Knowing Your Users by Heart: A Critical Examination of the Scientific Research on Emotions Conducted by Social Media Platforms

Since their inception, social media platforms have continuously developed innovations and discourses centered on affective and emotional dimensions. Platform owners invest heavily in research and development to both identify and elicit feelings and emotions. However, the research conducted by these companies on this topic remains understudied. In this study, we present a systematic and critical analysis of the scientific literature published by the research teams of Meta (Facebook, Instagram) and Google (YouTube) regarding the analysis and processing of their users’ emotions. We first demonstrate that this literature relies on schematic definitions of emotions, thereby simplifying the complexity of affective phenomena for facilitating their automated detection. We then highlight the objectives pursued by these research efforts, the majority of which aim to promote users’ well-being through the instrumentalization of what affects them. Finally, we examine the various methods employed to identify emotions, which primarily serve to feed these platforms’ machine learning algorithms with data labeled as emotional. Beyond their scientific contributions, we consider these studies as narratives targeted at the scientific community and investors, serving to legitimize and support the development of specific innovations. Analyzing these works enables us to examine how the techniques, methods, and applications developed by these two companies commodify emotions and actively shape users’ online behaviors.