Only thirty years ago, the field of emotion research was almost unchartered territory: an almost desert island to which very few people were interested to visit and even less to stay. Neuroscientists consider the frontal lobes mostly a source of conflict from people with mental diseases, so prescription of lobotomies were quite common; emotions were, for cognitive pscychologists just a source of irrationality so nobody did pay much attention to it. So if you picked up randomly three different books of introduction to cognitive psychology, they would give you three different lists of emotional states: the first book would include hunger and sexual desire as emotions, but wouldn't include surprise. The second book included surprise and sexual desire but consider hunger just a "drive", and the third included a list of forty different emotions, most of them not mentioned in the first two. And, if the AI people were interested at all in emotions that was probably just as a trick to feign emotions in a computer in order to help it to pass a future Turing Test.
Fortunately, these perceptions have changed dramatically since then, and now the study of emotions is a very active and respected field. This change of perception is mainly due to changes in methodology and new experimental results from those three main disciplines we were just discussing.
1) Neuroscience has supplied lots of empirical data as well as some functioning models on the key role that emotion plays when we humans take decisions. Since the studies of the patient -now almost a legend- Phineas Gage we have come to realise that emotions, far from being a nuisance, are an undissociable part of most of our mental states, playing a role both in cognition and perception
2) The evolution by means of procuring better and more detailed models, as well as the collection of empirical confirmation from very different sources has turned the cognitive model of emotions from a curiosity to the more accepted and influencing model of what emotions are nowadays. This has helped greatly to introduce the concept of emotion into several research fields of cognitive psychology from which they were missing.
3) The development of affective computing has proved to be a very valuable field in AI, offering new interacting and theoretical models on what it means to be "intelligent" and how important are emotions in order to improve communications between humans and computers. At the same time, bottom-up approach towards the creation of autonomous artificial creatures, are including more an more emotional elements to their prototypes.
All this promising results as well as the emerging paradigm of affective computing has motivated us to produce this volume on synthetic emotions and social robots which you are reading now. We strongly believe that research in this field is not only important to develop better computer applications which are more able to communicate with humans or fulfill their tasks. We also argue that synthetic emotions can help not only neuroscience or cognitive psychology to test their models "in silicon"; we can gain a lot much more from this research; it can take the form of interdisciplinary models on the relationships between emotions and cognition; and it can even lead us to rethink classical philosophical problems, like the question of the reality of qualia and to consider "the hard problem of consciousness" from another point of view.
We have divided the book into the following sections, which we consider the more promising now in this field: recognizing emotions, emotional social robots, philosophical questions, modeling emotions, applied artificial emotions and ambient emotion.
State of the art papers on how we can create artificial systems to recognize emotional states in humans and use them to interact better with them, or to make predictions on groups behavior.
In Emotional Modeling in an interactive Robotic Head authors Deniz, et al describe the emotional model and implementation of CASIMIRO, a prototype social robot built by the authors. CASIMIRO is a complex robot with multimodal capabilities defined by a number of software modules, a social robot able to recognize emocions.
Automatic Detection of Emotions in Music presents the work from Laurier, Herrera in order to detect emotion in music from audio content, describing a machine learning method to do so.
This section ends with chapter Facial Expression Analysis, Modeling and Synthesis: Overcoming the LImitations of Artificial Intelligence with the Art of the Soluble by Bartneck and Lyons to review the situation in HCI with regards to the human face, and to discuss strategies, which could bring more slowly developing areas up to speed.
Emotional Social Robots
Description of methodologies, developments and theories on how to use artificial emotions in order to facilitate the development of social robots; autonomous artificial systems which are able to cooperate among them in order to fulfil specific tasks.
Multirobot Team work with Benevolent Characters: The Roles of Emotions. by Banik, et al describes an emotional model and stratgey in order to make cooperative robots to work between there better, including the concept of "benevolent character" in the modelling, while Affective Goal and Task selection for Social Robots written by Scheutz and Schemerhorn present a DIARC or distributed integrated affect cognition and re?ection architecture designed to help social robots in order to take decisions.
Authors Lee-Johson and Carnegie present Robotic Emotions: Navigation with feeling present a mobile robot navigation system that employs affect and emotion as adaptation mechanisms. The robot’s emotions can arise from hard-coded interpretations of local stimuli, as well as from learned associations stored in global maps.
