Poster C11, Friday, August 17, 10:30 am – 12:15 pm, Room 2000AB
Neural evidence of voice processing: the distinction between familiar and unknown voices
Julien Plante-Hébert1, Boutheina Jemel2, Victor Boucher1;1Laboratoire de sciences phonetiques de l'Universite de Montreal, 2Laboratoire de recherche en electrophysiologie cognitive, Hopital Riviere-des-Prairies
From an evolutionary perspective, familiar voice recognition, an ability humans share with many other species, was crucial to distinguish individuals representing a potential menace. However, discrimination amongst unknown voices presents a more general ability. On this basis, research has suggested that separate processes may underlie the recognition of familiar voices and the discrimination of unknown voices. It remains, however, unclear whether these two abilities present differing neural correlates. The primary objective of our study is to identify the EEG correlates associated with listeners’ recognition of a familiar voice and discrimination of unknown voices. In the present study, the ERPs of 13 participants were recorded as they passively listened to a set of 12 similar voices (same dialect, average F0 within 1 semitone, etc.). Within this set of voices, only one speaker was very familiar to the participants (e.g. the voice of a family member, a close friend, a partner, etc.) while 11 other voices were unknown. The ERPs were recorded over four blocks of 240 trials, each block representing a four-syllable utterance. The familiar voice and one unknown voice (“frequently heard voice” condition) were each presented on 33% of trials and the remaining 10 unknown voices (“rarely heard voice” condition) were each presented on approximately 3% of trials. Significant differences were observed between the familiar voice and both types of unknown voices (frequently and rarely presented) on a component peaking between 200 and 250 ms on right central electrodes (FC4, C4 and CP4). The familiar voice was also found to be significantly different from all the unknown voices on later-occurring components between 450 and 850 ms. Within this time window, the strongest differences were observed on most right temporal and left parietal sites between 550 and 650 ms. Analysis also revealed significant differences between the frequently presented voice and the rarely presented voices on most left frontal and right parietal sites only between 300 and 350 ms. We interpret the timeline of ERP responses as reflecting a two-step process of familiar voice recognition. First, as shown in the 200–250 ms time window, listeners recognize a voice as being familiar, compared to unknown voices. Second, as illustrated by the results of the later-occurring components (450–850 ms), listeners identify the voice by accessing a speaker-voice representation in long-term memory, which requires more processing time in more brain regions. Our results also support the idea that the process of unknown voice discrimination is distinct from familiar voice recognition. Indeed, the only significant difference between the frequent and the rare voices was observed in a different time window (300–350 ms). This difference suggests that listeners processed the frequent voice and the rare voices separately. However, since no contextual information about the frequent speaker was accessible to the listener, the voice remained unknown and yet remembered. Taken together, the results offer neurophysiological evidence of a complex processing of familiar voices as well as the presence of a parallel process implicated in unknown voice discrimination based on the creation of exemplars in memory.
Topic Area: Perception: Speech Perception and Audiovisual Integration