The Perception of Bodily Expressions of Emotion and the Implications for Computing

Abstract

Increasingly rapid technological developments in robotics, human-computer interaction (HCI), and the computer and video games industry have led to the development of quite sophisticated, intelligent robots and realistic, computer-generated characters. These robots and characters vary in their human-like qualities, from visual and behavioural replicas of humans, to androids, aliens or other beings that have some human characteristics, to creatures that have little in common with humans. The most common human-like qualities evidenced by these robots and characters are intelligence, looks, movements and language. Yet until very recently they were often not endowed with much in the way of emotions, and when they were they typically appeared emotionally underdeveloped, stilted, or false. For example, in James Cameron’s film Terminator 2, Arnold Schwarzenegger plays an emotionless but intelligent humanoid robot. The film character explains that it is equipped with a microchip based on connectionist neural-net architecture. He comes to understand the patterning of human emotional behaviour but, of course, he does not have feelings or emotions. This was also true of other animated or computer-generated characters in films. Even if the characters gave rise to highly emotional video game or film scenes, the users or audience were in agreement that the characters were emotionally impoverished or did not have emotions at all. In large part this was and still is a technological issue: advances made in endowing machines and computer-generated characters with realistic emotions have not kept pace with the rapid advances in endowing such machines and characters with realistic intelligence, looks, movements and language. Nonetheless, given that these robots and characters are designed to interact with humans, and that emotions are so central to successful social interactions, effective implementation of emotions in these robots and characters is vital. The latest developments in HCI include an area that is concerned with implementing emotions with mostly human-like characteristics into HCI technology. About 10 years ago, even a completely new field of computer science emerged that is based on such an attempt, namely “affective computing” (Picard, 1997). Since then, various attempts have been made either to implement emotions into machines or have machines that express emotional reactions. A close examination of these examples reveals two ways of implementing emotions and emotional reactions in HCI technology. First, emotions are seen as