Artificial shame models for machines?

© 2014 by Nova Science Publishers, Inc. All rights reserved. The purpose of this chapter is to provide a reliable conceptual model of shame for artificial cognitive systems which must or may not be able to interact with human beings. We are considering here the necessity or not of implementing shame into future social artificial systems, evaluating the usefulness of this emotion from the past evolutionary and social trends and projecting this information towards the future of AI. If it is true that affective computing, social robotics or Human-Robot Interaction (HRI) have been fields interested on creating models or cognitive architectures that allow better human-machine interactions, in no case these researches have considered the reliability of the emotional model or the number of implemented emotions that will be useful for a better performance of these machines. Emotional architectures have been focused on more basic models of emotions (for example Ekman's 6 basic emotions) and shame has been neglected in most of research cases by researchers of robotics, computer sciences or AI. Will be justified here the necessity of this emotion for artificial devices as well for the future of human-machine interactions.