# HICEM: A High-Coverage Emotion Model for Artificial Emotional Intelligence

**Authors:** Benjamin Wortman, James Z. Wang  
**arXiv:** [2206.07593](https://arxiv.org/abs/2206.07593)  
**Date:** 2022-06-15  
**Summary page:** https://www.summarizepaper.com/en/paper/2206.07593v1/

## Key points

- Affective computing is advancing as social robots and intelligent machines enter the home, leading to a demand for artificial emotional intelligence (AEI) that can facilitate deeper human-machine interactions.
- Next-generation AEI requires comprehensive human emotion models for training.
- Emotion models are descriptive tools that need robust coverage to be effective in facilitating communication between humans and machines.
- Existing emotion models are insufficient for real-world applications and require further refinement.
- Researchers developed a new data-driven approach leveraging existing large-scale emotion recognition datasets to assess the validity of existing emotion models in relation to human perception at scale.
- The resulting high-coverage cross-cultural model of emotion (HICEM) provides a more efficient model of human emotion that is foundational for developing real-world affective computing applications.
- The study has broad implications in social robotics, human-machine interaction, mental healthcare, and computational psychology.
- The paper also covers related work on basic emotions theory, continuous models, and componential models in relation to affective computing and AEI.
- This research was supported by generous gifts from the Amazon Research Awards program and computational resources from the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant No. ACI-1548562.

## Summary

The field of affective computing is rapidly advancing as social robots and other intelligent machines enter the home, leading to a growing demand for artificial emotional intelligence (AEI) that can facilitate deeper and more meaningful human-machine interactions. To achieve this, next-generation AEI requires comprehensive human emotion models for training. Unlike theory of emotion, which has been the historical focus in psychology, emotion models are descriptive tools that need robust coverage to be effective. This means defining the smallest core set of emotions from which all others can be derived. To address this challenge, a team of researchers developed a new data-driven approach leveraging the annotations of existing large-scale emotion recognition datasets to assess the validity of existing emotion models in relation to human perception at scale. The study found that existing models of emotion are insufficient for real-world applications and require further refinement. The research team used unsupervised clustering techniques to identify 15 components with minimal overlap and maximum coverage across six major languages: Arabic, Chinese, English, French, Spanish, and Russian. The resulting high-coverage cross-cultural model of emotion (HICEM) provides a more efficient model of human emotion that is foundational for developing real-world affective computing applications. The study has broad implications in social robotics, human-machine interaction, mental healthcare and computational psychology. It highlights the importance of comprehensive and robust emotion models in facilitating effective communication between humans and machines. The paper also covers related work on basic emotions theory, continuous models and componential models in relation to affective computing and AEI. This research was supported by generous gifts from the Amazon Research Awards program and computational resources from the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant No. ACI-1548562. The authors are grateful to Yu Luo for their contributions to this project.