Affective Computing Hernani Costa and Luis Macedo ATCM State of the Art, Theoretical Report CISUC, University of Coimbra Coimbra, Portugal {hpcosta, macedo}@dei.uc.pt Abstract—Affective Computing is currently one of the most active research topics, having increasingly intensive attention. This strong interest is driven by a wide spectrum of promising applications in many areas, such as perceptual interface, virtual reality, affective agents or recommender systems. Affective Computing concerns multidisciplinary background knowledge, such as psychology, cognitive and computer sciences. In this paper, we present an overview of some fields where AC has been applied, as well as some background knowledge in emotion theories. I. I NTRODUCTION The research on affect or emotion can be traced from nowadays to 19 century [1]. Traditionally, “affect” was seldom linked to lifeless machines, being normally studied by psychologists. It is quite new in the recent years that the affect features were captured and processed by the computer. The original version for Affective Computing (AC) was designed to recognise what users are experiencing, when they are interacting with the systems. The emotional responses from these interactions need to be modelled, and the resulting model used to modify the interaction (perhaps to create new emotional experiences or to fix problems like for users relieve frustrated or something that make the interaction needs to be adapted and changed). These ideas really came out from Artificial Intelligence (AI), from the strong idea of the possibility to understand people and to bring emotions. As one of those things that humans should understand in the system. But this was a while ago... The term “Affective Computing” was establish only in 1998, with the book “Affective Computing” by Rosalind Picard (see [2]). Since then, a lot of work was been done, and several branches/modifications of the original version (such as Affective Interaction, see section III) were created. AC can be seen as a branch of computing concerned with the theory and construction of machines which can detect, respond to, and simulate human emotional states. It is an interdisciplinary field spanning the computer sciences, psychology and cognitive science. Nowadays, AC is trying to assign computers the human-like capabilities of interpretation, observation and generation of affect features. An important topic for the harmonious human-computer interaction, with the purpose of increasing the quality of human-computer communication and improving the intelligence of the computer. There are several goals in AC. One is to sense and responder respectfully to human emotion. For instance, if a person is communicating with a technology and he is frustrated or confused, the technology needs to be able to respond differently to that person. If the humans respond differently to the technology, the technology should to do the same! Another goal of AC is to enable people to communicate emotions more clearly. Equally important is to look at the role that emotion plays in intelligence processing. AC is, in part, about understanding how emotions plays vital rules in us for regulating our intention, helping us make good decisions, changing the way we emphasise and prioritise things, organising or figuring out what matters. These roles of emotions are ones that the people do not usually think about as emotional, because usually we are not ‘emotional’ when we do that, it is just some background regular mechanisms that are important for functioning intelligently. With the purpose of increase the quality of human-computer communication, and also improve the intelligence of the computer, affective computing builds an “affect model” based on the various sensors-captured information. This way, it is possible to build a personalised computing system with the capability of perception, interpretation to human’s feeling as well as giving us intelligent, sensitive and friendly responses. AC is a rapidly developing field within industry and science. There is now a great drive to make technologies, such as robotic systems, avatars in service-related human computer interaction, e-learning, game characters, or companion devices more marketable by endowing agents, robots, among others. This new peace of technology brings the ability to recognise and adjust to the user’s feelings as well as the ability to communicate appropriate emotional signals. Computer science believes that the human intelligence can be described to the point that it can be simulated by a machine. Its aim is to design an AI that can process, recognise, interpret and simulate human affect. The ultimate goal being simulated empathy. Even though the machine does not feel emotion it must be able to express and interpret those emotions to interact better with us humans. This paper presents some of the theories behind AC, and its importance in other fields. The remainder of the paper is organised as follows. In section II it is presented theories about emotions. Section III focus in a new research branch of AC, Affective Interaction. Before concluding (section VII) we present three related topics where AC has been applied: artificial agents, multiagents systems and recommender systems (sections IV, V and VI, respectively).