Physiological Detection of Affective States in Children with Autism Spectrum Disorder

Abstract

Autism spectrum disorder (ASD) is associated with emotion processing difficulties, including limitations in understanding the emotional states of others and processing one's own internal experiences. The nature of these difficulties remains largely unknown. This is due, in part, to challenges in acquiring reliable self-reports of emotional experiences from this population. Automatically characterizing emotional states with the use of physiological signals is a potential means of overcoming this problem, as physiological signals can provide an objective and nonverbal method for assessing affective states. However, this approach has not been well considered with ASD to date. To this end, we investigated detection of autonomic responses to positive and negative stimuli in children with ASD using four physiological measurements. Electrocardiograms, respiration, skin conductance and temperature were measured while 15 children with ASD viewed standard images known to evoke varying levels of valence (positive and negative) and arousal (low and high intensity). Using an ensemble of classifiers, affective states induced by stimuli of positive and negative valence or high and low arousal was differentiated at average accuracies approaching or exceeding 80 percent. These results suggest the feasibility of discerning affective states in individuals with ASD objectively using physiological signals.