Figure 1.
Stimulus setup and trial structure.
(A) Subjects sat approximately 53 cm from a video of the speaker's face. Sounds were positioned at 13 different locations ranging from −36° to +36° in 6° increments. Subjects reported whether they perceived the presented sound to originate from the Same- or Different-Location as the speaker's mouth. Arrows point toward the perceived location and the corresponding response by the subject. (B) Trial structure. The first frame of each video was loaded 500 ms prior to starting the full video. Subjects were allowed 1.2 s to respond.
Table 1.
Confusion matrix for VCV identification. (% of Responses; mean ± SEM).
Figure 2.
HRTFs precisely capture binaural auditory cues.
Subjects indicated the perceived location of an HRTF filtered speech token (e.g. /aba/) at one of thirteen different locations via a pointing task. These indicated locations are plotted as a function of the intended (experimenter controlled) location. The data show a linear relationship between the Presented Angle and Reported Angle (1.25*X+0.176, R2 = 0.99). (mean ± SEM, N = 10).
Figure 3.
McGurk interference is spatially independent.
(A) McGurk interference rates are plotted as a function of stimulus condition and audiovisual pair. Responses indicative of McGurk interference significantly increased in the Incongruent condition compared to the Still Face and Congruent conditions. (B) Corrected McGurk interference rates are plotted as a function of location. The data suggest that speech related cues operate independently of spatial information. Corrected McGurk interference rates were calculated by subtracting the %McGurk Responses in the Still Face Condition (light-gray bars) from the Incongruent condition (dark-gray bars). (N = 11, means ± SEM).
Figure 4.
Speech cues affect audiovisual integration in space.
(A) The percentage of Same-Location responses is plotted as a function of location for each condition: Congruent (black triangles), Incongruent (dark gray, asterisk), and Still Face (light gray, squares). Importantly, the likelihood of spatial integration (i.e. responding Same-Location) decreases at the smallest disparities in the Incongruent relative to the Congruent condition. (N = 11, means ± SEM).