Research in affective computing examines the ways in which computers induce, recognize, represent, and express emotions. This volume contains first drafts of papers that were presented at a workshop held during the 1999 I3 conference in Sienna, Italy, as well as an interview with Rosalind Picard. The papers and interview provide an interesting and valuable look at a variety of approaches and theories currently being researched. Overall, the papers are well written and informative. This book will be valuable to anyone interested in affective computing.

The editor’s introduction, “Affective Interactions: Toward a New Generation of Computer Interfaces?” provides some background on affective computing and computer interactions, and describes the content of each article.

“Listen to Your Heart Rate: Counting the Cost of Media Quality,” by Wilson and Sasse, considers the optimum media quality for multimedia conferencing. The goal is to achieve good communications while holding down cost. The authors describe methods for assessing quality through the use of physiological indicators, and discuss the possibility of having an application modify itself based on physiological data.

“Effective Affective in Intelligent Systems--Building on Evidence of Empathy in Teaching and Learning,” by Cooper, Brna, and Martins, discusses the use of intelligent systems in classroom communications. The authors suggest that empathy can be an important factor in teaching and learning. Research suggests that it is possible to improve learning from an artificial tutor by giving the computing agent empathic characteristics. “The Communication of Meaningful Emotional Information for Children Interacting with Virtual Actors,” by George and McIllhagga, deals with research in a virtual play environment for four-to-eight-year-olds. It describes the use of facial expressions in cartoon-like computer agents to enable story construction. “Emotion and Facial Expression,” by Wehrle and Kaiser, presents research into automatic facial recognition techniques, in which the computing agent attempts to understand the user’s emotional state by analyzing facial expressions.

“A Cognitive Approach to Affective User Modeling,” by Martinho, Machado, and Paiva, describes research into cognitive-based affective user modeling. A central idea is that models of the user’s affective states can be used in reacting to user actions. In “Affective Appraisal versus Cognitive Evaluation in Social Emotions and Interactions,” Castelfranchi discusses the constituent elements of emotions. The analysis is based on a contrast between appraisal and evaluation. “An Emotion-Based ‘Conscious’ Software Agent Architecture,” by McCauley, Franklin, and Bogner, discusses the use of software agents to model a psychological theory of consciousness, including emotions. Emotion is seen as an aid in directing user attention.

“Redesigning the Agents’ Decision Machinery,” by Antunes and Coelho, describes an agent architecture based on beliefs, values, and goals. In this approach, emotion is used as a meta-level control mechanism in a setting that involves multiple computing agents interacting in a shared environment. “Artificial Emotion and Emotion Learning: Emotions as Value Judgments,” by Bozninovski, considers emotion as a learning rule in a connectionist learning architecture. The author describes the use of this model in a robot learning system.

“Integrating Models of Personality and Emotions into Lifelike Characters,” by André, Klesen, Gebhard, Allen, and Rist, considers the use of personality and emotions to enhance user-agent interactions. The authors describe three ongoing projects that use on-screen characters to make lifelike computer agents more believable. “Why Should Agents Be Emotional for Entertaining Users? A Critical Analysis,” by Rizzo, focuses on providing users with a richer experience by using emotion with artificial agents. The author presents an analysis and points to the need for empirical research. “Emotional Meaning and Expression in Animated Faces,” by Poggi and Pelachaud, describes research aimed at analyzing facial expressions of emotions by computing agents. The authors examine the idea that emotional content can be found in facial expressions that are not strictly emotional expressions.

Ball and Breese are interested in constructing convincing conversational computing agents. In “Relating Personality and Behavior: Posture and Gestures,” they focus on the use of postures and gestures by the agents, and describe a technique based on Bayesian networks. “Affective Natural Language Generation,” by de Rosis and Grasso, presents research into the use of emotion in natural language generation tools. The work described focuses on examples from the medical explanation domain. It looks at the possibilities for computing agents to recognize and influence user emotional states.

Finally, the interview with Rosalind Picard discusses the general field of affective computing, and some of the important current research areas. Picard also describes her research in this area.