Emotion Modelling

Emotion Modelling

Mehdi Dastani and John-Jules Ch. Meyer

Emotions are crucial to both and agent's believable behavior as well as making the right decisions.

Autonomous software agents are assumed to have different (possibly conflicting) objectives, able to sense their environments, update their states accordingly, and decide which actions to perform at any moment in time. The decision behavior of such software agents can be effective, practical and realistic only if they are able to continuously and adequately assess their (sensed) situation and update their states with relevant information and crucial objectives.

Emotion is generally understood as a (cognitive) mechanism that directs an agent’s thought and attention to what is relevant, important, and significant. Such a mechanism is crucial for the design of software agents that must operate in highly-dynamic, semi-predictable environments and which need mechanisms for allocating their computational resources efficiently. For example, a robot that aims at transporting a container may perceive its low battery charge. With an emotion-based decision module the robot may assess this as being relevant for its objective and decide to suspend its transportation, or to  focus on a less battery demanding transport objective. Besides this emotions are also crucial to obtain so-called believable behavior in applications such as serious games / training programs, where NPCs should react in a more realistic way, so that the trainee can learn to behave in a way that is better aligned with reality.

adapted from: Programming Agents with Emotion, Dastani et al. (2014)

Our group works on the foundations of intelligent agents, in particular logical theories to describe or specify software agents that are autonomous in the sense that they make their decisions based on their internal states and available options. Inspired by psychological theories of emotion, we develop formal and computational models of emotions, and integrate them in the software agent models. We investigate and formalize various aspects of emotions such as appraisal, experience and coping. We instantiate the emotion-based agent models to develop various applications such as emotional NPCs in serious games, e-coaches for lifestyle change/maintenance and treatment of diseases, and human-robot interactions.

For this we look out for sponsors and partners.