About thirty years after the first designs of personal computers, most information and communication (ICT) systems are still constructed according to the initial paradigm of a single-computer-to-user relationship, assuming and consuming full and undivided user attention. In combination with the ever-growing number of devices, services, and available amounts of data, these systems massively overload users with unrequested and unprocessable amounts of data, resulting in overwhelming information overload and attention scarcity. This work pursues an alternative, truly user-centered “\textit Such systems could elevate the role of users from the limiting bottleneck in information transfer to the central implicit and explicit control entity of interaction design as to, e.g., avoid information overload by adapting information distribution to perception capabilities, prevent attentional errors via the detection and handling of distraction, or actively support of users in paralleled, multi-task scenarios. This work addresses the first necessary step, the sensorial assessment of human attention as an input for computer systems via the analysis of observable indicators of human attention.
Exploring human expressions of attention as an input for interactive systems, the main contributions of this work are: (i) an encompassing, interdisciplinary review and evaluation of potential attention indicators regarding their application potentials in interactive systems and most of all (ii) the development of innovative approaches towards attention estimation, both on methodological and algorithmic levels. The focus in the selection of investigated attention indicators is set on candidates that are established in the literature of Cognitive Psychology regarding reliability and temporal resolution, and which have not been the focus of extensive research in the past. As a result, this work contributes advances in the field of online and real-world applicability of pupillometry as well as an innovative approach towards the analysis of executive behavior. The methodological focus in this work is set on the traditional process of (i) model design, (ii) algorithmic realization, and (iii) empirical, statistical validation (machine learning approaches, statistical analysis) based on test and training data derived from conducted experiments or available reference data. The contributions of this work are expected to encourage and facilitate the current trend of integrating user attention as a fundamental aspect of the design of interactive systems.