Internal models differentially implicated in audiovisual perception of non-native vowel contrasts

This study investigates neural processes underlying audiovisual perception of a non-native vowel contrast using fMRI. Eight native Japanese and eight native English speakers participated in the study. The two-alternative forced choice task was to identify which vowel was present in the CVC English word presented under audiovisual (AV) and audio only (AO) conditions. Three sessions were conducted with different vowel contrasts; two native contrasts for Japanese ([o-e)], [o-i]), and one nonnative ([o-U]). A block design was employed. Significantly greater activity between the native Japanese group and the native English group for the contrast of AV-AO for the non-native session minus the native sessions occurs in the cerebellum. A decrease in activity occurs in the left transverse temporal gyrus. Increased activity in the cerebellum is consistent with the hypothesis that audiovisual non-native phonetic perception utilizes internal models to a greater extent than audio processing alone or to audiovisual processing of native contrasts. Decreases of activity in the right transverse temporal gyrus suggests that multisensory integration may involve higher order mapping of phonetic categories.