from 12.5s (mono-sound) to 11.1s (3D-sound) (t(19)=2.69, p=.015). Furthermore, overall presence ratings were increased slightly but significantly. Note that vection onset times were not significantly affected (9.6s vs. 9.9s, p>.05).
We conclude that adding spatialized 3D-sound that moves concordantly with a visual self-motion simulation does not only increase overall presence, but also improves the self-motion sensation itself. The effect size for the vection measures was, however, rather small (about 15), which might be explained by a ceiling effect, as visually induced vection was already quite strong without the 3D-sound (9.9s vection onset time). Merely adding non-spatialized (mono) sound did not show any clear effects. These results have important implications for the understanding or multi-modal cue integration in general and self-motion simulations in Virtual Reality in particular.