Workshop on Embracing Ambiguity and Subjectivity in Emotion Research (EASE)

Emotion research and affective computing involve tasks like describing, predicting, and explaining emotional reactions to stimuli such as text, images, and audio. A prevalent assumption in this field is that emotional reactions can be adequately represented by a single numerical value or label. However, we contend that variability is common in this domain due to the “subjectivity” (individuals can have vastly different responses to the same stimulus) and “ambiguity” (the same response could be described in multiple, equally valid ways) inherent in emotional processes. These properties make assumptions of a singular true description for responses highly problematic for research and applications, such as emotion recognition, affective content analysis, or generating affective behavior in robots or virtual agents. Consequently, we are convinced that considering subjectivity and ambiguity is crucial and that doing so adequately will require a deeper understanding of these phenomena and the development of alternative practices (e.g., modeling responses as distributions over several viable outcomes, or using contextual variables for disambiguation).

This workshop aims to increase awareness of and appreciation for the importance of ambiguity and subjectivity in the affective computing community. In particular, we hope to facilitate focused discussions on relevant issues and support future research on these important concepts by providing a multi-disciplinary forum for interested researchers and practitioners.

Topics of interest for the workshop are position papers touching on one of the following KEY THEMES:

• Theoretical Foundations: Discussions and refinement of terminology related to ambiguity and subjectivity, as well as integrations of relevant perspectives from areas like psychology, statistics, and philosophy.
• Database and Annotation Design: Proposals, discussions, or comparisons of different approaches to data collection and annotation to facilitate the study of ambiguity and subjectivity, e.g., stimulus types, measurement instruments, annotation software, and emotion representations.
• Modeling and Evaluation Approaches: Proposals, discussions, or comparisons of different approaches to modeling and model evaluation that consider ambiguity or subjectivity.
• Future Challenges and Opportunities: Discussions of questions, directions, and applications for future research on ambiguity and subjectivity to explore.

Important Dates

All deadlines are at the end of the day in the GMT-12 timezone.

Program

Time slots will correspond to the local time in Glasgow, UK.

Location: TBA