The selfie, a digital self-portrait photo, has become a unique means of self-presentation, with the emergence of image-based social media platforms, such as Instagram. Social cues presented in selfies may influence viewers to form and evaluate impressions toward uploaders, resulting in viewers’ specific social behaviors, as in off-line face-to-face situations. The goal of this study is to understand how facial features, especially facial expressions of emotion, which are known as representative social cues of nonverbal communication, can affect social engagement. We analyze a large-scale set of selfies on Instagram by applying a deep learning-based image analysis algorithm to discover how facial emotions affect the extent to which selfies induce likes and comments.We also conducted in-depth interviews with 14 Instagram users to better understand the underlying motivations of social responses to selfies. Our findings are as follows: (1) Selfies with positive emotional valence invited more frequent likes and comments than those with negative emotional valence. (2) However, our findings also reveal that the effect of facial emotion on social engagement is more complex than mere valence alone. The effect of emotion on social response differed depending on the type of emotion. Happy faces got more likes and comments; surprised faces got fewer likes and comments. (3) Viewers’ implicit motivations for social feedback on selfies were revealed to be impression evaluation and relationship enhancement. Based on these findings, we discuss design implications for driving user response in image-based social media.
Seoul National University
College of Social Sciences