What did you study?
I studied how emotions can be modeled computationally using reinforcement learning. I developed short vignettes and experimental scenarios representing different appraisal patterns for humans to experience and rate their emotions, and compared those ratings with the model's predictions.
What were the results of your study or what is its main finding?
The computational model’s predictions closely matched human emotional responses. The findings show that emotions can be modeled as evaluations of a situation. In the model, human goals are represented as rewards, allowing it to calculate appraisals such as goal relevance, goal conduciveness, suddenness, agency, and power, etc. Based on these appraisals, the model can classify a situation into its emotional categories. My study provides a computational account of how emotions may arise from people’s cognitive evaluations of events.
How can the results be applied? What new insights did the research contribute to the topic?
The findings could support the development of social robots and interactive systems that respond more sensitively to people’s emotional situations. They may also help designers create user interfaces that take users’ emotional experiences into account. More broadly, the research shows that emotions are not simply irrational reactions, but structured and meaningful evaluations of human situations.
M.Sc. Jiayi Eurus Zhang defends their doctoral dissertation in Cognitive Science ”Appraisal-Based Computational Emotion Modeling with Reinforcement Learning” on 23rd of October at 12PM in room H320. Opponent is Professor Giulio Jacucci (University of Helsinki) and Custos is Associate Professor Jussi P. P. Jokinen.
The language of the event is English.
