Accents and Confidence Modulate Brain Activity during Speech Perception

Post by: Shireen Parimoo

What's the science?

In-group favoritism refers to people’s tendency to hold more positive views about members of their own social group compared to members of an out-group. Accents act as indicators of social belonging, and a speaker’s voice also conveys information about their trustworthiness; they are more likely to be believed when their tone is confident rather than doubtful. We don’t know how voice-based perception of a speaker is processed in the brain and how level of confidence may influence the perception of an out-group speaker. This week in NeuroImage, Jiang and colleagues investigated patterns of brain activity in participants listening to in-group and out-group speakers with different accents and varying degrees of confidence.

How did they do it?

Twenty-six young adults listened to and rated the believability of English statements spoken in their native accent (Canadian-English; in-group), in a regional accent (Québécois-French; out-group/regional), and in a foreign accent (Australian; out-group/foreign). The statements also varied in whether they sounded confident, neutral, or doubtful. Participants rated how intelligible each accent sounded to them and how much they liked each accent. Neural activity was recorded using functional magnetic resonance imaging (fMRI) to understand whether in-group and out-group accents differentially activated the brain while participants judged believability. The authors used psycho-physiological interaction (PPI) analysis to determine if activity in regions involved in confidence judgments, such as the right superior temporal gyrus and the inferior frontal gyrus, was correlated with activity in other brain regions while participants made believability judgments. Finally, they examined whether these functional connections in the brain were modulated by the speaker’s accent.

What did they find?

Overall, participants were more likely to believe statements spoken in their native accent than those spoken in a regional or foreign accent. Listening to speakers with out-group accents activated temporal regions of the brain (e.g. superior temporal gyrus), whereas listening to in-group accents activated frontal regions (e.g. superior frontal gyrus). Participants also rated neutral and confidently spoken statements as more believable than doubtful statements; statement confidence was related to increased activity in the superior temporal gyrus, while statements made in a doubtful voice were associated with greater activation of temporal areas. Although confident statements were rated as equally believable across all accents, brain activity was greater in the caudate, cuneus, and fusiform regions when confident statements were heard in out-group accents. This could be because judging the believability of out-group speakers requires more in-depth processing of the vocal and acoustic characteristics of speech. PPI analysis revealed that listening to out-group accents resulted in increased connectivity between regions involved in decoding confidence – such as the superior temporal gyrus – and regions involved in inferring meaning of sentences, such as the inferior frontal gyrus. Statement believability was associated with greater connectivity between the superior temporal gyrus and the middle temporal gyrus and between the superior temporal gyrus and the lingual gyrus/middle occipital gyrus when listening to out-group speakers. These findings suggest that neural activity associated with processing voice characteristics and associated biases towards out-group speakers can be modified by the level of confidence of the speaker.


What's the impact?

The study demonstrates how distinct neural mechanisms are involved in carrying out the same decision-making process in speech perception depending on the speaker’s accent and degree of confidence. These results provide a deeper understanding of how different regions of the brain are involved in categorizing speakers based on group membership, and the effect this has on social inference from speech.

Jiang et al. Neural architecture underlying person perception from in-group and out-group voices. NeuroImage. 2018.  Access the original scientific publication here.