The sensing and measurement of frustration with computers

Abstract

By giving users a way to vent, we transform their frustration into a valuable source of information for adapting interfaces. Drawing from psychophysiology and tactile sensing, we present frustration sensors as a way of incorporating user feedback into interface design processes. This thesis documents the development of designs for several sensors aimed at detecting user frustration with computers. Additionally the thesis explores the design space between active sensors that facilitate the communication of frustration and passive sensors that detect frustration without demanding the user's attention. During evaluations we learned several things: -- Participants liked having devices to communicate frustration. -- The data that was collected during active and passive user interactions can be used for redesigning and adapting systems (either by hand, or automatically). -- User behaved differently during usability problems. In a comparative study of three active designs (Frustrometer, Squeezemouse, and traditional feedback web page) we found that users prefer the Frustrometer to a web feedback page. Preliminary results suggest that frustration-stimulated behavior can also be detected through passive sensors. When combined with other contextual information, these sensors provide a crucial building block in systems that interact and adapt to human behavior by indicating where and when change is needed.