Creating Synthetic Emotions through Technological and Robotic Advancements

Creating Synthetic Emotions through Technological and Robotic Advancements

Jordi Vallverdú (Universitat Autònoma de Barcelona, Spain)
Release Date: May, 2012|Copyright: © 2012 |Pages: 233
ISBN13: 9781466615953|ISBN10: 1466615958|EISBN13: 9781466615960|DOI: 10.4018/978-1-4666-1595-3

Description

As humans interact more often and more intimately with computers, and as computational systems become an ever more important element of our society, playing roles in education, the production of culture and goods, and management, it is inevitable that we should seek to interact with these systems in ways that take advantage of our powerful emotional capabilities.

Creating Synthetic Emotions through Technological and Robotic Advancements compiles progressive research in the emerging and groundbreaking fields of artificial emotions, affective computing, and sociable robotics that allow humans to begin the once impossible-seeming task of interacting with robots, systems, devices, and agents. This landmark volume brings together expert international researchers to expound upon these topics as synthetic emotions move toward becoming a daily reality.

Topics Covered

The many academic areas covered in this publication include, but are not limited to:

  • Affective computing
  • Cognitive Science
  • Emotional computing
  • Engineering approaches to emotions
  • Evolutionary approaches to the mind
  • Humanoid Robotics
  • Language and emotions
  • Modeling artificial emotions
  • Sociable robotics
  • Synthetic Emotions

Reviews and Testimonials

Every published work contains rich and different approaches to artificial emotions. Somebody could think that this diversity could lead to a lack of agreement or that this is the example of the dominant confusion in this field. On the contrary, each contribution sheds light upon the multiple faces of the studies on synthetic emotions; all the authors, although not connected directly, contribute to the improvement of the understanding and implementation of synthetic emotions. This is a collective creation of knowledge in which all the involved researchers are designing the basis of a shared future independently.

– Jordi Vallverdú, Universitat Autònoma de Barcelona, Spain

For anyone who is forward-thinking, bold and just excited about opening up new frontiers, this is the book to read. Much is revealed in it that is new and unearthed through research.

– Nano Khilnani, BizIndia Book Reviews

Table of Contents and List of Contributors

Search this Book:
Reset

Preface

“SUBSUMING OR EMBODYING EMOTIONS?”

STILL…WHAT IS A SYNTHETIC EMOTION?

There is a question with which I’m usually faced when I give talks on emotions and cognitive systems: “what is a synthetic emotion?” Everybody knows about feeling and emotion, but the doubts start when one tries to understand the bodily meaning, evolutionary reasons, or physiological details of emotions, and even more, what is the relationship between emotions and machines? Another important aspect is that the kind of audiences interested on the field of artificial emotions is very broad: AI experts, engineers, electronic artists, neurologists, robot producers, game programmers, computational theoreticians, psychologists....and despite their interest on how to relate emotions, social interaction, and intelligence (or intelligent task-solving designs), there has never been a clear idea about the history and highlights of this very short historical research field. The author of this preface can understand this confusion; emotions have not been completely implemented into the analysis or design of cognitive entities. And even in these cases in which something like emotions have been taken into account, they have been added to the system, instead of being the structural backbone of the cognitive system.

As a consequence of all the previously explained, it can be affirmed that the research on artificial emotions has two main and limited interests:

a) recognize and imitate human emotions;
b) consider the benefits of a emotion-like architecture or functioning (neural nets) for a cognitive system.

The truth is that the high complexity of the cognitive brain processes makes impossible to think about, emulate, or even to simulate those activities. Nevertheless, it is obvious that at the end, the real goal of the research on artificial emotions is to create machines that are able to feel. Not to feel as human beings, in the same way that airplanes don’t fly like birds…but much faster and carrying heavy things, like cargo aircrafts. The future of synthetic emotions cannot be to design human-like robots able to interact in human environments, or AI programs with advanced cognitive skills as a consequence of some implemented emotional architecture. Machines will need to have emotions in order to survive, to fulfil the deep meaning of human existence, to know and to feel fine with this (partially) satisfied emotion. Allow the preface to explain briefly how this author thinks this process can start.

SUBSUMING OR EMBODYING EMOTIONS?

