AFFECTIVE COMPUTING IN THE AREA OF AUTISM

The prevalence rate of Autism Spectrum Disorders (ASD) is increasing at an alarming rate (1 in 68 children). With this increase comes the need of early diagnosis of ASD, timely intervention, and understanding the conditions that could be comorbid to ASD. Understanding co-morbid anxiety and its interaction with emotion comprehension and production in ASD is a growing and multifaceted area of research. Recognizing and producing contingent emotional expressions is a complex task, which is even more difficult for individuals with ASD. First, I investigate the arousal experienced by adolescents with ASD in a group therapy setting. In this study I identify the instances in which the physiological arousal is experienced by adolescents with ASD (“have-it”), see if the facial expressions of these adolescents indicate their arousal (“show-it”), and determine if the adolescents are self-aware of this arousal or not (“know-it”). In order to establish a relationship across these three components of emotion expression and recognition, a multi-modal approach for data collection is utilized. Machine learning techniques are used to determine whether still video images of facial expressions could be used to predict Electrodermal Activity (EDA) data. Implications for the understandin