Supporting the Anticipation Feedback Loop
Marcus Plach1 and Dieter Wallach2
1 Saarland University, 2 University of Basel
Designing a user interface is a demanding cognitive task. In order to develop effective methods and software-based tools which could support and assist designers, a better understanding of the mental activities involved in design is needed ( Carroll 1996a). In investigating cognitive processes in design, researchers have traditionally focussed on the perspective of the individual mind. However, essential constraints of design problems emerge from the need to share and anticipate users' knowledge. In particular, during design sessions, the designer has to take the perspective of the user. That is, the designer has to anticipate the user's (sub)goals, assumptions, and understanding of the information presented at any point in the interaction sequence. Whereas there exist well known formal methods for user modeling in the human-computer interaction literature, e.g. GOMS ( Card, Moranand Newell 1983) or TAG ( Payne and Green 1986), we are concerned with the informal and heuristic way designers deal with this aspect of interface design. For this heuristic mode of mental activity we adopt the term anticipation-feedback loop (AFL) which was originally introduced in the context of intelligent user interfaces by Wahlster ( 1991).
In usability engineering the users' perspective formally comes into play in the form of design or prototype evaluation methods. Besides empirical usability testing, discount usability methods have been proposed to evaluate the usability of (prototype) interfaces (see Nielsen and Mack 1994). In general, two different classes of such methods can be distinguished. On the one hand there are guideline-oriented methods like heuristic evaluation ( Molich and Nielsen 1990). On the other hand, there exist methods like cognitive walkthrough which represent a model-based evaluation technique (e.g. Wharton, Rieman, Lewis, and Polson 1994).