On the Affective Bias of Emotional and Physiological States

The emerging paradigm of ubiquitous computing can be better realized in the presence of intelligent machines that adjust their operation based on the user’s affectively driven behavior or disposition. In this work we provide a method that can be used to investigate how two types of affective attributes influence certain types of affective computing systems. Specifically, our method compares the suitability of emotion-based attributes versus the suitability of physiology-based attributes in affective computing systems that handle media content. We implement our method and use it to evaluate affective computing systems built for a variety of relevant affective attributes. Our findings indicate that, under certain conditions, the emotion-biased systems are more sensitive than the systems which are driven by purely physiology-based attributes.