Using Emotion Simulation to Influence User Attitudes and Behaviour

Humans have a natural tendency to unconsciously treat computers as though they are social actors. Many of the human-human social rules that have been well researched and defined in social psychology also seem to apply in human-computer interaction (HCI). This presents an opportunity for computers to use many of the principles and rules from human-human interaction and utilise them for enhancing HCI. One such set of principles that have been extensively investigated and examined in social psychology are influence strategies. A number of different approaches and techniques can be used to influence another’s attitudes and behaviour. These persuasive techniques could potentially be used by computers to influence or change user attitudes and behaviour for beneficial purposes (e.g. smoking cessation or improving exercise behaviour). In this project I will investigate how we respond to interface agents which make use of influence strategies when attempting to change our attitudes and behaviour toward eating. The manipulation of emotions is an essential aspect of attitude and behaviour change and it is proposed that the role of emotion in HCI will also be investigated. Interface designers often incorporate emotional expressions and statements into interface agents, through the use of textual content, synthetic or recorded speech, embodied characters, or multimedia video, without an understanding of the psychological impact on users. Moreover, it is largely unknown how strong user responses are to the synthetic emotional expressions made by computers. I will attempt to address these issues through three different investigations. Firstly, an examination of how users respond to the emotional expressions of different types of interface agents will be undertaken. Secondly, an investigation will be conducted to help understand whether agents that make use of influence strategies are able to influence or change user attitudes and behaviour more effectively than agents which do not. Finally, I will test whether emotional interface agents (which employ influence tactics) are able to influence or change user attitudes and behaviour more effectively than unemotional agents.