There is a very important problem (or, better, pending question) in the current research on artificial emotions: when one looks at the bibliography on this topic, they can observe a common tendency to consider emotions like something added to the artificial cognitive systems, but never natural or prewired to them. In most of published research on artificial emotions, the situation is like this: “well…if we implement an emotional-drive mechanism this machine will work better” or “with emotion recognition and expression this robot will perform better human interactions,” because those artificial systems are not considered autonomous entities. Instead, those devices are designed as machines that perform tasks or run programs, even evolutionary ones, but they are never considered as real entities. Emotions are then subsumed into some secondary conceptual layer. Human physical structure determines whole behaviour: what can be thought, what can be touched or felt, where to go, and so on. Only thanks to extended instruments (telescope, computers, cars...) can humans go beyond the limits imposed by natural evolution. Even in the case of extended parts of human minds and bodies, people still project their physical structure over them, or human emotional nature. People live for several things: be loved, feel good, satisfy physical necessities, avoid pain, understand one’s surrounding nature….and humans are driven by these intentional directions. Although all throughout man’s lives, he tries to minimize pain and maximize pleasure, he cannot choose pain or pleasure: the body is prewired in these modes. Man cannot escape from hunger, fear, or future death. Perhaps humanity can learn to assume some emotional moods thanks to symbolic processes (death is nothing, because I can only feel me while I’m alive, as Epicurus said), but even in that case, the discussion is about emotions inside cognitive-processed behaviours. Any social activity involves several types of emotional regulations. If the aim is to understand at which point this is important, look at people with Asperger’s syndrome or autism, and it is apparent how emotions are absolutely necessary for normal human daily activity: from performing scientific research (so often in coordinated teams) to social interactions (friendship, relationship, hobbies, ….).

Despite all the previous statements, this author can understand the reason of the lack of understanding, that is, denial of emotions into general research fields: on the one hand, science still doesn’t understand the true nature and role of emotions in human cognition, and therefore, man cannot reproduce them exactly; on the other hand, and despite the last two decades of empirical neurological evidence towards the active involvement of emotions into cognitive tasks, emotions are still considered by hundreds of experts as neither necessary nor primordial. Consequently, emotions are secondary aspects in the main design of any kind of an action regulation system.

Consequently, a change is necessary. Emotions cannot be something that should be embedded into a pre-existing artificial device, but they must be embodied within them. Emotions are at the same time part of the physical body and the informational process of the body sensors. Human bodily structure allows man to feel the world under a specific meaning route, the emotional one. Embodying emotions is the real task, future, and goal of the whole research field. At the end, the field needs to create feeling machines. Yes, machine learning, integration of perceptive data, data categorization, action, and goal selection or communication among agents, among others, are important and unsolved problems of the contemporary research on cognitive AI. Because of the impossibility of solving the whole problem of human cognition, researchers have fragmented it into smaller, more tractable pieces, trying to achieve easier solutions to them. And this is the current state of the research, necessary but unsatisfactory. The convergence of all these fields led to the artificial conscious existence, something for which will be absolutely necessary in creating synthetic emotions.

IJSE’S ROLE AND CURRENT STATE

The International Journal of Synthetic Emotions is the first and leading journal in the world devoted specifically to the artificial or synthetic emotions researches. As it states on the website, the mission of IJSE is to provide a forum for the advancement of knowledge and methods necessary for the creation of artificial devices with emotions. IJSE approaches the field of synthetic emotions, offering a unique interdisciplinary platform for all international researchers on this topic. The journal presents a new common space of the richest and best ideas about synthetic emotions. Also discussed is a conceptual framework that enables a synergy and symbiosis among computer scientists, cognitive scientists, robot and synthetic agent designers, as well as psychologists, neuroscientists, and philosophers. 

After two years of the journal, it can be affirmed that this editorial project is a solid reality and that IJSE accomplishes a new role as the conceptual meeting point between the several types of experts of this interdisciplinary research area. Every published work contains rich and different approaches to artificial emotions. Somebody could think that this diversity could lead to a lack of agreement or that this is the example of the dominant confusion in this field. On the contrary, each contribution sheds light upon the multiple faces of the studies on synthetic emotions; all the authors, although not connected directly, contribute to the improvement of the understanding and implementation of synthetic emotions. This is a collective creation of knowledge in which all the involved researchers are designing the basis of a shared future independently.

The journal’s first year had great contributions that the author of this preface comments on briefly below.

(a) Modelling the Experience of Emotion (pages 1-17) was written by Joost Broekens (Man-Machine Interaction Group, Delft University of Technology, The Netherlands). Accepting the challenge of thinking about a new paradigm for synthetic emotions, Prof. Broekens made a critic review of the basic prevailing ideas on the field and suggested how emotions could be modelled. This is a brave attempt to face the pending problems of the field as well to clarify conceptual problems that could otherwise add confusion. Finally, he reinforced the idea of considering emotion as an experience, and consequently, presented ideas on how computational modelling of emotion cold help to better understand the experience of emotion, providing evidence that several computational models of emotion already addressed the issue.

(b) Review of Kansei Research in Japan (pages 18-29) was written by Seiji Inokuchi (Takarazuka University of Arts and Design, Japan). This paper gave light to a forgotten and key aspect of the engineering implementation of emotional aspects: Japanese Kansei research. Despite the continuous and excellent relationships between Western and Japanese AI & Robotics researchers (for example I’ve enjoyed recently from a JSPS fellowship), the general literature on synthetic emotions neglects any direct reference to Kansei Engineering. This paper was a first step towards a true connection between Eastern and Western specialists.

