Beyond intention - pushing boundaries with incidental interaction

Abstract

Traditional interaction is purposeful, but interactions based on sensing and environmental context are of lower salience and require less explicit attention. The extreme end of this spectrum, but becoming more common, is implicit interaction, which occurs when a user action intended for one purpose is interpreted to achieve some other goal. Because it is an extreme point, incidental interaction acts as a touchstone allowing us to explore the limits of traditional interaction models, task analysis, cognition and software architectures. We find that all either break down or require rethinking in the face of these radically novel interactions. 1 . introduction In traditional computer applications a user was expected to approach the system with a clear intention to perform some activity or achieve some goal. The actions were purposeful and direct and the results were explicitly attended to and evaluated. The design emphasis is on making the affordances of interaction unambiguous and available and ensuring that system feedback and state are clearly visible. However, in many areas of human computer interaction we have seen a growing number of systems and interaction paradigms where user attention and intention is lower. In CSCW the concept of awareness has been central for many years and similarly ambient interfaces emphasise low salience displays of background information. Paradigms such as calm interfaces, tacit and implicit interaction all emphasis output that is non-intrusive and ecologically natural forms of input/control. Whereas the traditional interface was based around controls and input devices, these low attention and natural input paradigms are more closely related to sensing technologies and contextual interpretations. At the extreme end of the spectrum are interactions where actors (users) perform actions for some purpose (say opening a door to enter a room) and the system senses this and uses it for some purpose of which the actors are unaware (perhaps adjusting the air conditioning), but which affects their future interactions with the environment or system. I call this incidental interaction. In this paper we'll look briefly at the definition of incidental interaction and some examples. We'll then see how this challenges many areas of traditional interaction design: the fundamental execution–evaluation cycle implicit in much of HCI, the notion of task and task analysis, the limits of cognitive abilities and the impact of software architecture. Incidental interaction is being used to highlight these issues, but it is evident that these are problematic in many forms of low attention and low intention interaction.