Variation over space of Audio-Visual (AV) spatial fusion has been investigated in darkness and light conditions (Hartnagel et al. 2007, Roumes et al. 2004). Those experiments revealed a gaze shift effect, indicating a reference frame of AV fusion space being neither head- nor eye-centered. Results in vision research have shown influence of visual allocentric reference frame on visual localization. Schmidt et al. (2003) have shown local distortion effect of visual landmark and experiments on the Reolof effect have shown shift of localization relative to the asymmetric surrounding display (Dassonville et al., 2004). Our experiment investigates visual allocentric effects on AV fusion. An hemi-cylindrical screen hiding 21 loudspeakers in a 2D arrangement, a projector displays on the screen a green permanent rectangular large background (135°H x 80°V); participant (head and body aligned) was sideways-oriented so the display appeared shifted 15° to the right relative to straight ahead and the surrounding visual frame was asymmetrical (frame offset). In each trial a vertical line providing a visual landmark randomly either straight ahead (head 0°) or 15° to the right (mid-display), a broadband noise burst and a 1° spot of light, 500ms duration, were simultaneously presented with random 2D spatial disparity. The task was about perception of spatial unity of the bimodal stimulus (fusion). Results showed that AV fusion depends mainly on egocentric reference frames relative position (gaze and head) and that local allocentric reference frame has no significant effect. Comparisons with previous results confirm the importance of surrounding visual display.

Audio; Visual; Fusion; Reference frames; Space