Human emotion simulation in a dynamic environment

Abstract

The aim of this work is to contribute to the believability of the simulated emotions for virtual entities to allow them display human like features. Endowing virtual entities with such features requires an appropriate architecture and model. For that, a study of emotional models from different perspective is undertaken. The fields include Psychology, Organic Components, Attention study and Computing. Two contributions are provided to reach the aim. The first one is a computational emotional model based on Scherer’s theory (K. Scherer, 2001). This contribution allows to generate a series of modifications in the affective state from one event by contrast to the existing solutions where one emotion is mapped to one single event. Several theories are used to make the model concrete. The second contribution make use of attention theories to build a paradigm in the execution of tasks in parallel. An algorithm is proposed to assess the available resources and allocate them to tasks for their execution. The algorithm is based on the multiple resources theory by Wickens (Wickens, 2008). The two contributions are combined into one architecture to produce a dynamic emotional system that allows its components to work in parallel. The first contribution was evaluated using a questionnaire. The results showed that mapping one event into a series of modifications in the affective state can enhance the believability of the simulation. The results also showed that people who develop more variations in the affective state are more perceived to be feminine.

Date of Award	Jun 2018
Original language	English
Awarding Institution	University of Northampton
Supervisor	Scott Turner (Supervisor), Philip Picton (Supervisor) & Kamal Bechkoum (Supervisor)