(c) Simplifying the Design of Human-Like Behaviour: Emotions as Durative Dynamic State for Action Selection (pages 30-50) was written by Joanna J. Bryson (University of Bath, UK) and Emmanuel Tanguy (University of Bath, UK). The authors created a new model, the Dynamic Emotion Representation (DER), which integrates emotional responses and keeps track of emotional intensities changing over time. Besides demonstrating their system with a virtual actor, they also demonstrate how even a simplified version of this representation can improve goal arbitration in autonomous agents; i.e., more simple, more reliable.

(d) Emotion as a Significant Change in Neural Activity (pages 51-67) was written by Karla Parussel (University of Stirling, Scotland). Working on neural nets, specifically feed-forward networks of leaky integrate-and-fire neurons, Prof. Parussel tested a hypothesis according to which neuromodulators act as signals to actions than can be labelled as driving or satisfying emotions (two classes of emotions). This paper is a good example of how theoretical literature can be experimentally checked by computational means.

(e) Automatic, Dimensional, and Continuous Emotion Recognition (pages 68-99) was written by Hatice Gunes (Imperial College London, UK) and Maja Pantic (Imperial College London, UK and University of Twente, EEMCS, The Netherlands). One of the most important aspects of the interest of human emotions is the emotion recognition process, which makes possible a true HRI. This paper shows how natural emotions (and not a simplistic and short idealized model of them) can be detected and processed, from the help of several techniques they analyze: dimensional and continuous affect modelling, sensing, and automatic recognition from visual, audio, tactile, and brain-wave modalities.

(f) Emotion in the Pursuit of Understanding (pages 1-11) was written by Daniel S. Levine (University of Texas at Arlington, USA) and Leonid I. Perlovsky (Harvard University, USA). From the knowledge of their neural and cognitive modelling expertise, the authors suggest a neural network model in which emotional and rational processes are intertwined. At the same time, they suggest from the existing literature, a beautiful, powerful, and simple hypothesis: humans have a knowledge instinct that drives them to make sense of their environments, that is, a genetic aesthetic driving force behind cognitive processes.

(g) Chatterbox Challenge as a Test-Bed for Synthetic Emotions (pages 12-37) was written by Jordi Vallverdú (Universitat Autònoma de Barcelona, Spain), Huma Shah (Universitat Autònoma de Barcelona, Spain), and David Casacuberta (Universitat Autònoma de Barcelona, Spain). The Turing test involves a semantic knowledge of the world, but the common mistake of all researchers who try to win the Turing test with their chatbots is to consider semantics as a linguistic property of things, instead of the emotional interaction between words and feelings (from specific bodies). This paper showed the unsatisfactory results of checking the performances of several chatbots and how the emotional drive of these artificial entities must be emotionally designed in order achieve any human-like appearance.

(h) Effects of Polite Behaviors Expressed by Robots: A Psychological Experiment in Japan (pages 38-52), by Tatsuya Nomura (Ryukoku University, Japan), and Kazuma Saeki (Ryukoku University, Japan). Going one step beyond the habitual difficulties involved into robotics research, humanoid robotics is faced with several problems, most of them the result of the complexity of human social interactions. This experiment analyzes how a robot can express polite behaviours and which is the human answer to this action. 

All these excellent pieces of research are a perfect example of the latest and most innovative approaches to the multidimensionality of artificial emotions, and in some cases are very brave and critical with their own fields. The authors are always looking for a better research future, always creating new roads for knowledge.

OBJECTIVES

The objectives of IJSE Editorial Board for the future will be to maintain this journal as an open intellectual platform for all researchers who think on how natural or artificial entities make decisions and act. As this author wrote in the preface of the first issue “In a world of super-specialised experts, this journal aims to be a bridge and meeting space for the different researchers who work in the field of synthetic emotions: neurologists, computer scientists, robotic engineers, philosophers, artists, and so forth. The advancement of science comes from those who are brave, who look for new thinking spaces, who create new ideas. We are at the frontier of true knowledge and inspired work. The suggested diversity of approaches will not imply a lack of depth of analysis, nor a simplification of the ideas or concepts involved. We wish to build a complex editorial project, which facilitates such a meeting point between different kinds of experts.”

