Toward Cognitive Load Inference for Attention Management in Ubiquitous Systems

From not disturbing a focused programmer to entertaining a restless commuter waiting for a train, personal ubiquitous computing devices could greatly enhance their interaction with humans, should these devices only be aware of their users’ cognitive engagement. Despite impressive advances in the inference of human movement, physical activity, routines, and other behavioral aspects, inferring cognitive load remains challenging due to the subtle manifestations of users’ mental engagements via vital signal reactions. These signals are traditionally captured with expensive, obtrusive, and purpose-built equipment, preventing seamless cognitive load inference for human–computer interaction adaptation. In this article, we present our achievements toward enabling large-scale unobtrusive cognitive load inference. Our approaches rely on mining sensor data collected by commodity wearable devices, and software-defined radio-based wireless radars. We also discuss further related research avenues, as well as ethical issues surrounding automatic cognitive load inference.