Similar Representations of Emotions across Faces and Voices. / Kuhn, Lisa; Wydell, Taeko; Lavan, Nadine; McGettigan, Carolyn; Garrido, Lucia.

In: Emotion, 02.01.2017.

Research output: Contribution to journalArticle

Forthcoming

Documents

Abstract

Emotions are a vital component of social communication, carried across a range of modalities and via different perceptual signals such as specific muscle contractions in the face and in the upper respiratory system. Previous studies have found that emotion recognition impairments after brain damage depend on the modality of presentation: recognition from faces may be impaired whilst recognition from voices remains preserved, and vice versa. On the other hand, there is also evidence for shared neural activation during emotion processing in both modalities. In a behavioural study, we investigated whether there are shared representations in the recognition of emotions from faces and voices. We used a within-subjects design in which participants rated the intensity of facial expressions and non-verbal vocalisations for each of the six basic emotion labels. For each participant and each modality, we then computed a representation matrix with the intensity ratings of each emotion. These matrices allowed us to examine the patterns of confusions between emotions and to characterise the representations of emotions within each modality. We then compared the representations across modalities by computing the correlations of the representation matrices across faces and voices. We found highly correlated matrices across modalities, which suggest similar representations of emotions across faces and voices. We also showed that these results could not be explained by commonalities between low-level visual and acoustic properties of the stimuli. We thus propose that there are similar or shared coding mechanisms for emotions which may act independently of modality, despite their distinct perceptual inputs.
Original languageEnglish
JournalEmotion
StateAccepted/In press - 2 Jan 2017
This open access research output is licenced under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.

ID: 27560060