The author of this preface further posits is possible to observe repeatedly a general behaviour when making talks to several audiences (graduate, undergraduate, postdocs) from several backgrounds (engineering, humanities, social sciences, cognitive,..): 

  1. At the beginning the same faces of agreement or boredom... “yes, I know that there is a relationship between emotions and rationality, but what the H*** can a philosopher show me?”
  2. In the middle of a lecture there are plenty of interested persons following along with the ideas and bad jokes, some others who clearly are fighting themselves about the meaning of some new ideas they’ve heard, and finally, a few who have experienced an illumination about some weird aspect of their research for which they think they’ve found a new solution. Obviously, the sleeping students in the population are not counted under this analysis, as well as those skeptics who, because a feeling of disgust towards bearded philosophers like me, (and following a new fallacy (ad bearded philosopher hominem) have not followed any of the presented ideas, considering that they have all the knowledge they need to solve their academic problems.
  3. At the end: some quickly abandon the room, some others discuss some of the exposed ideas, another group take the opportunity of an office visit to talk pleasantly with their colleagues about their last weekend, and a few come to me to ask for PPT slides and to establish a short discussion about some part of the talk (surely, they think that they need to offer some words to justify getting the full files for the lecture). Some hours later, emails will be received asking for some details about my talk. And then, back again to the silence of one’s desk.

The students are all following specific and traceable emotional dispositions that are translated into different kinds of attitudes and cognitive tendencies. It is difficult to see when talking about emotions that students and the professor are feeling them all throughout the process. There is no distance between the world, the words we use to express it, and the self. Emotions are the force that bridges these domains of reality. People are faced with the embodied nature of emotions, as the end point of understanding, to the fulfilment of knowledge necessities. 

For all the previous aspects, it is necessary to maintain a true interdisciplinary journal in which experts from several domains of expertise can share, learn, discuss, or imagine new ways of creating artificial emotions. This is a complex project; perhaps researchers in this field swim against the contemporary tide of overspecialization, but it is necessary to be open to different ideas to solve the complexities of the emotional design.

FUTURE TRENDS

The future of IJSE is in the hand of the massive community of researchers who are more or less involved into the analysis of emotions, especially those who are trying to design more humanoid and/or intelligent devices. Nevertheless, as Editor-in-Chief, the author of this preface has some bureaucratic duties as well as research goals. Among these, the editor in chief must encourage two different kinds of issues, published alternatively: a) normal issues with invited or open calls collected papers; b) specialized numbers about one topic with a guest editor.

It is very important to be open to new ideas, new debates, or new points of view from young, senior, acclaimed, or unknown researchers in pursuit of the highest quality and the most innovative approaches, because synthetic emotions are a very complex problem. The truth is that complexity, although present, it is not the most important aspect of synthetic emotions: there is the pervasiveness of emotions as a qualitative and subjective experience, and there also is the difficulty of creating the main architectural design of an emotional device. Qualia and emergence are perhaps the two keystones of the future research of emotions, because in order to create emotions, it is not enough to talk about simulating some facial expressions, but about how to create a machine that feels the world according to its own structure. This is the real meaning of embodiment. In the process of creating such a machine, it is vital to start from the scratch reproducing basic cognitive systems (plants, insects…) and increase the complexity progressively, looking for the emergence of more complex emotions, and consequently, of better cognitive systems and consciousness itself. To be is to feel, in any kind of reality that your brain wishes to offer to you. Nervous system cognition and emotions are several faces of the same phenomena: conscious existence. Even in the case of the lack of self-consciousness, emotions drive all throughout the cognitive and social processes. They are the frame from which humans understand the world. And their understanding is not a mental activity, but also a physical one: man knows the laws of the universe through his body. To conceptualize them is a different step.

Meanwhile, and coming back to the real research, the journal devotes specialized issues of different aspects of the research: computational neuroscience and emotions, robot musicians and emotional performance, anthropology of emotions, human-robot interaction, emotions’ physical measurement, electronic art and new ways to think about bodies and emotions, emotional architectures design….there are plenty of topics that must be analyzed in order to make an advance into the field. The journal must also pay attention to any specific technology or technological advance that can improve the understanding of emotions as well as their simulation.

As a final and concluding remark, this author wants to thank to this community of experts for devoting their researches to such a difficult, fascinating, and beautiful topic. Emotions create the meaning of the world, and by working in the future of artificial emotions researchers are at the same time looking for new meanings of reality. The author hopes for a long future for IJSE.

Author(s)/Editor(s) Biography

Jordi Vallverdú, B.Phil, P. Mus, MSc, Ph.D., is Lecturer Professor at UAB (Catalonia, Spain), where he teaches Philosophy and History of Science and Computing. His research is dedicated to the epistemological and cognitive aspects of Philosophy of Computing and Science. He is Editor-in-Chief of the International Journal of Synthetic Emotions (IJSE), and as researcher is member of the EuCogIII and the Convergent Science Network of Biomimetic and Biohybrid Systems Net. He has written several books on computer epistemology and artificial emotions. Very recently, he won a Japanese JSPS fellowship for his studies on HRI at Nishidalab (Kyoto University).

Indices