Evaluating Empathy in Artificial Agents

Abstract

Empathy is a complex socio-emotional behavior that results from the interaction between affective and cognitive mechanisms. Equipping embodied conversational agents (ECAs) with empathic capacity can benefit from the integration and evaluation of these low and high level capabilities in a hierarchical manner. Following the theoretical background on empathic behavior in humans, this paper presents a framework to equip ECAs with real time multi-modal empathic interaction capabilities. We present the implementation of this framework, which includes basic dialogue capabilities as well as three levels of empathic behavior in a conversational scenario. Our approach is an inclusive stand on modeling levels of empathy and provides a baseline behavior for empathic interaction.