Digital Labor

Reading Faces in the Crowd: Postcolonial Algorithms of Affective Computing

Sat, November 15
04:30 PM - 05:15 PM
Kellen Auditorium (101)
66 5th Ave., 1st Floor


Reading Faces in the Crowd: Postcolonial Algorithms of Affective Computing
There has been a growing interest in implementing crowdsourcing technologies for affective computing problems. Crowdsourcing systems provide low-cost global cognitive labor force for training machine learning programs for various tasks including image recognition and natural language analysis. These cognitive labor apparatuses are an extension of what Bernard Stiegler calls “grammatization” through “cognitive and affective proletarianization.” This paper first traces the history of the grammatization of affect starting from the debate between Margaret Mead and Paul Ekman whose emotion classification system has been a central influence in contemporary affective computing applications ranging from video surveillance systems to sentiment analysis of consumer reviews. In contrast to Margaret Mead, Paul Ekman suggests that human emotions are universal as there is no culture-specific aspect to facial expressions according to his study of “stone-age cultures in New Guinea.” Today, Ekman’s “universal” taxonomy has been reified into affective computing agents through various mechanisms including crowdsourcing platforms such as Amazon Mechanical Turk. This standardization takes crucial characteristics as most of the crowdsourcing platforms derive the required cognitive labor from a global work force across many cultures and traditions. I argue that what is activated in these apparatuses of affective proleterianization is the postcolonial premise of locating global subject as a target of surveillance and control as well as a statistically predictable consumer/worker.