Following the X-Phi or Experimental Philosophy paradigm, this chapter is devoted to show how classical and recent philosophical conundrums can be shed a new light when they are considered from the of view of the artificial emotions field.
According to Gros in Emotions, diffusive emotional control and the motivational problem for autonomous cognitive systems intelligent systems need to include motivational procedures, showing therefore the importance of emotional control for the design of artificial intelligences and synthetic cognitive systems.
Robots React, but Can they Feel? allows author MacLennan to analyse the famous "hard problem of consciousness" within the context of synthetic emotions, trying to understand whether it is possible to consider a robot able to feel them.
The chapter Personality and Emotions in Robotics from the Gender Perspective by García-Ordaz et al seek to apply the gender perspective in the analysis of some emotional features to be taken into account before they are applied to the field of robotics.
Philosopher Gomila presents in Moral emotions for autonomous agents the moral implications related to the idea of an autonomous robot and how to deal with the nightmare of the "evil robot" that looses control.
Besides being useful to solve or improve certain tasks within the AI domain, artificial emotions are also an important instrument in order to simulate emotional processes and develop better models of what an emotion is which can be of great help for several cognitive science disciplines.
An emotional perspective for agent based computational economics by Cipresso, et al present an analytical model of hyper-inflated economies and develop a computational model that permits us to consider expectations of the levels of future prices following emotional rules and strategies.
Aubé show us in Unfolding commitments management: A systemic view of emotions a model of emotion which is more realistic that the usual ones, in order to develop artificial emotional applications that are really functional.
In A Cognitive Apprasial Based Approach for Emotional Representation author Rodríguez et al presents a model for synthetic emotions based on the cognition/apprasial theory, while Raïevsky and Michaud present the usefulness of the signaling function of emotion for situated agents and an artificial model of anger and fear based on mismatch theories of emotion in their chapter Emotion Generation Based on a Mismatch Theory of Emotions for Situated Agents.
Macedo at all presents in Artificial Surprise a review of the models of surprise, comparing several of them, indicating future research and possible practical applications. Based on Arabic word roots, author Adi describes a linguistically based theory of emotions in A theory of Emotions Based on Natural Language Semantics.
Applied artificial emotions
Recently, we have seen an important increase in the use of emotion in order to help computers to solve simple tasks: from passing the Turing Test to create more friendly and useful learning environments, as we can see from the materials of this chapter.
In this chapter by Shah, A downward trend for machines in recent Loebner Prizes argues on how important is to include some ability to use and recognize emotions if we want an AI to pass the Turing test, and The Use of Artificial Emoitional Intelligence in Virtual Creatures by Ramos, et al argues on the importance of emotions in order to design a proper avatar in order to model ts perception, learning, decision process, behavior and other cognitive functions.
The chapter by Psycharis Physics and Cognitive-Emotional-Metacognitive Variables-Learning Performance in the Environment of CTAT applies artificial emotion to teach mechanics, while author Gorga in the paper Computer-Based Learning Environments with emotional agents purpose discusses conceptual issues and challenges related to the integration of emotional agents in the design of computer-based learning environments and to propose a framework for the discussion of future research. Finally, Emotional Memory and Adaptive Personalities by Francis et al present an artificial intelligence model inspired by these psychological results in which an emotion model triggers case-based emotional preference learning and behavioral adaptation guided by personality models.
Both from a theoretical as well as from a practical point of view, the field of Ambient Intelligence -or how to include some sort of intelligent ability to interact with humans and solve specific tasks in a distributed system- can gain a lot by means of including artificial emotions in their modeling. Here we explain how this can be done.
In Emotional Ambient Media Lugmayr et at introduces the reader to a technical oriented view towards recognizing, simulating, and binding emotions in ambient media systems, as well as presenting a case-study for an emotion recognition and response system.The Modelling Hardwired Synthetic Emotions: TPR 2.0 by the editors of this book describes an ambient intelligence system which uses protoemotions in order to respond to specific actions fromo the user. The book ends with Invisibility and Visibility: the shadows of artificial intelligence by Crutzen and Hein which analyses how the mental, physical, and methodical invisibility of artificial intelligent tools and environments will have an effect on the relation between the activities of both, users and designers.