Watching Feeling: Emotional Data from Cybernetics to Social Media

This dissertation tells the story of how emotion was made computable. It traces a series of exchanges between the psychological sciences and computer science from the mid-20th century to the present, ranging from debates around cybernetic models of emotion at post-war interdisciplinary conferences, through early applications of computing to mental health at state hospitals, to the development of empathic AI agents at the MIT media lab. The history uncovered here tracks the development and mainstreaming of emotional data, by which is meant the result of the combination of psychological models of emotion and methods for recording emotional experiences, and paradigms for automating and algorithmically analyzing those recordings. The reconfiguration of emotion into data subject to algorithmic observation and intervention has ushered in a new biopolitical regime, one where these data give rise to new forms of knowledge, power, and value. The deep history in this dissertation reveals that emotion was a technical object long before contemporary algorithms learned to recognize it. At the same time, it demonstrates that the pursuit of artificial intelligence was an emotional endeavor from the earliest days of computer science, a field often considered the epitome of technical rationality.