Learning to recognize unfamiliar talkers: Listeners rapidly form representations of facial dynamic signatures

Seeing the motion of a talking face can be sufficient to recognize personally highly familiar speakers, suggesting that dynamic facial information is stored in long-term representations for familiar speakers. In the present study, we tested whether talking-related facial dynamic information can guide the learning of unfamiliar speakers. Participants were asked to identify speakers from configuration-normalized point-light displays showing only the biological motion that speakers produced while saying short sentences. During an initial learning phase, feedback was given. During test, listeners identified speakers from point-light displays of the training sentences and of new sentences. Listeners learned to identify two speakers, and four speakers in another experiment, from visual dynamic information alone. Learning was evident already after very little exposure. Furthermore, listeners formed abstract representations of visual dynamic signatures that allowed them to recognize speakers at test even from new linguistic materials. Control experiments showed that any potentially remaining static information in the point-light displays was not sufficient to guide learning and that listeners learned to recognize the identity, rather than the sex, of the speakers, as learning was also found when speakers were of the same sex. Overall, these results demonstrate that listeners can learn to identify unfamiliar speakers from the motion they produce during talking. Listeners thus establish abstract representations of the talking-related dynamic facial motion signatures of unfamiliar speakers already from limited exposure.