Simulating the Emotion Dynamics of a Multimodal Conversational Agent

Abstract

Abstract. We describe an implemented system for the simulation and visualisation of the emotional state of a multimodal conversational agent called Max. The focus of the presented work lies on modeling a coherent course of emotions over time. The basic idea of the underlying emotion system is the linkage of two interrelated psychological concepts: an emotion axisrepresenting short-time system states -and an orthogonal mood axis that stands for an undirected, longer lasting system state. A third axis was added to realize a dimension of boredom. To enhance the believability and lifelikeness of Max, the emotion system has been integrated in the agent's architecture. In result, Max's facial expression, gesture, speech, and secondary behaviors as well as his cognitive functions are modulated by the emotional system that, in turn, is affected by information arising at various levels within the agent's architecture.

A Conversational Agent as Museum Guide – Design and Evaluation of a Real-World Application

“…The emotional state is available anytime both in continuous terms of valence and arousal as well as a categorized emotion, e.g. happy, sad or angry, along with an intensity value (see [ 1]). The continuous values modulate subtle aspects of the agent's behaviors, namely, the pitch and speech rate of his voice and the rates of breathing and eye blink.…”

“…The emotion system, in turn, receives input from both the perception (e.g., seeing a person immediately causes positive stimulus) and the deliberative component. For example, obscene or politically incorrect wordings ("nowords") in the user input leads to negative impulses on Max's emotional system (see [ 1]). Since subsequent stimuli in the same direction accumulate in the emotion system, repeated insults will put the agent in an extremely bad mood, which in turn can eventually result in Max leaving the scene, an effect introduced to de-escalate rude visitor behavior.…”

Embodied Cooperative Systems: From Tool to Partnership

“…; internal: goal achievement or failure, resulting in positive resp. negative emotions), and which defines the agent's explicit emotional state over time in pleasure-arousal-dominance (PAD) space (Becker et al 2004). The agent is controlled by a cognitive architecture (Sect.…”

“…2) is negatively influenced by inputs containing ungracious or politically incorrect wordings ("no-words") which, when repeated, can eventually trigger a plan causing the agent to leave the display and stay away until the emotion has returned to a balanced state (an effect introduced to deescalate rude visitor behavior in the museum). The period of absence can either be shortened by complimenting Max or extended by insulting him again, see (Becker et al 2004). Altogether, this kind of behavior of Max may be taken as beginnings of moral judgement.…”

„Ich, Max“ – Kommunikation mit Künstlicher Intelligenz

“…Parallel dazu werden die Stimmungsvalenzen, die kontinuierlich in einem dreidimensionalen abstrakten Raum verlaufen, kategorisiert und als explizite Überzeugungen (beliefs) symbolisch repräsentiert; so können sie bei der Auswahl zwischen Handlungsoptionen zum Tragen kommen (vgl. Becker et al 2004). Die symbolisch repräsentierten Emotionszustände kann Max auch verbal äußern ("jetzt bin ich ärgerlich"); in diesem Sinne scheint sich Max ihrer "bewusst" zu sein.…”