Abstract
We can infer a voice of an unfamiliar person from his/her face. One hypothesis attributes this to common information conveyed by voices and faces. However, the nature of this critical information remains unclear (Mavica & Barenholz, 2013). Some recent reports have shown that auditory and visual stimuli, which do not have any direct relations (e.g. music – color, timbre – visual texture), nonetheless appeared to be related based on certain perceptual and emotional properties (Palmer et al, 2013; Peterson et al., 2014). The present study examined whether supra-modal information, such as perceptual and emotional properties and personality traits can mediate such inferential links between voice and face. In this experiment, voices conveyed neutrally spoken sentences and faces were neutral visual pictures of male/female faces. To investigate indirect relationships between voice and face, models of pictures were different people than the speakers. In the first task, one voice was presented simultaneously with multiple faces, and participants had to select the first, second and third faces that corresponded to the presented voice. In a second, then a third task, the voice and face were independently presented along with 18 pairs of bipolar adjectives. Using an eight point scale, participants rated the likelihood of a voice or face matching a given pole. Adjectives described perceptual properties (e.g. smooth - rough), emotional properties (e.g. happy - sad), and personality traits (e.g. passive - dominant). For each adjective pair, a weighted average of ratings of faces selected as corresponding to each voice was calculated. Results showed that weighted averages were strongly correlated with ratings for the voice itself, especially for certain perceptual adjectives (e.g. r = .74, glossy – matte). This suggests that the correspondence between neutral voices and faces are mediated mainly by perceptual properties.
Meeting abstract presented at VSS 2015