Emotion in the Neutral Face: Applications for Computer Vision and Aesthetics

Oftentimes “neutral” is classified as a baseline for other emotive categories such as angry, happy, or sad. Yet, neutrality has been the least studied amongst these discrete emotions. In this chapter, we focus on how neutrality is extracted, represented, and influences face perceptionFace perception and draw inferences to emotion perception in general. Neutral faces, or at least non-expressive faces, appear to be the most ubiquitous faces one encounters. Rarely does one experience a prototypical emotional face outside of a sterile lab or posed environment. This chapter explores the perceptual uniqueness of a neutral visage and the importance that understanding such an ambiguous face has across fields such as psychology and computer vision. We trace the psychological and computer science evolution of appreciating how a neutral face can inform and enhance the classification, detection, and judgment of faces that vary in characteristics such as ageAge, raceRace, and genderGender as well as incidental cues such as emotion, health, and attractiveness. This chapter summarizes classical and contemporary computer vision techniques as applied to (neutral) faces and discusses how a deeper understanding of non-emotionality is paramount to furthering the study of aesthetics, emotionality, and artistic value of people, objects, and scenes across multiple domains.