Exploring Contextual Cues for Audio-visual Emotion Recognition
What are specific cues and traits within the sender’s context that shape human emotion expression and how do these cues influence and bias emotion perception?
G.R. Ataya (TU Delft - Electrical Engineering, Mathematics and Computer Science)
S. Mukherjee – Mentor (TU Delft - Electrical Engineering, Mathematics and Computer Science)
B.J.W. Dudzik – Mentor (TU Delft - Electrical Engineering, Mathematics and Computer Science)
S. Tan – Graduation committee member (TU Delft - Electrical Engineering, Mathematics and Computer Science)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Computational systems are increasingly integrated into people’s daily lives. As this continues, proper emotional detection is needed for human-compatible intelligent systems (Russell, 2021). As the universality theory of emotions—the idea that human emotional expressions are consistent and universal across cultures—becomes less dominant in scientific discourse, context is increasingly recognised as important for emotion recognition.
Using the SALSA methodology, this systematic literature review explores sender context and attempts to answer the following questions: What specific cues and traits within the sender’s context shape emotional displays? Through which mechanisms do these cues and traits influence how emotion is perceived?
We anticipated that the four most impactful context categories emerging from the literature would be age, culture, gender, and personality. We found substantial evidence of the impact of age, culture, and gender on emotional expression and recognition through four key mechanisms: emotion suppression and display rules, in-group advantage, socially acquired stereotypes, and physiological differences. Personality was excluded due to insufficient literature. We discuss the implications of these findings for psychology and the development of context-aware emotion recognition (CAER) systems.