Slide Slam F13
Rapid pre-attentive voice recognition of a famous speaker: Electrophysiological effects of Angela Merkel’s voice
Paula Rinke1, Kjartan Beier1, Ramona Kaul1, Tatjana Schmidt1, Mathias Scharinger1; 1Philipps-University Marburg
Introduction: The recognition of human speakers by their voice is a remarkable cognitive ability with parallels to human face recognition. Previous research has established a voice area in the right temporal cortex that helps extracting relevant acoustic features while listening to speech. The integration of these features appears to be accomplished rapidly, indicating a pre-attentive processing of voices. With contrasting brain responses to familiar and unfamiliar voices, a previous MMN (Mismatch Negativity) study reported an effect of voice familiarity and recognition on general voice processing: stronger activations were generated by familiar voices. The familiar voices used for this study were extracted from the participants’ personal environment. The present study aims to further investigate this effect for publicly but not personally known voices. Therefore, the voice of German chancellor Angela Merkel was contrasted with acoustically matched-voices while an electroencephalography was carried out to record the correlating neurophysiological patterns. Method: A classic passive oddball paradigm contrasted two two-syllable German utterances (Kinder ‘children’ and Tochter ‘daughter’) by Merkel with the same words uttered by two unknown female speakers with matched voice characteristics (age, regional background and voice quality). As Merkel has been in office for over 15 years, it can be assumed that she should be easily recognized by German listeners. Twenty-one participants (21-43y) were presented with standard-deviant trains of these word stimuli while their electroencephalogram was recorded from 32 active electrodes. Electrophysiological voice processing indices were quantified as identity mismatch negativities (MMNs) and P3 differences, and cortical sources of both difference wave forms were estimated with variable resolution electromagnetic tomography. Results: The results showed amplitude and latency effect for both MMN and P3: Merkel’s voice elicited a smaller but earlier MMN than the control voices. The P3, by contrast, was both larger and later in response to Merkel. Importantly, the MMN amplitude effect depended on speaker recognition. Recognition could be predicted by MMN amplitudes and participant age. Younger participants’ recognition was indexed by higher amplitudes, while older participants’ recognition was indexed by lower amplitudes. Finally, the MMN in response to Merkel originated from right-hemispheric regions in temporal cortex, overlapping with the voice area, while the MMN in response to the controls originated from left superior temporal gyrus. P3 responses to Merkel had sources in left superior temporal and middle frontal gyrus, while P3 responses to the controls had sources in right temporal and occipital areas. Conclusion: These results suggest unique recognition patterns of very famous voices by electrophysiological responses. Speaker recognition by voice seems thus to rely on pre-attentive right temporal processing within the first 150 ms of the acoustic signal. Further, the amplitude reduction in response to Merkel’s voice is compatible with predictive coding accounts that assume reduced prediction error if there are strong priors.