Toward affective brain-computer interfaces : exploring the neurophysiology of affect during human media interaction

Affective Brain-Computer Interfaces (aBCI), the sensing of emotions from brain activity, seems a fantasy from the realm of science fiction. But unlike faster-than-light travel or teleportation, aBCI seems almost within reach due to novel sensor technologies, the advancement of neuroscience, and the refinement of machine learning techniques. However, as for so many novel technologies before, the challenges for aBCI become more obvious as we get closer to the seemingly tangible goal. One of the primary challenges on the road toward aBCI is the identification of neurophysiological signals that can reliably differentiate affective states in the complex, multimodal environments in which aBCIs are supposed to work. Specifically, we are concerned with the nature of affect during the interaction with media, such as computer games, music videos, pictures, and sounds. In this thesis we present three studies, in which we employ a variety of methods to shed light on the neurophysiology of affect in the context of human media interaction as measured by electroencephalography (EEG): we evaluate active affective human-computer interaction (HCI) using a neurophysiological indicator identified from the literature, we explore correlates of affect induced by natural, multimodal media stimuli, and we study the separability of complex affective responses into context-specific and general components.