Computational Simulation of How Emotions are Constructed in Our Brain According to the Theory of Constructed Emotion

Abstract

Understanding emotion is one of the most difficult tasks for computers. Not only is it difficult for computers, but humankind itself even has different theories of how emotions are constructed in our mind and how exactly the brain constructs feelings. Most theories about emotions say that basic emotions are genetically endowed, whereas the theory of constructed emotion states that our brain constantly uses past experience to guide our actions and generate emotional feelings and, in each situation a new instance of emotion is generated. This takes away the generic approach to classify emotions and allows us to describe emotions in multidimensional values – valence, arousal, and dominance. By describing emotional sensations on these dimensions, we can start comparing emotional states based on input and reflect the emotional state on a computer and thus improve the current state of Human-Computer Interaction (HCI). This allows creating a simulation that is not trying to find the best output from learned data by finding common features of the input and expected output but is emotionally intelligent and generates output based on its current emotional state. This thesis analyses the possibilities of the given theory about emotions and its possibilities of implementing it in HCI applications. By improving the emotional intelligence of these applications, computer programs can become better at evaluating our emotional state and act accordingly.